From 996986bed5f2dd97a3d76f584d8eddc1203f8396 Mon Sep 17 00:00:00 2001 From: vanhauser-thc Date: Sat, 5 Sep 2020 12:11:48 +0200 Subject: first batch of changes --- GNUmakefile.gcc_plugin | 167 +++ GNUmakefile.llvm | 499 +++++++ gcc_plugin/GNUmakefile | 170 --- gcc_plugin/Makefile | 159 -- gcc_plugin/README.instrument_list.md | 73 - gcc_plugin/README.md | 158 -- gcc_plugin/afl-gcc-fast.c | 406 ----- gcc_plugin/afl-gcc-pass.so.cc | 601 -------- gcc_plugin/afl-gcc-rt.o.c | 315 ---- instrumentation/LLVMInsTrim.so.cc | 598 ++++++++ instrumentation/MarkNodes.cc | 481 ++++++ instrumentation/MarkNodes.h | 12 + instrumentation/README.cmplog.md | 42 + instrumentation/README.ctx.md | 22 + instrumentation/README.gcc_plugin.md | 158 ++ instrumentation/README.instrim.md | 30 + instrumentation/README.instrument_list.md | 87 ++ instrumentation/README.laf-intel.md | 56 + instrumentation/README.llvm.md | 194 +++ instrumentation/README.lto.md | 290 ++++ instrumentation/README.neverzero.md | 35 + instrumentation/README.ngram.md | 28 + instrumentation/README.persistent_mode.md | 209 +++ instrumentation/README.snapshot.md | 16 + instrumentation/SanitizerCoverageLTO.so.cc | 1503 +++++++++++++++++++ instrumentation/afl-compiler-rt.o.c | 1254 ++++++++++++++++ instrumentation/afl-gcc-pass.so.cc | 601 ++++++++ instrumentation/afl-llvm-common.cc | 575 ++++++++ instrumentation/afl-llvm-common.h | 52 + instrumentation/afl-llvm-dict2file.so.cc | 599 ++++++++ instrumentation/afl-llvm-lto-instrumentation.so.cc | 957 ++++++++++++ instrumentation/afl-llvm-lto-instrumentlist.so.cc | 147 ++ instrumentation/afl-llvm-pass.so.cc | 654 +++++++++ instrumentation/afl-llvm-rt-lto.o.c | 27 + instrumentation/cmplog-instructions-pass.cc | 292 ++++ instrumentation/cmplog-routines-pass.cc | 212 +++ instrumentation/compare-transform-pass.so.cc | 587 ++++++++ instrumentation/llvm-ngram-coverage.h | 18 + instrumentation/split-compares-pass.so.cc | 1356 +++++++++++++++++ instrumentation/split-switches-pass.so.cc | 447 ++++++ llvm_mode/GNUmakefile | 480 ------ llvm_mode/LLVMInsTrim.so.cc | 598 -------- llvm_mode/Makefile | 2 - llvm_mode/MarkNodes.cc | 481 ------ llvm_mode/MarkNodes.h | 12 - llvm_mode/README.cmplog.md | 42 - llvm_mode/README.ctx.md | 22 - llvm_mode/README.instrim.md | 25 - llvm_mode/README.instrument_list.md | 86 -- llvm_mode/README.laf-intel.md | 42 - llvm_mode/README.lto.md | 293 ---- llvm_mode/README.md | 186 --- llvm_mode/README.neverzero.md | 35 - llvm_mode/README.ngram.md | 28 - llvm_mode/README.persistent_mode.md | 209 --- llvm_mode/README.snapshot.md | 16 - llvm_mode/SanitizerCoverageLTO.so.cc | 1503 ------------------- llvm_mode/afl-clang-fast.c | 1143 --------------- llvm_mode/afl-ld-lto.c | 358 ----- llvm_mode/afl-llvm-common.cc | 575 -------- llvm_mode/afl-llvm-common.h | 52 - llvm_mode/afl-llvm-lto-instrumentation.so.cc | 957 ------------ llvm_mode/afl-llvm-lto-instrumentlist.so.cc | 147 -- llvm_mode/afl-llvm-pass.so.cc | 654 --------- llvm_mode/afl-llvm-rt-lto.o.c | 27 - llvm_mode/afl-llvm-rt.o.c | 1244 ---------------- llvm_mode/cmplog-instructions-pass.cc | 292 ---- llvm_mode/cmplog-routines-pass.cc | 212 --- llvm_mode/compare-transform-pass.so.cc | 587 -------- llvm_mode/llvm-ngram-coverage.h | 18 - llvm_mode/split-compares-pass.so.cc | 1356 ----------------- llvm_mode/split-switches-pass.so.cc | 447 ------ src/afl-cc.c | 1544 ++++++++++++++++++++ src/afl-gcc.c | 488 ------- src/afl-ld-lto.c | 358 +++++ 75 files changed, 14107 insertions(+), 14499 deletions(-) create mode 100644 GNUmakefile.gcc_plugin create mode 100644 GNUmakefile.llvm delete mode 100644 gcc_plugin/GNUmakefile delete mode 100644 gcc_plugin/Makefile delete mode 100644 gcc_plugin/README.instrument_list.md delete mode 100644 gcc_plugin/README.md delete mode 100644 gcc_plugin/afl-gcc-fast.c delete mode 100644 gcc_plugin/afl-gcc-pass.so.cc delete mode 100644 gcc_plugin/afl-gcc-rt.o.c create mode 100644 instrumentation/LLVMInsTrim.so.cc create mode 100644 instrumentation/MarkNodes.cc create mode 100644 instrumentation/MarkNodes.h create mode 100644 instrumentation/README.cmplog.md create mode 100644 instrumentation/README.ctx.md create mode 100644 instrumentation/README.gcc_plugin.md create mode 100644 instrumentation/README.instrim.md create mode 100644 instrumentation/README.instrument_list.md create mode 100644 instrumentation/README.laf-intel.md create mode 100644 instrumentation/README.llvm.md create mode 100644 instrumentation/README.lto.md create mode 100644 instrumentation/README.neverzero.md create mode 100644 instrumentation/README.ngram.md create mode 100644 instrumentation/README.persistent_mode.md create mode 100644 instrumentation/README.snapshot.md create mode 100644 instrumentation/SanitizerCoverageLTO.so.cc create mode 100644 instrumentation/afl-compiler-rt.o.c create mode 100644 instrumentation/afl-gcc-pass.so.cc create mode 100644 instrumentation/afl-llvm-common.cc create mode 100644 instrumentation/afl-llvm-common.h create mode 100644 instrumentation/afl-llvm-dict2file.so.cc create mode 100644 instrumentation/afl-llvm-lto-instrumentation.so.cc create mode 100644 instrumentation/afl-llvm-lto-instrumentlist.so.cc create mode 100644 instrumentation/afl-llvm-pass.so.cc create mode 100644 instrumentation/afl-llvm-rt-lto.o.c create mode 100644 instrumentation/cmplog-instructions-pass.cc create mode 100644 instrumentation/cmplog-routines-pass.cc create mode 100644 instrumentation/compare-transform-pass.so.cc create mode 100644 instrumentation/llvm-ngram-coverage.h create mode 100644 instrumentation/split-compares-pass.so.cc create mode 100644 instrumentation/split-switches-pass.so.cc delete mode 100644 llvm_mode/GNUmakefile delete mode 100644 llvm_mode/LLVMInsTrim.so.cc delete mode 100644 llvm_mode/Makefile delete mode 100644 llvm_mode/MarkNodes.cc delete mode 100644 llvm_mode/MarkNodes.h delete mode 100644 llvm_mode/README.cmplog.md delete mode 100644 llvm_mode/README.ctx.md delete mode 100644 llvm_mode/README.instrim.md delete mode 100644 llvm_mode/README.instrument_list.md delete mode 100644 llvm_mode/README.laf-intel.md delete mode 100644 llvm_mode/README.lto.md delete mode 100644 llvm_mode/README.md delete mode 100644 llvm_mode/README.neverzero.md delete mode 100644 llvm_mode/README.ngram.md delete mode 100644 llvm_mode/README.persistent_mode.md delete mode 100644 llvm_mode/README.snapshot.md delete mode 100644 llvm_mode/SanitizerCoverageLTO.so.cc delete mode 100644 llvm_mode/afl-clang-fast.c delete mode 100644 llvm_mode/afl-ld-lto.c delete mode 100644 llvm_mode/afl-llvm-common.cc delete mode 100644 llvm_mode/afl-llvm-common.h delete mode 100644 llvm_mode/afl-llvm-lto-instrumentation.so.cc delete mode 100644 llvm_mode/afl-llvm-lto-instrumentlist.so.cc delete mode 100644 llvm_mode/afl-llvm-pass.so.cc delete mode 100644 llvm_mode/afl-llvm-rt-lto.o.c delete mode 100644 llvm_mode/afl-llvm-rt.o.c delete mode 100644 llvm_mode/cmplog-instructions-pass.cc delete mode 100644 llvm_mode/cmplog-routines-pass.cc delete mode 100644 llvm_mode/compare-transform-pass.so.cc delete mode 100644 llvm_mode/llvm-ngram-coverage.h delete mode 100644 llvm_mode/split-compares-pass.so.cc delete mode 100644 llvm_mode/split-switches-pass.so.cc create mode 100644 src/afl-cc.c delete mode 100644 src/afl-gcc.c create mode 100644 src/afl-ld-lto.c diff --git a/GNUmakefile.gcc_plugin b/GNUmakefile.gcc_plugin new file mode 100644 index 00000000..aeb1ef16 --- /dev/null +++ b/GNUmakefile.gcc_plugin @@ -0,0 +1,167 @@ +# +# american fuzzy lop++ - GCC plugin instrumentation +# ----------------------------------------------- +# +# Written by Austin Seipp and +# Laszlo Szekeres and +# Michal Zalewski and +# Heiko Eißfeldt +# +# GCC integration design is based on the LLVM design, which comes +# from Laszlo Szekeres. +# +# Copyright 2015 Google Inc. All rights reserved. +# Copyright 2019-2020 AFLplusplus Project. All rights reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at: +# +# http://www.apache.org/licenses/LICENSE-2.0 +# + +PREFIX ?= /usr/local +HELPER_PATH ?= $(PREFIX)/lib/afl +BIN_PATH ?= $(PREFIX)/bin +DOC_PATH ?= $(PREFIX)/share/doc/afl +MAN_PATH ?= $(PREFIX)/share/man/man8 + +VERSION = $(shell grep '^$(HASH)define VERSION ' ./config.h | cut -d '"' -f2) + +CFLAGS ?= -O3 -g -funroll-loops -D_FORTIFY_SOURCE=2 +CFLAGS_SAFE := -Wall -Iinclude -Wno-pointer-sign \ + -DAFL_PATH=\"$(HELPER_PATH)\" -DBIN_PATH=\"$(BIN_PATH)\" \ + -DGCC_VERSION=\"$(GCCVER)\" -DGCC_BINDIR=\"$(GCCBINDIR)\" \ + -Wno-unused-function +override CFLAGS += $(CFLAGS_SAFE) + +CXXFLAGS ?= -O3 -g -funroll-loops -D_FORTIFY_SOURCE=2 +CXXEFLAGS := $(CXXFLAGS) -Wall + +CC ?= gcc +CXX ?= g++ + +ifeq "clang" "$(CC)" + CC = gcc + CXX = g++ +endif + +ifeq "clang++" "$(CXX)" + CC = gcc + CXX = g++ +endif + +PLUGIN_FLAGS = -fPIC -fno-rtti -I"$(shell $(CC) -print-file-name=plugin)/include" +HASH=\# + +GCCVER = $(shell $(CC) --version 2>/dev/null | awk 'NR == 1 {print $$NF}') +GCCBINDIR = $(shell dirname `command -v $(CC)` 2>/dev/null ) + +ifeq "$(shell echo '$(HASH)include @$(HASH)include @int main() { int _id = shmget(IPC_PRIVATE, 65536, IPC_CREAT | IPC_EXCL | 0600); shmctl(_id, IPC_RMID, 0); return 0;}' | tr @ '\n' | $(CC) -x c - -o .test2 2>/dev/null && echo 1 || echo 0 ; rm -f .test2 )" "1" + SHMAT_OK=1 +else + SHMAT_OK=0 + override CFLAGS += -DUSEMMAP=1 +endif + +ifeq "$(TEST_MMAP)" "1" + SHMAT_OK=0 + override CFLAGS += -DUSEMMAP=1 +endif + +ifneq "$(shell uname -s)" "Haiku" + LDFLAGS += -lrt +else + CFLAGS_SAFE += -DUSEMMAP=1 +endif + +ifeq "$(shell uname -s)" "SunOS" + PLUGIN_FLAGS += -I/usr/include/gmp +endif + + +PROGS = ./afl-gcc-pass.so + +.PHONY: all +all: test_shm test_deps $(PROGS) test_build all_done + +.PHONY: test_shm +ifeq "$(SHMAT_OK)" "1" +test_shm: + @echo "[+] shmat seems to be working." + @rm -f .test2 +else +test_shm: + @echo "[-] shmat seems not to be working, switching to mmap implementation" +endif + +.PHONY: test_deps +test_deps: + @echo "[*] Checking for working '$(CC)'..." + @type $(CC) >/dev/null 2>&1 || ( echo "[-] Oops, can't find '$(CC)'. Make sure that it's in your \$$PATH (or set \$$CC and \$$CXX)."; exit 1 ) +# @echo "[*] Checking for gcc for plugin support..." +# @$(CC) -v 2>&1 | grep -q -- --enable-plugin || ( echo "[-] Oops, this gcc has not been configured with plugin support."; exit 1 ) + @echo "[*] Checking for gcc plugin development header files..." + @test -d `$(CC) -print-file-name=plugin`/include || ( echo "[-] Oops, can't find gcc header files. Be sure to install 'gcc-X-plugin-dev'."; exit 1 ) + @echo "[*] Checking for './afl-showmap'..." + @test -f ./afl-showmap || ( echo "[-] Oops, can't find './afl-showmap'. Be sure to compile AFL first."; exit 1 ) + @echo "[+] All set and ready to build." + +afl-common.o: ./src/afl-common.c + $(CC) $(CFLAGS) $(CPPFLAGS) -c $< -o $@ $(LDFLAGS) + +./afl-gcc-pass.so: instrumentation/afl-gcc-pass.so.cc | test_deps + $(CXX) $(CXXEFLAGS) $(PLUGIN_FLAGS) -shared $< -o $@ + ln -sf afl-cc afl-gcc-fast + ln -sf afl-cc afl-g++-fast + ln -sf afl-cc.8 afl-gcc-fast.8 + ln -sf afl-cc.8 afl-g++-fast.8 + +.PHONY: test_build +test_build: $(PROGS) + @echo "[*] Testing the CC wrapper and instrumentation output..." + unset AFL_USE_ASAN AFL_USE_MSAN; AFL_QUIET=1 AFL_INST_RATIO=100 AFL_PATH=. AFL_CC=$(CC) ./afl-gcc-fast $(CFLAGS) $(CPPFLAGS) ./test-instr.c -o test-instr $(LDFLAGS) + ASAN_OPTIONS=detect_leaks=0 ./afl-showmap -m none -q -o .test-instr0 ./test-instr ./$@ + @echo .SH NAME >> ./$@ + @echo .B $* >> ./$@ + @echo >> ./$@ + @echo .SH SYNOPSIS >> ./$@ + @./$* -h 2>&1 | head -n 3 | tail -n 1 | sed 's/^\.\///' >> ./$@ + @echo >> ./$@ + @echo .SH OPTIONS >> ./$@ + @echo .nf >> ./$@ + @./$* -h 2>&1 | tail -n +4 >> ./$@ + @echo >> ./$@ + @echo .SH AUTHOR >> ./$@ + @echo "afl++ was written by Michal \"lcamtuf\" Zalewski and is maintained by Marc \"van Hauser\" Heuse , Heiko \"hexcoder-\" Eissfeldt , Andrea Fioraldi and Dominik Maier " >> ./$@ + @echo The homepage of afl++ is: https://github.com/AFLplusplus/AFLplusplus >> ./$@ + @echo >> ./$@ + @echo .SH LICENSE >> ./$@ + @echo Apache License Version 2.0, January 2004 >> ./$@ + ln -sf afl-cc.8 ./afl-g++-fast.8 + +.PHONY: install +install: all + ln -sf afl-cc $${DESTDIR}$(BIN_PATH)/afl-gcc-fast + ln -sf afl-c++ $${DESTDIR}$(BIN_PATH)/afl-g++-fast + install -m 755 ./afl-gcc-pass.so $${DESTDIR}$(HELPER_PATH) + install -m 644 -T instrumentation/README.gcc_plugin.md $${DESTDIR}$(DOC_PATH)/README.gcc_plugin.md + +.PHONY: clean +clean: + rm -f *.o *.so *~ a.out core core.[1-9][0-9]* test-instr .test-instr0 .test-instr1 .test2 + rm -f $(PROGS) afl-common.o ./afl-g++-fast ./afl-g*-fast.8 instrumentation/*.o diff --git a/GNUmakefile.llvm b/GNUmakefile.llvm new file mode 100644 index 00000000..d4502319 --- /dev/null +++ b/GNUmakefile.llvm @@ -0,0 +1,499 @@ +# american fuzzy lop++ - LLVM instrumentation +# ----------------------------------------- +# +# Written by Laszlo Szekeres and +# Michal Zalewski +# +# LLVM integration design comes from Laszlo Szekeres. +# +# Copyright 2015, 2016 Google Inc. All rights reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at: +# +# http://www.apache.org/licenses/LICENSE-2.0 +# + +# For Heiko: +#TEST_MMAP=1 +HASH=\# + +PREFIX ?= /usr/local +HELPER_PATH ?= $(PREFIX)/lib/afl +BIN_PATH ?= $(PREFIX)/bin +DOC_PATH ?= $(PREFIX)/share/doc/afl +MISC_PATH ?= $(PREFIX)/share/afl +MAN_PATH ?= $(PREFIX)/share/man/man8 + +BUILD_DATE ?= $(shell date -u -d "@$(SOURCE_DATE_EPOCH)" "+%Y-%m-%d" 2>/dev/null || date -u -r "$(SOURCE_DATE_EPOCH)" "+%Y-%m-%d" 2>/dev/null || date -u "+%Y-%m-%d") + +VERSION = $(shell grep '^$(HASH)define VERSION ' ./config.h | cut -d '"' -f2) + +ifeq "$(shell uname)" "OpenBSD" + LLVM_CONFIG ?= $(BIN_PATH)/llvm-config + HAS_OPT = $(shell test -x $(BIN_PATH)/opt && echo 0 || echo 1) + ifeq "$(HAS_OPT)" "1" + $(error llvm_mode needs a complete llvm installation (versions 3.4 up to 12) -> e.g. "pkg_add llvm-7.0.1p9") + endif +else + LLVM_CONFIG ?= llvm-config +endif + +LLVMVER = $(shell $(LLVM_CONFIG) --version 2>/dev/null | sed 's/git//' ) +LLVM_MAJOR = $(shell $(LLVM_CONFIG) --version 2>/dev/null | sed 's/\..*//' ) +LLVM_UNSUPPORTED = $(shell $(LLVM_CONFIG) --version 2>/dev/null | egrep -q '^3\.[0-3]|^19' && echo 1 || echo 0 ) +LLVM_NEW_API = $(shell $(LLVM_CONFIG) --version 2>/dev/null | egrep -q '^1[0-9]' && echo 1 || echo 0 ) +LLVM_HAVE_LTO = $(shell $(LLVM_CONFIG) --version 2>/dev/null | egrep -q '^1[1-9]' && echo 1 || echo 0 ) +LLVM_MAJOR = $(shell $(LLVM_CONFIG) --version 2>/dev/null | sed 's/\..*//') +LLVM_BINDIR = $(shell $(LLVM_CONFIG) --bindir 2>/dev/null) +LLVM_LIBDIR = $(shell $(LLVM_CONFIG) --libdir 2>/dev/null) +LLVM_STDCXX = gnu++11 +LLVM_APPLE_XCODE = $(shell clang -v 2>&1 | grep -q Apple && echo 1 || echo 0) +LLVM_LTO = 0 + +ifeq "$(LLVMVER)" "" + $(warning [!] llvm_mode needs llvm-config, which was not found) +endif + +ifeq "$(LLVM_UNSUPPORTED)" "1" + $(warning llvm_mode only supports llvm versions 3.4 up to 12) +endif + +LLVM_TOO_OLD=1 + +ifeq "$(LLVM_MAJOR)" "9" + $(info [+] llvm_mode detected llvm 9, enabling neverZero implementation) + LLVM_TOO_OLD=0 +endif + +ifeq "$(LLVM_NEW_API)" "1" + $(info [+] llvm_mode detected llvm 10+, enabling neverZero implementation and c++14) + LLVM_STDCXX = c++14 + LLVM_TOO_OLD=0 +endif + +ifeq "$(LLVM_TOO_OLD)" "1" + $(info [!] llvm_mode detected an old version of llvm, upgrade to at least 9 or preferable 11!) + $(shell sleep 1) +endif + +ifeq "$(LLVM_HAVE_LTO)" "1" + $(info [+] llvm_mode detected llvm 11+, enabling afl-lto LTO implementation) + LLVM_LTO = 1 + #TEST_MMAP = 1 +endif + +ifeq "$(LLVM_LTO)" "0" + $(info [+] llvm_mode detected llvm < 11, afl-lto LTO will not be build.) +endif + +ifeq "$(LLVM_APPLE_XCODE)" "1" + $(warning llvm_mode will not compile with Xcode clang...) +endif + +# We were using llvm-config --bindir to get the location of clang, but +# this seems to be busted on some distros, so using the one in $PATH is +# probably better. + +CC = $(LLVM_BINDIR)/clang +CXX = $(LLVM_BINDIR)/clang++ + +# llvm-config --bindir may not providing a valid path, so ... +ifeq "$(shell test -e $(CC) || echo 1 )" "1" + # however we must ensure that this is not a "CC=gcc make" + ifeq "$(shell command -v $(CC) 2> /dev/null)" "" + # we do not have a valid CC variable so we try alternatives + ifeq "$(shell test -e '$(BIN_DIR)/clang' && echo 1)" "1" + # we found one in the local install directory, lets use these + CC = $(BIN_DIR)/clang + else + # hope for the best + $(warning we have trouble finding clang - llvm-config is not helping us) + CC = clang + endif + endif +endif +# llvm-config --bindir may not providing a valid path, so ... +ifeq "$(shell test -e $(CXX) || echo 1 )" "1" + # however we must ensure that this is not a "CC=gcc make" + ifeq "$(shell command -v $(CXX) 2> /dev/null)" "" + # we do not have a valid CC variable so we try alternatives + ifeq "$(shell test -e '$(BIN_DIR)/clang++' && echo 1)" "1" + # we found one in the local install directory, lets use these + CXX = $(BIN_DIR)/clang++ + else + # hope for the best + $(warning we have trouble finding clang++ - llvm-config is not helping us) + CXX = clang++ + endif + endif +endif + +# sanity check. +# Are versions of clang --version and llvm-config --version equal? +CLANGVER = $(shell $(CC) --version | sed -E -ne '/^.*version\ (1?[0-9]\.[0-9]\.[0-9]).*/s//\1/p') + +# I disable this because it does not make sense with what we did before (marc) +# We did exactly set these 26 lines above with these values, and it would break +# "CC=gcc make" etc. usages +ifeq "$(findstring clang, $(shell $(CC) --version 2>/dev/null))" "" + CC_SAVE := $(LLVM_BINDIR)/clang +else + CC_SAVE := $(CC) +endif +ifeq "$(findstring clang, $(shell $(CXX) --version 2>/dev/null))" "" + CXX_SAVE := $(LLVM_BINDIR)/clang++ +else + CXX_SAVE := $(CXX) +endif + +CLANG_BIN := $(CC_SAVE) +CLANGPP_BIN := $(CXX_SAVE) + +ifeq "$(CC_SAVE)" "$(LLVM_BINDIR)/clang" + USE_BINDIR = 1 +else + ifeq "$(CXX_SAVE)" "$(LLVM_BINDIR)/clang++" + USE_BINDIR = 1 + else + USE_BINDIR = 0 + endif +endif + +# On old platform we cannot compile with clang because std++ libraries are too +# old. For these we need to use gcc/g++, so if we find REAL_CC and REAL_CXX +# variable we override the compiler variables here +ifneq "$(REAL_CC)" "" +CC = $(REAL_CC) +endif +ifneq "$(REAL_CXX)" "" +CXX = $(REAL_CXX) +endif + +# After we set CC/CXX we can start makefile magic tests + +#ifeq "$(shell echo 'int main() {return 0; }' | $(CC) -x c - -march=native -o .test 2>/dev/null && echo 1 || echo 0 ; rm -f .test )" "1" +# CFLAGS_OPT = -march=native +#endif + +ifeq "$(shell echo 'int main() {return 0; }' | $(CLANG_BIN) -x c - -flto=full -o .test 2>/dev/null && echo 1 || echo 0 ; rm -f .test )" "1" + AFL_CLANG_FLTO ?= -flto=full +else + ifeq "$(shell echo 'int main() {return 0; }' | $(CLANG_BIN) -x c - -flto=thin -o .test 2>/dev/null && echo 1 || echo 0 ; rm -f .test )" "1" + AFL_CLANG_FLTO ?= -flto=thin + else + ifeq "$(shell echo 'int main() {return 0; }' | $(CLANG_BIN) -x c - -flto -o .test 2>/dev/null && echo 1 || echo 0 ; rm -f .test )" "1" + AFL_CLANG_FLTO ?= -flto + endif + endif +endif + +ifeq "$(LLVM_LTO)" "1" + ifneq "$(AFL_CLANG_FLTO)" "" + ifeq "$(AFL_REAL_LD)" "" + ifneq "$(shell readlink $(LLVM_BINDIR)/ld.lld 2>&1)" "" + AFL_REAL_LD = $(LLVM_BINDIR)/ld.lld + else + $(warn ld.lld not found, cannot enable LTO mode) + LLVM_LTO = 0 + endif + endif + else + $(warn clang option -flto is not working - maybe LLVMgold.so not found - cannot enable LTO mode) + LLVM_LTO = 0 + endif +endif + +AFL_CLANG_FUSELD= +ifeq "$(LLVM_LTO)" "1" + ifeq "$(shell echo 'int main() {return 0; }' | $(CLANG_BIN) -x c - -fuse-ld=`command -v ld` -o .test 2>/dev/null && echo 1 || echo 0 ; rm -f .test )" "1" + AFL_CLANG_FUSELD=1 + ifeq "$(shell echo 'int main() {return 0; }' | $(CLANG_BIN) -x c - -fuse-ld=ld.lld --ld-path=$(LLVM_BINDIR)/ld.lld -o .test 2>/dev/null && echo 1 || echo 0 ; rm -f .test )" "1" + AFL_CLANG_LDPATH=1 + endif + else + $(warn -fuse-ld is not working, cannot enable LTO mode) + LLVM_LTO = 0 + endif +endif + +ifeq "$(shell echo 'int main() {return 0; }' | $(CLANG_BIN) -x c - -fdebug-prefix-map=$(CURDIR)=llvm_mode -o .test 2>/dev/null && echo 1 || echo 0 ; rm -f .test )" "1" + AFL_CLANG_DEBUG_PREFIX = -fdebug-prefix-map="$(CURDIR)=llvm_mode" +else + AFL_CLANG_DEBUG_PREFIX = "" +endif + +CFLAGS ?= -O3 -funroll-loops -fPIC -D_FORTIFY_SOURCE=2 +CFLAGS_SAFE := -Wall -g -Wno-pointer-sign -I ./include/ -I ./instrumentation/ \ + -DAFL_PATH=\"$(HELPER_PATH)\" -DBIN_PATH=\"$(BIN_PATH)\" \ + -DLLVM_BINDIR=\"$(LLVM_BINDIR)\" -DVERSION=\"$(VERSION)\" \ + -DLLVM_LIBDIR=\"$(LLVM_LIBDIR)\" -DLLVM_VERSION=\"$(LLVMVER)\" \ + -DAFL_CLANG_FLTO=\"$(AFL_CLANG_FLTO)\" \ + -DAFL_REAL_LD=\"$(AFL_REAL_LD)\" \ + -DAFL_CLANG_LDPATH=\"$(AFL_CLANG_LDPATH)\" \ + -DAFL_CLANG_FUSELD=\"$(AFL_CLANG_FUSELD)\" \ + -DCLANG_BIN=\"$(CLANG_BIN)\" -DCLANGPP_BIN=\"$(CLANGPP_BIN)\" -DUSE_BINDIR=$(USE_BINDIR) -Wno-unused-function \ + $(AFL_CLANG_DEBUG_PREFIX) +override CFLAGS += $(CFLAGS_SAFE) + +ifdef AFL_TRACE_PC + $(info Compile option AFL_TRACE_PC is deprecated, just set AFL_LLVM_INSTRUMENT=PCGUARD to activate when compiling targets ) +endif + +CXXFLAGS ?= -O3 -funroll-loops -fPIC -D_FORTIFY_SOURCE=2 +override CXXFLAGS += -Wall -g -I ./include/ \ + -DVERSION=\"$(VERSION)\" -Wno-variadic-macros + +ifneq "$(shell $(LLVM_CONFIG) --includedir) 2> /dev/null" "" + CLANG_CFL = -I$(shell $(LLVM_CONFIG) --includedir) +endif +ifneq "$(LLVM_CONFIG)" "" + CLANG_CFL += -I$(shell dirname $(LLVM_CONFIG))/../include +endif +CLANG_CPPFL = `$(LLVM_CONFIG) --cxxflags` -fno-rtti -fPIC $(CXXFLAGS) -Wno-deprecated-declarations +CLANG_LFL = `$(LLVM_CONFIG) --ldflags` $(LDFLAGS) + + +# User teor2345 reports that this is required to make things work on MacOS X. +ifeq "$(shell uname)" "Darwin" + CLANG_LFL += -Wl,-flat_namespace -Wl,-undefined,suppress +else + CLANG_CPPFL += -Wl,-znodelete +endif + +ifeq "$(shell uname)" "OpenBSD" + CLANG_LFL += `$(LLVM_CONFIG) --libdir`/libLLVM.so + CLANG_CPPFL += -mno-retpoline + CFLAGS += -mno-retpoline + # Needed for unwind symbols + LDFLAGS += -lc++abi +endif + +ifeq "$(shell echo '$(HASH)include @$(HASH)include @int main() { int _id = shmget(IPC_PRIVATE, 65536, IPC_CREAT | IPC_EXCL | 0600); shmctl(_id, IPC_RMID, 0); return 0;}' | tr @ '\n' | $(CC) -x c - -o .test2 2>/dev/null && echo 1 || echo 0 ; rm -f .test2 )" "1" + SHMAT_OK=1 +else + SHMAT_OK=0 + #CFLAGS+=-DUSEMMAP=1 + LDFLAGS += -Wno-deprecated-declarations +endif + +ifeq "$(TEST_MMAP)" "1" + SHMAT_OK=0 + CFLAGS+=-DUSEMMAP=1 + LDFLAGS += -Wno-deprecated-declarations +endif + +PROGS_ALWAYS = ./afl-cc ./afl-compiler-rt.o ./afl-compiler-rt-32.o ./afl-compiler-rt-64.o +PROGS = $(PROGS_ALWAYS) ./afl-llvm-pass.so ./split-compares-pass.so ./split-switches-pass.so ./cmplog-routines-pass.so ./cmplog-instructions-pass.so ./afl-llvm-dict2file.so ./compare-transform-pass.so ./libLLVMInsTrim.so ./afl-ld-lto ./afl-llvm-lto-instrumentlist.so ./afl-llvm-lto-instrumentation.so ./SanitizerCoverageLTO.so + +# If prerequisites are not given, warn, do not build anything, and exit with code 0 +ifeq "$(LLVMVER)" "" + NO_BUILD = 1 +endif + +ifneq "$(LLVM_UNSUPPORTED)$(LLVM_APPLE_XCODE)" "00" + NO_BUILD = 1 +endif + +ifeq "$(NO_BUILD)" "1" + TARGETS = test_shm $(PROGS_ALWAYS) afl-cc.8 +else + TARGETS = test_shm test_deps $(PROGS) afl-cc.8 test_build all_done +endif + +LLVM_MIN_4_0_1 = $(shell awk 'function tonum(ver, a) {split(ver,a,"."); return a[1]*1000000+a[2]*1000+a[3]} BEGIN { exit tonum(ARGV[1]) >= tonum(ARGV[2]) }' $(LLVMVER) 4.0.1; echo $$?) + +.PHONY: all +all: $(TARGETS) + +.PHONY: test_shm +ifeq "$(SHMAT_OK)" "1" +test_shm: + @echo "[+] shmat seems to be working." + @rm -f .test2 +else +test_shm: + @echo "[-] shmat seems not to be working, switching to mmap implementation" +endif + +.PHONY: no_build +no_build: + @printf "%b\\n" "\\033[0;31mPrerequisites are not met, skipping build llvm_mode\\033[0m" + +.PHONY: test_deps +test_deps: + @echo "[*] Checking for working 'llvm-config'..." + ifneq "$(LLVM_APPLE_XCODE)" "1" + @type $(LLVM_CONFIG) >/dev/null 2>&1 || ( echo "[-] Oops, can't find 'llvm-config'. Install clang or set \$$LLVM_CONFIG or \$$PATH beforehand."; echo " (Sometimes, the binary will be named llvm-config-3.5 or something like that.)"; exit 1 ) + endif + @echo "[*] Checking for working '$(CC)'..." + @type $(CC) >/dev/null 2>&1 || ( echo "[-] Oops, can't find '$(CC)'. Make sure that it's in your \$$PATH (or set \$$CC and \$$CXX)."; exit 1 ) + @echo "[*] Checking for matching versions of '$(CC)' and '$(LLVM_CONFIG)'" +ifneq "$(CLANGVER)" "$(LLVMVER)" + @echo "[!] WARNING: we have llvm-config version $(LLVMVER) and a clang version $(CLANGVER)" +else + @echo "[*] We have llvm-config version $(LLVMVER) with a clang version $(CLANGVER), good." +endif + @echo "[*] Checking for './afl-showmap'..." + @test -f ./afl-showmap || ( echo "[-] Oops, can't find './afl-showmap'. Be sure to compile AFL first."; exit 1 ) + @echo "[+] All set and ready to build." + +instrumentation/afl-common.o: ./src/afl-common.c + $(CC) $(CFLAGS) $(CPPFLAGS) -c $< -o $@ $(LDFLAGS) + +./afl-cc: src/afl-cc.c instrumentation/afl-common.o | test_deps + $(CC) $(CLANG_CFL) $(CFLAGS) $(CPPFLAGS) $< instrumentation/afl-common.o -o $@ -DLLVM_MAJOR=$(LLVM_MAJOR) $(LDFLAGS) -DCFLAGS_OPT=\"$(CFLAGS_OPT)\" + @ln -sf afl-cc ./afl-c++ + @ln -sf afl-cc ./afl-gcc + @ln -sf afl-cc ./afl-g++ + @ln -sf afl-cc ./afl-clang-fast + @ln -sf afl-cc ./afl-clang-fast++ +ifneq "$(AFL_CLANG_FLTO)" "" +ifeq "$(LLVM_LTO)" "1" + @ln -sf afl-cc ./afl-clang-lto + @ln -sf afl-cc ./afl-clang-lto++ + @ln -sf afl-cc ./afl-lto + @ln -sf afl-cc ./afl-lto++ +endif +endif + +instrumentation/afl-llvm-common.o: instrumentation/afl-llvm-common.cc instrumentation/afl-llvm-common.h + $(CXX) $(CFLAGS) $(CPPFLAGS) `$(LLVM_CONFIG) --cxxflags` -fno-rtti -fPIC -std=$(LLVM_STDCXX) -c $< -o $@ + +./libLLVMInsTrim.so: instrumentation/LLVMInsTrim.so.cc instrumentation/MarkNodes.cc instrumentation/afl-llvm-common.o | test_deps + -$(CXX) $(CLANG_CPPFL) -DLLVMInsTrim_EXPORTS -fno-rtti -fPIC -std=$(LLVM_STDCXX) -shared $< instrumentation/MarkNodes.cc -o $@ $(CLANG_LFL) instrumentation/afl-llvm-common.o + +./afl-llvm-pass.so: instrumentation/afl-llvm-pass.so.cc instrumentation/afl-llvm-common.o | test_deps +ifeq "$(LLVM_MIN_4_0_1)" "0" + $(info [!] N-gram branch coverage instrumentation is not available for llvm version $(LLVMVER)) +endif + $(CXX) $(CLANG_CPPFL) -DLLVMInsTrim_EXPORTS -fno-rtti -fPIC -std=$(LLVM_STDCXX) -shared $< -o $@ $(CLANG_LFL) instrumentation/afl-llvm-common.o + +./afl-llvm-lto-instrumentlist.so: instrumentation/afl-llvm-lto-instrumentlist.so.cc instrumentation/afl-llvm-common.o +ifeq "$(LLVM_LTO)" "1" + $(CXX) $(CLANG_CPPFL) -fno-rtti -fPIC -std=$(LLVM_STDCXX) -shared $< -o $@ $(CLANG_LFL) instrumentation/afl-llvm-common.o +endif + +./afl-ld-lto: src/afl-ld-lto.c +ifeq "$(LLVM_LTO)" "1" + $(CC) $(CFLAGS) $(CPPFLAGS) $< -o $@ +endif + +./SanitizerCoverageLTO.so: instrumentation/SanitizerCoverageLTO.so.cc +ifeq "$(LLVM_LTO)" "1" + $(CXX) $(CLANG_CPPFL) -Wno-writable-strings -fno-rtti -fPIC -std=$(LLVM_STDCXX) -shared $< -o $@ $(CLANG_LFL) instrumentation/afl-llvm-common.o +endif + +./afl-llvm-lto-instrumentation.so: instrumentation/afl-llvm-lto-instrumentation.so.cc instrumentation/afl-llvm-common.o +ifeq "$(LLVM_LTO)" "1" + $(CXX) $(CLANG_CPPFL) -Wno-writable-strings -fno-rtti -fPIC -std=$(LLVM_STDCXX) -shared $< -o $@ $(CLANG_LFL) instrumentation/afl-llvm-common.o + $(CLANG_BIN) $(CFLAGS_SAFE) $(CPPFLAGS) -Wno-unused-result -O0 $(AFL_CLANG_FLTO) -fPIC -c instrumentation/afl-llvm-rt-lto.o.c -o ./afl-llvm-rt-lto.o + @$(CLANG_BIN) $(CFLAGS_SAFE) $(CPPFLAGS) -Wno-unused-result -O0 $(AFL_CLANG_FLTO) -m64 -fPIC -c instrumentation/afl-llvm-rt-lto.o.c -o ./afl-llvm-rt-lto-64.o 2>/dev/null; if [ "$$?" = "0" ]; then : ; fi + @$(CLANG_BIN) $(CFLAGS_SAFE) $(CPPFLAGS) -Wno-unused-result -O0 $(AFL_CLANG_FLTO) -m32 -fPIC -c instrumentation/afl-llvm-rt-lto.o.c -o ./afl-llvm-rt-lto-32.o 2>/dev/null; if [ "$$?" = "0" ]; then : ; fi +endif + +# laf +./split-switches-pass.so: instrumentation/split-switches-pass.so.cc instrumentation/afl-llvm-common.o | test_deps + $(CXX) $(CLANG_CPPFL) -shared $< -o $@ $(CLANG_LFL) instrumentation/afl-llvm-common.o +./compare-transform-pass.so: instrumentation/compare-transform-pass.so.cc instrumentation/afl-llvm-common.o | test_deps + $(CXX) $(CLANG_CPPFL) -shared $< -o $@ $(CLANG_LFL) instrumentation/afl-llvm-common.o +./split-compares-pass.so: instrumentation/split-compares-pass.so.cc instrumentation/afl-llvm-common.o | test_deps + $(CXX) $(CLANG_CPPFL) -shared $< -o $@ $(CLANG_LFL) instrumentation/afl-llvm-common.o +# /laf + +./cmplog-routines-pass.so: instrumentation/cmplog-routines-pass.cc instrumentation/afl-llvm-common.o | test_deps + $(CXX) $(CLANG_CPPFL) -shared $< -o $@ $(CLANG_LFL) instrumentation/afl-llvm-common.o + +./cmplog-instructions-pass.so: instrumentation/cmplog-instructions-pass.cc instrumentation/afl-llvm-common.o | test_deps + $(CXX) $(CLANG_CPPFL) -shared $< -o $@ $(CLANG_LFL) instrumentation/afl-llvm-common.o + +afl-llvm-dict2file.so: instrumentation/afl-llvm-dict2file.so.cc instrumentation/afl-llvm-common.o | test_deps + $(CXX) $(CLANG_CPPFL) -shared $< -o $@ $(CLANG_LFL) instrumentation/afl-llvm-common.o + +.PHONY: document +document: + $(CLANG_BIN) -D_AFL_DOCUMENT_MUTATIONS $(CFLAGS_SAFE) $(CPPFLAGS) $(CLANG_CFL) -O3 -Wno-unused-result -fPIC -c instrumentation/afl-compiler-rt.o.c -o ./afl-compiler-rt.o + @$(CLANG_BIN) -D_AFL_DOCUMENT_MUTATIONS $(CFLAGS_SAFE) $(CPPFLAGS) $(CLANG_CFL) -O3 -Wno-unused-result -m32 -fPIC -c instrumentation/afl-compiler-rt.o.c -o ./afl-compiler-rt-32.o 2>/dev/null; if [ "$$?" = "0" ]; then echo "success!"; else echo "failed (that's fine)"; fi + @$(CLANG_BIN) -D_AFL_DOCUMENT_MUTATIONS $(CFLAGS_SAFE) $(CPPFLAGS) $(CLANG_CFL) -O3 -Wno-unused-result -m64 -fPIC -c instrumentation/afl-compiler-rt.o.c -o ./afl-compiler-rt-64.o 2>/dev/null; if [ "$$?" = "0" ]; then echo "success!"; else echo "failed (that's fine)"; fi + +./afl-compiler-rt.o: instrumentation/afl-compiler-rt.o.c | test_deps + $(CLANG_BIN) $(CLANG_CFL) $(CFLAGS_SAFE) $(CPPFLAGS) -O3 -Wno-unused-result -fPIC -c $< -o $@ + ln -sf ./afl-compiler-rt.o ./afl-llvm-rt.o + +./afl-compiler-rt-32.o: instrumentation/afl-compiler-rt.o.c | test_deps + @printf "[*] Building 32-bit variant of the runtime (-m32)... " + @$(CLANG_BIN) $(CLANG_CFL) $(CFLAGS_SAFE) $(CPPFLAGS) -O3 -Wno-unused-result -m32 -fPIC -c $< -o $@ 2>/dev/null; if [ "$$?" = "0" ]; then echo "success!"; else echo "failed (that's fine)"; fi + @test -e ./afl-compiler-rt-32.o && ln -sf ./afl-compiler-rt-32.o ./afl-llvm-rt-32.o + +./afl-compiler-rt-64.o: instrumentation/afl-compiler-rt.o.c | test_deps + @printf "[*] Building 64-bit variant of the runtime (-m64)... " + @$(CLANG_BIN) $(CLANG_CFL) $(CFLAGS_SAFE) $(CPPFLAGS) -O3 -Wno-unused-result -m64 -fPIC -c $< -o $@ 2>/dev/null; if [ "$$?" = "0" ]; then echo "success!"; else echo "failed (that's fine)"; fi + @test -e ./afl-compiler-rt-64.o && ln -sf ./afl-compiler-rt-64.o ./afl-llvm-rt-64.o + +.PHONY: test_build +test_build: $(PROGS) + @echo "[*] Testing the CC wrapper and instrumentation output..." + unset AFL_USE_ASAN AFL_USE_MSAN AFL_INST_RATIO; AFL_QUIET=1 AFL_PATH=. AFL_LLVM_LAF_ALL=1 ./afl-cc $(CFLAGS) $(CPPFLAGS) ./test-instr.c -o test-instr $(LDFLAGS) + ASAN_OPTIONS=detect_leaks=0 ./afl-showmap -m none -q -o .test-instr0 ./test-instr < /dev/null + echo 1 | ASAN_OPTIONS=detect_leaks=0 ./afl-showmap -m none -q -o .test-instr1 ./test-instr + @rm -f test-instr + @cmp -s .test-instr0 .test-instr1; DR="$$?"; rm -f .test-instr0 .test-instr1; if [ "$$DR" = "0" ]; then echo; echo "Oops, the instrumentation does not seem to be behaving correctly!"; echo; echo "Please post to https://github.com/AFLplusplus/AFLplusplus/issues to troubleshoot the issue."; echo; exit 1; fi + @echo "[+] All right, the instrumentation seems to be working!" + +.PHONY: all_done +all_done: test_build + @echo "[+] All done! You can now use './afl-cc' to compile programs." + +.NOTPARALLEL: clean + +.PHONY: install +install: all + @install -d -m 755 $${DESTDIR}$(BIN_PATH) $${DESTDIR}$(HELPER_PATH) $${DESTDIR}$(DOC_PATH) $${DESTDIR}$(MISC_PATH) + @if [ -f ./afl-cc ]; then set -e; install -m 755 ./afl-cc $${DESTDIR}$(BIN_PATH); ln -sf afl-cc $${DESTDIR}$(BIN_PATH)/afl-c++; fi + @if [ -f ./afl-compiler-rt.o ]; then set -e; install -m 755 ./afl-compiler-rt.o $${DESTDIR}$(HELPER_PATH); fi + @if [ -f ./afl-lto ]; then set -e; ln -sf afl-cc $${DESTDIR}$(BIN_PATH)/afl-lto; ln -sf afl-cc $${DESTDIR}$(BIN_PATH)/afl-lto++; ln -sf afl-cc $${DESTDIR}$(BIN_PATH)/afl-clang-lto; ln -sf afl-cc $${DESTDIR}$(BIN_PATH)/afl-clang-lto++; install -m 755 ./afl-llvm-lto-instrumentation.so ./afl-llvm-rt-lto*.o ./afl-llvm-lto-instrumentlist.so $${DESTDIR}$(HELPER_PATH); fi + @if [ -f ./afl-ld-lto ]; then set -e; install -m 755 ./afl-ld-lto $${DESTDIR}$(BIN_PATH); fi + @if [ -f ./afl-compiler-rt-32.o ]; then set -e; install -m 755 ./afl-compiler-rt-32.o $${DESTDIR}$(HELPER_PATH); ln -sf afl-compiler-rt-32.o $${DESTDIR}$(HELPER_PATH)/afl-llvm-rt-32.o; fi + @if [ -f ./afl-compiler-rt-64.o ]; then set -e; install -m 755 ./afl-compiler-rt-64.o $${DESTDIR}$(HELPER_PATH); ln -sf afl-compiler-rt-64.o $${DESTDIR}$(HELPER_PATH)/afl-llvm-rt-64.o; fi + @if [ -f ./compare-transform-pass.so ]; then set -e; install -m 755 ./*.so $${DESTDIR}$(HELPER_PATH); fi + @if [ -f ./compare-transform-pass.so ]; then set -e; ln -sf afl-cc $${DESTDIR}$(BIN_PATH)/afl-clang-fast ; ln -sf ./afl-c++ $${DESTDIR}$(BIN_PATH)/afl-clang-fast++ ; ln -sf afl-cc $${DESTDIR}$(BIN_PATH)/afl-clang ; ln -sf ./afl-c++ $${DESTDIR}$(BIN_PATH)/afl-clang++ ; fi + @if [ -f ./SanitizerCoverageLTO.so ]; then set -e; ln -sf afl-cc $${DESTDIR}$(BIN_PATH)/afl-clang-lto ; ln -sf ./afl-c++ $${DESTDIR}$(BIN_PATH)/afl-clang-lto++ ; fi + set -e; install -m 644 ./dynamic_list.txt $${DESTDIR}$(HELPER_PATH) + install -m 644 instrumentation/README.*.md $${DESTDIR}$(DOC_PATH)/ + +vpath % .. +%.8: % + @echo .TH $* 8 $(BUILD_DATE) "afl++" > ./$@ + @echo .SH NAME >> ./$@ + @printf ".B $* \- " >> ../$@ + @./$* -h 2>&1 | head -n 1 | sed -e "s/$$(printf '\e')[^m]*m//g" >> ../$@ + @echo .B $* >> ./$@ + @echo >> ./$@ + @echo .SH SYNOPSIS >> ./$@ + @./$* -h 2>&1 | head -n 3 | tail -n 1 | sed 's/^\.\///' >> ./$@ + @echo >> ./$@ + @echo .SH OPTIONS >> ./$@ + @echo .nf >> ./$@ + @./$* -h 2>&1 | tail -n +4 >> ./$@ + @echo >> ./$@ + @echo .SH AUTHOR >> ./$@ + @echo "afl++ was written by Michal \"lcamtuf\" Zalewski and is maintained by Marc \"van Hauser\" Heuse , Heiko \"hexcoder-\" Eissfeldt , Andrea Fioraldi and Dominik Maier " >> ./$@ + @echo The homepage of afl++ is: https://github.com/AFLplusplus/AFLplusplus >> ./$@ + @echo >> ./$@ + @echo .SH LICENSE >> ./$@ + @echo Apache License Version 2.0, January 2004 >> ./$@ + @ln -sf afl-cc.8 ./afl-c++.8 +ifneq "$(AFL_CLANG_FLTO)" "" +ifeq "$(LLVM_LTO)" "1" + @ln -sf afl-cc.8 ./afl-clang-lto.8 + @ln -sf afl-cc.8 ./afl-clang-lto++.8 + @ln -sf afl-cc.8 ./afl-lto.8 + @ln -sf afl-cc.8 ./afl-lto++.8 +endif +endif + +.PHONY: clean +clean: + rm -f *.o *.so *~ a.out core core.[1-9][0-9]* .test2 test-instr .test-instr0 .test-instr1 *.dwo + rm -f $(PROGS) afl-common.o ./afl-c++ ./afl-lto ./afl-lto++ ./afl-clang-lto* ./afl-clang-fast* ./afl-clang*.8 ./ld ./afl-ld ./afl-llvm-rt*.o instrumentation/*.o diff --git a/gcc_plugin/GNUmakefile b/gcc_plugin/GNUmakefile deleted file mode 100644 index 625b55fb..00000000 --- a/gcc_plugin/GNUmakefile +++ /dev/null @@ -1,170 +0,0 @@ -# -# american fuzzy lop++ - GCC plugin instrumentation -# ----------------------------------------------- -# -# Written by Austin Seipp and -# Laszlo Szekeres and -# Michal Zalewski and -# Heiko Eißfeldt -# -# GCC integration design is based on the LLVM design, which comes -# from Laszlo Szekeres. -# -# Copyright 2015 Google Inc. All rights reserved. -# Copyright 2019-2020 AFLplusplus Project. All rights reserved. -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at: -# -# http://www.apache.org/licenses/LICENSE-2.0 -# - -PREFIX ?= /usr/local -HELPER_PATH ?= $(PREFIX)/lib/afl -BIN_PATH ?= $(PREFIX)/bin -DOC_PATH ?= $(PREFIX)/share/doc/afl -MAN_PATH ?= $(PREFIX)/share/man/man8 - -VERSION = $(shell grep '^$(HASH)define VERSION ' ../config.h | cut -d '"' -f2) - -CFLAGS ?= -O3 -g -funroll-loops -D_FORTIFY_SOURCE=2 -CFLAGS_SAFE := -Wall -I../include -Wno-pointer-sign \ - -DAFL_PATH=\"$(HELPER_PATH)\" -DBIN_PATH=\"$(BIN_PATH)\" \ - -DGCC_VERSION=\"$(GCCVER)\" -DGCC_BINDIR=\"$(GCCBINDIR)\" \ - -Wno-unused-function -override CFLAGS += $(CFLAGS_SAFE) - -CXXFLAGS ?= -O3 -g -funroll-loops -D_FORTIFY_SOURCE=2 -CXXEFLAGS := $(CXXFLAGS) -Wall - -CC ?= gcc -CXX ?= g++ - -ifeq "clang" "$(CC)" - CC = gcc - CXX = g++ -endif - -ifeq "clang++" "$(CXX)" - CC = gcc - CXX = g++ -endif - -PLUGIN_FLAGS = -fPIC -fno-rtti -I"$(shell $(CC) -print-file-name=plugin)/include" -HASH=\# - -GCCVER = $(shell $(CC) --version 2>/dev/null | awk 'NR == 1 {print $$NF}') -GCCBINDIR = $(shell dirname `command -v $(CC)` 2>/dev/null ) - -ifeq "$(shell echo '$(HASH)include @$(HASH)include @int main() { int _id = shmget(IPC_PRIVATE, 65536, IPC_CREAT | IPC_EXCL | 0600); shmctl(_id, IPC_RMID, 0); return 0;}' | tr @ '\n' | $(CC) -x c - -o .test2 2>/dev/null && echo 1 || echo 0 ; rm -f .test2 )" "1" - SHMAT_OK=1 -else - SHMAT_OK=0 - override CFLAGS += -DUSEMMAP=1 -endif - -ifeq "$(TEST_MMAP)" "1" - SHMAT_OK=0 - override CFLAGS += -DUSEMMAP=1 -endif - -ifneq "$(shell uname -s)" "Haiku" - LDFLAGS += -lrt -else - CFLAGS_SAFE += -DUSEMMAP=1 -endif - -ifeq "$(shell uname -s)" "SunOS" - PLUGIN_FLAGS += -I/usr/include/gmp -endif - - -PROGS = ../afl-gcc-fast ../afl-gcc-pass.so ../afl-gcc-rt.o - - -all: test_shm test_deps $(PROGS) afl-gcc-fast.8 test_build all_done - -ifeq "$(SHMAT_OK)" "1" - -test_shm: - @echo "[+] shmat seems to be working." - @rm -f .test2 - -else - -test_shm: - @echo "[-] shmat seems not to be working, switching to mmap implementation" - -endif - - -test_deps: - @echo "[*] Checking for working '$(CC)'..." - @type $(CC) >/dev/null 2>&1 || ( echo "[-] Oops, can't find '$(CC)'. Make sure that it's in your \$$PATH (or set \$$CC and \$$CXX)."; exit 1 ) -# @echo "[*] Checking for gcc for plugin support..." -# @$(CC) -v 2>&1 | grep -q -- --enable-plugin || ( echo "[-] Oops, this gcc has not been configured with plugin support."; exit 1 ) - @echo "[*] Checking for gcc plugin development header files..." - @test -d `$(CC) -print-file-name=plugin`/include || ( echo "[-] Oops, can't find gcc header files. Be sure to install 'gcc-X-plugin-dev'."; exit 1 ) - @echo "[*] Checking for '../afl-showmap'..." - @test -f ../afl-showmap || ( echo "[-] Oops, can't find '../afl-showmap'. Be sure to compile AFL first."; exit 1 ) - @echo "[+] All set and ready to build." - -afl-common.o: ../src/afl-common.c - $(CC) $(CFLAGS) $(CPPFLAGS) -c $< -o $@ $(LDFLAGS) - -../afl-gcc-fast: afl-gcc-fast.c afl-common.o | test_deps - $(CC) -DAFL_GCC_CC=\"$(CC)\" -DAFL_GCC_CXX=\"$(CXX)\" $(CFLAGS) $(CPPFLAGS) $< afl-common.o -o $@ $(LDFLAGS) - ln -sf afl-gcc-fast ../afl-g++-fast - -../afl-gcc-pass.so: afl-gcc-pass.so.cc | test_deps - $(CXX) $(CXXEFLAGS) $(PLUGIN_FLAGS) -shared $< -o $@ - -../afl-gcc-rt.o: afl-gcc-rt.o.c | test_deps - $(CC) $(CFLAGS_SAFE) $(CPPFLAGS) -fPIC -c $< -o $@ - -test_build: $(PROGS) - @echo "[*] Testing the CC wrapper and instrumentation output..." - unset AFL_USE_ASAN AFL_USE_MSAN; AFL_QUIET=1 AFL_INST_RATIO=100 AFL_PATH=. AFL_CC=$(CC) ../afl-gcc-fast $(CFLAGS) $(CPPFLAGS) ../test-instr.c -o test-instr $(LDFLAGS) -# unset AFL_USE_ASAN AFL_USE_MSAN; AFL_INST_RATIO=100 AFL_PATH=. AFL_CC=$(CC) ../afl-gcc-fast $(CFLAGS) ../test-instr.c -o test-instr $(LDFLAGS) - ASAN_OPTIONS=detect_leaks=0 ../afl-showmap -m none -q -o .test-instr0 ./test-instr ../$@ - @echo .SH NAME >> ../$@ - @echo .B $* >> ../$@ - @echo >> ../$@ - @echo .SH SYNOPSIS >> ../$@ - @../$* -h 2>&1 | head -n 3 | tail -n 1 | sed 's/^\.\///' >> ../$@ - @echo >> ../$@ - @echo .SH OPTIONS >> ../$@ - @echo .nf >> ../$@ - @../$* -h 2>&1 | tail -n +4 >> ../$@ - @echo >> ../$@ - @echo .SH AUTHOR >> ../$@ - @echo "afl++ was written by Michal \"lcamtuf\" Zalewski and is maintained by Marc \"van Hauser\" Heuse , Heiko \"hexcoder-\" Eissfeldt , Andrea Fioraldi and Dominik Maier " >> ../$@ - @echo The homepage of afl++ is: https://github.com/AFLplusplus/AFLplusplus >> ../$@ - @echo >> ../$@ - @echo .SH LICENSE >> ../$@ - @echo Apache License Version 2.0, January 2004 >> ../$@ - ln -sf afl-gcc-fast.8 ../afl-g++-fast.8 - -install: all - install -m 755 ../afl-gcc-fast $${DESTDIR}$(BIN_PATH) - install -m 755 ../afl-gcc-pass.so ../afl-gcc-rt.o $${DESTDIR}$(HELPER_PATH) - install -m 644 -T README.md $${DESTDIR}$(DOC_PATH)/README.gcc_plugin.md - install -m 644 -T README.instrument_list.md $${DESTDIR}$(DOC_PATH)/README.gcc_plugin.instrument_file.md - -clean: - rm -f *.o *.so *~ a.out core core.[1-9][0-9]* test-instr .test-instr0 .test-instr1 .test2 - rm -f $(PROGS) afl-common.o ../afl-g++-fast ../afl-g*-fast.8 diff --git a/gcc_plugin/Makefile b/gcc_plugin/Makefile deleted file mode 100644 index 23477e22..00000000 --- a/gcc_plugin/Makefile +++ /dev/null @@ -1,159 +0,0 @@ -# -# american fuzzy lop++ - GCC plugin instrumentation -# ----------------------------------------------- -# -# Written by Austin Seipp and -# Laszlo Szekeres and -# Michal Zalewski and -# Heiko Eißfeldt -# -# GCC integration design is based on the LLVM design, which comes -# from Laszlo Szekeres. -# -# Copyright 2015 Google Inc. All rights reserved. -# Copyright 2019-2020 AFLplusplus Project. All rights reserved. -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at: -# -# http://www.apache.org/licenses/LICENSE-2.0 -# - -PREFIX ?= /usr/local -HELPER_PATH ?= $(PREFIX)/lib/afl -BIN_PATH ?= $(PREFIX)/bin -DOC_PATH ?= $(PREFIX)/share/doc/afl -MAN_PATH ?= $(PREFIX)/share/man/man8 - -VERSION = $(shell grep '^$(HASH)define VERSION ' ../config.h | cut -d '"' -f2) -VERSION:sh= grep '^$(HASH)define VERSION ' ../config.h | cut -d '"' -f2 - -CFLAGS ?= -O3 -g -funroll-loops -D_FORTIFY_SOURCE=2 -CFLAGS = -Wall -I../include -Wno-pointer-sign \ - -DAFL_PATH=\"$(HELPER_PATH)\" -DBIN_PATH=\"$(BIN_PATH)\" \ - -DGCC_VERSION=\"$(GCCVER)\" -DGCC_BINDIR=\"$(GCCBINDIR)\" \ - -Wno-unused-function - -CXXFLAGS = -O3 -g -funroll-loops -D_FORTIFY_SOURCE=2 -CXXEFLAGS = $(CXXFLAGS) -Wall - -CC = gcc -CXX = g++ - -MYCC=$(CC:clang=gcc) -MYCXX=$(CXX:clang++=g++) - -PLUGIN_PATH = $(shell $(MYCC) -print-file-name=plugin) -PLUGIN_PATH:sh= $(MYCC) -print-file-name=plugin -PLUGIN_FLAGS = -fPIC -fno-rtti -I"$(PLUGIN_PATH)/include" -HASH=\# - -GCCVER = $(shell $(MYCC) --version 2>/dev/null | awk 'NR == 1 {print $$NF}') -GCCVER:sh= gcc --version 2>/dev/null | awk 'NR == 1 {print $$NF}' -GCCBINDIR = $(shell dirname `command -v $(MYCC)` 2>/dev/null ) -GCCBINDIR:sh= dirname `command -v $(MYCC)` 2>/dev/null - -_SHMAT_OK= $(shell echo '$(HASH)include @$(HASH)include @int main() { int _id = shmget(IPC_PRIVATE, 65536, IPC_CREAT | IPC_EXCL | 0600); shmctl(_id, IPC_RMID, 0); return 0;}' | tr @ '\n' | $(MYCC) -x c - -o .test2 2>/dev/null && echo 1 || echo 0 ; rm -f .test2 ) -_SHMAT_OK:sh= echo '$(HASH)include @$(HASH)include @int main() { int _id = shmget(IPC_PRIVATE, 65536, IPC_CREAT | IPC_EXCL | 0600); shmctl(_id, IPC_RMID, 0); return 0;}' | tr @ '\n' | $(MYCC) -x c - -o .test2 2>/dev/null && echo 1 || echo 0 ; rm -f .test2 - -IGNORE_MMAP=$(TEST_MMAP:1=0) -__SHMAT_OK=$(_SHMAT_OK)$(IGNORE_MMAP) -___SHMAT_OK=$(__SHMAT_OK:10=0) -SHMAT_OK=$(___SHMAT_OK:1=1) -_CFLAGS_ADD=$(SHMAT_OK:1=) -CFLAGS_ADD=$(_CFLAGS_ADD:0=-DUSEMMAP=1) - -_LDFLAGS_ADD=$(SHMAT_OK:1=) -LDFLAGS_ADD=$(_LDFLAGS_ADD:0=-lrt) - -CFLAGS += $(CFLAGS_ADD) -LDFLAGS += $(LDFLAGS_ADD) - -PROGS = ../afl-gcc-pass.so ../afl-gcc-fast ../afl-gcc-rt.o - -all: test_shm test_deps $(PROGS) ../afl-gcc-fast.8 test_build all_done - -debug: - @echo _SHMAT_OK = $(_SHMAT_OK) - @echo IGNORE_MMAP = $(IGNORE_MMAP) - @echo __SHMAT_OK = $(__SHMAT_OK) - @echo ___SHMAT_OK = $(___SHMAT_OK) - @echo SHMAT_OK = $(SHMAT_OK) - -test_shm: - @if [ "$(SHMAT_OK)" == "1" ]; then \ - echo "[+] shmat seems to be working."; \ - rm -f .test2; \ - else \ - echo "[-] shmat seems not to be working, switching to mmap implementation"; \ - fi - -test_deps: - @echo "[*] Checking for working '$(MYCC)'..." - @type $(MYCC) >/dev/null 2>&1 || ( echo "[-] Oops, can't find '$(MYCC)'. Make sure that it's in your \$$PATH (or set \$$CC and \$$CXX)."; exit 1 ) -# @echo "[*] Checking for gcc for plugin support..." -# @$(MYCC) -v 2>&1 | grep -q -- --enable-plugin || ( echo "[-] Oops, this gcc has not been configured with plugin support."; exit 1 ) - @echo "[*] Checking for gcc plugin development header files..." - @test -d `$(MYCC) -print-file-name=plugin`/include || ( echo "[-] Oops, can't find gcc header files. Be sure to install 'gcc-X-plugin-dev'."; exit 1 ) - @echo "[*] Checking for '../afl-showmap'..." - @test -f ../afl-showmap || ( echo "[-] Oops, can't find '../afl-showmap'. Be sure to compile AFL first."; exit 1 ) - @echo "[+] All set and ready to build." - -afl-common.o: ../src/afl-common.c - $(MYCC) $(CFLAGS) -c $< -o $@ $(LDFLAGS) - -../afl-gcc-fast: afl-gcc-fast.c afl-common.o - $(MYCC) -DAFL_GCC_CC=\"$(MYCC)\" -DAFL_GCC_CXX=\"$(MYCXX)\" $(CFLAGS) afl-gcc-fast.c afl-common.o -o $@ $(LDFLAGS) - ln -sf afl-gcc-fast ../afl-g++-fast - -../afl-gcc-pass.so: afl-gcc-pass.so.cc - $(MYCXX) $(CXXEFLAGS) $(PLUGIN_FLAGS) -shared afl-gcc-pass.so.cc -o $@ - -../afl-gcc-rt.o: afl-gcc-rt.o.c - $(MYCC) $(CFLAGS) -fPIC -c afl-gcc-rt.o.c -o $@ - -test_build: $(PROGS) - @echo "[*] Testing the CC wrapper and instrumentation output..." - @unset AFL_USE_ASAN AFL_USE_MSAN; AFL_QUIET=1 AFL_INST_RATIO=100 AFL_PATH=. AFL_CC=$(CC) ../afl-gcc-fast $(CFLAGS) ../test-instr.c -o test-instr $(LDFLAGS) -# unset AFL_USE_ASAN AFL_USE_MSAN; AFL_INST_RATIO=100 AFL_PATH=. AFL_CC=$(CC) ../afl-gcc-fast $(CFLAGS) ../test-instr.c -o test-instr $(LDFLAGS) - @ASAN_OPTIONS=detect_leaks=0 ../afl-showmap -m none -q -o .test-instr0 ./test-instr ../$@ - @echo .SH NAME >> ../$@ - @echo .B $* >> ../$@ - @echo >> ../$@ - @echo .SH SYNOPSIS >> ../$@ - @../$* -h 2>&1 | head -n 3 | tail -n 1 | sed 's/^\.\///' >> ../$@ - @echo >> ../$@ - @echo .SH OPTIONS >> ../$@ - @echo .nf >> ../$@ - @../$* -h 2>&1 | tail -n +4 >> ../$@ - @echo >> ../$@ - @echo .SH AUTHOR >> ../$@ - @echo "afl++ was written by Michal \"lcamtuf\" Zalewski and is maintained by Marc \"van Hauser\" Heuse , Heiko \"hexcoder-\" Eissfeldt , Andrea Fioraldi and Dominik Maier " >> ../$@ - @echo The homepage of afl++ is: https://github.com/AFLplusplus/AFLplusplus >> ../$@ - @echo >> ../$@ - @echo .SH LICENSE >> ../$@ - @echo Apache License Version 2.0, January 2004 >> ../$@ - ln -sf afl-gcc-fast.8 ../afl-g++-fast.8 - -install: all - install -m 755 ../afl-gcc-fast $${DESTDIR}$(BIN_PATH) - install -m 755 ../afl-gcc-pass.so ../afl-gcc-rt.o $${DESTDIR}$(HELPER_PATH) - install -m 644 -T README.md $${DESTDIR}$(DOC_PATH)/README.gcc_plugin.md - install -m 644 -T README.instrument_list.md $${DESTDIR}$(DOC_PATH)/README.gcc_plugin.instrument_file.md - -clean: - rm -f *.o *.so *~ a.out core core.[1-9][0-9]* test-instr .test-instr0 .test-instr1 .test2 - rm -f $(PROGS) afl-common.o ../afl-g++-fast ../afl-g*-fast.8 diff --git a/gcc_plugin/README.instrument_list.md b/gcc_plugin/README.instrument_list.md deleted file mode 100644 index d0eaf6ff..00000000 --- a/gcc_plugin/README.instrument_list.md +++ /dev/null @@ -1,73 +0,0 @@ -======================================== -Using afl++ with partial instrumentation -======================================== - - This file describes how you can selectively instrument only the source files - that are interesting to you using the gcc instrumentation provided by - afl++. - - Plugin by hexcoder-. - - -## 1) Description and purpose - -When building and testing complex programs where only a part of the program is -the fuzzing target, it often helps to only instrument the necessary parts of -the program, leaving the rest uninstrumented. This helps to focus the fuzzer -on the important parts of the program, avoiding undesired noise and -disturbance by uninteresting code being exercised. - -For this purpose, I have added a "partial instrumentation" support to the gcc -plugin of AFLFuzz that allows you to specify on a source file level which files -should be compiled with or without instrumentation. - - -## 2) Building the gcc plugin - -The new code is part of the existing afl++ gcc plugin in the gcc_plugin/ -subdirectory. There is nothing specifically to do :) - - -## 3) How to use the partial instrumentation mode - -In order to build with partial instrumentation, you need to build with -afl-gcc-fast and afl-g++-fast respectively. The only required change is -that you need to set the environment variable AFL_GCC_INSTRUMENT_FILE when calling -the compiler. - -The environment variable must point to a file containing all the filenames -that should be instrumented. For matching, the filename that is being compiled -must end in the filename entry contained in this instrument list (to avoid breaking -the matching when absolute paths are used during compilation). - -For example if your source tree looks like this: - -``` -project/ -project/feature_a/a1.cpp -project/feature_a/a2.cpp -project/feature_b/b1.cpp -project/feature_b/b2.cpp -``` - -and you only want to test feature_a, then create a instrument list file containing: - -``` -feature_a/a1.cpp -feature_a/a2.cpp -``` - -However if the instrument list file contains only this, it works as well: - -``` -a1.cpp -a2.cpp -``` - -but it might lead to files being unwantedly instrumented if the same filename -exists somewhere else in the project directories. - -The created instrument list file is then set to AFL_GCC_INSTRUMENT_FILE when you compile -your program. For each file that didn't match the instrument list, the compiler will -issue a warning at the end stating that no blocks were instrumented. If you -didn't intend to instrument that file, then you can safely ignore that warning. diff --git a/gcc_plugin/README.md b/gcc_plugin/README.md deleted file mode 100644 index f762131e..00000000 --- a/gcc_plugin/README.md +++ /dev/null @@ -1,158 +0,0 @@ -# GCC-based instrumentation for afl-fuzz - - (See [../README.md](../README.md) for the general instruction manual.) - (See [../llvm_mode/README.md](../llvm_mode/README.md) for the LLVM-based instrumentation.) - -!!! TODO items are: -!!! => inline instrumentation has to work! -!!! - - -## 1) Introduction - -The code in this directory allows you to instrument programs for AFL using -true compiler-level instrumentation, instead of the more crude -assembly-level rewriting approach taken by afl-gcc and afl-clang. This has -several interesting properties: - - - The compiler can make many optimizations that are hard to pull off when - manually inserting assembly. As a result, some slow, CPU-bound programs will - run up to around faster. - - The gains are less pronounced for fast binaries, where the speed is limited - chiefly by the cost of creating new processes. In such cases, the gain will - probably stay within 10%. - - - The instrumentation is CPU-independent. At least in principle, you should - be able to rely on it to fuzz programs on non-x86 architectures (after - building afl-fuzz with AFL_NOX86=1). - - - Because the feature relies on the internals of GCC, it is gcc-specific - and will *not* work with LLVM (see ../llvm_mode for an alternative). - -Once this implementation is shown to be sufficiently robust and portable, it -will probably replace afl-gcc. For now, it can be built separately and -co-exists with the original code. - -The idea and much of the implementation comes from Laszlo Szekeres. - -## 2) How to use - -In order to leverage this mechanism, you need to have modern enough GCC -(>= version 4.5.0) and the plugin headers installed on your system. That -should be all you need. On Debian machines, these headers can be acquired by -installing the `gcc--plugin-dev` packages. - -To build the instrumentation itself, type 'make'. This will generate binaries -called afl-gcc-fast and afl-g++-fast in the parent directory. -If the CC/CXX have been overridden, those compilers will be used from -those wrappers without using AFL_CXX/AFL_CC settings. -Once this is done, you can instrument third-party code in a way similar to the -standard operating mode of AFL, e.g.: - - CC=/path/to/afl/afl-gcc-fast ./configure [...options...] - make - -Be sure to also include CXX set to afl-g++-fast for C++ code. - -The tool honors roughly the same environmental variables as afl-gcc (see -[env_variables.md](../docs/env_variables.md). This includes AFL_INST_RATIO, AFL_USE_ASAN, -AFL_HARDEN, and AFL_DONT_OPTIMIZE. - -Note: if you want the GCC plugin to be installed on your system for all -users, you need to build it before issuing 'make install' in the parent -directory. - -## 3) Gotchas, feedback, bugs - -This is an early-stage mechanism, so field reports are welcome. You can send bug -reports to . - -## 4) Bonus feature #1: deferred initialization - -AFL tries to optimize performance by executing the targeted binary just once, -stopping it just before main(), and then cloning this "main" process to get -a steady supply of targets to fuzz. - -Although this approach eliminates much of the OS-, linker- and libc-level -costs of executing the program, it does not always help with binaries that -perform other time-consuming initialization steps - say, parsing a large config -file before getting to the fuzzed data. - -In such cases, it's beneficial to initialize the forkserver a bit later, once -most of the initialization work is already done, but before the binary attempts -to read the fuzzed input and parse it; in some cases, this can offer a 10x+ -performance gain. You can implement delayed initialization in LLVM mode in a -fairly simple way. - -First, locate a suitable location in the code where the delayed cloning can -take place. This needs to be done with *extreme* care to avoid breaking the -binary. In particular, the program will probably malfunction if you select -a location after: - - - The creation of any vital threads or child processes - since the forkserver - can't clone them easily. - - - The initialization of timers via setitimer() or equivalent calls. - - - The creation of temporary files, network sockets, offset-sensitive file - descriptors, and similar shared-state resources - but only provided that - their state meaningfully influences the behavior of the program later on. - - - Any access to the fuzzed input, including reading the metadata about its - size. - -With the location selected, add this code in the appropriate spot: - -``` -#ifdef __AFL_HAVE_MANUAL_CONTROL - __AFL_INIT(); -#endif -``` - -You don't need the #ifdef guards, but they will make the program still work as -usual when compiled with a tool other than afl-gcc-fast/afl-clang-fast. - -Finally, recompile the program with afl-gcc-fast (afl-gcc or afl-clang will -*not* generate a deferred-initialization binary) - and you should be all set! - -## 5) Bonus feature #2: persistent mode - -Some libraries provide APIs that are stateless, or whose state can be reset in -between processing different input files. When such a reset is performed, a -single long-lived process can be reused to try out multiple test cases, -eliminating the need for repeated fork() calls and the associated OS overhead. - -The basic structure of the program that does this would be: - -``` - while (__AFL_LOOP(1000)) { - - /* Read input data. */ - /* Call library code to be fuzzed. */ - /* Reset state. */ - - } - - /* Exit normally */ -``` - -The numerical value specified within the loop controls the maximum number -of iterations before AFL will restart the process from scratch. This minimizes -the impact of memory leaks and similar glitches; 1000 is a good starting point. - -A more detailed template is shown in ../examples/persistent_demo/. -Similarly to the previous mode, the feature works only with afl-gcc-fast or -afl-clang-fast; #ifdef guards can be used to suppress it when using other -compilers. - -Note that as with the previous mode, the feature is easy to misuse; if you -do not reset the critical state fully, you may end up with false positives or -waste a whole lot of CPU power doing nothing useful at all. Be particularly -wary of memory leaks and the state of file descriptors. - -When running in this mode, the execution paths will inherently vary a bit -depending on whether the input loop is being entered for the first time or -executed again. To avoid spurious warnings, the feature implies -AFL_NO_VAR_CHECK and hides the "variable path" warnings in the UI. - diff --git a/gcc_plugin/afl-gcc-fast.c b/gcc_plugin/afl-gcc-fast.c deleted file mode 100644 index b1bacfbd..00000000 --- a/gcc_plugin/afl-gcc-fast.c +++ /dev/null @@ -1,406 +0,0 @@ -/* - american fuzzy lop++ - GCC wrapper for GCC plugin - ------------------------------------------------ - - Written by Austin Seipp and - Laszlo Szekeres and - Michal Zalewski - - GCC integration design is based on the LLVM design, which comes - from Laszlo Szekeres. - - Copyright 2015 Google Inc. All rights reserved. - - Licensed under the Apache License, Version 2.0 (the "License"); - you may not use this file except in compliance with the License. - You may obtain a copy of the License at: - - http://www.apache.org/licenses/LICENSE-2.0 - - This program is a drop-in replacement for gcc, similar in most - respects to ../afl-gcc, but with compiler instrumentation through a - plugin. It tries to figure out compilation mode, adds a bunch of - flags, and then calls the real compiler. - - */ - -#define AFL_MAIN - -#include "config.h" -#include "types.h" -#include "debug.h" -#include "common.h" -#include "alloc-inl.h" - -#include -#include -#include -#include - -static u8 * obj_path; /* Path to runtime libraries */ -static u8 **cc_params; /* Parameters passed to the real CC */ -static u32 cc_par_cnt = 1; /* Param count, including argv0 */ -u8 use_stdin = 0; /* dummy */ - -/* Try to find the runtime libraries. If that fails, abort. */ - -static void find_obj(u8 *argv0) { - - u8 *afl_path = getenv("AFL_PATH"); - u8 *slash, *tmp; - - if (afl_path) { - - tmp = alloc_printf("%s/afl-gcc-rt.o", afl_path); - - if (!access(tmp, R_OK)) { - - obj_path = afl_path; - ck_free(tmp); - return; - - } - - ck_free(tmp); - - } - - slash = strrchr(argv0, '/'); - - if (slash) { - - u8 *dir; - - *slash = 0; - dir = ck_strdup(argv0); - *slash = '/'; - - tmp = alloc_printf("%s/afl-gcc-rt.o", dir); - - if (!access(tmp, R_OK)) { - - obj_path = dir; - ck_free(tmp); - return; - - } - - ck_free(tmp); - ck_free(dir); - - } - - if (!access(AFL_PATH "/afl-gcc-rt.o", R_OK)) { - - obj_path = AFL_PATH; - return; - - } - - FATAL( - "Unable to find 'afl-gcc-rt.o' or 'afl-gcc-pass.so'. Please set " - "AFL_PATH"); - -} - -/* Copy argv to cc_params, making the necessary edits. */ - -static void edit_params(u32 argc, char **argv) { - - u8 fortify_set = 0, asan_set = 0, x_set = 0, maybe_linking = 1; - u8 *name; - - cc_params = ck_alloc((argc + 128) * sizeof(u8 *)); - - name = strrchr(argv[0], '/'); - if (!name) - name = argv[0]; - else - ++name; - - if (!strcmp(name, "afl-g++-fast")) { - - u8 *alt_cxx = getenv("AFL_CXX"); - cc_params[0] = alt_cxx && *alt_cxx ? alt_cxx : (u8 *)AFL_GCC_CXX; - - } else if (!strcmp(name, "afl-gcc-fast")) { - - u8 *alt_cc = getenv("AFL_CC"); - cc_params[0] = alt_cc && *alt_cc ? alt_cc : (u8 *)AFL_GCC_CC; - - } else { - - fprintf(stderr, "Name of the binary: %s\n", argv[0]); - FATAL( - "Name of the binary is not a known name, expected afl-(gcc|g++)-fast"); - - } - - char *fplugin_arg = alloc_printf("-fplugin=%s/afl-gcc-pass.so", obj_path); - cc_params[cc_par_cnt++] = fplugin_arg; - - /* Detect stray -v calls from ./configure scripts. */ - - if (argc == 1 && !strcmp(argv[1], "-v")) maybe_linking = 0; - - while (--argc) { - - u8 *cur = *(++argv); - -#if defined(__x86_64__) - if (!strcmp(cur, "-m32")) FATAL("-m32 is not supported"); -#endif - - if (!strcmp(cur, "-x")) x_set = 1; - - if (!strcmp(cur, "-c") || !strcmp(cur, "-S") || !strcmp(cur, "-E") || - !strcmp(cur, "-v")) - maybe_linking = 0; - - if (!strcmp(cur, "-fsanitize=address") || !strcmp(cur, "-fsanitize=memory")) - asan_set = 1; - - if (strstr(cur, "FORTIFY_SOURCE")) fortify_set = 1; - - if (!strcmp(cur, "-shared")) maybe_linking = 0; - - cc_params[cc_par_cnt++] = cur; - - } - - if (getenv("AFL_HARDEN")) { - - cc_params[cc_par_cnt++] = "-fstack-protector-all"; - - if (!fortify_set) cc_params[cc_par_cnt++] = "-D_FORTIFY_SOURCE=2"; - - } - - if (!asan_set) { - - if (getenv("AFL_USE_ASAN")) { - - if (getenv("AFL_USE_MSAN")) FATAL("ASAN and MSAN are mutually exclusive"); - - if (getenv("AFL_HARDEN")) - FATAL("ASAN and AFL_HARDEN are mutually exclusive"); - - cc_params[cc_par_cnt++] = "-U_FORTIFY_SOURCE"; - cc_params[cc_par_cnt++] = "-fsanitize=address"; - - } else if (getenv("AFL_USE_MSAN")) { - - if (getenv("AFL_USE_ASAN")) FATAL("ASAN and MSAN are mutually exclusive"); - - if (getenv("AFL_HARDEN")) - FATAL("MSAN and AFL_HARDEN are mutually exclusive"); - - cc_params[cc_par_cnt++] = "-U_FORTIFY_SOURCE"; - cc_params[cc_par_cnt++] = "-fsanitize=memory"; - - } - - } - - if (getenv("AFL_USE_UBSAN")) { - - cc_params[cc_par_cnt++] = "-fsanitize=undefined"; - cc_params[cc_par_cnt++] = "-fsanitize-undefined-trap-on-error"; - cc_params[cc_par_cnt++] = "-fno-sanitize-recover=all"; - - } - - if (!getenv("AFL_DONT_OPTIMIZE")) { - - cc_params[cc_par_cnt++] = "-g"; - cc_params[cc_par_cnt++] = "-O3"; - cc_params[cc_par_cnt++] = "-funroll-loops"; - - } - - if (getenv("AFL_NO_BUILTIN")) { - - cc_params[cc_par_cnt++] = "-fno-builtin-strcmp"; - cc_params[cc_par_cnt++] = "-fno-builtin-strncmp"; - cc_params[cc_par_cnt++] = "-fno-builtin-strcasecmp"; - cc_params[cc_par_cnt++] = "-fno-builtin-strncasecmp"; - cc_params[cc_par_cnt++] = "-fno-builtin-memcmp"; - cc_params[cc_par_cnt++] = "-fno-builtin-bcmp"; - cc_params[cc_par_cnt++] = "-fno-builtin-strstr"; - cc_params[cc_par_cnt++] = "-fno-builtin-strcasestr"; - - } - -#if defined(USEMMAP) && !defined(__HAIKU__) - cc_params[cc_par_cnt++] = "-lrt"; -#endif - - cc_params[cc_par_cnt++] = "-D__AFL_HAVE_MANUAL_CONTROL=1"; - cc_params[cc_par_cnt++] = "-D__AFL_COMPILER=1"; - cc_params[cc_par_cnt++] = "-DFUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION=1"; - - /* When the user tries to use persistent or deferred forkserver modes by - appending a single line to the program, we want to reliably inject a - signature into the binary (to be picked up by afl-fuzz) and we want - to call a function from the runtime .o file. This is unnecessarily - painful for three reasons: - - 1) We need to convince the compiler not to optimize out the signature. - This is done with __attribute__((used)). - - 2) We need to convince the linker, when called with -Wl,--gc-sections, - not to do the same. This is done by forcing an assignment to a - 'volatile' pointer. - - 3) We need to declare __afl_persistent_loop() in the global namespace, - but doing this within a method in a class is hard - :: and extern "C" - are forbidden and __attribute__((alias(...))) doesn't work. Hence the - __asm__ aliasing trick. - - */ - - cc_params[cc_par_cnt++] = - "-D__AFL_LOOP(_A)=" - "({ static volatile char *_B __attribute__((used)); " - " _B = (char*)\"" PERSIST_SIG - "\"; " -#ifdef __APPLE__ - "int _L(unsigned int) __asm__(\"___afl_persistent_loop\"); " -#else - "int _L(unsigned int) __asm__(\"__afl_persistent_loop\"); " -#endif /* ^__APPLE__ */ - "_L(_A); })"; - - cc_params[cc_par_cnt++] = - "-D__AFL_INIT()=" - "do { static volatile char *_A __attribute__((used)); " - " _A = (char*)\"" DEFER_SIG - "\"; " -#ifdef __APPLE__ - "void _I(void) __asm__(\"___afl_manual_init\"); " -#else - "void _I(void) __asm__(\"__afl_manual_init\"); " -#endif /* ^__APPLE__ */ - "_I(); } while (0)"; - - if (maybe_linking) { - - if (x_set) { - - cc_params[cc_par_cnt++] = "-x"; - cc_params[cc_par_cnt++] = "none"; - - } - - cc_params[cc_par_cnt++] = alloc_printf("%s/afl-gcc-rt.o", obj_path); - - } - - cc_params[cc_par_cnt] = NULL; - -} - -/* Main entry point */ - -int main(int argc, char **argv, char **envp) { - - if (argc < 2 || strcmp(argv[1], "-h") == 0) { - - printf(cCYA - "afl-gcc-fast" VERSION cRST - " initially by , maintainer: hexcoder-\n" - "\n" - "afl-gcc-fast [options]\n" - "\n" - "This is a helper application for afl-fuzz. It serves as a drop-in " - "replacement\n" - "for gcc, letting you recompile third-party code with the required " - "runtime\n" - "instrumentation. A common use pattern would be one of the " - "following:\n\n" - - " CC=%s/afl-gcc-fast ./configure\n" - " CXX=%s/afl-g++-fast ./configure\n\n" - - "In contrast to the traditional afl-gcc tool, this version is " - "implemented as\n" - "a GCC plugin and tends to offer improved performance with slow " - "programs\n" - "(similarly to the LLVM plugin used by afl-clang-fast).\n\n" - - "Environment variables used:\n" - "AFL_CC: path to the C compiler to use\n" - "AFL_CXX: path to the C++ compiler to use\n" - "AFL_PATH: path to instrumenting pass and runtime (afl-gcc-rt.*o)\n" - "AFL_DONT_OPTIMIZE: disable optimization instead of -O3\n" - "AFL_NO_BUILTIN: compile for use with libtokencap.so\n" - "AFL_INST_RATIO: percentage of branches to instrument\n" - "AFL_QUIET: suppress verbose output\n" - "AFL_DEBUG: enable developer debugging output\n" - "AFL_HARDEN: adds code hardening to catch memory bugs\n" - "AFL_USE_ASAN: activate address sanitizer\n" - "AFL_USE_MSAN: activate memory sanitizer\n" - "AFL_USE_UBSAN: activate undefined behaviour sanitizer\n" - "AFL_GCC_INSTRUMENT_FILE: enable selective instrumentation by " - "filename\n" - - "\nafl-gcc-fast was built for gcc %s with the gcc binary path of " - "\"%s\".\n\n", - BIN_PATH, BIN_PATH, GCC_VERSION, GCC_BINDIR); - - exit(1); - - } else if ((isatty(2) && !getenv("AFL_QUIET")) || - - getenv("AFL_DEBUG") != NULL) { - - SAYF(cCYA "afl-gcc-fast" VERSION cRST - " initially by , maintainer: hexcoder-\n"); - - if (getenv("AFL_GCC_INSTRUMENT_FILE") == NULL && - getenv("AFL_GCC_WHITELIST") == NULL) { - - SAYF( - cYEL - "Warning:" cRST - " using afl-gcc-fast without using AFL_GCC_INSTRUMENT_FILE currently " - "produces worse results than afl-gcc. Even better, use " - "llvm_mode for now.\n"); - - } - - } else - - be_quiet = 1; - - u8 *ptr; - if (!be_quiet && - ((ptr = getenv("AFL_MAP_SIZE")) || (ptr = getenv("AFL_MAPSIZE")))) { - - u32 map_size = atoi(ptr); - if (map_size != MAP_SIZE) - WARNF("AFL_MAP_SIZE is not supported by afl-gcc-fast"); - - } - - check_environment_vars(envp); - - find_obj(argv[0]); - - edit_params(argc, argv); - /*if (isatty(2) && !getenv("AFL_QUIET")) { - - printf("Calling \"%s\" with:\n", cc_params[0]); - for(int i=1; i with bits from - Emese Revfy - - Fixed by Heiko Eißfeldt 2019-2020 for AFL++ - - GCC integration design is based on the LLVM design, which comes - from Laszlo Szekeres. Some of the boilerplate code below for - afl_pass to adapt to different GCC versions was taken from Emese - Revfy's Size Overflow plugin for GCC, licensed under the GPLv2/v3. - - (NOTE: this plugin code is under GPLv3, in order to comply with the - GCC runtime library exception, which states that you may distribute - "Target Code" from the compiler under a license of your choice, as - long as the "Compilation Process" is "Eligible", and contains no - GPL-incompatible software in GCC "during the process of - transforming high level code to target code". In this case, the - plugin will be used to generate "Target Code" during the - "Compilation Process", and thus it must be GPLv3 to be "eligible".) - - Copyright (C) 2015 Austin Seipp - - This program is free software: you can redistribute it and/or modify - it under the terms of the GNU General Public License as published by - the Free Software Foundation, either version 3 of the License, or - (at your option) any later version. - - This program is distributed in the hope that it will be useful, - but WITHOUT ANY WARRANTY; without even the implied warranty of - MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - GNU General Public License for more details. - - You should have received a copy of the GNU General Public License - along with this program. If not, see . - - */ - -#define BUILD_INLINE_INST - -#include "../include/config.h" -#include "../include/debug.h" - -/* clear helper macros AFL types pull in, which intervene with gcc-plugin - * headers from GCC-8 */ -#ifdef likely - #undef likely -#endif -#ifdef unlikely - #undef unlikely -#endif - -#include -#include -#include - -#include -#include -#include - -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include - -/* -------------------------------------------------------------------------- */ -/* -- AFL instrumentation pass ---------------------------------------------- */ - -static int be_quiet = 0; -static unsigned int inst_ratio = 100; -static bool inst_ext = true; -static std::list myInstrumentList; - -static unsigned int ext_call_instrument(function *fun) { - - /* Instrument all the things! */ - basic_block bb; - unsigned finst_blocks = 0; - unsigned fcnt_blocks = 0; - - tree fntype = build_function_type_list(void_type_node, /* return */ - uint32_type_node, /* args */ - NULL_TREE); /* done */ - tree fndecl = build_fn_decl("__afl_trace", fntype); - TREE_STATIC(fndecl) = 1; /* Defined elsewhere */ - TREE_PUBLIC(fndecl) = 1; /* Public */ - DECL_EXTERNAL(fndecl) = 1; /* External linkage */ - DECL_ARTIFICIAL(fndecl) = 1; /* Injected by compiler */ - - FOR_EACH_BB_FN(bb, fun) { - - gimple_seq fcall; - gimple_seq seq = NULL; - gimple_stmt_iterator bentry; - ++fcnt_blocks; - - // only instrument if this basic block is the destination of a previous - // basic block that has multiple successors - // this gets rid of ~5-10% of instrumentations that are unnecessary - // result: a little more speed and less map pollution - - int more_than_one = -1; - edge ep; - edge_iterator eip; - - FOR_EACH_EDGE(ep, eip, bb->preds) { - - int count = 0; - if (more_than_one == -1) more_than_one = 0; - - basic_block Pred = ep->src; - edge es; - edge_iterator eis; - FOR_EACH_EDGE(es, eis, Pred->succs) { - - basic_block Succ = es->dest; - if (Succ != NULL) count++; - - } - - if (count > 1) more_than_one = 1; - - } - - if (more_than_one != 1) continue; - - /* Bail on this block if we trip the specified ratio */ - if (R(100) >= inst_ratio) continue; - - /* Make up cur_loc */ - unsigned int rand_loc = R(MAP_SIZE); - tree cur_loc = build_int_cst(uint32_type_node, rand_loc); - - /* Update bitmap via external call */ - /* to quote: - * /+ Trace a basic block with some ID +/ - * void __afl_trace(u32 x); - */ - - fcall = gimple_build_call( - fndecl, 1, - cur_loc); /* generate the function _call_ to above built reference, with - *1* parameter -> the random const for the location */ - gimple_seq_add_stmt(&seq, fcall); /* and insert into a sequence */ - - /* Done - grab the entry to the block and insert sequence */ - bentry = gsi_after_labels(bb); - gsi_insert_seq_before(&bentry, seq, GSI_SAME_STMT); - - ++finst_blocks; - - } - - /* Say something nice. */ - if (!be_quiet) { - - if (!finst_blocks) - WARNF(G_("No instrumentation targets found in " cBRI "%s" cRST), - function_name(fun)); - else if (finst_blocks < fcnt_blocks) - OKF(G_("Instrumented %2u /%2u locations in " cBRI "%s" cRST), - finst_blocks, fcnt_blocks, function_name(fun)); - else - OKF(G_("Instrumented %2u locations in " cBRI "%s" cRST), finst_blocks, - function_name(fun)); - - } - - return 0; - -} - -static unsigned int inline_instrument(function *fun) { - - /* Instrument all the things! */ - basic_block bb; - unsigned finst_blocks = 0; - unsigned fcnt_blocks = 0; - tree one = build_int_cst(unsigned_char_type_node, 1); - // tree zero = build_int_cst(unsigned_char_type_node, 0); - - /* Set up global type declarations */ - tree map_type = build_pointer_type(unsigned_char_type_node); - tree map_ptr_g = - build_decl(UNKNOWN_LOCATION, VAR_DECL, - get_identifier_with_length("__afl_area_ptr", 14), map_type); - TREE_USED(map_ptr_g) = 1; - TREE_STATIC(map_ptr_g) = 1; /* Defined elsewhere */ - DECL_EXTERNAL(map_ptr_g) = 1; /* External linkage */ - DECL_PRESERVE_P(map_ptr_g) = 1; - DECL_ARTIFICIAL(map_ptr_g) = 1; /* Injected by compiler */ - rest_of_decl_compilation(map_ptr_g, 1, 0); - - tree prev_loc_g = build_decl(UNKNOWN_LOCATION, VAR_DECL, - get_identifier_with_length("__afl_prev_loc", 14), - uint32_type_node); - TREE_USED(prev_loc_g) = 1; - TREE_STATIC(prev_loc_g) = 1; /* Defined elsewhere */ - DECL_EXTERNAL(prev_loc_g) = 1; /* External linkage */ - DECL_PRESERVE_P(prev_loc_g) = 1; - DECL_ARTIFICIAL(prev_loc_g) = 1; /* Injected by compiler */ - set_decl_tls_model(prev_loc_g, TLS_MODEL_REAL); /* TLS attribute */ - rest_of_decl_compilation(prev_loc_g, 1, 0); - - FOR_EACH_BB_FN(bb, fun) { - - gimple_seq seq = NULL; - gimple_stmt_iterator bentry; - ++fcnt_blocks; - - // only instrument if this basic block is the destination of a previous - // basic block that has multiple successors - // this gets rid of ~5-10% of instrumentations that are unnecessary - // result: a little more speed and less map pollution - - int more_than_one = -1; - edge ep; - edge_iterator eip; - FOR_EACH_EDGE(ep, eip, bb->preds) { - - int count = 0; - if (more_than_one == -1) more_than_one = 0; - - basic_block Pred = ep->src; - edge es; - edge_iterator eis; - FOR_EACH_EDGE(es, eis, Pred->succs) { - - basic_block Succ = es->dest; - if (Succ != NULL) count++; - - } - - if (count > 1) more_than_one = 1; - - } - - if (more_than_one != 1) continue; - - /* Bail on this block if we trip the specified ratio */ - if (R(100) >= inst_ratio) continue; - - /* Make up cur_loc */ - - unsigned int rand_loc = R(MAP_SIZE); - tree cur_loc = build_int_cst(uint32_type_node, rand_loc); - - /* Load prev_loc, xor with cur_loc */ - // gimple_assign - tree prev_loc = create_tmp_var_raw(uint32_type_node, "prev_loc"); - gassign *g = gimple_build_assign(prev_loc, VAR_DECL, prev_loc_g); - gimple_seq_add_stmt(&seq, g); // load prev_loc - update_stmt(g); - - // gimple_assign - tree area_off = create_tmp_var_raw(uint32_type_node, "area_off"); - g = gimple_build_assign(area_off, BIT_XOR_EXPR, prev_loc, cur_loc); - gimple_seq_add_stmt(&seq, g); // area_off = prev_loc ^ cur_loc - update_stmt(g); - - /* Update bitmap */ - - // gimple_assign - tree map_ptr = create_tmp_var(map_type, "map_ptr"); - tree map_ptr2 = create_tmp_var(map_type, "map_ptr2"); - - g = gimple_build_assign(map_ptr, map_ptr_g); - gimple_seq_add_stmt(&seq, g); // map_ptr = __afl_area_ptr - update_stmt(g); - -#if 1 - #if 0 - tree addr = build2(ADDR_EXPR, map_type, map_ptr, area_off); - g = gimple_build_assign(map_ptr2, MODIFY_EXPR, addr); - gimple_seq_add_stmt(&seq, g); // map_ptr2 = map_ptr + area_off - update_stmt(g); - #else - g = gimple_build_assign(map_ptr2, PLUS_EXPR, map_ptr, area_off); - gimple_seq_add_stmt(&seq, g); // map_ptr2 = map_ptr + area_off - update_stmt(g); - #endif - - // gimple_assign - tree tmp1 = create_tmp_var_raw(unsigned_char_type_node, "tmp1"); - g = gimple_build_assign(tmp1, MEM_REF, map_ptr2); - gimple_seq_add_stmt(&seq, g); // tmp1 = *map_ptr2 - update_stmt(g); -#else - tree atIndex = build2(PLUS_EXPR, uint32_type_node, map_ptr, area_off); - tree array_address = build1(ADDR_EXPR, map_type, atIndex); - tree array_access = build1(INDIRECT_REF, map_type, array_address); - tree tmp1 = create_tmp_var(unsigned_char_type_node, "tmp1"); - g = gimple_build_assign(tmp1, array_access); - gimple_seq_add_stmt(&seq, g); // tmp1 = *(map_ptr + area_off) - update_stmt(g); -#endif - // gimple_assign - tree tmp2 = create_tmp_var_raw(unsigned_char_type_node, "tmp2"); - g = gimple_build_assign(tmp2, PLUS_EXPR, tmp1, one); - gimple_seq_add_stmt(&seq, g); // tmp2 = tmp1 + 1 - update_stmt(g); - - // TODO: neverZero: here we have to check if tmp3 == 0 - // and add 1 if so - - // gimple_assign - // tree map_ptr3 = create_tmp_var_raw(map_type, "map_ptr3"); - g = gimple_build_assign(map_ptr2, INDIRECT_REF, tmp2); - gimple_seq_add_stmt(&seq, g); // *map_ptr2 = tmp2 - update_stmt(g); - - /* Set prev_loc to cur_loc >> 1 */ - - // gimple_assign - tree shifted_loc = build_int_cst(TREE_TYPE(prev_loc_g), rand_loc >> 1); - tree prev_loc2 = create_tmp_var_raw(uint32_type_node, "prev_loc2"); - g = gimple_build_assign(prev_loc2, shifted_loc); - gimple_seq_add_stmt(&seq, g); // __afl_prev_loc = cur_loc >> 1 - update_stmt(g); - g = gimple_build_assign(prev_loc_g, prev_loc2); - gimple_seq_add_stmt(&seq, g); // __afl_prev_loc = cur_loc >> 1 - update_stmt(g); - - /* Done - grab the entry to the block and insert sequence */ - - bentry = gsi_after_labels(bb); - gsi_insert_seq_before(&bentry, seq, GSI_NEW_STMT); - - ++finst_blocks; - - } - - /* Say something nice. */ - if (!be_quiet) { - - if (!finst_blocks) - WARNF(G_("No instrumentation targets found in " cBRI "%s" cRST), - function_name(fun)); - else if (finst_blocks < fcnt_blocks) - OKF(G_("Instrumented %2u /%2u locations in " cBRI "%s" cRST), - finst_blocks, fcnt_blocks, function_name(fun)); - else - OKF(G_("Instrumented %2u locations in " cBRI "%s" cRST), finst_blocks, - function_name(fun)); - - } - - return 0; - -} - -/* -------------------------------------------------------------------------- */ -/* -- Boilerplate and initialization ---------------------------------------- */ - -static const struct pass_data afl_pass_data = { - - .type = GIMPLE_PASS, - .name = "afl-inst", - .optinfo_flags = OPTGROUP_NONE, - - .tv_id = TV_NONE, - .properties_required = 0, - .properties_provided = 0, - .properties_destroyed = 0, - .todo_flags_start = 0, - // NOTE(aseipp): it's very, very important to include - // at least 'TODO_update_ssa' here so that GCC will - // properly update the resulting SSA form, e.g., to - // include new PHI nodes for newly added symbols or - // names. Do not remove this. Do not taunt Happy Fun - // Ball. - .todo_flags_finish = TODO_update_ssa | TODO_verify_il | TODO_cleanup_cfg, - -}; - -namespace { - -class afl_pass : public gimple_opt_pass { - - private: - bool do_ext_call; - - public: - afl_pass(bool ext_call, gcc::context *g) - : gimple_opt_pass(afl_pass_data, g), do_ext_call(ext_call) { - - } - - unsigned int execute(function *fun) override { - - if (!myInstrumentList.empty()) { - - bool instrumentBlock = false; - std::string instFilename; - unsigned int instLine = 0; - - /* EXPR_FILENAME - This macro returns the name of the file in which the entity was declared, - as a char*. For an entity declared implicitly by the compiler (like - __builtin_ memcpy), this will be the string "". - */ - const char *fname = DECL_SOURCE_FILE(fun->decl); - - if (0 != strncmp("", fname, 10) && - 0 != strncmp("", fname, 10)) { - - instFilename = fname; - instLine = DECL_SOURCE_LINE(fun->decl); - - /* Continue only if we know where we actually are */ - if (!instFilename.empty()) { - - for (std::list::iterator it = myInstrumentList.begin(); - it != myInstrumentList.end(); ++it) { - - /* We don't check for filename equality here because - * filenames might actually be full paths. Instead we - * check that the actual filename ends in the filename - * specified in the list. */ - if (instFilename.length() >= it->length()) { - - if (instFilename.compare(instFilename.length() - it->length(), - it->length(), *it) == 0) { - - instrumentBlock = true; - break; - - } - - } - - } - - } - - } - - /* Either we couldn't figure out our location or the location is - * not in the instrument list, so we skip instrumentation. */ - if (!instrumentBlock) { - - if (!be_quiet) { - - if (!instFilename.empty()) - SAYF(cYEL "[!] " cBRI - "Not in instrument list, skipping %s line %u...\n", - instFilename.c_str(), instLine); - else - SAYF(cYEL "[!] " cBRI "No filename information found, skipping it"); - - } - - return 0; - - } - - } - - return do_ext_call ? ext_call_instrument(fun) : inline_instrument(fun); - - } - -}; /* class afl_pass */ - -} // namespace - -static struct opt_pass *make_afl_pass(bool ext_call, gcc::context *ctxt) { - - return new afl_pass(ext_call, ctxt); - -} - -/* -------------------------------------------------------------------------- */ -/* -- Initialization -------------------------------------------------------- */ - -int plugin_is_GPL_compatible = 1; - -static struct plugin_info afl_plugin_info = { - - .version = "20200519", - .help = "AFL++ gcc plugin\n", - -}; - -int plugin_init(struct plugin_name_args * plugin_info, - struct plugin_gcc_version *version) { - - struct register_pass_info afl_pass_info; - struct timeval tv; - struct timezone tz; - u32 rand_seed; - - /* Setup random() so we get Actually Random(TM) outputs from R() */ - gettimeofday(&tv, &tz); - rand_seed = tv.tv_sec ^ tv.tv_usec ^ getpid(); - SR(rand_seed); - - /* Pass information */ - afl_pass_info.pass = make_afl_pass(inst_ext, g); - afl_pass_info.reference_pass_name = "ssa"; - afl_pass_info.ref_pass_instance_number = 1; - afl_pass_info.pos_op = PASS_POS_INSERT_AFTER; - - if (!plugin_default_version_check(version, &gcc_version)) { - - FATAL(G_("Incompatible gcc/plugin versions! Expected GCC %d.%d"), - GCCPLUGIN_VERSION_MAJOR, GCCPLUGIN_VERSION_MINOR); - - } - - /* Show a banner */ - if ((isatty(2) && !getenv("AFL_QUIET")) || getenv("AFL_DEBUG") != NULL) { - - SAYF(G_(cCYA "afl-gcc-pass" VERSION cRST - " initially by , maintainer: hexcoder-\n")); - - } else - - be_quiet = 1; - - /* Decide instrumentation ratio */ - char *inst_ratio_str = getenv("AFL_INST_RATIO"); - - if (inst_ratio_str) { - - if (sscanf(inst_ratio_str, "%u", &inst_ratio) != 1 || !inst_ratio || - inst_ratio > 100) - FATAL(G_("Bad value of AFL_INST_RATIO (must be between 1 and 100)")); - else { - - if (!be_quiet) - ACTF(G_("%s instrumentation at ratio of %u%% in %s mode."), - inst_ext ? G_("Call-based") : G_("Inline"), inst_ratio, - getenv("AFL_HARDEN") ? G_("hardened") : G_("non-hardened")); - - } - - } - - char *instInstrumentListFilename = getenv("AFL_GCC_INSTRUMENT_FILE"); - if (!instInstrumentListFilename) - instInstrumentListFilename = getenv("AFL_GCC_WHITELIST"); - if (instInstrumentListFilename) { - - std::string line; - std::ifstream fileStream; - fileStream.open(instInstrumentListFilename); - if (!fileStream) PFATAL("Unable to open AFL_GCC_INSTRUMENT_FILE"); - getline(fileStream, line); - while (fileStream) { - - myInstrumentList.push_back(line); - getline(fileStream, line); - - } - - } else if (!be_quiet && (getenv("AFL_LLVM_WHITELIST") || - - getenv("AFL_LLVM_INSTRUMENT_FILE"))) { - - SAYF(cYEL "[-] " cRST - "AFL_LLVM_INSTRUMENT_FILE environment variable detected - did " - "you mean AFL_GCC_INSTRUMENT_FILE?\n"); - - } - - /* Go go gadget */ - register_callback(plugin_info->base_name, PLUGIN_INFO, NULL, - &afl_plugin_info); - register_callback(plugin_info->base_name, PLUGIN_PASS_MANAGER_SETUP, NULL, - &afl_pass_info); - return 0; - -} - diff --git a/gcc_plugin/afl-gcc-rt.o.c b/gcc_plugin/afl-gcc-rt.o.c deleted file mode 100644 index 49a03cae..00000000 --- a/gcc_plugin/afl-gcc-rt.o.c +++ /dev/null @@ -1,315 +0,0 @@ -/* - american fuzzy lop++ - GCC plugin instrumentation bootstrap - --------------------------------------------------------- - - Written by Austin Seipp and - Laszlo Szekeres and - Michal Zalewski - - GCC integration design is based on the LLVM design, which comes - from Laszlo Szekeres. - - Licensed under the Apache License, Version 2.0 (the "License"); - you may not use this file except in compliance with the License. - You may obtain a copy of the License at: - - http://www.apache.org/licenses/LICENSE-2.0 - - This code is the rewrite of afl-as.h's main_payload. - -*/ - -#ifdef __ANDROID__ - #include "android-ashmem.h" -#endif -#include "../config.h" -#include "../types.h" - -#ifdef USEMMAP - #include -#endif -#include -#include -#include -#include -#include - -#include -#ifndef USEMMAP - #include -#endif -#include -#include - -#include -#include - -/* Globals needed by the injected instrumentation. The __afl_area_initial region - is used for instrumentation output before __afl_map_shm() has a chance to - run. It will end up as .comm, so it shouldn't be too wasteful. */ - -u8 __afl_area_initial[MAP_SIZE]; -u8 *__afl_area_ptr = __afl_area_initial; - -#ifdef __ANDROID__ -u32 __afl_prev_loc; -u32 __afl_final_loc; -#else -__thread u32 __afl_prev_loc; -__thread u32 __afl_final_loc; -#endif - -/* Trace a basic block with some ID */ -void __afl_trace(const u32 x) { - -#if 1 /* enable for neverZero feature. */ - __afl_area_ptr[__afl_prev_loc ^ x] += - 1 + ((u8)(1 + __afl_area_ptr[__afl_prev_loc ^ x]) == 0); -#else - ++__afl_area_ptr[__afl_prev_loc ^ x]; -#endif - - __afl_prev_loc = (x >> 1); - return; - -} - -/* Running in persistent mode? */ - -static u8 is_persistent; - -/* SHM setup. */ - -static void __afl_map_shm(void) { - - u8 *id_str = getenv(SHM_ENV_VAR); - - /* If we're running under AFL, attach to the appropriate region, replacing the - early-stage __afl_area_initial region that is needed to allow some really - hacky .init code to work correctly in projects such as OpenSSL. */ - - if (id_str) { - -#ifdef USEMMAP - const char * shm_file_path = id_str; - int shm_fd = -1; - unsigned char *shm_base = NULL; - - /* create the shared memory segment as if it was a file */ - shm_fd = shm_open(shm_file_path, O_RDWR, 0600); - if (shm_fd == -1) { - - fprintf(stderr, "shm_open() failed\n"); - exit(1); - - } - - /* map the shared memory segment to the address space of the process */ - shm_base = mmap(0, MAP_SIZE, PROT_READ | PROT_WRITE, MAP_SHARED, shm_fd, 0); - if (shm_base == MAP_FAILED) { - - close(shm_fd); - shm_fd = -1; - - fprintf(stderr, "mmap() failed\n"); - exit(2); - - } - - __afl_area_ptr = shm_base; -#else - u32 shm_id = atoi(id_str); - - __afl_area_ptr = shmat(shm_id, NULL, 0); -#endif - - /* Whooooops. */ - - if (__afl_area_ptr == (void *)-1) exit(1); - - /* Write something into the bitmap so that even with low AFL_INST_RATIO, - our parent doesn't give up on us. */ - - __afl_area_ptr[0] = 1; - - } - -} - -/* Fork server logic. */ - -static void __afl_start_forkserver(void) { - - u8 tmp[4] = {0, 0, 0, 0}; - u32 map_size = MAP_SIZE; - s32 child_pid; - - u8 child_stopped = 0; - - void (*old_sigchld_handler)(int) = signal(SIGCHLD, SIG_DFL); - - /* Phone home and tell the parent that we're OK. If parent isn't there, - assume we're not running in forkserver mode and just execute program. */ - - if (MAP_SIZE <= 0x800000) { - - map_size = (FS_OPT_ENABLED | FS_OPT_MAPSIZE | FS_OPT_SET_MAPSIZE(MAP_SIZE)); - memcpy(tmp, &map_size, 4); - - } - - if (write(FORKSRV_FD + 1, tmp, 4) != 4) return; - - while (1) { - - u32 was_killed; - int status; - - /* Wait for parent by reading from the pipe. Abort if read fails. */ - - if (read(FORKSRV_FD, &was_killed, 4) != 4) exit(1); - - /* If we stopped the child in persistent mode, but there was a race - condition and afl-fuzz already issued SIGKILL, write off the old - process. */ - - if (child_stopped && was_killed) { - - child_stopped = 0; - if (waitpid(child_pid, &status, 0) < 0) exit(1); - - } - - if (!child_stopped) { - - /* Once woken up, create a clone of our process. */ - - child_pid = fork(); - if (child_pid < 0) exit(1); - - /* In child process: close fds, resume execution. */ - - if (!child_pid) { - - signal(SIGCHLD, old_sigchld_handler); - - close(FORKSRV_FD); - close(FORKSRV_FD + 1); - return; - - } - - } else { - - /* Special handling for persistent mode: if the child is alive but - currently stopped, simply restart it with SIGCONT. */ - - kill(child_pid, SIGCONT); - child_stopped = 0; - - } - - /* In parent process: write PID to pipe, then wait for child. */ - - if (write(FORKSRV_FD + 1, &child_pid, 4) != 4) exit(1); - - if (waitpid(child_pid, &status, is_persistent ? WUNTRACED : 0) < 0) exit(1); - - /* In persistent mode, the child stops itself with SIGSTOP to indicate - a successful run. In this case, we want to wake it up without forking - again. */ - - if (WIFSTOPPED(status)) child_stopped = 1; - - /* Relay wait status to pipe, then loop back. */ - - if (write(FORKSRV_FD + 1, &status, 4) != 4) exit(1); - - } - -} - -/* A simplified persistent mode handler, used as explained in README.md. */ - -int __afl_persistent_loop(unsigned int max_cnt) { - - static u8 first_pass = 1; - static u32 cycle_cnt; - - if (first_pass) { - - /* Make sure that every iteration of __AFL_LOOP() starts with a clean slate. - On subsequent calls, the parent will take care of that, but on the first - iteration, it's our job to erase any trace of whatever happened - before the loop. */ - - if (is_persistent) { - - memset(__afl_area_ptr, 0, MAP_SIZE); - __afl_area_ptr[0] = 1; - __afl_prev_loc = 0; - - } - - cycle_cnt = max_cnt; - first_pass = 0; - return 1; - - } - - if (is_persistent) { - - if (--cycle_cnt) { - - raise(SIGSTOP); - - __afl_area_ptr[0] = 1; - __afl_prev_loc = 0; - - return 1; - - } else { - - /* When exiting __AFL_LOOP(), make sure that the subsequent code that - follows the loop is not traced. We do that by pivoting back to the - dummy output region. */ - - __afl_area_ptr = __afl_area_initial; - - } - - } - - return 0; - -} - -/* This one can be called from user code when deferred forkserver mode - is enabled. */ - -void __afl_manual_init(void) { - - static u8 init_done; - - if (!init_done) { - - __afl_map_shm(); - __afl_start_forkserver(); - init_done = 1; - - } - -} - -/* Proper initialization routine. */ - -__attribute__((constructor(101))) void __afl_auto_init(void) { - - is_persistent = !!getenv(PERSIST_ENV_VAR); - - if (getenv(DEFER_ENV_VAR)) return; - - __afl_manual_init(); - -} - diff --git a/instrumentation/LLVMInsTrim.so.cc b/instrumentation/LLVMInsTrim.so.cc new file mode 100644 index 00000000..61a420ba --- /dev/null +++ b/instrumentation/LLVMInsTrim.so.cc @@ -0,0 +1,598 @@ +#include +#include +#include +#include + +#include "llvm/Config/llvm-config.h" +#if LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR < 5 +typedef long double max_align_t; +#endif + +#include "llvm/ADT/DenseMap.h" +#include "llvm/ADT/DenseSet.h" +#if LLVM_VERSION_MAJOR > 3 || \ + (LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR > 4) + #include "llvm/IR/CFG.h" + #include "llvm/IR/Dominators.h" + #include "llvm/IR/DebugInfo.h" +#else + #include "llvm/Support/CFG.h" + #include "llvm/Analysis/Dominators.h" + #include "llvm/DebugInfo.h" +#endif +#include "llvm/IR/IRBuilder.h" +#include "llvm/IR/Instructions.h" +#include "llvm/IR/LegacyPassManager.h" +#include "llvm/IR/Module.h" +#include "llvm/Pass.h" +#include "llvm/Support/raw_ostream.h" +#include "llvm/Support/CommandLine.h" +#include "llvm/Transforms/IPO/PassManagerBuilder.h" +#include "llvm/Transforms/Utils/BasicBlockUtils.h" +#include "llvm/IR/BasicBlock.h" +#include +#include +#include +#include +#include + +#include "MarkNodes.h" +#include "afl-llvm-common.h" +#include "llvm-ngram-coverage.h" + +#include "config.h" +#include "debug.h" + +using namespace llvm; + +static cl::opt MarkSetOpt("markset", cl::desc("MarkSet"), + cl::init(false)); +static cl::opt LoopHeadOpt("loophead", cl::desc("LoopHead"), + cl::init(false)); + +namespace { + +struct InsTrim : public ModulePass { + + protected: + uint32_t function_minimum_size = 1; + char * skip_nozero = NULL; + + private: + std::mt19937 generator; + int total_instr = 0; + + unsigned int genLabel() { + + return generator() & (MAP_SIZE - 1); + + } + + public: + static char ID; + + InsTrim() : ModulePass(ID), generator(0) { + + initInstrumentList(); + + } + + void getAnalysisUsage(AnalysisUsage &AU) const override { + + AU.addRequired(); + + } + +#if LLVM_VERSION_MAJOR < 4 + const char * +#else + StringRef +#endif + getPassName() const override { + + return "InstTrim Instrumentation"; + + } + +#if LLVM_VERSION_MAJOR > 4 || \ + (LLVM_VERSION_MAJOR == 4 && LLVM_VERSION_PATCH >= 1) + #define AFL_HAVE_VECTOR_INTRINSICS 1 +#endif + + bool runOnModule(Module &M) override { + + setvbuf(stdout, NULL, _IONBF, 0); + + if ((isatty(2) && !getenv("AFL_QUIET")) || getenv("AFL_DEBUG") != NULL) { + + SAYF(cCYA "LLVMInsTrim" VERSION cRST " by csienslab\n"); + + } else + + be_quiet = 1; + + if (getenv("AFL_DEBUG") != NULL) debug = 1; + + LLVMContext &C = M.getContext(); + + IntegerType *Int8Ty = IntegerType::getInt8Ty(C); + IntegerType *Int32Ty = IntegerType::getInt32Ty(C); + +#if LLVM_VERSION_MAJOR < 9 + char *neverZero_counters_str; + if ((neverZero_counters_str = getenv("AFL_LLVM_NOT_ZERO")) != NULL) + if (!be_quiet) OKF("LLVM neverZero activated (by hexcoder)\n"); +#endif + skip_nozero = getenv("AFL_LLVM_SKIP_NEVERZERO"); + + if (getenv("AFL_LLVM_INSTRIM_LOOPHEAD") != NULL || + getenv("LOOPHEAD") != NULL) { + + LoopHeadOpt = true; + + } + + unsigned int PrevLocSize = 0; + char * ngram_size_str = getenv("AFL_LLVM_NGRAM_SIZE"); + if (!ngram_size_str) ngram_size_str = getenv("AFL_NGRAM_SIZE"); + char *ctx_str = getenv("AFL_LLVM_CTX"); + +#ifdef AFL_HAVE_VECTOR_INTRINSICS + unsigned int ngram_size = 0; + /* Decide previous location vector size (must be a power of two) */ + VectorType *PrevLocTy = NULL; + + if (ngram_size_str) + if (sscanf(ngram_size_str, "%u", &ngram_size) != 1 || ngram_size < 2 || + ngram_size > NGRAM_SIZE_MAX) + FATAL( + "Bad value of AFL_NGRAM_SIZE (must be between 2 and NGRAM_SIZE_MAX " + "(%u))", + NGRAM_SIZE_MAX); + + if (ngram_size) + PrevLocSize = ngram_size - 1; + else +#else + if (ngram_size_str) + #ifdef LLVM_VERSION_STRING + FATAL( + "Sorry, NGRAM branch coverage is not supported with llvm version %s!", + LLVM_VERSION_STRING); + #else + #ifndef LLVM_VERSION_PATCH + FATAL( + "Sorry, NGRAM branch coverage is not supported with llvm version " + "%d.%d.%d!", + LLVM_VERSION_MAJOR, LLVM_VERSION_MINOR, 0); + #else + FATAL( + "Sorry, NGRAM branch coverage is not supported with llvm version " + "%d.%d.%d!", + LLVM_VERSION_MAJOR, LLVM_VERSION_MINOR, LLVM_VERISON_PATCH); + #endif + #endif +#endif + PrevLocSize = 1; + +#ifdef AFL_HAVE_VECTOR_INTRINSICS + // IntegerType *Int64Ty = IntegerType::getInt64Ty(C); + int PrevLocVecSize = PowerOf2Ceil(PrevLocSize); + IntegerType *IntLocTy = + IntegerType::getIntNTy(C, sizeof(PREV_LOC_T) * CHAR_BIT); + if (ngram_size) + PrevLocTy = VectorType::get(IntLocTy, PrevLocVecSize + #if LLVM_VERSION_MAJOR >= 12 + , + false + #endif + ); +#endif + + /* Get globals for the SHM region and the previous location. Note that + __afl_prev_loc is thread-local. */ + + GlobalVariable *AFLMapPtr = + new GlobalVariable(M, PointerType::get(Int8Ty, 0), false, + GlobalValue::ExternalLinkage, 0, "__afl_area_ptr"); + GlobalVariable *AFLPrevLoc; + GlobalVariable *AFLContext = NULL; + LoadInst * PrevCtx = NULL; // for CTX sensitive coverage + + if (ctx_str) +#ifdef __ANDROID__ + AFLContext = new GlobalVariable( + M, Int32Ty, false, GlobalValue::ExternalLinkage, 0, "__afl_prev_ctx"); +#else + AFLContext = new GlobalVariable( + M, Int32Ty, false, GlobalValue::ExternalLinkage, 0, "__afl_prev_ctx", + 0, GlobalVariable::GeneralDynamicTLSModel, 0, false); +#endif + +#ifdef AFL_HAVE_VECTOR_INTRINSICS + if (ngram_size) + #ifdef __ANDROID__ + AFLPrevLoc = new GlobalVariable( + M, PrevLocTy, /* isConstant */ false, GlobalValue::ExternalLinkage, + /* Initializer */ nullptr, "__afl_prev_loc"); + #else + AFLPrevLoc = new GlobalVariable( + M, PrevLocTy, /* isConstant */ false, GlobalValue::ExternalLinkage, + /* Initializer */ nullptr, "__afl_prev_loc", + /* InsertBefore */ nullptr, GlobalVariable::GeneralDynamicTLSModel, + /* AddressSpace */ 0, /* IsExternallyInitialized */ false); + #endif + else +#endif +#ifdef __ANDROID__ + AFLPrevLoc = new GlobalVariable( + M, Int32Ty, false, GlobalValue::ExternalLinkage, 0, "__afl_prev_loc"); +#else + AFLPrevLoc = new GlobalVariable( + M, Int32Ty, false, GlobalValue::ExternalLinkage, 0, "__afl_prev_loc", 0, + GlobalVariable::GeneralDynamicTLSModel, 0, false); +#endif + +#ifdef AFL_HAVE_VECTOR_INTRINSICS + /* Create the vector shuffle mask for updating the previous block history. + Note that the first element of the vector will store cur_loc, so just set + it to undef to allow the optimizer to do its thing. */ + + SmallVector PrevLocShuffle = {UndefValue::get(Int32Ty)}; + + for (unsigned I = 0; I < PrevLocSize - 1; ++I) + PrevLocShuffle.push_back(ConstantInt::get(Int32Ty, I)); + + for (int I = PrevLocSize; I < PrevLocVecSize; ++I) + PrevLocShuffle.push_back(ConstantInt::get(Int32Ty, PrevLocSize)); + + Constant *PrevLocShuffleMask = ConstantVector::get(PrevLocShuffle); +#endif + + // this is our default + MarkSetOpt = true; + + ConstantInt *Zero = ConstantInt::get(Int8Ty, 0); + ConstantInt *One = ConstantInt::get(Int8Ty, 1); + + u64 total_rs = 0; + u64 total_hs = 0; + + scanForDangerousFunctions(&M); + + for (Function &F : M) { + + if (debug) { + + uint32_t bb_cnt = 0; + + for (auto &BB : F) + if (BB.size() > 0) ++bb_cnt; + SAYF(cMGN "[D] " cRST "Function %s size %zu %u\n", + F.getName().str().c_str(), F.size(), bb_cnt); + + } + + if (!isInInstrumentList(&F)) continue; + + // if the function below our minimum size skip it (1 or 2) + if (F.size() < function_minimum_size) { continue; } + + std::unordered_set MS; + if (!MarkSetOpt) { + + for (auto &BB : F) { + + MS.insert(&BB); + + } + + total_rs += F.size(); + + } else { + + auto Result = markNodes(&F); + auto RS = Result.first; + auto HS = Result.second; + + MS.insert(RS.begin(), RS.end()); + if (!LoopHeadOpt) { + + MS.insert(HS.begin(), HS.end()); + total_rs += MS.size(); + + } else { + + DenseSet> EdgeSet; + DominatorTreeWrapperPass * DTWP = + &getAnalysis(F); + auto DT = &DTWP->getDomTree(); + + total_rs += RS.size(); + total_hs += HS.size(); + + for (BasicBlock *BB : HS) { + + bool Inserted = false; + for (auto BI = pred_begin(BB), BE = pred_end(BB); BI != BE; ++BI) { + + auto Edge = BasicBlockEdge(*BI, BB); + if (Edge.isSingleEdge() && DT->dominates(Edge, BB)) { + + EdgeSet.insert({*BI, BB}); + Inserted = true; + break; + + } + + } + + if (!Inserted) { + + MS.insert(BB); + total_rs += 1; + total_hs -= 1; + + } + + } + + for (auto I = EdgeSet.begin(), E = EdgeSet.end(); I != E; ++I) { + + auto PredBB = I->first; + auto SuccBB = I->second; + auto NewBB = + SplitBlockPredecessors(SuccBB, {PredBB}, ".split", DT, nullptr, +#if LLVM_VERSION_MAJOR >= 8 + nullptr, +#endif + false); + MS.insert(NewBB); + + } + + } + + for (BasicBlock &BB : F) { + + if (MS.find(&BB) == MS.end()) { continue; } + IRBuilder<> IRB(&*BB.getFirstInsertionPt()); + +#ifdef AFL_HAVE_VECTOR_INTRINSICS + if (ngram_size) { + + LoadInst *PrevLoc = IRB.CreateLoad(AFLPrevLoc); + PrevLoc->setMetadata(M.getMDKindID("nosanitize"), + MDNode::get(C, None)); + + Value *ShuffledPrevLoc = IRB.CreateShuffleVector( + PrevLoc, UndefValue::get(PrevLocTy), PrevLocShuffleMask); + Value *UpdatedPrevLoc = IRB.CreateInsertElement( + ShuffledPrevLoc, ConstantInt::get(Int32Ty, genLabel()), + (uint64_t)0); + + IRB.CreateStore(UpdatedPrevLoc, AFLPrevLoc) + ->setMetadata(M.getMDKindID("nosanitize"), + MDNode::get(C, None)); + + } else + +#endif + { + + IRB.CreateStore(ConstantInt::get(Int32Ty, genLabel()), AFLPrevLoc); + + } + + } + + } + + int has_calls = 0; + for (BasicBlock &BB : F) { + + auto PI = pred_begin(&BB); + auto PE = pred_end(&BB); + IRBuilder<> IRB(&*BB.getFirstInsertionPt()); + Value * L = NULL; + unsigned int cur_loc; + + // Context sensitive coverage + if (ctx_str && &BB == &F.getEntryBlock()) { + + PrevCtx = IRB.CreateLoad(AFLContext); + PrevCtx->setMetadata(M.getMDKindID("nosanitize"), + MDNode::get(C, None)); + + // does the function have calls? and is any of the calls larger than + // one basic block? + has_calls = 0; + for (auto &BB : F) { + + if (has_calls) break; + for (auto &IN : BB) { + + CallInst *callInst = nullptr; + if ((callInst = dyn_cast(&IN))) { + + Function *Callee = callInst->getCalledFunction(); + if (!Callee || Callee->size() < function_minimum_size) + continue; + else { + + has_calls = 1; + break; + + } + + } + + } + + } + + // if yes we store a context ID for this function in the global var + if (has_calls) { + + ConstantInt *NewCtx = ConstantInt::get(Int32Ty, genLabel()); + StoreInst * StoreCtx = IRB.CreateStore(NewCtx, AFLContext); + StoreCtx->setMetadata(M.getMDKindID("nosanitize"), + MDNode::get(C, None)); + + } + + } // END of ctx_str + + if (MarkSetOpt && MS.find(&BB) == MS.end()) { continue; } + + if (PI == PE) { + + cur_loc = genLabel(); + L = ConstantInt::get(Int32Ty, cur_loc); + + } else { + + auto *PN = PHINode::Create(Int32Ty, 0, "", &*BB.begin()); + DenseMap PredMap; + for (auto PI = pred_begin(&BB), PE = pred_end(&BB); PI != PE; ++PI) { + + BasicBlock *PBB = *PI; + auto It = PredMap.insert({PBB, genLabel()}); + unsigned Label = It.first->second; + cur_loc = Label; + PN->addIncoming(ConstantInt::get(Int32Ty, Label), PBB); + + } + + L = PN; + + } + + /* Load prev_loc */ + LoadInst *PrevLoc = IRB.CreateLoad(AFLPrevLoc); + PrevLoc->setMetadata(M.getMDKindID("nosanitize"), MDNode::get(C, None)); + Value *PrevLocTrans; + +#ifdef AFL_HAVE_VECTOR_INTRINSICS + /* "For efficiency, we propose to hash the tuple as a key into the + hit_count map as (prev_block_trans << 1) ^ curr_block_trans, where + prev_block_trans = (block_trans_1 ^ ... ^ block_trans_(n-1)" */ + + if (ngram_size) + PrevLocTrans = + IRB.CreateZExt(IRB.CreateXorReduce(PrevLoc), IRB.getInt32Ty()); + else +#endif + PrevLocTrans = IRB.CreateZExt(PrevLoc, IRB.getInt32Ty()); + + if (ctx_str) + PrevLocTrans = + IRB.CreateZExt(IRB.CreateXor(PrevLocTrans, PrevCtx), Int32Ty); + + /* Load SHM pointer */ + LoadInst *MapPtr = IRB.CreateLoad(AFLMapPtr); + MapPtr->setMetadata(M.getMDKindID("nosanitize"), MDNode::get(C, None)); + Value *MapPtrIdx; +#ifdef AFL_HAVE_VECTOR_INTRINSICS + if (ngram_size) + MapPtrIdx = IRB.CreateGEP( + MapPtr, IRB.CreateZExt(IRB.CreateXor(PrevLocTrans, L), Int32Ty)); + else +#endif + MapPtrIdx = IRB.CreateGEP(MapPtr, IRB.CreateXor(PrevLocTrans, L)); + + /* Update bitmap */ + LoadInst *Counter = IRB.CreateLoad(MapPtrIdx); + Counter->setMetadata(M.getMDKindID("nosanitize"), MDNode::get(C, None)); + + Value *Incr = IRB.CreateAdd(Counter, One); + +#if LLVM_VERSION_MAJOR < 9 + if (neverZero_counters_str != + NULL) // with llvm 9 we make this the default as the bug in llvm is + // then fixed +#else + if (!skip_nozero) +#endif + { + + /* hexcoder: Realize a counter that skips zero during overflow. + * Once this counter reaches its maximum value, it next increments to + * 1 + * + * Instead of + * Counter + 1 -> Counter + * we inject now this + * Counter + 1 -> {Counter, OverflowFlag} + * Counter + OverflowFlag -> Counter + */ + auto cf = IRB.CreateICmpEQ(Incr, Zero); + auto carry = IRB.CreateZExt(cf, Int8Ty); + Incr = IRB.CreateAdd(Incr, carry); + + } + + IRB.CreateStore(Incr, MapPtrIdx) + ->setMetadata(M.getMDKindID("nosanitize"), MDNode::get(C, None)); + + if (ctx_str && has_calls) { + + // in CTX mode we have to restore the original context for the + // caller - she might be calling other functions which need the + // correct CTX + Instruction *Inst = BB.getTerminator(); + if (isa(Inst) || isa(Inst)) { + + IRBuilder<> Post_IRB(Inst); + StoreInst * RestoreCtx = Post_IRB.CreateStore(PrevCtx, AFLContext); + RestoreCtx->setMetadata(M.getMDKindID("nosanitize"), + MDNode::get(C, None)); + + } + + } + + total_instr++; + + } + + } + + if (!be_quiet) { + + char modeline[100]; + snprintf(modeline, sizeof(modeline), "%s%s%s%s%s", + getenv("AFL_HARDEN") ? "hardened" : "non-hardened", + getenv("AFL_USE_ASAN") ? ", ASAN" : "", + getenv("AFL_USE_MSAN") ? ", MSAN" : "", + getenv("AFL_USE_CFISAN") ? ", CFISAN" : "", + getenv("AFL_USE_UBSAN") ? ", UBSAN" : ""); + + OKF("Instrumented %u locations (%llu, %llu) (%s mode)\n", total_instr, + total_rs, total_hs, modeline); + + } + + return false; + + } + +}; // end of struct InsTrim + +} // end of anonymous namespace + +char InsTrim::ID = 0; + +static void registerAFLPass(const PassManagerBuilder &, + legacy::PassManagerBase &PM) { + + PM.add(new InsTrim()); + +} + +static RegisterStandardPasses RegisterAFLPass( + PassManagerBuilder::EP_OptimizerLast, registerAFLPass); + +static RegisterStandardPasses RegisterAFLPass0( + PassManagerBuilder::EP_EnabledOnOptLevel0, registerAFLPass); + diff --git a/instrumentation/MarkNodes.cc b/instrumentation/MarkNodes.cc new file mode 100644 index 00000000..20a7df35 --- /dev/null +++ b/instrumentation/MarkNodes.cc @@ -0,0 +1,481 @@ +#include +#include +#include +#include +#include + +#include "llvm/Config/llvm-config.h" +#if LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR < 5 +typedef long double max_align_t; +#endif + +#include "llvm/ADT/DenseMap.h" +#include "llvm/ADT/DenseSet.h" +#include "llvm/ADT/SmallVector.h" +#include "llvm/IR/BasicBlock.h" +#if LLVM_VERSION_MAJOR > 3 || \ + (LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR > 4) + #include "llvm/IR/CFG.h" +#else + #include "llvm/Support/CFG.h" +#endif +#include "llvm/IR/Constants.h" +#include "llvm/IR/Function.h" +#include "llvm/IR/IRBuilder.h" +#include "llvm/IR/Instructions.h" +#include "llvm/IR/Module.h" +#include "llvm/Pass.h" +#include "llvm/Support/Debug.h" +#include "llvm/Support/raw_ostream.h" + +using namespace llvm; + +DenseMap LMap; +std::vector Blocks; +std::set Marked, Markabove; +std::vector > Succs, Preds; + +void reset() { + + LMap.clear(); + Blocks.clear(); + Marked.clear(); + Markabove.clear(); + +} + +uint32_t start_point; + +void labelEachBlock(Function *F) { + + // Fake single endpoint; + LMap[NULL] = Blocks.size(); + Blocks.push_back(NULL); + + // Assign the unique LabelID to each block; + for (auto I = F->begin(), E = F->end(); I != E; ++I) { + + BasicBlock *BB = &*I; + LMap[BB] = Blocks.size(); + Blocks.push_back(BB); + + } + + start_point = LMap[&F->getEntryBlock()]; + +} + +void buildCFG(Function *F) { + + Succs.resize(Blocks.size()); + Preds.resize(Blocks.size()); + for (size_t i = 0; i < Succs.size(); i++) { + + Succs[i].clear(); + Preds[i].clear(); + + } + + for (auto S = F->begin(), E = F->end(); S != E; ++S) { + + BasicBlock *BB = &*S; + uint32_t MyID = LMap[BB]; + + for (auto I = succ_begin(BB), E = succ_end(BB); I != E; ++I) { + + Succs[MyID].push_back(LMap[*I]); + + } + + } + +} + +std::vector > tSuccs; +std::vector tag, indfs; + +void DFStree(size_t now_id) { + + if (tag[now_id]) return; + tag[now_id] = true; + indfs[now_id] = true; + for (auto succ : tSuccs[now_id]) { + + if (tag[succ] and indfs[succ]) { + + Marked.insert(succ); + Markabove.insert(succ); + continue; + + } + + Succs[now_id].push_back(succ); + Preds[succ].push_back(now_id); + DFStree(succ); + + } + + indfs[now_id] = false; + +} + +void turnCFGintoDAG() { + + tSuccs = Succs; + tag.resize(Blocks.size()); + indfs.resize(Blocks.size()); + for (size_t i = 0; i < Blocks.size(); ++i) { + + Succs[i].clear(); + tag[i] = false; + indfs[i] = false; + + } + + DFStree(start_point); + for (size_t i = 0; i < Blocks.size(); ++i) + if (Succs[i].empty()) { + + Succs[i].push_back(0); + Preds[0].push_back(i); + + } + +} + +uint32_t timeStamp; +namespace DominatorTree { + +std::vector > cov; +std::vector dfn, nfd, par, sdom, idom, mom, mn; + +bool Compare(uint32_t u, uint32_t v) { + + return dfn[u] < dfn[v]; + +} + +uint32_t eval(uint32_t u) { + + if (mom[u] == u) return u; + uint32_t res = eval(mom[u]); + if (Compare(sdom[mn[mom[u]]], sdom[mn[u]])) { mn[u] = mn[mom[u]]; } + return mom[u] = res; + +} + +void DFS(uint32_t now) { + + timeStamp += 1; + dfn[now] = timeStamp; + nfd[timeStamp - 1] = now; + for (auto succ : Succs[now]) { + + if (dfn[succ] == 0) { + + par[succ] = now; + DFS(succ); + + } + + } + +} + +void DominatorTree() { + + if (Blocks.empty()) return; + uint32_t s = start_point; + + // Initialization + mn.resize(Blocks.size()); + cov.resize(Blocks.size()); + dfn.resize(Blocks.size()); + nfd.resize(Blocks.size()); + par.resize(Blocks.size()); + mom.resize(Blocks.size()); + sdom.resize(Blocks.size()); + idom.resize(Blocks.size()); + + for (uint32_t i = 0; i < Blocks.size(); i++) { + + dfn[i] = 0; + nfd[i] = Blocks.size(); + cov[i].clear(); + idom[i] = mom[i] = mn[i] = sdom[i] = i; + + } + + timeStamp = 0; + DFS(s); + + for (uint32_t i = Blocks.size() - 1; i >= 1u; i--) { + + uint32_t now = nfd[i]; + if (now == Blocks.size()) { continue; } + for (uint32_t pre : Preds[now]) { + + if (dfn[pre]) { + + eval(pre); + if (Compare(sdom[mn[pre]], sdom[now])) { sdom[now] = sdom[mn[pre]]; } + + } + + } + + cov[sdom[now]].push_back(now); + mom[now] = par[now]; + for (uint32_t x : cov[par[now]]) { + + eval(x); + if (Compare(sdom[mn[x]], par[now])) { + + idom[x] = mn[x]; + + } else { + + idom[x] = par[now]; + + } + + } + + } + + for (uint32_t i = 1; i < Blocks.size(); i += 1) { + + uint32_t now = nfd[i]; + if (now == Blocks.size()) { continue; } + if (idom[now] != sdom[now]) idom[now] = idom[idom[now]]; + + } + +} + +} // namespace DominatorTree + +std::vector Visited, InStack; +std::vector TopoOrder, InDeg; +std::vector > t_Succ, t_Pred; + +void Go(uint32_t now, uint32_t tt) { + + if (now == tt) return; + Visited[now] = InStack[now] = timeStamp; + + for (uint32_t nxt : Succs[now]) { + + if (Visited[nxt] == timeStamp and InStack[nxt] == timeStamp) { + + Marked.insert(nxt); + + } + + t_Succ[now].push_back(nxt); + t_Pred[nxt].push_back(now); + InDeg[nxt] += 1; + if (Visited[nxt] == timeStamp) { continue; } + Go(nxt, tt); + + } + + InStack[now] = 0; + +} + +void TopologicalSort(uint32_t ss, uint32_t tt) { + + timeStamp += 1; + + Go(ss, tt); + + TopoOrder.clear(); + std::queue wait; + wait.push(ss); + while (not wait.empty()) { + + uint32_t now = wait.front(); + wait.pop(); + TopoOrder.push_back(now); + for (uint32_t nxt : t_Succ[now]) { + + InDeg[nxt] -= 1; + if (InDeg[nxt] == 0u) { wait.push(nxt); } + + } + + } + +} + +std::vector > NextMarked; +bool Indistinguish(uint32_t node1, uint32_t node2) { + + if (NextMarked[node1].size() > NextMarked[node2].size()) { + + uint32_t _swap = node1; + node1 = node2; + node2 = _swap; + + } + + for (uint32_t x : NextMarked[node1]) { + + if (NextMarked[node2].find(x) != NextMarked[node2].end()) { return true; } + + } + + return false; + +} + +void MakeUniq(uint32_t now) { + + bool StopFlag = false; + if (Marked.find(now) == Marked.end()) { + + for (uint32_t pred1 : t_Pred[now]) { + + for (uint32_t pred2 : t_Pred[now]) { + + if (pred1 == pred2) continue; + if (Indistinguish(pred1, pred2)) { + + Marked.insert(now); + StopFlag = true; + break; + + } + + } + + if (StopFlag) { break; } + + } + + } + + if (Marked.find(now) != Marked.end()) { + + NextMarked[now].insert(now); + + } else { + + for (uint32_t pred : t_Pred[now]) { + + for (uint32_t x : NextMarked[pred]) { + + NextMarked[now].insert(x); + + } + + } + + } + +} + +bool MarkSubGraph(uint32_t ss, uint32_t tt) { + + TopologicalSort(ss, tt); + if (TopoOrder.empty()) return false; + + for (uint32_t i : TopoOrder) { + + NextMarked[i].clear(); + + } + + NextMarked[TopoOrder[0]].insert(TopoOrder[0]); + for (uint32_t i = 1; i < TopoOrder.size(); i += 1) { + + MakeUniq(TopoOrder[i]); + + } + + // Check if there is an empty path. + if (NextMarked[tt].count(TopoOrder[0]) > 0) return true; + return false; + +} + +void MarkVertice() { + + uint32_t s = start_point; + + InDeg.resize(Blocks.size()); + Visited.resize(Blocks.size()); + InStack.resize(Blocks.size()); + t_Succ.resize(Blocks.size()); + t_Pred.resize(Blocks.size()); + NextMarked.resize(Blocks.size()); + + for (uint32_t i = 0; i < Blocks.size(); i += 1) { + + Visited[i] = InStack[i] = InDeg[i] = 0; + t_Succ[i].clear(); + t_Pred[i].clear(); + + } + + timeStamp = 0; + uint32_t t = 0; + bool emptyPathExists = true; + + while (s != t) { + + emptyPathExists &= MarkSubGraph(DominatorTree::idom[t], t); + t = DominatorTree::idom[t]; + + } + + if (emptyPathExists) { + + // Mark all exit blocks to catch the empty path. + Marked.insert(t_Pred[0].begin(), t_Pred[0].end()); + + } + +} + +// return {marked nodes} +std::pair, std::vector > markNodes( + Function *F) { + + assert(F->size() > 0 && "Function can not be empty"); + + reset(); + labelEachBlock(F); + buildCFG(F); + turnCFGintoDAG(); + DominatorTree::DominatorTree(); + MarkVertice(); + + std::vector Result, ResultAbove; + for (uint32_t x : Markabove) { + + auto it = Marked.find(x); + if (it != Marked.end()) Marked.erase(it); + if (x) ResultAbove.push_back(Blocks[x]); + + } + + for (uint32_t x : Marked) { + + if (x == 0) { + + continue; + + } else { + + Result.push_back(Blocks[x]); + + } + + } + + return {Result, ResultAbove}; + +} + diff --git a/instrumentation/MarkNodes.h b/instrumentation/MarkNodes.h new file mode 100644 index 00000000..8ddc978d --- /dev/null +++ b/instrumentation/MarkNodes.h @@ -0,0 +1,12 @@ +#ifndef __MARK_NODES__ +#define __MARK_NODES__ + +#include "llvm/IR/BasicBlock.h" +#include "llvm/IR/Function.h" +#include + +std::pair, std::vector> +markNodes(llvm::Function *F); + +#endif + diff --git a/instrumentation/README.cmplog.md b/instrumentation/README.cmplog.md new file mode 100644 index 00000000..5f855e1f --- /dev/null +++ b/instrumentation/README.cmplog.md @@ -0,0 +1,42 @@ +# CmpLog instrumentation + +The CmpLog instrumentation enables the logging of the comparisons operands in a +shared memory. + +These values can be used by various mutators built on top of it. +At the moment we support the RedQueen mutator (input-2-state instructions only). + +## Build + +To use CmpLog, you have to build two versions of the instrumented target +program. + +The first version is built using the regular AFL++ instrumentation. + +The second one, the CmpLog binary, with setting AFL_LLVM_CMPLOG during the compilation. + +For example: + +``` +./configure --cc=~/path/to/afl-clang-fast +make +cp ./program ./program.afl +make clean +export AFL_LLVM_CMPLOG=1 +./configure --cc=~/path/to/afl-clang-fast +make +cp ./program ./program.cmplog +``` + +## Use + +AFL++ has the new -c option that needs to be used to specify the CmpLog binary (the second +build). + +For example: + +``` +afl-fuzz -i input -o output -c ./program.cmplog -m none -- ./program.afl @@ +``` + +Be sure to use `-m none` because CmpLog can map a lot of pages. diff --git a/instrumentation/README.ctx.md b/instrumentation/README.ctx.md new file mode 100644 index 00000000..caf2c09a --- /dev/null +++ b/instrumentation/README.ctx.md @@ -0,0 +1,22 @@ +# AFL Context Sensitive Branch Coverage + +## What is this? + +This is an LLVM-based implementation of the context sensitive branch coverage. + +Basically every function gets its own ID and that ID is combined with the +edges of the called functions. + +So if both function A and function B call a function C, the coverage +collected in C will be different. + +In math the coverage is collected as follows: +`map[current_location_ID ^ previous_location_ID >> 1 ^ previous_callee_ID] += 1` + +## Usage + +Set the `AFL_LLVM_INSTRUMENT=CTX` or `AFL_LLVM_CTX=1` environment variable. + +It is highly recommended to increase the MAP_SIZE_POW2 definition in +config.h to at least 18 and maybe up to 20 for this as otherwise too +many map collisions occur. diff --git a/instrumentation/README.gcc_plugin.md b/instrumentation/README.gcc_plugin.md new file mode 100644 index 00000000..9d6bc200 --- /dev/null +++ b/instrumentation/README.gcc_plugin.md @@ -0,0 +1,158 @@ +# GCC-based instrumentation for afl-fuzz + + (See [../README.md](../README.md) for the general instruction manual.) + (See [README.llvm.md](README.llvm.md) for the LLVM-based instrumentation.) + +!!! TODO items are: +!!! => inline instrumentation has to work! +!!! + + +## 1) Introduction + +The code in this directory allows you to instrument programs for AFL using +true compiler-level instrumentation, instead of the more crude +assembly-level rewriting approach taken by afl-gcc and afl-clang. This has +several interesting properties: + + - The compiler can make many optimizations that are hard to pull off when + manually inserting assembly. As a result, some slow, CPU-bound programs will + run up to around faster. + + The gains are less pronounced for fast binaries, where the speed is limited + chiefly by the cost of creating new processes. In such cases, the gain will + probably stay within 10%. + + - The instrumentation is CPU-independent. At least in principle, you should + be able to rely on it to fuzz programs on non-x86 architectures (after + building afl-fuzz with AFL_NOX86=1). + + - Because the feature relies on the internals of GCC, it is gcc-specific + and will *not* work with LLVM (see ../llvm_mode for an alternative). + +Once this implementation is shown to be sufficiently robust and portable, it +will probably replace afl-gcc. For now, it can be built separately and +co-exists with the original code. + +The idea and much of the implementation comes from Laszlo Szekeres. + +## 2) How to use + +In order to leverage this mechanism, you need to have modern enough GCC +(>= version 4.5.0) and the plugin headers installed on your system. That +should be all you need. On Debian machines, these headers can be acquired by +installing the `gcc--plugin-dev` packages. + +To build the instrumentation itself, type 'make'. This will generate binaries +called afl-gcc-fast and afl-g++-fast in the parent directory. +If the CC/CXX have been overridden, those compilers will be used from +those wrappers without using AFL_CXX/AFL_CC settings. +Once this is done, you can instrument third-party code in a way similar to the +standard operating mode of AFL, e.g.: + + CC=/path/to/afl/afl-gcc-fast ./configure [...options...] + make + +Be sure to also include CXX set to afl-g++-fast for C++ code. + +The tool honors roughly the same environmental variables as afl-gcc (see +[env_variables.md](../docs/env_variables.md). This includes AFL_INST_RATIO, AFL_USE_ASAN, +AFL_HARDEN, and AFL_DONT_OPTIMIZE. + +Note: if you want the GCC plugin to be installed on your system for all +users, you need to build it before issuing 'make install' in the parent +directory. + +## 3) Gotchas, feedback, bugs + +This is an early-stage mechanism, so field reports are welcome. You can send bug +reports to . + +## 4) Bonus feature #1: deferred initialization + +AFL tries to optimize performance by executing the targeted binary just once, +stopping it just before main(), and then cloning this "main" process to get +a steady supply of targets to fuzz. + +Although this approach eliminates much of the OS-, linker- and libc-level +costs of executing the program, it does not always help with binaries that +perform other time-consuming initialization steps - say, parsing a large config +file before getting to the fuzzed data. + +In such cases, it's beneficial to initialize the forkserver a bit later, once +most of the initialization work is already done, but before the binary attempts +to read the fuzzed input and parse it; in some cases, this can offer a 10x+ +performance gain. You can implement delayed initialization in LLVM mode in a +fairly simple way. + +First, locate a suitable location in the code where the delayed cloning can +take place. This needs to be done with *extreme* care to avoid breaking the +binary. In particular, the program will probably malfunction if you select +a location after: + + - The creation of any vital threads or child processes - since the forkserver + can't clone them easily. + + - The initialization of timers via setitimer() or equivalent calls. + + - The creation of temporary files, network sockets, offset-sensitive file + descriptors, and similar shared-state resources - but only provided that + their state meaningfully influences the behavior of the program later on. + + - Any access to the fuzzed input, including reading the metadata about its + size. + +With the location selected, add this code in the appropriate spot: + +``` +#ifdef __AFL_HAVE_MANUAL_CONTROL + __AFL_INIT(); +#endif +``` + +You don't need the #ifdef guards, but they will make the program still work as +usual when compiled with a tool other than afl-gcc-fast/afl-clang-fast. + +Finally, recompile the program with afl-gcc-fast (afl-gcc or afl-clang will +*not* generate a deferred-initialization binary) - and you should be all set! + +## 5) Bonus feature #2: persistent mode + +Some libraries provide APIs that are stateless, or whose state can be reset in +between processing different input files. When such a reset is performed, a +single long-lived process can be reused to try out multiple test cases, +eliminating the need for repeated fork() calls and the associated OS overhead. + +The basic structure of the program that does this would be: + +``` + while (__AFL_LOOP(1000)) { + + /* Read input data. */ + /* Call library code to be fuzzed. */ + /* Reset state. */ + + } + + /* Exit normally */ +``` + +The numerical value specified within the loop controls the maximum number +of iterations before AFL will restart the process from scratch. This minimizes +the impact of memory leaks and similar glitches; 1000 is a good starting point. + +A more detailed template is shown in ../examples/persistent_demo/. +Similarly to the previous mode, the feature works only with afl-gcc-fast or +afl-clang-fast; #ifdef guards can be used to suppress it when using other +compilers. + +Note that as with the previous mode, the feature is easy to misuse; if you +do not reset the critical state fully, you may end up with false positives or +waste a whole lot of CPU power doing nothing useful at all. Be particularly +wary of memory leaks and the state of file descriptors. + +When running in this mode, the execution paths will inherently vary a bit +depending on whether the input loop is being entered for the first time or +executed again. To avoid spurious warnings, the feature implies +AFL_NO_VAR_CHECK and hides the "variable path" warnings in the UI. + diff --git a/instrumentation/README.instrim.md b/instrumentation/README.instrim.md new file mode 100644 index 00000000..99f6477a --- /dev/null +++ b/instrumentation/README.instrim.md @@ -0,0 +1,30 @@ +# InsTrim + +InsTrim: Lightweight Instrumentation for Coverage-guided Fuzzing + +## Introduction + +InsTrim is the work of Chin-Chia Hsu, Che-Yu Wu, Hsu-Chun Hsiao and Shih-Kun Huang. + +It uses a CFG (call flow graph) and markers to instrument just what +is necessary in the binary (ie less than llvm_mode). As a result the binary is +about 10-15% faster compared to normal llvm_mode however with some coverage loss. +It requires at least llvm version 3.8.0 to build. +If you have LLVM 7+ we recommend PCGUARD instead. + +## Usage + +Set the environment variable `AFL_LLVM_INSTRUMENT=CFG` or `AFL_LLVM_INSTRIM=1` +during compilation of the target. + +There is also special mode which instruments loops in a way so that +afl-fuzz can see which loop path has been selected but not being able to +see how often the loop has been rerun. +This again is a tradeoff for speed for less path information. +To enable this mode set `AFL_LLVM_INSTRIM_LOOPHEAD=1`. + +## Background + +The paper from Chin-Chia Hsu, Che-Yu Wu, Hsu-Chun Hsiao and Shih-Kun Huang: +[InsTrim: Lightweight Instrumentation for Coverage-guided Fuzzing] +(https://www.ndss-symposium.org/wp-content/uploads/2018/07/bar2018_14_Hsu_paper.pdf) diff --git a/instrumentation/README.instrument_list.md b/instrumentation/README.instrument_list.md new file mode 100644 index 00000000..60474ec6 --- /dev/null +++ b/instrumentation/README.instrument_list.md @@ -0,0 +1,87 @@ +# Using afl++ with partial instrumentation + + This file describes how to selectively instrument only source files + or functions that are of interest to you using the LLVM instrumentation + provided by afl++. + +## 1) Description and purpose + +When building and testing complex programs where only a part of the program is +the fuzzing target, it often helps to only instrument the necessary parts of +the program, leaving the rest uninstrumented. This helps to focus the fuzzer +on the important parts of the program, avoiding undesired noise and +disturbance by uninteresting code being exercised. + +For this purpose, a "partial instrumentation" support en par with llvm sancov +is provided by afl++ that allows to specify on a source file and function +level which function should be compiled with or without instrumentation. + +Note: When using PCGUARD mode - and llvm 12+ - you can use this instead: +https://clang.llvm.org/docs/SanitizerCoverage.html#partially-disabling-instrumentation + +The llvm sancov list format is fully supported by afl++, however afl++ has +more flexibility. + +## 2) Building the LLVM module + +The new code is part of the existing afl++ LLVM module in the instrumentation/ +subdirectory. There is nothing specifically to do for the build :) + +## 3) How to use the partial instrumentation mode + +In order to build with partial instrumentation, you need to build with +afl-clang-fast/afl-clang-fast++ or afl-clang-lto/afl-clang-lto++. +The only required change is that you need to set either the environment variable +AFL_LLVM_ALLOWLIST or AFL_LLVM_DENYLIST set with a filename. + +That file should contain the file names or functions that are to be instrumented +(AFL_LLVM_ALLOWLIST) or are specifically NOT to be instrumented (AFL_LLVM_DENYLIST). + +For matching to succeed, the function/file name that is being compiled must end in the +function/file name entry contained in this instrument file list. That is to avoid +breaking the match when absolute paths are used during compilation. + +**NOTE:** In builds with optimization enabled, functions might be inlined and would not match! + +For example if your source tree looks like this: +``` +project/ +project/feature_a/a1.cpp +project/feature_a/a2.cpp +project/feature_b/b1.cpp +project/feature_b/b2.cpp +``` + +and you only want to test feature_a, then create an "instrument file list" file containing: +``` +feature_a/a1.cpp +feature_a/a2.cpp +``` + +However if the "instrument file list" file contains only this, it works as well: +``` +a1.cpp +a2.cpp +``` +but it might lead to files being unwantedly instrumented if the same filename +exists somewhere else in the project directories. + +You can also specify function names. Note that for C++ the function names +must be mangled to match! `nm` can print these names. + +afl++ is able to identify whether an entry is a filename or a function. +However if you want to be sure (and compliant to the sancov allow/blocklist +format), you can specify source file entries like this: +``` +src: *malloc.c +``` +and function entries like this: +``` +fun: MallocFoo +``` +Note that whitespace is ignored and comments (`# foo`) are supported. + +## 4) UNIX-style pattern matching + +You can add UNIX-style pattern matching in the "instrument file list" entries. +See `man fnmatch` for the syntax. We do not set any of the `fnmatch` flags. diff --git a/instrumentation/README.laf-intel.md b/instrumentation/README.laf-intel.md new file mode 100644 index 00000000..c50a6979 --- /dev/null +++ b/instrumentation/README.laf-intel.md @@ -0,0 +1,56 @@ +# laf-intel instrumentation + +## Introduction + +This originally is the work of an individual nicknamed laf-intel. +His blog [Circumventing Fuzzing Roadblocks with Compiler Transformations] +(https://lafintel.wordpress.com/) and gitlab repo [laf-llvm-pass] +(https://gitlab.com/laf-intel/laf-llvm-pass/) +describe some code transformations that +help afl++ to enter conditional blocks, where conditions consist of +comparisons of large values. + +## Usage + +By default these passes will not run when you compile programs using +afl-clang-fast. Hence, you can use AFL as usual. +To enable the passes you must set environment variables before you +compile the target project. + +The following options exist: + +`export AFL_LLVM_LAF_SPLIT_SWITCHES=1` + +Enables the split-switches pass. + +`export AFL_LLVM_LAF_TRANSFORM_COMPARES=1` + +Enables the transform-compares pass (strcmp, memcmp, strncmp, +strcasecmp, strncasecmp). + +`export AFL_LLVM_LAF_SPLIT_COMPARES=1` + +Enables the split-compares pass. +By default it will +1. simplify operators >= (and <=) into chains of > (<) and == comparisons +2. change signed integer comparisons to a chain of sign-only comparison +and unsigned integer comparisons +3. split all unsigned integer comparisons with bit widths of +64, 32 or 16 bits to chains of 8 bits comparisons. + +You can change the behaviour of the last step by setting +`export AFL_LLVM_LAF_SPLIT_COMPARES_BITW=`, where +bit_width may be 64, 32 or 16. For example, a bit_width of 16 +would split larger comparisons down to 16 bit comparisons. + +A new experimental feature is splitting floating point comparisons into a +series of sign, exponent and mantissa comparisons followed by splitting each +of them into 8 bit comparisons when necessary. +It is activated with the `AFL_LLVM_LAF_SPLIT_FLOATS` setting. +Please note that full IEEE 754 functionality is not preserved, that is +values of nan and infinity will probably behave differently. + +Note that setting this automatically activates `AFL_LLVM_LAF_SPLIT_COMPARES` + +You can also set `AFL_LLVM_LAF_ALL` and have all of the above enabled :-) + diff --git a/instrumentation/README.llvm.md b/instrumentation/README.llvm.md new file mode 100644 index 00000000..51e9995b --- /dev/null +++ b/instrumentation/README.llvm.md @@ -0,0 +1,194 @@ +# Fast LLVM-based instrumentation for afl-fuzz + + (See [../README.md](../README.md) for the general instruction manual.) + + (See [README.gcc_plugon.md](../README.gcc_plugin.md) for the GCC-based instrumentation.) + +## 1) Introduction + +! llvm_mode works with llvm versions 3.4 up to 12 ! + +The code in this directory allows you to instrument programs for AFL using +true compiler-level instrumentation, instead of the more crude +assembly-level rewriting approach taken by afl-gcc and afl-clang. This has +several interesting properties: + + - The compiler can make many optimizations that are hard to pull off when + manually inserting assembly. As a result, some slow, CPU-bound programs will + run up to around 2x faster. + + The gains are less pronounced for fast binaries, where the speed is limited + chiefly by the cost of creating new processes. In such cases, the gain will + probably stay within 10%. + + - The instrumentation is CPU-independent. At least in principle, you should + be able to rely on it to fuzz programs on non-x86 architectures (after + building afl-fuzz with AFL_NO_X86=1). + + - The instrumentation can cope a bit better with multi-threaded targets. + + - Because the feature relies on the internals of LLVM, it is clang-specific + and will *not* work with GCC (see ../gcc_plugin/ for an alternative once + it is available). + +Once this implementation is shown to be sufficiently robust and portable, it +will probably replace afl-clang. For now, it can be built separately and +co-exists with the original code. + +The idea and much of the intial implementation came from Laszlo Szekeres. + +## 2a) How to use this - short + +Set the `LLVM_CONFIG` variable to the clang version you want to use, e.g. +``` +LLVM_CONFIG=llvm-config-9 make +``` +In case you have your own compiled llvm version specify the full path: +``` +LLVM_CONFIG=~/llvm-project/build/bin/llvm-config make +``` +If you try to use a new llvm version on an old Linux this can fail because of +old c++ libraries. In this case usually switching to gcc/g++ to compile +llvm_mode will work: +``` +LLVM_CONFIG=llvm-config-7 REAL_CC=gcc REAL_CXX=g++ make +``` +It is highly recommended to use the newest clang version you can put your +hands on :) + +Then look at [README.persistent_mode.md](README.persistent_mode.md). + +## 2b) How to use this - long + +In order to leverage this mechanism, you need to have clang installed on your +system. You should also make sure that the llvm-config tool is in your path +(or pointed to via LLVM_CONFIG in the environment). + +Note that if you have several LLVM versions installed, pointing LLVM_CONFIG +to the version you want to use will switch compiling to this specific +version - if you installation is set up correctly :-) + +Unfortunately, some systems that do have clang come without llvm-config or the +LLVM development headers; one example of this is FreeBSD. FreeBSD users will +also run into problems with clang being built statically and not being able to +load modules (you'll see "Service unavailable" when loading afl-llvm-pass.so). + +To solve all your problems, you can grab pre-built binaries for your OS from: + + http://llvm.org/releases/download.html + +...and then put the bin/ directory from the tarball at the beginning of your +$PATH when compiling the feature and building packages later on. You don't need +to be root for that. + +To build the instrumentation itself, type 'make'. This will generate binaries +called afl-clang-fast and afl-clang-fast++ in the parent directory. Once this +is done, you can instrument third-party code in a way similar to the standard +operating mode of AFL, e.g.: + +``` + CC=/path/to/afl/afl-clang-fast ./configure [...options...] + make +``` + +Be sure to also include CXX set to afl-clang-fast++ for C++ code. + +Note that afl-clang-fast/afl-clang-fast++ are just pointers to afl-cc. +You can also use afl-cc/afl-c++ and instead direct it to use LLVM +instrumentation by either setting `AFL_CC_COMPILER=LLVM` or pass the parameter +`--afl-llvm` via CFLAGS/CXXFLAGS/CPPFLAGS. + +The tool honors roughly the same environmental variables as afl-gcc (see +[docs/env_variables.md](../docs/env_variables.md)). This includes AFL_USE_ASAN, +AFL_HARDEN, and AFL_DONT_OPTIMIZE. However AFL_INST_RATIO is not honored +as it does not serve a good purpose with the more effective PCGUARD, LTO and + instrim CFG analysis. + +## 3) Options + +Several options are present to make llvm_mode faster or help it rearrange +the code to make afl-fuzz path discovery easier. + +If you need just to instrument specific parts of the code, you can the instrument file list +which C/C++ files to actually instrument. See [README.instrument_list.md](README.instrument_list.md) + +For splitting memcmp, strncmp, etc. please see [README.laf-intel.md](README.laf-intel.md) + +Then there are different ways of instrumenting the target: + +1. There is an optimized instrumentation strategy that uses CFGs and +markers to just instrument what is needed. This increases speed by 10-15% +without any disadvantages +If you want to use this, set AFL_LLVM_INSTRUMENT=CFG or AFL_LLVM_INSTRIM=1 +See [README.instrim.md](README.instrim.md) + +2. An even better instrumentation strategy uses LTO and link time +instrumentation. Note that not all targets can compile in this mode, however +if it works it is the best option you can use. +Simply use afl-clang-lto/afl-clang-lto++ to use this option. +See [README.lto.md](README.lto.md) + +3. Alternativly you can choose a completely different coverage method: + +3a. N-GRAM coverage - which combines the previous visited edges with the +current one. This explodes the map but on the other hand has proven to be +effective for fuzzing. +See [README.ngram.md](README.ngram.md) + +3b. Context sensitive coverage - which combines the visited edges with an +individual caller ID (the function that called the current one) +[README.ctx.md](README.ctx.md) + +Then - additionally to one of the instrumentation options above - there is +a very effective new instrumentation option called CmpLog as an alternative to +laf-intel that allow AFL++ to apply mutations similar to Redqueen. +See [README.cmplog.md](README.cmplog.md) + +Finally if your llvm version is 8 or lower, you can activate a mode that +prevents that a counter overflow result in a 0 value. This is good for +path discovery, but the llvm implementation for x86 for this functionality +is not optimal and was only fixed in llvm 9. +You can set this with AFL_LLVM_NOT_ZERO=1 +See [README.neverzero.md](README.neverzero.md) + +## 4) Snapshot feature + +To speed up fuzzing you can use a linux loadable kernel module which enables +a snapshot feature. +See [README.snapshot.md](README.snapshot.md) + +## 5) Gotchas, feedback, bugs + +This is an early-stage mechanism, so field reports are welcome. You can send bug +reports to . + +## 6) deferred initialization, persistent mode, shared memory fuzzing + +This is the most powerful and effective fuzzing you can do. +Please see [README.persistent_mode.md](README.persistent_mode.md) for a +full explanation. + +## 7) Bonus feature: 'trace-pc-guard' mode + +LLVM is shipping with a built-in execution tracing feature +that provides AFL with the necessary tracing data without the need to +post-process the assembly or install any compiler plugins. See: + + http://clang.llvm.org/docs/SanitizerCoverage.html#tracing-pcs-with-guards + +If you have not an outdated compiler and want to give it a try, build +targets this way: + +``` +AFL_LLVM_INSTRUMENT=PCGUARD make +``` + +Note that this us currently the default if you use LLVM >= 7, as it is the best +mode. Recommended is LLVM >= 9. +If you have llvm 11+ and compiled afl-clang-lto - this is the only better mode. + +## 8) Bonus feature: 'dict2file' pass + +Just specify `AFL_LLVM_DICT2FILE=/absolute/path/file.txt` and during compilation +all constant string compare parameters will be written to this file to be +used with afl-fuzz' `-x` option. diff --git a/instrumentation/README.lto.md b/instrumentation/README.lto.md new file mode 100644 index 00000000..abdbd2ac --- /dev/null +++ b/instrumentation/README.lto.md @@ -0,0 +1,290 @@ +# afl-clang-lto - collision free instrumentation at link time + +## TLDR; + +This version requires a current llvm 11+ compiled from the github master. + +1. Use afl-clang-lto/afl-clang-lto++ because it is faster and gives better + coverage than anything else that is out there in the AFL world + +2. You can use it together with llvm_mode: laf-intel and the instrument file listing + features and can be combined with cmplog/Redqueen + +3. It only works with llvm 11+ + +4. AUTODICTIONARY feature! see below + +5. If any problems arise be sure to set `AR=llvm-ar RANLIB=llvm-ranlib`. + Some targets might need `LD=afl-clang-lto` and others `LD=afl-ld-lto`. + +## Introduction and problem description + +A big issue with how afl/afl++ works is that the basic block IDs that are +set during compilation are random - and hence naturally the larger the number +of instrumented locations, the higher the number of edge collisions are in the +map. This can result in not discovering new paths and therefore degrade the +efficiency of the fuzzing process. + +*This issue is underestimated in the fuzzing community!* +With a 2^16 = 64kb standard map at already 256 instrumented blocks there is +on average one collision. On average a target has 10.000 to 50.000 +instrumented blocks hence the real collisions are between 750-18.000! + +To reach a solution that prevents any collisions took several approaches +and many dead ends until we got to this: + + * We instrument at link time when we have all files pre-compiled + * To instrument at link time we compile in LTO (link time optimization) mode + * Our compiler (afl-clang-lto/afl-clang-lto++) takes care of setting the + correct LTO options and runs our own afl-ld linker instead of the system + linker + * The LLVM linker collects all LTO files to link and instruments them so that + we have non-colliding edge overage + * We use a new (for afl) edge coverage - which is the same as in llvm + -fsanitize=coverage edge coverage mode :) + +The result: + * 10-25% speed gain compared to llvm_mode + * guaranteed non-colliding edge coverage :-) + * The compile time especially for binaries to an instrumented library can be + much longer + +Example build output from a libtiff build: +``` +libtool: link: afl-clang-lto -g -O2 -Wall -W -o thumbnail thumbnail.o ../libtiff/.libs/libtiff.a ../port/.libs/libport.a -llzma -ljbig -ljpeg -lz -lm +afl-clang-lto++2.63d by Marc "vanHauser" Heuse in mode LTO +afl-llvm-lto++2.63d by Marc "vanHauser" Heuse +AUTODICTIONARY: 11 strings found +[+] Instrumented 12071 locations with no collisions (on average 1046 collisions would be in afl-gcc/afl-clang-fast) (non-hardened mode). +``` + +## Getting llvm 11+ + +### Installing llvm from the llvm repository (version 11) + +Installing the llvm snapshot builds is easy and mostly painless: + +In the follow line change `NAME` for your Debian or Ubuntu release name +(e.g. buster, focal, eon, etc.): +``` +echo deb http://apt.llvm.org/NAME/ llvm-toolchain-NAME NAME >> /etc/apt/sources.list +``` +then add the pgp key of llvm and install the packages: +``` +wget -O - https://apt.llvm.org/llvm-snapshot.gpg.key | apt-key add - +apt-get update && apt-get upgrade -y +apt-get install -y clang-11 clang-tools-11 libc++1-11 libc++-11-dev \ + libc++abi1-11 libc++abi-11-dev libclang1-11 libclang-11-dev \ + libclang-common-11-dev libclang-cpp11 libclang-cpp11-dev liblld-11 \ + liblld-11-dev liblldb-11 liblldb-11-dev libllvm11 libomp-11-dev \ + libomp5-11 lld-11 lldb-11 llvm-11 llvm-11-dev llvm-11-runtime llvm-11-tools +``` + +### Building llvm yourself (version 12) + +Building llvm from github takes quite some long time and is not painless: +``` +sudo apt install binutils-dev # this is *essential*! +git clone https://github.com/llvm/llvm-project +cd llvm-project +mkdir build +cd build +cmake -DLLVM_ENABLE_PROJECTS='clang;clang-tools-extra;compiler-rt;libclc;libcxx;libcxxabi;libunwind;lld' -DCMAKE_BUILD_TYPE=Release -DLLVM_BINUTILS_INCDIR=/usr/include/ ../llvm/ +make -j $(nproc) +export PATH=`pwd`/bin:$PATH +export LLVM_CONFIG=`pwd`/bin/llvm-config +cd /path/to/AFLplusplus/ +make +sudo make install +``` + +## How to use afl-clang-lto + +Just use afl-clang-lto like you did with afl-clang-fast or afl-gcc. + +Also the instrument file listing (AFL_LLVM_ALLOWLIST/AFL_LLVM_DENYLIST -> [README.instrument_list.md](README.instrument_list.md)) and +laf-intel/compcov (AFL_LLVM_LAF_* -> [README.laf-intel.md](README.laf-intel.md)) work. + +Example: +``` +CC=afl-clang-lto CXX=afl-clang-lto++ RANLIB=llvm-ranlib AR=llvm-ar ./configure +make +``` + +NOTE: some targets also need to set the linker, try both `afl-clang-lto` and +`afl-ld-lto` for `LD=` before `configure`. + +## AUTODICTIONARY feature + +While compiling, a dictionary based on string comparisons is automatically +generated and put into the target binary. This dictionary is transfered to afl-fuzz +on start. This improves coverage statistically by 5-10% :) + +## Fixed memory map + +To speed up fuzzing, it is possible to set a fixed shared memory map. +Recommended is the value 0x10000. +In most cases this will work without any problems. However if a target uses +early constructors, ifuncs or a deferred forkserver this can crash the target. +On unusual operating systems/processors/kernels or weird libraries this might +fail so to change the fixed address at compile time set +AFL_LLVM_MAP_ADDR with a better value (a value of 0 or empty sets the map address +to be dynamic - the original afl way, which is slower). + +## Document edge IDs + +Setting `export AFL_LLVM_DOCUMENT_IDS=file` will document in a file which edge +ID was given to which function. This helps to identify functions with variable +bytes or which functions were touched by an input. + +## Solving difficult targets + +Some targets are difficult because the configure script does unusual stuff that +is unexpected for afl. See the next chapter `Potential issues` for how to solve +these. + +### Example: ffmpeg + +An example of a hard to solve target is ffmpeg. Here is how to successfully +instrument it: + +1. Get and extract the current ffmpeg and change to its directory + +2. Running configure with --cc=clang fails and various other items will fail + when compiling, so we have to trick configure: + +``` +./configure --enable-lto --disable-shared --disable-inline-asm +``` + +3. Now the configuration is done - and we edit the settings in `./ffbuild/config.mak` + (-: the original line, +: what to change it into): +``` +-CC=gcc ++CC=afl-clang-lto +-CXX=g++ ++CXX=afl-clang-lto++ +-AS=gcc ++AS=llvm-as +-LD=gcc ++LD=afl-clang-lto++ +-DEPCC=gcc ++DEPCC=afl-clang-lto +-DEPAS=gcc ++DEPAS=afl-clang-lto++ +-AR=ar ++AR=llvm-ar +-AR_CMD=ar ++AR_CMD=llvm-ar +-NM_CMD=nm -g ++NM_CMD=llvm-nm -g +-RANLIB=ranlib -D ++RANLIB=llvm-ranlib -D +``` + +4. Then type make, wait for a long time and you are done :) + +### Example: WebKit jsc + +Building jsc is difficult as the build script has bugs. + +1. checkout Webkit: +``` +svn checkout https://svn.webkit.org/repository/webkit/trunk WebKit +cd WebKit +``` + +2. Fix the build environment: +``` +mkdir -p WebKitBuild/Release +cd WebKitBuild/Release +ln -s ../../../../../usr/bin/llvm-ar-12 llvm-ar-12 +ln -s ../../../../../usr/bin/llvm-ranlib-12 llvm-ranlib-12 +cd ../.. +``` + +3. Build :) + +``` +Tools/Scripts/build-jsc --jsc-only --cli --cmakeargs="-DCMAKE_AR='llvm-ar-12' -DCMAKE_RANLIB='llvm-ranlib-12' -DCMAKE_VERBOSE_MAKEFILE:BOOL=ON -DCMAKE_CC_FLAGS='-O3 -lrt' -DCMAKE_CXX_FLAGS='-O3 -lrt' -DIMPORTED_LOCATION='/lib/x86_64-linux-gnu/' -DCMAKE_CC=afl-clang-lto -DCMAKE_CXX=afl-clang-lto++ -DENABLE_STATIC_JSC=ON" +``` + +## Potential issues + +### compiling libraries fails + +If you see this message: +``` +/bin/ld: libfoo.a: error adding symbols: archive has no index; run ranlib to add one +``` +This is because usually gnu gcc ranlib is being called which cannot deal with clang LTO files. +The solution is simple: when you ./configure you also have to set RANLIB=llvm-ranlib and AR=llvm-ar + +Solution: +``` +AR=llvm-ar RANLIB=llvm-ranlib CC=afl-clang-lto CXX=afl-clang-lto++ ./configure --disable-shared +``` +and on some targets you have to set AR=/RANLIB= even for make as the configure script does not save it. +Other targets ignore environment variables and need the parameters set via +`./configure --cc=... --cxx= --ranlib= ...` etc. (I am looking at you ffmpeg!). + + +If you see this message +``` +assembler command failed ... +``` +then try setting `llvm-as` for configure: +``` +AS=llvm-as ... +``` + +### compiling programs still fail + +afl-clang-lto is still work in progress. + +Known issues: + * Anything that llvm 11+ cannot compile, afl-clang-lto cannot compile either - obviously + * Anything that does not compile with LTO, afl-clang-lto cannot compile either - obviously + +Hence if building a target with afl-clang-lto fails try to build it with llvm12 +and LTO enabled (`CC=clang-12` `CXX=clang++-12` `CFLAGS=-flto=full` and +`CXXFLAGS=-flto=full`). + +If this succeeeds then there is an issue with afl-clang-lto. Please report at +[https://github.com/AFLplusplus/AFLplusplus/issues/226](https://github.com/AFLplusplus/AFLplusplus/issues/226) + +Even some targets where clang-12 fails can be build if the fail is just in +`./configure`, see `Solving difficult targets` above. + +## History + +This was originally envisioned by hexcoder- in Summer 2019, however we saw no +way to create a pass that is run at link time - although there is a option +for this in the PassManager: EP_FullLinkTimeOptimizationLast +("Fun" info - nobody knows what this is doing. And the developer who +implemented this didn't respond to emails.) + +In December then came the idea to implement this as a pass that is run via +the llvm "opt" program, which is performed via an own linker that afterwards +calls the real linker. +This was first implemented in January and work ... kinda. +The LTO time instrumentation worked, however "how" the basic blocks were +instrumented was a problem, as reducing duplicates turned out to be very, +very difficult with a program that has so many paths and therefore so many +dependencies. A lot of strategies were implemented - and failed. +And then sat solvers were tried, but with over 10.000 variables that turned +out to be a dead-end too. + +The final idea to solve this came from domenukk who proposed to insert a block +into an edge and then just use incremental counters ... and this worked! +After some trials and errors to implement this vanhauser-thc found out that +there is actually an llvm function for this: SplitEdge() :-) + +Still more problems came up though as this only works without bugs from +llvm 9 onwards, and with high optimization the link optimization ruins +the instrumented control flow graph. + +This is all now fixed with llvm 11+. The llvm's own linker is now able to +load passes and this bypasses all problems we had. + +Happy end :) diff --git a/instrumentation/README.neverzero.md b/instrumentation/README.neverzero.md new file mode 100644 index 00000000..5c894d6e --- /dev/null +++ b/instrumentation/README.neverzero.md @@ -0,0 +1,35 @@ +# NeverZero counters for LLVM instrumentation + +## Usage + +In larger, complex or reiterative programs the byte sized counters that collect +the edge coverage can easily fill up and wrap around. +This is not that much of an issue - unless by chance it wraps just to a value +of zero when the program execution ends. +In this case afl-fuzz is not able to see that the edge has been accessed and +will ignore it. + +NeverZero prevents this behaviour. If a counter wraps, it jumps over the value +0 directly to a 1. This improves path discovery (by a very little amount) +at a very little cost (one instruction per edge). + +(The alternative of saturated counters has been tested also and proved to be +inferior in terms of path discovery.) + +This is implemented in afl-gcc, however for llvm_mode this is optional if +the llvm version is below 9 - as there is a perfomance bug that is only fixed +in version 9 and onwards. + +If you want to enable this for llvm versions below 9 then set + +``` +export AFL_LLVM_NOT_ZERO=1 +``` + +In case you are on llvm 9 or greater and you do not want this behaviour then +you can set: +``` +AFL_LLVM_SKIP_NEVERZERO=1 +``` +If the target does not have extensive loops or functions that are called +a lot then this can give a small performance boost. diff --git a/instrumentation/README.ngram.md b/instrumentation/README.ngram.md new file mode 100644 index 00000000..de3ba432 --- /dev/null +++ b/instrumentation/README.ngram.md @@ -0,0 +1,28 @@ +# AFL N-Gram Branch Coverage + +## Source + +This is an LLVM-based implementation of the n-gram branch coverage proposed in +the paper ["Be Sensitive and Collaborative: Analzying Impact of Coverage Metrics +in Greybox Fuzzing"](https://www.usenix.org/system/files/raid2019-wang-jinghan.pdf), +by Jinghan Wang, et. al. + +Note that the original implementation (available +[here](https://github.com/bitsecurerlab/afl-sensitive)) +is built on top of AFL's QEMU mode. +This is essentially a port that uses LLVM vectorized instructions to achieve +the same results when compiling source code. + +In math the branch coverage is performed as follows: +`map[current_location ^ prev_location[0] >> 1 ^ prev_location[1] >> 1 ^ ... up to n-1`] += 1` + +## Usage + +The size of `n` (i.e., the number of branches to remember) is an option +that is specified either in the `AFL_LLVM_INSTRUMENT=NGRAM-{value}` or the +`AFL_LLVM_NGRAM_SIZE` environment variable. +Good values are 2, 4 or 8, valid are 2-16. + +It is highly recommended to increase the MAP_SIZE_POW2 definition in +config.h to at least 18 and maybe up to 20 for this as otherwise too +many map collisions occur. diff --git a/instrumentation/README.persistent_mode.md b/instrumentation/README.persistent_mode.md new file mode 100644 index 00000000..e095f036 --- /dev/null +++ b/instrumentation/README.persistent_mode.md @@ -0,0 +1,209 @@ +# llvm_mode persistent mode + +## 1) Introduction + +The most effective way is to fuzz in persistent mode, as the speed can easily +be x10 or x20 times faster without any disadvanges. +*All professional fuzzing is using this mode.* + +This requires that the target can be called in a (or several) function(s), +and that its state can be resetted so that multiple calls can be performed +without resource leaks and former runs having no impact on following runs +(this can be seen by the `stability` indicator in the `afl-fuzz` UI). + +Examples can be found in [examples/persistent_mode](../examples/persistent_mode). + +## 2) TLDR; + +Example `fuzz_target.c`: +``` +#include "what_you_need_for_your_target.h" + +__AFL_FUZZ_INIT(); + +main() { + +#ifdef __AFL_HAVE_MANUAL_CONTROL + __AFL_INIT(); +#endif + + unsigned char *buf = __AFL_FUZZ_TESTCASE_BUF; // must be after __AFL_INIT + + while (__AFL_LOOP(10000)) { + + int len = __AFL_FUZZ_TESTCASE_LEN; + if (len < 8) continue; // check for a required/useful minimum input length + + /* Setup function call, e.g. struct target *tmp = libtarget_init() */ + /* Call function to be fuzzed, e.g.: */ + target_function(buf, len); + /* Reset state. e.g. libtarget_free(tmp) */ + + } + + return 0; + +} +``` +And then compile: +``` +afl-clang-fast -o fuzz_target fuzz_target.c -lwhat_you_need_for_your_target +``` +And that is it! +The speed increase is usually x10 to x20. + +If you want to be able to compile the target without afl-clang-fast/lto then +add this just after the includes: + +``` +#ifndef __AFL_FUZZ_TESTCASE_LEN + ssize_t fuzz_len; + #define __AFL_FUZZ_TESTCASE_LEN fuzz_len + unsigned char fuzz_buf[1024000]; + #define __AFL_FUZZ_TESTCASE_BUF fuzz_buf + #define __AFL_FUZZ_INIT() void sync(void); + #define __AFL_LOOP(x) ((fuzz_len = read(0, fuzz_buf, sizeof(fuzz_buf))) > 0 ? + #define __AFL_INIT() sync() +#endif +``` + +## 3) Deferred initialization + +AFL tries to optimize performance by executing the targeted binary just once, +stopping it just before main(), and then cloning this "main" process to get +a steady supply of targets to fuzz. + +Although this approach eliminates much of the OS-, linker- and libc-level +costs of executing the program, it does not always help with binaries that +perform other time-consuming initialization steps - say, parsing a large config +file before getting to the fuzzed data. + +In such cases, it's beneficial to initialize the forkserver a bit later, once +most of the initialization work is already done, but before the binary attempts +to read the fuzzed input and parse it; in some cases, this can offer a 10x+ +performance gain. You can implement delayed initialization in LLVM mode in a +fairly simple way. + +First, find a suitable location in the code where the delayed cloning can +take place. This needs to be done with *extreme* care to avoid breaking the +binary. In particular, the program will probably malfunction if you select +a location after: + + - The creation of any vital threads or child processes - since the forkserver + can't clone them easily. + + - The initialization of timers via setitimer() or equivalent calls. + + - The creation of temporary files, network sockets, offset-sensitive file + descriptors, and similar shared-state resources - but only provided that + their state meaningfully influences the behavior of the program later on. + + - Any access to the fuzzed input, including reading the metadata about its + size. + +With the location selected, add this code in the appropriate spot: + +```c +#ifdef __AFL_HAVE_MANUAL_CONTROL + __AFL_INIT(); +#endif +``` + +You don't need the #ifdef guards, but including them ensures that the program +will keep working normally when compiled with a tool other than afl-clang-fast. + +Finally, recompile the program with afl-clang-fast/lto (afl-gcc or afl-clang will +*not* generate a deferred-initialization binary) - and you should be all set! + +*NOTE:* In the code between `main` and `__AFL_INIT()` should not be any code +run that is instrumented - otherwise a crash might occure. +In case this is useful (e.g. for expensive one time initialization) you can +try to do the following: + +Add after the includes: +``` +extern unsigned char *__afl_area_ptr; +#define MAX_DUMMY_SIZE 256000 + +__attribute__((constructor(1))) void __afl_protect(void) { +#ifdef MAP_FIXED_NOREPLACE + __afl_area_ptr = (unsigned char*) mmap((void *)0x10000, MAX_DUMMY_SIZE, PROT_READ | PROT_WRITE, MAP_FIXED_NOREPLACE | MAP_SHARED | MAP_ANONYMOUS, -1, 0); + if ((uint64_t)__afl_area_ptr == -1) +#endif + __afl_area_ptr = (unsigned char*) mmap((void *)0x10000, MAX_DUMMY_SIZE, PROT_READ | PROT_WRITE, MAP_SHARED | MAP_ANONYMOUS, -1, 0); + if ((uint64_t)__afl_area_ptr == -1) + __afl_area_ptr = (unsigned char*) mmap(NULL, MAX_DUMMY_SIZE, PROT_READ | PROT_WRITE, MAP_SHARED | MAP_ANONYMOUS, -1, 0); +} + +``` +and just before `__AFL_INIT()`: +``` + munmap(__afl_area_ptr, MAX_DUMMY_SIZE); + __afl_area_ptr = NULL; +``` + +## 4) Persistent mode + +Some libraries provide APIs that are stateless, or whose state can be reset in +between processing different input files. When such a reset is performed, a +single long-lived process can be reused to try out multiple test cases, +eliminating the need for repeated fork() calls and the associated OS overhead. + +The basic structure of the program that does this would be: + +```c + while (__AFL_LOOP(1000)) { + + /* Read input data. */ + /* Call library code to be fuzzed. */ + /* Reset state. */ + + } + + /* Exit normally */ +``` + +The numerical value specified within the loop controls the maximum number +of iterations before AFL will restart the process from scratch. This minimizes +the impact of memory leaks and similar glitches; 1000 is a good starting point, +and going much higher increases the likelihood of hiccups without giving you +any real performance benefits. + +A more detailed template is shown in ../examples/persistent_demo/. +Similarly to the previous mode, the feature works only with afl-clang-fast; #ifdef +guards can be used to suppress it when using other compilers. + +Note that as with the previous mode, the feature is easy to misuse; if you +do not fully reset the critical state, you may end up with false positives or +waste a whole lot of CPU power doing nothing useful at all. Be particularly +wary of memory leaks and of the state of file descriptors. + +PS. Because there are task switches still involved, the mode isn't as fast as +"pure" in-process fuzzing offered, say, by LLVM's LibFuzzer; but it is a lot +faster than the normal fork() model, and compared to in-process fuzzing, +should be a lot more robust. + +## 5) Shared memory fuzzing + +You can speed up the fuzzing process even more by receiving the fuzzing data +via shared memory instead of stdin or files. +This is a further speed multiplier of about 2x. + +Setting this up is very easy: + +After the includes set the following macro: + +``` +__AFL_FUZZ_INIT(); +``` +Directly at the start of main - or if you are using the deferred forkserver +with `__AFL_INIT()` then *after* `__AFL_INIT? : +``` + unsigned char *buf = __AFL_FUZZ_TESTCASE_BUF; +``` + +Then as first line after the `__AFL_LOOP` while loop: +``` + int len = __AFL_FUZZ_TESTCASE_LEN; +``` +and that is all! diff --git a/instrumentation/README.snapshot.md b/instrumentation/README.snapshot.md new file mode 100644 index 00000000..c40a956a --- /dev/null +++ b/instrumentation/README.snapshot.md @@ -0,0 +1,16 @@ +# AFL++ snapshot feature + +Snapshotting is a feature that makes a snapshot from a process and then +restores its state, which is faster then forking it again. + +All targets compiled with llvm_mode are automatically enabled for the +snapshot feature. + +To use the snapshot feature for fuzzing compile and load this kernel +module: [https://github.com/AFLplusplus/AFL-Snapshot-LKM](https://github.com/AFLplusplus/AFL-Snapshot-LKM) + +Note that is has little value for persistent (__AFL_LOOP) fuzzing. + +## Notes + +Snapshot does not work with multithreaded targets yet. Still in WIP, it is now usable only for single threaded applications. diff --git a/instrumentation/SanitizerCoverageLTO.so.cc b/instrumentation/SanitizerCoverageLTO.so.cc new file mode 100644 index 00000000..1dd65188 --- /dev/null +++ b/instrumentation/SanitizerCoverageLTO.so.cc @@ -0,0 +1,1503 @@ +/* SanitizeCoverage.cpp ported to afl++ LTO :-) */ + +#define AFL_LLVM_PASS + +#include +#include +#include +#include +#include + +#include +#include +#include +#include + +#include "llvm/Transforms/Instrumentation/SanitizerCoverage.h" +#include "llvm/ADT/ArrayRef.h" +#include "llvm/ADT/SmallVector.h" +#include "llvm/Analysis/EHPersonalities.h" +#include "llvm/Analysis/PostDominators.h" +#include "llvm/Analysis/ValueTracking.h" +#include "llvm/IR/BasicBlock.h" +#include "llvm/IR/CFG.h" +#include "llvm/IR/Constant.h" +#include "llvm/IR/DataLayout.h" +#include "llvm/IR/DebugInfo.h" +#include "llvm/IR/Dominators.h" +#include "llvm/IR/Function.h" +#include "llvm/IR/GlobalVariable.h" +#include "llvm/IR/IRBuilder.h" +#include "llvm/IR/InlineAsm.h" +#include "llvm/IR/Instructions.h" +#include "llvm/IR/IntrinsicInst.h" +#include "llvm/IR/Intrinsics.h" +#include "llvm/IR/LegacyPassManager.h" +#include "llvm/IR/LLVMContext.h" +#include "llvm/IR/MDBuilder.h" +#include "llvm/IR/Mangler.h" +#include "llvm/IR/Module.h" +#include "llvm/IR/Type.h" +#include "llvm/InitializePasses.h" +#include "llvm/Pass.h" +#include "llvm/Support/CommandLine.h" +#include "llvm/Support/Debug.h" +#include "llvm/Support/SpecialCaseList.h" +#include "llvm/Support/VirtualFileSystem.h" +#include "llvm/Support/raw_ostream.h" +#include "llvm/Transforms/Instrumentation.h" +#include "llvm/Transforms/IPO/PassManagerBuilder.h" +#include "llvm/Transforms/Utils/BasicBlockUtils.h" +#include "llvm/Transforms/Utils/BasicBlockUtils.h" +#include "llvm/Transforms/Utils/ModuleUtils.h" + +#include "config.h" +#include "debug.h" +#include "afl-llvm-common.h" + +using namespace llvm; + +#define DEBUG_TYPE "sancov" + +static const char *const SanCovTracePCIndirName = + "__sanitizer_cov_trace_pc_indir"; +static const char *const SanCovTracePCName = "__sanitizer_cov_trace_pc"; +// static const char *const SanCovTracePCGuardName = +// "__sanitizer_cov_trace_pc_guard"; +static const char *const SanCovGuardsSectionName = "sancov_guards"; +static const char *const SanCovCountersSectionName = "sancov_cntrs"; +static const char *const SanCovBoolFlagSectionName = "sancov_bools"; +static const char *const SanCovPCsSectionName = "sancov_pcs"; + +static cl::opt ClCoverageLevel( + "lto-coverage-level", + cl::desc("Sanitizer Coverage. 0: none, 1: entry block, 2: all blocks, " + "3: all blocks and critical edges"), + cl::Hidden, cl::init(3)); + +static cl::opt ClTracePC("lto-coverage-trace-pc", + cl::desc("Experimental pc tracing"), cl::Hidden, + cl::init(false)); + +static cl::opt ClTracePCGuard("lto-coverage-trace-pc-guard", + cl::desc("pc tracing with a guard"), + cl::Hidden, cl::init(false)); + +// If true, we create a global variable that contains PCs of all instrumented +// BBs, put this global into a named section, and pass this section's bounds +// to __sanitizer_cov_pcs_init. +// This way the coverage instrumentation does not need to acquire the PCs +// at run-time. Works with trace-pc-guard, inline-8bit-counters, and +// inline-bool-flag. +static cl::opt ClCreatePCTable("lto-coverage-pc-table", + cl::desc("create a static PC table"), + cl::Hidden, cl::init(false)); + +static cl::opt ClInline8bitCounters( + "lto-coverage-inline-8bit-counters", + cl::desc("increments 8-bit counter for every edge"), cl::Hidden, + cl::init(false)); + +static cl::opt ClInlineBoolFlag( + "lto-coverage-inline-bool-flag", + cl::desc("sets a boolean flag for every edge"), cl::Hidden, + cl::init(false)); + +static cl::opt ClPruneBlocks( + "lto-coverage-prune-blocks", + cl::desc("Reduce the number of instrumented blocks"), cl::Hidden, + cl::init(true)); + +namespace { + +SanitizerCoverageOptions getOptions(int LegacyCoverageLevel) { + + SanitizerCoverageOptions Res; + switch (LegacyCoverageLevel) { + + case 0: + Res.CoverageType = SanitizerCoverageOptions::SCK_None; + break; + case 1: + Res.CoverageType = SanitizerCoverageOptions::SCK_Function; + break; + case 2: + Res.CoverageType = SanitizerCoverageOptions::SCK_BB; + break; + case 3: + Res.CoverageType = SanitizerCoverageOptions::SCK_Edge; + break; + case 4: + Res.CoverageType = SanitizerCoverageOptions::SCK_Edge; + Res.IndirectCalls = true; + break; + + } + + return Res; + +} + +SanitizerCoverageOptions OverrideFromCL(SanitizerCoverageOptions Options) { + + // Sets CoverageType and IndirectCalls. + SanitizerCoverageOptions CLOpts = getOptions(ClCoverageLevel); + Options.CoverageType = std::max(Options.CoverageType, CLOpts.CoverageType); + Options.IndirectCalls |= CLOpts.IndirectCalls; + Options.TracePC |= ClTracePC; + Options.TracePCGuard |= ClTracePCGuard; + Options.Inline8bitCounters |= ClInline8bitCounters; + Options.InlineBoolFlag |= ClInlineBoolFlag; + Options.PCTable |= ClCreatePCTable; + Options.NoPrune |= !ClPruneBlocks; + if (!Options.TracePCGuard && !Options.TracePC && + !Options.Inline8bitCounters && !Options.InlineBoolFlag) + Options.TracePCGuard = true; // TracePCGuard is default. + return Options; + +} + +using DomTreeCallback = function_ref; +using PostDomTreeCallback = + function_ref; + +class ModuleSanitizerCoverage { + + public: + ModuleSanitizerCoverage( + const SanitizerCoverageOptions &Options = SanitizerCoverageOptions()) + : Options(OverrideFromCL(Options)) { + + /* , + const SpecialCaseList * Allowlist = nullptr, + const SpecialCaseList * Blocklist = nullptr) + , + Allowlist(Allowlist), + Blocklist(Blocklist) { + + */ + + } + + bool instrumentModule(Module &M, DomTreeCallback DTCallback, + PostDomTreeCallback PDTCallback); + + private: + void instrumentFunction(Function &F, DomTreeCallback DTCallback, + PostDomTreeCallback PDTCallback); + void InjectCoverageForIndirectCalls(Function & F, + ArrayRef IndirCalls); + bool InjectCoverage(Function &F, ArrayRef AllBlocks, + bool IsLeafFunc = true); + GlobalVariable *CreateFunctionLocalArrayInSection(size_t NumElements, + Function &F, Type *Ty, + const char *Section); + GlobalVariable *CreatePCArray(Function &F, ArrayRef AllBlocks); + void CreateFunctionLocalArrays(Function &F, ArrayRef AllBlocks); + void InjectCoverageAtBlock(Function &F, BasicBlock &BB, size_t Idx, + bool IsLeafFunc = true); + std::pair CreateSecStartEnd(Module &M, const char *Section, + Type *Ty); + + void SetNoSanitizeMetadata(Instruction *I) { + + I->setMetadata(I->getModule()->getMDKindID("nosanitize"), + MDNode::get(*C, None)); + + } + + std::string getSectionName(const std::string &Section) const; + std::string getSectionStart(const std::string &Section) const; + std::string getSectionEnd(const std::string &Section) const; + FunctionCallee SanCovTracePCIndir; + FunctionCallee SanCovTracePC /*, SanCovTracePCGuard*/; + Type *IntptrTy, *IntptrPtrTy, *Int64Ty, *Int64PtrTy, *Int32Ty, *Int32PtrTy, + *Int16Ty, *Int8Ty, *Int8PtrTy, *Int1Ty, *Int1PtrTy; + Module * CurModule; + std::string CurModuleUniqueId; + Triple TargetTriple; + LLVMContext * C; + const DataLayout *DL; + + GlobalVariable *FunctionGuardArray; // for trace-pc-guard. + GlobalVariable *Function8bitCounterArray; // for inline-8bit-counters. + GlobalVariable *FunctionBoolArray; // for inline-bool-flag. + GlobalVariable *FunctionPCsArray; // for pc-table. + SmallVector GlobalsToAppendToUsed; + SmallVector GlobalsToAppendToCompilerUsed; + + SanitizerCoverageOptions Options; + + // afl++ START + // const SpecialCaseList * Allowlist; + // const SpecialCaseList * Blocklist; + uint32_t autodictionary = 1; + uint32_t inst = 0; + uint32_t afl_global_id = 0; + uint64_t map_addr = 0; + char * skip_nozero = NULL; + std::vector BlockList; + DenseMap valueMap; + std::vector dictionary; + IntegerType * Int8Tyi = NULL; + IntegerType * Int32Tyi = NULL; + IntegerType * Int64Tyi = NULL; + ConstantInt * Zero = NULL; + ConstantInt * One = NULL; + LLVMContext * Ct = NULL; + Module * Mo = NULL; + GlobalVariable * AFLMapPtr = NULL; + Value * MapPtrFixed = NULL; + FILE * documentFile = NULL; + // afl++ END + +}; + +class ModuleSanitizerCoverageLegacyPass : public ModulePass { + + public: + static char ID; + StringRef getPassName() const override { + + return "sancov"; + + } + + void getAnalysisUsage(AnalysisUsage &AU) const override { + + AU.addRequired(); + AU.addRequired(); + + } + + ModuleSanitizerCoverageLegacyPass( + const SanitizerCoverageOptions &Options = SanitizerCoverageOptions()) + : ModulePass(ID), Options(Options) { + + /* , + const std::vector &AllowlistFiles = + std::vector(), + const std::vector &BlocklistFiles = + std::vector()) + if (AllowlistFiles.size() > 0) + Allowlist = SpecialCaseList::createOrDie(AllowlistFiles, + *vfs::getRealFileSystem()); + if (BlocklistFiles.size() > 0) + Blocklist = SpecialCaseList::createOrDie(BlocklistFiles, + *vfs::getRealFileSystem()); + */ + initializeModuleSanitizerCoverageLegacyPassPass( + *PassRegistry::getPassRegistry()); + + } + + bool runOnModule(Module &M) override { + + ModuleSanitizerCoverage ModuleSancov(Options); + // , Allowlist.get(), Blocklist.get()); + auto DTCallback = [this](Function &F) -> const DominatorTree * { + + return &this->getAnalysis(F).getDomTree(); + + }; + + auto PDTCallback = [this](Function &F) -> const PostDominatorTree * { + + return &this->getAnalysis(F) + .getPostDomTree(); + + }; + + return ModuleSancov.instrumentModule(M, DTCallback, PDTCallback); + + } + + private: + SanitizerCoverageOptions Options; + + // std::unique_ptr Allowlist; + // std::unique_ptr Blocklist; + +}; + +} // namespace + +PreservedAnalyses ModuleSanitizerCoveragePass::run(Module & M, + ModuleAnalysisManager &MAM) { + + ModuleSanitizerCoverage ModuleSancov(Options); + // Allowlist.get(), Blocklist.get()); + auto &FAM = MAM.getResult(M).getManager(); + auto DTCallback = [&FAM](Function &F) -> const DominatorTree * { + + return &FAM.getResult(F); + + }; + + auto PDTCallback = [&FAM](Function &F) -> const PostDominatorTree * { + + return &FAM.getResult(F); + + }; + + if (ModuleSancov.instrumentModule(M, DTCallback, PDTCallback)) + return PreservedAnalyses::none(); + + return PreservedAnalyses::all(); + +} + +std::pair ModuleSanitizerCoverage::CreateSecStartEnd( + Module &M, const char *Section, Type *Ty) { + + GlobalVariable *SecStart = + new GlobalVariable(M, Ty, false, GlobalVariable::ExternalLinkage, nullptr, + getSectionStart(Section)); + SecStart->setVisibility(GlobalValue::HiddenVisibility); + GlobalVariable *SecEnd = + new GlobalVariable(M, Ty, false, GlobalVariable::ExternalLinkage, nullptr, + getSectionEnd(Section)); + SecEnd->setVisibility(GlobalValue::HiddenVisibility); + IRBuilder<> IRB(M.getContext()); + Value * SecEndPtr = IRB.CreatePointerCast(SecEnd, Ty); + if (!TargetTriple.isOSBinFormatCOFF()) + return std::make_pair(IRB.CreatePointerCast(SecStart, Ty), SecEndPtr); + + // Account for the fact that on windows-msvc __start_* symbols actually + // point to a uint64_t before the start of the array. + auto SecStartI8Ptr = IRB.CreatePointerCast(SecStart, Int8PtrTy); + auto GEP = IRB.CreateGEP(Int8Ty, SecStartI8Ptr, + ConstantInt::get(IntptrTy, sizeof(uint64_t))); + return std::make_pair(IRB.CreatePointerCast(GEP, Ty), SecEndPtr); + +} + +bool ModuleSanitizerCoverage::instrumentModule( + Module &M, DomTreeCallback DTCallback, PostDomTreeCallback PDTCallback) { + + if (Options.CoverageType == SanitizerCoverageOptions::SCK_None) return false; + /* + if (Allowlist && + !Allowlist->inSection("coverage", "src", M.getSourceFileName())) + return false; + if (Blocklist && + Blocklist->inSection("coverage", "src", M.getSourceFileName())) + return false; + */ + BlockList.clear(); + valueMap.clear(); + dictionary.clear(); + C = &(M.getContext()); + DL = &M.getDataLayout(); + CurModule = &M; + CurModuleUniqueId = getUniqueModuleId(CurModule); + TargetTriple = Triple(M.getTargetTriple()); + FunctionGuardArray = nullptr; + Function8bitCounterArray = nullptr; + FunctionBoolArray = nullptr; + FunctionPCsArray = nullptr; + IntptrTy = Type::getIntNTy(*C, DL->getPointerSizeInBits()); + IntptrPtrTy = PointerType::getUnqual(IntptrTy); + Type * VoidTy = Type::getVoidTy(*C); + IRBuilder<> IRB(*C); + Int64PtrTy = PointerType::getUnqual(IRB.getInt64Ty()); + Int32PtrTy = PointerType::getUnqual(IRB.getInt32Ty()); + Int8PtrTy = PointerType::getUnqual(IRB.getInt8Ty()); + Int1PtrTy = PointerType::getUnqual(IRB.getInt1Ty()); + Int64Ty = IRB.getInt64Ty(); + Int32Ty = IRB.getInt32Ty(); + Int16Ty = IRB.getInt16Ty(); + Int8Ty = IRB.getInt8Ty(); + Int1Ty = IRB.getInt1Ty(); + + /* afl++ START */ + char * ptr; + LLVMContext &Ctx = M.getContext(); + Ct = &Ctx; + Int8Tyi = IntegerType::getInt8Ty(Ctx); + Int32Tyi = IntegerType::getInt32Ty(Ctx); + Int64Tyi = IntegerType::getInt64Ty(Ctx); + + /* Show a banner */ + setvbuf(stdout, NULL, _IONBF, 0); + if (getenv("AFL_DEBUG")) debug = 1; + + if ((isatty(2) && !getenv("AFL_QUIET")) || debug) { + + SAYF(cCYA "afl-llvm-lto" VERSION cRST + " by Marc \"vanHauser\" Heuse \n"); + + } else + + be_quiet = 1; + + skip_nozero = getenv("AFL_LLVM_SKIP_NEVERZERO"); + + if ((ptr = getenv("AFL_LLVM_LTO_STARTID")) != NULL) + if ((afl_global_id = atoi(ptr)) < 0) + FATAL("AFL_LLVM_LTO_STARTID value of \"%s\" is negative\n", ptr); + + if ((ptr = getenv("AFL_LLVM_DOCUMENT_IDS")) != NULL) { + + if ((documentFile = fopen(ptr, "a")) == NULL) + WARNF("Cannot access document file %s", ptr); + + } + + // we make this the default as the fixed map has problems with + // defered forkserver, early constructors, ifuncs and maybe more + /*if (getenv("AFL_LLVM_MAP_DYNAMIC"))*/ + map_addr = 0; + + if ((ptr = getenv("AFL_LLVM_MAP_ADDR"))) { + + uint64_t val; + if (!*ptr || !strcmp(ptr, "0") || !strcmp(ptr, "0x0")) { + + map_addr = 0; + + } else if (getenv("AFL_LLVM_MAP_DYNAMIC")) { + + FATAL( + "AFL_LLVM_MAP_ADDR and AFL_LLVM_MAP_DYNAMIC cannot be used together"); + + } else if (strncmp(ptr, "0x", 2) != 0) { + + map_addr = 0x10000; // the default + + } else { + + val = strtoull(ptr, NULL, 16); + if (val < 0x100 || val > 0xffffffff00000000) { + + FATAL( + "AFL_LLVM_MAP_ADDR must be a value between 0x100 and " + "0xffffffff00000000"); + + } + + map_addr = val; + + } + + } + + /* Get/set the globals for the SHM region. */ + + if (!map_addr) { + + AFLMapPtr = + new GlobalVariable(M, PointerType::get(Int8Tyi, 0), false, + GlobalValue::ExternalLinkage, 0, "__afl_area_ptr"); + + } else { + + ConstantInt *MapAddr = ConstantInt::get(Int64Tyi, map_addr); + MapPtrFixed = + ConstantExpr::getIntToPtr(MapAddr, PointerType::getUnqual(Int8Tyi)); + + } + + Zero = ConstantInt::get(Int8Tyi, 0); + One = ConstantInt::get(Int8Tyi, 1); + + scanForDangerousFunctions(&M); + Mo = &M; + + if (autodictionary) { + + for (auto &F : M) { + + for (auto &BB : F) { + + for (auto &IN : BB) { + + CallInst *callInst = nullptr; + + if ((callInst = dyn_cast(&IN))) { + + bool isStrcmp = true; + bool isMemcmp = true; + bool isStrncmp = true; + bool isStrcasecmp = true; + bool isStrncasecmp = true; + bool isIntMemcpy = true; + bool addedNull = false; + size_t optLen = 0; + + Function *Callee = callInst->getCalledFunction(); + if (!Callee) continue; + if (callInst->getCallingConv() != llvm::CallingConv::C) continue; + std::string FuncName = Callee->getName().str(); + isStrcmp &= !FuncName.compare("strcmp"); + isMemcmp &= !FuncName.compare("memcmp"); + isStrncmp &= !FuncName.compare("strncmp"); + isStrcasecmp &= !FuncName.compare("strcasecmp"); + isStrncasecmp &= !FuncName.compare("strncasecmp"); + isIntMemcpy &= !FuncName.compare("llvm.memcpy.p0i8.p0i8.i64"); + + /* we do something different here, putting this BB and the + successors in a block map */ + if (!FuncName.compare("__afl_persistent_loop")) { + + BlockList.push_back(&BB); + for (succ_iterator SI = succ_begin(&BB), SE = succ_end(&BB); + SI != SE; ++SI) { + + BasicBlock *succ = *SI; + BlockList.push_back(succ); + + } + + } + + if (!isStrcmp && !isMemcmp && !isStrncmp && !isStrcasecmp && + !isStrncasecmp && !isIntMemcpy) + continue; + + /* Verify the strcmp/memcmp/strncmp/strcasecmp/strncasecmp function + * prototype */ + FunctionType *FT = Callee->getFunctionType(); + + isStrcmp &= FT->getNumParams() == 2 && + FT->getReturnType()->isIntegerTy(32) && + FT->getParamType(0) == FT->getParamType(1) && + FT->getParamType(0) == + IntegerType::getInt8PtrTy(M.getContext()); + isStrcasecmp &= FT->getNumParams() == 2 && + FT->getReturnType()->isIntegerTy(32) && + FT->getParamType(0) == FT->getParamType(1) && + FT->getParamType(0) == + IntegerType::getInt8PtrTy(M.getContext()); + isMemcmp &= FT->getNumParams() == 3 && + FT->getReturnType()->isIntegerTy(32) && + FT->getParamType(0)->isPointerTy() && + FT->getParamType(1)->isPointerTy() && + FT->getParamType(2)->isIntegerTy(); + isStrncmp &= FT->getNumParams() == 3 && + FT->getReturnType()->isIntegerTy(32) && + FT->getParamType(0) == FT->getParamType(1) && + FT->getParamType(0) == + IntegerType::getInt8PtrTy(M.getContext()) && + FT->getParamType(2)->isIntegerTy(); + isStrncasecmp &= FT->getNumParams() == 3 && + FT->getReturnType()->isIntegerTy(32) && + FT->getParamType(0) == FT->getParamType(1) && + FT->getParamType(0) == + IntegerType::getInt8PtrTy(M.getContext()) && + FT->getParamType(2)->isIntegerTy(); + + if (!isStrcmp && !isMemcmp && !isStrncmp && !isStrcasecmp && + !isStrncasecmp && !isIntMemcpy) + continue; + + /* is a str{n,}{case,}cmp/memcmp, check if we have + * str{case,}cmp(x, "const") or str{case,}cmp("const", x) + * strn{case,}cmp(x, "const", ..) or strn{case,}cmp("const", x, ..) + * memcmp(x, "const", ..) or memcmp("const", x, ..) */ + Value *Str1P = callInst->getArgOperand(0), + *Str2P = callInst->getArgOperand(1); + std::string Str1, Str2; + StringRef TmpStr; + bool HasStr1 = getConstantStringInfo(Str1P, TmpStr); + if (TmpStr.empty()) + HasStr1 = false; + else + Str1 = TmpStr.str(); + bool HasStr2 = getConstantStringInfo(Str2P, TmpStr); + if (TmpStr.empty()) + HasStr2 = false; + else + Str2 = TmpStr.str(); + + if (debug) + fprintf(stderr, "F:%s %p(%s)->\"%s\"(%s) %p(%s)->\"%s\"(%s)\n", + FuncName.c_str(), Str1P, Str1P->getName().str().c_str(), + Str1.c_str(), HasStr1 == true ? "true" : "false", Str2P, + Str2P->getName().str().c_str(), Str2.c_str(), + HasStr2 == true ? "true" : "false"); + + // we handle the 2nd parameter first because of llvm memcpy + if (!HasStr2) { + + auto *Ptr = dyn_cast(Str2P); + if (Ptr && Ptr->isGEPWithNoNotionalOverIndexing()) { + + if (auto *Var = dyn_cast(Ptr->getOperand(0))) { + + if (Var->hasInitializer()) { + + if (auto *Array = dyn_cast( + Var->getInitializer())) { + + HasStr2 = true; + Str2 = Array->getAsString().str(); + + } + + } + + } + + } + + } + + // for the internal memcpy routine we only care for the second + // parameter and are not reporting anything. + if (isIntMemcpy == true) { + + if (HasStr2 == true) { + + Value * op2 = callInst->getArgOperand(2); + ConstantInt *ilen = dyn_cast(op2); + if (ilen) { + + uint64_t literalLength = Str2.size(); + uint64_t optLength = ilen->getZExtValue(); + if (literalLength + 1 == optLength) { + + Str2.append("\0", 1); // add null byte + addedNull = true; + + } + + } + + valueMap[Str1P] = new std::string(Str2); + + if (debug) + fprintf(stderr, "Saved: %s for %p\n", Str2.c_str(), Str1P); + continue; + + } + + continue; + + } + + // Neither a literal nor a global variable? + // maybe it is a local variable that we saved + if (!HasStr2) { + + std::string *strng = valueMap[Str2P]; + if (strng && !strng->empty()) { + + Str2 = *strng; + HasStr2 = true; + if (debug) + fprintf(stderr, "Filled2: %s for %p\n", strng->c_str(), + Str2P); + + } + + } + + if (!HasStr1) { + + auto Ptr = dyn_cast(Str1P); + + if (Ptr && Ptr->isGEPWithNoNotionalOverIndexing()) { + + if (auto *Var = dyn_cast(Ptr->getOperand(0))) { + + if (Var->hasInitializer()) { + + if (auto *Array = dyn_cast( + Var->getInitializer())) { + + HasStr1 = true; + Str1 = Array->getAsString().str(); + + } + + } + + } + + } + + } + + // Neither a literal nor a global variable? + // maybe it is a local variable that we saved + if (!HasStr1) { + + std::string *strng = valueMap[Str1P]; + if (strng && !strng->empty()) { + + Str1 = *strng; + HasStr1 = true; + if (debug) + fprintf(stderr, "Filled1: %s for %p\n", strng->c_str(), + Str1P); + + } + + } + + /* handle cases of one string is const, one string is variable */ + if (!(HasStr1 ^ HasStr2)) continue; + + std::string thestring; + + if (HasStr1) + thestring = Str1; + else + thestring = Str2; + + optLen = thestring.length(); + + if (isMemcmp || isStrncmp || isStrncasecmp) { + + Value * op2 = callInst->getArgOperand(2); + ConstantInt *ilen = dyn_cast(op2); + if (ilen) { + + uint64_t literalLength = optLen; + optLen = ilen->getZExtValue(); + if (literalLength + 1 == optLen) { // add null byte + thestring.append("\0", 1); + addedNull = true; + + } + + } + + } + + // add null byte if this is a string compare function and a null + // was not already added + if (!isMemcmp) { + + if (addedNull == false) { + + thestring.append("\0", 1); // add null byte + optLen++; + + } + + // ensure we do not have garbage + size_t offset = thestring.find('\0', 0); + if (offset + 1 < optLen) optLen = offset + 1; + thestring = thestring.substr(0, optLen); + + } + + if (!be_quiet) { + + std::string outstring; + fprintf(stderr, "%s: length %zu/%zu \"", FuncName.c_str(), optLen, + thestring.length()); + for (uint8_t i = 0; i < thestring.length(); i++) { + + uint8_t c = thestring[i]; + if (c <= 32 || c >= 127) + fprintf(stderr, "\\x%02x", c); + else + fprintf(stderr, "%c", c); + + } + + fprintf(stderr, "\"\n"); + + } + + // we take the longer string, even if the compare was to a + // shorter part. Note that depending on the optimizer of the + // compiler this can be wrong, but it is more likely that this + // is helping the fuzzer + if (optLen != thestring.length()) optLen = thestring.length(); + if (optLen > MAX_AUTO_EXTRA) optLen = MAX_AUTO_EXTRA; + if (optLen < MIN_AUTO_EXTRA) // too short? skip + continue; + + dictionary.push_back(thestring.substr(0, optLen)); + + } + + } + + } + + } + + } + + // afl++ END + + SanCovTracePCIndir = + M.getOrInsertFunction(SanCovTracePCIndirName, VoidTy, IntptrTy); + // Make sure smaller parameters are zero-extended to i64 as required by the + // x86_64 ABI. + AttributeList SanCovTraceCmpZeroExtAL; + if (TargetTriple.getArch() == Triple::x86_64) { + + SanCovTraceCmpZeroExtAL = + SanCovTraceCmpZeroExtAL.addParamAttribute(*C, 0, Attribute::ZExt); + SanCovTraceCmpZeroExtAL = + SanCovTraceCmpZeroExtAL.addParamAttribute(*C, 1, Attribute::ZExt); + + } + + SanCovTracePC = M.getOrInsertFunction(SanCovTracePCName, VoidTy); + + // SanCovTracePCGuard = + // M.getOrInsertFunction(SanCovTracePCGuardName, VoidTy, Int32PtrTy); + + for (auto &F : M) + instrumentFunction(F, DTCallback, PDTCallback); + + // afl++ START + if (documentFile) { + + fclose(documentFile); + documentFile = NULL; + + } + + if (!getenv("AFL_LLVM_LTO_DONTWRITEID") || dictionary.size() || map_addr) { + + // yes we could create our own function, insert it into ctors ... + // but this would be a pain in the butt ... so we use afl-llvm-rt-lto.o + + Function *f = M.getFunction("__afl_auto_init_globals"); + + if (!f) { + + fprintf(stderr, + "Error: init function could not be found (this should not " + "happen)\n"); + exit(-1); + + } + + BasicBlock *bb = &f->getEntryBlock(); + if (!bb) { + + fprintf(stderr, + "Error: init function does not have an EntryBlock (this should " + "not happen)\n"); + exit(-1); + + } + + BasicBlock::iterator IP = bb->getFirstInsertionPt(); + IRBuilder<> IRB(&(*IP)); + + if (map_addr) { + + GlobalVariable *AFLMapAddrFixed = new GlobalVariable( + M, Int64Tyi, true, GlobalValue::ExternalLinkage, 0, "__afl_map_addr"); + ConstantInt *MapAddr = ConstantInt::get(Int64Tyi, map_addr); + StoreInst * StoreMapAddr = IRB.CreateStore(MapAddr, AFLMapAddrFixed); + StoreMapAddr->setMetadata(M.getMDKindID("nosanitize"), + MDNode::get(Ctx, None)); + + } + + if (getenv("AFL_LLVM_LTO_DONTWRITEID") == NULL) { + + uint32_t write_loc = afl_global_id; + + if (afl_global_id % 8) write_loc = (((afl_global_id + 8) >> 3) << 3); + + GlobalVariable *AFLFinalLoc = + new GlobalVariable(M, Int32Tyi, true, GlobalValue::ExternalLinkage, 0, + "__afl_final_loc"); + ConstantInt *const_loc = ConstantInt::get(Int32Tyi, write_loc); + StoreInst * StoreFinalLoc = IRB.CreateStore(const_loc, AFLFinalLoc); + StoreFinalLoc->setMetadata(M.getMDKindID("nosanitize"), + MDNode::get(Ctx, None)); + + } + + if (dictionary.size()) { + + size_t memlen = 0, count = 0, offset = 0; + char * ptr; + + for (auto token : dictionary) { + + memlen += token.length(); + count++; + + } + + if (!be_quiet) + printf("AUTODICTIONARY: %lu string%s found\n", count, + count == 1 ? "" : "s"); + + if (count) { + + if ((ptr = (char *)malloc(memlen + count)) == NULL) { + + fprintf(stderr, "Error: malloc for %lu bytes failed!\n", + memlen + count); + exit(-1); + + } + + count = 0; + + for (auto token : dictionary) { + + if (offset + token.length() < 0xfffff0 && count < MAX_AUTO_EXTRAS) { + + ptr[offset++] = (uint8_t)token.length(); + memcpy(ptr + offset, token.c_str(), token.length()); + offset += token.length(); + count++; + + } + + } + + GlobalVariable *AFLDictionaryLen = + new GlobalVariable(M, Int32Tyi, false, GlobalValue::ExternalLinkage, + 0, "__afl_dictionary_len"); + ConstantInt *const_len = ConstantInt::get(Int32Tyi, offset); + StoreInst *StoreDictLen = IRB.CreateStore(const_len, AFLDictionaryLen); + StoreDictLen->setMetadata(M.getMDKindID("nosanitize"), + MDNode::get(Ctx, None)); + + ArrayType *ArrayTy = ArrayType::get(IntegerType::get(Ctx, 8), offset); + GlobalVariable *AFLInternalDictionary = new GlobalVariable( + M, ArrayTy, true, GlobalValue::ExternalLinkage, + ConstantDataArray::get(Ctx, + *(new ArrayRef((char *)ptr, offset))), + "__afl_internal_dictionary"); + AFLInternalDictionary->setInitializer(ConstantDataArray::get( + Ctx, *(new ArrayRef((char *)ptr, offset)))); + AFLInternalDictionary->setConstant(true); + + GlobalVariable *AFLDictionary = new GlobalVariable( + M, PointerType::get(Int8Tyi, 0), false, + GlobalValue::ExternalLinkage, 0, "__afl_dictionary"); + + Value *AFLDictOff = IRB.CreateGEP(AFLInternalDictionary, Zero); + Value *AFLDictPtr = + IRB.CreatePointerCast(AFLDictOff, PointerType::get(Int8Tyi, 0)); + StoreInst *StoreDict = IRB.CreateStore(AFLDictPtr, AFLDictionary); + StoreDict->setMetadata(M.getMDKindID("nosanitize"), + MDNode::get(Ctx, None)); + + } + + } + + } + + /* Say something nice. */ + + if (!be_quiet) { + + if (!inst) + WARNF("No instrumentation targets found."); + else { + + char modeline[100]; + snprintf(modeline, sizeof(modeline), "%s%s%s%s%s", + getenv("AFL_HARDEN") ? "hardened" : "non-hardened", + getenv("AFL_USE_ASAN") ? ", ASAN" : "", + getenv("AFL_USE_MSAN") ? ", MSAN" : "", + getenv("AFL_USE_CFISAN") ? ", CFISAN" : "", + getenv("AFL_USE_UBSAN") ? ", UBSAN" : ""); + OKF("Instrumented %u locations with no collisions (on average %llu " + "collisions would be in afl-gcc/afl-clang-fast) (%s mode).", + inst, calculateCollisions(inst), modeline); + + } + + } + + // afl++ END + + // We don't reference these arrays directly in any of our runtime functions, + // so we need to prevent them from being dead stripped. + if (TargetTriple.isOSBinFormatMachO()) appendToUsed(M, GlobalsToAppendToUsed); + appendToCompilerUsed(M, GlobalsToAppendToCompilerUsed); + return true; + +} + +// True if block has successors and it dominates all of them. +static bool isFullDominator(const BasicBlock *BB, const DominatorTree *DT) { + + if (succ_begin(BB) == succ_end(BB)) return false; + + for (const BasicBlock *SUCC : make_range(succ_begin(BB), succ_end(BB))) { + + if (!DT->dominates(BB, SUCC)) return false; + + } + + return true; + +} + +// True if block has predecessors and it postdominates all of them. +static bool isFullPostDominator(const BasicBlock * BB, + const PostDominatorTree *PDT) { + + if (pred_begin(BB) == pred_end(BB)) return false; + + for (const BasicBlock *PRED : make_range(pred_begin(BB), pred_end(BB))) { + + if (!PDT->dominates(BB, PRED)) return false; + + } + + return true; + +} + +static bool shouldInstrumentBlock(const Function &F, const BasicBlock *BB, + const DominatorTree * DT, + const PostDominatorTree * PDT, + const SanitizerCoverageOptions &Options) { + + // Don't insert coverage for blocks containing nothing but unreachable: we + // will never call __sanitizer_cov() for them, so counting them in + // NumberOfInstrumentedBlocks() might complicate calculation of code coverage + // percentage. Also, unreachable instructions frequently have no debug + // locations. + if (isa(BB->getFirstNonPHIOrDbgOrLifetime())) return false; + + // Don't insert coverage into blocks without a valid insertion point + // (catchswitch blocks). + if (BB->getFirstInsertionPt() == BB->end()) return false; + + // afl++ START + if (!Options.NoPrune && &F.getEntryBlock() == BB && F.size() > 1) + return false; + // afl++ END + + if (Options.NoPrune || &F.getEntryBlock() == BB) return true; + + if (Options.CoverageType == SanitizerCoverageOptions::SCK_Function && + &F.getEntryBlock() != BB) + return false; + + // Do not instrument full dominators, or full post-dominators with multiple + // predecessors. + return !isFullDominator(BB, DT) && + !(isFullPostDominator(BB, PDT) && !BB->getSinglePredecessor()); + +} + +void ModuleSanitizerCoverage::instrumentFunction( + Function &F, DomTreeCallback DTCallback, PostDomTreeCallback PDTCallback) { + + if (F.empty()) return; + if (F.getName().find(".module_ctor") != std::string::npos) + return; // Should not instrument sanitizer init functions. + if (F.getName().startswith("__sanitizer_")) + return; // Don't instrument __sanitizer_* callbacks. + // Don't touch available_externally functions, their actual body is elewhere. + if (F.getLinkage() == GlobalValue::AvailableExternallyLinkage) return; + // Don't instrument MSVC CRT configuration helpers. They may run before normal + // initialization. + if (F.getName() == "__local_stdio_printf_options" || + F.getName() == "__local_stdio_scanf_options") + return; + if (isa(F.getEntryBlock().getTerminator())) return; + // Don't instrument functions using SEH for now. Splitting basic blocks like + // we do for coverage breaks WinEHPrepare. + // FIXME: Remove this when SEH no longer uses landingpad pattern matching. + if (F.hasPersonalityFn() && + isAsynchronousEHPersonality(classifyEHPersonality(F.getPersonalityFn()))) + return; + // if (Allowlist && !Allowlist->inSection("coverage", "fun", F.getName())) + // return; + // if (Blocklist && Blocklist->inSection("coverage", "fun", F.getName())) + // return; + + // afl++ START + if (!F.size()) return; + if (isIgnoreFunction(&F)) return; + // afl++ END + + if (Options.CoverageType >= SanitizerCoverageOptions::SCK_Edge) + SplitAllCriticalEdges( + F, CriticalEdgeSplittingOptions().setIgnoreUnreachableDests()); + SmallVector IndirCalls; + SmallVector BlocksToInstrument; + + const DominatorTree * DT = DTCallback(F); + const PostDominatorTree *PDT = PDTCallback(F); + bool IsLeafFunc = true; + + for (auto &BB : F) { + + if (shouldInstrumentBlock(F, &BB, DT, PDT, Options)) + BlocksToInstrument.push_back(&BB); + for (auto &Inst : BB) { + + if (Options.IndirectCalls) { + + CallBase *CB = dyn_cast(&Inst); + if (CB && !CB->getCalledFunction()) IndirCalls.push_back(&Inst); + + } + + } + + } + + InjectCoverage(F, BlocksToInstrument, IsLeafFunc); + InjectCoverageForIndirectCalls(F, IndirCalls); + +} + +GlobalVariable *ModuleSanitizerCoverage::CreateFunctionLocalArrayInSection( + size_t NumElements, Function &F, Type *Ty, const char *Section) { + + ArrayType *ArrayTy = ArrayType::get(Ty, NumElements); + auto Array = new GlobalVariable( + *CurModule, ArrayTy, false, GlobalVariable::PrivateLinkage, + Constant::getNullValue(ArrayTy), "__sancov_gen_"); + + if (TargetTriple.supportsCOMDAT() && !F.isInterposable()) + if (auto Comdat = + GetOrCreateFunctionComdat(F, TargetTriple, CurModuleUniqueId)) + Array->setComdat(Comdat); + Array->setSection(getSectionName(Section)); + Array->setAlignment(Align(DL->getTypeStoreSize(Ty).getFixedSize())); + GlobalsToAppendToUsed.push_back(Array); + GlobalsToAppendToCompilerUsed.push_back(Array); + MDNode *MD = MDNode::get(F.getContext(), ValueAsMetadata::get(&F)); + Array->addMetadata(LLVMContext::MD_associated, *MD); + + return Array; + +} + +GlobalVariable *ModuleSanitizerCoverage::CreatePCArray( + Function &F, ArrayRef AllBlocks) { + + size_t N = AllBlocks.size(); + assert(N); + SmallVector PCs; + IRBuilder<> IRB(&*F.getEntryBlock().getFirstInsertionPt()); + for (size_t i = 0; i < N; i++) { + + if (&F.getEntryBlock() == AllBlocks[i]) { + + PCs.push_back((Constant *)IRB.CreatePointerCast(&F, IntptrPtrTy)); + PCs.push_back((Constant *)IRB.CreateIntToPtr( + ConstantInt::get(IntptrTy, 1), IntptrPtrTy)); + + } else { + + PCs.push_back((Constant *)IRB.CreatePointerCast( + BlockAddress::get(AllBlocks[i]), IntptrPtrTy)); + PCs.push_back((Constant *)IRB.CreateIntToPtr( + ConstantInt::get(IntptrTy, 0), IntptrPtrTy)); + + } + + } + + auto *PCArray = CreateFunctionLocalArrayInSection(N * 2, F, IntptrPtrTy, + SanCovPCsSectionName); + PCArray->setInitializer( + ConstantArray::get(ArrayType::get(IntptrPtrTy, N * 2), PCs)); + PCArray->setConstant(true); + + return PCArray; + +} + +void ModuleSanitizerCoverage::CreateFunctionLocalArrays( + Function &F, ArrayRef AllBlocks) { + + if (Options.TracePCGuard) + FunctionGuardArray = CreateFunctionLocalArrayInSection( + AllBlocks.size(), F, Int32Ty, SanCovGuardsSectionName); + if (Options.Inline8bitCounters) + Function8bitCounterArray = CreateFunctionLocalArrayInSection( + AllBlocks.size(), F, Int8Ty, SanCovCountersSectionName); + if (Options.InlineBoolFlag) + FunctionBoolArray = CreateFunctionLocalArrayInSection( + AllBlocks.size(), F, Int1Ty, SanCovBoolFlagSectionName); + if (Options.PCTable) FunctionPCsArray = CreatePCArray(F, AllBlocks); + +} + +bool ModuleSanitizerCoverage::InjectCoverage(Function & F, + ArrayRef AllBlocks, + bool IsLeafFunc) { + + if (AllBlocks.empty()) return false; + CreateFunctionLocalArrays(F, AllBlocks); + for (size_t i = 0, N = AllBlocks.size(); i < N; i++) { + + // afl++ START + if (BlockList.size()) { + + int skip = 0; + for (uint32_t k = 0; k < BlockList.size(); k++) { + + if (AllBlocks[i] == BlockList[k]) { + + if (debug) + fprintf(stderr, + "DEBUG: Function %s skipping BB with/after __afl_loop\n", + F.getName().str().c_str()); + skip = 1; + + } + + } + + if (skip) continue; + + } + + // afl++ END + + InjectCoverageAtBlock(F, *AllBlocks[i], i, IsLeafFunc); + + } + + return true; + +} + +// On every indirect call we call a run-time function +// __sanitizer_cov_indir_call* with two parameters: +// - callee address, +// - global cache array that contains CacheSize pointers (zero-initialized). +// The cache is used to speed up recording the caller-callee pairs. +// The address of the caller is passed implicitly via caller PC. +// CacheSize is encoded in the name of the run-time function. +void ModuleSanitizerCoverage::InjectCoverageForIndirectCalls( + Function &F, ArrayRef IndirCalls) { + + if (IndirCalls.empty()) return; + assert(Options.TracePC || Options.TracePCGuard || + Options.Inline8bitCounters || Options.InlineBoolFlag); + for (auto I : IndirCalls) { + + IRBuilder<> IRB(I); + CallBase & CB = cast(*I); + Value * Callee = CB.getCalledOperand(); + if (isa(Callee)) continue; + IRB.CreateCall(SanCovTracePCIndir, IRB.CreatePointerCast(Callee, IntptrTy)); + + } + +} + +void ModuleSanitizerCoverage::InjectCoverageAtBlock(Function &F, BasicBlock &BB, + size_t Idx, + bool IsLeafFunc) { + + BasicBlock::iterator IP = BB.getFirstInsertionPt(); + bool IsEntryBB = &BB == &F.getEntryBlock(); + DebugLoc EntryLoc; + if (IsEntryBB) { + + if (auto SP = F.getSubprogram()) + EntryLoc = DebugLoc::get(SP->getScopeLine(), 0, SP); + // Keep static allocas and llvm.localescape calls in the entry block. Even + // if we aren't splitting the block, it's nice for allocas to be before + // calls. + IP = PrepareToSplitEntryBlock(BB, IP); + + } else { + + EntryLoc = IP->getDebugLoc(); + + } + + IRBuilder<> IRB(&*IP); + IRB.SetCurrentDebugLocation(EntryLoc); + if (Options.TracePC) { + + IRB.CreateCall(SanCovTracePC) +#if LLVM_VERSION_MAJOR < 12 + ->cannotMerge(); // gets the PC using GET_CALLER_PC. +#else + ->setCannotMerge(); // gets the PC using GET_CALLER_PC. +#endif + + } + + if (Options.TracePCGuard) { + + // afl++ START + ++afl_global_id; + + if (documentFile) { + + unsigned long long int moduleID = + (((unsigned long long int)(rand() & 0xffffffff)) << 32) | getpid(); + fprintf(documentFile, "ModuleID=%llu Function=%s edgeID=%u\n", moduleID, + F.getName().str().c_str(), afl_global_id); + + } + + /* Set the ID of the inserted basic block */ + + ConstantInt *CurLoc = ConstantInt::get(Int32Tyi, afl_global_id); + + /* Load SHM pointer */ + + Value *MapPtrIdx; + + if (map_addr) { + + MapPtrIdx = IRB.CreateGEP(MapPtrFixed, CurLoc); + + } else { + + LoadInst *MapPtr = IRB.CreateLoad(AFLMapPtr); + MapPtr->setMetadata(Mo->getMDKindID("nosanitize"), + MDNode::get(*Ct, None)); + MapPtrIdx = IRB.CreateGEP(MapPtr, CurLoc); + + } + + /* Update bitmap */ + + LoadInst *Counter = IRB.CreateLoad(MapPtrIdx); + Counter->setMetadata(Mo->getMDKindID("nosanitize"), MDNode::get(*Ct, None)); + + Value *Incr = IRB.CreateAdd(Counter, One); + + if (skip_nozero == NULL) { + + auto cf = IRB.CreateICmpEQ(Incr, Zero); + auto carry = IRB.CreateZExt(cf, Int8Tyi); + Incr = IRB.CreateAdd(Incr, carry); + + } + + IRB.CreateStore(Incr, MapPtrIdx) + ->setMetadata(Mo->getMDKindID("nosanitize"), MDNode::get(*Ct, None)); + + // done :) + + inst++; + // afl++ END + + /* + XXXXXXXXXXXXXXXXXXX + + auto GuardPtr = IRB.CreateIntToPtr( + IRB.CreateAdd(IRB.CreatePointerCast(FunctionGuardArray, IntptrTy), + ConstantInt::get(IntptrTy, Idx * 4)), + Int32PtrTy); + + IRB.CreateCall(SanCovTracePCGuard, GuardPtr)->setCannotMerge(); + */ + + } + + if (Options.Inline8bitCounters) { + + auto CounterPtr = IRB.CreateGEP( + Function8bitCounterArray->getValueType(), Function8bitCounterArray, + {ConstantInt::get(IntptrTy, 0), ConstantInt::get(IntptrTy, Idx)}); + auto Load = IRB.CreateLoad(Int8Ty, CounterPtr); + auto Inc = IRB.CreateAdd(Load, ConstantInt::get(Int8Ty, 1)); + auto Store = IRB.CreateStore(Inc, CounterPtr); + SetNoSanitizeMetadata(Load); + SetNoSanitizeMetadata(Store); + + } + + if (Options.InlineBoolFlag) { + + auto FlagPtr = IRB.CreateGEP( + FunctionBoolArray->getValueType(), FunctionBoolArray, + {ConstantInt::get(IntptrTy, 0), ConstantInt::get(IntptrTy, Idx)}); + auto Load = IRB.CreateLoad(Int1Ty, FlagPtr); + auto ThenTerm = + SplitBlockAndInsertIfThen(IRB.CreateIsNull(Load), &*IP, false); + IRBuilder<> ThenIRB(ThenTerm); + auto Store = ThenIRB.CreateStore(ConstantInt::getTrue(Int1Ty), FlagPtr); + SetNoSanitizeMetadata(Load); + SetNoSanitizeMetadata(Store); + + } + +} + +std::string ModuleSanitizerCoverage::getSectionName( + const std::string &Section) const { + + if (TargetTriple.isOSBinFormatCOFF()) { + + if (Section == SanCovCountersSectionName) return ".SCOV$CM"; + if (Section == SanCovBoolFlagSectionName) return ".SCOV$BM"; + if (Section == SanCovPCsSectionName) return ".SCOVP$M"; + return ".SCOV$GM"; // For SanCovGuardsSectionName. + + } + + if (TargetTriple.isOSBinFormatMachO()) return "__DATA,__" + Section; + return "__" + Section; + +} + +std::string ModuleSanitizerCoverage::getSectionStart( + const std::string &Section) const { + + if (TargetTriple.isOSBinFormatMachO()) + return "\1section$start$__DATA$__" + Section; + return "__start___" + Section; + +} + +std::string ModuleSanitizerCoverage::getSectionEnd( + const std::string &Section) const { + + if (TargetTriple.isOSBinFormatMachO()) + return "\1section$end$__DATA$__" + Section; + return "__stop___" + Section; + +} + +char ModuleSanitizerCoverageLegacyPass::ID = 0; + +INITIALIZE_PASS_BEGIN(ModuleSanitizerCoverageLegacyPass, "sancov", + "Pass for instrumenting coverage on functions", false, + false) +INITIALIZE_PASS_DEPENDENCY(DominatorTreeWrapperPass) +INITIALIZE_PASS_DEPENDENCY(PostDominatorTreeWrapperPass) +INITIALIZE_PASS_END(ModuleSanitizerCoverageLegacyPass, "sancov", + "Pass for instrumenting coverage on functions", false, + false) + +ModulePass *llvm::createModuleSanitizerCoverageLegacyPassPass( + const SanitizerCoverageOptions &Options, + const std::vector &AllowlistFiles, + const std::vector &BlocklistFiles) { + + return new ModuleSanitizerCoverageLegacyPass(Options); + //, AllowlistFiles, BlocklistFiles); + +} + +static void registerLTOPass(const PassManagerBuilder &, + legacy::PassManagerBase &PM) { + + auto p = new ModuleSanitizerCoverageLegacyPass(); + PM.add(p); + +} + +static RegisterStandardPasses RegisterCompTransPass( + PassManagerBuilder::EP_OptimizerLast, registerLTOPass); + +static RegisterStandardPasses RegisterCompTransPass0( + PassManagerBuilder::EP_EnabledOnOptLevel0, registerLTOPass); + +#if LLVM_VERSION_MAJOR >= 11 +static RegisterStandardPasses RegisterCompTransPassLTO( + PassManagerBuilder::EP_FullLinkTimeOptimizationLast, registerLTOPass); +#endif + diff --git a/instrumentation/afl-compiler-rt.o.c b/instrumentation/afl-compiler-rt.o.c new file mode 100644 index 00000000..a3d75b15 --- /dev/null +++ b/instrumentation/afl-compiler-rt.o.c @@ -0,0 +1,1254 @@ +/* + american fuzzy lop++ - instrumentation bootstrap + ------------------------------------------------ + + Copyright 2015, 2016 Google Inc. All rights reserved. + Copyright 2019-2020 AFLplusplus Project. All rights reserved. + + Licensed under the Apache License, Version 2.0 (the "License"); + you may not use this file except in compliance with the License. + You may obtain a copy of the License at: + + http://www.apache.org/licenses/LICENSE-2.0 + + +*/ + +#ifdef __ANDROID__ + #include "android-ashmem.h" +#endif +#include "config.h" +#include "types.h" +#include "cmplog.h" +#include "llvm-ngram-coverage.h" + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include +#include +#include +#include + +#include "llvm/Config/llvm-config.h" + +#ifdef __linux__ + #include "snapshot-inl.h" +#endif + +/* This is a somewhat ugly hack for the experimental 'trace-pc-guard' mode. + Basically, we need to make sure that the forkserver is initialized after + the LLVM-generated runtime initialization pass, not before. */ + +#ifndef MAP_FIXED_NOREPLACE + #ifdef MAP_EXCL + #define MAP_FIXED_NOREPLACE MAP_EXCL | MAP_FIXED + #else + #define MAP_FIXED_NOREPLACE MAP_FIXED + #endif +#endif + +#define CTOR_PRIO 3 + +#include +#include + +/* Globals needed by the injected instrumentation. The __afl_area_initial region + is used for instrumentation output before __afl_map_shm() has a chance to + run. It will end up as .comm, so it shouldn't be too wasteful. */ + +#if MAP_SIZE <= 65536 + #define MAP_INITIAL_SIZE 256000 +#else + #define MAP_INITIAL_SIZE MAP_SIZE +#endif + +u8 __afl_area_initial[MAP_INITIAL_SIZE]; +u8 * __afl_area_ptr = __afl_area_initial; +u8 * __afl_dictionary; +u8 * __afl_fuzz_ptr; +u32 __afl_fuzz_len_dummy; +u32 *__afl_fuzz_len = &__afl_fuzz_len_dummy; + +u32 __afl_final_loc; +u32 __afl_map_size = MAP_SIZE; +u32 __afl_dictionary_len; +u64 __afl_map_addr; + +#ifdef __ANDROID__ +PREV_LOC_T __afl_prev_loc[NGRAM_SIZE_MAX]; +u32 __afl_prev_ctx; +u32 __afl_cmp_counter; +#else +__thread PREV_LOC_T __afl_prev_loc[NGRAM_SIZE_MAX]; +__thread u32 __afl_prev_ctx; +__thread u32 __afl_cmp_counter; +#endif + +int __afl_sharedmem_fuzzing __attribute__((weak)); + +struct cmp_map *__afl_cmp_map; + +/* Running in persistent mode? */ + +static u8 is_persistent; + +/* Are we in sancov mode? */ + +static u8 _is_sancov; + +/* Uninspired gcc plugin instrumentation */ + +void __afl_trace(const u32 x) { + +#if 1 /* enable for neverZero feature. */ + __afl_area_ptr[__afl_prev_loc[0] ^ x] += + 1 + ((u8)(1 + __afl_area_ptr[__afl_prev_loc[0] ^ x]) == 0); +#else + ++__afl_area_ptr[__afl_prev_loc[0] ^ x]; +#endif + + __afl_prev_loc[0] = (x >> 1); + return; + +} + +/* Error reporting to forkserver controller */ + +void send_forkserver_error(int error) { + + u32 status; + if (!error || error > 0xffff) return; + status = (FS_OPT_ERROR | FS_OPT_SET_ERROR(error)); + if (write(FORKSRV_FD + 1, (char *)&status, 4) != 4) return; + +} + +/* SHM fuzzing setup. */ + +static void __afl_map_shm_fuzz() { + + char *id_str = getenv(SHM_FUZZ_ENV_VAR); + + if (id_str) { + + u8 *map = NULL; + +#ifdef USEMMAP + const char * shm_file_path = id_str; + int shm_fd = -1; + unsigned char *shm_base = NULL; + + /* create the shared memory segment as if it was a file */ + shm_fd = shm_open(shm_file_path, O_RDWR, 0600); + if (shm_fd == -1) { + + fprintf(stderr, "shm_open() failed for fuzz\n"); + send_forkserver_error(FS_ERROR_SHM_OPEN); + exit(1); + + } + + map = + (u8 *)mmap(0, MAX_FILE + sizeof(u32), PROT_READ, MAP_SHARED, shm_fd, 0); + +#else + u32 shm_id = atoi(id_str); + map = (u8 *)shmat(shm_id, NULL, 0); + +#endif + + /* Whooooops. */ + + if (!map || map == (void *)-1) { + + perror("Could not access fuzzign shared memory"); + exit(1); + + } + + __afl_fuzz_len = (u32 *)map; + __afl_fuzz_ptr = map + sizeof(u32); + + if (getenv("AFL_DEBUG")) { + + fprintf(stderr, "DEBUG: successfully got fuzzing shared memory\n"); + + } + + } else { + + fprintf(stderr, "Error: variable for fuzzing shared memory is not set\n"); + exit(1); + + } + +} + +/* SHM setup. */ + +static void __afl_map_shm(void) { + + // we we are not running in afl ensure the map exists + if (!__afl_area_ptr) { __afl_area_ptr = __afl_area_initial; } + + char *id_str = getenv(SHM_ENV_VAR); + + if (__afl_final_loc) { + + if (__afl_final_loc % 8) + __afl_final_loc = (((__afl_final_loc + 7) >> 3) << 3); + __afl_map_size = __afl_final_loc; + + if (__afl_final_loc > MAP_SIZE) { + + char *ptr; + u32 val = 0; + if ((ptr = getenv("AFL_MAP_SIZE")) != NULL) val = atoi(ptr); + if (val < __afl_final_loc) { + + if (__afl_final_loc > FS_OPT_MAX_MAPSIZE) { + + if (!getenv("AFL_QUIET")) + fprintf(stderr, + "Error: AFL++ tools *require* to set AFL_MAP_SIZE to %u " + "to be able to run this instrumented program!\n", + __afl_final_loc); + + if (id_str) { + + send_forkserver_error(FS_ERROR_MAP_SIZE); + exit(-1); + + } + + } else { + + if (!getenv("AFL_QUIET")) + fprintf(stderr, + "Warning: AFL++ tools will need to set AFL_MAP_SIZE to %u " + "to be able to run this instrumented program!\n", + __afl_final_loc); + + } + + } + + } + + } + + /* If we're running under AFL, attach to the appropriate region, replacing the + early-stage __afl_area_initial region that is needed to allow some really + hacky .init code to work correctly in projects such as OpenSSL. */ + + if (getenv("AFL_DEBUG")) + fprintf(stderr, + "DEBUG: id_str %s, __afl_area_ptr %p, __afl_area_initial %p, " + "__afl_map_addr 0x%llx, MAP_SIZE %u, __afl_final_loc %u, " + "max_size_forkserver %u/0x%x\n", + id_str == NULL ? "" : id_str, __afl_area_ptr, + __afl_area_initial, __afl_map_addr, MAP_SIZE, __afl_final_loc, + FS_OPT_MAX_MAPSIZE, FS_OPT_MAX_MAPSIZE); + + if (id_str) { + + if (__afl_area_ptr && __afl_area_ptr != __afl_area_initial) { + + if (__afl_map_addr) + munmap((void *)__afl_map_addr, __afl_final_loc); + else + free(__afl_area_ptr); + __afl_area_ptr = __afl_area_initial; + + } + +#ifdef USEMMAP + const char * shm_file_path = id_str; + int shm_fd = -1; + unsigned char *shm_base = NULL; + + /* create the shared memory segment as if it was a file */ + shm_fd = shm_open(shm_file_path, O_RDWR, 0600); + if (shm_fd == -1) { + + fprintf(stderr, "shm_open() failed\n"); + send_forkserver_error(FS_ERROR_SHM_OPEN); + exit(1); + + } + + /* map the shared memory segment to the address space of the process */ + if (__afl_map_addr) { + + shm_base = + mmap((void *)__afl_map_addr, __afl_map_size, PROT_READ | PROT_WRITE, + MAP_FIXED_NOREPLACE | MAP_SHARED, shm_fd, 0); + + } else { + + shm_base = mmap(0, __afl_map_size, PROT_READ | PROT_WRITE, MAP_SHARED, + shm_fd, 0); + + } + + if (shm_base == MAP_FAILED) { + + close(shm_fd); + shm_fd = -1; + + fprintf(stderr, "mmap() failed\n"); + if (__afl_map_addr) + send_forkserver_error(FS_ERROR_MAP_ADDR); + else + send_forkserver_error(FS_ERROR_MMAP); + exit(2); + + } + + __afl_area_ptr = shm_base; +#else + u32 shm_id = atoi(id_str); + + __afl_area_ptr = shmat(shm_id, (void *)__afl_map_addr, 0); + +#endif + + /* Whooooops. */ + + if (__afl_area_ptr == (void *)-1) { + + if (__afl_map_addr) + send_forkserver_error(FS_ERROR_MAP_ADDR); + else + send_forkserver_error(FS_ERROR_SHMAT); + _exit(1); + + } + + /* Write something into the bitmap so that even with low AFL_INST_RATIO, + our parent doesn't give up on us. */ + + __afl_area_ptr[0] = 1; + + } else if ((!__afl_area_ptr || __afl_area_ptr == __afl_area_initial) && + + __afl_map_addr) { + + __afl_area_ptr = + mmap((void *)__afl_map_addr, __afl_map_size, PROT_READ | PROT_WRITE, + MAP_FIXED_NOREPLACE | MAP_SHARED | MAP_ANONYMOUS, -1, 0); + + if (__afl_area_ptr == MAP_FAILED) { + + fprintf(stderr, "can not acquire mmap for address %p\n", + (void *)__afl_map_addr); + exit(1); + + } + + } else if (_is_sancov && __afl_area_ptr != __afl_area_initial) { + + free(__afl_area_ptr); + __afl_area_ptr = NULL; + if (__afl_final_loc > MAP_INITIAL_SIZE) + __afl_area_ptr = malloc(__afl_final_loc); + if (!__afl_area_ptr) __afl_area_ptr = __afl_area_initial; + + } + + id_str = getenv(CMPLOG_SHM_ENV_VAR); + + if (getenv("AFL_DEBUG")) { + + fprintf(stderr, "DEBUG: cmplog id_str %s\n", + id_str == NULL ? "" : id_str); + + } + + if (id_str) { + +#ifdef USEMMAP + const char * shm_file_path = id_str; + int shm_fd = -1; + unsigned char *shm_base = NULL; + + /* create the shared memory segment as if it was a file */ + shm_fd = shm_open(shm_file_path, O_RDWR, 0600); + if (shm_fd == -1) { + + fprintf(stderr, "shm_open() failed\n"); + exit(1); + + } + + /* map the shared memory segment to the address space of the process */ + shm_base = mmap(0, sizeof(struct cmp_map), PROT_READ | PROT_WRITE, + MAP_SHARED, shm_fd, 0); + if (shm_base == MAP_FAILED) { + + close(shm_fd); + shm_fd = -1; + + fprintf(stderr, "mmap() failed\n"); + exit(2); + + } + + __afl_cmp_map = shm_base; +#else + u32 shm_id = atoi(id_str); + + __afl_cmp_map = shmat(shm_id, NULL, 0); +#endif + + if (__afl_cmp_map == (void *)-1) _exit(1); + + } + +} + +#ifdef __linux__ +static void __afl_start_snapshots(void) { + + static u8 tmp[4] = {0, 0, 0, 0}; + s32 child_pid; + u32 status = 0; + u32 already_read_first = 0; + u32 was_killed; + + u8 child_stopped = 0; + + void (*old_sigchld_handler)(int) = 0; // = signal(SIGCHLD, SIG_DFL); + + /* Phone home and tell the parent that we're OK. If parent isn't there, + assume we're not running in forkserver mode and just execute program. */ + + status |= (FS_OPT_ENABLED | FS_OPT_SNAPSHOT); + if (__afl_sharedmem_fuzzing != 0) status |= FS_OPT_SHDMEM_FUZZ; + if (__afl_map_size <= FS_OPT_MAX_MAPSIZE) + status |= (FS_OPT_SET_MAPSIZE(__afl_map_size) | FS_OPT_MAPSIZE); + if (__afl_dictionary_len && __afl_dictionary) status |= FS_OPT_AUTODICT; + memcpy(tmp, &status, 4); + + if (write(FORKSRV_FD + 1, tmp, 4) != 4) return; + + if (__afl_sharedmem_fuzzing || (__afl_dictionary_len && __afl_dictionary)) { + + if (read(FORKSRV_FD, &was_killed, 4) != 4) _exit(1); + + if (getenv("AFL_DEBUG")) { + + fprintf(stderr, "target forkserver recv: %08x\n", was_killed); + + } + + if ((was_killed & (FS_OPT_ENABLED | FS_OPT_SHDMEM_FUZZ)) == + (FS_OPT_ENABLED | FS_OPT_SHDMEM_FUZZ)) { + + __afl_map_shm_fuzz(); + + } + + if ((was_killed & (FS_OPT_ENABLED | FS_OPT_AUTODICT)) == + (FS_OPT_ENABLED | FS_OPT_AUTODICT)) { + + // great lets pass the dictionary through the forkserver FD + u32 len = __afl_dictionary_len, offset = 0; + s32 ret; + + if (write(FORKSRV_FD + 1, &len, 4) != 4) { + + write(2, "Error: could not send dictionary len\n", + strlen("Error: could not send dictionary len\n")); + _exit(1); + + } + + while (len != 0) { + + ret = write(FORKSRV_FD + 1, __afl_dictionary + offset, len); + + if (ret < 1) { + + write(2, "Error: could not send dictionary\n", + strlen("Error: could not send dictionary\n")); + _exit(1); + + } + + len -= ret; + offset += ret; + + } + + } else { + + // uh this forkserver does not understand extended option passing + // or does not want the dictionary + if (!__afl_fuzz_ptr) already_read_first = 1; + + } + + } + + while (1) { + + int status; + + if (already_read_first) { + + already_read_first = 0; + + } else { + + /* Wait for parent by reading from the pipe. Abort if read fails. */ + if (read(FORKSRV_FD, &was_killed, 4) != 4) _exit(1); + + } + + #ifdef _AFL_DOCUMENT_MUTATIONS + if (__afl_fuzz_ptr) { + + static uint32_t counter = 0; + char fn[32]; + sprintf(fn, "%09u:forkserver", counter); + s32 fd_doc = open(fn, O_WRONLY | O_CREAT | O_TRUNC, 0600); + if (fd_doc >= 0) { + + if (write(fd_doc, __afl_fuzz_ptr, *__afl_fuzz_len) != *__afl_fuzz_len) { + + fprintf(stderr, "write of mutation file failed: %s\n", fn); + unlink(fn); + + } + + close(fd_doc); + + } + + counter++; + + } + + #endif + + /* If we stopped the child in persistent mode, but there was a race + condition and afl-fuzz already issued SIGKILL, write off the old + process. */ + + if (child_stopped && was_killed) { + + child_stopped = 0; + if (waitpid(child_pid, &status, 0) < 0) _exit(1); + + } + + if (!child_stopped) { + + /* Once woken up, create a clone of our process. */ + + child_pid = fork(); + if (child_pid < 0) _exit(1); + + /* In child process: close fds, resume execution. */ + + if (!child_pid) { + + //(void)nice(-20); // does not seem to improve + + signal(SIGCHLD, old_sigchld_handler); + + close(FORKSRV_FD); + close(FORKSRV_FD + 1); + + if (!afl_snapshot_take(AFL_SNAPSHOT_MMAP | AFL_SNAPSHOT_FDS | + AFL_SNAPSHOT_REGS | AFL_SNAPSHOT_EXIT)) { + + raise(SIGSTOP); + + } + + __afl_area_ptr[0] = 1; + memset(__afl_prev_loc, 0, NGRAM_SIZE_MAX * sizeof(PREV_LOC_T)); + + return; + + } + + } else { + + /* Special handling for persistent mode: if the child is alive but + currently stopped, simply restart it with SIGCONT. */ + + kill(child_pid, SIGCONT); + child_stopped = 0; + + } + + /* In parent process: write PID to pipe, then wait for child. */ + + if (write(FORKSRV_FD + 1, &child_pid, 4) != 4) _exit(1); + + if (waitpid(child_pid, &status, WUNTRACED) < 0) _exit(1); + + /* In persistent mode, the child stops itself with SIGSTOP to indicate + a successful run. In this case, we want to wake it up without forking + again. */ + + if (WIFSTOPPED(status)) child_stopped = 1; + + /* Relay wait status to pipe, then loop back. */ + + if (write(FORKSRV_FD + 1, &status, 4) != 4) _exit(1); + + } + +} + +#endif + +/* Fork server logic. */ + +static void __afl_start_forkserver(void) { + +#ifdef __linux__ + if (/*!is_persistent &&*/ !__afl_cmp_map && !getenv("AFL_NO_SNAPSHOT") && + afl_snapshot_init() >= 0) { + + __afl_start_snapshots(); + return; + + } + +#endif + + u8 tmp[4] = {0, 0, 0, 0}; + s32 child_pid; + u32 status = 0; + u32 already_read_first = 0; + u32 was_killed; + + u8 child_stopped = 0; + + void (*old_sigchld_handler)(int) = 0; // = signal(SIGCHLD, SIG_DFL); + + if (__afl_map_size <= FS_OPT_MAX_MAPSIZE) + status |= (FS_OPT_SET_MAPSIZE(__afl_map_size) | FS_OPT_MAPSIZE); + if (__afl_dictionary_len && __afl_dictionary) status |= FS_OPT_AUTODICT; + if (__afl_sharedmem_fuzzing != 0) status |= FS_OPT_SHDMEM_FUZZ; + if (status) status |= (FS_OPT_ENABLED); + memcpy(tmp, &status, 4); + + /* Phone home and tell the parent that we're OK. If parent isn't there, + assume we're not running in forkserver mode and just execute program. */ + + if (write(FORKSRV_FD + 1, tmp, 4) != 4) return; + + if (__afl_sharedmem_fuzzing || (__afl_dictionary_len && __afl_dictionary)) { + + if (read(FORKSRV_FD, &was_killed, 4) != 4) _exit(1); + + if (getenv("AFL_DEBUG")) { + + fprintf(stderr, "target forkserver recv: %08x\n", was_killed); + + } + + if ((was_killed & (FS_OPT_ENABLED | FS_OPT_SHDMEM_FUZZ)) == + (FS_OPT_ENABLED | FS_OPT_SHDMEM_FUZZ)) { + + __afl_map_shm_fuzz(); + + } + + if ((was_killed & (FS_OPT_ENABLED | FS_OPT_AUTODICT)) == + (FS_OPT_ENABLED | FS_OPT_AUTODICT)) { + + // great lets pass the dictionary through the forkserver FD + u32 len = __afl_dictionary_len, offset = 0; + s32 ret; + + if (write(FORKSRV_FD + 1, &len, 4) != 4) { + + write(2, "Error: could not send dictionary len\n", + strlen("Error: could not send dictionary len\n")); + _exit(1); + + } + + while (len != 0) { + + ret = write(FORKSRV_FD + 1, __afl_dictionary + offset, len); + + if (ret < 1) { + + write(2, "Error: could not send dictionary\n", + strlen("Error: could not send dictionary\n")); + _exit(1); + + } + + len -= ret; + offset += ret; + + } + + } else { + + // uh this forkserver does not understand extended option passing + // or does not want the dictionary + if (!__afl_fuzz_ptr) already_read_first = 1; + + } + + } + + while (1) { + + int status; + + /* Wait for parent by reading from the pipe. Abort if read fails. */ + + if (already_read_first) { + + already_read_first = 0; + + } else { + + if (read(FORKSRV_FD, &was_killed, 4) != 4) _exit(1); + + } + +#ifdef _AFL_DOCUMENT_MUTATIONS + if (__afl_fuzz_ptr) { + + static uint32_t counter = 0; + char fn[32]; + sprintf(fn, "%09u:forkserver", counter); + s32 fd_doc = open(fn, O_WRONLY | O_CREAT | O_TRUNC, 0600); + if (fd_doc >= 0) { + + if (write(fd_doc, __afl_fuzz_ptr, *__afl_fuzz_len) != *__afl_fuzz_len) { + + fprintf(stderr, "write of mutation file failed: %s\n", fn); + unlink(fn); + + } + + close(fd_doc); + + } + + counter++; + + } + +#endif + + /* If we stopped the child in persistent mode, but there was a race + condition and afl-fuzz already issued SIGKILL, write off the old + process. */ + + if (child_stopped && was_killed) { + + child_stopped = 0; + if (waitpid(child_pid, &status, 0) < 0) _exit(1); + + } + + if (!child_stopped) { + + /* Once woken up, create a clone of our process. */ + + child_pid = fork(); + if (child_pid < 0) _exit(1); + + /* In child process: close fds, resume execution. */ + + if (!child_pid) { + + //(void)nice(-20); + + signal(SIGCHLD, old_sigchld_handler); + + close(FORKSRV_FD); + close(FORKSRV_FD + 1); + return; + + } + + } else { + + /* Special handling for persistent mode: if the child is alive but + currently stopped, simply restart it with SIGCONT. */ + + kill(child_pid, SIGCONT); + child_stopped = 0; + + } + + /* In parent process: write PID to pipe, then wait for child. */ + + if (write(FORKSRV_FD + 1, &child_pid, 4) != 4) _exit(1); + + if (waitpid(child_pid, &status, is_persistent ? WUNTRACED : 0) < 0) + _exit(1); + + /* In persistent mode, the child stops itself with SIGSTOP to indicate + a successful run. In this case, we want to wake it up without forking + again. */ + + if (WIFSTOPPED(status)) child_stopped = 1; + + /* Relay wait status to pipe, then loop back. */ + + if (write(FORKSRV_FD + 1, &status, 4) != 4) _exit(1); + + } + +} + +/* A simplified persistent mode handler, used as explained in + * README.llvm.md. */ + +int __afl_persistent_loop(unsigned int max_cnt) { + + static u8 first_pass = 1; + static u32 cycle_cnt; + + if (first_pass) { + + /* Make sure that every iteration of __AFL_LOOP() starts with a clean slate. + On subsequent calls, the parent will take care of that, but on the first + iteration, it's our job to erase any trace of whatever happened + before the loop. */ + + if (is_persistent) { + + memset(__afl_area_ptr, 0, __afl_map_size); + __afl_area_ptr[0] = 1; + memset(__afl_prev_loc, 0, NGRAM_SIZE_MAX * sizeof(PREV_LOC_T)); + + } + + cycle_cnt = max_cnt; + first_pass = 0; + return 1; + + } + + if (is_persistent) { + + if (--cycle_cnt) { + + raise(SIGSTOP); + + __afl_area_ptr[0] = 1; + memset(__afl_prev_loc, 0, NGRAM_SIZE_MAX * sizeof(PREV_LOC_T)); + + return 1; + + } else { + + /* When exiting __AFL_LOOP(), make sure that the subsequent code that + follows the loop is not traced. We do that by pivoting back to the + dummy output region. */ + + __afl_area_ptr = __afl_area_initial; + + } + + } + + return 0; + +} + +/* This one can be called from user code when deferred forkserver mode + is enabled. */ + +void __afl_manual_init(void) { + + static u8 init_done; + + if (getenv("AFL_DISABLE_LLVM_INSTRUMENTATION")) { + + init_done = 1; + is_persistent = 0; + __afl_sharedmem_fuzzing = 0; + if (__afl_area_ptr == NULL) __afl_area_ptr = __afl_area_initial; + + if (getenv("AFL_DEBUG")) + fprintf(stderr, + "DEBUG: disabled instrumentation because of " + "AFL_DISABLE_LLVM_INSTRUMENTATION\n"); + + } + + if (!init_done) { + + __afl_start_forkserver(); + init_done = 1; + + } + +} + +/* Initialization of the forkserver - latest possible */ + +__attribute__((constructor())) void __afl_auto_init(void) { + + if (getenv("AFL_DISABLE_LLVM_INSTRUMENTATION")) return; + + if (getenv(DEFER_ENV_VAR)) return; + + __afl_manual_init(); + +} + +/* Initialization of the shmem - earliest possible because of LTO fixed mem. */ + +__attribute__((constructor(CTOR_PRIO))) void __afl_auto_early(void) { + + is_persistent = !!getenv(PERSIST_ENV_VAR); + + if (getenv("AFL_DISABLE_LLVM_INSTRUMENTATION")) return; + + __afl_map_shm(); + +} + +/* preset __afl_area_ptr #2 */ + +__attribute__((constructor(1))) void __afl_auto_second(void) { + + if (getenv("AFL_DISABLE_LLVM_INSTRUMENTATION")) return; + u8 *ptr; + + if (__afl_final_loc) { + + if (__afl_area_ptr && __afl_area_ptr != __afl_area_initial) + free(__afl_area_ptr); + + if (__afl_map_addr) + ptr = (u8 *)mmap((void *)__afl_map_addr, __afl_final_loc, + PROT_READ | PROT_WRITE, + MAP_FIXED_NOREPLACE | MAP_SHARED | MAP_ANONYMOUS, -1, 0); + else + ptr = (u8 *)malloc(__afl_final_loc); + + if (ptr && (ssize_t)ptr != -1) __afl_area_ptr = ptr; + + } + +} + +/* preset __afl_area_ptr #1 - at constructor level 0 global variables have + not been set */ + +__attribute__((constructor(0))) void __afl_auto_first(void) { + + if (getenv("AFL_DISABLE_LLVM_INSTRUMENTATION")) return; + u8 *ptr; + + ptr = (u8 *)malloc(1024000); + + if (ptr && (ssize_t)ptr != -1) __afl_area_ptr = ptr; + +} + +/* The following stuff deals with supporting -fsanitize-coverage=trace-pc-guard. + It remains non-operational in the traditional, plugin-backed LLVM mode. + For more info about 'trace-pc-guard', see README.llvm.md. + + The first function (__sanitizer_cov_trace_pc_guard) is called back on every + edge (as opposed to every basic block). */ + +void __sanitizer_cov_trace_pc_guard(uint32_t *guard) { + + // For stability analysis, if you want to know to which function unstable + // edge IDs belong - uncomment, recompile+install llvm_mode, recompile + // the target. libunwind and libbacktrace are better solutions. + // Set AFL_DEBUG_CHILD_OUTPUT=1 and run afl-fuzz with 2>file to capture + // the backtrace output + /* + uint32_t unstable[] = { ... unstable edge IDs }; + uint32_t idx; + char bt[1024]; + for (idx = 0; i < sizeof(unstable)/sizeof(uint32_t); i++) { + + if (unstable[idx] == __afl_area_ptr[*guard]) { + + int bt_size = backtrace(bt, 256); + if (bt_size > 0) { + + char **bt_syms = backtrace_symbols(bt, bt_size); + if (bt_syms) { + + fprintf(stderr, "DEBUG: edge=%u caller=%s\n", unstable[idx], + bt_syms[0]); + free(bt_syms); + + } + + } + + } + + } + + */ + +#if (LLVM_VERSION_MAJOR < 9) + + __afl_area_ptr[*guard]++; + +#else + + __afl_area_ptr[*guard] = + __afl_area_ptr[*guard] + 1 + (__afl_area_ptr[*guard] == 255 ? 1 : 0); + +#endif + +} + +/* Init callback. Populates instrumentation IDs. Note that we're using + ID of 0 as a special value to indicate non-instrumented bits. That may + still touch the bitmap, but in a fairly harmless way. */ + +void __sanitizer_cov_trace_pc_guard_init(uint32_t *start, uint32_t *stop) { + + u32 inst_ratio = 100; + char *x; + + _is_sancov = 1; + + if (getenv("AFL_DEBUG")) { + + fprintf(stderr, "Running __sanitizer_cov_trace_pc_guard_init: %p-%p\n", + start, stop); + + } + + if (start == stop || *start) return; + + x = getenv("AFL_INST_RATIO"); + if (x) inst_ratio = (u32)atoi(x); + + if (!inst_ratio || inst_ratio > 100) { + + fprintf(stderr, "[-] ERROR: Invalid AFL_INST_RATIO (must be 1-100).\n"); + abort(); + + } + + /* Make sure that the first element in the range is always set - we use that + to avoid duplicate calls (which can happen as an artifact of the underlying + implementation in LLVM). */ + + *(start++) = R(MAP_SIZE - 1) + 1; + + while (start < stop) { + + if (R(100) < inst_ratio) + *start = ++__afl_final_loc; + else + *start = 0; + + start++; + + } + +} + +///// CmpLog instrumentation + +void __cmplog_ins_hook1(uint8_t arg1, uint8_t arg2) { + + if (unlikely(!__afl_cmp_map)) return; + + uintptr_t k = (uintptr_t)__builtin_return_address(0); + k = (k >> 4) ^ (k << 8); + k &= CMP_MAP_W - 1; + + __afl_cmp_map->headers[k].type = CMP_TYPE_INS; + + u32 hits = __afl_cmp_map->headers[k].hits; + __afl_cmp_map->headers[k].hits = hits + 1; + // if (!__afl_cmp_map->headers[k].cnt) + // __afl_cmp_map->headers[k].cnt = __afl_cmp_counter++; + + __afl_cmp_map->headers[k].shape = 0; + + hits &= CMP_MAP_H - 1; + __afl_cmp_map->log[k][hits].v0 = arg1; + __afl_cmp_map->log[k][hits].v1 = arg2; + +} + +void __cmplog_ins_hook2(uint16_t arg1, uint16_t arg2) { + + if (unlikely(!__afl_cmp_map)) return; + + uintptr_t k = (uintptr_t)__builtin_return_address(0); + k = (k >> 4) ^ (k << 8); + k &= CMP_MAP_W - 1; + + __afl_cmp_map->headers[k].type = CMP_TYPE_INS; + + u32 hits = __afl_cmp_map->headers[k].hits; + __afl_cmp_map->headers[k].hits = hits + 1; + + __afl_cmp_map->headers[k].shape = 1; + + hits &= CMP_MAP_H - 1; + __afl_cmp_map->log[k][hits].v0 = arg1; + __afl_cmp_map->log[k][hits].v1 = arg2; + +} + +void __cmplog_ins_hook4(uint32_t arg1, uint32_t arg2) { + + if (unlikely(!__afl_cmp_map)) return; + + uintptr_t k = (uintptr_t)__builtin_return_address(0); + k = (k >> 4) ^ (k << 8); + k &= CMP_MAP_W - 1; + + __afl_cmp_map->headers[k].type = CMP_TYPE_INS; + + u32 hits = __afl_cmp_map->headers[k].hits; + __afl_cmp_map->headers[k].hits = hits + 1; + + __afl_cmp_map->headers[k].shape = 3; + + hits &= CMP_MAP_H - 1; + __afl_cmp_map->log[k][hits].v0 = arg1; + __afl_cmp_map->log[k][hits].v1 = arg2; + +} + +void __cmplog_ins_hook8(uint64_t arg1, uint64_t arg2) { + + if (unlikely(!__afl_cmp_map)) return; + + uintptr_t k = (uintptr_t)__builtin_return_address(0); + k = (k >> 4) ^ (k << 8); + k &= CMP_MAP_W - 1; + + __afl_cmp_map->headers[k].type = CMP_TYPE_INS; + + u32 hits = __afl_cmp_map->headers[k].hits; + __afl_cmp_map->headers[k].hits = hits + 1; + + __afl_cmp_map->headers[k].shape = 7; + + hits &= CMP_MAP_H - 1; + __afl_cmp_map->log[k][hits].v0 = arg1; + __afl_cmp_map->log[k][hits].v1 = arg2; + +} + +#if defined(__APPLE__) + #pragma weak __sanitizer_cov_trace_const_cmp1 = __cmplog_ins_hook1 + #pragma weak __sanitizer_cov_trace_const_cmp2 = __cmplog_ins_hook2 + #pragma weak __sanitizer_cov_trace_const_cmp4 = __cmplog_ins_hook4 + #pragma weak __sanitizer_cov_trace_const_cmp8 = __cmplog_ins_hook8 + + #pragma weak __sanitizer_cov_trace_cmp1 = __cmplog_ins_hook1 + #pragma weak __sanitizer_cov_trace_cmp2 = __cmplog_ins_hook2 + #pragma weak __sanitizer_cov_trace_cmp4 = __cmplog_ins_hook4 + #pragma weak __sanitizer_cov_trace_cmp8 = __cmplog_ins_hook8 +#else +void __sanitizer_cov_trace_const_cmp1(uint8_t arg1, uint8_t arg2) + __attribute__((alias("__cmplog_ins_hook1"))); +void __sanitizer_cov_trace_const_cmp2(uint16_t arg1, uint16_t arg2) + __attribute__((alias("__cmplog_ins_hook2"))); +void __sanitizer_cov_trace_const_cmp4(uint32_t arg1, uint32_t arg2) + __attribute__((alias("__cmplog_ins_hook4"))); +void __sanitizer_cov_trace_const_cmp8(uint64_t arg1, uint64_t arg2) + __attribute__((alias("__cmplog_ins_hook8"))); + +void __sanitizer_cov_trace_cmp1(uint8_t arg1, uint8_t arg2) + __attribute__((alias("__cmplog_ins_hook1"))); +void __sanitizer_cov_trace_cmp2(uint16_t arg1, uint16_t arg2) + __attribute__((alias("__cmplog_ins_hook2"))); +void __sanitizer_cov_trace_cmp4(uint32_t arg1, uint32_t arg2) + __attribute__((alias("__cmplog_ins_hook4"))); +void __sanitizer_cov_trace_cmp8(uint64_t arg1, uint64_t arg2) + __attribute__((alias("__cmplog_ins_hook8"))); +#endif /* defined(__APPLE__) */ + +void __sanitizer_cov_trace_switch(uint64_t val, uint64_t *cases) { + + if (unlikely(!__afl_cmp_map)) return; + + for (uint64_t i = 0; i < cases[0]; i++) { + + uintptr_t k = (uintptr_t)__builtin_return_address(0) + i; + k = (k >> 4) ^ (k << 8); + k &= CMP_MAP_W - 1; + + __afl_cmp_map->headers[k].type = CMP_TYPE_INS; + + u32 hits = __afl_cmp_map->headers[k].hits; + __afl_cmp_map->headers[k].hits = hits + 1; + + __afl_cmp_map->headers[k].shape = 7; + + hits &= CMP_MAP_H - 1; + __afl_cmp_map->log[k][hits].v0 = val; + __afl_cmp_map->log[k][hits].v1 = cases[i + 2]; + + } + +} + +// POSIX shenanigan to see if an area is mapped. +// If it is mapped as X-only, we have a problem, so maybe we should add a check +// to avoid to call it on .text addresses +static int area_is_mapped(void *ptr, size_t len) { + + char *p = ptr; + char *page = (char *)((uintptr_t)p & ~(sysconf(_SC_PAGE_SIZE) - 1)); + + int r = msync(page, (p - page) + len, MS_ASYNC); + if (r < 0) return errno != ENOMEM; + return 1; + +} + +void __cmplog_rtn_hook(u8 *ptr1, u8 *ptr2) { + + if (unlikely(!__afl_cmp_map)) return; + + if (!area_is_mapped(ptr1, 32) || !area_is_mapped(ptr2, 32)) return; + + uintptr_t k = (uintptr_t)__builtin_return_address(0); + k = (k >> 4) ^ (k << 8); + k &= CMP_MAP_W - 1; + + __afl_cmp_map->headers[k].type = CMP_TYPE_RTN; + + u32 hits = __afl_cmp_map->headers[k].hits; + __afl_cmp_map->headers[k].hits = hits + 1; + + __afl_cmp_map->headers[k].shape = 31; + + hits &= CMP_MAP_RTN_H - 1; + __builtin_memcpy(((struct cmpfn_operands *)__afl_cmp_map->log[k])[hits].v0, + ptr1, 32); + __builtin_memcpy(((struct cmpfn_operands *)__afl_cmp_map->log[k])[hits].v1, + ptr2, 32); + +} + diff --git a/instrumentation/afl-gcc-pass.so.cc b/instrumentation/afl-gcc-pass.so.cc new file mode 100644 index 00000000..c5614aca --- /dev/null +++ b/instrumentation/afl-gcc-pass.so.cc @@ -0,0 +1,601 @@ +// +// There are some TODOs in this file: +// - fix instrumentation via external call +// - fix inline instrumentation +// - implement instrument list feature +// - dont instrument blocks that are uninteresting +// - implement neverZero +// + +/* + american fuzzy lop++ - GCC instrumentation pass + --------------------------------------------- + + Written by Austin Seipp with bits from + Emese Revfy + + Fixed by Heiko Eißfeldt 2019-2020 for AFL++ + + GCC integration design is based on the LLVM design, which comes + from Laszlo Szekeres. Some of the boilerplate code below for + afl_pass to adapt to different GCC versions was taken from Emese + Revfy's Size Overflow plugin for GCC, licensed under the GPLv2/v3. + + (NOTE: this plugin code is under GPLv3, in order to comply with the + GCC runtime library exception, which states that you may distribute + "Target Code" from the compiler under a license of your choice, as + long as the "Compilation Process" is "Eligible", and contains no + GPL-incompatible software in GCC "during the process of + transforming high level code to target code". In this case, the + plugin will be used to generate "Target Code" during the + "Compilation Process", and thus it must be GPLv3 to be "eligible".) + + Copyright (C) 2015 Austin Seipp + + This program is free software: you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation, either version 3 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with this program. If not, see . + + */ + +#define BUILD_INLINE_INST + +#include "../include/config.h" +#include "../include/debug.h" + +/* clear helper macros AFL types pull in, which intervene with gcc-plugin + * headers from GCC-8 */ +#ifdef likely + #undef likely +#endif +#ifdef unlikely + #undef unlikely +#endif + +#include +#include +#include + +#include +#include +#include + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +/* -------------------------------------------------------------------------- */ +/* -- AFL instrumentation pass ---------------------------------------------- */ + +static int be_quiet = 0; +static unsigned int inst_ratio = 100; +static bool inst_ext = true; +static std::list myInstrumentList; + +static unsigned int ext_call_instrument(function *fun) { + + /* Instrument all the things! */ + basic_block bb; + unsigned finst_blocks = 0; + unsigned fcnt_blocks = 0; + + tree fntype = build_function_type_list(void_type_node, /* return */ + uint32_type_node, /* args */ + NULL_TREE); /* done */ + tree fndecl = build_fn_decl("__afl_trace", fntype); + TREE_STATIC(fndecl) = 1; /* Defined elsewhere */ + TREE_PUBLIC(fndecl) = 1; /* Public */ + DECL_EXTERNAL(fndecl) = 1; /* External linkage */ + DECL_ARTIFICIAL(fndecl) = 1; /* Injected by compiler */ + + FOR_EACH_BB_FN(bb, fun) { + + gimple_seq fcall; + gimple_seq seq = NULL; + gimple_stmt_iterator bentry; + ++fcnt_blocks; + + // only instrument if this basic block is the destination of a previous + // basic block that has multiple successors + // this gets rid of ~5-10% of instrumentations that are unnecessary + // result: a little more speed and less map pollution + + int more_than_one = -1; + edge ep; + edge_iterator eip; + + FOR_EACH_EDGE(ep, eip, bb->preds) { + + int count = 0; + if (more_than_one == -1) more_than_one = 0; + + basic_block Pred = ep->src; + edge es; + edge_iterator eis; + FOR_EACH_EDGE(es, eis, Pred->succs) { + + basic_block Succ = es->dest; + if (Succ != NULL) count++; + + } + + if (count > 1) more_than_one = 1; + + } + + if (more_than_one != 1) continue; + + /* Bail on this block if we trip the specified ratio */ + if (R(100) >= inst_ratio) continue; + + /* Make up cur_loc */ + unsigned int rand_loc = R(MAP_SIZE); + tree cur_loc = build_int_cst(uint32_type_node, rand_loc); + + /* Update bitmap via external call */ + /* to quote: + * /+ Trace a basic block with some ID +/ + * void __afl_trace(u32 x); + */ + + fcall = gimple_build_call( + fndecl, 1, + cur_loc); /* generate the function _call_ to above built reference, with + *1* parameter -> the random const for the location */ + gimple_seq_add_stmt(&seq, fcall); /* and insert into a sequence */ + + /* Done - grab the entry to the block and insert sequence */ + bentry = gsi_after_labels(bb); + gsi_insert_seq_before(&bentry, seq, GSI_SAME_STMT); + + ++finst_blocks; + + } + + /* Say something nice. */ + if (!be_quiet) { + + if (!finst_blocks) + WARNF(G_("No instrumentation targets found in " cBRI "%s" cRST), + function_name(fun)); + else if (finst_blocks < fcnt_blocks) + OKF(G_("Instrumented %2u /%2u locations in " cBRI "%s" cRST), + finst_blocks, fcnt_blocks, function_name(fun)); + else + OKF(G_("Instrumented %2u locations in " cBRI "%s" cRST), finst_blocks, + function_name(fun)); + + } + + return 0; + +} + +static unsigned int inline_instrument(function *fun) { + + /* Instrument all the things! */ + basic_block bb; + unsigned finst_blocks = 0; + unsigned fcnt_blocks = 0; + tree one = build_int_cst(unsigned_char_type_node, 1); + // tree zero = build_int_cst(unsigned_char_type_node, 0); + + /* Set up global type declarations */ + tree map_type = build_pointer_type(unsigned_char_type_node); + tree map_ptr_g = + build_decl(UNKNOWN_LOCATION, VAR_DECL, + get_identifier_with_length("__afl_area_ptr", 14), map_type); + TREE_USED(map_ptr_g) = 1; + TREE_STATIC(map_ptr_g) = 1; /* Defined elsewhere */ + DECL_EXTERNAL(map_ptr_g) = 1; /* External linkage */ + DECL_PRESERVE_P(map_ptr_g) = 1; + DECL_ARTIFICIAL(map_ptr_g) = 1; /* Injected by compiler */ + rest_of_decl_compilation(map_ptr_g, 1, 0); + + tree prev_loc_g = build_decl(UNKNOWN_LOCATION, VAR_DECL, + get_identifier_with_length("__afl_prev_loc", 14), + uint32_type_node); + TREE_USED(prev_loc_g) = 1; + TREE_STATIC(prev_loc_g) = 1; /* Defined elsewhere */ + DECL_EXTERNAL(prev_loc_g) = 1; /* External linkage */ + DECL_PRESERVE_P(prev_loc_g) = 1; + DECL_ARTIFICIAL(prev_loc_g) = 1; /* Injected by compiler */ + set_decl_tls_model(prev_loc_g, TLS_MODEL_REAL); /* TLS attribute */ + rest_of_decl_compilation(prev_loc_g, 1, 0); + + FOR_EACH_BB_FN(bb, fun) { + + gimple_seq seq = NULL; + gimple_stmt_iterator bentry; + ++fcnt_blocks; + + // only instrument if this basic block is the destination of a previous + // basic block that has multiple successors + // this gets rid of ~5-10% of instrumentations that are unnecessary + // result: a little more speed and less map pollution + + int more_than_one = -1; + edge ep; + edge_iterator eip; + FOR_EACH_EDGE(ep, eip, bb->preds) { + + int count = 0; + if (more_than_one == -1) more_than_one = 0; + + basic_block Pred = ep->src; + edge es; + edge_iterator eis; + FOR_EACH_EDGE(es, eis, Pred->succs) { + + basic_block Succ = es->dest; + if (Succ != NULL) count++; + + } + + if (count > 1) more_than_one = 1; + + } + + if (more_than_one != 1) continue; + + /* Bail on this block if we trip the specified ratio */ + if (R(100) >= inst_ratio) continue; + + /* Make up cur_loc */ + + unsigned int rand_loc = R(MAP_SIZE); + tree cur_loc = build_int_cst(uint32_type_node, rand_loc); + + /* Load prev_loc, xor with cur_loc */ + // gimple_assign + tree prev_loc = create_tmp_var_raw(uint32_type_node, "prev_loc"); + gassign *g = gimple_build_assign(prev_loc, VAR_DECL, prev_loc_g); + gimple_seq_add_stmt(&seq, g); // load prev_loc + update_stmt(g); + + // gimple_assign + tree area_off = create_tmp_var_raw(uint32_type_node, "area_off"); + g = gimple_build_assign(area_off, BIT_XOR_EXPR, prev_loc, cur_loc); + gimple_seq_add_stmt(&seq, g); // area_off = prev_loc ^ cur_loc + update_stmt(g); + + /* Update bitmap */ + + // gimple_assign + tree map_ptr = create_tmp_var(map_type, "map_ptr"); + tree map_ptr2 = create_tmp_var(map_type, "map_ptr2"); + + g = gimple_build_assign(map_ptr, map_ptr_g); + gimple_seq_add_stmt(&seq, g); // map_ptr = __afl_area_ptr + update_stmt(g); + +#if 1 + #if 0 + tree addr = build2(ADDR_EXPR, map_type, map_ptr, area_off); + g = gimple_build_assign(map_ptr2, MODIFY_EXPR, addr); + gimple_seq_add_stmt(&seq, g); // map_ptr2 = map_ptr + area_off + update_stmt(g); + #else + g = gimple_build_assign(map_ptr2, PLUS_EXPR, map_ptr, area_off); + gimple_seq_add_stmt(&seq, g); // map_ptr2 = map_ptr + area_off + update_stmt(g); + #endif + + // gimple_assign + tree tmp1 = create_tmp_var_raw(unsigned_char_type_node, "tmp1"); + g = gimple_build_assign(tmp1, MEM_REF, map_ptr2); + gimple_seq_add_stmt(&seq, g); // tmp1 = *map_ptr2 + update_stmt(g); +#else + tree atIndex = build2(PLUS_EXPR, uint32_type_node, map_ptr, area_off); + tree array_address = build1(ADDR_EXPR, map_type, atIndex); + tree array_access = build1(INDIRECT_REF, map_type, array_address); + tree tmp1 = create_tmp_var(unsigned_char_type_node, "tmp1"); + g = gimple_build_assign(tmp1, array_access); + gimple_seq_add_stmt(&seq, g); // tmp1 = *(map_ptr + area_off) + update_stmt(g); +#endif + // gimple_assign + tree tmp2 = create_tmp_var_raw(unsigned_char_type_node, "tmp2"); + g = gimple_build_assign(tmp2, PLUS_EXPR, tmp1, one); + gimple_seq_add_stmt(&seq, g); // tmp2 = tmp1 + 1 + update_stmt(g); + + // TODO: neverZero: here we have to check if tmp3 == 0 + // and add 1 if so + + // gimple_assign + // tree map_ptr3 = create_tmp_var_raw(map_type, "map_ptr3"); + g = gimple_build_assign(map_ptr2, INDIRECT_REF, tmp2); + gimple_seq_add_stmt(&seq, g); // *map_ptr2 = tmp2 + update_stmt(g); + + /* Set prev_loc to cur_loc >> 1 */ + + // gimple_assign + tree shifted_loc = build_int_cst(TREE_TYPE(prev_loc_g), rand_loc >> 1); + tree prev_loc2 = create_tmp_var_raw(uint32_type_node, "prev_loc2"); + g = gimple_build_assign(prev_loc2, shifted_loc); + gimple_seq_add_stmt(&seq, g); // __afl_prev_loc = cur_loc >> 1 + update_stmt(g); + g = gimple_build_assign(prev_loc_g, prev_loc2); + gimple_seq_add_stmt(&seq, g); // __afl_prev_loc = cur_loc >> 1 + update_stmt(g); + + /* Done - grab the entry to the block and insert sequence */ + + bentry = gsi_after_labels(bb); + gsi_insert_seq_before(&bentry, seq, GSI_NEW_STMT); + + ++finst_blocks; + + } + + /* Say something nice. */ + if (!be_quiet) { + + if (!finst_blocks) + WARNF(G_("No instrumentation targets found in " cBRI "%s" cRST), + function_name(fun)); + else if (finst_blocks < fcnt_blocks) + OKF(G_("Instrumented %2u /%2u locations in " cBRI "%s" cRST), + finst_blocks, fcnt_blocks, function_name(fun)); + else + OKF(G_("Instrumented %2u locations in " cBRI "%s" cRST), finst_blocks, + function_name(fun)); + + } + + return 0; + +} + +/* -------------------------------------------------------------------------- */ +/* -- Boilerplate and initialization ---------------------------------------- */ + +static const struct pass_data afl_pass_data = { + + .type = GIMPLE_PASS, + .name = "afl-inst", + .optinfo_flags = OPTGROUP_NONE, + + .tv_id = TV_NONE, + .properties_required = 0, + .properties_provided = 0, + .properties_destroyed = 0, + .todo_flags_start = 0, + // NOTE(aseipp): it's very, very important to include + // at least 'TODO_update_ssa' here so that GCC will + // properly update the resulting SSA form, e.g., to + // include new PHI nodes for newly added symbols or + // names. Do not remove this. Do not taunt Happy Fun + // Ball. + .todo_flags_finish = TODO_update_ssa | TODO_verify_il | TODO_cleanup_cfg, + +}; + +namespace { + +class afl_pass : public gimple_opt_pass { + + private: + bool do_ext_call; + + public: + afl_pass(bool ext_call, gcc::context *g) + : gimple_opt_pass(afl_pass_data, g), do_ext_call(ext_call) { + + } + + unsigned int execute(function *fun) override { + + if (!myInstrumentList.empty()) { + + bool instrumentBlock = false; + std::string instFilename; + unsigned int instLine = 0; + + /* EXPR_FILENAME + This macro returns the name of the file in which the entity was declared, + as a char*. For an entity declared implicitly by the compiler (like + __builtin_ memcpy), this will be the string "". + */ + const char *fname = DECL_SOURCE_FILE(fun->decl); + + if (0 != strncmp("", fname, 10) && + 0 != strncmp("", fname, 10)) { + + instFilename = fname; + instLine = DECL_SOURCE_LINE(fun->decl); + + /* Continue only if we know where we actually are */ + if (!instFilename.empty()) { + + for (std::list::iterator it = myInstrumentList.begin(); + it != myInstrumentList.end(); ++it) { + + /* We don't check for filename equality here because + * filenames might actually be full paths. Instead we + * check that the actual filename ends in the filename + * specified in the list. */ + if (instFilename.length() >= it->length()) { + + if (instFilename.compare(instFilename.length() - it->length(), + it->length(), *it) == 0) { + + instrumentBlock = true; + break; + + } + + } + + } + + } + + } + + /* Either we couldn't figure out our location or the location is + * not in the instrument list, so we skip instrumentation. */ + if (!instrumentBlock) { + + if (!be_quiet) { + + if (!instFilename.empty()) + SAYF(cYEL "[!] " cBRI + "Not in instrument list, skipping %s line %u...\n", + instFilename.c_str(), instLine); + else + SAYF(cYEL "[!] " cBRI "No filename information found, skipping it"); + + } + + return 0; + + } + + } + + return do_ext_call ? ext_call_instrument(fun) : inline_instrument(fun); + + } + +}; /* class afl_pass */ + +} // namespace + +static struct opt_pass *make_afl_pass(bool ext_call, gcc::context *ctxt) { + + return new afl_pass(ext_call, ctxt); + +} + +/* -------------------------------------------------------------------------- */ +/* -- Initialization -------------------------------------------------------- */ + +int plugin_is_GPL_compatible = 1; + +static struct plugin_info afl_plugin_info = { + + .version = "20200519", + .help = "AFL++ gcc plugin\n", + +}; + +int plugin_init(struct plugin_name_args * plugin_info, + struct plugin_gcc_version *version) { + + struct register_pass_info afl_pass_info; + struct timeval tv; + struct timezone tz; + u32 rand_seed; + + /* Setup random() so we get Actually Random(TM) outputs from R() */ + gettimeofday(&tv, &tz); + rand_seed = tv.tv_sec ^ tv.tv_usec ^ getpid(); + SR(rand_seed); + + /* Pass information */ + afl_pass_info.pass = make_afl_pass(inst_ext, g); + afl_pass_info.reference_pass_name = "ssa"; + afl_pass_info.ref_pass_instance_number = 1; + afl_pass_info.pos_op = PASS_POS_INSERT_AFTER; + + if (!plugin_default_version_check(version, &gcc_version)) { + + FATAL(G_("Incompatible gcc/plugin versions! Expected GCC %d.%d"), + GCCPLUGIN_VERSION_MAJOR, GCCPLUGIN_VERSION_MINOR); + + } + + /* Show a banner */ + if ((isatty(2) && !getenv("AFL_QUIET")) || getenv("AFL_DEBUG") != NULL) { + + SAYF(G_(cCYA "afl-gcc-pass" VERSION cRST + " initially by , maintainer: hexcoder-\n")); + + } else + + be_quiet = 1; + + /* Decide instrumentation ratio */ + char *inst_ratio_str = getenv("AFL_INST_RATIO"); + + if (inst_ratio_str) { + + if (sscanf(inst_ratio_str, "%u", &inst_ratio) != 1 || !inst_ratio || + inst_ratio > 100) + FATAL(G_("Bad value of AFL_INST_RATIO (must be between 1 and 100)")); + else { + + if (!be_quiet) + ACTF(G_("%s instrumentation at ratio of %u%% in %s mode."), + inst_ext ? G_("Call-based") : G_("Inline"), inst_ratio, + getenv("AFL_HARDEN") ? G_("hardened") : G_("non-hardened")); + + } + + } + + char *instInstrumentListFilename = getenv("AFL_GCC_INSTRUMENT_FILE"); + if (!instInstrumentListFilename) + instInstrumentListFilename = getenv("AFL_GCC_WHITELIST"); + if (instInstrumentListFilename) { + + std::string line; + std::ifstream fileStream; + fileStream.open(instInstrumentListFilename); + if (!fileStream) PFATAL("Unable to open AFL_GCC_INSTRUMENT_FILE"); + getline(fileStream, line); + while (fileStream) { + + myInstrumentList.push_back(line); + getline(fileStream, line); + + } + + } else if (!be_quiet && (getenv("AFL_LLVM_WHITELIST") || + + getenv("AFL_LLVM_INSTRUMENT_FILE"))) { + + SAYF(cYEL "[-] " cRST + "AFL_LLVM_INSTRUMENT_FILE environment variable detected - did " + "you mean AFL_GCC_INSTRUMENT_FILE?\n"); + + } + + /* Go go gadget */ + register_callback(plugin_info->base_name, PLUGIN_INFO, NULL, + &afl_plugin_info); + register_callback(plugin_info->base_name, PLUGIN_PASS_MANAGER_SETUP, NULL, + &afl_pass_info); + return 0; + +} + diff --git a/instrumentation/afl-llvm-common.cc b/instrumentation/afl-llvm-common.cc new file mode 100644 index 00000000..189b4ec6 --- /dev/null +++ b/instrumentation/afl-llvm-common.cc @@ -0,0 +1,575 @@ +#define AFL_LLVM_PASS + +#include "config.h" +#include "debug.h" + +#include +#include +#include +#include +#include + +#include +#include +#include + +#include + +#define IS_EXTERN extern +#include "afl-llvm-common.h" + +using namespace llvm; + +static std::list allowListFiles; +static std::list allowListFunctions; +static std::list denyListFiles; +static std::list denyListFunctions; + +char *getBBName(const llvm::BasicBlock *BB) { + + static char *name; + + if (!BB->getName().empty()) { + + name = strdup(BB->getName().str().c_str()); + return name; + + } + + std::string Str; + raw_string_ostream OS(Str); + +#if LLVM_VERSION_MAJOR >= 4 || \ + (LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR >= 7) + BB->printAsOperand(OS, false); +#endif + name = strdup(OS.str().c_str()); + return name; + +} + +/* Function that we never instrument or analyze */ +/* Note: this ignore check is also called in isInInstrumentList() */ +bool isIgnoreFunction(const llvm::Function *F) { + + // Starting from "LLVMFuzzer" these are functions used in libfuzzer based + // fuzzing campaign installations, e.g. oss-fuzz + + static const char *ignoreList[] = { + + "asan.", + "llvm.", + "sancov.", + "__ubsan_", + "ign.", + "__afl_", + "_fini", + "__libc_csu", + "__asan", + "__msan", + "__cmplog", + "__sancov", + "msan.", + "LLVMFuzzer", + "__decide_deferred", + "maybe_duplicate_stderr", + "discard_output", + "close_stdout", + "dup_and_close_stderr", + "maybe_close_fd_mask", + "ExecuteFilesOnyByOne" + + }; + + for (auto const &ignoreListFunc : ignoreList) { + + if (F->getName().startswith(ignoreListFunc)) { return true; } + + } + + return false; + +} + +void initInstrumentList() { + + char *allowlist = getenv("AFL_LLVM_ALLOWLIST"); + if (!allowlist) allowlist = getenv("AFL_LLVM_INSTRUMENT_FILE"); + if (!allowlist) allowlist = getenv("AFL_LLVM_WHITELIST"); + char *denylist = getenv("AFL_LLVM_DENYLIST"); + if (!denylist) denylist = getenv("AFL_LLVM_BLOCKLIST"); + + if (allowlist && denylist) + FATAL( + "You can only specify either AFL_LLVM_ALLOWLIST or AFL_LLVM_DENYLIST " + "but not both!"); + + if (allowlist) { + + std::string line; + std::ifstream fileStream; + fileStream.open(allowlist); + if (!fileStream) report_fatal_error("Unable to open AFL_LLVM_ALLOWLIST"); + getline(fileStream, line); + + while (fileStream) { + + int is_file = -1; + std::size_t npos; + std::string original_line = line; + + line.erase(std::remove_if(line.begin(), line.end(), ::isspace), + line.end()); + + // remove # and following + if ((npos = line.find("#")) != std::string::npos) + line = line.substr(0, npos); + + if (line.compare(0, 4, "fun:") == 0) { + + is_file = 0; + line = line.substr(4); + + } else if (line.compare(0, 9, "function:") == 0) { + + is_file = 0; + line = line.substr(9); + + } else if (line.compare(0, 4, "src:") == 0) { + + is_file = 1; + line = line.substr(4); + + } else if (line.compare(0, 7, "source:") == 0) { + + is_file = 1; + line = line.substr(7); + + } + + if (line.find(":") != std::string::npos) { + + FATAL("invalid line in AFL_LLVM_ALLOWLIST: %s", original_line.c_str()); + + } + + if (line.length() > 0) { + + // if the entry contains / or . it must be a file + if (is_file == -1) + if (line.find("/") != std::string::npos || + line.find(".") != std::string::npos) + is_file = 1; + // otherwise it is a function + + if (is_file == 1) + allowListFiles.push_back(line); + else + allowListFunctions.push_back(line); + getline(fileStream, line); + + } + + } + + if (debug) + SAYF(cMGN "[D] " cRST + "loaded allowlist with %zu file and %zu function entries\n", + allowListFiles.size(), allowListFunctions.size()); + + } + + if (denylist) { + + std::string line; + std::ifstream fileStream; + fileStream.open(denylist); + if (!fileStream) report_fatal_error("Unable to open AFL_LLVM_DENYLIST"); + getline(fileStream, line); + + while (fileStream) { + + int is_file = -1; + std::size_t npos; + std::string original_line = line; + + line.erase(std::remove_if(line.begin(), line.end(), ::isspace), + line.end()); + + // remove # and following + if ((npos = line.find("#")) != std::string::npos) + line = line.substr(0, npos); + + if (line.compare(0, 4, "fun:") == 0) { + + is_file = 0; + line = line.substr(4); + + } else if (line.compare(0, 9, "function:") == 0) { + + is_file = 0; + line = line.substr(9); + + } else if (line.compare(0, 4, "src:") == 0) { + + is_file = 1; + line = line.substr(4); + + } else if (line.compare(0, 7, "source:") == 0) { + + is_file = 1; + line = line.substr(7); + + } + + if (line.find(":") != std::string::npos) { + + FATAL("invalid line in AFL_LLVM_DENYLIST: %s", original_line.c_str()); + + } + + if (line.length() > 0) { + + // if the entry contains / or . it must be a file + if (is_file == -1) + if (line.find("/") != std::string::npos || + line.find(".") != std::string::npos) + is_file = 1; + // otherwise it is a function + + if (is_file == 1) + denyListFiles.push_back(line); + else + denyListFunctions.push_back(line); + getline(fileStream, line); + + } + + } + + if (debug) + SAYF(cMGN "[D] " cRST + "loaded denylist with %zu file and %zu function entries\n", + denyListFiles.size(), denyListFunctions.size()); + + } + +} + +void scanForDangerousFunctions(llvm::Module *M) { + + if (!M) return; + +#if LLVM_VERSION_MAJOR > 3 || \ + (LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR >= 9) + + for (GlobalIFunc &IF : M->ifuncs()) { + + StringRef ifunc_name = IF.getName(); + Constant *r = IF.getResolver(); + StringRef r_name = cast(r->getOperand(0))->getName(); + if (!be_quiet) + fprintf(stderr, + "Info: Found an ifunc with name %s that points to resolver " + "function %s, we will not instrument this, putting it into the " + "block list.\n", + ifunc_name.str().c_str(), r_name.str().c_str()); + denyListFunctions.push_back(r_name.str()); + + } + + GlobalVariable *GV = M->getNamedGlobal("llvm.global_ctors"); + if (GV && !GV->isDeclaration() && !GV->hasLocalLinkage()) { + + ConstantArray *InitList = dyn_cast(GV->getInitializer()); + + if (InitList) { + + for (unsigned i = 0, e = InitList->getNumOperands(); i != e; ++i) { + + if (ConstantStruct *CS = + dyn_cast(InitList->getOperand(i))) { + + if (CS->getNumOperands() >= 2) { + + if (CS->getOperand(1)->isNullValue()) + break; // Found a null terminator, stop here. + + ConstantInt *CI = dyn_cast(CS->getOperand(0)); + int Priority = CI ? CI->getSExtValue() : 0; + + Constant *FP = CS->getOperand(1); + if (ConstantExpr *CE = dyn_cast(FP)) + if (CE->isCast()) FP = CE->getOperand(0); + if (Function *F = dyn_cast(FP)) { + + if (!F->isDeclaration() && + strncmp(F->getName().str().c_str(), "__afl", 5) != 0) { + + if (!be_quiet) + fprintf(stderr, + "Info: Found constructor function %s with prio " + "%u, we will not instrument this, putting it into a " + "block list.\n", + F->getName().str().c_str(), Priority); + denyListFunctions.push_back(F->getName().str()); + + } + + } + + } + + } + + } + + } + + } + +#endif + +} + +static std::string getSourceName(llvm::Function *F) { + + // let's try to get the filename for the function + auto bb = &F->getEntryBlock(); + BasicBlock::iterator IP = bb->getFirstInsertionPt(); + IRBuilder<> IRB(&(*IP)); + DebugLoc Loc = IP->getDebugLoc(); + +#if LLVM_VERSION_MAJOR >= 4 || \ + (LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR >= 7) + if (Loc) { + + StringRef instFilename; + DILocation *cDILoc = dyn_cast(Loc.getAsMDNode()); + + if (cDILoc) { instFilename = cDILoc->getFilename(); } + + if (instFilename.str().empty()) { + + /* If the original location is empty, try using the inlined location + */ + DILocation *oDILoc = cDILoc->getInlinedAt(); + if (oDILoc) { instFilename = oDILoc->getFilename(); } + + } + + return instFilename.str(); + + } + +#else + if (!Loc.isUnknown()) { + + DILocation cDILoc(Loc.getAsMDNode(F->getContext())); + + StringRef instFilename = cDILoc.getFilename(); + + /* Continue only if we know where we actually are */ + return instFilename.str(); + + } + +#endif + + return std::string(""); + +} + +bool isInInstrumentList(llvm::Function *F) { + + bool return_default = true; + + // is this a function with code? If it is external we don't instrument it + // anyway and it can't be in the instrument file list. Or if it is it is + // ignored. + if (!F->size() || isIgnoreFunction(F)) return false; + + if (!denyListFiles.empty() || !denyListFunctions.empty()) { + + if (!denyListFunctions.empty()) { + + std::string instFunction = F->getName().str(); + + for (std::list::iterator it = denyListFunctions.begin(); + it != denyListFunctions.end(); ++it) { + + /* We don't check for filename equality here because + * filenames might actually be full paths. Instead we + * check that the actual filename ends in the filename + * specified in the list. We also allow UNIX-style pattern + * matching */ + + if (instFunction.length() >= it->length()) { + + if (fnmatch(("*" + *it).c_str(), instFunction.c_str(), 0) == 0) { + + if (debug) + SAYF(cMGN "[D] " cRST + "Function %s is in the deny function list, " + "not instrumenting ... \n", + instFunction.c_str()); + return false; + + } + + } + + } + + } + + if (!denyListFiles.empty()) { + + std::string source_file = getSourceName(F); + + if (!source_file.empty()) { + + for (std::list::iterator it = denyListFiles.begin(); + it != denyListFiles.end(); ++it) { + + /* We don't check for filename equality here because + * filenames might actually be full paths. Instead we + * check that the actual filename ends in the filename + * specified in the list. We also allow UNIX-style pattern + * matching */ + + if (source_file.length() >= it->length()) { + + if (fnmatch(("*" + *it).c_str(), source_file.c_str(), 0) == 0) { + + return false; + + } + + } + + } + + } else { + + // we could not find out the location. in this case we say it is not + // in the instrument file list + if (!be_quiet) + WARNF( + "No debug information found for function %s, will be " + "instrumented (recompile with -g -O[1-3]).", + F->getName().str().c_str()); + + } + + } + + } + + // if we do not have a instrument file list return true + if (!allowListFiles.empty() || !allowListFunctions.empty()) { + + return_default = false; + + if (!allowListFunctions.empty()) { + + std::string instFunction = F->getName().str(); + + for (std::list::iterator it = allowListFunctions.begin(); + it != allowListFunctions.end(); ++it) { + + /* We don't check for filename equality here because + * filenames might actually be full paths. Instead we + * check that the actual filename ends in the filename + * specified in the list. We also allow UNIX-style pattern + * matching */ + + if (instFunction.length() >= it->length()) { + + if (fnmatch(("*" + *it).c_str(), instFunction.c_str(), 0) == 0) { + + if (debug) + SAYF(cMGN "[D] " cRST + "Function %s is in the allow function list, " + "instrumenting ... \n", + instFunction.c_str()); + return true; + + } + + } + + } + + } + + if (!allowListFiles.empty()) { + + std::string source_file = getSourceName(F); + + if (!source_file.empty()) { + + for (std::list::iterator it = allowListFiles.begin(); + it != allowListFiles.end(); ++it) { + + /* We don't check for filename equality here because + * filenames might actually be full paths. Instead we + * check that the actual filename ends in the filename + * specified in the list. We also allow UNIX-style pattern + * matching */ + + if (source_file.length() >= it->length()) { + + if (fnmatch(("*" + *it).c_str(), source_file.c_str(), 0) == 0) { + + if (debug) + SAYF(cMGN "[D] " cRST + "Function %s is in the allowlist (%s), " + "instrumenting ... \n", + F->getName().str().c_str(), source_file.c_str()); + return true; + + } + + } + + } + + } else { + + // we could not find out the location. In this case we say it is not + // in the instrument file list + if (!be_quiet) + WARNF( + "No debug information found for function %s, will not be " + "instrumented (recompile with -g -O[1-3]).", + F->getName().str().c_str()); + return false; + + } + + } + + } + + return return_default; + +} + +// Calculate the number of average collisions that would occur if all +// location IDs would be assigned randomly (like normal afl/afl++). +// This uses the "balls in bins" algorithm. +unsigned long long int calculateCollisions(uint32_t edges) { + + double bins = MAP_SIZE; + double balls = edges; + double step1 = 1 - (1 / bins); + double step2 = pow(step1, balls); + double step3 = bins * step2; + double step4 = round(step3); + unsigned long long int empty = step4; + unsigned long long int collisions = edges - (MAP_SIZE - empty); + return collisions; + +} + diff --git a/instrumentation/afl-llvm-common.h b/instrumentation/afl-llvm-common.h new file mode 100644 index 00000000..a1561d9c --- /dev/null +++ b/instrumentation/afl-llvm-common.h @@ -0,0 +1,52 @@ +#ifndef __AFLLLVMCOMMON_H +#define __AFLLLVMCOMMON_H + +#include +#include +#include + +#include +#include +#include +#include + +#include "llvm/Config/llvm-config.h" +#if LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR < 5 +typedef long double max_align_t; +#endif + +#include "llvm/IR/IRBuilder.h" +#include "llvm/IR/LegacyPassManager.h" +#include "llvm/IR/BasicBlock.h" +#include "llvm/IR/Module.h" +#include "llvm/Support/Debug.h" +#include "llvm/Support/MathExtras.h" +#include "llvm/Transforms/IPO/PassManagerBuilder.h" + +#if LLVM_VERSION_MAJOR > 3 || \ + (LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR > 4) + #include "llvm/IR/DebugInfo.h" + #include "llvm/IR/CFG.h" +#else + #include "llvm/DebugInfo.h" + #include "llvm/Support/CFG.h" +#endif + +char * getBBName(const llvm::BasicBlock *BB); +bool isIgnoreFunction(const llvm::Function *F); +void initInstrumentList(); +bool isInInstrumentList(llvm::Function *F); +unsigned long long int calculateCollisions(uint32_t edges); +void scanForDangerousFunctions(llvm::Module *M); + +#ifndef IS_EXTERN + #define IS_EXTERN +#endif + +IS_EXTERN int debug; +IS_EXTERN int be_quiet; + +#undef IS_EXTERN + +#endif + diff --git a/instrumentation/afl-llvm-dict2file.so.cc b/instrumentation/afl-llvm-dict2file.so.cc new file mode 100644 index 00000000..e87ecce8 --- /dev/null +++ b/instrumentation/afl-llvm-dict2file.so.cc @@ -0,0 +1,599 @@ +/* + american fuzzy lop++ - LLVM LTO instrumentation pass + ---------------------------------------------------- + + Written by Marc Heuse + + Copyright 2019-2020 AFLplusplus Project. All rights reserved. + + Licensed under the Apache License, Version 2.0 (the "License"); + you may not use this file except in compliance with the License. + You may obtain a copy of the License at: + + http://www.apache.org/licenses/LICENSE-2.0 + + This library is plugged into LLVM when invoking clang through afl-clang-lto. + + */ + +#define AFL_LLVM_PASS + +#include "config.h" +#include "debug.h" + +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include +#include +#include +#include + +#include "llvm/Config/llvm-config.h" +#include "llvm/ADT/Statistic.h" +#include "llvm/IR/IRBuilder.h" +#include "llvm/IR/LegacyPassManager.h" +#include "llvm/IR/BasicBlock.h" +#include "llvm/IR/Module.h" +#include "llvm/IR/DebugInfo.h" +#include "llvm/IR/CFG.h" +#include "llvm/IR/Verifier.h" +#include "llvm/Support/Debug.h" +#include "llvm/Support/raw_ostream.h" +#include "llvm/Transforms/IPO/PassManagerBuilder.h" +#include "llvm/Transforms/Utils/BasicBlockUtils.h" +#include "llvm/Analysis/LoopInfo.h" +#include "llvm/Analysis/MemorySSAUpdater.h" +#include "llvm/Analysis/ValueTracking.h" +#include "llvm/Pass.h" +#include "llvm/IR/Constants.h" + +#include "afl-llvm-common.h" + +#ifndef O_DSYNC + #define O_DSYNC O_SYNC +#endif + +using namespace llvm; + +namespace { + +class AFLdict2filePass : public ModulePass { + + public: + static char ID; + + AFLdict2filePass() : ModulePass(ID) { + + if (getenv("AFL_DEBUG")) debug = 1; + + } + + bool runOnModule(Module &M) override; + +}; + +} // namespace + +void dict2file(int fd, u8 *mem, u32 len) { + + int i, j, binary = 0; + char line[MAX_AUTO_EXTRA * 8], tmp[8]; + + strcpy(line, "\""); + j = 1; + for (i = 0; i < len; i++) { + + if (isprint(mem[i])) { + + line[j++] = mem[i]; + + } else { + + if (i + 1 != len || mem[i] != 0 || binary || len == 4 || len == 8) { + + line[j] = 0; + sprintf(tmp, "\\x%02x", (u8)mem[i]); + strcat(line, tmp); + j = strlen(line); + + } + + binary = 1; + + } + + } + + line[j] = 0; + strcat(line, "\"\n"); + if (write(fd, line, strlen(line)) <= 0) + PFATAL("Could not write to dictionary file"); + fsync(fd); + + if (!be_quiet) fprintf(stderr, "Found dictionary token: %s", line); + +} + +bool AFLdict2filePass::runOnModule(Module &M) { + + DenseMap valueMap; + char * ptr; + int fd, found = 0; + + /* Show a banner */ + setvbuf(stdout, NULL, _IONBF, 0); + + if ((isatty(2) && !getenv("AFL_QUIET")) || debug) { + + SAYF(cCYA "afl-llvm-dict2file" VERSION cRST + " by Marc \"vanHauser\" Heuse \n"); + + } else + + be_quiet = 1; + + scanForDangerousFunctions(&M); + + ptr = getenv("AFL_LLVM_DICT2FILE"); + + if (!ptr || *ptr != '/') + FATAL("AFL_LLVM_DICT2FILE is not set to an absolute path: %s", ptr); + + if ((fd = open(ptr, O_WRONLY | O_APPEND | O_CREAT | O_DSYNC, 0644)) < 0) + PFATAL("Could not open/create %s.", ptr); + + /* Instrument all the things! */ + + for (auto &F : M) { + + if (isIgnoreFunction(&F)) continue; + + /* Some implementation notes. + * + * We try to handle 3 cases: + * - memcmp("foo", arg, 3) <- literal string + * - static char globalvar[] = "foo"; + * memcmp(globalvar, arg, 3) <- global variable + * - char localvar[] = "foo"; + * memcmp(locallvar, arg, 3) <- local variable + * + * The local variable case is the hardest. We can only detect that + * case if there is no reassignment or change in the variable. + * And it might not work across llvm version. + * What we do is hooking the initializer function for local variables + * (llvm.memcpy.p0i8.p0i8.i64) and note the string and the assigned + * variable. And if that variable is then used in a compare function + * we use that noted string. + * This seems not to work for tokens that have a size <= 4 :-( + * + * - if the compared length is smaller than the string length we + * save the full string. This is likely better for fuzzing but + * might be wrong in a few cases depending on optimizers + * + * - not using StringRef because there is a bug in the llvm 11 + * checkout I am using which sometimes points to wrong strings + * + * Over and out. Took me a full day. damn. mh/vh + */ + + for (auto &BB : F) { + + for (auto &IN : BB) { + + CallInst *callInst = nullptr; + CmpInst * cmpInst = nullptr; + + if ((cmpInst = dyn_cast(&IN))) { + + Value * op = cmpInst->getOperand(1); + ConstantInt *ilen = dyn_cast(op); + + if (ilen) { + + u64 val2 = 0, val = ilen->getZExtValue(); + u32 len = 0; + if (val > 0x10000 && val < 0xffffffff) len = 4; + if (val > 0x100000001 && val < 0xffffffffffffffff) len = 8; + + if (len) { + + auto c = cmpInst->getPredicate(); + + switch (c) { + + case CmpInst::FCMP_OGT: // fall through + case CmpInst::FCMP_OLE: // fall through + case CmpInst::ICMP_SLE: // fall through + case CmpInst::ICMP_SGT: + + // signed comparison and it is a negative constant + if ((len == 4 && (val & 80000000)) || + (len == 8 && (val & 8000000000000000))) { + + if ((val & 0xffff) != 1) val2 = val - 1; + break; + + } + + // fall through + + case CmpInst::FCMP_UGT: // fall through + case CmpInst::FCMP_ULE: // fall through + case CmpInst::ICMP_UGT: // fall through + case CmpInst::ICMP_ULE: + if ((val & 0xffff) != 0xfffe) val2 = val + 1; + break; + + case CmpInst::FCMP_OLT: // fall through + case CmpInst::FCMP_OGE: // fall through + case CmpInst::ICMP_SLT: // fall through + case CmpInst::ICMP_SGE: + + // signed comparison and it is a negative constant + if ((len == 4 && (val & 80000000)) || + (len == 8 && (val & 8000000000000000))) { + + if ((val & 0xffff) != 1) val2 = val - 1; + break; + + } + + // fall through + + case CmpInst::FCMP_ULT: // fall through + case CmpInst::FCMP_UGE: // fall through + case CmpInst::ICMP_ULT: // fall through + case CmpInst::ICMP_UGE: + if ((val & 0xffff) != 1) val2 = val - 1; + break; + + default: + val2 = 0; + + } + + dict2file(fd, (u8 *)&val, len); + found++; + if (val2) { + + dict2file(fd, (u8 *)&val2, len); + found++; + + } + + } + + } + + } + + if ((callInst = dyn_cast(&IN))) { + + bool isStrcmp = true; + bool isMemcmp = true; + bool isStrncmp = true; + bool isStrcasecmp = true; + bool isStrncasecmp = true; + bool isIntMemcpy = true; + bool addedNull = false; + size_t optLen = 0; + + Function *Callee = callInst->getCalledFunction(); + if (!Callee) continue; + if (callInst->getCallingConv() != llvm::CallingConv::C) continue; + std::string FuncName = Callee->getName().str(); + isStrcmp &= !FuncName.compare("strcmp"); + isMemcmp &= !FuncName.compare("memcmp"); + isStrncmp &= !FuncName.compare("strncmp"); + isStrcasecmp &= !FuncName.compare("strcasecmp"); + isStrncasecmp &= !FuncName.compare("strncasecmp"); + isIntMemcpy &= !FuncName.compare("llvm.memcpy.p0i8.p0i8.i64"); + + if (!isStrcmp && !isMemcmp && !isStrncmp && !isStrcasecmp && + !isStrncasecmp && !isIntMemcpy) + continue; + + /* Verify the strcmp/memcmp/strncmp/strcasecmp/strncasecmp function + * prototype */ + FunctionType *FT = Callee->getFunctionType(); + + isStrcmp &= + FT->getNumParams() == 2 && FT->getReturnType()->isIntegerTy(32) && + FT->getParamType(0) == FT->getParamType(1) && + FT->getParamType(0) == IntegerType::getInt8PtrTy(M.getContext()); + isStrcasecmp &= + FT->getNumParams() == 2 && FT->getReturnType()->isIntegerTy(32) && + FT->getParamType(0) == FT->getParamType(1) && + FT->getParamType(0) == IntegerType::getInt8PtrTy(M.getContext()); + isMemcmp &= FT->getNumParams() == 3 && + FT->getReturnType()->isIntegerTy(32) && + FT->getParamType(0)->isPointerTy() && + FT->getParamType(1)->isPointerTy() && + FT->getParamType(2)->isIntegerTy(); + isStrncmp &= FT->getNumParams() == 3 && + FT->getReturnType()->isIntegerTy(32) && + FT->getParamType(0) == FT->getParamType(1) && + FT->getParamType(0) == + IntegerType::getInt8PtrTy(M.getContext()) && + FT->getParamType(2)->isIntegerTy(); + isStrncasecmp &= FT->getNumParams() == 3 && + FT->getReturnType()->isIntegerTy(32) && + FT->getParamType(0) == FT->getParamType(1) && + FT->getParamType(0) == + IntegerType::getInt8PtrTy(M.getContext()) && + FT->getParamType(2)->isIntegerTy(); + + if (!isStrcmp && !isMemcmp && !isStrncmp && !isStrcasecmp && + !isStrncasecmp && !isIntMemcpy) + continue; + + /* is a str{n,}{case,}cmp/memcmp, check if we have + * str{case,}cmp(x, "const") or str{case,}cmp("const", x) + * strn{case,}cmp(x, "const", ..) or strn{case,}cmp("const", x, ..) + * memcmp(x, "const", ..) or memcmp("const", x, ..) */ + Value *Str1P = callInst->getArgOperand(0), + *Str2P = callInst->getArgOperand(1); + std::string Str1, Str2; + StringRef TmpStr; + bool HasStr1 = getConstantStringInfo(Str1P, TmpStr); + if (TmpStr.empty()) { + + HasStr1 = false; + + } else { + + HasStr1 = true; + Str1 = TmpStr.str(); + + } + + bool HasStr2 = getConstantStringInfo(Str2P, TmpStr); + if (TmpStr.empty()) { + + HasStr2 = false; + + } else { + + HasStr2 = true; + Str2 = TmpStr.str(); + + } + + if (debug) + fprintf(stderr, "F:%s %p(%s)->\"%s\"(%s) %p(%s)->\"%s\"(%s)\n", + FuncName.c_str(), Str1P, Str1P->getName().str().c_str(), + Str1.c_str(), HasStr1 == true ? "true" : "false", Str2P, + Str2P->getName().str().c_str(), Str2.c_str(), + HasStr2 == true ? "true" : "false"); + + // we handle the 2nd parameter first because of llvm memcpy + if (!HasStr2) { + + auto *Ptr = dyn_cast(Str2P); + if (Ptr && Ptr->isGEPWithNoNotionalOverIndexing()) { + + if (auto *Var = dyn_cast(Ptr->getOperand(0))) { + + if (Var->hasInitializer()) { + + if (auto *Array = + dyn_cast(Var->getInitializer())) { + + HasStr2 = true; + Str2 = Array->getAsString().str(); + + } + + } + + } + + } + + } + + // for the internal memcpy routine we only care for the second + // parameter and are not reporting anything. + if (isIntMemcpy == true) { + + if (HasStr2 == true) { + + Value * op2 = callInst->getArgOperand(2); + ConstantInt *ilen = dyn_cast(op2); + if (ilen) { + + uint64_t literalLength = Str2.size(); + uint64_t optLength = ilen->getZExtValue(); + if (literalLength + 1 == optLength) { + + Str2.append("\0", 1); // add null byte + addedNull = true; + + } + + } + + valueMap[Str1P] = new std::string(Str2); + + if (debug) + fprintf(stderr, "Saved: %s for %p\n", Str2.c_str(), Str1P); + continue; + + } + + continue; + + } + + // Neither a literal nor a global variable? + // maybe it is a local variable that we saved + if (!HasStr2) { + + std::string *strng = valueMap[Str2P]; + if (strng && !strng->empty()) { + + Str2 = *strng; + HasStr2 = true; + if (debug) + fprintf(stderr, "Filled2: %s for %p\n", strng->c_str(), Str2P); + + } + + } + + if (!HasStr1) { + + auto Ptr = dyn_cast(Str1P); + + if (Ptr && Ptr->isGEPWithNoNotionalOverIndexing()) { + + if (auto *Var = dyn_cast(Ptr->getOperand(0))) { + + if (Var->hasInitializer()) { + + if (auto *Array = + dyn_cast(Var->getInitializer())) { + + HasStr1 = true; + Str1 = Array->getAsString().str(); + + } + + } + + } + + } + + } + + // Neither a literal nor a global variable? + // maybe it is a local variable that we saved + if (!HasStr1) { + + std::string *strng = valueMap[Str1P]; + if (strng && !strng->empty()) { + + Str1 = *strng; + HasStr1 = true; + if (debug) + fprintf(stderr, "Filled1: %s for %p\n", strng->c_str(), Str1P); + + } + + } + + /* handle cases of one string is const, one string is variable */ + if (!(HasStr1 ^ HasStr2)) continue; + + std::string thestring; + + if (HasStr1) + thestring = Str1; + else + thestring = Str2; + + optLen = thestring.length(); + + if (isMemcmp || isStrncmp || isStrncasecmp) { + + Value * op2 = callInst->getArgOperand(2); + ConstantInt *ilen = dyn_cast(op2); + if (ilen) { + + uint64_t literalLength = optLen; + optLen = ilen->getZExtValue(); + if (literalLength + 1 == optLen) { // add null byte + thestring.append("\0", 1); + addedNull = true; + + } + + } + + } + + // add null byte if this is a string compare function and a null + // was not already added + if (!isMemcmp) { + + if (addedNull == false) { + + thestring.append("\0", 1); // add null byte + optLen++; + + } + + // ensure we do not have garbage + size_t offset = thestring.find('\0', 0); + if (offset + 1 < optLen) optLen = offset + 1; + thestring = thestring.substr(0, optLen); + + } + + // we take the longer string, even if the compare was to a + // shorter part. Note that depending on the optimizer of the + // compiler this can be wrong, but it is more likely that this + // is helping the fuzzer + if (optLen != thestring.length()) optLen = thestring.length(); + if (optLen > MAX_AUTO_EXTRA) optLen = MAX_AUTO_EXTRA; + if (optLen < 3) // too short? skip + continue; + + ptr = (char *)thestring.c_str(); + + dict2file(fd, (u8 *)ptr, optLen); + found++; + + } + + } + + } + + } + + close(fd); + + /* Say something nice. */ + + if (!be_quiet) { + + if (!found) + OKF("No entries for a dictionary found."); + else + OKF("Wrote %d entries to the dictionary file.\n", found); + + } + + return true; + +} + +char AFLdict2filePass::ID = 0; + +static void registerAFLdict2filePass(const PassManagerBuilder &, + legacy::PassManagerBase &PM) { + + PM.add(new AFLdict2filePass()); + +} + +static RegisterPass X("afl-dict2file", + "afl++ dict2file instrumentation pass", + false, false); + +static RegisterStandardPasses RegisterAFLdict2filePass( + PassManagerBuilder::EP_OptimizerLast, registerAFLdict2filePass); + +static RegisterStandardPasses RegisterAFLdict2filePass0( + PassManagerBuilder::EP_EnabledOnOptLevel0, registerAFLdict2filePass); + diff --git a/instrumentation/afl-llvm-lto-instrumentation.so.cc b/instrumentation/afl-llvm-lto-instrumentation.so.cc new file mode 100644 index 00000000..125db229 --- /dev/null +++ b/instrumentation/afl-llvm-lto-instrumentation.so.cc @@ -0,0 +1,957 @@ +/* + american fuzzy lop++ - LLVM LTO instrumentation pass + ---------------------------------------------------- + + Written by Marc Heuse + + Copyright 2019-2020 AFLplusplus Project. All rights reserved. + + Licensed under the Apache License, Version 2.0 (the "License"); + you may not use this file except in compliance with the License. + You may obtain a copy of the License at: + + http://www.apache.org/licenses/LICENSE-2.0 + + This library is plugged into LLVM when invoking clang through afl-clang-lto. + + */ + +#define AFL_LLVM_PASS + +#include "config.h" +#include "debug.h" + +#include +#include +#include +#include +#include + +#include +#include +#include +#include + +#include "llvm/Config/llvm-config.h" +#include "llvm/ADT/Statistic.h" +#include "llvm/IR/IRBuilder.h" +#include "llvm/IR/LegacyPassManager.h" +#include "llvm/IR/BasicBlock.h" +#include "llvm/IR/Module.h" +#include "llvm/IR/DebugInfo.h" +#include "llvm/IR/CFG.h" +#include "llvm/IR/Verifier.h" +#include "llvm/Support/Debug.h" +#include "llvm/Support/raw_ostream.h" +#include "llvm/Transforms/IPO/PassManagerBuilder.h" +#include "llvm/Transforms/Utils/BasicBlockUtils.h" +#include "llvm/Analysis/LoopInfo.h" +#include "llvm/Analysis/MemorySSAUpdater.h" +#include "llvm/Analysis/ValueTracking.h" +#include "llvm/Pass.h" +#include "llvm/IR/Constants.h" + +#include "afl-llvm-common.h" + +using namespace llvm; + +namespace { + +class AFLLTOPass : public ModulePass { + + public: + static char ID; + + AFLLTOPass() : ModulePass(ID) { + + char *ptr; + + if (getenv("AFL_DEBUG")) debug = 1; + if ((ptr = getenv("AFL_LLVM_LTO_STARTID")) != NULL) + if ((afl_global_id = atoi(ptr)) < 0 || afl_global_id >= MAP_SIZE) + FATAL("AFL_LLVM_LTO_STARTID value of \"%s\" is not between 0 and %d\n", + ptr, MAP_SIZE - 1); + + skip_nozero = getenv("AFL_LLVM_SKIP_NEVERZERO"); + + } + + void getAnalysisUsage(AnalysisUsage &AU) const override { + + ModulePass::getAnalysisUsage(AU); + AU.addRequired(); + AU.addRequired(); + + } + + bool runOnModule(Module &M) override; + + protected: + int afl_global_id = 1, autodictionary = 1; + uint32_t function_minimum_size = 1; + uint32_t inst_blocks = 0, inst_funcs = 0, total_instr = 0; + uint64_t map_addr = 0x10000; + char * skip_nozero = NULL; + +}; + +} // namespace + +bool AFLLTOPass::runOnModule(Module &M) { + + LLVMContext & C = M.getContext(); + std::vector dictionary; + std::vector calls; + DenseMap valueMap; + std::vector BlockList; + char * ptr; + FILE * documentFile = NULL; + + srand((unsigned int)time(NULL)); + + unsigned long long int moduleID = + (((unsigned long long int)(rand() & 0xffffffff)) << 32) | getpid(); + + IntegerType *Int8Ty = IntegerType::getInt8Ty(C); + IntegerType *Int32Ty = IntegerType::getInt32Ty(C); + IntegerType *Int64Ty = IntegerType::getInt64Ty(C); + + /* Show a banner */ + setvbuf(stdout, NULL, _IONBF, 0); + + if ((isatty(2) && !getenv("AFL_QUIET")) || debug) { + + SAYF(cCYA "afl-llvm-lto" VERSION cRST + " by Marc \"vanHauser\" Heuse \n"); + + } else + + be_quiet = 1; + + if ((ptr = getenv("AFL_LLVM_DOCUMENT_IDS")) != NULL) { + + if ((documentFile = fopen(ptr, "a")) == NULL) + WARNF("Cannot access document file %s", ptr); + + } + + // we make this the default as the fixed map has problems with + // defered forkserver, early constructors, ifuncs and maybe more + /*if (getenv("AFL_LLVM_MAP_DYNAMIC"))*/ + map_addr = 0; + + if ((ptr = getenv("AFL_LLVM_MAP_ADDR"))) { + + uint64_t val; + if (!*ptr || !strcmp(ptr, "0") || !strcmp(ptr, "0x0")) { + + map_addr = 0; + + } else if (getenv("AFL_LLVM_MAP_DYNAMIC")) { + + FATAL( + "AFL_LLVM_MAP_ADDR and AFL_LLVM_MAP_DYNAMIC cannot be used together"); + + } else if (strncmp(ptr, "0x", 2) != 0) { + + map_addr = 0x10000; // the default + + } else { + + val = strtoull(ptr, NULL, 16); + if (val < 0x100 || val > 0xffffffff00000000) { + + FATAL( + "AFL_LLVM_MAP_ADDR must be a value between 0x100 and " + "0xffffffff00000000"); + + } + + map_addr = val; + + } + + } + + if (debug) { fprintf(stderr, "map address is 0x%lx\n", map_addr); } + + /* Get/set the globals for the SHM region. */ + + GlobalVariable *AFLMapPtr = NULL; + Value * MapPtrFixed = NULL; + + if (!map_addr) { + + AFLMapPtr = + new GlobalVariable(M, PointerType::get(Int8Ty, 0), false, + GlobalValue::ExternalLinkage, 0, "__afl_area_ptr"); + + } else { + + ConstantInt *MapAddr = ConstantInt::get(Int64Ty, map_addr); + MapPtrFixed = + ConstantExpr::getIntToPtr(MapAddr, PointerType::getUnqual(Int8Ty)); + + } + + ConstantInt *Zero = ConstantInt::get(Int8Ty, 0); + ConstantInt *One = ConstantInt::get(Int8Ty, 1); + + // This dumps all inialized global strings - might be useful in the future + /* + for (auto G=M.getGlobalList().begin(); G!=M.getGlobalList().end(); G++) { + + GlobalVariable &GV=*G; + if (!GV.getName().str().empty()) { + + fprintf(stderr, "Global Variable: %s", GV.getName().str().c_str()); + if (GV.hasInitializer()) + if (auto *Val = dyn_cast(GV.getInitializer())) + fprintf(stderr, " Value: \"%s\"", Val->getAsString().str().c_str()); + fprintf(stderr, "\n"); + + } + + } + + */ + + scanForDangerousFunctions(&M); + + /* Instrument all the things! */ + + int inst_blocks = 0; + + for (auto &F : M) { + + /*For debugging + AttributeSet X = F.getAttributes().getFnAttributes(); + fprintf(stderr, "DEBUG: Module %s Function %s attributes %u\n", + M.getName().str().c_str(), F.getName().str().c_str(), + X.getNumAttributes()); + */ + + if (F.size() < function_minimum_size) continue; + if (isIgnoreFunction(&F)) continue; + + // the instrument file list check + AttributeList Attrs = F.getAttributes(); + if (Attrs.hasAttribute(-1, StringRef("skipinstrument"))) { + + if (debug) + fprintf(stderr, + "DEBUG: Function %s is not in a source file that was specified " + "in the instrument file list\n", + F.getName().str().c_str()); + continue; + + } + + std::vector InsBlocks; + + if (autodictionary) { + + /* Some implementation notes. + * + * We try to handle 3 cases: + * - memcmp("foo", arg, 3) <- literal string + * - static char globalvar[] = "foo"; + * memcmp(globalvar, arg, 3) <- global variable + * - char localvar[] = "foo"; + * memcmp(locallvar, arg, 3) <- local variable + * + * The local variable case is the hardest. We can only detect that + * case if there is no reassignment or change in the variable. + * And it might not work across llvm version. + * What we do is hooking the initializer function for local variables + * (llvm.memcpy.p0i8.p0i8.i64) and note the string and the assigned + * variable. And if that variable is then used in a compare function + * we use that noted string. + * This seems not to work for tokens that have a size <= 4 :-( + * + * - if the compared length is smaller than the string length we + * save the full string. This is likely better for fuzzing but + * might be wrong in a few cases depending on optimizers + * + * - not using StringRef because there is a bug in the llvm 11 + * checkout I am using which sometimes points to wrong strings + * + * Over and out. Took me a full day. damn. mh/vh + */ + + for (auto &BB : F) { + + for (auto &IN : BB) { + + CallInst *callInst = nullptr; + + if ((callInst = dyn_cast(&IN))) { + + bool isStrcmp = true; + bool isMemcmp = true; + bool isStrncmp = true; + bool isStrcasecmp = true; + bool isStrncasecmp = true; + bool isIntMemcpy = true; + bool addedNull = false; + size_t optLen = 0; + + Function *Callee = callInst->getCalledFunction(); + if (!Callee) continue; + if (callInst->getCallingConv() != llvm::CallingConv::C) continue; + std::string FuncName = Callee->getName().str(); + isStrcmp &= !FuncName.compare("strcmp"); + isMemcmp &= !FuncName.compare("memcmp"); + isStrncmp &= !FuncName.compare("strncmp"); + isStrcasecmp &= !FuncName.compare("strcasecmp"); + isStrncasecmp &= !FuncName.compare("strncasecmp"); + isIntMemcpy &= !FuncName.compare("llvm.memcpy.p0i8.p0i8.i64"); + + /* we do something different here, putting this BB and the + successors in a block map */ + if (!FuncName.compare("__afl_persistent_loop")) { + + BlockList.push_back(&BB); + /* + for (succ_iterator SI = succ_begin(&BB), SE = + succ_end(&BB); SI != SE; ++SI) { + + BasicBlock *succ = *SI; + BlockList.push_back(succ); + + } + + */ + + } + + if (!isStrcmp && !isMemcmp && !isStrncmp && !isStrcasecmp && + !isStrncasecmp && !isIntMemcpy) + continue; + + /* Verify the strcmp/memcmp/strncmp/strcasecmp/strncasecmp function + * prototype */ + FunctionType *FT = Callee->getFunctionType(); + + isStrcmp &= FT->getNumParams() == 2 && + FT->getReturnType()->isIntegerTy(32) && + FT->getParamType(0) == FT->getParamType(1) && + FT->getParamType(0) == + IntegerType::getInt8PtrTy(M.getContext()); + isStrcasecmp &= FT->getNumParams() == 2 && + FT->getReturnType()->isIntegerTy(32) && + FT->getParamType(0) == FT->getParamType(1) && + FT->getParamType(0) == + IntegerType::getInt8PtrTy(M.getContext()); + isMemcmp &= FT->getNumParams() == 3 && + FT->getReturnType()->isIntegerTy(32) && + FT->getParamType(0)->isPointerTy() && + FT->getParamType(1)->isPointerTy() && + FT->getParamType(2)->isIntegerTy(); + isStrncmp &= FT->getNumParams() == 3 && + FT->getReturnType()->isIntegerTy(32) && + FT->getParamType(0) == FT->getParamType(1) && + FT->getParamType(0) == + IntegerType::getInt8PtrTy(M.getContext()) && + FT->getParamType(2)->isIntegerTy(); + isStrncasecmp &= FT->getNumParams() == 3 && + FT->getReturnType()->isIntegerTy(32) && + FT->getParamType(0) == FT->getParamType(1) && + FT->getParamType(0) == + IntegerType::getInt8PtrTy(M.getContext()) && + FT->getParamType(2)->isIntegerTy(); + + if (!isStrcmp && !isMemcmp && !isStrncmp && !isStrcasecmp && + !isStrncasecmp && !isIntMemcpy) + continue; + + /* is a str{n,}{case,}cmp/memcmp, check if we have + * str{case,}cmp(x, "const") or str{case,}cmp("const", x) + * strn{case,}cmp(x, "const", ..) or strn{case,}cmp("const", x, ..) + * memcmp(x, "const", ..) or memcmp("const", x, ..) */ + Value *Str1P = callInst->getArgOperand(0), + *Str2P = callInst->getArgOperand(1); + std::string Str1, Str2; + StringRef TmpStr; + bool HasStr1 = getConstantStringInfo(Str1P, TmpStr); + if (TmpStr.empty()) { + + HasStr1 = false; + + } else { + + HasStr1 = true; + Str1 = TmpStr.str(); + + } + + bool HasStr2 = getConstantStringInfo(Str2P, TmpStr); + if (TmpStr.empty()) { + + HasStr2 = false; + + } else { + + HasStr2 = true; + Str2 = TmpStr.str(); + + } + + if (debug) + fprintf(stderr, "F:%s %p(%s)->\"%s\"(%s) %p(%s)->\"%s\"(%s)\n", + FuncName.c_str(), Str1P, Str1P->getName().str().c_str(), + Str1.c_str(), HasStr1 == true ? "true" : "false", Str2P, + Str2P->getName().str().c_str(), Str2.c_str(), + HasStr2 == true ? "true" : "false"); + + // we handle the 2nd parameter first because of llvm memcpy + if (!HasStr2) { + + auto *Ptr = dyn_cast(Str2P); + if (Ptr && Ptr->isGEPWithNoNotionalOverIndexing()) { + + if (auto *Var = dyn_cast(Ptr->getOperand(0))) { + + if (Var->hasInitializer()) { + + if (auto *Array = dyn_cast( + Var->getInitializer())) { + + HasStr2 = true; + Str2 = Array->getAsString().str(); + + } + + } + + } + + } + + } + + // for the internal memcpy routine we only care for the second + // parameter and are not reporting anything. + if (isIntMemcpy == true) { + + if (HasStr2 == true) { + + Value * op2 = callInst->getArgOperand(2); + ConstantInt *ilen = dyn_cast(op2); + if (ilen) { + + uint64_t literalLength = Str2.size(); + uint64_t optLength = ilen->getZExtValue(); + if (literalLength + 1 == optLength) { + + Str2.append("\0", 1); // add null byte + addedNull = true; + + } + + } + + valueMap[Str1P] = new std::string(Str2); + + if (debug) + fprintf(stderr, "Saved: %s for %p\n", Str2.c_str(), Str1P); + continue; + + } + + continue; + + } + + // Neither a literal nor a global variable? + // maybe it is a local variable that we saved + if (!HasStr2) { + + std::string *strng = valueMap[Str2P]; + if (strng && !strng->empty()) { + + Str2 = *strng; + HasStr2 = true; + if (debug) + fprintf(stderr, "Filled2: %s for %p\n", strng->c_str(), + Str2P); + + } + + } + + if (!HasStr1) { + + auto Ptr = dyn_cast(Str1P); + + if (Ptr && Ptr->isGEPWithNoNotionalOverIndexing()) { + + if (auto *Var = dyn_cast(Ptr->getOperand(0))) { + + if (Var->hasInitializer()) { + + if (auto *Array = dyn_cast( + Var->getInitializer())) { + + HasStr1 = true; + Str1 = Array->getAsString().str(); + + } + + } + + } + + } + + } + + // Neither a literal nor a global variable? + // maybe it is a local variable that we saved + if (!HasStr1) { + + std::string *strng = valueMap[Str1P]; + if (strng && !strng->empty()) { + + Str1 = *strng; + HasStr1 = true; + if (debug) + fprintf(stderr, "Filled1: %s for %p\n", strng->c_str(), + Str1P); + + } + + } + + /* handle cases of one string is const, one string is variable */ + if (!(HasStr1 ^ HasStr2)) continue; + + std::string thestring; + + if (HasStr1) + thestring = Str1; + else + thestring = Str2; + + optLen = thestring.length(); + + if (isMemcmp || isStrncmp || isStrncasecmp) { + + Value * op2 = callInst->getArgOperand(2); + ConstantInt *ilen = dyn_cast(op2); + if (ilen) { + + uint64_t literalLength = optLen; + optLen = ilen->getZExtValue(); + if (literalLength + 1 == optLen) { // add null byte + thestring.append("\0", 1); + addedNull = true; + + } + + } + + } + + // add null byte if this is a string compare function and a null + // was not already added + if (!isMemcmp) { + + if (addedNull == false) { + + thestring.append("\0", 1); // add null byte + optLen++; + + } + + // ensure we do not have garbage + size_t offset = thestring.find('\0', 0); + if (offset + 1 < optLen) optLen = offset + 1; + thestring = thestring.substr(0, optLen); + + } + + if (!be_quiet) { + + std::string outstring; + fprintf(stderr, "%s: length %zu/%zu \"", FuncName.c_str(), optLen, + thestring.length()); + for (uint8_t i = 0; i < thestring.length(); i++) { + + uint8_t c = thestring[i]; + if (c <= 32 || c >= 127) + fprintf(stderr, "\\x%02x", c); + else + fprintf(stderr, "%c", c); + + } + + fprintf(stderr, "\"\n"); + + } + + // we take the longer string, even if the compare was to a + // shorter part. Note that depending on the optimizer of the + // compiler this can be wrong, but it is more likely that this + // is helping the fuzzer + if (optLen != thestring.length()) optLen = thestring.length(); + if (optLen > MAX_AUTO_EXTRA) optLen = MAX_AUTO_EXTRA; + if (optLen < MIN_AUTO_EXTRA) // too short? skip + continue; + + dictionary.push_back(thestring.substr(0, optLen)); + + } + + } + + } + + } + + for (auto &BB : F) { + + if (F.size() == 1) { + + InsBlocks.push_back(&BB); + continue; + + } + + uint32_t succ = 0; + for (succ_iterator SI = succ_begin(&BB), SE = succ_end(&BB); SI != SE; + ++SI) + if ((*SI)->size() > 0) succ++; + if (succ < 2) // no need to instrument + continue; + + if (BlockList.size()) { + + int skip = 0; + for (uint32_t k = 0; k < BlockList.size(); k++) { + + if (&BB == BlockList[k]) { + + if (debug) + fprintf(stderr, + "DEBUG: Function %s skipping BB with/after __afl_loop\n", + F.getName().str().c_str()); + skip = 1; + + } + + } + + if (skip) continue; + + } + + InsBlocks.push_back(&BB); + + } + + if (InsBlocks.size() > 0) { + + uint32_t i = InsBlocks.size(); + + do { + + --i; + BasicBlock * newBB = NULL; + BasicBlock * origBB = &(*InsBlocks[i]); + std::vector Successors; + Instruction * TI = origBB->getTerminator(); + uint32_t fs = origBB->getParent()->size(); + uint32_t countto; + + for (succ_iterator SI = succ_begin(origBB), SE = succ_end(origBB); + SI != SE; ++SI) { + + BasicBlock *succ = *SI; + Successors.push_back(succ); + + } + + if (fs == 1) { + + newBB = origBB; + countto = 1; + + } else { + + if (TI == NULL || TI->getNumSuccessors() < 2) continue; + countto = Successors.size(); + + } + + // if (Successors.size() != TI->getNumSuccessors()) + // FATAL("Different successor numbers %lu <-> %u\n", Successors.size(), + // TI->getNumSuccessors()); + + for (uint32_t j = 0; j < countto; j++) { + + if (fs != 1) newBB = llvm::SplitEdge(origBB, Successors[j]); + + if (!newBB) { + + if (!be_quiet) WARNF("Split failed!"); + continue; + + } + + if (documentFile) { + + fprintf(documentFile, "ModuleID=%llu Function=%s edgeID=%u\n", + moduleID, F.getName().str().c_str(), afl_global_id); + + } + + BasicBlock::iterator IP = newBB->getFirstInsertionPt(); + IRBuilder<> IRB(&(*IP)); + + /* Set the ID of the inserted basic block */ + + ConstantInt *CurLoc = ConstantInt::get(Int32Ty, afl_global_id++); + + /* Load SHM pointer */ + + Value *MapPtrIdx; + + if (map_addr) { + + MapPtrIdx = IRB.CreateGEP(MapPtrFixed, CurLoc); + + } else { + + LoadInst *MapPtr = IRB.CreateLoad(AFLMapPtr); + MapPtr->setMetadata(M.getMDKindID("nosanitize"), + MDNode::get(C, None)); + MapPtrIdx = IRB.CreateGEP(MapPtr, CurLoc); + + } + + /* Update bitmap */ + + LoadInst *Counter = IRB.CreateLoad(MapPtrIdx); + Counter->setMetadata(M.getMDKindID("nosanitize"), + MDNode::get(C, None)); + + Value *Incr = IRB.CreateAdd(Counter, One); + + if (skip_nozero == NULL) { + + auto cf = IRB.CreateICmpEQ(Incr, Zero); + auto carry = IRB.CreateZExt(cf, Int8Ty); + Incr = IRB.CreateAdd(Incr, carry); + + } + + IRB.CreateStore(Incr, MapPtrIdx) + ->setMetadata(M.getMDKindID("nosanitize"), MDNode::get(C, None)); + + // done :) + + inst_blocks++; + + } + + } while (i > 0); + + } + + } + + if (documentFile) fclose(documentFile); + documentFile = NULL; + + // save highest location ID to global variable + // do this after each function to fail faster + if (!be_quiet && afl_global_id > MAP_SIZE && + afl_global_id > FS_OPT_MAX_MAPSIZE) { + + uint32_t pow2map = 1, map = afl_global_id; + while ((map = map >> 1)) + pow2map++; + WARNF( + "We have %u blocks to instrument but the map size is only %u. Either " + "edit config.h and set MAP_SIZE_POW2 from %u to %u, then recompile " + "afl-fuzz and llvm_mode and then make this target - or set " + "AFL_MAP_SIZE with at least size %u when running afl-fuzz with this " + "target.", + afl_global_id, MAP_SIZE, MAP_SIZE_POW2, pow2map, afl_global_id); + + } + + if (!getenv("AFL_LLVM_LTO_DONTWRITEID") || dictionary.size() || map_addr) { + + // yes we could create our own function, insert it into ctors ... + // but this would be a pain in the butt ... so we use afl-llvm-rt-lto.o + + Function *f = M.getFunction("__afl_auto_init_globals"); + + if (!f) { + + fprintf(stderr, + "Error: init function could not be found (this should not " + "happen)\n"); + exit(-1); + + } + + BasicBlock *bb = &f->getEntryBlock(); + if (!bb) { + + fprintf(stderr, + "Error: init function does not have an EntryBlock (this should " + "not happen)\n"); + exit(-1); + + } + + BasicBlock::iterator IP = bb->getFirstInsertionPt(); + IRBuilder<> IRB(&(*IP)); + + if (map_addr) { + + GlobalVariable *AFLMapAddrFixed = new GlobalVariable( + M, Int64Ty, true, GlobalValue::ExternalLinkage, 0, "__afl_map_addr"); + ConstantInt *MapAddr = ConstantInt::get(Int64Ty, map_addr); + StoreInst * StoreMapAddr = IRB.CreateStore(MapAddr, AFLMapAddrFixed); + StoreMapAddr->setMetadata(M.getMDKindID("nosanitize"), + MDNode::get(C, None)); + + } + + if (getenv("AFL_LLVM_LTO_DONTWRITEID") == NULL) { + + uint32_t write_loc = afl_global_id; + + if (afl_global_id % 8) write_loc = (((afl_global_id + 8) >> 3) << 3); + + GlobalVariable *AFLFinalLoc = new GlobalVariable( + M, Int32Ty, true, GlobalValue::ExternalLinkage, 0, "__afl_final_loc"); + ConstantInt *const_loc = ConstantInt::get(Int32Ty, write_loc); + StoreInst * StoreFinalLoc = IRB.CreateStore(const_loc, AFLFinalLoc); + StoreFinalLoc->setMetadata(M.getMDKindID("nosanitize"), + MDNode::get(C, None)); + + } + + if (dictionary.size()) { + + size_t memlen = 0, count = 0, offset = 0; + char * ptr; + + for (auto token : dictionary) { + + memlen += token.length(); + count++; + + } + + if (!be_quiet) + printf("AUTODICTIONARY: %lu string%s found\n", count, + count == 1 ? "" : "s"); + + if (count) { + + if ((ptr = (char *)malloc(memlen + count)) == NULL) { + + fprintf(stderr, "Error: malloc for %lu bytes failed!\n", + memlen + count); + exit(-1); + + } + + count = 0; + + for (auto token : dictionary) { + + if (offset + token.length() < 0xfffff0 && count < MAX_AUTO_EXTRAS) { + + ptr[offset++] = (uint8_t)token.length(); + memcpy(ptr + offset, token.c_str(), token.length()); + offset += token.length(); + count++; + + } + + } + + GlobalVariable *AFLDictionaryLen = + new GlobalVariable(M, Int32Ty, false, GlobalValue::ExternalLinkage, + 0, "__afl_dictionary_len"); + ConstantInt *const_len = ConstantInt::get(Int32Ty, offset); + StoreInst *StoreDictLen = IRB.CreateStore(const_len, AFLDictionaryLen); + StoreDictLen->setMetadata(M.getMDKindID("nosanitize"), + MDNode::get(C, None)); + + ArrayType *ArrayTy = ArrayType::get(IntegerType::get(C, 8), offset); + GlobalVariable *AFLInternalDictionary = new GlobalVariable( + M, ArrayTy, true, GlobalValue::ExternalLinkage, + ConstantDataArray::get(C, + *(new ArrayRef((char *)ptr, offset))), + "__afl_internal_dictionary"); + AFLInternalDictionary->setInitializer(ConstantDataArray::get( + C, *(new ArrayRef((char *)ptr, offset)))); + AFLInternalDictionary->setConstant(true); + + GlobalVariable *AFLDictionary = new GlobalVariable( + M, PointerType::get(Int8Ty, 0), false, GlobalValue::ExternalLinkage, + 0, "__afl_dictionary"); + + Value *AFLDictOff = IRB.CreateGEP(AFLInternalDictionary, Zero); + Value *AFLDictPtr = + IRB.CreatePointerCast(AFLDictOff, PointerType::get(Int8Ty, 0)); + StoreInst *StoreDict = IRB.CreateStore(AFLDictPtr, AFLDictionary); + StoreDict->setMetadata(M.getMDKindID("nosanitize"), + MDNode::get(C, None)); + + } + + } + + } + + /* Say something nice. */ + + if (!be_quiet) { + + if (!inst_blocks) + WARNF("No instrumentation targets found."); + else { + + char modeline[100]; + snprintf(modeline, sizeof(modeline), "%s%s%s%s%s", + getenv("AFL_HARDEN") ? "hardened" : "non-hardened", + getenv("AFL_USE_ASAN") ? ", ASAN" : "", + getenv("AFL_USE_MSAN") ? ", MSAN" : "", + getenv("AFL_USE_CFISAN") ? ", CFISAN" : "", + getenv("AFL_USE_UBSAN") ? ", UBSAN" : ""); + OKF("Instrumented %u locations with no collisions (on average %llu " + "collisions would be in afl-gcc/afl-clang-fast) (%s mode).", + inst_blocks, calculateCollisions(inst_blocks), modeline); + + } + + } + + return true; + +} + +char AFLLTOPass::ID = 0; + +static void registerAFLLTOPass(const PassManagerBuilder &, + legacy::PassManagerBase &PM) { + + PM.add(new AFLLTOPass()); + +} + +static RegisterPass X("afl-lto", "afl++ LTO instrumentation pass", + false, false); + +static RegisterStandardPasses RegisterAFLLTOPass( + PassManagerBuilder::EP_FullLinkTimeOptimizationLast, registerAFLLTOPass); + diff --git a/instrumentation/afl-llvm-lto-instrumentlist.so.cc b/instrumentation/afl-llvm-lto-instrumentlist.so.cc new file mode 100644 index 00000000..a7331444 --- /dev/null +++ b/instrumentation/afl-llvm-lto-instrumentlist.so.cc @@ -0,0 +1,147 @@ +/* + american fuzzy lop++ - LLVM-mode instrumentation pass + --------------------------------------------------- + + Written by Laszlo Szekeres and + Michal Zalewski + + LLVM integration design comes from Laszlo Szekeres. C bits copied-and-pasted + from afl-as.c are Michal's fault. + + Copyright 2015, 2016 Google Inc. All rights reserved. + Copyright 2019-2020 AFLplusplus Project. All rights reserved. + + Licensed under the Apache License, Version 2.0 (the "License"); + you may not use this file except in compliance with the License. + You may obtain a copy of the License at: + + http://www.apache.org/licenses/LICENSE-2.0 + + This library is plugged into LLVM when invoking clang through afl-clang-fast. + It tells the compiler to add code roughly equivalent to the bits discussed + in ../afl-as.h. + + */ + +#define AFL_LLVM_PASS + +#include "config.h" +#include "debug.h" + +#include +#include +#include + +#include +#include +#include +#include +#include + +#include "llvm/IR/DebugInfo.h" +#include "llvm/IR/BasicBlock.h" +#include "llvm/IR/IRBuilder.h" +#include "llvm/IR/LegacyPassManager.h" +#include "llvm/IR/Module.h" +#include "llvm/Support/Debug.h" +#include "llvm/Transforms/IPO/PassManagerBuilder.h" +#include "llvm/IR/CFG.h" + +#include "afl-llvm-common.h" + +using namespace llvm; + +namespace { + +class AFLcheckIfInstrument : public ModulePass { + + public: + static char ID; + AFLcheckIfInstrument() : ModulePass(ID) { + + if (getenv("AFL_DEBUG")) debug = 1; + + initInstrumentList(); + + } + + bool runOnModule(Module &M) override; + + // StringRef getPassName() const override { + + // return "American Fuzzy Lop Instrumentation"; + // } + + protected: + std::list myInstrumentList; + +}; + +} // namespace + +char AFLcheckIfInstrument::ID = 0; + +bool AFLcheckIfInstrument::runOnModule(Module &M) { + + /* Show a banner */ + + setvbuf(stdout, NULL, _IONBF, 0); + + if ((isatty(2) && !getenv("AFL_QUIET")) || getenv("AFL_DEBUG") != NULL) { + + SAYF(cCYA "afl-llvm-lto-instrumentlist" VERSION cRST + " by Marc \"vanHauser\" Heuse \n"); + + } else if (getenv("AFL_QUIET")) + + be_quiet = 1; + + for (auto &F : M) { + + if (F.size() < 1) continue; + + // fprintf(stderr, "F:%s\n", F.getName().str().c_str()); + + if (isInInstrumentList(&F)) { + + if (debug) + SAYF(cMGN "[D] " cRST "function %s is in the instrument file list\n", + F.getName().str().c_str()); + + } else { + + if (debug) + SAYF(cMGN "[D] " cRST + "function %s is NOT in the instrument file list\n", + F.getName().str().c_str()); + + auto & Ctx = F.getContext(); + AttributeList Attrs = F.getAttributes(); + AttrBuilder NewAttrs; + NewAttrs.addAttribute("skipinstrument"); + F.setAttributes( + Attrs.addAttributes(Ctx, AttributeList::FunctionIndex, NewAttrs)); + + } + + } + + return true; + +} + +static void registerAFLcheckIfInstrumentpass(const PassManagerBuilder &, + legacy::PassManagerBase &PM) { + + PM.add(new AFLcheckIfInstrument()); + +} + +static RegisterStandardPasses RegisterAFLcheckIfInstrumentpass( + PassManagerBuilder::EP_ModuleOptimizerEarly, + registerAFLcheckIfInstrumentpass); + +static RegisterStandardPasses RegisterAFLcheckIfInstrumentpass0( + PassManagerBuilder::EP_EnabledOnOptLevel0, + registerAFLcheckIfInstrumentpass); + diff --git a/instrumentation/afl-llvm-pass.so.cc b/instrumentation/afl-llvm-pass.so.cc new file mode 100644 index 00000000..8c8c987a --- /dev/null +++ b/instrumentation/afl-llvm-pass.so.cc @@ -0,0 +1,654 @@ +/* + american fuzzy lop++ - LLVM-mode instrumentation pass + --------------------------------------------------- + + Written by Laszlo Szekeres , + Adrian Herrera , + Michal Zalewski + + LLVM integration design comes from Laszlo Szekeres. C bits copied-and-pasted + from afl-as.c are Michal's fault. + + NGRAM previous location coverage comes from Adrian Herrera. + + Copyright 2015, 2016 Google Inc. All rights reserved. + Copyright 2019-2020 AFLplusplus Project. All rights reserved. + + Licensed under the Apache License, Version 2.0 (the "License"); + you may not use this file except in compliance with the License. + You may obtain a copy of the License at: + + http://www.apache.org/licenses/LICENSE-2.0 + + This library is plugged into LLVM when invoking clang through afl-clang-fast. + It tells the compiler to add code roughly equivalent to the bits discussed + in ../afl-as.h. + + */ + +#define AFL_LLVM_PASS + +#include "config.h" +#include "debug.h" +#include +#include +#include + +#include +#include +#include +#include + +#include "llvm/Config/llvm-config.h" +#if LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR < 5 +typedef long double max_align_t; +#endif + +#include "llvm/IR/IRBuilder.h" +#include "llvm/IR/LegacyPassManager.h" +#include "llvm/IR/BasicBlock.h" +#include "llvm/IR/Module.h" +#include "llvm/Support/Debug.h" +#include "llvm/Support/MathExtras.h" +#include "llvm/Transforms/IPO/PassManagerBuilder.h" + +#if LLVM_VERSION_MAJOR > 3 || \ + (LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR > 4) + #include "llvm/IR/DebugInfo.h" + #include "llvm/IR/CFG.h" +#else + #include "llvm/DebugInfo.h" + #include "llvm/Support/CFG.h" +#endif + +#include "afl-llvm-common.h" +#include "llvm-ngram-coverage.h" + +using namespace llvm; + +namespace { + +class AFLCoverage : public ModulePass { + + public: + static char ID; + AFLCoverage() : ModulePass(ID) { + + initInstrumentList(); + + } + + bool runOnModule(Module &M) override; + + protected: + uint32_t ngram_size = 0; + uint32_t map_size = MAP_SIZE; + uint32_t function_minimum_size = 1; + char * ctx_str = NULL, *skip_nozero = NULL; + +}; + +} // namespace + +char AFLCoverage::ID = 0; + +/* needed up to 3.9.0 */ +#if LLVM_VERSION_MAJOR == 3 && \ + (LLVM_VERSION_MINOR < 9 || \ + (LLVM_VERSION_MINOR == 9 && LLVM_VERSION_PATCH < 1)) +uint64_t PowerOf2Ceil(unsigned in) { + + uint64_t in64 = in - 1; + in64 |= (in64 >> 1); + in64 |= (in64 >> 2); + in64 |= (in64 >> 4); + in64 |= (in64 >> 8); + in64 |= (in64 >> 16); + in64 |= (in64 >> 32); + return in64 + 1; + +} + +#endif + +/* #if LLVM_VERSION_STRING >= "4.0.1" */ +#if LLVM_VERSION_MAJOR > 4 || \ + (LLVM_VERSION_MAJOR == 4 && LLVM_VERSION_PATCH >= 1) + #define AFL_HAVE_VECTOR_INTRINSICS 1 +#endif +bool AFLCoverage::runOnModule(Module &M) { + + LLVMContext &C = M.getContext(); + + IntegerType *Int8Ty = IntegerType::getInt8Ty(C); + IntegerType *Int32Ty = IntegerType::getInt32Ty(C); +#ifdef AFL_HAVE_VECTOR_INTRINSICS + IntegerType *IntLocTy = + IntegerType::getIntNTy(C, sizeof(PREV_LOC_T) * CHAR_BIT); +#endif + struct timeval tv; + struct timezone tz; + u32 rand_seed; + unsigned int cur_loc = 0; + + /* Setup random() so we get Actually Random(TM) outputs from AFL_R() */ + gettimeofday(&tv, &tz); + rand_seed = tv.tv_sec ^ tv.tv_usec ^ getpid(); + AFL_SR(rand_seed); + + /* Show a banner */ + + setvbuf(stdout, NULL, _IONBF, 0); + + if (getenv("AFL_DEBUG")) debug = 1; + + if ((isatty(2) && !getenv("AFL_QUIET")) || getenv("AFL_DEBUG") != NULL) { + + SAYF(cCYA "afl-llvm-pass" VERSION cRST + " by and \n"); + + } else + + be_quiet = 1; + + /* + char *ptr; + if ((ptr = getenv("AFL_MAP_SIZE")) || (ptr = getenv("AFL_MAPSIZE"))) { + + map_size = atoi(ptr); + if (map_size < 8 || map_size > (1 << 29)) + FATAL("illegal AFL_MAP_SIZE %u, must be between 2^3 and 2^30", + map_size); if (map_size % 8) map_size = (((map_size >> 3) + 1) << 3); + + } + + */ + + /* Decide instrumentation ratio */ + + char * inst_ratio_str = getenv("AFL_INST_RATIO"); + unsigned int inst_ratio = 100; + + if (inst_ratio_str) { + + if (sscanf(inst_ratio_str, "%u", &inst_ratio) != 1 || !inst_ratio || + inst_ratio > 100) + FATAL("Bad value of AFL_INST_RATIO (must be between 1 and 100)"); + + } + +#if LLVM_VERSION_MAJOR < 9 + char *neverZero_counters_str = getenv("AFL_LLVM_NOT_ZERO"); +#endif + skip_nozero = getenv("AFL_LLVM_SKIP_NEVERZERO"); + + unsigned PrevLocSize = 0; + + char *ngram_size_str = getenv("AFL_LLVM_NGRAM_SIZE"); + if (!ngram_size_str) ngram_size_str = getenv("AFL_NGRAM_SIZE"); + ctx_str = getenv("AFL_LLVM_CTX"); + +#ifdef AFL_HAVE_VECTOR_INTRINSICS + /* Decide previous location vector size (must be a power of two) */ + VectorType *PrevLocTy = NULL; + + if (ngram_size_str) + if (sscanf(ngram_size_str, "%u", &ngram_size) != 1 || ngram_size < 2 || + ngram_size > NGRAM_SIZE_MAX) + FATAL( + "Bad value of AFL_NGRAM_SIZE (must be between 2 and NGRAM_SIZE_MAX " + "(%u))", + NGRAM_SIZE_MAX); + + if (ngram_size == 1) ngram_size = 0; + if (ngram_size) + PrevLocSize = ngram_size - 1; + else +#else + if (ngram_size_str) + #ifndef LLVM_VERSION_PATCH + FATAL( + "Sorry, NGRAM branch coverage is not supported with llvm version " + "%d.%d.%d!", + LLVM_VERSION_MAJOR, LLVM_VERSION_MINOR, 0); + #else + FATAL( + "Sorry, NGRAM branch coverage is not supported with llvm version " + "%d.%d.%d!", + LLVM_VERSION_MAJOR, LLVM_VERSION_MINOR, LLVM_VERSION_PATCH); + #endif +#endif + PrevLocSize = 1; + +#ifdef AFL_HAVE_VECTOR_INTRINSICS + int PrevLocVecSize = PowerOf2Ceil(PrevLocSize); + if (ngram_size) + PrevLocTy = VectorType::get(IntLocTy, PrevLocVecSize + #if LLVM_VERSION_MAJOR >= 12 + , + false + #endif + ); +#endif + + /* Get globals for the SHM region and the previous location. Note that + __afl_prev_loc is thread-local. */ + + GlobalVariable *AFLMapPtr = + new GlobalVariable(M, PointerType::get(Int8Ty, 0), false, + GlobalValue::ExternalLinkage, 0, "__afl_area_ptr"); + GlobalVariable *AFLPrevLoc; + GlobalVariable *AFLContext = NULL; + + if (ctx_str) +#ifdef __ANDROID__ + AFLContext = new GlobalVariable( + M, Int32Ty, false, GlobalValue::ExternalLinkage, 0, "__afl_prev_ctx"); +#else + AFLContext = new GlobalVariable( + M, Int32Ty, false, GlobalValue::ExternalLinkage, 0, "__afl_prev_ctx", 0, + GlobalVariable::GeneralDynamicTLSModel, 0, false); +#endif + +#ifdef AFL_HAVE_VECTOR_INTRINSICS + if (ngram_size) + #ifdef __ANDROID__ + AFLPrevLoc = new GlobalVariable( + M, PrevLocTy, /* isConstant */ false, GlobalValue::ExternalLinkage, + /* Initializer */ nullptr, "__afl_prev_loc"); + #else + AFLPrevLoc = new GlobalVariable( + M, PrevLocTy, /* isConstant */ false, GlobalValue::ExternalLinkage, + /* Initializer */ nullptr, "__afl_prev_loc", + /* InsertBefore */ nullptr, GlobalVariable::GeneralDynamicTLSModel, + /* AddressSpace */ 0, /* IsExternallyInitialized */ false); + #endif + else +#endif +#ifdef __ANDROID__ + AFLPrevLoc = new GlobalVariable( + M, Int32Ty, false, GlobalValue::ExternalLinkage, 0, "__afl_prev_loc"); +#else + AFLPrevLoc = new GlobalVariable( + M, Int32Ty, false, GlobalValue::ExternalLinkage, 0, "__afl_prev_loc", 0, + GlobalVariable::GeneralDynamicTLSModel, 0, false); +#endif + +#ifdef AFL_HAVE_VECTOR_INTRINSICS + /* Create the vector shuffle mask for updating the previous block history. + Note that the first element of the vector will store cur_loc, so just set + it to undef to allow the optimizer to do its thing. */ + + SmallVector PrevLocShuffle = {UndefValue::get(Int32Ty)}; + + for (unsigned I = 0; I < PrevLocSize - 1; ++I) + PrevLocShuffle.push_back(ConstantInt::get(Int32Ty, I)); + + for (int I = PrevLocSize; I < PrevLocVecSize; ++I) + PrevLocShuffle.push_back(ConstantInt::get(Int32Ty, PrevLocSize)); + + Constant *PrevLocShuffleMask = ConstantVector::get(PrevLocShuffle); +#endif + + // other constants we need + ConstantInt *Zero = ConstantInt::get(Int8Ty, 0); + ConstantInt *One = ConstantInt::get(Int8Ty, 1); + + LoadInst *PrevCtx = NULL; // CTX sensitive coverage + + /* Instrument all the things! */ + + int inst_blocks = 0; + scanForDangerousFunctions(&M); + + for (auto &F : M) { + + int has_calls = 0; + if (debug) + fprintf(stderr, "FUNCTION: %s (%zu)\n", F.getName().str().c_str(), + F.size()); + + if (!isInInstrumentList(&F)) continue; + + if (F.size() < function_minimum_size) continue; + + for (auto &BB : F) { + + BasicBlock::iterator IP = BB.getFirstInsertionPt(); + IRBuilder<> IRB(&(*IP)); + + // Context sensitive coverage + if (ctx_str && &BB == &F.getEntryBlock()) { + + // load the context ID of the previous function and write to to a local + // variable on the stack + PrevCtx = IRB.CreateLoad(AFLContext); + PrevCtx->setMetadata(M.getMDKindID("nosanitize"), MDNode::get(C, None)); + + // does the function have calls? and is any of the calls larger than one + // basic block? + for (auto &BB : F) { + + if (has_calls) break; + for (auto &IN : BB) { + + CallInst *callInst = nullptr; + if ((callInst = dyn_cast(&IN))) { + + Function *Callee = callInst->getCalledFunction(); + if (!Callee || Callee->size() < function_minimum_size) + continue; + else { + + has_calls = 1; + break; + + } + + } + + } + + } + + // if yes we store a context ID for this function in the global var + if (has_calls) { + + ConstantInt *NewCtx = ConstantInt::get(Int32Ty, AFL_R(map_size)); + StoreInst * StoreCtx = IRB.CreateStore(NewCtx, AFLContext); + StoreCtx->setMetadata(M.getMDKindID("nosanitize"), + MDNode::get(C, None)); + + } + + } + + if (AFL_R(100) >= inst_ratio) continue; + + /* Make up cur_loc */ + + // cur_loc++; + cur_loc = AFL_R(map_size); + +/* There is a problem with Ubuntu 18.04 and llvm 6.0 (see issue #63). + The inline function successors() is not inlined and also not found at runtime + :-( As I am unable to detect Ubuntu18.04 heree, the next best thing is to + disable this optional optimization for LLVM 6.0.0 and Linux */ +#if !(LLVM_VERSION_MAJOR == 6 && LLVM_VERSION_MINOR == 0) || !defined __linux__ + // only instrument if this basic block is the destination of a previous + // basic block that has multiple successors + // this gets rid of ~5-10% of instrumentations that are unnecessary + // result: a little more speed and less map pollution + int more_than_one = -1; + // fprintf(stderr, "BB %u: ", cur_loc); + for (pred_iterator PI = pred_begin(&BB), E = pred_end(&BB); PI != E; + ++PI) { + + BasicBlock *Pred = *PI; + + int count = 0; + if (more_than_one == -1) more_than_one = 0; + // fprintf(stderr, " %p=>", Pred); + + for (succ_iterator SI = succ_begin(Pred), E = succ_end(Pred); SI != E; + ++SI) { + + BasicBlock *Succ = *SI; + + // if (count > 0) + // fprintf(stderr, "|"); + if (Succ != NULL) count++; + // fprintf(stderr, "%p", Succ); + + } + + if (count > 1) more_than_one = 1; + + } + + // fprintf(stderr, " == %d\n", more_than_one); + if (F.size() > 1 && more_than_one != 1) { + + // in CTX mode we have to restore the original context for the caller - + // she might be calling other functions which need the correct CTX + if (ctx_str && has_calls) { + + Instruction *Inst = BB.getTerminator(); + if (isa(Inst) || isa(Inst)) { + + IRBuilder<> Post_IRB(Inst); + StoreInst * RestoreCtx = Post_IRB.CreateStore(PrevCtx, AFLContext); + RestoreCtx->setMetadata(M.getMDKindID("nosanitize"), + MDNode::get(C, None)); + + } + + } + + continue; + + } + +#endif + + ConstantInt *CurLoc; + +#ifdef AFL_HAVE_VECTOR_INTRINSICS + if (ngram_size) + CurLoc = ConstantInt::get(IntLocTy, cur_loc); + else +#endif + CurLoc = ConstantInt::get(Int32Ty, cur_loc); + + /* Load prev_loc */ + + LoadInst *PrevLoc = IRB.CreateLoad(AFLPrevLoc); + PrevLoc->setMetadata(M.getMDKindID("nosanitize"), MDNode::get(C, None)); + Value *PrevLocTrans; + +#ifdef AFL_HAVE_VECTOR_INTRINSICS + /* "For efficiency, we propose to hash the tuple as a key into the + hit_count map as (prev_block_trans << 1) ^ curr_block_trans, where + prev_block_trans = (block_trans_1 ^ ... ^ block_trans_(n-1)" */ + + if (ngram_size) + PrevLocTrans = + IRB.CreateZExt(IRB.CreateXorReduce(PrevLoc), IRB.getInt32Ty()); + else +#endif + PrevLocTrans = PrevLoc; + + if (ctx_str) + PrevLocTrans = + IRB.CreateZExt(IRB.CreateXor(PrevLocTrans, PrevCtx), Int32Ty); + else + PrevLocTrans = IRB.CreateZExt(PrevLocTrans, IRB.getInt32Ty()); + + /* Load SHM pointer */ + + LoadInst *MapPtr = IRB.CreateLoad(AFLMapPtr); + MapPtr->setMetadata(M.getMDKindID("nosanitize"), MDNode::get(C, None)); + + Value *MapPtrIdx; +#ifdef AFL_HAVE_VECTOR_INTRINSICS + if (ngram_size) + MapPtrIdx = IRB.CreateGEP( + MapPtr, + IRB.CreateZExt( + IRB.CreateXor(PrevLocTrans, IRB.CreateZExt(CurLoc, Int32Ty)), + Int32Ty)); + else +#endif + MapPtrIdx = IRB.CreateGEP(MapPtr, IRB.CreateXor(PrevLocTrans, CurLoc)); + + /* Update bitmap */ + + LoadInst *Counter = IRB.CreateLoad(MapPtrIdx); + Counter->setMetadata(M.getMDKindID("nosanitize"), MDNode::get(C, None)); + + Value *Incr = IRB.CreateAdd(Counter, One); + +#if LLVM_VERSION_MAJOR < 9 + if (neverZero_counters_str != + NULL) { // with llvm 9 we make this the default as the bug in llvm is + // then fixed +#else + if (!skip_nozero) { + +#endif + /* hexcoder: Realize a counter that skips zero during overflow. + * Once this counter reaches its maximum value, it next increments to 1 + * + * Instead of + * Counter + 1 -> Counter + * we inject now this + * Counter + 1 -> {Counter, OverflowFlag} + * Counter + OverflowFlag -> Counter + */ + + auto cf = IRB.CreateICmpEQ(Incr, Zero); + auto carry = IRB.CreateZExt(cf, Int8Ty); + Incr = IRB.CreateAdd(Incr, carry); + + } + + IRB.CreateStore(Incr, MapPtrIdx) + ->setMetadata(M.getMDKindID("nosanitize"), MDNode::get(C, None)); + + /* Update prev_loc history vector (by placing cur_loc at the head of the + vector and shuffle the other elements back by one) */ + + StoreInst *Store; + +#ifdef AFL_HAVE_VECTOR_INTRINSICS + if (ngram_size) { + + Value *ShuffledPrevLoc = IRB.CreateShuffleVector( + PrevLoc, UndefValue::get(PrevLocTy), PrevLocShuffleMask); + Value *UpdatedPrevLoc = IRB.CreateInsertElement( + ShuffledPrevLoc, IRB.CreateLShr(CurLoc, (uint64_t)1), (uint64_t)0); + + Store = IRB.CreateStore(UpdatedPrevLoc, AFLPrevLoc); + Store->setMetadata(M.getMDKindID("nosanitize"), MDNode::get(C, None)); + + } else + +#endif + { + + Store = IRB.CreateStore(ConstantInt::get(Int32Ty, cur_loc >> 1), + AFLPrevLoc); + + } + + // in CTX mode we have to restore the original context for the caller - + // she might be calling other functions which need the correct CTX. + // Currently this is only needed for the Ubuntu clang-6.0 bug + if (ctx_str && has_calls) { + + Instruction *Inst = BB.getTerminator(); + if (isa(Inst) || isa(Inst)) { + + IRBuilder<> Post_IRB(Inst); + StoreInst * RestoreCtx = Post_IRB.CreateStore(PrevCtx, AFLContext); + RestoreCtx->setMetadata(M.getMDKindID("nosanitize"), + MDNode::get(C, None)); + + } + + } + + inst_blocks++; + + } + + } + + /* + // This is currently disabled because we not only need to create/insert a + // function (easy), but also add it as a constructor with an ID < 5 + + if (getenv("AFL_LLVM_DONTWRITEID") == NULL) { + + // yes we could create our own function, insert it into ctors ... + // but this would be a pain in the butt ... so we use afl-llvm-rt.o + + Function *f = ... + + if (!f) { + + fprintf(stderr, + "Error: init function could not be created (this should not + happen)\n"); exit(-1); + + } + + ... constructor for f = 4 + + BasicBlock *bb = &f->getEntryBlock(); + if (!bb) { + + fprintf(stderr, + "Error: init function does not have an EntryBlock (this should + not happen)\n"); exit(-1); + + } + + BasicBlock::iterator IP = bb->getFirstInsertionPt(); + IRBuilder<> IRB(&(*IP)); + + if (map_size <= 0x800000) { + + GlobalVariable *AFLFinalLoc = new GlobalVariable( + M, Int32Ty, true, GlobalValue::ExternalLinkage, 0, + "__afl_final_loc"); + ConstantInt *const_loc = ConstantInt::get(Int32Ty, map_size); + StoreInst * StoreFinalLoc = IRB.CreateStore(const_loc, AFLFinalLoc); + StoreFinalLoc->setMetadata(M.getMDKindID("nosanitize"), + MDNode::get(C, None)); + + } + + } + + */ + + /* Say something nice. */ + + if (!be_quiet) { + + if (!inst_blocks) + WARNF("No instrumentation targets found."); + else { + + char modeline[100]; + snprintf(modeline, sizeof(modeline), "%s%s%s%s%s", + getenv("AFL_HARDEN") ? "hardened" : "non-hardened", + getenv("AFL_USE_ASAN") ? ", ASAN" : "", + getenv("AFL_USE_MSAN") ? ", MSAN" : "", + getenv("AFL_USE_CFISAN") ? ", CFISAN" : "", + getenv("AFL_USE_UBSAN") ? ", UBSAN" : ""); + OKF("Instrumented %u locations (%s mode, ratio %u%%).", inst_blocks, + modeline, inst_ratio); + + } + + } + + return true; + +} + +static void registerAFLPass(const PassManagerBuilder &, + legacy::PassManagerBase &PM) { + + PM.add(new AFLCoverage()); + +} + +static RegisterStandardPasses RegisterAFLPass( + PassManagerBuilder::EP_OptimizerLast, registerAFLPass); + +static RegisterStandardPasses RegisterAFLPass0( + PassManagerBuilder::EP_EnabledOnOptLevel0, registerAFLPass); + diff --git a/instrumentation/afl-llvm-rt-lto.o.c b/instrumentation/afl-llvm-rt-lto.o.c new file mode 100644 index 00000000..e53785ff --- /dev/null +++ b/instrumentation/afl-llvm-rt-lto.o.c @@ -0,0 +1,27 @@ +/* + american fuzzy lop++ - LLVM instrumentation bootstrap + ----------------------------------------------------- + + Licensed under the Apache License, Version 2.0 (the "License"); + you may not use this file except in compliance with the License. + You may obtain a copy of the License at: + + http://www.apache.org/licenses/LICENSE-2.0 + +*/ + +#include +#include + +// to prevent the function from being removed +unsigned char __afl_lto_mode = 0; + +/* Proper initialization routine. */ + +__attribute__((constructor(0))) void __afl_auto_init_globals(void) { + + if (getenv("AFL_DEBUG")) fprintf(stderr, "[__afl_auto_init_globals]\n"); + __afl_lto_mode = 1; + +} + diff --git a/instrumentation/cmplog-instructions-pass.cc b/instrumentation/cmplog-instructions-pass.cc new file mode 100644 index 00000000..d5de3dbb --- /dev/null +++ b/instrumentation/cmplog-instructions-pass.cc @@ -0,0 +1,292 @@ +/* + american fuzzy lop++ - LLVM CmpLog instrumentation + -------------------------------------------------- + + Written by Andrea Fioraldi + + Copyright 2015, 2016 Google Inc. All rights reserved. + Copyright 2019-2020 AFLplusplus Project. All rights reserved. + + Licensed under the Apache License, Version 2.0 (the "License"); + you may not use this file except in compliance with the License. + You may obtain a copy of the License at: + + http://www.apache.org/licenses/LICENSE-2.0 + +*/ + +#include +#include +#include + +#include +#include +#include +#include +#include "llvm/Config/llvm-config.h" + +#include "llvm/ADT/Statistic.h" +#include "llvm/IR/IRBuilder.h" +#include "llvm/IR/LegacyPassManager.h" +#include "llvm/IR/Module.h" +#include "llvm/Support/Debug.h" +#include "llvm/Support/raw_ostream.h" +#include "llvm/Transforms/IPO/PassManagerBuilder.h" +#include "llvm/Transforms/Utils/BasicBlockUtils.h" +#include "llvm/Pass.h" +#include "llvm/Analysis/ValueTracking.h" + +#if LLVM_VERSION_MAJOR > 3 || \ + (LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR > 4) + #include "llvm/IR/Verifier.h" + #include "llvm/IR/DebugInfo.h" +#else + #include "llvm/Analysis/Verifier.h" + #include "llvm/DebugInfo.h" + #define nullptr 0 +#endif + +#include +#include "afl-llvm-common.h" + +using namespace llvm; + +namespace { + +class CmpLogInstructions : public ModulePass { + + public: + static char ID; + CmpLogInstructions() : ModulePass(ID) { + + initInstrumentList(); + + } + + bool runOnModule(Module &M) override; + +#if LLVM_VERSION_MAJOR < 4 + const char *getPassName() const override { + +#else + StringRef getPassName() const override { + +#endif + return "cmplog instructions"; + + } + + private: + bool hookInstrs(Module &M); + +}; + +} // namespace + +char CmpLogInstructions::ID = 0; + +bool CmpLogInstructions::hookInstrs(Module &M) { + + std::vector icomps; + LLVMContext & C = M.getContext(); + + Type * VoidTy = Type::getVoidTy(C); + IntegerType *Int8Ty = IntegerType::getInt8Ty(C); + IntegerType *Int16Ty = IntegerType::getInt16Ty(C); + IntegerType *Int32Ty = IntegerType::getInt32Ty(C); + IntegerType *Int64Ty = IntegerType::getInt64Ty(C); + +#if LLVM_VERSION_MAJOR < 9 + Constant * +#else + FunctionCallee +#endif + c1 = M.getOrInsertFunction("__cmplog_ins_hook1", VoidTy, Int8Ty, Int8Ty +#if LLVM_VERSION_MAJOR < 5 + , + NULL +#endif + ); +#if LLVM_VERSION_MAJOR < 9 + Function *cmplogHookIns1 = cast(c1); +#else + FunctionCallee cmplogHookIns1 = c1; +#endif + +#if LLVM_VERSION_MAJOR < 9 + Constant * +#else + FunctionCallee +#endif + c2 = M.getOrInsertFunction("__cmplog_ins_hook2", VoidTy, Int16Ty, Int16Ty +#if LLVM_VERSION_MAJOR < 5 + , + NULL +#endif + ); +#if LLVM_VERSION_MAJOR < 9 + Function *cmplogHookIns2 = cast(c2); +#else + FunctionCallee cmplogHookIns2 = c2; +#endif + +#if LLVM_VERSION_MAJOR < 9 + Constant * +#else + FunctionCallee +#endif + c4 = M.getOrInsertFunction("__cmplog_ins_hook4", VoidTy, Int32Ty, Int32Ty +#if LLVM_VERSION_MAJOR < 5 + , + NULL +#endif + ); +#if LLVM_VERSION_MAJOR < 9 + Function *cmplogHookIns4 = cast(c4); +#else + FunctionCallee cmplogHookIns4 = c4; +#endif + +#if LLVM_VERSION_MAJOR < 9 + Constant * +#else + FunctionCallee +#endif + c8 = M.getOrInsertFunction("__cmplog_ins_hook8", VoidTy, Int64Ty, Int64Ty +#if LLVM_VERSION_MAJOR < 5 + , + NULL +#endif + ); +#if LLVM_VERSION_MAJOR < 9 + Function *cmplogHookIns8 = cast(c8); +#else + FunctionCallee cmplogHookIns8 = c8; +#endif + + /* iterate over all functions, bbs and instruction and add suitable calls */ + for (auto &F : M) { + + if (!isInInstrumentList(&F)) continue; + + for (auto &BB : F) { + + for (auto &IN : BB) { + + CmpInst *selectcmpInst = nullptr; + + if ((selectcmpInst = dyn_cast(&IN))) { + + if (selectcmpInst->getPredicate() == CmpInst::ICMP_EQ || + selectcmpInst->getPredicate() == CmpInst::ICMP_NE || + selectcmpInst->getPredicate() == CmpInst::ICMP_UGT || + selectcmpInst->getPredicate() == CmpInst::ICMP_SGT || + selectcmpInst->getPredicate() == CmpInst::ICMP_ULT || + selectcmpInst->getPredicate() == CmpInst::ICMP_SLT || + selectcmpInst->getPredicate() == CmpInst::ICMP_UGE || + selectcmpInst->getPredicate() == CmpInst::ICMP_SGE || + selectcmpInst->getPredicate() == CmpInst::ICMP_ULE || + selectcmpInst->getPredicate() == CmpInst::ICMP_SLE) { + + auto op0 = selectcmpInst->getOperand(0); + auto op1 = selectcmpInst->getOperand(1); + + IntegerType *intTyOp0 = dyn_cast(op0->getType()); + IntegerType *intTyOp1 = dyn_cast(op1->getType()); + + /* this is probably not needed but we do it anyway */ + if (!intTyOp0 || !intTyOp1) { continue; } + + icomps.push_back(selectcmpInst); + + } + + } + + } + + } + + } + + if (!icomps.size()) return false; + if (!be_quiet) errs() << "Hooking " << icomps.size() << " cmp instructions\n"; + + for (auto &selectcmpInst : icomps) { + + IRBuilder<> IRB(selectcmpInst->getParent()); + IRB.SetInsertPoint(selectcmpInst); + + auto op0 = selectcmpInst->getOperand(0); + auto op1 = selectcmpInst->getOperand(1); + + IntegerType *intTyOp0 = dyn_cast(op0->getType()); + IntegerType *intTyOp1 = dyn_cast(op1->getType()); + + unsigned max_size = intTyOp0->getBitWidth() > intTyOp1->getBitWidth() + ? intTyOp0->getBitWidth() + : intTyOp1->getBitWidth(); + + std::vector args; + args.push_back(op0); + args.push_back(op1); + + switch (max_size) { + + case 8: + IRB.CreateCall(cmplogHookIns1, args); + break; + case 16: + IRB.CreateCall(cmplogHookIns2, args); + break; + case 32: + IRB.CreateCall(cmplogHookIns4, args); + break; + case 64: + IRB.CreateCall(cmplogHookIns8, args); + break; + default: + break; + + } + + } + + return true; + +} + +bool CmpLogInstructions::runOnModule(Module &M) { + + if (getenv("AFL_QUIET") == NULL) + llvm::errs() + << "Running cmplog-instructions-pass by andreafioraldi@gmail.com\n"; + else + be_quiet = 1; + hookInstrs(M); + verifyModule(M); + + return true; + +} + +static void registerCmpLogInstructionsPass(const PassManagerBuilder &, + legacy::PassManagerBase &PM) { + + auto p = new CmpLogInstructions(); + PM.add(p); + +} + +static RegisterStandardPasses RegisterCmpLogInstructionsPass( + PassManagerBuilder::EP_OptimizerLast, registerCmpLogInstructionsPass); + +static RegisterStandardPasses RegisterCmpLogInstructionsPass0( + PassManagerBuilder::EP_EnabledOnOptLevel0, registerCmpLogInstructionsPass); + +#if LLVM_VERSION_MAJOR >= 11 +static RegisterStandardPasses RegisterCmpLogInstructionsPassLTO( + PassManagerBuilder::EP_FullLinkTimeOptimizationLast, + registerCmpLogInstructionsPass); +#endif + diff --git a/instrumentation/cmplog-routines-pass.cc b/instrumentation/cmplog-routines-pass.cc new file mode 100644 index 00000000..c44f38c4 --- /dev/null +++ b/instrumentation/cmplog-routines-pass.cc @@ -0,0 +1,212 @@ +/* + american fuzzy lop++ - LLVM CmpLog instrumentation + -------------------------------------------------- + + Written by Andrea Fioraldi + + Copyright 2015, 2016 Google Inc. All rights reserved. + Copyright 2019-2020 AFLplusplus Project. All rights reserved. + + Licensed under the Apache License, Version 2.0 (the "License"); + you may not use this file except in compliance with the License. + You may obtain a copy of the License at: + + http://www.apache.org/licenses/LICENSE-2.0 + +*/ + +#include +#include +#include + +#include +#include +#include +#include +#include "llvm/Config/llvm-config.h" + +#include "llvm/ADT/Statistic.h" +#include "llvm/IR/IRBuilder.h" +#include "llvm/IR/LegacyPassManager.h" +#include "llvm/IR/Module.h" +#include "llvm/Support/Debug.h" +#include "llvm/Support/raw_ostream.h" +#include "llvm/Transforms/IPO/PassManagerBuilder.h" +#include "llvm/Transforms/Utils/BasicBlockUtils.h" +#include "llvm/Pass.h" +#include "llvm/Analysis/ValueTracking.h" + +#if LLVM_VERSION_MAJOR > 3 || \ + (LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR > 4) + #include "llvm/IR/Verifier.h" + #include "llvm/IR/DebugInfo.h" +#else + #include "llvm/Analysis/Verifier.h" + #include "llvm/DebugInfo.h" + #define nullptr 0 +#endif + +#include +#include "afl-llvm-common.h" + +using namespace llvm; + +namespace { + +class CmpLogRoutines : public ModulePass { + + public: + static char ID; + CmpLogRoutines() : ModulePass(ID) { + + initInstrumentList(); + + } + + bool runOnModule(Module &M) override; + +#if LLVM_VERSION_MAJOR < 4 + const char *getPassName() const override { + +#else + StringRef getPassName() const override { + +#endif + return "cmplog routines"; + + } + + private: + bool hookRtns(Module &M); + +}; + +} // namespace + +char CmpLogRoutines::ID = 0; + +bool CmpLogRoutines::hookRtns(Module &M) { + + std::vector calls; + LLVMContext & C = M.getContext(); + + Type *VoidTy = Type::getVoidTy(C); + // PointerType *VoidPtrTy = PointerType::get(VoidTy, 0); + IntegerType *Int8Ty = IntegerType::getInt8Ty(C); + PointerType *i8PtrTy = PointerType::get(Int8Ty, 0); + +#if LLVM_VERSION_MAJOR < 9 + Constant * +#else + FunctionCallee +#endif + c = M.getOrInsertFunction("__cmplog_rtn_hook", VoidTy, i8PtrTy, i8PtrTy +#if LLVM_VERSION_MAJOR < 5 + , + NULL +#endif + ); +#if LLVM_VERSION_MAJOR < 9 + Function *cmplogHookFn = cast(c); +#else + FunctionCallee cmplogHookFn = c; +#endif + + /* iterate over all functions, bbs and instruction and add suitable calls */ + for (auto &F : M) { + + if (!isInInstrumentList(&F)) continue; + + for (auto &BB : F) { + + for (auto &IN : BB) { + + CallInst *callInst = nullptr; + + if ((callInst = dyn_cast(&IN))) { + + Function *Callee = callInst->getCalledFunction(); + if (!Callee) continue; + if (callInst->getCallingConv() != llvm::CallingConv::C) continue; + + FunctionType *FT = Callee->getFunctionType(); + + bool isPtrRtn = FT->getNumParams() >= 2 && + !FT->getReturnType()->isVoidTy() && + FT->getParamType(0) == FT->getParamType(1) && + FT->getParamType(0)->isPointerTy(); + + if (!isPtrRtn) continue; + + calls.push_back(callInst); + + } + + } + + } + + } + + if (!calls.size()) return false; + if (!be_quiet) + errs() << "Hooking " << calls.size() + << " calls with pointers as arguments\n"; + + for (auto &callInst : calls) { + + Value *v1P = callInst->getArgOperand(0), *v2P = callInst->getArgOperand(1); + + IRBuilder<> IRB(callInst->getParent()); + IRB.SetInsertPoint(callInst); + + std::vector args; + Value * v1Pcasted = IRB.CreatePointerCast(v1P, i8PtrTy); + Value * v2Pcasted = IRB.CreatePointerCast(v2P, i8PtrTy); + args.push_back(v1Pcasted); + args.push_back(v2Pcasted); + + IRB.CreateCall(cmplogHookFn, args); + + // errs() << callInst->getCalledFunction()->getName() << "\n"; + + } + + return true; + +} + +bool CmpLogRoutines::runOnModule(Module &M) { + + if (getenv("AFL_QUIET") == NULL) + llvm::errs() + << "Running cmplog-routines-pass by andreafioraldi@gmail.com\n"; + else + be_quiet = 1; + hookRtns(M); + verifyModule(M); + + return true; + +} + +static void registerCmpLogRoutinesPass(const PassManagerBuilder &, + legacy::PassManagerBase &PM) { + + auto p = new CmpLogRoutines(); + PM.add(p); + +} + +static RegisterStandardPasses RegisterCmpLogRoutinesPass( + PassManagerBuilder::EP_OptimizerLast, registerCmpLogRoutinesPass); + +static RegisterStandardPasses RegisterCmpLogRoutinesPass0( + PassManagerBuilder::EP_EnabledOnOptLevel0, registerCmpLogRoutinesPass); + +#if LLVM_VERSION_MAJOR >= 11 +static RegisterStandardPasses RegisterCmpLogRoutinesPassLTO( + PassManagerBuilder::EP_FullLinkTimeOptimizationLast, + registerCmpLogRoutinesPass); +#endif + diff --git a/instrumentation/compare-transform-pass.so.cc b/instrumentation/compare-transform-pass.so.cc new file mode 100644 index 00000000..acdd0f3b --- /dev/null +++ b/instrumentation/compare-transform-pass.so.cc @@ -0,0 +1,587 @@ +/* + * Copyright 2016 laf-intel + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#include +#include +#include + +#include +#include +#include +#include +#include "llvm/Config/llvm-config.h" + +#include "llvm/ADT/Statistic.h" +#include "llvm/IR/IRBuilder.h" +#include "llvm/IR/LegacyPassManager.h" +#include "llvm/IR/Module.h" +#include "llvm/Support/Debug.h" +#include "llvm/Support/raw_ostream.h" +#include "llvm/Transforms/IPO/PassManagerBuilder.h" +#include "llvm/Transforms/Utils/BasicBlockUtils.h" +#include "llvm/Pass.h" +#include "llvm/Analysis/ValueTracking.h" + +#if LLVM_VERSION_MAJOR > 3 || \ + (LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR > 4) + #include "llvm/IR/Verifier.h" + #include "llvm/IR/DebugInfo.h" +#else + #include "llvm/Analysis/Verifier.h" + #include "llvm/DebugInfo.h" + #define nullptr 0 +#endif + +#include +#include "afl-llvm-common.h" + +using namespace llvm; + +namespace { + +class CompareTransform : public ModulePass { + + public: + static char ID; + CompareTransform() : ModulePass(ID) { + + initInstrumentList(); + + } + + bool runOnModule(Module &M) override; + +#if LLVM_VERSION_MAJOR < 4 + const char *getPassName() const override { + +#else + StringRef getPassName() const override { + +#endif + return "transforms compare functions"; + + } + + private: + bool transformCmps(Module &M, const bool processStrcmp, + const bool processMemcmp, const bool processStrncmp, + const bool processStrcasecmp, + const bool processStrncasecmp); + +}; + +} // namespace + +char CompareTransform::ID = 0; + +bool CompareTransform::transformCmps(Module &M, const bool processStrcmp, + const bool processMemcmp, + const bool processStrncmp, + const bool processStrcasecmp, + const bool processStrncasecmp) { + + DenseMap valueMap; + std::vector calls; + LLVMContext & C = M.getContext(); + IntegerType * Int8Ty = IntegerType::getInt8Ty(C); + IntegerType * Int32Ty = IntegerType::getInt32Ty(C); + IntegerType * Int64Ty = IntegerType::getInt64Ty(C); + +#if LLVM_VERSION_MAJOR < 9 + Constant * +#else + FunctionCallee +#endif + c = M.getOrInsertFunction("tolower", Int32Ty, Int32Ty +#if LLVM_VERSION_MAJOR < 5 + , + NULL +#endif + ); +#if LLVM_VERSION_MAJOR < 9 + Function *tolowerFn = cast(c); +#else + FunctionCallee tolowerFn = c; +#endif + + /* iterate over all functions, bbs and instruction and add suitable calls to + * strcmp/memcmp/strncmp/strcasecmp/strncasecmp */ + for (auto &F : M) { + + if (!isInInstrumentList(&F)) continue; + + for (auto &BB : F) { + + for (auto &IN : BB) { + + CallInst *callInst = nullptr; + + if ((callInst = dyn_cast(&IN))) { + + bool isStrcmp = processStrcmp; + bool isMemcmp = processMemcmp; + bool isStrncmp = processStrncmp; + bool isStrcasecmp = processStrcasecmp; + bool isStrncasecmp = processStrncasecmp; + bool isIntMemcpy = true; + + Function *Callee = callInst->getCalledFunction(); + if (!Callee) continue; + if (callInst->getCallingConv() != llvm::CallingConv::C) continue; + StringRef FuncName = Callee->getName(); + isStrcmp &= !FuncName.compare(StringRef("strcmp")); + isMemcmp &= !FuncName.compare(StringRef("memcmp")); + isStrncmp &= !FuncName.compare(StringRef("strncmp")); + isStrcasecmp &= !FuncName.compare(StringRef("strcasecmp")); + isStrncasecmp &= !FuncName.compare(StringRef("strncasecmp")); + isIntMemcpy &= !FuncName.compare("llvm.memcpy.p0i8.p0i8.i64"); + + if (!isStrcmp && !isMemcmp && !isStrncmp && !isStrcasecmp && + !isStrncasecmp && !isIntMemcpy) + continue; + + /* Verify the strcmp/memcmp/strncmp/strcasecmp/strncasecmp function + * prototype */ + FunctionType *FT = Callee->getFunctionType(); + + isStrcmp &= + FT->getNumParams() == 2 && FT->getReturnType()->isIntegerTy(32) && + FT->getParamType(0) == FT->getParamType(1) && + FT->getParamType(0) == IntegerType::getInt8PtrTy(M.getContext()); + isStrcasecmp &= + FT->getNumParams() == 2 && FT->getReturnType()->isIntegerTy(32) && + FT->getParamType(0) == FT->getParamType(1) && + FT->getParamType(0) == IntegerType::getInt8PtrTy(M.getContext()); + isMemcmp &= FT->getNumParams() == 3 && + FT->getReturnType()->isIntegerTy(32) && + FT->getParamType(0)->isPointerTy() && + FT->getParamType(1)->isPointerTy() && + FT->getParamType(2)->isIntegerTy(); + isStrncmp &= FT->getNumParams() == 3 && + FT->getReturnType()->isIntegerTy(32) && + FT->getParamType(0) == FT->getParamType(1) && + FT->getParamType(0) == + IntegerType::getInt8PtrTy(M.getContext()) && + FT->getParamType(2)->isIntegerTy(); + isStrncasecmp &= FT->getNumParams() == 3 && + FT->getReturnType()->isIntegerTy(32) && + FT->getParamType(0) == FT->getParamType(1) && + FT->getParamType(0) == + IntegerType::getInt8PtrTy(M.getContext()) && + FT->getParamType(2)->isIntegerTy(); + + if (!isStrcmp && !isMemcmp && !isStrncmp && !isStrcasecmp && + !isStrncasecmp && !isIntMemcpy) + continue; + + /* is a str{n,}{case,}cmp/memcmp, check if we have + * str{case,}cmp(x, "const") or str{case,}cmp("const", x) + * strn{case,}cmp(x, "const", ..) or strn{case,}cmp("const", x, ..) + * memcmp(x, "const", ..) or memcmp("const", x, ..) */ + Value *Str1P = callInst->getArgOperand(0), + *Str2P = callInst->getArgOperand(1); + StringRef Str1, Str2; + bool HasStr1 = getConstantStringInfo(Str1P, Str1); + bool HasStr2 = getConstantStringInfo(Str2P, Str2); + + if (isIntMemcpy && HasStr2) { + + valueMap[Str1P] = new std::string(Str2.str()); + // fprintf(stderr, "saved %s for %p\n", Str2.str().c_str(), Str1P); + continue; + + } + + // not literal? maybe global or local variable + if (!(HasStr1 || HasStr2)) { + + auto *Ptr = dyn_cast(Str2P); + if (Ptr && Ptr->isGEPWithNoNotionalOverIndexing()) { + + if (auto *Var = dyn_cast(Ptr->getOperand(0))) { + + if (Var->hasInitializer()) { + + if (auto *Array = + dyn_cast(Var->getInitializer())) { + + HasStr2 = true; + Str2 = Array->getAsString(); + valueMap[Str2P] = new std::string(Str2.str()); + fprintf(stderr, "glo2 %s\n", Str2.str().c_str()); + + } + + } + + } + + } + + if (!HasStr2) { + + auto *Ptr = dyn_cast(Str1P); + if (Ptr && Ptr->isGEPWithNoNotionalOverIndexing()) { + + if (auto *Var = dyn_cast(Ptr->getOperand(0))) { + + if (Var->hasInitializer()) { + + if (auto *Array = dyn_cast( + Var->getInitializer())) { + + HasStr1 = true; + Str1 = Array->getAsString(); + valueMap[Str1P] = new std::string(Str1.str()); + // fprintf(stderr, "glo1 %s\n", Str1.str().c_str()); + + } + + } + + } + + } + + } else if (isIntMemcpy) { + + valueMap[Str1P] = new std::string(Str2.str()); + // fprintf(stderr, "saved\n"); + + } + + } + + if (isIntMemcpy) continue; + + if (!(HasStr1 || HasStr2)) { + + // do we have a saved local variable initialization? + std::string *val = valueMap[Str1P]; + if (val && !val->empty()) { + + Str1 = StringRef(*val); + HasStr1 = true; + // fprintf(stderr, "loaded1 %s\n", Str1.str().c_str()); + + } else { + + val = valueMap[Str2P]; + if (val && !val->empty()) { + + Str2 = StringRef(*val); + HasStr2 = true; + // fprintf(stderr, "loaded2 %s\n", Str2.str().c_str()); + + } + + } + + } + + /* handle cases of one string is const, one string is variable */ + if (!(HasStr1 || HasStr2)) continue; + + if (isMemcmp || isStrncmp || isStrncasecmp) { + + /* check if third operand is a constant integer + * strlen("constStr") and sizeof() are treated as constant */ + Value * op2 = callInst->getArgOperand(2); + ConstantInt *ilen = dyn_cast(op2); + if (ilen) { + + uint64_t len = ilen->getZExtValue(); + // if len is zero this is a pointless call but allow real + // implementation to worry about that + if (!len) continue; + + if (isMemcmp) { + + // if size of compare is larger than constant string this is + // likely a bug but allow real implementation to worry about + // that + uint64_t literalLength = HasStr1 ? Str1.size() : Str2.size(); + if (literalLength + 1 < ilen->getZExtValue()) continue; + + } + + } else if (isMemcmp) + + // this *may* supply a len greater than the constant string at + // runtime so similarly we don't want to have to handle that + continue; + + } + + calls.push_back(callInst); + + } + + } + + } + + } + + if (!calls.size()) return false; + if (!be_quiet) + errs() << "Replacing " << calls.size() + << " calls to strcmp/memcmp/strncmp/strcasecmp/strncasecmp\n"; + + for (auto &callInst : calls) { + + Value *Str1P = callInst->getArgOperand(0), + *Str2P = callInst->getArgOperand(1); + StringRef Str1, Str2, ConstStr; + std::string TmpConstStr; + Value * VarStr; + bool HasStr1 = getConstantStringInfo(Str1P, Str1); + bool HasStr2 = getConstantStringInfo(Str2P, Str2); + uint64_t constStrLen, unrollLen, constSizedLen = 0; + bool isMemcmp = + !callInst->getCalledFunction()->getName().compare(StringRef("memcmp")); + bool isSizedcmp = isMemcmp || + !callInst->getCalledFunction()->getName().compare( + StringRef("strncmp")) || + !callInst->getCalledFunction()->getName().compare( + StringRef("strncasecmp")); + Value *sizedValue = isSizedcmp ? callInst->getArgOperand(2) : NULL; + bool isConstSized = sizedValue && isa(sizedValue); + bool isCaseInsensitive = !callInst->getCalledFunction()->getName().compare( + StringRef("strcasecmp")) || + !callInst->getCalledFunction()->getName().compare( + StringRef("strncasecmp")); + + if (!(HasStr1 || HasStr2)) { + + // do we have a saved local or global variable initialization? + std::string *val = valueMap[Str1P]; + if (val && !val->empty()) { + + Str1 = StringRef(*val); + HasStr1 = true; + + } else { + + val = valueMap[Str2P]; + if (val && !val->empty()) { + + Str2 = StringRef(*val); + HasStr2 = true; + + } + + } + + } + + if (isConstSized) { + + constSizedLen = dyn_cast(sizedValue)->getZExtValue(); + + } + + if (HasStr1) { + + TmpConstStr = Str1.str(); + VarStr = Str2P; + + } else { + + TmpConstStr = Str2.str(); + VarStr = Str1P; + + } + + // add null termination character implicit in c strings + TmpConstStr.append("\0", 1); + + // in the unusual case the const str has embedded null + // characters, the string comparison functions should terminate + // at the first null + if (!isMemcmp) + TmpConstStr.assign(TmpConstStr, 0, TmpConstStr.find('\0') + 1); + + constStrLen = TmpConstStr.length(); + // prefer use of StringRef (in comparison to std::string a StringRef has + // built-in runtime bounds checking, which makes debugging easier) + ConstStr = StringRef(TmpConstStr); + + if (isConstSized) + unrollLen = constSizedLen < constStrLen ? constSizedLen : constStrLen; + else + unrollLen = constStrLen; + + if (!be_quiet) + errs() << callInst->getCalledFunction()->getName() << ": unroll len " + << unrollLen + << ((isSizedcmp && !isConstSized) ? ", variable n" : "") << ": " + << ConstStr << "\n"; + + /* split before the call instruction */ + BasicBlock *bb = callInst->getParent(); + BasicBlock *end_bb = bb->splitBasicBlock(BasicBlock::iterator(callInst)); + + BasicBlock *next_lenchk_bb = NULL; + if (isSizedcmp && !isConstSized) { + + next_lenchk_bb = + BasicBlock::Create(C, "len_check", end_bb->getParent(), end_bb); + BranchInst::Create(end_bb, next_lenchk_bb); + + } + + BasicBlock *next_cmp_bb = + BasicBlock::Create(C, "cmp_added", end_bb->getParent(), end_bb); + BranchInst::Create(end_bb, next_cmp_bb); + PHINode *PN = PHINode::Create( + Int32Ty, (next_lenchk_bb ? 2 : 1) * unrollLen + 1, "cmp_phi"); + +#if LLVM_VERSION_MAJOR < 8 + TerminatorInst *term = bb->getTerminator(); +#else + Instruction *term = bb->getTerminator(); +#endif + BranchInst::Create(next_lenchk_bb ? next_lenchk_bb : next_cmp_bb, bb); + term->eraseFromParent(); + + for (uint64_t i = 0; i < unrollLen; i++) { + + BasicBlock * cur_cmp_bb = next_cmp_bb, *cur_lenchk_bb = next_lenchk_bb; + unsigned char c; + + if (cur_lenchk_bb) { + + IRBuilder<> cur_lenchk_IRB(&*(cur_lenchk_bb->getFirstInsertionPt())); + Value * icmp = cur_lenchk_IRB.CreateICmpEQ( + sizedValue, ConstantInt::get(sizedValue->getType(), i)); + cur_lenchk_IRB.CreateCondBr(icmp, end_bb, cur_cmp_bb); + cur_lenchk_bb->getTerminator()->eraseFromParent(); + + PN->addIncoming(ConstantInt::get(Int32Ty, 0), cur_lenchk_bb); + + } + + if (isCaseInsensitive) + c = (unsigned char)(tolower((int)ConstStr[i]) & 0xff); + else + c = (unsigned char)ConstStr[i]; + + IRBuilder<> cur_cmp_IRB(&*(cur_cmp_bb->getFirstInsertionPt())); + + Value *v = ConstantInt::get(Int64Ty, i); + Value *ele = cur_cmp_IRB.CreateInBoundsGEP(VarStr, v, "empty"); + Value *load = cur_cmp_IRB.CreateLoad(ele); + + if (isCaseInsensitive) { + + // load >= 'A' && load <= 'Z' ? load | 0x020 : load + load = cur_cmp_IRB.CreateZExt(load, Int32Ty); + std::vector args; + args.push_back(load); + load = cur_cmp_IRB.CreateCall(tolowerFn, args); + load = cur_cmp_IRB.CreateTrunc(load, Int8Ty); + + } + + Value *isub; + if (HasStr1) + isub = cur_cmp_IRB.CreateSub(ConstantInt::get(Int8Ty, c), load); + else + isub = cur_cmp_IRB.CreateSub(load, ConstantInt::get(Int8Ty, c)); + + Value *sext = cur_cmp_IRB.CreateSExt(isub, Int32Ty); + PN->addIncoming(sext, cur_cmp_bb); + + if (i < unrollLen - 1) { + + if (cur_lenchk_bb) { + + next_lenchk_bb = + BasicBlock::Create(C, "len_check", end_bb->getParent(), end_bb); + BranchInst::Create(end_bb, next_lenchk_bb); + + } + + next_cmp_bb = + BasicBlock::Create(C, "cmp_added", end_bb->getParent(), end_bb); + BranchInst::Create(end_bb, next_cmp_bb); + + Value *icmp = + cur_cmp_IRB.CreateICmpEQ(isub, ConstantInt::get(Int8Ty, 0)); + cur_cmp_IRB.CreateCondBr( + icmp, next_lenchk_bb ? next_lenchk_bb : next_cmp_bb, end_bb); + cur_cmp_bb->getTerminator()->eraseFromParent(); + + } else { + + // IRB.CreateBr(end_bb); + + } + + // add offset to varstr + // create load + // create signed isub + // create icmp + // create jcc + // create next_bb + + } + + /* since the call is the first instruction of the bb it is safe to + * replace it with a phi instruction */ + BasicBlock::iterator ii(callInst); + ReplaceInstWithInst(callInst->getParent()->getInstList(), ii, PN); + + } + + return true; + +} + +bool CompareTransform::runOnModule(Module &M) { + + if ((isatty(2) && getenv("AFL_QUIET") == NULL) || getenv("AFL_DEBUG") != NULL) + llvm::errs() << "Running compare-transform-pass by laf.intel@gmail.com, " + "extended by heiko@hexco.de\n"; + else + be_quiet = 1; + transformCmps(M, true, true, true, true, true); + verifyModule(M); + + return true; + +} + +static void registerCompTransPass(const PassManagerBuilder &, + legacy::PassManagerBase &PM) { + + auto p = new CompareTransform(); + PM.add(p); + +} + +static RegisterStandardPasses RegisterCompTransPass( + PassManagerBuilder::EP_OptimizerLast, registerCompTransPass); + +static RegisterStandardPasses RegisterCompTransPass0( + PassManagerBuilder::EP_EnabledOnOptLevel0, registerCompTransPass); + +#if LLVM_VERSION_MAJOR >= 11 +static RegisterStandardPasses RegisterCompTransPassLTO( + PassManagerBuilder::EP_FullLinkTimeOptimizationLast, registerCompTransPass); +#endif + diff --git a/instrumentation/llvm-ngram-coverage.h b/instrumentation/llvm-ngram-coverage.h new file mode 100644 index 00000000..12b666e9 --- /dev/null +++ b/instrumentation/llvm-ngram-coverage.h @@ -0,0 +1,18 @@ +#ifndef AFL_NGRAM_CONFIG_H +#define AFL_NGRAM_CONFIG_H + +#include "../config.h" + +#if (MAP_SIZE_POW2 <= 16) +typedef u16 PREV_LOC_T; +#elif (MAP_SIZE_POW2 <= 32) +typedef u32 PREV_LOC_T; +#else +typedef u64 PREV_LOC_T; +#endif + +/* Maximum ngram size */ +#define NGRAM_SIZE_MAX 16U + +#endif + diff --git a/instrumentation/split-compares-pass.so.cc b/instrumentation/split-compares-pass.so.cc new file mode 100644 index 00000000..2fb90e5e --- /dev/null +++ b/instrumentation/split-compares-pass.so.cc @@ -0,0 +1,1356 @@ +/* + * Copyright 2016 laf-intel + * extended for floating point by Heiko Eißfeldt + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#include +#include +#include + +#include +#include +#include +#include + +#include "llvm/Config/llvm-config.h" + +#include "llvm/Pass.h" +#include "llvm/Support/raw_ostream.h" +#include "llvm/IR/LegacyPassManager.h" +#include "llvm/Transforms/IPO/PassManagerBuilder.h" +#include "llvm/Transforms/Utils/BasicBlockUtils.h" +#include "llvm/IR/Module.h" + +#include "llvm/IR/IRBuilder.h" +#if LLVM_VERSION_MAJOR > 3 || \ + (LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR > 4) + #include "llvm/IR/Verifier.h" + #include "llvm/IR/DebugInfo.h" +#else + #include "llvm/Analysis/Verifier.h" + #include "llvm/DebugInfo.h" + #define nullptr 0 +#endif + +using namespace llvm; +#include "afl-llvm-common.h" + +namespace { + +class SplitComparesTransform : public ModulePass { + + public: + static char ID; + SplitComparesTransform() : ModulePass(ID) { + + initInstrumentList(); + + } + + bool runOnModule(Module &M) override; +#if LLVM_VERSION_MAJOR >= 4 + StringRef getPassName() const override { + +#else + const char *getPassName() const override { + +#endif + return "simplifies and splits ICMP instructions"; + + } + + private: + int enableFPSplit; + + size_t splitIntCompares(Module &M, unsigned bitw); + size_t splitFPCompares(Module &M); + bool simplifyCompares(Module &M); + bool simplifyFPCompares(Module &M); + bool simplifyIntSignedness(Module &M); + size_t nextPowerOfTwo(size_t in); + +}; + +} // namespace + +char SplitComparesTransform::ID = 0; + +/* This function splits FCMP instructions with xGE or xLE predicates into two + * FCMP instructions with predicate xGT or xLT and EQ */ +bool SplitComparesTransform::simplifyFPCompares(Module &M) { + + LLVMContext & C = M.getContext(); + std::vector fcomps; + IntegerType * Int1Ty = IntegerType::getInt1Ty(C); + + /* iterate over all functions, bbs and instruction and add + * all integer comparisons with >= and <= predicates to the icomps vector */ + for (auto &F : M) { + + if (!isInInstrumentList(&F)) continue; + + for (auto &BB : F) { + + for (auto &IN : BB) { + + CmpInst *selectcmpInst = nullptr; + + if ((selectcmpInst = dyn_cast(&IN))) { + + if (enableFPSplit && + (selectcmpInst->getPredicate() == CmpInst::FCMP_OGE || + selectcmpInst->getPredicate() == CmpInst::FCMP_UGE || + selectcmpInst->getPredicate() == CmpInst::FCMP_OLE || + selectcmpInst->getPredicate() == CmpInst::FCMP_ULE)) { + + auto op0 = selectcmpInst->getOperand(0); + auto op1 = selectcmpInst->getOperand(1); + + Type *TyOp0 = op0->getType(); + Type *TyOp1 = op1->getType(); + + /* this is probably not needed but we do it anyway */ + if (TyOp0 != TyOp1) { continue; } + + if (TyOp0->isArrayTy() || TyOp0->isVectorTy()) { continue; } + + fcomps.push_back(selectcmpInst); + + } + + } + + } + + } + + } + + if (!fcomps.size()) { return false; } + + /* transform for floating point */ + for (auto &FcmpInst : fcomps) { + + BasicBlock *bb = FcmpInst->getParent(); + + auto op0 = FcmpInst->getOperand(0); + auto op1 = FcmpInst->getOperand(1); + + /* find out what the new predicate is going to be */ + auto pred = dyn_cast(FcmpInst)->getPredicate(); + CmpInst::Predicate new_pred; + switch (pred) { + + case CmpInst::FCMP_UGE: + new_pred = CmpInst::FCMP_UGT; + break; + case CmpInst::FCMP_OGE: + new_pred = CmpInst::FCMP_OGT; + break; + case CmpInst::FCMP_ULE: + new_pred = CmpInst::FCMP_ULT; + break; + case CmpInst::FCMP_OLE: + new_pred = CmpInst::FCMP_OLT; + break; + default: // keep the compiler happy + continue; + + } + + /* split before the fcmp instruction */ + BasicBlock *end_bb = bb->splitBasicBlock(BasicBlock::iterator(FcmpInst)); + + /* the old bb now contains a unconditional jump to the new one (end_bb) + * we need to delete it later */ + + /* create the FCMP instruction with new_pred and add it to the old basic + * block bb it is now at the position where the old FcmpInst was */ + Instruction *fcmp_np; + fcmp_np = CmpInst::Create(Instruction::FCmp, new_pred, op0, op1); + bb->getInstList().insert(BasicBlock::iterator(bb->getTerminator()), + fcmp_np); + + /* create a new basic block which holds the new EQ fcmp */ + Instruction *fcmp_eq; + /* insert middle_bb before end_bb */ + BasicBlock *middle_bb = + BasicBlock::Create(C, "injected", end_bb->getParent(), end_bb); + fcmp_eq = CmpInst::Create(Instruction::FCmp, CmpInst::FCMP_OEQ, op0, op1); + middle_bb->getInstList().push_back(fcmp_eq); + /* add an unconditional branch to the end of middle_bb with destination + * end_bb */ + BranchInst::Create(end_bb, middle_bb); + + /* replace the uncond branch with a conditional one, which depends on the + * new_pred fcmp. True goes to end, false to the middle (injected) bb */ + auto term = bb->getTerminator(); + BranchInst::Create(end_bb, middle_bb, fcmp_np, bb); + term->eraseFromParent(); + + /* replace the old FcmpInst (which is the first inst in end_bb) with a PHI + * inst to wire up the loose ends */ + PHINode *PN = PHINode::Create(Int1Ty, 2, ""); + /* the first result depends on the outcome of fcmp_eq */ + PN->addIncoming(fcmp_eq, middle_bb); + /* if the source was the original bb we know that the fcmp_np yielded true + * hence we can hardcode this value */ + PN->addIncoming(ConstantInt::get(Int1Ty, 1), bb); + /* replace the old FcmpInst with our new and shiny PHI inst */ + BasicBlock::iterator ii(FcmpInst); + ReplaceInstWithInst(FcmpInst->getParent()->getInstList(), ii, PN); + + } + + return true; + +} + +/* This function splits ICMP instructions with xGE or xLE predicates into two + * ICMP instructions with predicate xGT or xLT and EQ */ +bool SplitComparesTransform::simplifyCompares(Module &M) { + + LLVMContext & C = M.getContext(); + std::vector icomps; + IntegerType * Int1Ty = IntegerType::getInt1Ty(C); + + /* iterate over all functions, bbs and instruction and add + * all integer comparisons with >= and <= predicates to the icomps vector */ + for (auto &F : M) { + + if (!isInInstrumentList(&F)) continue; + + for (auto &BB : F) { + + for (auto &IN : BB) { + + CmpInst *selectcmpInst = nullptr; + + if ((selectcmpInst = dyn_cast(&IN))) { + + if (selectcmpInst->getPredicate() == CmpInst::ICMP_UGE || + selectcmpInst->getPredicate() == CmpInst::ICMP_SGE || + selectcmpInst->getPredicate() == CmpInst::ICMP_ULE || + selectcmpInst->getPredicate() == CmpInst::ICMP_SLE) { + + auto op0 = selectcmpInst->getOperand(0); + auto op1 = selectcmpInst->getOperand(1); + + IntegerType *intTyOp0 = dyn_cast(op0->getType()); + IntegerType *intTyOp1 = dyn_cast(op1->getType()); + + /* this is probably not needed but we do it anyway */ + if (!intTyOp0 || !intTyOp1) { continue; } + + icomps.push_back(selectcmpInst); + + } + + } + + } + + } + + } + + if (!icomps.size()) { return false; } + + for (auto &IcmpInst : icomps) { + + BasicBlock *bb = IcmpInst->getParent(); + + auto op0 = IcmpInst->getOperand(0); + auto op1 = IcmpInst->getOperand(1); + + /* find out what the new predicate is going to be */ + auto pred = dyn_cast(IcmpInst)->getPredicate(); + CmpInst::Predicate new_pred; + switch (pred) { + + case CmpInst::ICMP_UGE: + new_pred = CmpInst::ICMP_UGT; + break; + case CmpInst::ICMP_SGE: + new_pred = CmpInst::ICMP_SGT; + break; + case CmpInst::ICMP_ULE: + new_pred = CmpInst::ICMP_ULT; + break; + case CmpInst::ICMP_SLE: + new_pred = CmpInst::ICMP_SLT; + break; + default: // keep the compiler happy + continue; + + } + + /* split before the icmp instruction */ + BasicBlock *end_bb = bb->splitBasicBlock(BasicBlock::iterator(IcmpInst)); + + /* the old bb now contains a unconditional jump to the new one (end_bb) + * we need to delete it later */ + + /* create the ICMP instruction with new_pred and add it to the old basic + * block bb it is now at the position where the old IcmpInst was */ + Instruction *icmp_np; + icmp_np = CmpInst::Create(Instruction::ICmp, new_pred, op0, op1); + bb->getInstList().insert(BasicBlock::iterator(bb->getTerminator()), + icmp_np); + + /* create a new basic block which holds the new EQ icmp */ + Instruction *icmp_eq; + /* insert middle_bb before end_bb */ + BasicBlock *middle_bb = + BasicBlock::Create(C, "injected", end_bb->getParent(), end_bb); + icmp_eq = CmpInst::Create(Instruction::ICmp, CmpInst::ICMP_EQ, op0, op1); + middle_bb->getInstList().push_back(icmp_eq); + /* add an unconditional branch to the end of middle_bb with destination + * end_bb */ + BranchInst::Create(end_bb, middle_bb); + + /* replace the uncond branch with a conditional one, which depends on the + * new_pred icmp. True goes to end, false to the middle (injected) bb */ + auto term = bb->getTerminator(); + BranchInst::Create(end_bb, middle_bb, icmp_np, bb); + term->eraseFromParent(); + + /* replace the old IcmpInst (which is the first inst in end_bb) with a PHI + * inst to wire up the loose ends */ + PHINode *PN = PHINode::Create(Int1Ty, 2, ""); + /* the first result depends on the outcome of icmp_eq */ + PN->addIncoming(icmp_eq, middle_bb); + /* if the source was the original bb we know that the icmp_np yielded true + * hence we can hardcode this value */ + PN->addIncoming(ConstantInt::get(Int1Ty, 1), bb); + /* replace the old IcmpInst with our new and shiny PHI inst */ + BasicBlock::iterator ii(IcmpInst); + ReplaceInstWithInst(IcmpInst->getParent()->getInstList(), ii, PN); + + } + + return true; + +} + +/* this function transforms signed compares to equivalent unsigned compares */ +bool SplitComparesTransform::simplifyIntSignedness(Module &M) { + + LLVMContext & C = M.getContext(); + std::vector icomps; + IntegerType * Int1Ty = IntegerType::getInt1Ty(C); + + /* iterate over all functions, bbs and instructions and add + * all signed compares to icomps vector */ + for (auto &F : M) { + + if (!isInInstrumentList(&F)) continue; + + for (auto &BB : F) { + + for (auto &IN : BB) { + + CmpInst *selectcmpInst = nullptr; + + if ((selectcmpInst = dyn_cast(&IN))) { + + if (selectcmpInst->getPredicate() == CmpInst::ICMP_SGT || + selectcmpInst->getPredicate() == CmpInst::ICMP_SLT) { + + auto op0 = selectcmpInst->getOperand(0); + auto op1 = selectcmpInst->getOperand(1); + + IntegerType *intTyOp0 = dyn_cast(op0->getType()); + IntegerType *intTyOp1 = dyn_cast(op1->getType()); + + /* see above */ + if (!intTyOp0 || !intTyOp1) { continue; } + + /* i think this is not possible but to lazy to look it up */ + if (intTyOp0->getBitWidth() != intTyOp1->getBitWidth()) { + + continue; + + } + + icomps.push_back(selectcmpInst); + + } + + } + + } + + } + + } + + if (!icomps.size()) { return false; } + + for (auto &IcmpInst : icomps) { + + BasicBlock *bb = IcmpInst->getParent(); + + auto op0 = IcmpInst->getOperand(0); + auto op1 = IcmpInst->getOperand(1); + + IntegerType *intTyOp0 = dyn_cast(op0->getType()); + unsigned bitw = intTyOp0->getBitWidth(); + IntegerType *IntType = IntegerType::get(C, bitw); + + /* get the new predicate */ + auto pred = dyn_cast(IcmpInst)->getPredicate(); + CmpInst::Predicate new_pred; + if (pred == CmpInst::ICMP_SGT) { + + new_pred = CmpInst::ICMP_UGT; + + } else { + + new_pred = CmpInst::ICMP_ULT; + + } + + BasicBlock *end_bb = bb->splitBasicBlock(BasicBlock::iterator(IcmpInst)); + + /* create a 1 bit compare for the sign bit. to do this shift and trunc + * the original operands so only the first bit remains.*/ + Instruction *s_op0, *t_op0, *s_op1, *t_op1, *icmp_sign_bit; + + s_op0 = BinaryOperator::Create(Instruction::LShr, op0, + ConstantInt::get(IntType, bitw - 1)); + bb->getInstList().insert(BasicBlock::iterator(bb->getTerminator()), s_op0); + t_op0 = new TruncInst(s_op0, Int1Ty); + bb->getInstList().insert(BasicBlock::iterator(bb->getTerminator()), t_op0); + + s_op1 = BinaryOperator::Create(Instruction::LShr, op1, + ConstantInt::get(IntType, bitw - 1)); + bb->getInstList().insert(BasicBlock::iterator(bb->getTerminator()), s_op1); + t_op1 = new TruncInst(s_op1, Int1Ty); + bb->getInstList().insert(BasicBlock::iterator(bb->getTerminator()), t_op1); + + /* compare of the sign bits */ + icmp_sign_bit = + CmpInst::Create(Instruction::ICmp, CmpInst::ICMP_EQ, t_op0, t_op1); + bb->getInstList().insert(BasicBlock::iterator(bb->getTerminator()), + icmp_sign_bit); + + /* create a new basic block which is executed if the signedness bit is + * different */ + Instruction *icmp_inv_sig_cmp; + BasicBlock * sign_bb = + BasicBlock::Create(C, "sign", end_bb->getParent(), end_bb); + if (pred == CmpInst::ICMP_SGT) { + + /* if we check for > and the op0 positive and op1 negative then the final + * result is true. if op0 negative and op1 pos, the cmp must result + * in false + */ + icmp_inv_sig_cmp = + CmpInst::Create(Instruction::ICmp, CmpInst::ICMP_ULT, t_op0, t_op1); + + } else { + + /* just the inverse of the above statement */ + icmp_inv_sig_cmp = + CmpInst::Create(Instruction::ICmp, CmpInst::ICMP_UGT, t_op0, t_op1); + + } + + sign_bb->getInstList().push_back(icmp_inv_sig_cmp); + BranchInst::Create(end_bb, sign_bb); + + /* create a new bb which is executed if signedness is equal */ + Instruction *icmp_usign_cmp; + BasicBlock * middle_bb = + BasicBlock::Create(C, "injected", end_bb->getParent(), end_bb); + /* we can do a normal unsigned compare now */ + icmp_usign_cmp = CmpInst::Create(Instruction::ICmp, new_pred, op0, op1); + middle_bb->getInstList().push_back(icmp_usign_cmp); + BranchInst::Create(end_bb, middle_bb); + + auto term = bb->getTerminator(); + /* if the sign is eq do a normal unsigned cmp, else we have to check the + * signedness bit */ + BranchInst::Create(middle_bb, sign_bb, icmp_sign_bit, bb); + term->eraseFromParent(); + + PHINode *PN = PHINode::Create(Int1Ty, 2, ""); + + PN->addIncoming(icmp_usign_cmp, middle_bb); + PN->addIncoming(icmp_inv_sig_cmp, sign_bb); + + BasicBlock::iterator ii(IcmpInst); + ReplaceInstWithInst(IcmpInst->getParent()->getInstList(), ii, PN); + + } + + return true; + +} + +size_t SplitComparesTransform::nextPowerOfTwo(size_t in) { + + --in; + in |= in >> 1; + in |= in >> 2; + in |= in >> 4; + // in |= in >> 8; + // in |= in >> 16; + return in + 1; + +} + +/* splits fcmps into two nested fcmps with sign compare and the rest */ +size_t SplitComparesTransform::splitFPCompares(Module &M) { + + size_t count = 0; + + LLVMContext &C = M.getContext(); + +#if LLVM_VERSION_MAJOR > 3 || \ + (LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR > 7) + const DataLayout &dl = M.getDataLayout(); + + /* define unions with floating point and (sign, exponent, mantissa) triples + */ + if (dl.isLittleEndian()) { + + } else if (dl.isBigEndian()) { + + } else { + + return count; + + } + +#endif + + std::vector fcomps; + + /* get all EQ, NE, GT, and LT fcmps. if the other two + * functions were executed only these four predicates should exist */ + for (auto &F : M) { + + if (!isInInstrumentList(&F)) continue; + + for (auto &BB : F) { + + for (auto &IN : BB) { + + CmpInst *selectcmpInst = nullptr; + + if ((selectcmpInst = dyn_cast(&IN))) { + + if (selectcmpInst->getPredicate() == CmpInst::FCMP_OEQ || + selectcmpInst->getPredicate() == CmpInst::FCMP_ONE || + selectcmpInst->getPredicate() == CmpInst::FCMP_UNE || + selectcmpInst->getPredicate() == CmpInst::FCMP_UGT || + selectcmpInst->getPredicate() == CmpInst::FCMP_OGT || + selectcmpInst->getPredicate() == CmpInst::FCMP_ULT || + selectcmpInst->getPredicate() == CmpInst::FCMP_OLT) { + + auto op0 = selectcmpInst->getOperand(0); + auto op1 = selectcmpInst->getOperand(1); + + Type *TyOp0 = op0->getType(); + Type *TyOp1 = op1->getType(); + + if (TyOp0 != TyOp1) { continue; } + + if (TyOp0->isArrayTy() || TyOp0->isVectorTy()) { continue; } + + fcomps.push_back(selectcmpInst); + + } + + } + + } + + } + + } + + if (!fcomps.size()) { return count; } + + IntegerType *Int1Ty = IntegerType::getInt1Ty(C); + + for (auto &FcmpInst : fcomps) { + + BasicBlock *bb = FcmpInst->getParent(); + + auto op0 = FcmpInst->getOperand(0); + auto op1 = FcmpInst->getOperand(1); + + unsigned op_size; + op_size = op0->getType()->getPrimitiveSizeInBits(); + + if (op_size != op1->getType()->getPrimitiveSizeInBits()) { continue; } + + const unsigned int sizeInBits = op0->getType()->getPrimitiveSizeInBits(); + const unsigned int precision = + sizeInBits == 32 + ? 24 + : sizeInBits == 64 + ? 53 + : sizeInBits == 128 ? 113 + : sizeInBits == 16 ? 11 + /* sizeInBits == 80 */ + : 65; + + const unsigned shiftR_exponent = precision - 1; + const unsigned long long mask_fraction = + (1ULL << (shiftR_exponent - 1)) | ((1ULL << (shiftR_exponent - 1)) - 1); + const unsigned long long mask_exponent = + (1ULL << (sizeInBits - precision)) - 1; + + // round up sizes to the next power of two + // this should help with integer compare splitting + size_t exTySizeBytes = ((sizeInBits - precision + 7) >> 3); + size_t frTySizeBytes = ((precision - 1ULL + 7) >> 3); + + IntegerType *IntExponentTy = + IntegerType::get(C, nextPowerOfTwo(exTySizeBytes) << 3); + IntegerType *IntFractionTy = + IntegerType::get(C, nextPowerOfTwo(frTySizeBytes) << 3); + + // errs() << "Fractions: IntFractionTy size " << + // IntFractionTy->getPrimitiveSizeInBits() << ", op_size " << op_size << + // ", mask " << mask_fraction << + // ", precision " << precision << "\n"; + + BasicBlock *end_bb = bb->splitBasicBlock(BasicBlock::iterator(FcmpInst)); + + /* create the integers from floats directly */ + Instruction *b_op0, *b_op1; + b_op0 = CastInst::Create(Instruction::BitCast, op0, + IntegerType::get(C, op_size)); + bb->getInstList().insert(BasicBlock::iterator(bb->getTerminator()), b_op0); + + b_op1 = CastInst::Create(Instruction::BitCast, op1, + IntegerType::get(C, op_size)); + bb->getInstList().insert(BasicBlock::iterator(bb->getTerminator()), b_op1); + + /* isolate signs of value of floating point type */ + + /* create a 1 bit compare for the sign bit. to do this shift and trunc + * the original operands so only the first bit remains.*/ + Instruction *s_s0, *t_s0, *s_s1, *t_s1, *icmp_sign_bit; + + s_s0 = + BinaryOperator::Create(Instruction::LShr, b_op0, + ConstantInt::get(b_op0->getType(), op_size - 1)); + bb->getInstList().insert(BasicBlock::iterator(bb->getTerminator()), s_s0); + t_s0 = new TruncInst(s_s0, Int1Ty); + bb->getInstList().insert(BasicBlock::iterator(bb->getTerminator()), t_s0); + + s_s1 = + BinaryOperator::Create(Instruction::LShr, b_op1, + ConstantInt::get(b_op1->getType(), op_size - 1)); + bb->getInstList().insert(BasicBlock::iterator(bb->getTerminator()), s_s1); + t_s1 = new TruncInst(s_s1, Int1Ty); + bb->getInstList().insert(BasicBlock::iterator(bb->getTerminator()), t_s1); + + /* compare of the sign bits */ + icmp_sign_bit = + CmpInst::Create(Instruction::ICmp, CmpInst::ICMP_EQ, t_s0, t_s1); + bb->getInstList().insert(BasicBlock::iterator(bb->getTerminator()), + icmp_sign_bit); + + /* create a new basic block which is executed if the signedness bits are + * equal */ + BasicBlock *signequal_bb = + BasicBlock::Create(C, "signequal", end_bb->getParent(), end_bb); + + BranchInst::Create(end_bb, signequal_bb); + + /* create a new bb which is executed if exponents are satisfying the compare + */ + BasicBlock *middle_bb = + BasicBlock::Create(C, "injected", end_bb->getParent(), end_bb); + + BranchInst::Create(end_bb, middle_bb); + + auto term = bb->getTerminator(); + /* if the signs are different goto end_bb else to signequal_bb */ + BranchInst::Create(signequal_bb, end_bb, icmp_sign_bit, bb); + term->eraseFromParent(); + + /* insert code for equal signs */ + + /* isolate the exponents */ + Instruction *s_e0, *m_e0, *t_e0, *s_e1, *m_e1, *t_e1; + + s_e0 = BinaryOperator::Create( + Instruction::LShr, b_op0, + ConstantInt::get(b_op0->getType(), shiftR_exponent)); + s_e1 = BinaryOperator::Create( + Instruction::LShr, b_op1, + ConstantInt::get(b_op1->getType(), shiftR_exponent)); + signequal_bb->getInstList().insert( + BasicBlock::iterator(signequal_bb->getTerminator()), s_e0); + signequal_bb->getInstList().insert( + BasicBlock::iterator(signequal_bb->getTerminator()), s_e1); + + t_e0 = new TruncInst(s_e0, IntExponentTy); + t_e1 = new TruncInst(s_e1, IntExponentTy); + signequal_bb->getInstList().insert( + BasicBlock::iterator(signequal_bb->getTerminator()), t_e0); + signequal_bb->getInstList().insert( + BasicBlock::iterator(signequal_bb->getTerminator()), t_e1); + + if (sizeInBits - precision < exTySizeBytes * 8) { + + m_e0 = BinaryOperator::Create( + Instruction::And, t_e0, + ConstantInt::get(t_e0->getType(), mask_exponent)); + m_e1 = BinaryOperator::Create( + Instruction::And, t_e1, + ConstantInt::get(t_e1->getType(), mask_exponent)); + signequal_bb->getInstList().insert( + BasicBlock::iterator(signequal_bb->getTerminator()), m_e0); + signequal_bb->getInstList().insert( + BasicBlock::iterator(signequal_bb->getTerminator()), m_e1); + + } else { + + m_e0 = t_e0; + m_e1 = t_e1; + + } + + /* compare the exponents of the operands */ + Instruction *icmp_exponents_equal; + Instruction *icmp_exponent_result; + BasicBlock * signequal2_bb = signequal_bb; + switch (FcmpInst->getPredicate()) { + + case CmpInst::FCMP_OEQ: + icmp_exponent_result = + CmpInst::Create(Instruction::ICmp, CmpInst::ICMP_EQ, m_e0, m_e1); + break; + case CmpInst::FCMP_ONE: + case CmpInst::FCMP_UNE: + icmp_exponent_result = + CmpInst::Create(Instruction::ICmp, CmpInst::ICMP_NE, m_e0, m_e1); + break; + /* compare the exponents of the operands (signs are equal) + * if exponents are equal -> proceed to mantissa comparison + * else get result depending on sign + */ + case CmpInst::FCMP_OGT: + case CmpInst::FCMP_UGT: + Instruction *icmp_exponent; + icmp_exponents_equal = + CmpInst::Create(Instruction::ICmp, CmpInst::ICMP_EQ, m_e0, m_e1); + signequal_bb->getInstList().insert( + BasicBlock::iterator(signequal_bb->getTerminator()), + icmp_exponents_equal); + + // shortcut for unequal exponents + signequal2_bb = signequal_bb->splitBasicBlock( + BasicBlock::iterator(signequal_bb->getTerminator())); + + /* if the exponents are equal goto middle_bb else to signequal2_bb */ + term = signequal_bb->getTerminator(); + BranchInst::Create(middle_bb, signequal2_bb, icmp_exponents_equal, + signequal_bb); + term->eraseFromParent(); + + icmp_exponent = + CmpInst::Create(Instruction::ICmp, CmpInst::ICMP_UGT, m_e0, m_e1); + signequal2_bb->getInstList().insert( + BasicBlock::iterator(signequal2_bb->getTerminator()), + icmp_exponent); + icmp_exponent_result = + BinaryOperator::Create(Instruction::Xor, icmp_exponent, t_s0); + break; + case CmpInst::FCMP_OLT: + case CmpInst::FCMP_ULT: + icmp_exponents_equal = + CmpInst::Create(Instruction::ICmp, CmpInst::ICMP_EQ, m_e0, m_e1); + signequal_bb->getInstList().insert( + BasicBlock::iterator(signequal_bb->getTerminator()), + icmp_exponents_equal); + + // shortcut for unequal exponents + signequal2_bb = signequal_bb->splitBasicBlock( + BasicBlock::iterator(signequal_bb->getTerminator())); + + /* if the exponents are equal goto middle_bb else to signequal2_bb */ + term = signequal_bb->getTerminator(); + BranchInst::Create(middle_bb, signequal2_bb, icmp_exponents_equal, + signequal_bb); + term->eraseFromParent(); + + icmp_exponent = + CmpInst::Create(Instruction::ICmp, CmpInst::ICMP_ULT, m_e0, m_e1); + signequal2_bb->getInstList().insert( + BasicBlock::iterator(signequal2_bb->getTerminator()), + icmp_exponent); + icmp_exponent_result = + BinaryOperator::Create(Instruction::Xor, icmp_exponent, t_s0); + break; + default: + continue; + + } + + signequal2_bb->getInstList().insert( + BasicBlock::iterator(signequal2_bb->getTerminator()), + icmp_exponent_result); + + { + + term = signequal2_bb->getTerminator(); + + switch (FcmpInst->getPredicate()) { + + case CmpInst::FCMP_OEQ: + /* if the exponents are satifying the compare do a fraction cmp in + * middle_bb */ + BranchInst::Create(middle_bb, end_bb, icmp_exponent_result, + signequal2_bb); + break; + case CmpInst::FCMP_ONE: + case CmpInst::FCMP_UNE: + /* if the exponents are satifying the compare do a fraction cmp in + * middle_bb */ + BranchInst::Create(end_bb, middle_bb, icmp_exponent_result, + signequal2_bb); + break; + case CmpInst::FCMP_OGT: + case CmpInst::FCMP_UGT: + case CmpInst::FCMP_OLT: + case CmpInst::FCMP_ULT: + BranchInst::Create(end_bb, signequal2_bb); + break; + default: + continue; + + } + + term->eraseFromParent(); + + } + + /* isolate the mantissa aka fraction */ + Instruction *t_f0, *t_f1; + bool needTrunc = IntFractionTy->getPrimitiveSizeInBits() < op_size; + + if (precision - 1 < frTySizeBytes * 8) { + + Instruction *m_f0, *m_f1; + m_f0 = BinaryOperator::Create( + Instruction::And, b_op0, + ConstantInt::get(b_op0->getType(), mask_fraction)); + m_f1 = BinaryOperator::Create( + Instruction::And, b_op1, + ConstantInt::get(b_op1->getType(), mask_fraction)); + middle_bb->getInstList().insert( + BasicBlock::iterator(middle_bb->getTerminator()), m_f0); + middle_bb->getInstList().insert( + BasicBlock::iterator(middle_bb->getTerminator()), m_f1); + + if (needTrunc) { + + t_f0 = new TruncInst(m_f0, IntFractionTy); + t_f1 = new TruncInst(m_f1, IntFractionTy); + middle_bb->getInstList().insert( + BasicBlock::iterator(middle_bb->getTerminator()), t_f0); + middle_bb->getInstList().insert( + BasicBlock::iterator(middle_bb->getTerminator()), t_f1); + + } else { + + t_f0 = m_f0; + t_f1 = m_f1; + + } + + } else { + + if (needTrunc) { + + t_f0 = new TruncInst(b_op0, IntFractionTy); + t_f1 = new TruncInst(b_op1, IntFractionTy); + middle_bb->getInstList().insert( + BasicBlock::iterator(middle_bb->getTerminator()), t_f0); + middle_bb->getInstList().insert( + BasicBlock::iterator(middle_bb->getTerminator()), t_f1); + + } else { + + t_f0 = b_op0; + t_f1 = b_op1; + + } + + } + + /* compare the fractions of the operands */ + Instruction *icmp_fraction_result; + Instruction *icmp_fraction_result2; + BasicBlock * middle2_bb = middle_bb; + PHINode * PN2 = nullptr; + switch (FcmpInst->getPredicate()) { + + case CmpInst::FCMP_OEQ: + icmp_fraction_result = + CmpInst::Create(Instruction::ICmp, CmpInst::ICMP_EQ, t_f0, t_f1); + middle2_bb->getInstList().insert( + BasicBlock::iterator(middle2_bb->getTerminator()), + icmp_fraction_result); + + break; + case CmpInst::FCMP_UNE: + case CmpInst::FCMP_ONE: + icmp_fraction_result = + CmpInst::Create(Instruction::ICmp, CmpInst::ICMP_NE, t_f0, t_f1); + middle2_bb->getInstList().insert( + BasicBlock::iterator(middle2_bb->getTerminator()), + icmp_fraction_result); + + break; + case CmpInst::FCMP_OGT: + case CmpInst::FCMP_UGT: + case CmpInst::FCMP_OLT: + case CmpInst::FCMP_ULT: { + + middle2_bb = middle_bb->splitBasicBlock( + BasicBlock::iterator(middle_bb->getTerminator())); + + BasicBlock *negative_bb = BasicBlock::Create( + C, "negative_value", middle2_bb->getParent(), middle2_bb); + BasicBlock *positive_bb = BasicBlock::Create( + C, "positive_value", negative_bb->getParent(), negative_bb); + + if (FcmpInst->getPredicate() == CmpInst::FCMP_OGT || + FcmpInst->getPredicate() == CmpInst::FCMP_UGT) { + + negative_bb->getInstList().push_back( + icmp_fraction_result = CmpInst::Create( + Instruction::ICmp, CmpInst::ICMP_ULT, t_f0, t_f1)); + positive_bb->getInstList().push_back( + icmp_fraction_result2 = CmpInst::Create( + Instruction::ICmp, CmpInst::ICMP_UGT, t_f0, t_f1)); + + } else { + + negative_bb->getInstList().push_back( + icmp_fraction_result = CmpInst::Create( + Instruction::ICmp, CmpInst::ICMP_UGT, t_f0, t_f1)); + positive_bb->getInstList().push_back( + icmp_fraction_result2 = CmpInst::Create( + Instruction::ICmp, CmpInst::ICMP_ULT, t_f0, t_f1)); + + } + + BranchInst::Create(middle2_bb, negative_bb); + BranchInst::Create(middle2_bb, positive_bb); + + term = middle_bb->getTerminator(); + BranchInst::Create(negative_bb, positive_bb, t_s0, middle_bb); + term->eraseFromParent(); + + PN2 = PHINode::Create(Int1Ty, 2, ""); + PN2->addIncoming(icmp_fraction_result, negative_bb); + PN2->addIncoming(icmp_fraction_result2, positive_bb); + middle2_bb->getInstList().insert( + BasicBlock::iterator(middle2_bb->getTerminator()), PN2); + + } break; + + default: + continue; + + } + + PHINode *PN = PHINode::Create(Int1Ty, 3, ""); + + switch (FcmpInst->getPredicate()) { + + case CmpInst::FCMP_OEQ: + /* unequal signs cannot be equal values */ + /* goto false branch */ + PN->addIncoming(ConstantInt::get(Int1Ty, 0), bb); + /* unequal exponents cannot be equal values, too */ + PN->addIncoming(ConstantInt::get(Int1Ty, 0), signequal_bb); + /* fractions comparison */ + PN->addIncoming(icmp_fraction_result, middle2_bb); + break; + case CmpInst::FCMP_ONE: + case CmpInst::FCMP_UNE: + /* unequal signs are unequal values */ + /* goto true branch */ + PN->addIncoming(ConstantInt::get(Int1Ty, 1), bb); + /* unequal exponents are unequal values, too */ + PN->addIncoming(icmp_exponent_result, signequal_bb); + /* fractions comparison */ + PN->addIncoming(icmp_fraction_result, middle2_bb); + break; + case CmpInst::FCMP_OGT: + case CmpInst::FCMP_UGT: + /* if op1 is negative goto true branch, + else go on comparing */ + PN->addIncoming(t_s1, bb); + PN->addIncoming(icmp_exponent_result, signequal2_bb); + PN->addIncoming(PN2, middle2_bb); + break; + case CmpInst::FCMP_OLT: + case CmpInst::FCMP_ULT: + /* if op0 is negative goto true branch, + else go on comparing */ + PN->addIncoming(t_s0, bb); + PN->addIncoming(icmp_exponent_result, signequal2_bb); + PN->addIncoming(PN2, middle2_bb); + break; + default: + continue; + + } + + BasicBlock::iterator ii(FcmpInst); + ReplaceInstWithInst(FcmpInst->getParent()->getInstList(), ii, PN); + ++count; + + } + + return count; + +} + +/* splits icmps of size bitw into two nested icmps with bitw/2 size each */ +size_t SplitComparesTransform::splitIntCompares(Module &M, unsigned bitw) { + + size_t count = 0; + + LLVMContext &C = M.getContext(); + + IntegerType *Int1Ty = IntegerType::getInt1Ty(C); + IntegerType *OldIntType = IntegerType::get(C, bitw); + IntegerType *NewIntType = IntegerType::get(C, bitw / 2); + + std::vector icomps; + + if (bitw % 2) { return 0; } + + /* not supported yet */ + if (bitw > 64) { return 0; } + + /* get all EQ, NE, UGT, and ULT icmps of width bitw. if the + * functions simplifyCompares() and simplifyIntSignedness() + * were executed only these four predicates should exist */ + for (auto &F : M) { + + if (!isInInstrumentList(&F)) continue; + + for (auto &BB : F) { + + for (auto &IN : BB) { + + CmpInst *selectcmpInst = nullptr; + + if ((selectcmpInst = dyn_cast(&IN))) { + + if (selectcmpInst->getPredicate() == CmpInst::ICMP_EQ || + selectcmpInst->getPredicate() == CmpInst::ICMP_NE || + selectcmpInst->getPredicate() == CmpInst::ICMP_UGT || + selectcmpInst->getPredicate() == CmpInst::ICMP_ULT) { + + auto op0 = selectcmpInst->getOperand(0); + auto op1 = selectcmpInst->getOperand(1); + + IntegerType *intTyOp0 = dyn_cast(op0->getType()); + IntegerType *intTyOp1 = dyn_cast(op1->getType()); + + if (!intTyOp0 || !intTyOp1) { continue; } + + /* check if the bitwidths are the one we are looking for */ + if (intTyOp0->getBitWidth() != bitw || + intTyOp1->getBitWidth() != bitw) { + + continue; + + } + + icomps.push_back(selectcmpInst); + + } + + } + + } + + } + + } + + if (!icomps.size()) { return 0; } + + for (auto &IcmpInst : icomps) { + + BasicBlock *bb = IcmpInst->getParent(); + + auto op0 = IcmpInst->getOperand(0); + auto op1 = IcmpInst->getOperand(1); + + auto pred = dyn_cast(IcmpInst)->getPredicate(); + + BasicBlock *end_bb = bb->splitBasicBlock(BasicBlock::iterator(IcmpInst)); + + /* create the comparison of the top halves of the original operands */ + Instruction *s_op0, *op0_high, *s_op1, *op1_high, *icmp_high; + + s_op0 = BinaryOperator::Create(Instruction::LShr, op0, + ConstantInt::get(OldIntType, bitw / 2)); + bb->getInstList().insert(BasicBlock::iterator(bb->getTerminator()), s_op0); + op0_high = new TruncInst(s_op0, NewIntType); + bb->getInstList().insert(BasicBlock::iterator(bb->getTerminator()), + op0_high); + + s_op1 = BinaryOperator::Create(Instruction::LShr, op1, + ConstantInt::get(OldIntType, bitw / 2)); + bb->getInstList().insert(BasicBlock::iterator(bb->getTerminator()), s_op1); + op1_high = new TruncInst(s_op1, NewIntType); + bb->getInstList().insert(BasicBlock::iterator(bb->getTerminator()), + op1_high); + + icmp_high = CmpInst::Create(Instruction::ICmp, pred, op0_high, op1_high); + bb->getInstList().insert(BasicBlock::iterator(bb->getTerminator()), + icmp_high); + + /* now we have to destinguish between == != and > < */ + if (pred == CmpInst::ICMP_EQ || pred == CmpInst::ICMP_NE) { + + /* transformation for == and != icmps */ + + /* create a compare for the lower half of the original operands */ + Instruction *op0_low, *op1_low, *icmp_low; + BasicBlock * cmp_low_bb = + BasicBlock::Create(C, "injected", end_bb->getParent(), end_bb); + + op0_low = new TruncInst(op0, NewIntType); + cmp_low_bb->getInstList().push_back(op0_low); + + op1_low = new TruncInst(op1, NewIntType); + cmp_low_bb->getInstList().push_back(op1_low); + + icmp_low = CmpInst::Create(Instruction::ICmp, pred, op0_low, op1_low); + cmp_low_bb->getInstList().push_back(icmp_low); + BranchInst::Create(end_bb, cmp_low_bb); + + /* dependent on the cmp of the high parts go to the end or go on with + * the comparison */ + auto term = bb->getTerminator(); + if (pred == CmpInst::ICMP_EQ) { + + BranchInst::Create(cmp_low_bb, end_bb, icmp_high, bb); + + } else { + + /* CmpInst::ICMP_NE */ + BranchInst::Create(end_bb, cmp_low_bb, icmp_high, bb); + + } + + term->eraseFromParent(); + + /* create the PHI and connect the edges accordingly */ + PHINode *PN = PHINode::Create(Int1Ty, 2, ""); + PN->addIncoming(icmp_low, cmp_low_bb); + if (pred == CmpInst::ICMP_EQ) { + + PN->addIncoming(ConstantInt::get(Int1Ty, 0), bb); + + } else { + + /* CmpInst::ICMP_NE */ + PN->addIncoming(ConstantInt::get(Int1Ty, 1), bb); + + } + + /* replace the old icmp with the new PHI */ + BasicBlock::iterator ii(IcmpInst); + ReplaceInstWithInst(IcmpInst->getParent()->getInstList(), ii, PN); + + } else { + + /* CmpInst::ICMP_UGT and CmpInst::ICMP_ULT */ + /* transformations for < and > */ + + /* create a basic block which checks for the inverse predicate. + * if this is true we can go to the end if not we have to go to the + * bb which checks the lower half of the operands */ + Instruction *icmp_inv_cmp, *op0_low, *op1_low, *icmp_low; + BasicBlock * inv_cmp_bb = + BasicBlock::Create(C, "inv_cmp", end_bb->getParent(), end_bb); + if (pred == CmpInst::ICMP_UGT) { + + icmp_inv_cmp = CmpInst::Create(Instruction::ICmp, CmpInst::ICMP_ULT, + op0_high, op1_high); + + } else { + + icmp_inv_cmp = CmpInst::Create(Instruction::ICmp, CmpInst::ICMP_UGT, + op0_high, op1_high); + + } + + inv_cmp_bb->getInstList().push_back(icmp_inv_cmp); + + auto term = bb->getTerminator(); + term->eraseFromParent(); + BranchInst::Create(end_bb, inv_cmp_bb, icmp_high, bb); + + /* create a bb which handles the cmp of the lower halves */ + BasicBlock *cmp_low_bb = + BasicBlock::Create(C, "injected", end_bb->getParent(), end_bb); + op0_low = new TruncInst(op0, NewIntType); + cmp_low_bb->getInstList().push_back(op0_low); + op1_low = new TruncInst(op1, NewIntType); + cmp_low_bb->getInstList().push_back(op1_low); + + icmp_low = CmpInst::Create(Instruction::ICmp, pred, op0_low, op1_low); + cmp_low_bb->getInstList().push_back(icmp_low); + BranchInst::Create(end_bb, cmp_low_bb); + + BranchInst::Create(end_bb, cmp_low_bb, icmp_inv_cmp, inv_cmp_bb); + + PHINode *PN = PHINode::Create(Int1Ty, 3); + PN->addIncoming(icmp_low, cmp_low_bb); + PN->addIncoming(ConstantInt::get(Int1Ty, 1), bb); + PN->addIncoming(ConstantInt::get(Int1Ty, 0), inv_cmp_bb); + + BasicBlock::iterator ii(IcmpInst); + ReplaceInstWithInst(IcmpInst->getParent()->getInstList(), ii, PN); + + } + + ++count; + + } + + return count; + +} + +bool SplitComparesTransform::runOnModule(Module &M) { + + int bitw = 64; + size_t count; + + char *bitw_env = getenv("AFL_LLVM_LAF_SPLIT_COMPARES_BITW"); + if (!bitw_env) bitw_env = getenv("LAF_SPLIT_COMPARES_BITW"); + if (bitw_env) { bitw = atoi(bitw_env); } + + enableFPSplit = getenv("AFL_LLVM_LAF_SPLIT_FLOATS") != NULL; + + if ((isatty(2) && getenv("AFL_QUIET") == NULL) || + getenv("AFL_DEBUG") != NULL) { + + errs() << "Split-compare-pass by laf.intel@gmail.com, extended by " + "heiko@hexco.de\n"; + + } else { + + be_quiet = 1; + + } + + if (enableFPSplit) { + + count = splitFPCompares(M); + + if (!be_quiet) { + + errs() << "Split-floatingpoint-compare-pass: " << count + << " FP comparisons split\n"; + + } + + simplifyFPCompares(M); + + } + + simplifyCompares(M); + + simplifyIntSignedness(M); + + switch (bitw) { + + case 64: + count = splitIntCompares(M, bitw); + if (!be_quiet) + errs() << "Split-integer-compare-pass " << bitw << "bit: " << count + << " split\n"; + + bitw >>= 1; +#if LLVM_VERSION_MAJOR > 3 || \ + (LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR > 7) + [[clang::fallthrough]]; /*FALLTHRU*/ /* FALLTHROUGH */ +#endif + case 32: + count = splitIntCompares(M, bitw); + if (!be_quiet) + errs() << "Split-integer-compare-pass " << bitw << "bit: " << count + << " split\n"; + + bitw >>= 1; +#if LLVM_VERSION_MAJOR > 3 || \ + (LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR > 7) + [[clang::fallthrough]]; /*FALLTHRU*/ /* FALLTHROUGH */ +#endif + case 16: + count = splitIntCompares(M, bitw); + if (!be_quiet) + errs() << "Split-integer-compare-pass " << bitw << "bit: " << count + << " split\n"; + + bitw >>= 1; + break; + + default: + if (!be_quiet) errs() << "NOT Running split-compare-pass \n"; + return false; + break; + + } + + verifyModule(M); + return true; + +} + +static void registerSplitComparesPass(const PassManagerBuilder &, + legacy::PassManagerBase &PM) { + + PM.add(new SplitComparesTransform()); + +} + +static RegisterStandardPasses RegisterSplitComparesPass( + PassManagerBuilder::EP_OptimizerLast, registerSplitComparesPass); + +static RegisterStandardPasses RegisterSplitComparesTransPass0( + PassManagerBuilder::EP_EnabledOnOptLevel0, registerSplitComparesPass); + +#if LLVM_VERSION_MAJOR >= 11 +static RegisterStandardPasses RegisterSplitComparesTransPassLTO( + PassManagerBuilder::EP_FullLinkTimeOptimizationLast, + registerSplitComparesPass); +#endif + diff --git a/instrumentation/split-switches-pass.so.cc b/instrumentation/split-switches-pass.so.cc new file mode 100644 index 00000000..a79d4114 --- /dev/null +++ b/instrumentation/split-switches-pass.so.cc @@ -0,0 +1,447 @@ +/* + * Copyright 2016 laf-intel + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#include +#include +#include + +#include +#include +#include +#include + +#include "llvm/Config/llvm-config.h" + +#include "llvm/ADT/Statistic.h" +#include "llvm/IR/IRBuilder.h" +#include "llvm/IR/LegacyPassManager.h" +#include "llvm/IR/Module.h" +#include "llvm/Support/Debug.h" +#include "llvm/Support/raw_ostream.h" +#include "llvm/Transforms/IPO/PassManagerBuilder.h" +#include "llvm/Transforms/Utils/BasicBlockUtils.h" +#include "llvm/Pass.h" +#include "llvm/Analysis/ValueTracking.h" + +#include "llvm/IR/IRBuilder.h" +#if LLVM_VERSION_MAJOR > 3 || \ + (LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR > 4) + #include "llvm/IR/Verifier.h" + #include "llvm/IR/DebugInfo.h" +#else + #include "llvm/Analysis/Verifier.h" + #include "llvm/DebugInfo.h" + #define nullptr 0 +#endif + +#include +#include "afl-llvm-common.h" + +using namespace llvm; + +namespace { + +class SplitSwitchesTransform : public ModulePass { + + public: + static char ID; + SplitSwitchesTransform() : ModulePass(ID) { + + initInstrumentList(); + + } + + bool runOnModule(Module &M) override; + +#if LLVM_VERSION_MAJOR >= 4 + StringRef getPassName() const override { + +#else + const char *getPassName() const override { + +#endif + return "splits switch constructs"; + + } + + struct CaseExpr { + + ConstantInt *Val; + BasicBlock * BB; + + CaseExpr(ConstantInt *val = nullptr, BasicBlock *bb = nullptr) + : Val(val), BB(bb) { + + } + + }; + + typedef std::vector CaseVector; + + private: + bool splitSwitches(Module &M); + bool transformCmps(Module &M, const bool processStrcmp, + const bool processMemcmp); + BasicBlock *switchConvert(CaseVector Cases, std::vector bytesChecked, + BasicBlock *OrigBlock, BasicBlock *NewDefault, + Value *Val, unsigned level); + +}; + +} // namespace + +char SplitSwitchesTransform::ID = 0; + +/* switchConvert - Transform simple list of Cases into list of CaseRange's */ +BasicBlock *SplitSwitchesTransform::switchConvert( + CaseVector Cases, std::vector bytesChecked, BasicBlock *OrigBlock, + BasicBlock *NewDefault, Value *Val, unsigned level) { + + unsigned ValTypeBitWidth = Cases[0].Val->getBitWidth(); + IntegerType *ValType = + IntegerType::get(OrigBlock->getContext(), ValTypeBitWidth); + IntegerType * ByteType = IntegerType::get(OrigBlock->getContext(), 8); + unsigned BytesInValue = bytesChecked.size(); + std::vector setSizes; + std::vector > byteSets(BytesInValue, std::set()); + + assert(ValTypeBitWidth >= 8 && ValTypeBitWidth <= 64); + + /* for each of the possible cases we iterate over all bytes of the values + * build a set of possible values at each byte position in byteSets */ + for (CaseExpr &Case : Cases) { + + for (unsigned i = 0; i < BytesInValue; i++) { + + uint8_t byte = (Case.Val->getZExtValue() >> (i * 8)) & 0xFF; + byteSets[i].insert(byte); + + } + + } + + /* find the index of the first byte position that was not yet checked. then + * save the number of possible values at that byte position */ + unsigned smallestIndex = 0; + unsigned smallestSize = 257; + for (unsigned i = 0; i < byteSets.size(); i++) { + + if (bytesChecked[i]) continue; + if (byteSets[i].size() < smallestSize) { + + smallestIndex = i; + smallestSize = byteSets[i].size(); + + } + + } + + assert(bytesChecked[smallestIndex] == false); + + /* there are only smallestSize different bytes at index smallestIndex */ + + Instruction *Shift, *Trunc; + Function * F = OrigBlock->getParent(); + BasicBlock * NewNode = BasicBlock::Create(Val->getContext(), "NodeBlock", F); + Shift = BinaryOperator::Create(Instruction::LShr, Val, + ConstantInt::get(ValType, smallestIndex * 8)); + NewNode->getInstList().push_back(Shift); + + if (ValTypeBitWidth > 8) { + + Trunc = new TruncInst(Shift, ByteType); + NewNode->getInstList().push_back(Trunc); + + } else { + + /* not necessary to trunc */ + Trunc = Shift; + + } + + /* this is a trivial case, we can directly check for the byte, + * if the byte is not found go to default. if the byte was found + * mark the byte as checked. if this was the last byte to check + * we can finally execute the block belonging to this case */ + + if (smallestSize == 1) { + + uint8_t byte = *(byteSets[smallestIndex].begin()); + + /* insert instructions to check whether the value we are switching on is + * equal to byte */ + ICmpInst *Comp = + new ICmpInst(ICmpInst::ICMP_EQ, Trunc, ConstantInt::get(ByteType, byte), + "byteMatch"); + NewNode->getInstList().push_back(Comp); + + bytesChecked[smallestIndex] = true; + bool allBytesAreChecked = true; + + for (std::vector::iterator BCI = bytesChecked.begin(), + E = bytesChecked.end(); + BCI != E; ++BCI) { + + if (!*BCI) { + + allBytesAreChecked = false; + break; + + } + + } + + // if (std::all_of(bytesChecked.begin(), bytesChecked.end(), + // [](bool b) { return b; })) { + + if (allBytesAreChecked) { + + assert(Cases.size() == 1); + BranchInst::Create(Cases[0].BB, NewDefault, Comp, NewNode); + + /* we have to update the phi nodes! */ + for (BasicBlock::iterator I = Cases[0].BB->begin(); + I != Cases[0].BB->end(); ++I) { + + if (!isa(&*I)) { continue; } + PHINode *PN = cast(I); + + /* Only update the first occurrence. */ + unsigned Idx = 0, E = PN->getNumIncomingValues(); + for (; Idx != E; ++Idx) { + + if (PN->getIncomingBlock(Idx) == OrigBlock) { + + PN->setIncomingBlock(Idx, NewNode); + break; + + } + + } + + } + + } else { + + BasicBlock *BB = switchConvert(Cases, bytesChecked, OrigBlock, NewDefault, + Val, level + 1); + BranchInst::Create(BB, NewDefault, Comp, NewNode); + + } + + } + + /* there is no byte which we can directly check on, split the tree */ + else { + + std::vector byteVector; + std::copy(byteSets[smallestIndex].begin(), byteSets[smallestIndex].end(), + std::back_inserter(byteVector)); + std::sort(byteVector.begin(), byteVector.end()); + uint8_t pivot = byteVector[byteVector.size() / 2]; + + /* we already chose to divide the cases based on the value of byte at index + * smallestIndex the pivot value determines the threshold for the decicion; + * if a case value + * is smaller at this byte index move it to the LHS vector, otherwise to the + * RHS vector */ + + CaseVector LHSCases, RHSCases; + + for (CaseExpr &Case : Cases) { + + uint8_t byte = (Case.Val->getZExtValue() >> (smallestIndex * 8)) & 0xFF; + + if (byte < pivot) { + + LHSCases.push_back(Case); + + } else { + + RHSCases.push_back(Case); + + } + + } + + BasicBlock *LBB, *RBB; + LBB = switchConvert(LHSCases, bytesChecked, OrigBlock, NewDefault, Val, + level + 1); + RBB = switchConvert(RHSCases, bytesChecked, OrigBlock, NewDefault, Val, + level + 1); + + /* insert instructions to check whether the value we are switching on is + * equal to byte */ + ICmpInst *Comp = + new ICmpInst(ICmpInst::ICMP_ULT, Trunc, + ConstantInt::get(ByteType, pivot), "byteMatch"); + NewNode->getInstList().push_back(Comp); + BranchInst::Create(LBB, RBB, Comp, NewNode); + + } + + return NewNode; + +} + +bool SplitSwitchesTransform::splitSwitches(Module &M) { + +#if (LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR < 7) + LLVMContext &C = M.getContext(); +#endif + + std::vector switches; + + /* iterate over all functions, bbs and instruction and add + * all switches to switches vector for later processing */ + for (auto &F : M) { + + if (!isInInstrumentList(&F)) continue; + + for (auto &BB : F) { + + SwitchInst *switchInst = nullptr; + + if ((switchInst = dyn_cast(BB.getTerminator()))) { + + if (switchInst->getNumCases() < 1) continue; + switches.push_back(switchInst); + + } + + } + + } + + if (!switches.size()) return false; + if (!be_quiet) + errs() << "Rewriting " << switches.size() << " switch statements " + << "\n"; + + for (auto &SI : switches) { + + BasicBlock *CurBlock = SI->getParent(); + BasicBlock *OrigBlock = CurBlock; + Function * F = CurBlock->getParent(); + /* this is the value we are switching on */ + Value * Val = SI->getCondition(); + BasicBlock *Default = SI->getDefaultDest(); + unsigned bitw = Val->getType()->getIntegerBitWidth(); + + if (!be_quiet) + errs() << "switch: " << SI->getNumCases() << " cases " << bitw + << " bit\n"; + + /* If there is only the default destination or the condition checks 8 bit or + * less, don't bother with the code below. */ + if (!SI->getNumCases() || bitw <= 8) { + + if (!be_quiet) errs() << "skip trivial switch..\n"; + continue; + + } + + /* Create a new, empty default block so that the new hierarchy of + * if-then statements go to this and the PHI nodes are happy. + * if the default block is set as an unreachable we avoid creating one + * because will never be a valid target.*/ + BasicBlock *NewDefault = nullptr; + NewDefault = BasicBlock::Create(SI->getContext(), "NewDefault", F, Default); + BranchInst::Create(Default, NewDefault); + + /* Prepare cases vector. */ + CaseVector Cases; + for (SwitchInst::CaseIt i = SI->case_begin(), e = SI->case_end(); i != e; + ++i) +#if LLVM_VERSION_MAJOR < 5 + Cases.push_back(CaseExpr(i.getCaseValue(), i.getCaseSuccessor())); +#else + Cases.push_back(CaseExpr(i->getCaseValue(), i->getCaseSuccessor())); +#endif + /* bugfix thanks to pbst + * round up bytesChecked (in case getBitWidth() % 8 != 0) */ + std::vector bytesChecked((7 + Cases[0].Val->getBitWidth()) / 8, + false); + BasicBlock * SwitchBlock = + switchConvert(Cases, bytesChecked, OrigBlock, NewDefault, Val, 0); + + /* Branch to our shiny new if-then stuff... */ + BranchInst::Create(SwitchBlock, OrigBlock); + + /* We are now done with the switch instruction, delete it. */ + CurBlock->getInstList().erase(SI); + + /* we have to update the phi nodes! */ + for (BasicBlock::iterator I = Default->begin(); I != Default->end(); ++I) { + + if (!isa(&*I)) { continue; } + PHINode *PN = cast(I); + + /* Only update the first occurrence. */ + unsigned Idx = 0, E = PN->getNumIncomingValues(); + for (; Idx != E; ++Idx) { + + if (PN->getIncomingBlock(Idx) == OrigBlock) { + + PN->setIncomingBlock(Idx, NewDefault); + break; + + } + + } + + } + + } + + verifyModule(M); + return true; + +} + +bool SplitSwitchesTransform::runOnModule(Module &M) { + + if ((isatty(2) && getenv("AFL_QUIET") == NULL) || getenv("AFL_DEBUG") != NULL) + llvm::errs() << "Running split-switches-pass by laf.intel@gmail.com\n"; + else + be_quiet = 1; + splitSwitches(M); + verifyModule(M); + + return true; + +} + +static void registerSplitSwitchesTransPass(const PassManagerBuilder &, + legacy::PassManagerBase &PM) { + + auto p = new SplitSwitchesTransform(); + PM.add(p); + +} + +static RegisterStandardPasses RegisterSplitSwitchesTransPass( + PassManagerBuilder::EP_OptimizerLast, registerSplitSwitchesTransPass); + +static RegisterStandardPasses RegisterSplitSwitchesTransPass0( + PassManagerBuilder::EP_EnabledOnOptLevel0, registerSplitSwitchesTransPass); + +#if LLVM_VERSION_MAJOR >= 11 +static RegisterStandardPasses RegisterSplitSwitchesTransPassLTO( + PassManagerBuilder::EP_FullLinkTimeOptimizationLast, + registerSplitSwitchesTransPass); +#endif + diff --git a/llvm_mode/GNUmakefile b/llvm_mode/GNUmakefile deleted file mode 100644 index c14e8b4e..00000000 --- a/llvm_mode/GNUmakefile +++ /dev/null @@ -1,480 +0,0 @@ -# american fuzzy lop++ - LLVM instrumentation -# ----------------------------------------- -# -# Written by Laszlo Szekeres and -# Michal Zalewski -# -# LLVM integration design comes from Laszlo Szekeres. -# -# Copyright 2015, 2016 Google Inc. All rights reserved. -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at: -# -# http://www.apache.org/licenses/LICENSE-2.0 -# - -# For Heiko: -#TEST_MMAP=1 -HASH=\# - -PREFIX ?= /usr/local -HELPER_PATH ?= $(PREFIX)/lib/afl -BIN_PATH ?= $(PREFIX)/bin -DOC_PATH ?= $(PREFIX)/share/doc/afl -MISC_PATH ?= $(PREFIX)/share/afl -MAN_PATH ?= $(PREFIX)/share/man/man8 - -VERSION = $(shell grep '^$(HASH)define VERSION ' ../config.h | cut -d '"' -f2) - -BUILD_DATE ?= $(shell date -u -d "@$(SOURCE_DATE_EPOCH)" "+%Y-%m-%d" 2>/dev/null || date -u -r "$(SOURCE_DATE_EPOCH)" "+%Y-%m-%d" 2>/dev/null || date -u "+%Y-%m-%d") - -ifeq "$(shell uname)" "OpenBSD" - LLVM_CONFIG ?= $(BIN_PATH)/llvm-config - HAS_OPT = $(shell test -x $(BIN_PATH)/opt && echo 0 || echo 1) - ifeq "$(HAS_OPT)" "1" - $(error llvm_mode needs a complete llvm installation (versions 3.4 up to 12) -> e.g. "pkg_add llvm-7.0.1p9") - endif -else - LLVM_CONFIG ?= llvm-config -endif - -LLVMVER = $(shell $(LLVM_CONFIG) --version 2>/dev/null | sed 's/git//' ) -LLVM_UNSUPPORTED = $(shell $(LLVM_CONFIG) --version 2>/dev/null | egrep -q '^3\.[0-3]|^19' && echo 1 || echo 0 ) -LLVM_NEW_API = $(shell $(LLVM_CONFIG) --version 2>/dev/null | egrep -q '^1[0-9]' && echo 1 || echo 0 ) -LLVM_HAVE_LTO = $(shell $(LLVM_CONFIG) --version 2>/dev/null | egrep -q '^1[1-9]' && echo 1 || echo 0 ) -LLVM_MAJOR = $(shell $(LLVM_CONFIG) --version 2>/dev/null | sed 's/\..*//') -LLVM_BINDIR = $(shell $(LLVM_CONFIG) --bindir 2>/dev/null) -LLVM_LIBDIR = $(shell $(LLVM_CONFIG) --libdir 2>/dev/null) -LLVM_STDCXX = gnu++11 -LLVM_APPLE_XCODE = $(shell clang -v 2>&1 | grep -q Apple && echo 1 || echo 0) -LLVM_LTO = 0 - -ifeq "$(LLVMVER)" "" - $(warning [!] llvm_mode needs llvm-config, which was not found) -endif - -ifeq "$(LLVM_UNSUPPORTED)" "1" - $(warning llvm_mode only supports llvm versions 3.4 up to 12) -endif - -LLVM_TOO_OLD=1 - -ifeq "$(LLVM_MAJOR)" "9" - $(info [+] llvm_mode detected llvm 9, enabling neverZero implementation) - LLVM_TOO_OLD=0 -endif - -ifeq "$(LLVM_NEW_API)" "1" - $(info [+] llvm_mode detected llvm 10+, enabling neverZero implementation and c++14) - LLVM_STDCXX = c++14 - LLVM_TOO_OLD=0 -endif - -ifeq "$(LLVM_TOO_OLD)" "1" - $(info [!] llvm_mode detected an old version of llvm, upgrade to at least 9 or preferable 11!) - $(shell sleep 1) -endif - -ifeq "$(LLVM_HAVE_LTO)" "1" - $(info [+] llvm_mode detected llvm 11+, enabling afl-clang-lto LTO implementation) - LLVM_LTO = 1 - #TEST_MMAP = 1 -endif - -ifeq "$(LLVM_LTO)" "0" - $(info [+] llvm_mode detected llvm < 11, afl-clang-lto LTO will not be build.) -endif - -ifeq "$(LLVM_APPLE_XCODE)" "1" - $(warning llvm_mode will not compile with Xcode clang...) -endif - -# We were using llvm-config --bindir to get the location of clang, but -# this seems to be busted on some distros, so using the one in $PATH is -# probably better. - -CC = $(LLVM_BINDIR)/clang -CXX = $(LLVM_BINDIR)/clang++ - -# llvm-config --bindir may not providing a valid path, so ... -ifeq "$(shell test -e $(CC) || echo 1 )" "1" - # however we must ensure that this is not a "CC=gcc make" - ifeq "$(shell command -v $(CC) 2> /dev/null)" "" - # we do not have a valid CC variable so we try alternatives - ifeq "$(shell test -e '$(BIN_DIR)/clang' && echo 1)" "1" - # we found one in the local install directory, lets use these - CC = $(BIN_DIR)/clang - else - # hope for the best - $(warning we have trouble finding clang - llvm-config is not helping us) - CC = clang - endif - endif -endif -# llvm-config --bindir may not providing a valid path, so ... -ifeq "$(shell test -e $(CXX) || echo 1 )" "1" - # however we must ensure that this is not a "CC=gcc make" - ifeq "$(shell command -v $(CXX) 2> /dev/null)" "" - # we do not have a valid CC variable so we try alternatives - ifeq "$(shell test -e '$(BIN_DIR)/clang++' && echo 1)" "1" - # we found one in the local install directory, lets use these - CXX = $(BIN_DIR)/clang++ - else - # hope for the best - $(warning we have trouble finding clang++ - llvm-config is not helping us) - CXX = clang++ - endif - endif -endif - -# sanity check. -# Are versions of clang --version and llvm-config --version equal? -CLANGVER = $(shell $(CC) --version | sed -E -ne '/^.*version\ (1?[0-9]\.[0-9]\.[0-9]).*/s//\1/p') - -# I disable this because it does not make sense with what we did before (marc) -# We did exactly set these 26 lines above with these values, and it would break -# "CC=gcc make" etc. usages -ifeq "$(findstring clang, $(shell $(CC) --version 2>/dev/null))" "" - CC_SAVE := $(LLVM_BINDIR)/clang -else - CC_SAVE := $(CC) -endif -ifeq "$(findstring clang, $(shell $(CXX) --version 2>/dev/null))" "" - CXX_SAVE := $(LLVM_BINDIR)/clang++ -else - CXX_SAVE := $(CXX) -endif - -CLANG_BIN := $(CC_SAVE) -CLANGPP_BIN := $(CXX_SAVE) - -ifeq "$(CC_SAVE)" "$(LLVM_BINDIR)/clang" - USE_BINDIR = 1 -else - ifeq "$(CXX_SAVE)" "$(LLVM_BINDIR)/clang++" - USE_BINDIR = 1 - else - USE_BINDIR = 0 - endif -endif - -# On old platform we cannot compile with clang because std++ libraries are too -# old. For these we need to use gcc/g++, so if we find REAL_CC and REAL_CXX -# variable we override the compiler variables here -ifneq "$(REAL_CC)" "" -CC = $(REAL_CC) -endif -ifneq "$(REAL_CXX)" "" -CXX = $(REAL_CXX) -endif - -# After we set CC/CXX we can start makefile magic tests - -#ifeq "$(shell echo 'int main() {return 0; }' | $(CC) -x c - -march=native -o .test 2>/dev/null && echo 1 || echo 0 ; rm -f .test )" "1" -# CFLAGS_OPT = -march=native -#endif - -ifeq "$(shell echo 'int main() {return 0; }' | $(CLANG_BIN) -x c - -flto=full -o .test 2>/dev/null && echo 1 || echo 0 ; rm -f .test )" "1" - AFL_CLANG_FLTO ?= -flto=full -else - ifeq "$(shell echo 'int main() {return 0; }' | $(CLANG_BIN) -x c - -flto=thin -o .test 2>/dev/null && echo 1 || echo 0 ; rm -f .test )" "1" - AFL_CLANG_FLTO ?= -flto=thin - else - ifeq "$(shell echo 'int main() {return 0; }' | $(CLANG_BIN) -x c - -flto -o .test 2>/dev/null && echo 1 || echo 0 ; rm -f .test )" "1" - AFL_CLANG_FLTO ?= -flto - endif - endif -endif - -ifeq "$(LLVM_LTO)" "1" - ifneq "$(AFL_CLANG_FLTO)" "" - ifeq "$(AFL_REAL_LD)" "" - ifneq "$(shell readlink $(LLVM_BINDIR)/ld.lld 2>&1)" "" - AFL_REAL_LD = $(LLVM_BINDIR)/ld.lld - else - $(warn ld.lld not found, cannot enable LTO mode) - LLVM_LTO = 0 - endif - endif - else - $(warn clang option -flto is not working - maybe LLVMgold.so not found - cannot enable LTO mode) - LLVM_LTO = 0 - endif -endif - -AFL_CLANG_FUSELD= -ifeq "$(LLVM_LTO)" "1" - ifeq "$(shell echo 'int main() {return 0; }' | $(CLANG_BIN) -x c - -fuse-ld=`command -v ld` -o .test 2>/dev/null && echo 1 || echo 0 ; rm -f .test )" "1" - AFL_CLANG_FUSELD=1 - ifeq "$(shell echo 'int main() {return 0; }' | $(CLANG_BIN) -x c - -fuse-ld=ld.lld --ld-path=$(LLVM_BINDIR)/ld.lld -o .test 2>/dev/null && echo 1 || echo 0 ; rm -f .test )" "1" - AFL_CLANG_LDPATH=1 - endif - else - $(warn -fuse-ld is not working, cannot enable LTO mode) - LLVM_LTO = 0 - endif -endif - -ifeq "$(shell echo 'int main() {return 0; }' | $(CLANG_BIN) -x c - -fdebug-prefix-map=$(CURDIR)=llvm_mode -o .test 2>/dev/null && echo 1 || echo 0 ; rm -f .test )" "1" - AFL_CLANG_DEBUG_PREFIX = -fdebug-prefix-map="$(CURDIR)=llvm_mode" -else - AFL_CLANG_DEBUG_PREFIX = "" -endif - -CFLAGS ?= -O3 -funroll-loops -fPIC -D_FORTIFY_SOURCE=2 -CFLAGS_SAFE := -Wall -g -Wno-pointer-sign -I ../include/ \ - -DAFL_PATH=\"$(HELPER_PATH)\" -DBIN_PATH=\"$(BIN_PATH)\" \ - -DLLVM_BINDIR=\"$(LLVM_BINDIR)\" -DVERSION=\"$(VERSION)\" \ - -DLLVM_LIBDIR=\"$(LLVM_LIBDIR)\" -DLLVM_VERSION=\"$(LLVMVER)\" \ - -DAFL_CLANG_FLTO=\"$(AFL_CLANG_FLTO)\" \ - -DAFL_REAL_LD=\"$(AFL_REAL_LD)\" \ - -DAFL_CLANG_LDPATH=\"$(AFL_CLANG_LDPATH)\" \ - -DAFL_CLANG_FUSELD=\"$(AFL_CLANG_FUSELD)\" \ - -DCLANG_BIN=\"$(CLANG_BIN)\" -DCLANGPP_BIN=\"$(CLANGPP_BIN)\" -DUSE_BINDIR=$(USE_BINDIR) -Wno-unused-function \ - $(AFL_CLANG_DEBUG_PREFIX) -override CFLAGS += $(CFLAGS_SAFE) - -ifdef AFL_TRACE_PC - $(info Compile option AFL_TRACE_PC is deprecated, just set AFL_LLVM_INSTRUMENT=PCGUARD to activate when compiling targets ) -endif - -CXXFLAGS ?= -O3 -funroll-loops -fPIC -D_FORTIFY_SOURCE=2 -override CXXFLAGS += -Wall -g -I ../include/ \ - -DVERSION=\"$(VERSION)\" -Wno-variadic-macros - -ifneq "$(shell $(LLVM_CONFIG) --includedir) 2> /dev/null" "" - CLANG_CFL = -I$(shell $(LLVM_CONFIG) --includedir) -endif -ifneq "$(LLVM_CONFIG)" "" - CLANG_CFL += -I$(shell dirname $(LLVM_CONFIG))/../include -endif -CLANG_CPPFL = `$(LLVM_CONFIG) --cxxflags` -fno-rtti -fPIC $(CXXFLAGS) -Wno-deprecated-declarations -CLANG_LFL = `$(LLVM_CONFIG) --ldflags` $(LDFLAGS) - - -# User teor2345 reports that this is required to make things work on MacOS X. -ifeq "$(shell uname)" "Darwin" - CLANG_LFL += -Wl,-flat_namespace -Wl,-undefined,suppress -else - CLANG_CPPFL += -Wl,-znodelete -endif - -ifeq "$(shell uname)" "OpenBSD" - CLANG_LFL += `$(LLVM_CONFIG) --libdir`/libLLVM.so - CLANG_CPPFL += -mno-retpoline - CFLAGS += -mno-retpoline - # Needed for unwind symbols - LDFLAGS += -lc++abi -endif - -ifeq "$(shell echo '$(HASH)include @$(HASH)include @int main() { int _id = shmget(IPC_PRIVATE, 65536, IPC_CREAT | IPC_EXCL | 0600); shmctl(_id, IPC_RMID, 0); return 0;}' | tr @ '\n' | $(CC) -x c - -o .test2 2>/dev/null && echo 1 || echo 0 ; rm -f .test2 )" "1" - SHMAT_OK=1 -else - SHMAT_OK=0 - #CFLAGS+=-DUSEMMAP=1 - LDFLAGS += -Wno-deprecated-declarations -endif - -ifeq "$(TEST_MMAP)" "1" - SHMAT_OK=0 - CFLAGS+=-DUSEMMAP=1 - LDFLAGS += -Wno-deprecated-declarations -endif - -PROGS = ../afl-clang-fast ../afl-llvm-pass.so ../afl-ld-lto ../afl-llvm-lto-instrumentlist.so ../afl-llvm-lto-instrumentation.so ../libLLVMInsTrim.so ../afl-llvm-rt.o ../afl-llvm-rt-32.o ../afl-llvm-rt-64.o ../compare-transform-pass.so ../split-compares-pass.so ../split-switches-pass.so ../cmplog-routines-pass.so ../cmplog-instructions-pass.so ../SanitizerCoverageLTO.so - -# If prerequisites are not given, warn, do not build anything, and exit with code 0 -ifeq "$(LLVMVER)" "" - NO_BUILD = 1 -endif - -ifneq "$(LLVM_UNSUPPORTED)$(LLVM_APPLE_XCODE)" "00" - NO_BUILD = 1 -endif - -ifeq "$(NO_BUILD)" "1" - TARGETS = no_build -else - TARGETS = test_shm test_deps $(PROGS) afl-clang-fast.8 test_build all_done -endif - -LLVM_MIN_4_0_1 = $(shell awk 'function tonum(ver, a) {split(ver,a,"."); return a[1]*1000000+a[2]*1000+a[3]} BEGIN { exit tonum(ARGV[1]) >= tonum(ARGV[2]) }' $(LLVMVER) 4.0.1; echo $$?) - -all: $(TARGETS) - -ifeq "$(SHMAT_OK)" "1" - -test_shm: - @echo "[+] shmat seems to be working." - @rm -f .test2 - -else - -test_shm: - @echo "[-] shmat seems not to be working, switching to mmap implementation" - -endif - -no_build: - @printf "%b\\n" "\\033[0;31mPrerequisites are not met, skipping build llvm_mode\\033[0m" - -test_deps: - @echo "[*] Checking for working 'llvm-config'..." - ifneq "$(LLVM_APPLE_XCODE)" "1" - @type $(LLVM_CONFIG) >/dev/null 2>&1 || ( echo "[-] Oops, can't find 'llvm-config'. Install clang or set \$$LLVM_CONFIG or \$$PATH beforehand."; echo " (Sometimes, the binary will be named llvm-config-3.5 or something like that.)"; exit 1 ) - endif - @echo "[*] Checking for working '$(CC)'..." - @type $(CC) >/dev/null 2>&1 || ( echo "[-] Oops, can't find '$(CC)'. Make sure that it's in your \$$PATH (or set \$$CC and \$$CXX)."; exit 1 ) - @echo "[*] Checking for matching versions of '$(CC)' and '$(LLVM_CONFIG)'" -ifneq "$(CLANGVER)" "$(LLVMVER)" - @echo "[!] WARNING: we have llvm-config version $(LLVMVER) and a clang version $(CLANGVER)" -else - @echo "[*] We have llvm-config version $(LLVMVER) with a clang version $(CLANGVER), good." -endif - @echo "[*] Checking for '../afl-showmap'..." - @test -f ../afl-showmap || ( echo "[-] Oops, can't find '../afl-showmap'. Be sure to compile AFL first."; exit 1 ) - @echo "[+] All set and ready to build." - -afl-common.o: ../src/afl-common.c - $(CC) $(CFLAGS) $(CPPFLAGS) -c $< -o $@ $(LDFLAGS) - -../afl-clang-fast: afl-clang-fast.c afl-common.o | test_deps - $(CC) $(CLANG_CFL) $(CFLAGS) $(CPPFLAGS) $< afl-common.o -o $@ $(LDFLAGS) -DCFLAGS_OPT=\"$(CFLAGS_OPT)\" - ln -sf afl-clang-fast ../afl-clang-fast++ -ifneq "$(AFL_CLANG_FLTO)" "" -ifeq "$(LLVM_LTO)" "1" - ln -sf afl-clang-fast ../afl-clang-lto - ln -sf afl-clang-fast ../afl-clang-lto++ -endif -endif - -afl-llvm-common.o: afl-llvm-common.cc afl-llvm-common.h - $(CXX) $(CFLAGS) $(CPPFLAGS) `$(LLVM_CONFIG) --cxxflags` -fno-rtti -fPIC -std=$(LLVM_STDCXX) -c $< -o $@ - -../libLLVMInsTrim.so: LLVMInsTrim.so.cc MarkNodes.cc afl-llvm-common.o | test_deps - -$(CXX) $(CLANG_CPPFL) -DLLVMInsTrim_EXPORTS -fno-rtti -fPIC -std=$(LLVM_STDCXX) -shared $< MarkNodes.cc -o $@ $(CLANG_LFL) afl-llvm-common.o - -../afl-llvm-pass.so: afl-llvm-pass.so.cc afl-llvm-common.o | test_deps -ifeq "$(LLVM_MIN_4_0_1)" "0" - $(info [!] N-gram branch coverage instrumentation is not available for llvm version $(LLVMVER)) -endif - $(CXX) $(CLANG_CPPFL) -DLLVMInsTrim_EXPORTS -fno-rtti -fPIC -std=$(LLVM_STDCXX) -shared $< -o $@ $(CLANG_LFL) afl-llvm-common.o - -../afl-llvm-lto-instrumentlist.so: afl-llvm-lto-instrumentlist.so.cc afl-llvm-common.o -ifeq "$(LLVM_LTO)" "1" - $(CXX) $(CLANG_CPPFL) -fno-rtti -fPIC -std=$(LLVM_STDCXX) -shared $< -o $@ $(CLANG_LFL) afl-llvm-common.o -endif - -../afl-ld-lto: afl-ld-lto.c -ifeq "$(LLVM_LTO)" "1" - $(CC) $(CFLAGS) $(CPPFLAGS) $< -o $@ -endif - -../SanitizerCoverageLTO.so: SanitizerCoverageLTO.so.cc -ifeq "$(LLVM_LTO)" "1" - $(CXX) $(CLANG_CPPFL) -Wno-writable-strings -fno-rtti -fPIC -std=$(LLVM_STDCXX) -shared $< -o $@ $(CLANG_LFL) afl-llvm-common.o -endif - -../afl-llvm-lto-instrumentation.so: afl-llvm-lto-instrumentation.so.cc afl-llvm-common.o -ifeq "$(LLVM_LTO)" "1" - $(CXX) $(CLANG_CPPFL) -Wno-writable-strings -fno-rtti -fPIC -std=$(LLVM_STDCXX) -shared $< -o $@ $(CLANG_LFL) afl-llvm-common.o - $(CLANG_BIN) $(CFLAGS_SAFE) $(CPPFLAGS) -Wno-unused-result -O0 $(AFL_CLANG_FLTO) -fPIC -c afl-llvm-rt-lto.o.c -o ../afl-llvm-rt-lto.o - @$(CLANG_BIN) $(CFLAGS_SAFE) $(CPPFLAGS) -Wno-unused-result -O0 $(AFL_CLANG_FLTO) -m64 -fPIC -c afl-llvm-rt-lto.o.c -o ../afl-llvm-rt-lto-64.o 2>/dev/null; if [ "$$?" = "0" ]; then : ; fi - @$(CLANG_BIN) $(CFLAGS_SAFE) $(CPPFLAGS) -Wno-unused-result -O0 $(AFL_CLANG_FLTO) -m32 -fPIC -c afl-llvm-rt-lto.o.c -o ../afl-llvm-rt-lto-32.o 2>/dev/null; if [ "$$?" = "0" ]; then : ; fi -endif - -# laf -../split-switches-pass.so: split-switches-pass.so.cc afl-llvm-common.o | test_deps - $(CXX) $(CLANG_CPPFL) -shared $< -o $@ $(CLANG_LFL) afl-llvm-common.o -../compare-transform-pass.so: compare-transform-pass.so.cc afl-llvm-common.o | test_deps - $(CXX) $(CLANG_CPPFL) -shared $< -o $@ $(CLANG_LFL) afl-llvm-common.o -../split-compares-pass.so: split-compares-pass.so.cc afl-llvm-common.o | test_deps - $(CXX) $(CLANG_CPPFL) -shared $< -o $@ $(CLANG_LFL) afl-llvm-common.o -# /laf - -../cmplog-routines-pass.so: cmplog-routines-pass.cc afl-llvm-common.o | test_deps - $(CXX) $(CLANG_CPPFL) -shared $< -o $@ $(CLANG_LFL) afl-llvm-common.o - -../cmplog-instructions-pass.so: cmplog-instructions-pass.cc afl-llvm-common.o | test_deps - $(CXX) $(CLANG_CPPFL) -shared $< -o $@ $(CLANG_LFL) afl-llvm-common.o - -document: - $(CLANG_BIN) -D_AFL_DOCUMENT_MUTATIONS $(CFLAGS_SAFE) $(CPPFLAGS) $(CLANG_CFL) -O3 -Wno-unused-result -fPIC -c afl-llvm-rt.o.c -o ../afl-llvm-rt.o - @$(CLANG_BIN) -D_AFL_DOCUMENT_MUTATIONS $(CFLAGS_SAFE) $(CPPFLAGS) $(CLANG_CFL) -O3 -Wno-unused-result -m32 -fPIC -c afl-llvm-rt.o.c -o ../afl-llvm-rt-32.o 2>/dev/null; if [ "$$?" = "0" ]; then echo "success!"; else echo "failed (that's fine)"; fi - @$(CLANG_BIN) -D_AFL_DOCUMENT_MUTATIONS $(CFLAGS_SAFE) $(CPPFLAGS) $(CLANG_CFL) -O3 -Wno-unused-result -m64 -fPIC -c afl-llvm-rt.o.c -o ../afl-llvm-rt-64.o 2>/dev/null; if [ "$$?" = "0" ]; then echo "success!"; else echo "failed (that's fine)"; fi - -../afl-llvm-rt.o: afl-llvm-rt.o.c | test_deps - $(CLANG_BIN) $(CLANG_CFL) $(CFLAGS_SAFE) $(CPPFLAGS) -O3 -Wno-unused-result -fPIC -c $< -o $@ - -../afl-llvm-rt-32.o: afl-llvm-rt.o.c | test_deps - @printf "[*] Building 32-bit variant of the runtime (-m32)... " - @$(CLANG_BIN) $(CLANG_CFL) $(CFLAGS_SAFE) $(CPPFLAGS) -O3 -Wno-unused-result -m32 -fPIC -c $< -o $@ 2>/dev/null; if [ "$$?" = "0" ]; then echo "success!"; else echo "failed (that's fine)"; fi - -../afl-llvm-rt-64.o: afl-llvm-rt.o.c | test_deps - @printf "[*] Building 64-bit variant of the runtime (-m64)... " - @$(CLANG_BIN) $(CLANG_CFL) $(CFLAGS_SAFE) $(CPPFLAGS) -O3 -Wno-unused-result -m64 -fPIC -c $< -o $@ 2>/dev/null; if [ "$$?" = "0" ]; then echo "success!"; else echo "failed (that's fine)"; fi - -test_build: $(PROGS) - @echo "[*] Testing the CC wrapper and instrumentation output..." - unset AFL_USE_ASAN AFL_USE_MSAN AFL_INST_RATIO; AFL_QUIET=1 AFL_PATH=. AFL_LLVM_LAF_SPLIT_SWITCHES=1 AFL_LLVM_LAF_TRANSFORM_COMPARES=1 AFL_LLVM_LAF_SPLIT_COMPARES=1 ../afl-clang-fast $(CFLAGS) ../test-instr.c -o test-instr $(LDFLAGS) - ASAN_OPTIONS=detect_leaks=0 ../afl-showmap -m none -q -o .test-instr0 ./test-instr < /dev/null - echo 1 | ASAN_OPTIONS=detect_leaks=0 ../afl-showmap -m none -q -o .test-instr1 ./test-instr - @rm -f test-instr - @cmp -s .test-instr0 .test-instr1; DR="$$?"; rm -f .test-instr0 .test-instr1; if [ "$$DR" = "0" ]; then echo; echo "Oops, the instrumentation does not seem to be behaving correctly!"; echo; echo "Please post to https://github.com/AFLplusplus/AFLplusplus/issues to troubleshoot the issue."; echo; exit 1; fi - @echo "[+] All right, the instrumentation seems to be working!" - -all_done: test_build - @echo "[+] All done! You can now use '../afl-clang-fast' to compile programs." - -.NOTPARALLEL: clean - -install: all - install -d -m 755 $${DESTDIR}$(BIN_PATH) $${DESTDIR}$(HELPER_PATH) $${DESTDIR}$(DOC_PATH) $${DESTDIR}$(MISC_PATH) - if [ -f ../afl-clang-fast -a -f ../libLLVMInsTrim.so -a -f ../afl-llvm-rt.o ]; then set -e; install -m 755 ../afl-clang-fast $${DESTDIR}$(BIN_PATH); ln -sf afl-clang-fast $${DESTDIR}$(BIN_PATH)/afl-clang-fast++; install -m 755 ../libLLVMInsTrim.so ../afl-llvm-pass.so ../afl-llvm-rt.o $${DESTDIR}$(HELPER_PATH); fi - if [ -f ../afl-clang-lto ]; then set -e; ln -sf afl-clang-fast $${DESTDIR}$(BIN_PATH)/afl-clang-lto; ln -sf afl-clang-fast $${DESTDIR}$(BIN_PATH)/afl-clang-lto++; install -m 755 ../afl-llvm-lto-instrumentation.so ../afl-llvm-rt-lto*.o ../afl-llvm-lto-instrumentlist.so $${DESTDIR}$(HELPER_PATH); fi - if [ -f ../afl-ld-lto ]; then set -e; install -m 755 ../afl-ld-lto $${DESTDIR}$(BIN_PATH); fi - if [ -f ../afl-llvm-rt-32.o ]; then set -e; install -m 755 ../afl-llvm-rt-32.o $${DESTDIR}$(HELPER_PATH); fi - if [ -f ../afl-llvm-rt-64.o ]; then set -e; install -m 755 ../afl-llvm-rt-64.o $${DESTDIR}$(HELPER_PATH); fi - if [ -f ../compare-transform-pass.so ]; then set -e; install -m 755 ../compare-transform-pass.so $${DESTDIR}$(HELPER_PATH); fi - if [ -f ../split-compares-pass.so ]; then set -e; install -m 755 ../split-compares-pass.so $${DESTDIR}$(HELPER_PATH); fi - if [ -f ../split-switches-pass.so ]; then set -e; install -m 755 ../split-switches-pass.so $${DESTDIR}$(HELPER_PATH); fi - if [ -f ../cmplog-instructions-pass.so ]; then set -e; install -m 755 ../cmplog-*-pass.so $${DESTDIR}$(HELPER_PATH); fi - if [ -f ../SanitizerCoverageLTO.so ]; then set -e; install -m 755 ../SanitizerCoverageLTO.so $${DESTDIR}$(HELPER_PATH); fi - set -e; install -m 644 ../dynamic_list.txt $${DESTDIR}$(HELPER_PATH) - set -e; if [ -f ../afl-clang-fast ] ; then ln -sf afl-clang-fast $${DESTDIR}$(BIN_PATH)/afl-clang ; ln -sf afl-clang-fast $${DESTDIR}$(BIN_PATH)/afl-clang++ ; else ln -sf afl-gcc $${DESTDIR}$(BIN_PATH)/afl-clang ; ln -sf afl-gcc $${DESTDIR}$(BIN_PATH)/afl-clang++; fi - install -m 644 README.*.md $${DESTDIR}$(DOC_PATH)/ - install -m 644 README.md $${DESTDIR}$(DOC_PATH)/README.llvm_mode.md - -vpath % .. -%.8: % - @echo .TH $* 8 $(BUILD_DATE) "afl++" > ../$@ - @echo .SH NAME >> ../$@ - @echo -n ".B $* \- " >> ../$@ - @../$* -h 2>&1 | head -n 1 | sed -e "s/$$(printf '\e')[^m]*m//g" >> ../$@ - @echo >> ../$@ - @echo .SH SYNOPSIS >> ../$@ - @../$* -h 2>&1 | head -n 3 | tail -n 1 | sed 's/^\.\///' >> ../$@ - @echo >> ../$@ - @echo .SH OPTIONS >> ../$@ - @echo .nf >> ../$@ - @../$* -h 2>&1 | tail -n +4 >> ../$@ - @echo >> ../$@ - @echo .SH AUTHOR >> ../$@ - @echo "afl++ was written by Michal \"lcamtuf\" Zalewski and is maintained by Marc \"van Hauser\" Heuse , Heiko \"hexcoder-\" Eissfeldt , Andrea Fioraldi and Dominik Maier " >> ../$@ - @echo The homepage of afl++ is: https://github.com/AFLplusplus/AFLplusplus >> ../$@ - @echo >> ../$@ - @echo .SH LICENSE >> ../$@ - @echo Apache License Version 2.0, January 2004 >> ../$@ - ln -sf afl-clang-fast.8 ../afl-clang-fast++.8 -ifneq "$(AFL_CLANG_FLTO)" "" -ifeq "$(LLVM_LTO)" "1" - ln -sf afl-clang-fast.8 ../afl-clang-lto.8 - ln -sf afl-clang-fast.8 ../afl-clang-lto++.8 -endif -endif - -clean: - rm -f *.o *.so *~ a.out core core.[1-9][0-9]* .test2 test-instr .test-instr0 .test-instr1 *.dwo - rm -f $(PROGS) afl-common.o ../afl-clang-fast++ ../afl-clang-lto ../afl-clang-lto++ ../afl-clang*.8 ../ld ../afl-ld ../afl-llvm-rt*.o diff --git a/llvm_mode/LLVMInsTrim.so.cc b/llvm_mode/LLVMInsTrim.so.cc deleted file mode 100644 index 61a420ba..00000000 --- a/llvm_mode/LLVMInsTrim.so.cc +++ /dev/null @@ -1,598 +0,0 @@ -#include -#include -#include -#include - -#include "llvm/Config/llvm-config.h" -#if LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR < 5 -typedef long double max_align_t; -#endif - -#include "llvm/ADT/DenseMap.h" -#include "llvm/ADT/DenseSet.h" -#if LLVM_VERSION_MAJOR > 3 || \ - (LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR > 4) - #include "llvm/IR/CFG.h" - #include "llvm/IR/Dominators.h" - #include "llvm/IR/DebugInfo.h" -#else - #include "llvm/Support/CFG.h" - #include "llvm/Analysis/Dominators.h" - #include "llvm/DebugInfo.h" -#endif -#include "llvm/IR/IRBuilder.h" -#include "llvm/IR/Instructions.h" -#include "llvm/IR/LegacyPassManager.h" -#include "llvm/IR/Module.h" -#include "llvm/Pass.h" -#include "llvm/Support/raw_ostream.h" -#include "llvm/Support/CommandLine.h" -#include "llvm/Transforms/IPO/PassManagerBuilder.h" -#include "llvm/Transforms/Utils/BasicBlockUtils.h" -#include "llvm/IR/BasicBlock.h" -#include -#include -#include -#include -#include - -#include "MarkNodes.h" -#include "afl-llvm-common.h" -#include "llvm-ngram-coverage.h" - -#include "config.h" -#include "debug.h" - -using namespace llvm; - -static cl::opt MarkSetOpt("markset", cl::desc("MarkSet"), - cl::init(false)); -static cl::opt LoopHeadOpt("loophead", cl::desc("LoopHead"), - cl::init(false)); - -namespace { - -struct InsTrim : public ModulePass { - - protected: - uint32_t function_minimum_size = 1; - char * skip_nozero = NULL; - - private: - std::mt19937 generator; - int total_instr = 0; - - unsigned int genLabel() { - - return generator() & (MAP_SIZE - 1); - - } - - public: - static char ID; - - InsTrim() : ModulePass(ID), generator(0) { - - initInstrumentList(); - - } - - void getAnalysisUsage(AnalysisUsage &AU) const override { - - AU.addRequired(); - - } - -#if LLVM_VERSION_MAJOR < 4 - const char * -#else - StringRef -#endif - getPassName() const override { - - return "InstTrim Instrumentation"; - - } - -#if LLVM_VERSION_MAJOR > 4 || \ - (LLVM_VERSION_MAJOR == 4 && LLVM_VERSION_PATCH >= 1) - #define AFL_HAVE_VECTOR_INTRINSICS 1 -#endif - - bool runOnModule(Module &M) override { - - setvbuf(stdout, NULL, _IONBF, 0); - - if ((isatty(2) && !getenv("AFL_QUIET")) || getenv("AFL_DEBUG") != NULL) { - - SAYF(cCYA "LLVMInsTrim" VERSION cRST " by csienslab\n"); - - } else - - be_quiet = 1; - - if (getenv("AFL_DEBUG") != NULL) debug = 1; - - LLVMContext &C = M.getContext(); - - IntegerType *Int8Ty = IntegerType::getInt8Ty(C); - IntegerType *Int32Ty = IntegerType::getInt32Ty(C); - -#if LLVM_VERSION_MAJOR < 9 - char *neverZero_counters_str; - if ((neverZero_counters_str = getenv("AFL_LLVM_NOT_ZERO")) != NULL) - if (!be_quiet) OKF("LLVM neverZero activated (by hexcoder)\n"); -#endif - skip_nozero = getenv("AFL_LLVM_SKIP_NEVERZERO"); - - if (getenv("AFL_LLVM_INSTRIM_LOOPHEAD") != NULL || - getenv("LOOPHEAD") != NULL) { - - LoopHeadOpt = true; - - } - - unsigned int PrevLocSize = 0; - char * ngram_size_str = getenv("AFL_LLVM_NGRAM_SIZE"); - if (!ngram_size_str) ngram_size_str = getenv("AFL_NGRAM_SIZE"); - char *ctx_str = getenv("AFL_LLVM_CTX"); - -#ifdef AFL_HAVE_VECTOR_INTRINSICS - unsigned int ngram_size = 0; - /* Decide previous location vector size (must be a power of two) */ - VectorType *PrevLocTy = NULL; - - if (ngram_size_str) - if (sscanf(ngram_size_str, "%u", &ngram_size) != 1 || ngram_size < 2 || - ngram_size > NGRAM_SIZE_MAX) - FATAL( - "Bad value of AFL_NGRAM_SIZE (must be between 2 and NGRAM_SIZE_MAX " - "(%u))", - NGRAM_SIZE_MAX); - - if (ngram_size) - PrevLocSize = ngram_size - 1; - else -#else - if (ngram_size_str) - #ifdef LLVM_VERSION_STRING - FATAL( - "Sorry, NGRAM branch coverage is not supported with llvm version %s!", - LLVM_VERSION_STRING); - #else - #ifndef LLVM_VERSION_PATCH - FATAL( - "Sorry, NGRAM branch coverage is not supported with llvm version " - "%d.%d.%d!", - LLVM_VERSION_MAJOR, LLVM_VERSION_MINOR, 0); - #else - FATAL( - "Sorry, NGRAM branch coverage is not supported with llvm version " - "%d.%d.%d!", - LLVM_VERSION_MAJOR, LLVM_VERSION_MINOR, LLVM_VERISON_PATCH); - #endif - #endif -#endif - PrevLocSize = 1; - -#ifdef AFL_HAVE_VECTOR_INTRINSICS - // IntegerType *Int64Ty = IntegerType::getInt64Ty(C); - int PrevLocVecSize = PowerOf2Ceil(PrevLocSize); - IntegerType *IntLocTy = - IntegerType::getIntNTy(C, sizeof(PREV_LOC_T) * CHAR_BIT); - if (ngram_size) - PrevLocTy = VectorType::get(IntLocTy, PrevLocVecSize - #if LLVM_VERSION_MAJOR >= 12 - , - false - #endif - ); -#endif - - /* Get globals for the SHM region and the previous location. Note that - __afl_prev_loc is thread-local. */ - - GlobalVariable *AFLMapPtr = - new GlobalVariable(M, PointerType::get(Int8Ty, 0), false, - GlobalValue::ExternalLinkage, 0, "__afl_area_ptr"); - GlobalVariable *AFLPrevLoc; - GlobalVariable *AFLContext = NULL; - LoadInst * PrevCtx = NULL; // for CTX sensitive coverage - - if (ctx_str) -#ifdef __ANDROID__ - AFLContext = new GlobalVariable( - M, Int32Ty, false, GlobalValue::ExternalLinkage, 0, "__afl_prev_ctx"); -#else - AFLContext = new GlobalVariable( - M, Int32Ty, false, GlobalValue::ExternalLinkage, 0, "__afl_prev_ctx", - 0, GlobalVariable::GeneralDynamicTLSModel, 0, false); -#endif - -#ifdef AFL_HAVE_VECTOR_INTRINSICS - if (ngram_size) - #ifdef __ANDROID__ - AFLPrevLoc = new GlobalVariable( - M, PrevLocTy, /* isConstant */ false, GlobalValue::ExternalLinkage, - /* Initializer */ nullptr, "__afl_prev_loc"); - #else - AFLPrevLoc = new GlobalVariable( - M, PrevLocTy, /* isConstant */ false, GlobalValue::ExternalLinkage, - /* Initializer */ nullptr, "__afl_prev_loc", - /* InsertBefore */ nullptr, GlobalVariable::GeneralDynamicTLSModel, - /* AddressSpace */ 0, /* IsExternallyInitialized */ false); - #endif - else -#endif -#ifdef __ANDROID__ - AFLPrevLoc = new GlobalVariable( - M, Int32Ty, false, GlobalValue::ExternalLinkage, 0, "__afl_prev_loc"); -#else - AFLPrevLoc = new GlobalVariable( - M, Int32Ty, false, GlobalValue::ExternalLinkage, 0, "__afl_prev_loc", 0, - GlobalVariable::GeneralDynamicTLSModel, 0, false); -#endif - -#ifdef AFL_HAVE_VECTOR_INTRINSICS - /* Create the vector shuffle mask for updating the previous block history. - Note that the first element of the vector will store cur_loc, so just set - it to undef to allow the optimizer to do its thing. */ - - SmallVector PrevLocShuffle = {UndefValue::get(Int32Ty)}; - - for (unsigned I = 0; I < PrevLocSize - 1; ++I) - PrevLocShuffle.push_back(ConstantInt::get(Int32Ty, I)); - - for (int I = PrevLocSize; I < PrevLocVecSize; ++I) - PrevLocShuffle.push_back(ConstantInt::get(Int32Ty, PrevLocSize)); - - Constant *PrevLocShuffleMask = ConstantVector::get(PrevLocShuffle); -#endif - - // this is our default - MarkSetOpt = true; - - ConstantInt *Zero = ConstantInt::get(Int8Ty, 0); - ConstantInt *One = ConstantInt::get(Int8Ty, 1); - - u64 total_rs = 0; - u64 total_hs = 0; - - scanForDangerousFunctions(&M); - - for (Function &F : M) { - - if (debug) { - - uint32_t bb_cnt = 0; - - for (auto &BB : F) - if (BB.size() > 0) ++bb_cnt; - SAYF(cMGN "[D] " cRST "Function %s size %zu %u\n", - F.getName().str().c_str(), F.size(), bb_cnt); - - } - - if (!isInInstrumentList(&F)) continue; - - // if the function below our minimum size skip it (1 or 2) - if (F.size() < function_minimum_size) { continue; } - - std::unordered_set MS; - if (!MarkSetOpt) { - - for (auto &BB : F) { - - MS.insert(&BB); - - } - - total_rs += F.size(); - - } else { - - auto Result = markNodes(&F); - auto RS = Result.first; - auto HS = Result.second; - - MS.insert(RS.begin(), RS.end()); - if (!LoopHeadOpt) { - - MS.insert(HS.begin(), HS.end()); - total_rs += MS.size(); - - } else { - - DenseSet> EdgeSet; - DominatorTreeWrapperPass * DTWP = - &getAnalysis(F); - auto DT = &DTWP->getDomTree(); - - total_rs += RS.size(); - total_hs += HS.size(); - - for (BasicBlock *BB : HS) { - - bool Inserted = false; - for (auto BI = pred_begin(BB), BE = pred_end(BB); BI != BE; ++BI) { - - auto Edge = BasicBlockEdge(*BI, BB); - if (Edge.isSingleEdge() && DT->dominates(Edge, BB)) { - - EdgeSet.insert({*BI, BB}); - Inserted = true; - break; - - } - - } - - if (!Inserted) { - - MS.insert(BB); - total_rs += 1; - total_hs -= 1; - - } - - } - - for (auto I = EdgeSet.begin(), E = EdgeSet.end(); I != E; ++I) { - - auto PredBB = I->first; - auto SuccBB = I->second; - auto NewBB = - SplitBlockPredecessors(SuccBB, {PredBB}, ".split", DT, nullptr, -#if LLVM_VERSION_MAJOR >= 8 - nullptr, -#endif - false); - MS.insert(NewBB); - - } - - } - - for (BasicBlock &BB : F) { - - if (MS.find(&BB) == MS.end()) { continue; } - IRBuilder<> IRB(&*BB.getFirstInsertionPt()); - -#ifdef AFL_HAVE_VECTOR_INTRINSICS - if (ngram_size) { - - LoadInst *PrevLoc = IRB.CreateLoad(AFLPrevLoc); - PrevLoc->setMetadata(M.getMDKindID("nosanitize"), - MDNode::get(C, None)); - - Value *ShuffledPrevLoc = IRB.CreateShuffleVector( - PrevLoc, UndefValue::get(PrevLocTy), PrevLocShuffleMask); - Value *UpdatedPrevLoc = IRB.CreateInsertElement( - ShuffledPrevLoc, ConstantInt::get(Int32Ty, genLabel()), - (uint64_t)0); - - IRB.CreateStore(UpdatedPrevLoc, AFLPrevLoc) - ->setMetadata(M.getMDKindID("nosanitize"), - MDNode::get(C, None)); - - } else - -#endif - { - - IRB.CreateStore(ConstantInt::get(Int32Ty, genLabel()), AFLPrevLoc); - - } - - } - - } - - int has_calls = 0; - for (BasicBlock &BB : F) { - - auto PI = pred_begin(&BB); - auto PE = pred_end(&BB); - IRBuilder<> IRB(&*BB.getFirstInsertionPt()); - Value * L = NULL; - unsigned int cur_loc; - - // Context sensitive coverage - if (ctx_str && &BB == &F.getEntryBlock()) { - - PrevCtx = IRB.CreateLoad(AFLContext); - PrevCtx->setMetadata(M.getMDKindID("nosanitize"), - MDNode::get(C, None)); - - // does the function have calls? and is any of the calls larger than - // one basic block? - has_calls = 0; - for (auto &BB : F) { - - if (has_calls) break; - for (auto &IN : BB) { - - CallInst *callInst = nullptr; - if ((callInst = dyn_cast(&IN))) { - - Function *Callee = callInst->getCalledFunction(); - if (!Callee || Callee->size() < function_minimum_size) - continue; - else { - - has_calls = 1; - break; - - } - - } - - } - - } - - // if yes we store a context ID for this function in the global var - if (has_calls) { - - ConstantInt *NewCtx = ConstantInt::get(Int32Ty, genLabel()); - StoreInst * StoreCtx = IRB.CreateStore(NewCtx, AFLContext); - StoreCtx->setMetadata(M.getMDKindID("nosanitize"), - MDNode::get(C, None)); - - } - - } // END of ctx_str - - if (MarkSetOpt && MS.find(&BB) == MS.end()) { continue; } - - if (PI == PE) { - - cur_loc = genLabel(); - L = ConstantInt::get(Int32Ty, cur_loc); - - } else { - - auto *PN = PHINode::Create(Int32Ty, 0, "", &*BB.begin()); - DenseMap PredMap; - for (auto PI = pred_begin(&BB), PE = pred_end(&BB); PI != PE; ++PI) { - - BasicBlock *PBB = *PI; - auto It = PredMap.insert({PBB, genLabel()}); - unsigned Label = It.first->second; - cur_loc = Label; - PN->addIncoming(ConstantInt::get(Int32Ty, Label), PBB); - - } - - L = PN; - - } - - /* Load prev_loc */ - LoadInst *PrevLoc = IRB.CreateLoad(AFLPrevLoc); - PrevLoc->setMetadata(M.getMDKindID("nosanitize"), MDNode::get(C, None)); - Value *PrevLocTrans; - -#ifdef AFL_HAVE_VECTOR_INTRINSICS - /* "For efficiency, we propose to hash the tuple as a key into the - hit_count map as (prev_block_trans << 1) ^ curr_block_trans, where - prev_block_trans = (block_trans_1 ^ ... ^ block_trans_(n-1)" */ - - if (ngram_size) - PrevLocTrans = - IRB.CreateZExt(IRB.CreateXorReduce(PrevLoc), IRB.getInt32Ty()); - else -#endif - PrevLocTrans = IRB.CreateZExt(PrevLoc, IRB.getInt32Ty()); - - if (ctx_str) - PrevLocTrans = - IRB.CreateZExt(IRB.CreateXor(PrevLocTrans, PrevCtx), Int32Ty); - - /* Load SHM pointer */ - LoadInst *MapPtr = IRB.CreateLoad(AFLMapPtr); - MapPtr->setMetadata(M.getMDKindID("nosanitize"), MDNode::get(C, None)); - Value *MapPtrIdx; -#ifdef AFL_HAVE_VECTOR_INTRINSICS - if (ngram_size) - MapPtrIdx = IRB.CreateGEP( - MapPtr, IRB.CreateZExt(IRB.CreateXor(PrevLocTrans, L), Int32Ty)); - else -#endif - MapPtrIdx = IRB.CreateGEP(MapPtr, IRB.CreateXor(PrevLocTrans, L)); - - /* Update bitmap */ - LoadInst *Counter = IRB.CreateLoad(MapPtrIdx); - Counter->setMetadata(M.getMDKindID("nosanitize"), MDNode::get(C, None)); - - Value *Incr = IRB.CreateAdd(Counter, One); - -#if LLVM_VERSION_MAJOR < 9 - if (neverZero_counters_str != - NULL) // with llvm 9 we make this the default as the bug in llvm is - // then fixed -#else - if (!skip_nozero) -#endif - { - - /* hexcoder: Realize a counter that skips zero during overflow. - * Once this counter reaches its maximum value, it next increments to - * 1 - * - * Instead of - * Counter + 1 -> Counter - * we inject now this - * Counter + 1 -> {Counter, OverflowFlag} - * Counter + OverflowFlag -> Counter - */ - auto cf = IRB.CreateICmpEQ(Incr, Zero); - auto carry = IRB.CreateZExt(cf, Int8Ty); - Incr = IRB.CreateAdd(Incr, carry); - - } - - IRB.CreateStore(Incr, MapPtrIdx) - ->setMetadata(M.getMDKindID("nosanitize"), MDNode::get(C, None)); - - if (ctx_str && has_calls) { - - // in CTX mode we have to restore the original context for the - // caller - she might be calling other functions which need the - // correct CTX - Instruction *Inst = BB.getTerminator(); - if (isa(Inst) || isa(Inst)) { - - IRBuilder<> Post_IRB(Inst); - StoreInst * RestoreCtx = Post_IRB.CreateStore(PrevCtx, AFLContext); - RestoreCtx->setMetadata(M.getMDKindID("nosanitize"), - MDNode::get(C, None)); - - } - - } - - total_instr++; - - } - - } - - if (!be_quiet) { - - char modeline[100]; - snprintf(modeline, sizeof(modeline), "%s%s%s%s%s", - getenv("AFL_HARDEN") ? "hardened" : "non-hardened", - getenv("AFL_USE_ASAN") ? ", ASAN" : "", - getenv("AFL_USE_MSAN") ? ", MSAN" : "", - getenv("AFL_USE_CFISAN") ? ", CFISAN" : "", - getenv("AFL_USE_UBSAN") ? ", UBSAN" : ""); - - OKF("Instrumented %u locations (%llu, %llu) (%s mode)\n", total_instr, - total_rs, total_hs, modeline); - - } - - return false; - - } - -}; // end of struct InsTrim - -} // end of anonymous namespace - -char InsTrim::ID = 0; - -static void registerAFLPass(const PassManagerBuilder &, - legacy::PassManagerBase &PM) { - - PM.add(new InsTrim()); - -} - -static RegisterStandardPasses RegisterAFLPass( - PassManagerBuilder::EP_OptimizerLast, registerAFLPass); - -static RegisterStandardPasses RegisterAFLPass0( - PassManagerBuilder::EP_EnabledOnOptLevel0, registerAFLPass); - diff --git a/llvm_mode/Makefile b/llvm_mode/Makefile deleted file mode 100644 index 3666a74d..00000000 --- a/llvm_mode/Makefile +++ /dev/null @@ -1,2 +0,0 @@ -all: - @gmake all || echo please install GNUmake diff --git a/llvm_mode/MarkNodes.cc b/llvm_mode/MarkNodes.cc deleted file mode 100644 index 20a7df35..00000000 --- a/llvm_mode/MarkNodes.cc +++ /dev/null @@ -1,481 +0,0 @@ -#include -#include -#include -#include -#include - -#include "llvm/Config/llvm-config.h" -#if LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR < 5 -typedef long double max_align_t; -#endif - -#include "llvm/ADT/DenseMap.h" -#include "llvm/ADT/DenseSet.h" -#include "llvm/ADT/SmallVector.h" -#include "llvm/IR/BasicBlock.h" -#if LLVM_VERSION_MAJOR > 3 || \ - (LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR > 4) - #include "llvm/IR/CFG.h" -#else - #include "llvm/Support/CFG.h" -#endif -#include "llvm/IR/Constants.h" -#include "llvm/IR/Function.h" -#include "llvm/IR/IRBuilder.h" -#include "llvm/IR/Instructions.h" -#include "llvm/IR/Module.h" -#include "llvm/Pass.h" -#include "llvm/Support/Debug.h" -#include "llvm/Support/raw_ostream.h" - -using namespace llvm; - -DenseMap LMap; -std::vector Blocks; -std::set Marked, Markabove; -std::vector > Succs, Preds; - -void reset() { - - LMap.clear(); - Blocks.clear(); - Marked.clear(); - Markabove.clear(); - -} - -uint32_t start_point; - -void labelEachBlock(Function *F) { - - // Fake single endpoint; - LMap[NULL] = Blocks.size(); - Blocks.push_back(NULL); - - // Assign the unique LabelID to each block; - for (auto I = F->begin(), E = F->end(); I != E; ++I) { - - BasicBlock *BB = &*I; - LMap[BB] = Blocks.size(); - Blocks.push_back(BB); - - } - - start_point = LMap[&F->getEntryBlock()]; - -} - -void buildCFG(Function *F) { - - Succs.resize(Blocks.size()); - Preds.resize(Blocks.size()); - for (size_t i = 0; i < Succs.size(); i++) { - - Succs[i].clear(); - Preds[i].clear(); - - } - - for (auto S = F->begin(), E = F->end(); S != E; ++S) { - - BasicBlock *BB = &*S; - uint32_t MyID = LMap[BB]; - - for (auto I = succ_begin(BB), E = succ_end(BB); I != E; ++I) { - - Succs[MyID].push_back(LMap[*I]); - - } - - } - -} - -std::vector > tSuccs; -std::vector tag, indfs; - -void DFStree(size_t now_id) { - - if (tag[now_id]) return; - tag[now_id] = true; - indfs[now_id] = true; - for (auto succ : tSuccs[now_id]) { - - if (tag[succ] and indfs[succ]) { - - Marked.insert(succ); - Markabove.insert(succ); - continue; - - } - - Succs[now_id].push_back(succ); - Preds[succ].push_back(now_id); - DFStree(succ); - - } - - indfs[now_id] = false; - -} - -void turnCFGintoDAG() { - - tSuccs = Succs; - tag.resize(Blocks.size()); - indfs.resize(Blocks.size()); - for (size_t i = 0; i < Blocks.size(); ++i) { - - Succs[i].clear(); - tag[i] = false; - indfs[i] = false; - - } - - DFStree(start_point); - for (size_t i = 0; i < Blocks.size(); ++i) - if (Succs[i].empty()) { - - Succs[i].push_back(0); - Preds[0].push_back(i); - - } - -} - -uint32_t timeStamp; -namespace DominatorTree { - -std::vector > cov; -std::vector dfn, nfd, par, sdom, idom, mom, mn; - -bool Compare(uint32_t u, uint32_t v) { - - return dfn[u] < dfn[v]; - -} - -uint32_t eval(uint32_t u) { - - if (mom[u] == u) return u; - uint32_t res = eval(mom[u]); - if (Compare(sdom[mn[mom[u]]], sdom[mn[u]])) { mn[u] = mn[mom[u]]; } - return mom[u] = res; - -} - -void DFS(uint32_t now) { - - timeStamp += 1; - dfn[now] = timeStamp; - nfd[timeStamp - 1] = now; - for (auto succ : Succs[now]) { - - if (dfn[succ] == 0) { - - par[succ] = now; - DFS(succ); - - } - - } - -} - -void DominatorTree() { - - if (Blocks.empty()) return; - uint32_t s = start_point; - - // Initialization - mn.resize(Blocks.size()); - cov.resize(Blocks.size()); - dfn.resize(Blocks.size()); - nfd.resize(Blocks.size()); - par.resize(Blocks.size()); - mom.resize(Blocks.size()); - sdom.resize(Blocks.size()); - idom.resize(Blocks.size()); - - for (uint32_t i = 0; i < Blocks.size(); i++) { - - dfn[i] = 0; - nfd[i] = Blocks.size(); - cov[i].clear(); - idom[i] = mom[i] = mn[i] = sdom[i] = i; - - } - - timeStamp = 0; - DFS(s); - - for (uint32_t i = Blocks.size() - 1; i >= 1u; i--) { - - uint32_t now = nfd[i]; - if (now == Blocks.size()) { continue; } - for (uint32_t pre : Preds[now]) { - - if (dfn[pre]) { - - eval(pre); - if (Compare(sdom[mn[pre]], sdom[now])) { sdom[now] = sdom[mn[pre]]; } - - } - - } - - cov[sdom[now]].push_back(now); - mom[now] = par[now]; - for (uint32_t x : cov[par[now]]) { - - eval(x); - if (Compare(sdom[mn[x]], par[now])) { - - idom[x] = mn[x]; - - } else { - - idom[x] = par[now]; - - } - - } - - } - - for (uint32_t i = 1; i < Blocks.size(); i += 1) { - - uint32_t now = nfd[i]; - if (now == Blocks.size()) { continue; } - if (idom[now] != sdom[now]) idom[now] = idom[idom[now]]; - - } - -} - -} // namespace DominatorTree - -std::vector Visited, InStack; -std::vector TopoOrder, InDeg; -std::vector > t_Succ, t_Pred; - -void Go(uint32_t now, uint32_t tt) { - - if (now == tt) return; - Visited[now] = InStack[now] = timeStamp; - - for (uint32_t nxt : Succs[now]) { - - if (Visited[nxt] == timeStamp and InStack[nxt] == timeStamp) { - - Marked.insert(nxt); - - } - - t_Succ[now].push_back(nxt); - t_Pred[nxt].push_back(now); - InDeg[nxt] += 1; - if (Visited[nxt] == timeStamp) { continue; } - Go(nxt, tt); - - } - - InStack[now] = 0; - -} - -void TopologicalSort(uint32_t ss, uint32_t tt) { - - timeStamp += 1; - - Go(ss, tt); - - TopoOrder.clear(); - std::queue wait; - wait.push(ss); - while (not wait.empty()) { - - uint32_t now = wait.front(); - wait.pop(); - TopoOrder.push_back(now); - for (uint32_t nxt : t_Succ[now]) { - - InDeg[nxt] -= 1; - if (InDeg[nxt] == 0u) { wait.push(nxt); } - - } - - } - -} - -std::vector > NextMarked; -bool Indistinguish(uint32_t node1, uint32_t node2) { - - if (NextMarked[node1].size() > NextMarked[node2].size()) { - - uint32_t _swap = node1; - node1 = node2; - node2 = _swap; - - } - - for (uint32_t x : NextMarked[node1]) { - - if (NextMarked[node2].find(x) != NextMarked[node2].end()) { return true; } - - } - - return false; - -} - -void MakeUniq(uint32_t now) { - - bool StopFlag = false; - if (Marked.find(now) == Marked.end()) { - - for (uint32_t pred1 : t_Pred[now]) { - - for (uint32_t pred2 : t_Pred[now]) { - - if (pred1 == pred2) continue; - if (Indistinguish(pred1, pred2)) { - - Marked.insert(now); - StopFlag = true; - break; - - } - - } - - if (StopFlag) { break; } - - } - - } - - if (Marked.find(now) != Marked.end()) { - - NextMarked[now].insert(now); - - } else { - - for (uint32_t pred : t_Pred[now]) { - - for (uint32_t x : NextMarked[pred]) { - - NextMarked[now].insert(x); - - } - - } - - } - -} - -bool MarkSubGraph(uint32_t ss, uint32_t tt) { - - TopologicalSort(ss, tt); - if (TopoOrder.empty()) return false; - - for (uint32_t i : TopoOrder) { - - NextMarked[i].clear(); - - } - - NextMarked[TopoOrder[0]].insert(TopoOrder[0]); - for (uint32_t i = 1; i < TopoOrder.size(); i += 1) { - - MakeUniq(TopoOrder[i]); - - } - - // Check if there is an empty path. - if (NextMarked[tt].count(TopoOrder[0]) > 0) return true; - return false; - -} - -void MarkVertice() { - - uint32_t s = start_point; - - InDeg.resize(Blocks.size()); - Visited.resize(Blocks.size()); - InStack.resize(Blocks.size()); - t_Succ.resize(Blocks.size()); - t_Pred.resize(Blocks.size()); - NextMarked.resize(Blocks.size()); - - for (uint32_t i = 0; i < Blocks.size(); i += 1) { - - Visited[i] = InStack[i] = InDeg[i] = 0; - t_Succ[i].clear(); - t_Pred[i].clear(); - - } - - timeStamp = 0; - uint32_t t = 0; - bool emptyPathExists = true; - - while (s != t) { - - emptyPathExists &= MarkSubGraph(DominatorTree::idom[t], t); - t = DominatorTree::idom[t]; - - } - - if (emptyPathExists) { - - // Mark all exit blocks to catch the empty path. - Marked.insert(t_Pred[0].begin(), t_Pred[0].end()); - - } - -} - -// return {marked nodes} -std::pair, std::vector > markNodes( - Function *F) { - - assert(F->size() > 0 && "Function can not be empty"); - - reset(); - labelEachBlock(F); - buildCFG(F); - turnCFGintoDAG(); - DominatorTree::DominatorTree(); - MarkVertice(); - - std::vector Result, ResultAbove; - for (uint32_t x : Markabove) { - - auto it = Marked.find(x); - if (it != Marked.end()) Marked.erase(it); - if (x) ResultAbove.push_back(Blocks[x]); - - } - - for (uint32_t x : Marked) { - - if (x == 0) { - - continue; - - } else { - - Result.push_back(Blocks[x]); - - } - - } - - return {Result, ResultAbove}; - -} - diff --git a/llvm_mode/MarkNodes.h b/llvm_mode/MarkNodes.h deleted file mode 100644 index 8ddc978d..00000000 --- a/llvm_mode/MarkNodes.h +++ /dev/null @@ -1,12 +0,0 @@ -#ifndef __MARK_NODES__ -#define __MARK_NODES__ - -#include "llvm/IR/BasicBlock.h" -#include "llvm/IR/Function.h" -#include - -std::pair, std::vector> -markNodes(llvm::Function *F); - -#endif - diff --git a/llvm_mode/README.cmplog.md b/llvm_mode/README.cmplog.md deleted file mode 100644 index 7f426ec8..00000000 --- a/llvm_mode/README.cmplog.md +++ /dev/null @@ -1,42 +0,0 @@ -# CmpLog instrumentation - -The CmpLog instrumentation enables the logging of the comparisons operands in a -shared memory. - -These values can be used by various mutators built on top of it. -At the moment we support the RedQueen mutator (input-2-state instructions only). - -## Build - -To use CmpLog, you have to build two versions of the instrumented target -program. - -The first version is built using the regular AFL++ instrumentation. - -The second one, the CmpLog binary, with setting AFL_LLVM_CMPLOG during the compilation. - -For example: - -``` -./configure --cc=~/path/to/afl-clang-fast -make -cp ./program ./program.afl -make clean -export AFL_LLVM_CMPLOG=1 -./configure --cc=~/path/to/afl-clang-fast -make -cp ./program ./program.cmplog -``` - -## Use - -AFL++ has the new -c option that can be used to specify a CmpLog binary (the second -build). - -For example: - -``` -afl-fuzz -i input -o output -c ./program.cmplog -m none -- ./program.afl @@ -``` - -Be careful to use -m none because CmpLog maps a lot of pages. diff --git a/llvm_mode/README.ctx.md b/llvm_mode/README.ctx.md deleted file mode 100644 index 14255313..00000000 --- a/llvm_mode/README.ctx.md +++ /dev/null @@ -1,22 +0,0 @@ -# AFL Context Sensitive Branch Coverage - -## What is this? - -This is an LLVM-based implementation of the context sensitive branch coverage. - -Basically every function gets it's own ID and that ID is combined with the -edges of the called functions. - -So if both function A and function B call a function C, the coverage -collected in C will be different. - -In math the coverage is collected as follows: -`map[current_location_ID ^ previous_location_ID >> 1 ^ previous_callee_ID] += 1` - -## Usage - -Set the `AFL_LLVM_INSTRUMENT=CTX` or `AFL_LLVM_CTX=1` environment variable. - -It is highly recommended to increase the MAP_SIZE_POW2 definition in -config.h to at least 18 and maybe up to 20 for this as otherwise too -many map collisions occur. diff --git a/llvm_mode/README.instrim.md b/llvm_mode/README.instrim.md deleted file mode 100644 index 7758091b..00000000 --- a/llvm_mode/README.instrim.md +++ /dev/null @@ -1,25 +0,0 @@ -# InsTrim - -InsTrim: Lightweight Instrumentation for Coverage-guided Fuzzing - -## Introduction - -InsTrim uses CFG and markers to instrument just what is necessary in the -binary in llvm_mode. It is about 10-15% faster without disadvantages. -It requires at least llvm version 3.8.0. - -## Usage - -Set the environment variable `AFL_LLVM_INSTRUMENT=CFG` or `AFL_LLVM_INSTRIM=1` -during compilation of the target. - -There is also an advanced mode which instruments loops in a way so that -afl-fuzz can see which loop path has been selected but not being able to -see how often the loop has been rerun. -This again is a tradeoff for speed for less path information. -To enable this mode set `AFL_LLVM_INSTRIM_LOOPHEAD=1`. - -## Background - -The paper: [InsTrim: Lightweight Instrumentation for Coverage-guided Fuzzing] -(https://www.ndss-symposium.org/wp-content/uploads/2018/07/bar2018_14_Hsu_paper.pdf) diff --git a/llvm_mode/README.instrument_list.md b/llvm_mode/README.instrument_list.md deleted file mode 100644 index 1fc06414..00000000 --- a/llvm_mode/README.instrument_list.md +++ /dev/null @@ -1,86 +0,0 @@ -# Using afl++ with partial instrumentation - - This file describes how you can selectively instrument only the source files - or functions that are interesting to you using the LLVM instrumentation - provided by afl++ - -## 1) Description and purpose - -When building and testing complex programs where only a part of the program is -the fuzzing target, it often helps to only instrument the necessary parts of -the program, leaving the rest uninstrumented. This helps to focus the fuzzer -on the important parts of the program, avoiding undesired noise and -disturbance by uninteresting code being exercised. - -For this purpose, a "partial instrumentation" support en par with llvm sancov -is provided by afl++ that allows you to specify on a source file and function -level which function should be compiled with or without instrumentation. - -Note: When using PCGUARD mode - and have llvm 12+ - you can use this instead: -https://clang.llvm.org/docs/SanitizerCoverage.html#partially-disabling-instrumentation - -The llvm sancov list format is fully supported by afl++, however afl++ has -more flexibility. - -## 2) Building the LLVM module - -The new code is part of the existing afl++ LLVM module in the llvm_mode/ -subdirectory. There is nothing specifically to do :) - -## 3) How to use the partial instrumentation mode - -In order to build with partial instrumentation, you need to build with -afl-clang-fast/afl-clang-fast++ or afl-clang-lto/afl-clang-lto++. -The only required change is that you need to set either the environment variable -AFL_LLVM_ALLOWLIST or AFL_LLVM_DENYLIST set with a filename. - -That file then contains the filenames or functions that should be instrumented -(AFL_LLVM_ALLOWLIST) or should specifically NOT be instrumented (AFL_LLVM_DENYLIST). - -For matching, the function/filename that is being compiled must end in the -function/filename entry contained in this instrument file list (to avoid -breaking the matching when absolute paths are used during compilation). - -**NOTE:** In builds with optimization enabled functions might be inlined and would not match! - -For example if your source tree looks like this: -``` -project/ -project/feature_a/a1.cpp -project/feature_a/a2.cpp -project/feature_b/b1.cpp -project/feature_b/b2.cpp -``` - -and you only want to test feature_a, then create a instrument file list file containing: -``` -feature_a/a1.cpp -feature_a/a2.cpp -``` - -However if the instrument file list file contains only this, it works as well: -``` -a1.cpp -a2.cpp -``` -but it might lead to files being unwantedly instrumented if the same filename -exists somewhere else in the project directories. - -You can also specify function names. Note that for C++ the function names -must be mangled to match! - -afl++ is able to identify if an entry is a filename or a function. -However if you want to be sure (and compliant to the sancov allow/blocklist -format), you can specify source file entries like this: -``` -src: *malloc.c -``` -and function entries like this: -``` -fun: MallocFoo -``` -Note that whitespace is ignored and comments (`# foo`) are supported. - -## 4) UNIX-style pattern matching -You can add UNIX-style pattern matching in the the instrument file list entries. -See `man fnmatch` for the syntax. We do not set any of the `fnmatch` flags. diff --git a/llvm_mode/README.laf-intel.md b/llvm_mode/README.laf-intel.md deleted file mode 100644 index f63ab2bb..00000000 --- a/llvm_mode/README.laf-intel.md +++ /dev/null @@ -1,42 +0,0 @@ -# laf-intel instrumentation - -## Usage - -By default these passes will not run when you compile programs using -afl-clang-fast. Hence, you can use AFL as usual. -To enable the passes you must set environment variables before you -compile the target project. - -The following options exist: - -`export AFL_LLVM_LAF_SPLIT_SWITCHES=1` - -Enables the split-switches pass. - -`export AFL_LLVM_LAF_TRANSFORM_COMPARES=1` - -Enables the transform-compares pass (strcmp, memcmp, strncmp, -strcasecmp, strncasecmp). - -`export AFL_LLVM_LAF_SPLIT_COMPARES=1` - -Enables the split-compares pass. -By default it will -1. simplify operators >= (and <=) into chains of > (<) and == comparisons -2. change signed integer comparisons to a chain of sign-only comparison -and unsigned comparisons -3. split all unsigned integer comparisons with bit widths of -64, 32 or 16 bits to chains of 8 bits comparisons. - -You can change the behaviour of the last step by setting -`export AFL_LLVM_LAF_SPLIT_COMPARES_BITW=`, where -bit_width may be 64, 32 or 16. - -A new experimental feature is splitting floating point comparisons into a -series of sign, exponent and mantissa comparisons followed by splitting each -of them into 8 bit comparisons when necessary. -It is activated with the `AFL_LLVM_LAF_SPLIT_FLOATS` setting. -Note that setting this automatically activates `AFL_LLVM_LAF_SPLIT_COMPARES` - -You can also set `AFL_LLVM_LAF_ALL` and have all of the above enabled :-) - diff --git a/llvm_mode/README.lto.md b/llvm_mode/README.lto.md deleted file mode 100644 index 9046c5a8..00000000 --- a/llvm_mode/README.lto.md +++ /dev/null @@ -1,293 +0,0 @@ -# afl-clang-lto - collision free instrumentation at link time - -## TLDR; - -This version requires a current llvm 11+ compiled from the github master. - -1. Use afl-clang-lto/afl-clang-lto++ because it is faster and gives better - coverage than anything else that is out there in the AFL world - -2. You can use it together with llvm_mode: laf-intel and the instrument file listing - features and can be combined with cmplog/Redqueen - -3. It only works with llvm 11+ - -4. AUTODICTIONARY feature! see below - -5. If any problems arise be sure to set `AR=llvm-ar RANLIB=llvm-ranlib`. - Some targets might need `LD=afl-clang-lto` and others `LD=afl-ld-lto`. - -## Introduction and problem description - -A big issue with how afl/afl++ works is that the basic block IDs that are -set during compilation are random - and hence naturally the larger the number -of instrumented locations, the higher the number of edge collisions are in the -map. This can result in not discovering new paths and therefore degrade the -efficiency of the fuzzing process. - -*This issue is underestimated in the fuzzing community!* -With a 2^16 = 64kb standard map at already 256 instrumented blocks there is -on average one collision. On average a target has 10.000 to 50.000 -instrumented blocks hence the real collisions are between 750-18.000! - -To reach a solution that prevents any collisions took several approaches -and many dead ends until we got to this: - - * We instrument at link time when we have all files pre-compiled - * To instrument at link time we compile in LTO (link time optimization) mode - * Our compiler (afl-clang-lto/afl-clang-lto++) takes care of setting the - correct LTO options and runs our own afl-ld linker instead of the system - linker - * The LLVM linker collects all LTO files to link and instruments them so that - we have non-colliding edge overage - * We use a new (for afl) edge coverage - which is the same as in llvm - -fsanitize=coverage edge coverage mode :) - -The result: - * 10-25% speed gain compared to llvm_mode - * guaranteed non-colliding edge coverage :-) - * The compile time especially for binaries to an instrumented library can be - much longer - -Example build output from a libtiff build: -``` -libtool: link: afl-clang-lto -g -O2 -Wall -W -o thumbnail thumbnail.o ../libtiff/.libs/libtiff.a ../port/.libs/libport.a -llzma -ljbig -ljpeg -lz -lm -afl-clang-lto++2.63d by Marc "vanHauser" Heuse in mode LTO -afl-llvm-lto++2.63d by Marc "vanHauser" Heuse -AUTODICTIONARY: 11 strings found -[+] Instrumented 12071 locations with no collisions (on average 1046 collisions would be in afl-gcc/afl-clang-fast) (non-hardened mode). -``` - -## Getting llvm 11+ - -### Installing llvm from the llvm repository (version 11) - -Installing the llvm snapshot builds is easy and mostly painless: - -In the follow line change `NAME` for your Debian or Ubuntu release name -(e.g. buster, focal, eon, etc.): -``` -echo deb http://apt.llvm.org/NAME/ llvm-toolchain-NAME NAME >> /etc/apt/sources.list -``` -then add the pgp key of llvm and install the packages: -``` -wget -O - https://apt.llvm.org/llvm-snapshot.gpg.key | apt-key add - -apt-get update && apt-get upgrade -y -apt-get install -y clang-11 clang-tools-11 libc++1-11 libc++-11-dev \ - libc++abi1-11 libc++abi-11-dev libclang1-11 libclang-11-dev \ - libclang-common-11-dev libclang-cpp11 libclang-cpp11-dev liblld-11 \ - liblld-11-dev liblldb-11 liblldb-11-dev libllvm11 libomp-11-dev \ - libomp5-11 lld-11 lldb-11 llvm-11 llvm-11-dev llvm-11-runtime llvm-11-tools -``` - -### Building llvm yourself (version 12) - -Building llvm from github takes quite some long time and is not painless: -``` -sudo apt install binutils-dev # this is *essential*! -git clone https://github.com/llvm/llvm-project -cd llvm-project -mkdir build -cd build -cmake -DLLVM_ENABLE_PROJECTS='clang;clang-tools-extra;compiler-rt;libclc;libcxx;libcxxabi;libunwind;lld' -DCMAKE_BUILD_TYPE=Release -DLLVM_BINUTILS_INCDIR=/usr/include/ ../llvm/ -make -j $(nproc) -export PATH=`pwd`/bin:$PATH -export LLVM_CONFIG=`pwd`/bin/llvm-config -cd /path/to/AFLplusplus/ -make -cd llvm_mode -make -cd .. -make install -``` - -## How to use afl-clang-lto - -Just use afl-clang-lto like you did with afl-clang-fast or afl-gcc. - -Also the instrument file listing (AFL_LLVM_ALLOWLIST/AFL_LLVM_DENYLIST -> [README.instrument_list.md](README.instrument_list.md)) and -laf-intel/compcov (AFL_LLVM_LAF_* -> [README.laf-intel.md](README.laf-intel.md)) work. - -Example: -``` -CC=afl-clang-lto CXX=afl-clang-lto++ RANLIB=llvm-ranlib AR=llvm-ar ./configure -make -``` - -NOTE: some targets also need to set the linker, try both `afl-clang-lto` and -`afl-ld-lto` for this for `LD=` for `configure`. - -## AUTODICTIONARY feature - -While compiling, automatically a dictionary based on string comparisons is -generated put into the target binary. This dictionary is transfered to afl-fuzz -on start. This improves coverage statistically by 5-10% :) - -## Fixed memory map - -To speed up fuzzing, it is possible to set a fixed shared memory map. -Recommened is the value 0x10000. -In most cases this will work without any problems. However if a target uses -early constructors, ifuncs or a deferred forkserver this can crash the target. -On unusual operating systems/processors/kernels or weird libraries this might -fail so to change the fixed address at compile time set -AFL_LLVM_MAP_ADDR with a better value (a value of 0 or empty sets the map address -to be dynamic - the original afl way, which is slower). - -## Document edge IDs - -Setting `export AFL_LLVM_DOCUMENT_IDS=file` will document to a file which edge -ID was given to which function. This helps to identify functions with variable -bytes or which functions were touched by an input. - -## Solving difficult targets - -Some targets are difficult because the configure script does unusual stuff that -is unexpected for afl. See the next chapter `Potential issues` how to solve -these. - -### Example: ffmpeg - -An example of a hard to solve target is ffmpeg. Here is how to successfully -instrument it: - -1. Get and extract the current ffmpeg and change to it's directory - -2. Running configure with --cc=clang fails and various other items will fail - when compiling, so we have to trick configure: - -``` -./configure --enable-lto --disable-shared --disable-inline-asm -``` - -3. Now the configuration is done - and we edit the settings in `./ffbuild/config.mak` - (-: the original line, +: what to change it into): -``` --CC=gcc -+CC=afl-clang-lto --CXX=g++ -+CXX=afl-clang-lto++ --AS=gcc -+AS=llvm-as --LD=gcc -+LD=afl-clang-lto++ --DEPCC=gcc -+DEPCC=afl-clang-lto --DEPAS=gcc -+DEPAS=afl-clang-lto++ --AR=ar -+AR=llvm-ar --AR_CMD=ar -+AR_CMD=llvm-ar --NM_CMD=nm -g -+NM_CMD=llvm-nm -g --RANLIB=ranlib -D -+RANLIB=llvm-ranlib -D -``` - -4. Then type make, wait for a long time and you are done :) - -### Example: WebKit jsc - -Building jsc is difficult as the build script has bugs. - -1. checkout Webkit: -``` -svn checkout https://svn.webkit.org/repository/webkit/trunk WebKit -cd WebKit -``` - -2. Fix the build environment: -``` -mkdir -p WebKitBuild/Release -cd WebKitBuild/Release -ln -s ../../../../../usr/bin/llvm-ar-12 llvm-ar-12 -ln -s ../../../../../usr/bin/llvm-ranlib-12 llvm-ranlib-12 -cd ../.. -``` - -3. Build :) - -``` -Tools/Scripts/build-jsc --jsc-only --cli --cmakeargs="-DCMAKE_AR='llvm-ar-12' -DCMAKE_RANLIB='llvm-ranlib-12' -DCMAKE_VERBOSE_MAKEFILE:BOOL=ON -DCMAKE_CC_FLAGS='-O3 -lrt' -DCMAKE_CXX_FLAGS='-O3 -lrt' -DIMPORTED_LOCATION='/lib/x86_64-linux-gnu/' -DCMAKE_CC=afl-clang-lto -DCMAKE_CXX=afl-clang-lto++ -DENABLE_STATIC_JSC=ON" -``` - -## Potential issues - -### compiling libraries fails - -If you see this message: -``` -/bin/ld: libfoo.a: error adding symbols: archive has no index; run ranlib to add one -``` -This is because usually gnu gcc ranlib is being called which cannot deal with clang LTO files. -The solution is simple: when you ./configure you have also have to set RANLIB=llvm-ranlib and AR=llvm-ar - -Solution: -``` -AR=llvm-ar RANLIB=llvm-ranlib CC=afl-clang-lto CXX=afl-clang-lto++ ./configure --disable-shared -``` -and on some target you have to to AR=/RANLIB= even for make as the configure script does not save it. -Other targets ignore environment variables and need the parameters set via -`./configure --cc=... --cxx= --ranlib= ...` etc. (I am looking at you ffmpeg!). - - -If you see this message -``` -assembler command failed ... -``` -then try setting `llvm-as` for configure: -``` -AS=llvm-as ... -``` - -### compiling programs still fail - -afl-clang-lto is still work in progress. - -Known issues: - * Anything that llvm 11+ cannot compile, afl-clang-lto can not compile either - obviously - * Anything that does not compile with LTO, afl-clang-lto can not compile either - obviously - -Hence if building a target with afl-clang-lto fails try to build it with llvm12 -and LTO enabled (`CC=clang-12` `CXX=clang++-12` `CFLAGS=-flto=full` and -`CXXFLAGS=-flto=full`). - -If this succeeeds then there is an issue with afl-clang-lto. Please report at -[https://github.com/AFLplusplus/AFLplusplus/issues/226](https://github.com/AFLplusplus/AFLplusplus/issues/226) - -Even some targets where clang-12 fails can be build if the fail is just in -`./configure`, see `Solving difficult targets` above. - -## History - -This was originally envisioned by hexcoder- in Summer 2019, however we saw no -way to create a pass that is run at link time - although there is a option -for this in the PassManager: EP_FullLinkTimeOptimizationLast -("Fun" info - nobody knows what this is doing. And the developer who -implemented this didn't respond to emails.) - -In December came then the idea to implement this as a pass that is run via -the llvm "opt" program, which is performed via an own linker that afterwards -calls the real linker. -This was first implemented in January and work ... kinda. -The LTO time instrumentation worked, however the "how" the basic blocks were -instrumented was a problem, as reducing duplicates turned out to be very, -very difficult with a program that has so many paths and therefore so many -dependencies. At lot of strategies were implemented - and failed. -And then sat solvers were tried, but with over 10.000 variables that turned -out to be a dead-end too. - -The final idea to solve this came from domenukk who proposed to insert a block -into an edge and then just use incremental counters ... and this worked! -After some trials and errors to implement this vanhauser-thc found out that -there is actually an llvm function for this: SplitEdge() :-) - -Still more problems came up though as this only works without bugs from -llvm 9 onwards, and with high optimization the link optimization ruins -the instrumented control flow graph. - -This is all now fixed with llvm 11+. The llvm's own linker is now able to -load passes and this bypasses all problems we had. - -Happy end :) diff --git a/llvm_mode/README.md b/llvm_mode/README.md deleted file mode 100644 index f23d7150..00000000 --- a/llvm_mode/README.md +++ /dev/null @@ -1,186 +0,0 @@ -# Fast LLVM-based instrumentation for afl-fuzz - - (See [../README](../README.md) for the general instruction manual.) - - (See [../gcc_plugin/README](../gcc_plugin/README.md) for the GCC-based instrumentation.) - -## 1) Introduction - -! llvm_mode works with llvm versions 3.4 up to 12 ! - -The code in this directory allows you to instrument programs for AFL using -true compiler-level instrumentation, instead of the more crude -assembly-level rewriting approach taken by afl-gcc and afl-clang. This has -several interesting properties: - - - The compiler can make many optimizations that are hard to pull off when - manually inserting assembly. As a result, some slow, CPU-bound programs will - run up to around 2x faster. - - The gains are less pronounced for fast binaries, where the speed is limited - chiefly by the cost of creating new processes. In such cases, the gain will - probably stay within 10%. - - - The instrumentation is CPU-independent. At least in principle, you should - be able to rely on it to fuzz programs on non-x86 architectures (after - building afl-fuzz with AFL_NO_X86=1). - - - The instrumentation can cope a bit better with multi-threaded targets. - - - Because the feature relies on the internals of LLVM, it is clang-specific - and will *not* work with GCC (see ../gcc_plugin/ for an alternative once - it is available). - -Once this implementation is shown to be sufficiently robust and portable, it -will probably replace afl-clang. For now, it can be built separately and -co-exists with the original code. - -The idea and much of the intial implementation came from Laszlo Szekeres. - -## 2a) How to use this - short - -Set the `LLVM_CONFIG` variable to the clang version you want to use, e.g. -``` -LLVM_CONFIG=llvm-config-9 make -``` -In case you have your own compiled llvm version specify the full path: -``` -LLVM_CONFIG=~/llvm-project/build/bin/llvm-config make -``` -If you try to use a new llvm version on an old Linux this can fail because of -old c++ libraries. In this case usually switching to gcc/g++ to compile -llvm_mode will work: -``` -LLVM_CONFIG=llvm-config-7 REAL_CC=gcc REAL_CXX=g++ make -``` -It is highly recommended to use the newest clang version you can put your -hands on :) - -Then look at [README.persistent_mode.md](README.persistent_mode.md). - -## 2b) How to use this - long - -In order to leverage this mechanism, you need to have clang installed on your -system. You should also make sure that the llvm-config tool is in your path -(or pointed to via LLVM_CONFIG in the environment). - -Note that if you have several LLVM versions installed, pointing LLVM_CONFIG -to the version you want to use will switch compiling to this specific -version - if you installation is set up correctly :-) - -Unfortunately, some systems that do have clang come without llvm-config or the -LLVM development headers; one example of this is FreeBSD. FreeBSD users will -also run into problems with clang being built statically and not being able to -load modules (you'll see "Service unavailable" when loading afl-llvm-pass.so). - -To solve all your problems, you can grab pre-built binaries for your OS from: - - http://llvm.org/releases/download.html - -...and then put the bin/ directory from the tarball at the beginning of your -$PATH when compiling the feature and building packages later on. You don't need -to be root for that. - -To build the instrumentation itself, type 'make'. This will generate binaries -called afl-clang-fast and afl-clang-fast++ in the parent directory. Once this -is done, you can instrument third-party code in a way similar to the standard -operating mode of AFL, e.g.: - -``` - CC=/path/to/afl/afl-clang-fast ./configure [...options...] - make -``` - -Be sure to also include CXX set to afl-clang-fast++ for C++ code. - -The tool honors roughly the same environmental variables as afl-gcc (see -[docs/env_variables.md](../docs/env_variables.md)). This includes AFL_USE_ASAN, -AFL_HARDEN, and AFL_DONT_OPTIMIZE. However AFL_INST_RATIO is not honored -as it does not serve a good purpose with the more effective instrim CFG -analysis. - -Note: if you want the LLVM helper to be installed on your system for all -users, you need to build it before issuing 'make install' in the parent -directory. - -## 3) Options - -Several options are present to make llvm_mode faster or help it rearrange -the code to make afl-fuzz path discovery easier. - -If you need just to instrument specific parts of the code, you can the instrument file list -which C/C++ files to actually instrument. See [README.instrument_list](README.instrument_list.md) - -For splitting memcmp, strncmp, etc. please see [README.laf-intel](README.laf-intel.md) - -Then there are different ways of instrumenting the target: - -1. There is an optimized instrumentation strategy that uses CFGs and -markers to just instrument what is needed. This increases speed by 10-15% -without any disadvantages -If you want to use this, set AFL_LLVM_INSTRUMENT=CFG or AFL_LLVM_INSTRIM=1 -See [README.instrim](README.instrim.md) - -2. An even better instrumentation strategy uses LTO and link time -instrumentation. Note that not all targets can compile in this mode, however -if it works it is the best option you can use. -Simply use afl-clang-lto/afl-clang-lto++ to use this option. -See [README.lto](README.lto.md) - -3. Alternativly you can choose a completely different coverage method: - -3a. N-GRAM coverage - which combines the previous visited edges with the -current one. This explodes the map but on the other hand has proven to be -effective for fuzzing. -See [README.ngram](README.ngram.md) - -3b. Context sensitive coverage - which combines the visited edges with an -individual caller ID (the function that called the current one) -[README.ctx](README.ctx.md) - -Then - additionally to one of the instrumentation options above - there is -a very effective new instrumentation option called CmpLog as an alternative to -laf-intel that allow AFL++ to apply mutations similar to Redqueen. -See [README.cmplog](README.cmplog.md) - -Finally if your llvm version is 8 or lower, you can activate a mode that -prevents that a counter overflow result in a 0 value. This is good for -path discovery, but the llvm implementation for x86 for this functionality -is not optimal and was only fixed in llvm 9. -You can set this with AFL_LLVM_NOT_ZERO=1 -See [README.neverzero](README.neverzero.md) - -## 4) Snapshot feature - -To speed up fuzzing you can use a linux loadable kernel module which enables -a snapshot feature. -See [README.snapshot](README.snapshot.md) - -## 5) Gotchas, feedback, bugs - -This is an early-stage mechanism, so field reports are welcome. You can send bug -reports to . - -## 6) deferred initialization, persistent mode, shared memory fuzzing - -This is the most powerful and effective fuzzing you can do. -Please see [README.persistent_mode.md](README.persistent_mode.md) for a -full explanation. - -## 7) Bonus feature: 'trace-pc-guard' mode - -LLVM is shipping with a built-in execution tracing feature -that provides AFL with the necessary tracing data without the need to -post-process the assembly or install any compiler plugins. See: - - http://clang.llvm.org/docs/SanitizerCoverage.html#tracing-pcs-with-guards - -If you have not an outdated compiler and want to give it a try, build -targets this way: - -``` -AFL_LLVM_INSTRUMENT=PCGUARD make -``` - -Note that this us currently the default, as it is the best mode. -If you have llvm 11+ and compiled afl-clang-lto - this is the only better mode. diff --git a/llvm_mode/README.neverzero.md b/llvm_mode/README.neverzero.md deleted file mode 100644 index 903e5bd3..00000000 --- a/llvm_mode/README.neverzero.md +++ /dev/null @@ -1,35 +0,0 @@ -# NeverZero counters for LLVM instrumentation - -## Usage - -In larger, complex or reiterative programs the counters that collect the edge -coverage can easily fill up and wrap around. -This is not that much of an issue - unless by chance it wraps just to a value -of zero when the program execution ends. -In this case afl-fuzz is not able to see that the edge has been accessed and -will ignore it. - -NeverZero prevents this behaviour. If a counter wraps, it jumps over the value -0 directly to a 1. This improves path discovery (by a very little amount) -at a very little cost (one instruction per edge). - -(The alternative of saturated counters has been tested also and proved to be -inferior in terms of path discovery.) - -This is implemented in afl-gcc, however for llvm_mode this is optional if -the llvm version is below 9 - as there is a perfomance bug that is only fixed -in version 9 and onwards. - -If you want to enable this for llvm versions below 9 then set - -``` -export AFL_LLVM_NOT_ZERO=1 -``` - -In case you are on llvm 9 or greater and you do not want this behaviour then -you can set: -``` -AFL_LLVM_SKIP_NEVERZERO=1 -``` -If the target does not have extensive loops or functions that are called -a lot then this can give a small performance boost. diff --git a/llvm_mode/README.ngram.md b/llvm_mode/README.ngram.md deleted file mode 100644 index de3ba432..00000000 --- a/llvm_mode/README.ngram.md +++ /dev/null @@ -1,28 +0,0 @@ -# AFL N-Gram Branch Coverage - -## Source - -This is an LLVM-based implementation of the n-gram branch coverage proposed in -the paper ["Be Sensitive and Collaborative: Analzying Impact of Coverage Metrics -in Greybox Fuzzing"](https://www.usenix.org/system/files/raid2019-wang-jinghan.pdf), -by Jinghan Wang, et. al. - -Note that the original implementation (available -[here](https://github.com/bitsecurerlab/afl-sensitive)) -is built on top of AFL's QEMU mode. -This is essentially a port that uses LLVM vectorized instructions to achieve -the same results when compiling source code. - -In math the branch coverage is performed as follows: -`map[current_location ^ prev_location[0] >> 1 ^ prev_location[1] >> 1 ^ ... up to n-1`] += 1` - -## Usage - -The size of `n` (i.e., the number of branches to remember) is an option -that is specified either in the `AFL_LLVM_INSTRUMENT=NGRAM-{value}` or the -`AFL_LLVM_NGRAM_SIZE` environment variable. -Good values are 2, 4 or 8, valid are 2-16. - -It is highly recommended to increase the MAP_SIZE_POW2 definition in -config.h to at least 18 and maybe up to 20 for this as otherwise too -many map collisions occur. diff --git a/llvm_mode/README.persistent_mode.md b/llvm_mode/README.persistent_mode.md deleted file mode 100644 index 7d2fd93b..00000000 --- a/llvm_mode/README.persistent_mode.md +++ /dev/null @@ -1,209 +0,0 @@ -# llvm_mode persistent mode - -## 1) Introduction - -The most effective way is to fuzz in persistent mode, as the speed can easily -be x10 or x20 times faster without any disadvanges. -*All professionel fuzzing is using this mode.* - -This requires that the target can be called in a (or several) function(s), -and that the state can be resetted so that multiple calls be be performed -without memory leaking and former runs having no impact on following runs -(this can be seen by the `stability` indicator in the `afl-fuzz` UI). - -Examples can be found in [examples/persistent_mode](../examples/persistent_mode). - -## 2) TLDR; - -Example `fuzz_target.c`: -``` -#include "what_you_need_for_your_target.h" - -__AFL_FUZZ_INIT(); - -main() { - -#ifdef __AFL_HAVE_MANUAL_CONTROL - __AFL_INIT(); -#endif - - unsigned char *buf = __AFL_FUZZ_TESTCASE_BUF; // must be after __AFL_INIT - - while (__AFL_LOOP(10000)) { - - int len = __AFL_FUZZ_TESTCASE_LEN; - if (len < 8) continue; // check for a required/useful minimum input length - - /* Setup function call, e.g. struct target *tmp = libtarget_init() */ - /* Call function to be fuzzed, e.g.: */ - target_function(buf, len); - /* Reset state. e.g. libtarget_free(tmp) */ - - } - - return 0; - -} -``` -And then compile: -``` -afl-clang-fast -o fuzz_target fuzz_target.c -lwhat_you_need_for_your_target -``` -And that is it! -The speed increase is usually x10 to x20. - -If you want to be able to compile the target without afl-clang-fast/lto then -add this just after the includes: - -``` -#ifndef __AFL_FUZZ_TESTCASE_LEN - ssize_t fuzz_len; - #define __AFL_FUZZ_TESTCASE_LEN fuzz_len - unsigned char fuzz_buf[1024000]; - #define __AFL_FUZZ_TESTCASE_BUF fuzz_buf - #define __AFL_FUZZ_INIT() void sync(void); - #define __AFL_LOOP(x) ((fuzz_len = read(0, fuzz_buf, sizeof(fuzz_buf))) > 0 ? - #define __AFL_INIT() sync() -#endif -``` - -## 3) deferred initialization - -AFL tries to optimize performance by executing the targeted binary just once, -stopping it just before main(), and then cloning this "main" process to get -a steady supply of targets to fuzz. - -Although this approach eliminates much of the OS-, linker- and libc-level -costs of executing the program, it does not always help with binaries that -perform other time-consuming initialization steps - say, parsing a large config -file before getting to the fuzzed data. - -In such cases, it's beneficial to initialize the forkserver a bit later, once -most of the initialization work is already done, but before the binary attempts -to read the fuzzed input and parse it; in some cases, this can offer a 10x+ -performance gain. You can implement delayed initialization in LLVM mode in a -fairly simple way. - -First, find a suitable location in the code where the delayed cloning can -take place. This needs to be done with *extreme* care to avoid breaking the -binary. In particular, the program will probably malfunction if you select -a location after: - - - The creation of any vital threads or child processes - since the forkserver - can't clone them easily. - - - The initialization of timers via setitimer() or equivalent calls. - - - The creation of temporary files, network sockets, offset-sensitive file - descriptors, and similar shared-state resources - but only provided that - their state meaningfully influences the behavior of the program later on. - - - Any access to the fuzzed input, including reading the metadata about its - size. - -With the location selected, add this code in the appropriate spot: - -```c -#ifdef __AFL_HAVE_MANUAL_CONTROL - __AFL_INIT(); -#endif -``` - -You don't need the #ifdef guards, but including them ensures that the program -will keep working normally when compiled with a tool other than afl-clang-fast. - -Finally, recompile the program with afl-clang-fast (afl-gcc or afl-clang will -*not* generate a deferred-initialization binary) - and you should be all set! - -*NOTE:* In the code between `main` and `__AFL_INIT()` should not be any code -run that is instrumented - otherwise a crash might occure. -In case this is useful (e.g. for expensive one time initialization) you can -try to do the following: - -Add after the includes: -``` -extern unsigned char *__afl_area_ptr; -#define MAX_DUMMY_SIZE 256000 - -__attribute__((constructor(1))) void __afl_protect(void) { -#ifdef MAP_FIXED_NOREPLACE - __afl_area_ptr = (unsigned char*) mmap((void *)0x10000, MAX_DUMMY_SIZE, PROT_READ | PROT_WRITE, MAP_FIXED_NOREPLACE | MAP_SHARED | MAP_ANONYMOUS, -1, 0); - if ((uint64_t)__afl_area_ptr == -1) -#endif - __afl_area_ptr = (unsigned char*) mmap((void *)0x10000, MAX_DUMMY_SIZE, PROT_READ | PROT_WRITE, MAP_SHARED | MAP_ANONYMOUS, -1, 0); - if ((uint64_t)__afl_area_ptr == -1) - __afl_area_ptr = (unsigned char*) mmap(NULL, MAX_DUMMY_SIZE, PROT_READ | PROT_WRITE, MAP_SHARED | MAP_ANONYMOUS, -1, 0); -} - -``` -and just before `__AFL_INIT()`: -``` - munmap(__afl_area_ptr, MAX_DUMMY_SIZE); - __afl_area_ptr = NULL; -``` - -## 4) persistent mode - -Some libraries provide APIs that are stateless, or whose state can be reset in -between processing different input files. When such a reset is performed, a -single long-lived process can be reused to try out multiple test cases, -eliminating the need for repeated fork() calls and the associated OS overhead. - -The basic structure of the program that does this would be: - -```c - while (__AFL_LOOP(1000)) { - - /* Read input data. */ - /* Call library code to be fuzzed. */ - /* Reset state. */ - - } - - /* Exit normally */ -``` - -The numerical value specified within the loop controls the maximum number -of iterations before AFL will restart the process from scratch. This minimizes -the impact of memory leaks and similar glitches; 1000 is a good starting point, -and going much higher increases the likelihood of hiccups without giving you -any real performance benefits. - -A more detailed template is shown in ../examples/persistent_demo/. -Similarly to the previous mode, the feature works only with afl-clang-fast; #ifdef -guards can be used to suppress it when using other compilers. - -Note that as with the previous mode, the feature is easy to misuse; if you -do not fully reset the critical state, you may end up with false positives or -waste a whole lot of CPU power doing nothing useful at all. Be particularly -wary of memory leaks and of the state of file descriptors. - -PS. Because there are task switches still involved, the mode isn't as fast as -"pure" in-process fuzzing offered, say, by LLVM's LibFuzzer; but it is a lot -faster than the normal fork() model, and compared to in-process fuzzing, -should be a lot more robust. - -## 5) shared memory fuzzing - -You can speed up the fuzzing process even more by receiving the fuzzing data -via shared memory instead of stdin or files. -This is a further speed multiplier of about 2x. - -Setting this up is very easy: - -After the includes set the following macro: - -``` -__AFL_FUZZ_INIT(); -``` -Directly at the start of main - or if you are using the deferred forkserver -with `__AFL_INIT()` then *after* `__AFL_INIT? : -``` - unsigned char *buf = __AFL_FUZZ_TESTCASE_BUF; -``` - -Then as first line after the `__AFL_LOOP` while loop: -``` - int len = __AFL_FUZZ_TESTCASE_LEN; -``` -and that is all! diff --git a/llvm_mode/README.snapshot.md b/llvm_mode/README.snapshot.md deleted file mode 100644 index 9c12a8ba..00000000 --- a/llvm_mode/README.snapshot.md +++ /dev/null @@ -1,16 +0,0 @@ -# AFL++ snapshot feature - -Snapshotting is a feature that makes a snapshot from a process and then -restores it's state, which is faster then forking it again. - -All targets compiled with llvm_mode are automatically enabled for the -snapshot feature. - -To use the snapshot feature for fuzzing compile and load this kernel -module: [https://github.com/AFLplusplus/AFL-Snapshot-LKM](https://github.com/AFLplusplus/AFL-Snapshot-LKM) - -Note that is has little value for persistent (__AFL_LOOP) fuzzing. - -## Notes - -Snapshot does not work with multithreaded targets yet. Still in WIP, it is now usable only for single threaded applications. diff --git a/llvm_mode/SanitizerCoverageLTO.so.cc b/llvm_mode/SanitizerCoverageLTO.so.cc deleted file mode 100644 index 1dd65188..00000000 --- a/llvm_mode/SanitizerCoverageLTO.so.cc +++ /dev/null @@ -1,1503 +0,0 @@ -/* SanitizeCoverage.cpp ported to afl++ LTO :-) */ - -#define AFL_LLVM_PASS - -#include -#include -#include -#include -#include - -#include -#include -#include -#include - -#include "llvm/Transforms/Instrumentation/SanitizerCoverage.h" -#include "llvm/ADT/ArrayRef.h" -#include "llvm/ADT/SmallVector.h" -#include "llvm/Analysis/EHPersonalities.h" -#include "llvm/Analysis/PostDominators.h" -#include "llvm/Analysis/ValueTracking.h" -#include "llvm/IR/BasicBlock.h" -#include "llvm/IR/CFG.h" -#include "llvm/IR/Constant.h" -#include "llvm/IR/DataLayout.h" -#include "llvm/IR/DebugInfo.h" -#include "llvm/IR/Dominators.h" -#include "llvm/IR/Function.h" -#include "llvm/IR/GlobalVariable.h" -#include "llvm/IR/IRBuilder.h" -#include "llvm/IR/InlineAsm.h" -#include "llvm/IR/Instructions.h" -#include "llvm/IR/IntrinsicInst.h" -#include "llvm/IR/Intrinsics.h" -#include "llvm/IR/LegacyPassManager.h" -#include "llvm/IR/LLVMContext.h" -#include "llvm/IR/MDBuilder.h" -#include "llvm/IR/Mangler.h" -#include "llvm/IR/Module.h" -#include "llvm/IR/Type.h" -#include "llvm/InitializePasses.h" -#include "llvm/Pass.h" -#include "llvm/Support/CommandLine.h" -#include "llvm/Support/Debug.h" -#include "llvm/Support/SpecialCaseList.h" -#include "llvm/Support/VirtualFileSystem.h" -#include "llvm/Support/raw_ostream.h" -#include "llvm/Transforms/Instrumentation.h" -#include "llvm/Transforms/IPO/PassManagerBuilder.h" -#include "llvm/Transforms/Utils/BasicBlockUtils.h" -#include "llvm/Transforms/Utils/BasicBlockUtils.h" -#include "llvm/Transforms/Utils/ModuleUtils.h" - -#include "config.h" -#include "debug.h" -#include "afl-llvm-common.h" - -using namespace llvm; - -#define DEBUG_TYPE "sancov" - -static const char *const SanCovTracePCIndirName = - "__sanitizer_cov_trace_pc_indir"; -static const char *const SanCovTracePCName = "__sanitizer_cov_trace_pc"; -// static const char *const SanCovTracePCGuardName = -// "__sanitizer_cov_trace_pc_guard"; -static const char *const SanCovGuardsSectionName = "sancov_guards"; -static const char *const SanCovCountersSectionName = "sancov_cntrs"; -static const char *const SanCovBoolFlagSectionName = "sancov_bools"; -static const char *const SanCovPCsSectionName = "sancov_pcs"; - -static cl::opt ClCoverageLevel( - "lto-coverage-level", - cl::desc("Sanitizer Coverage. 0: none, 1: entry block, 2: all blocks, " - "3: all blocks and critical edges"), - cl::Hidden, cl::init(3)); - -static cl::opt ClTracePC("lto-coverage-trace-pc", - cl::desc("Experimental pc tracing"), cl::Hidden, - cl::init(false)); - -static cl::opt ClTracePCGuard("lto-coverage-trace-pc-guard", - cl::desc("pc tracing with a guard"), - cl::Hidden, cl::init(false)); - -// If true, we create a global variable that contains PCs of all instrumented -// BBs, put this global into a named section, and pass this section's bounds -// to __sanitizer_cov_pcs_init. -// This way the coverage instrumentation does not need to acquire the PCs -// at run-time. Works with trace-pc-guard, inline-8bit-counters, and -// inline-bool-flag. -static cl::opt ClCreatePCTable("lto-coverage-pc-table", - cl::desc("create a static PC table"), - cl::Hidden, cl::init(false)); - -static cl::opt ClInline8bitCounters( - "lto-coverage-inline-8bit-counters", - cl::desc("increments 8-bit counter for every edge"), cl::Hidden, - cl::init(false)); - -static cl::opt ClInlineBoolFlag( - "lto-coverage-inline-bool-flag", - cl::desc("sets a boolean flag for every edge"), cl::Hidden, - cl::init(false)); - -static cl::opt ClPruneBlocks( - "lto-coverage-prune-blocks", - cl::desc("Reduce the number of instrumented blocks"), cl::Hidden, - cl::init(true)); - -namespace { - -SanitizerCoverageOptions getOptions(int LegacyCoverageLevel) { - - SanitizerCoverageOptions Res; - switch (LegacyCoverageLevel) { - - case 0: - Res.CoverageType = SanitizerCoverageOptions::SCK_None; - break; - case 1: - Res.CoverageType = SanitizerCoverageOptions::SCK_Function; - break; - case 2: - Res.CoverageType = SanitizerCoverageOptions::SCK_BB; - break; - case 3: - Res.CoverageType = SanitizerCoverageOptions::SCK_Edge; - break; - case 4: - Res.CoverageType = SanitizerCoverageOptions::SCK_Edge; - Res.IndirectCalls = true; - break; - - } - - return Res; - -} - -SanitizerCoverageOptions OverrideFromCL(SanitizerCoverageOptions Options) { - - // Sets CoverageType and IndirectCalls. - SanitizerCoverageOptions CLOpts = getOptions(ClCoverageLevel); - Options.CoverageType = std::max(Options.CoverageType, CLOpts.CoverageType); - Options.IndirectCalls |= CLOpts.IndirectCalls; - Options.TracePC |= ClTracePC; - Options.TracePCGuard |= ClTracePCGuard; - Options.Inline8bitCounters |= ClInline8bitCounters; - Options.InlineBoolFlag |= ClInlineBoolFlag; - Options.PCTable |= ClCreatePCTable; - Options.NoPrune |= !ClPruneBlocks; - if (!Options.TracePCGuard && !Options.TracePC && - !Options.Inline8bitCounters && !Options.InlineBoolFlag) - Options.TracePCGuard = true; // TracePCGuard is default. - return Options; - -} - -using DomTreeCallback = function_ref; -using PostDomTreeCallback = - function_ref; - -class ModuleSanitizerCoverage { - - public: - ModuleSanitizerCoverage( - const SanitizerCoverageOptions &Options = SanitizerCoverageOptions()) - : Options(OverrideFromCL(Options)) { - - /* , - const SpecialCaseList * Allowlist = nullptr, - const SpecialCaseList * Blocklist = nullptr) - , - Allowlist(Allowlist), - Blocklist(Blocklist) { - - */ - - } - - bool instrumentModule(Module &M, DomTreeCallback DTCallback, - PostDomTreeCallback PDTCallback); - - private: - void instrumentFunction(Function &F, DomTreeCallback DTCallback, - PostDomTreeCallback PDTCallback); - void InjectCoverageForIndirectCalls(Function & F, - ArrayRef IndirCalls); - bool InjectCoverage(Function &F, ArrayRef AllBlocks, - bool IsLeafFunc = true); - GlobalVariable *CreateFunctionLocalArrayInSection(size_t NumElements, - Function &F, Type *Ty, - const char *Section); - GlobalVariable *CreatePCArray(Function &F, ArrayRef AllBlocks); - void CreateFunctionLocalArrays(Function &F, ArrayRef AllBlocks); - void InjectCoverageAtBlock(Function &F, BasicBlock &BB, size_t Idx, - bool IsLeafFunc = true); - std::pair CreateSecStartEnd(Module &M, const char *Section, - Type *Ty); - - void SetNoSanitizeMetadata(Instruction *I) { - - I->setMetadata(I->getModule()->getMDKindID("nosanitize"), - MDNode::get(*C, None)); - - } - - std::string getSectionName(const std::string &Section) const; - std::string getSectionStart(const std::string &Section) const; - std::string getSectionEnd(const std::string &Section) const; - FunctionCallee SanCovTracePCIndir; - FunctionCallee SanCovTracePC /*, SanCovTracePCGuard*/; - Type *IntptrTy, *IntptrPtrTy, *Int64Ty, *Int64PtrTy, *Int32Ty, *Int32PtrTy, - *Int16Ty, *Int8Ty, *Int8PtrTy, *Int1Ty, *Int1PtrTy; - Module * CurModule; - std::string CurModuleUniqueId; - Triple TargetTriple; - LLVMContext * C; - const DataLayout *DL; - - GlobalVariable *FunctionGuardArray; // for trace-pc-guard. - GlobalVariable *Function8bitCounterArray; // for inline-8bit-counters. - GlobalVariable *FunctionBoolArray; // for inline-bool-flag. - GlobalVariable *FunctionPCsArray; // for pc-table. - SmallVector GlobalsToAppendToUsed; - SmallVector GlobalsToAppendToCompilerUsed; - - SanitizerCoverageOptions Options; - - // afl++ START - // const SpecialCaseList * Allowlist; - // const SpecialCaseList * Blocklist; - uint32_t autodictionary = 1; - uint32_t inst = 0; - uint32_t afl_global_id = 0; - uint64_t map_addr = 0; - char * skip_nozero = NULL; - std::vector BlockList; - DenseMap valueMap; - std::vector dictionary; - IntegerType * Int8Tyi = NULL; - IntegerType * Int32Tyi = NULL; - IntegerType * Int64Tyi = NULL; - ConstantInt * Zero = NULL; - ConstantInt * One = NULL; - LLVMContext * Ct = NULL; - Module * Mo = NULL; - GlobalVariable * AFLMapPtr = NULL; - Value * MapPtrFixed = NULL; - FILE * documentFile = NULL; - // afl++ END - -}; - -class ModuleSanitizerCoverageLegacyPass : public ModulePass { - - public: - static char ID; - StringRef getPassName() const override { - - return "sancov"; - - } - - void getAnalysisUsage(AnalysisUsage &AU) const override { - - AU.addRequired(); - AU.addRequired(); - - } - - ModuleSanitizerCoverageLegacyPass( - const SanitizerCoverageOptions &Options = SanitizerCoverageOptions()) - : ModulePass(ID), Options(Options) { - - /* , - const std::vector &AllowlistFiles = - std::vector(), - const std::vector &BlocklistFiles = - std::vector()) - if (AllowlistFiles.size() > 0) - Allowlist = SpecialCaseList::createOrDie(AllowlistFiles, - *vfs::getRealFileSystem()); - if (BlocklistFiles.size() > 0) - Blocklist = SpecialCaseList::createOrDie(BlocklistFiles, - *vfs::getRealFileSystem()); - */ - initializeModuleSanitizerCoverageLegacyPassPass( - *PassRegistry::getPassRegistry()); - - } - - bool runOnModule(Module &M) override { - - ModuleSanitizerCoverage ModuleSancov(Options); - // , Allowlist.get(), Blocklist.get()); - auto DTCallback = [this](Function &F) -> const DominatorTree * { - - return &this->getAnalysis(F).getDomTree(); - - }; - - auto PDTCallback = [this](Function &F) -> const PostDominatorTree * { - - return &this->getAnalysis(F) - .getPostDomTree(); - - }; - - return ModuleSancov.instrumentModule(M, DTCallback, PDTCallback); - - } - - private: - SanitizerCoverageOptions Options; - - // std::unique_ptr Allowlist; - // std::unique_ptr Blocklist; - -}; - -} // namespace - -PreservedAnalyses ModuleSanitizerCoveragePass::run(Module & M, - ModuleAnalysisManager &MAM) { - - ModuleSanitizerCoverage ModuleSancov(Options); - // Allowlist.get(), Blocklist.get()); - auto &FAM = MAM.getResult(M).getManager(); - auto DTCallback = [&FAM](Function &F) -> const DominatorTree * { - - return &FAM.getResult(F); - - }; - - auto PDTCallback = [&FAM](Function &F) -> const PostDominatorTree * { - - return &FAM.getResult(F); - - }; - - if (ModuleSancov.instrumentModule(M, DTCallback, PDTCallback)) - return PreservedAnalyses::none(); - - return PreservedAnalyses::all(); - -} - -std::pair ModuleSanitizerCoverage::CreateSecStartEnd( - Module &M, const char *Section, Type *Ty) { - - GlobalVariable *SecStart = - new GlobalVariable(M, Ty, false, GlobalVariable::ExternalLinkage, nullptr, - getSectionStart(Section)); - SecStart->setVisibility(GlobalValue::HiddenVisibility); - GlobalVariable *SecEnd = - new GlobalVariable(M, Ty, false, GlobalVariable::ExternalLinkage, nullptr, - getSectionEnd(Section)); - SecEnd->setVisibility(GlobalValue::HiddenVisibility); - IRBuilder<> IRB(M.getContext()); - Value * SecEndPtr = IRB.CreatePointerCast(SecEnd, Ty); - if (!TargetTriple.isOSBinFormatCOFF()) - return std::make_pair(IRB.CreatePointerCast(SecStart, Ty), SecEndPtr); - - // Account for the fact that on windows-msvc __start_* symbols actually - // point to a uint64_t before the start of the array. - auto SecStartI8Ptr = IRB.CreatePointerCast(SecStart, Int8PtrTy); - auto GEP = IRB.CreateGEP(Int8Ty, SecStartI8Ptr, - ConstantInt::get(IntptrTy, sizeof(uint64_t))); - return std::make_pair(IRB.CreatePointerCast(GEP, Ty), SecEndPtr); - -} - -bool ModuleSanitizerCoverage::instrumentModule( - Module &M, DomTreeCallback DTCallback, PostDomTreeCallback PDTCallback) { - - if (Options.CoverageType == SanitizerCoverageOptions::SCK_None) return false; - /* - if (Allowlist && - !Allowlist->inSection("coverage", "src", M.getSourceFileName())) - return false; - if (Blocklist && - Blocklist->inSection("coverage", "src", M.getSourceFileName())) - return false; - */ - BlockList.clear(); - valueMap.clear(); - dictionary.clear(); - C = &(M.getContext()); - DL = &M.getDataLayout(); - CurModule = &M; - CurModuleUniqueId = getUniqueModuleId(CurModule); - TargetTriple = Triple(M.getTargetTriple()); - FunctionGuardArray = nullptr; - Function8bitCounterArray = nullptr; - FunctionBoolArray = nullptr; - FunctionPCsArray = nullptr; - IntptrTy = Type::getIntNTy(*C, DL->getPointerSizeInBits()); - IntptrPtrTy = PointerType::getUnqual(IntptrTy); - Type * VoidTy = Type::getVoidTy(*C); - IRBuilder<> IRB(*C); - Int64PtrTy = PointerType::getUnqual(IRB.getInt64Ty()); - Int32PtrTy = PointerType::getUnqual(IRB.getInt32Ty()); - Int8PtrTy = PointerType::getUnqual(IRB.getInt8Ty()); - Int1PtrTy = PointerType::getUnqual(IRB.getInt1Ty()); - Int64Ty = IRB.getInt64Ty(); - Int32Ty = IRB.getInt32Ty(); - Int16Ty = IRB.getInt16Ty(); - Int8Ty = IRB.getInt8Ty(); - Int1Ty = IRB.getInt1Ty(); - - /* afl++ START */ - char * ptr; - LLVMContext &Ctx = M.getContext(); - Ct = &Ctx; - Int8Tyi = IntegerType::getInt8Ty(Ctx); - Int32Tyi = IntegerType::getInt32Ty(Ctx); - Int64Tyi = IntegerType::getInt64Ty(Ctx); - - /* Show a banner */ - setvbuf(stdout, NULL, _IONBF, 0); - if (getenv("AFL_DEBUG")) debug = 1; - - if ((isatty(2) && !getenv("AFL_QUIET")) || debug) { - - SAYF(cCYA "afl-llvm-lto" VERSION cRST - " by Marc \"vanHauser\" Heuse \n"); - - } else - - be_quiet = 1; - - skip_nozero = getenv("AFL_LLVM_SKIP_NEVERZERO"); - - if ((ptr = getenv("AFL_LLVM_LTO_STARTID")) != NULL) - if ((afl_global_id = atoi(ptr)) < 0) - FATAL("AFL_LLVM_LTO_STARTID value of \"%s\" is negative\n", ptr); - - if ((ptr = getenv("AFL_LLVM_DOCUMENT_IDS")) != NULL) { - - if ((documentFile = fopen(ptr, "a")) == NULL) - WARNF("Cannot access document file %s", ptr); - - } - - // we make this the default as the fixed map has problems with - // defered forkserver, early constructors, ifuncs and maybe more - /*if (getenv("AFL_LLVM_MAP_DYNAMIC"))*/ - map_addr = 0; - - if ((ptr = getenv("AFL_LLVM_MAP_ADDR"))) { - - uint64_t val; - if (!*ptr || !strcmp(ptr, "0") || !strcmp(ptr, "0x0")) { - - map_addr = 0; - - } else if (getenv("AFL_LLVM_MAP_DYNAMIC")) { - - FATAL( - "AFL_LLVM_MAP_ADDR and AFL_LLVM_MAP_DYNAMIC cannot be used together"); - - } else if (strncmp(ptr, "0x", 2) != 0) { - - map_addr = 0x10000; // the default - - } else { - - val = strtoull(ptr, NULL, 16); - if (val < 0x100 || val > 0xffffffff00000000) { - - FATAL( - "AFL_LLVM_MAP_ADDR must be a value between 0x100 and " - "0xffffffff00000000"); - - } - - map_addr = val; - - } - - } - - /* Get/set the globals for the SHM region. */ - - if (!map_addr) { - - AFLMapPtr = - new GlobalVariable(M, PointerType::get(Int8Tyi, 0), false, - GlobalValue::ExternalLinkage, 0, "__afl_area_ptr"); - - } else { - - ConstantInt *MapAddr = ConstantInt::get(Int64Tyi, map_addr); - MapPtrFixed = - ConstantExpr::getIntToPtr(MapAddr, PointerType::getUnqual(Int8Tyi)); - - } - - Zero = ConstantInt::get(Int8Tyi, 0); - One = ConstantInt::get(Int8Tyi, 1); - - scanForDangerousFunctions(&M); - Mo = &M; - - if (autodictionary) { - - for (auto &F : M) { - - for (auto &BB : F) { - - for (auto &IN : BB) { - - CallInst *callInst = nullptr; - - if ((callInst = dyn_cast(&IN))) { - - bool isStrcmp = true; - bool isMemcmp = true; - bool isStrncmp = true; - bool isStrcasecmp = true; - bool isStrncasecmp = true; - bool isIntMemcpy = true; - bool addedNull = false; - size_t optLen = 0; - - Function *Callee = callInst->getCalledFunction(); - if (!Callee) continue; - if (callInst->getCallingConv() != llvm::CallingConv::C) continue; - std::string FuncName = Callee->getName().str(); - isStrcmp &= !FuncName.compare("strcmp"); - isMemcmp &= !FuncName.compare("memcmp"); - isStrncmp &= !FuncName.compare("strncmp"); - isStrcasecmp &= !FuncName.compare("strcasecmp"); - isStrncasecmp &= !FuncName.compare("strncasecmp"); - isIntMemcpy &= !FuncName.compare("llvm.memcpy.p0i8.p0i8.i64"); - - /* we do something different here, putting this BB and the - successors in a block map */ - if (!FuncName.compare("__afl_persistent_loop")) { - - BlockList.push_back(&BB); - for (succ_iterator SI = succ_begin(&BB), SE = succ_end(&BB); - SI != SE; ++SI) { - - BasicBlock *succ = *SI; - BlockList.push_back(succ); - - } - - } - - if (!isStrcmp && !isMemcmp && !isStrncmp && !isStrcasecmp && - !isStrncasecmp && !isIntMemcpy) - continue; - - /* Verify the strcmp/memcmp/strncmp/strcasecmp/strncasecmp function - * prototype */ - FunctionType *FT = Callee->getFunctionType(); - - isStrcmp &= FT->getNumParams() == 2 && - FT->getReturnType()->isIntegerTy(32) && - FT->getParamType(0) == FT->getParamType(1) && - FT->getParamType(0) == - IntegerType::getInt8PtrTy(M.getContext()); - isStrcasecmp &= FT->getNumParams() == 2 && - FT->getReturnType()->isIntegerTy(32) && - FT->getParamType(0) == FT->getParamType(1) && - FT->getParamType(0) == - IntegerType::getInt8PtrTy(M.getContext()); - isMemcmp &= FT->getNumParams() == 3 && - FT->getReturnType()->isIntegerTy(32) && - FT->getParamType(0)->isPointerTy() && - FT->getParamType(1)->isPointerTy() && - FT->getParamType(2)->isIntegerTy(); - isStrncmp &= FT->getNumParams() == 3 && - FT->getReturnType()->isIntegerTy(32) && - FT->getParamType(0) == FT->getParamType(1) && - FT->getParamType(0) == - IntegerType::getInt8PtrTy(M.getContext()) && - FT->getParamType(2)->isIntegerTy(); - isStrncasecmp &= FT->getNumParams() == 3 && - FT->getReturnType()->isIntegerTy(32) && - FT->getParamType(0) == FT->getParamType(1) && - FT->getParamType(0) == - IntegerType::getInt8PtrTy(M.getContext()) && - FT->getParamType(2)->isIntegerTy(); - - if (!isStrcmp && !isMemcmp && !isStrncmp && !isStrcasecmp && - !isStrncasecmp && !isIntMemcpy) - continue; - - /* is a str{n,}{case,}cmp/memcmp, check if we have - * str{case,}cmp(x, "const") or str{case,}cmp("const", x) - * strn{case,}cmp(x, "const", ..) or strn{case,}cmp("const", x, ..) - * memcmp(x, "const", ..) or memcmp("const", x, ..) */ - Value *Str1P = callInst->getArgOperand(0), - *Str2P = callInst->getArgOperand(1); - std::string Str1, Str2; - StringRef TmpStr; - bool HasStr1 = getConstantStringInfo(Str1P, TmpStr); - if (TmpStr.empty()) - HasStr1 = false; - else - Str1 = TmpStr.str(); - bool HasStr2 = getConstantStringInfo(Str2P, TmpStr); - if (TmpStr.empty()) - HasStr2 = false; - else - Str2 = TmpStr.str(); - - if (debug) - fprintf(stderr, "F:%s %p(%s)->\"%s\"(%s) %p(%s)->\"%s\"(%s)\n", - FuncName.c_str(), Str1P, Str1P->getName().str().c_str(), - Str1.c_str(), HasStr1 == true ? "true" : "false", Str2P, - Str2P->getName().str().c_str(), Str2.c_str(), - HasStr2 == true ? "true" : "false"); - - // we handle the 2nd parameter first because of llvm memcpy - if (!HasStr2) { - - auto *Ptr = dyn_cast(Str2P); - if (Ptr && Ptr->isGEPWithNoNotionalOverIndexing()) { - - if (auto *Var = dyn_cast(Ptr->getOperand(0))) { - - if (Var->hasInitializer()) { - - if (auto *Array = dyn_cast( - Var->getInitializer())) { - - HasStr2 = true; - Str2 = Array->getAsString().str(); - - } - - } - - } - - } - - } - - // for the internal memcpy routine we only care for the second - // parameter and are not reporting anything. - if (isIntMemcpy == true) { - - if (HasStr2 == true) { - - Value * op2 = callInst->getArgOperand(2); - ConstantInt *ilen = dyn_cast(op2); - if (ilen) { - - uint64_t literalLength = Str2.size(); - uint64_t optLength = ilen->getZExtValue(); - if (literalLength + 1 == optLength) { - - Str2.append("\0", 1); // add null byte - addedNull = true; - - } - - } - - valueMap[Str1P] = new std::string(Str2); - - if (debug) - fprintf(stderr, "Saved: %s for %p\n", Str2.c_str(), Str1P); - continue; - - } - - continue; - - } - - // Neither a literal nor a global variable? - // maybe it is a local variable that we saved - if (!HasStr2) { - - std::string *strng = valueMap[Str2P]; - if (strng && !strng->empty()) { - - Str2 = *strng; - HasStr2 = true; - if (debug) - fprintf(stderr, "Filled2: %s for %p\n", strng->c_str(), - Str2P); - - } - - } - - if (!HasStr1) { - - auto Ptr = dyn_cast(Str1P); - - if (Ptr && Ptr->isGEPWithNoNotionalOverIndexing()) { - - if (auto *Var = dyn_cast(Ptr->getOperand(0))) { - - if (Var->hasInitializer()) { - - if (auto *Array = dyn_cast( - Var->getInitializer())) { - - HasStr1 = true; - Str1 = Array->getAsString().str(); - - } - - } - - } - - } - - } - - // Neither a literal nor a global variable? - // maybe it is a local variable that we saved - if (!HasStr1) { - - std::string *strng = valueMap[Str1P]; - if (strng && !strng->empty()) { - - Str1 = *strng; - HasStr1 = true; - if (debug) - fprintf(stderr, "Filled1: %s for %p\n", strng->c_str(), - Str1P); - - } - - } - - /* handle cases of one string is const, one string is variable */ - if (!(HasStr1 ^ HasStr2)) continue; - - std::string thestring; - - if (HasStr1) - thestring = Str1; - else - thestring = Str2; - - optLen = thestring.length(); - - if (isMemcmp || isStrncmp || isStrncasecmp) { - - Value * op2 = callInst->getArgOperand(2); - ConstantInt *ilen = dyn_cast(op2); - if (ilen) { - - uint64_t literalLength = optLen; - optLen = ilen->getZExtValue(); - if (literalLength + 1 == optLen) { // add null byte - thestring.append("\0", 1); - addedNull = true; - - } - - } - - } - - // add null byte if this is a string compare function and a null - // was not already added - if (!isMemcmp) { - - if (addedNull == false) { - - thestring.append("\0", 1); // add null byte - optLen++; - - } - - // ensure we do not have garbage - size_t offset = thestring.find('\0', 0); - if (offset + 1 < optLen) optLen = offset + 1; - thestring = thestring.substr(0, optLen); - - } - - if (!be_quiet) { - - std::string outstring; - fprintf(stderr, "%s: length %zu/%zu \"", FuncName.c_str(), optLen, - thestring.length()); - for (uint8_t i = 0; i < thestring.length(); i++) { - - uint8_t c = thestring[i]; - if (c <= 32 || c >= 127) - fprintf(stderr, "\\x%02x", c); - else - fprintf(stderr, "%c", c); - - } - - fprintf(stderr, "\"\n"); - - } - - // we take the longer string, even if the compare was to a - // shorter part. Note that depending on the optimizer of the - // compiler this can be wrong, but it is more likely that this - // is helping the fuzzer - if (optLen != thestring.length()) optLen = thestring.length(); - if (optLen > MAX_AUTO_EXTRA) optLen = MAX_AUTO_EXTRA; - if (optLen < MIN_AUTO_EXTRA) // too short? skip - continue; - - dictionary.push_back(thestring.substr(0, optLen)); - - } - - } - - } - - } - - } - - // afl++ END - - SanCovTracePCIndir = - M.getOrInsertFunction(SanCovTracePCIndirName, VoidTy, IntptrTy); - // Make sure smaller parameters are zero-extended to i64 as required by the - // x86_64 ABI. - AttributeList SanCovTraceCmpZeroExtAL; - if (TargetTriple.getArch() == Triple::x86_64) { - - SanCovTraceCmpZeroExtAL = - SanCovTraceCmpZeroExtAL.addParamAttribute(*C, 0, Attribute::ZExt); - SanCovTraceCmpZeroExtAL = - SanCovTraceCmpZeroExtAL.addParamAttribute(*C, 1, Attribute::ZExt); - - } - - SanCovTracePC = M.getOrInsertFunction(SanCovTracePCName, VoidTy); - - // SanCovTracePCGuard = - // M.getOrInsertFunction(SanCovTracePCGuardName, VoidTy, Int32PtrTy); - - for (auto &F : M) - instrumentFunction(F, DTCallback, PDTCallback); - - // afl++ START - if (documentFile) { - - fclose(documentFile); - documentFile = NULL; - - } - - if (!getenv("AFL_LLVM_LTO_DONTWRITEID") || dictionary.size() || map_addr) { - - // yes we could create our own function, insert it into ctors ... - // but this would be a pain in the butt ... so we use afl-llvm-rt-lto.o - - Function *f = M.getFunction("__afl_auto_init_globals"); - - if (!f) { - - fprintf(stderr, - "Error: init function could not be found (this should not " - "happen)\n"); - exit(-1); - - } - - BasicBlock *bb = &f->getEntryBlock(); - if (!bb) { - - fprintf(stderr, - "Error: init function does not have an EntryBlock (this should " - "not happen)\n"); - exit(-1); - - } - - BasicBlock::iterator IP = bb->getFirstInsertionPt(); - IRBuilder<> IRB(&(*IP)); - - if (map_addr) { - - GlobalVariable *AFLMapAddrFixed = new GlobalVariable( - M, Int64Tyi, true, GlobalValue::ExternalLinkage, 0, "__afl_map_addr"); - ConstantInt *MapAddr = ConstantInt::get(Int64Tyi, map_addr); - StoreInst * StoreMapAddr = IRB.CreateStore(MapAddr, AFLMapAddrFixed); - StoreMapAddr->setMetadata(M.getMDKindID("nosanitize"), - MDNode::get(Ctx, None)); - - } - - if (getenv("AFL_LLVM_LTO_DONTWRITEID") == NULL) { - - uint32_t write_loc = afl_global_id; - - if (afl_global_id % 8) write_loc = (((afl_global_id + 8) >> 3) << 3); - - GlobalVariable *AFLFinalLoc = - new GlobalVariable(M, Int32Tyi, true, GlobalValue::ExternalLinkage, 0, - "__afl_final_loc"); - ConstantInt *const_loc = ConstantInt::get(Int32Tyi, write_loc); - StoreInst * StoreFinalLoc = IRB.CreateStore(const_loc, AFLFinalLoc); - StoreFinalLoc->setMetadata(M.getMDKindID("nosanitize"), - MDNode::get(Ctx, None)); - - } - - if (dictionary.size()) { - - size_t memlen = 0, count = 0, offset = 0; - char * ptr; - - for (auto token : dictionary) { - - memlen += token.length(); - count++; - - } - - if (!be_quiet) - printf("AUTODICTIONARY: %lu string%s found\n", count, - count == 1 ? "" : "s"); - - if (count) { - - if ((ptr = (char *)malloc(memlen + count)) == NULL) { - - fprintf(stderr, "Error: malloc for %lu bytes failed!\n", - memlen + count); - exit(-1); - - } - - count = 0; - - for (auto token : dictionary) { - - if (offset + token.length() < 0xfffff0 && count < MAX_AUTO_EXTRAS) { - - ptr[offset++] = (uint8_t)token.length(); - memcpy(ptr + offset, token.c_str(), token.length()); - offset += token.length(); - count++; - - } - - } - - GlobalVariable *AFLDictionaryLen = - new GlobalVariable(M, Int32Tyi, false, GlobalValue::ExternalLinkage, - 0, "__afl_dictionary_len"); - ConstantInt *const_len = ConstantInt::get(Int32Tyi, offset); - StoreInst *StoreDictLen = IRB.CreateStore(const_len, AFLDictionaryLen); - StoreDictLen->setMetadata(M.getMDKindID("nosanitize"), - MDNode::get(Ctx, None)); - - ArrayType *ArrayTy = ArrayType::get(IntegerType::get(Ctx, 8), offset); - GlobalVariable *AFLInternalDictionary = new GlobalVariable( - M, ArrayTy, true, GlobalValue::ExternalLinkage, - ConstantDataArray::get(Ctx, - *(new ArrayRef((char *)ptr, offset))), - "__afl_internal_dictionary"); - AFLInternalDictionary->setInitializer(ConstantDataArray::get( - Ctx, *(new ArrayRef((char *)ptr, offset)))); - AFLInternalDictionary->setConstant(true); - - GlobalVariable *AFLDictionary = new GlobalVariable( - M, PointerType::get(Int8Tyi, 0), false, - GlobalValue::ExternalLinkage, 0, "__afl_dictionary"); - - Value *AFLDictOff = IRB.CreateGEP(AFLInternalDictionary, Zero); - Value *AFLDictPtr = - IRB.CreatePointerCast(AFLDictOff, PointerType::get(Int8Tyi, 0)); - StoreInst *StoreDict = IRB.CreateStore(AFLDictPtr, AFLDictionary); - StoreDict->setMetadata(M.getMDKindID("nosanitize"), - MDNode::get(Ctx, None)); - - } - - } - - } - - /* Say something nice. */ - - if (!be_quiet) { - - if (!inst) - WARNF("No instrumentation targets found."); - else { - - char modeline[100]; - snprintf(modeline, sizeof(modeline), "%s%s%s%s%s", - getenv("AFL_HARDEN") ? "hardened" : "non-hardened", - getenv("AFL_USE_ASAN") ? ", ASAN" : "", - getenv("AFL_USE_MSAN") ? ", MSAN" : "", - getenv("AFL_USE_CFISAN") ? ", CFISAN" : "", - getenv("AFL_USE_UBSAN") ? ", UBSAN" : ""); - OKF("Instrumented %u locations with no collisions (on average %llu " - "collisions would be in afl-gcc/afl-clang-fast) (%s mode).", - inst, calculateCollisions(inst), modeline); - - } - - } - - // afl++ END - - // We don't reference these arrays directly in any of our runtime functions, - // so we need to prevent them from being dead stripped. - if (TargetTriple.isOSBinFormatMachO()) appendToUsed(M, GlobalsToAppendToUsed); - appendToCompilerUsed(M, GlobalsToAppendToCompilerUsed); - return true; - -} - -// True if block has successors and it dominates all of them. -static bool isFullDominator(const BasicBlock *BB, const DominatorTree *DT) { - - if (succ_begin(BB) == succ_end(BB)) return false; - - for (const BasicBlock *SUCC : make_range(succ_begin(BB), succ_end(BB))) { - - if (!DT->dominates(BB, SUCC)) return false; - - } - - return true; - -} - -// True if block has predecessors and it postdominates all of them. -static bool isFullPostDominator(const BasicBlock * BB, - const PostDominatorTree *PDT) { - - if (pred_begin(BB) == pred_end(BB)) return false; - - for (const BasicBlock *PRED : make_range(pred_begin(BB), pred_end(BB))) { - - if (!PDT->dominates(BB, PRED)) return false; - - } - - return true; - -} - -static bool shouldInstrumentBlock(const Function &F, const BasicBlock *BB, - const DominatorTree * DT, - const PostDominatorTree * PDT, - const SanitizerCoverageOptions &Options) { - - // Don't insert coverage for blocks containing nothing but unreachable: we - // will never call __sanitizer_cov() for them, so counting them in - // NumberOfInstrumentedBlocks() might complicate calculation of code coverage - // percentage. Also, unreachable instructions frequently have no debug - // locations. - if (isa(BB->getFirstNonPHIOrDbgOrLifetime())) return false; - - // Don't insert coverage into blocks without a valid insertion point - // (catchswitch blocks). - if (BB->getFirstInsertionPt() == BB->end()) return false; - - // afl++ START - if (!Options.NoPrune && &F.getEntryBlock() == BB && F.size() > 1) - return false; - // afl++ END - - if (Options.NoPrune || &F.getEntryBlock() == BB) return true; - - if (Options.CoverageType == SanitizerCoverageOptions::SCK_Function && - &F.getEntryBlock() != BB) - return false; - - // Do not instrument full dominators, or full post-dominators with multiple - // predecessors. - return !isFullDominator(BB, DT) && - !(isFullPostDominator(BB, PDT) && !BB->getSinglePredecessor()); - -} - -void ModuleSanitizerCoverage::instrumentFunction( - Function &F, DomTreeCallback DTCallback, PostDomTreeCallback PDTCallback) { - - if (F.empty()) return; - if (F.getName().find(".module_ctor") != std::string::npos) - return; // Should not instrument sanitizer init functions. - if (F.getName().startswith("__sanitizer_")) - return; // Don't instrument __sanitizer_* callbacks. - // Don't touch available_externally functions, their actual body is elewhere. - if (F.getLinkage() == GlobalValue::AvailableExternallyLinkage) return; - // Don't instrument MSVC CRT configuration helpers. They may run before normal - // initialization. - if (F.getName() == "__local_stdio_printf_options" || - F.getName() == "__local_stdio_scanf_options") - return; - if (isa(F.getEntryBlock().getTerminator())) return; - // Don't instrument functions using SEH for now. Splitting basic blocks like - // we do for coverage breaks WinEHPrepare. - // FIXME: Remove this when SEH no longer uses landingpad pattern matching. - if (F.hasPersonalityFn() && - isAsynchronousEHPersonality(classifyEHPersonality(F.getPersonalityFn()))) - return; - // if (Allowlist && !Allowlist->inSection("coverage", "fun", F.getName())) - // return; - // if (Blocklist && Blocklist->inSection("coverage", "fun", F.getName())) - // return; - - // afl++ START - if (!F.size()) return; - if (isIgnoreFunction(&F)) return; - // afl++ END - - if (Options.CoverageType >= SanitizerCoverageOptions::SCK_Edge) - SplitAllCriticalEdges( - F, CriticalEdgeSplittingOptions().setIgnoreUnreachableDests()); - SmallVector IndirCalls; - SmallVector BlocksToInstrument; - - const DominatorTree * DT = DTCallback(F); - const PostDominatorTree *PDT = PDTCallback(F); - bool IsLeafFunc = true; - - for (auto &BB : F) { - - if (shouldInstrumentBlock(F, &BB, DT, PDT, Options)) - BlocksToInstrument.push_back(&BB); - for (auto &Inst : BB) { - - if (Options.IndirectCalls) { - - CallBase *CB = dyn_cast(&Inst); - if (CB && !CB->getCalledFunction()) IndirCalls.push_back(&Inst); - - } - - } - - } - - InjectCoverage(F, BlocksToInstrument, IsLeafFunc); - InjectCoverageForIndirectCalls(F, IndirCalls); - -} - -GlobalVariable *ModuleSanitizerCoverage::CreateFunctionLocalArrayInSection( - size_t NumElements, Function &F, Type *Ty, const char *Section) { - - ArrayType *ArrayTy = ArrayType::get(Ty, NumElements); - auto Array = new GlobalVariable( - *CurModule, ArrayTy, false, GlobalVariable::PrivateLinkage, - Constant::getNullValue(ArrayTy), "__sancov_gen_"); - - if (TargetTriple.supportsCOMDAT() && !F.isInterposable()) - if (auto Comdat = - GetOrCreateFunctionComdat(F, TargetTriple, CurModuleUniqueId)) - Array->setComdat(Comdat); - Array->setSection(getSectionName(Section)); - Array->setAlignment(Align(DL->getTypeStoreSize(Ty).getFixedSize())); - GlobalsToAppendToUsed.push_back(Array); - GlobalsToAppendToCompilerUsed.push_back(Array); - MDNode *MD = MDNode::get(F.getContext(), ValueAsMetadata::get(&F)); - Array->addMetadata(LLVMContext::MD_associated, *MD); - - return Array; - -} - -GlobalVariable *ModuleSanitizerCoverage::CreatePCArray( - Function &F, ArrayRef AllBlocks) { - - size_t N = AllBlocks.size(); - assert(N); - SmallVector PCs; - IRBuilder<> IRB(&*F.getEntryBlock().getFirstInsertionPt()); - for (size_t i = 0; i < N; i++) { - - if (&F.getEntryBlock() == AllBlocks[i]) { - - PCs.push_back((Constant *)IRB.CreatePointerCast(&F, IntptrPtrTy)); - PCs.push_back((Constant *)IRB.CreateIntToPtr( - ConstantInt::get(IntptrTy, 1), IntptrPtrTy)); - - } else { - - PCs.push_back((Constant *)IRB.CreatePointerCast( - BlockAddress::get(AllBlocks[i]), IntptrPtrTy)); - PCs.push_back((Constant *)IRB.CreateIntToPtr( - ConstantInt::get(IntptrTy, 0), IntptrPtrTy)); - - } - - } - - auto *PCArray = CreateFunctionLocalArrayInSection(N * 2, F, IntptrPtrTy, - SanCovPCsSectionName); - PCArray->setInitializer( - ConstantArray::get(ArrayType::get(IntptrPtrTy, N * 2), PCs)); - PCArray->setConstant(true); - - return PCArray; - -} - -void ModuleSanitizerCoverage::CreateFunctionLocalArrays( - Function &F, ArrayRef AllBlocks) { - - if (Options.TracePCGuard) - FunctionGuardArray = CreateFunctionLocalArrayInSection( - AllBlocks.size(), F, Int32Ty, SanCovGuardsSectionName); - if (Options.Inline8bitCounters) - Function8bitCounterArray = CreateFunctionLocalArrayInSection( - AllBlocks.size(), F, Int8Ty, SanCovCountersSectionName); - if (Options.InlineBoolFlag) - FunctionBoolArray = CreateFunctionLocalArrayInSection( - AllBlocks.size(), F, Int1Ty, SanCovBoolFlagSectionName); - if (Options.PCTable) FunctionPCsArray = CreatePCArray(F, AllBlocks); - -} - -bool ModuleSanitizerCoverage::InjectCoverage(Function & F, - ArrayRef AllBlocks, - bool IsLeafFunc) { - - if (AllBlocks.empty()) return false; - CreateFunctionLocalArrays(F, AllBlocks); - for (size_t i = 0, N = AllBlocks.size(); i < N; i++) { - - // afl++ START - if (BlockList.size()) { - - int skip = 0; - for (uint32_t k = 0; k < BlockList.size(); k++) { - - if (AllBlocks[i] == BlockList[k]) { - - if (debug) - fprintf(stderr, - "DEBUG: Function %s skipping BB with/after __afl_loop\n", - F.getName().str().c_str()); - skip = 1; - - } - - } - - if (skip) continue; - - } - - // afl++ END - - InjectCoverageAtBlock(F, *AllBlocks[i], i, IsLeafFunc); - - } - - return true; - -} - -// On every indirect call we call a run-time function -// __sanitizer_cov_indir_call* with two parameters: -// - callee address, -// - global cache array that contains CacheSize pointers (zero-initialized). -// The cache is used to speed up recording the caller-callee pairs. -// The address of the caller is passed implicitly via caller PC. -// CacheSize is encoded in the name of the run-time function. -void ModuleSanitizerCoverage::InjectCoverageForIndirectCalls( - Function &F, ArrayRef IndirCalls) { - - if (IndirCalls.empty()) return; - assert(Options.TracePC || Options.TracePCGuard || - Options.Inline8bitCounters || Options.InlineBoolFlag); - for (auto I : IndirCalls) { - - IRBuilder<> IRB(I); - CallBase & CB = cast(*I); - Value * Callee = CB.getCalledOperand(); - if (isa(Callee)) continue; - IRB.CreateCall(SanCovTracePCIndir, IRB.CreatePointerCast(Callee, IntptrTy)); - - } - -} - -void ModuleSanitizerCoverage::InjectCoverageAtBlock(Function &F, BasicBlock &BB, - size_t Idx, - bool IsLeafFunc) { - - BasicBlock::iterator IP = BB.getFirstInsertionPt(); - bool IsEntryBB = &BB == &F.getEntryBlock(); - DebugLoc EntryLoc; - if (IsEntryBB) { - - if (auto SP = F.getSubprogram()) - EntryLoc = DebugLoc::get(SP->getScopeLine(), 0, SP); - // Keep static allocas and llvm.localescape calls in the entry block. Even - // if we aren't splitting the block, it's nice for allocas to be before - // calls. - IP = PrepareToSplitEntryBlock(BB, IP); - - } else { - - EntryLoc = IP->getDebugLoc(); - - } - - IRBuilder<> IRB(&*IP); - IRB.SetCurrentDebugLocation(EntryLoc); - if (Options.TracePC) { - - IRB.CreateCall(SanCovTracePC) -#if LLVM_VERSION_MAJOR < 12 - ->cannotMerge(); // gets the PC using GET_CALLER_PC. -#else - ->setCannotMerge(); // gets the PC using GET_CALLER_PC. -#endif - - } - - if (Options.TracePCGuard) { - - // afl++ START - ++afl_global_id; - - if (documentFile) { - - unsigned long long int moduleID = - (((unsigned long long int)(rand() & 0xffffffff)) << 32) | getpid(); - fprintf(documentFile, "ModuleID=%llu Function=%s edgeID=%u\n", moduleID, - F.getName().str().c_str(), afl_global_id); - - } - - /* Set the ID of the inserted basic block */ - - ConstantInt *CurLoc = ConstantInt::get(Int32Tyi, afl_global_id); - - /* Load SHM pointer */ - - Value *MapPtrIdx; - - if (map_addr) { - - MapPtrIdx = IRB.CreateGEP(MapPtrFixed, CurLoc); - - } else { - - LoadInst *MapPtr = IRB.CreateLoad(AFLMapPtr); - MapPtr->setMetadata(Mo->getMDKindID("nosanitize"), - MDNode::get(*Ct, None)); - MapPtrIdx = IRB.CreateGEP(MapPtr, CurLoc); - - } - - /* Update bitmap */ - - LoadInst *Counter = IRB.CreateLoad(MapPtrIdx); - Counter->setMetadata(Mo->getMDKindID("nosanitize"), MDNode::get(*Ct, None)); - - Value *Incr = IRB.CreateAdd(Counter, One); - - if (skip_nozero == NULL) { - - auto cf = IRB.CreateICmpEQ(Incr, Zero); - auto carry = IRB.CreateZExt(cf, Int8Tyi); - Incr = IRB.CreateAdd(Incr, carry); - - } - - IRB.CreateStore(Incr, MapPtrIdx) - ->setMetadata(Mo->getMDKindID("nosanitize"), MDNode::get(*Ct, None)); - - // done :) - - inst++; - // afl++ END - - /* - XXXXXXXXXXXXXXXXXXX - - auto GuardPtr = IRB.CreateIntToPtr( - IRB.CreateAdd(IRB.CreatePointerCast(FunctionGuardArray, IntptrTy), - ConstantInt::get(IntptrTy, Idx * 4)), - Int32PtrTy); - - IRB.CreateCall(SanCovTracePCGuard, GuardPtr)->setCannotMerge(); - */ - - } - - if (Options.Inline8bitCounters) { - - auto CounterPtr = IRB.CreateGEP( - Function8bitCounterArray->getValueType(), Function8bitCounterArray, - {ConstantInt::get(IntptrTy, 0), ConstantInt::get(IntptrTy, Idx)}); - auto Load = IRB.CreateLoad(Int8Ty, CounterPtr); - auto Inc = IRB.CreateAdd(Load, ConstantInt::get(Int8Ty, 1)); - auto Store = IRB.CreateStore(Inc, CounterPtr); - SetNoSanitizeMetadata(Load); - SetNoSanitizeMetadata(Store); - - } - - if (Options.InlineBoolFlag) { - - auto FlagPtr = IRB.CreateGEP( - FunctionBoolArray->getValueType(), FunctionBoolArray, - {ConstantInt::get(IntptrTy, 0), ConstantInt::get(IntptrTy, Idx)}); - auto Load = IRB.CreateLoad(Int1Ty, FlagPtr); - auto ThenTerm = - SplitBlockAndInsertIfThen(IRB.CreateIsNull(Load), &*IP, false); - IRBuilder<> ThenIRB(ThenTerm); - auto Store = ThenIRB.CreateStore(ConstantInt::getTrue(Int1Ty), FlagPtr); - SetNoSanitizeMetadata(Load); - SetNoSanitizeMetadata(Store); - - } - -} - -std::string ModuleSanitizerCoverage::getSectionName( - const std::string &Section) const { - - if (TargetTriple.isOSBinFormatCOFF()) { - - if (Section == SanCovCountersSectionName) return ".SCOV$CM"; - if (Section == SanCovBoolFlagSectionName) return ".SCOV$BM"; - if (Section == SanCovPCsSectionName) return ".SCOVP$M"; - return ".SCOV$GM"; // For SanCovGuardsSectionName. - - } - - if (TargetTriple.isOSBinFormatMachO()) return "__DATA,__" + Section; - return "__" + Section; - -} - -std::string ModuleSanitizerCoverage::getSectionStart( - const std::string &Section) const { - - if (TargetTriple.isOSBinFormatMachO()) - return "\1section$start$__DATA$__" + Section; - return "__start___" + Section; - -} - -std::string ModuleSanitizerCoverage::getSectionEnd( - const std::string &Section) const { - - if (TargetTriple.isOSBinFormatMachO()) - return "\1section$end$__DATA$__" + Section; - return "__stop___" + Section; - -} - -char ModuleSanitizerCoverageLegacyPass::ID = 0; - -INITIALIZE_PASS_BEGIN(ModuleSanitizerCoverageLegacyPass, "sancov", - "Pass for instrumenting coverage on functions", false, - false) -INITIALIZE_PASS_DEPENDENCY(DominatorTreeWrapperPass) -INITIALIZE_PASS_DEPENDENCY(PostDominatorTreeWrapperPass) -INITIALIZE_PASS_END(ModuleSanitizerCoverageLegacyPass, "sancov", - "Pass for instrumenting coverage on functions", false, - false) - -ModulePass *llvm::createModuleSanitizerCoverageLegacyPassPass( - const SanitizerCoverageOptions &Options, - const std::vector &AllowlistFiles, - const std::vector &BlocklistFiles) { - - return new ModuleSanitizerCoverageLegacyPass(Options); - //, AllowlistFiles, BlocklistFiles); - -} - -static void registerLTOPass(const PassManagerBuilder &, - legacy::PassManagerBase &PM) { - - auto p = new ModuleSanitizerCoverageLegacyPass(); - PM.add(p); - -} - -static RegisterStandardPasses RegisterCompTransPass( - PassManagerBuilder::EP_OptimizerLast, registerLTOPass); - -static RegisterStandardPasses RegisterCompTransPass0( - PassManagerBuilder::EP_EnabledOnOptLevel0, registerLTOPass); - -#if LLVM_VERSION_MAJOR >= 11 -static RegisterStandardPasses RegisterCompTransPassLTO( - PassManagerBuilder::EP_FullLinkTimeOptimizationLast, registerLTOPass); -#endif - diff --git a/llvm_mode/afl-clang-fast.c b/llvm_mode/afl-clang-fast.c deleted file mode 100644 index ccdbca9d..00000000 --- a/llvm_mode/afl-clang-fast.c +++ /dev/null @@ -1,1143 +0,0 @@ -/* - american fuzzy lop++ - LLVM-mode wrapper for clang - ------------------------------------------------ - - Written by Laszlo Szekeres and - Michal Zalewski - - LLVM integration design comes from Laszlo Szekeres. - - Copyright 2015, 2016 Google Inc. All rights reserved. - Copyright 2019-2020 AFLplusplus Project. All rights reserved. - - Licensed under the Apache License, Version 2.0 (the "License"); - you may not use this file except in compliance with the License. - You may obtain a copy of the License at: - - http://www.apache.org/licenses/LICENSE-2.0 - - This program is a drop-in replacement for clang, similar in most respects - to ../afl-gcc. It tries to figure out compilation mode, adds a bunch - of flags, and then calls the real compiler. - - */ - -#define AFL_MAIN - -#include "common.h" -#include "config.h" -#include "types.h" -#include "debug.h" -#include "alloc-inl.h" -#include "llvm-ngram-coverage.h" - -#include -#include -#include -#include -#include -#include -#include - -#include "llvm/Config/llvm-config.h" - -static u8 * obj_path; /* Path to runtime libraries */ -static u8 **cc_params; /* Parameters passed to the real CC */ -static u32 cc_par_cnt = 1; /* Param count, including argv0 */ -static u8 llvm_fullpath[PATH_MAX]; -static u8 instrument_mode, instrument_opt_mode, ngram_size, lto_mode, cpp_mode; -static u8 *lto_flag = AFL_CLANG_FLTO; -static u8 debug; -static u8 cwd[4096]; -static u8 cmplog_mode; -u8 use_stdin = 0; /* dummy */ -// static u8 *march_opt = CFLAGS_OPT; - -enum { - - INSTURMENT_DEFAULT = 0, - INSTRUMENT_CLASSIC = 1, - INSTRUMENT_AFL = 1, - INSTRUMENT_PCGUARD = 2, - INSTRUMENT_INSTRIM = 3, - INSTRUMENT_CFG = 3, - INSTRUMENT_LTO = 4, - INSTRUMENT_OPT_CTX = 8, - INSTRUMENT_OPT_NGRAM = 16 - -}; - -char instrument_mode_string[18][18] = { - - "DEFAULT", "CLASSIC", "PCGUARD", "CFG", "LTO", "", "", "", "CTX", "", - "", "", "", "", "", "", "NGRAM", "" - -}; - -u8 *getthecwd() { - - static u8 fail[] = ""; - if (getcwd(cwd, sizeof(cwd)) == NULL) return fail; - return cwd; - -} - -/* Try to find the runtime libraries. If that fails, abort. */ - -static void find_obj(u8 *argv0) { - - u8 *afl_path = getenv("AFL_PATH"); - u8 *slash, *tmp; - - if (afl_path) { - -#ifdef __ANDROID__ - tmp = alloc_printf("%s/afl-llvm-rt.so", afl_path); -#else - tmp = alloc_printf("%s/afl-llvm-rt.o", afl_path); -#endif - - if (!access(tmp, R_OK)) { - - obj_path = afl_path; - ck_free(tmp); - return; - - } - - ck_free(tmp); - - } - - slash = strrchr(argv0, '/'); - - if (slash) { - - u8 *dir; - - *slash = 0; - dir = ck_strdup(argv0); - *slash = '/'; - -#ifdef __ANDROID__ - tmp = alloc_printf("%s/afl-llvm-rt.so", dir); -#else - tmp = alloc_printf("%s/afl-llvm-rt.o", dir); -#endif - - if (!access(tmp, R_OK)) { - - obj_path = dir; - ck_free(tmp); - return; - - } - - ck_free(tmp); - ck_free(dir); - - } - -#ifdef __ANDROID__ - if (!access(AFL_PATH "/afl-llvm-rt.so", R_OK)) { - -#else - if (!access(AFL_PATH "/afl-llvm-rt.o", R_OK)) { - -#endif - - obj_path = AFL_PATH; - return; - - } - - FATAL( - "Unable to find 'afl-llvm-rt.o' or 'afl-llvm-pass.so'. Please set " - "AFL_PATH"); - -} - -/* Copy argv to cc_params, making the necessary edits. */ - -static void edit_params(u32 argc, char **argv, char **envp) { - - u8 fortify_set = 0, asan_set = 0, x_set = 0, bit_mode = 0, shared_linking = 0, - preprocessor_only = 0; - u8 have_pic = 0; - u8 *name; - - cc_params = ck_alloc((argc + 128) * sizeof(u8 *)); - - name = strrchr(argv[0], '/'); - if (!name) - name = argv[0]; - else - ++name; - - if (lto_mode) - if (lto_flag[0] != '-') - FATAL( - "Using afl-clang-lto is not possible because Makefile magic did not " - "identify the correct -flto flag"); - - if (!strcmp(name, "afl-clang-fast++") || !strcmp(name, "afl-clang-lto++") || - !strcmp(name, "afl-clang++")) { - - u8 *alt_cxx = getenv("AFL_CXX"); - if (USE_BINDIR) - snprintf(llvm_fullpath, sizeof(llvm_fullpath), "%s/clang++", LLVM_BINDIR); - else - sprintf(llvm_fullpath, CLANGPP_BIN); - cc_params[0] = alt_cxx && *alt_cxx ? alt_cxx : (u8 *)llvm_fullpath; - cpp_mode = 1; - - } else if (!strcmp(name, "afl-clang-fast") || - - !strcmp(name, "afl-clang-lto") || !strcmp(name, "afl-clang")) { - - u8 *alt_cc = getenv("AFL_CC"); - if (USE_BINDIR) - snprintf(llvm_fullpath, sizeof(llvm_fullpath), "%s/clang", LLVM_BINDIR); - else - sprintf(llvm_fullpath, CLANG_BIN); - cc_params[0] = alt_cc && *alt_cc ? alt_cc : (u8 *)llvm_fullpath; - - } else { - - fprintf(stderr, "Name of the binary: %s\n", argv[0]); - FATAL( - "Name of the binary is not a known name, expected afl-clang-fast(++) " - "or afl-clang-lto(++)"); - - } - - cc_params[cc_par_cnt++] = "-Wno-unused-command-line-argument"; - - if (lto_mode && cpp_mode) - cc_params[cc_par_cnt++] = "-lc++"; // needed by fuzzbench, early - - /* There are several ways to compile with afl-clang-fast. In the traditional - mode, we use afl-llvm-pass.so, then there is libLLVMInsTrim.so which is - faster and creates less map pollution. - Then there is the 'trace-pc-guard' mode, we use native LLVM - instrumentation callbacks instead. For trace-pc-guard see: - http://clang.llvm.org/docs/SanitizerCoverage.html#tracing-pcs-with-guards - The best instrumentatation is with the LTO modes, the classic and - InsTrimLTO, the latter is faster. The LTO modes are activated by using - afl-clang-lto(++) - */ - - if (lto_mode) { - - if (getenv("AFL_LLVM_INSTRUMENT_FILE") != NULL || - getenv("AFL_LLVM_WHITELIST") || getenv("AFL_LLVM_ALLOWLIST") || - getenv("AFL_LLVM_DENYLIST") || getenv("AFL_LLVM_BLOCKLIST")) { - - cc_params[cc_par_cnt++] = "-Xclang"; - cc_params[cc_par_cnt++] = "-load"; - cc_params[cc_par_cnt++] = "-Xclang"; - cc_params[cc_par_cnt++] = - alloc_printf("%s/afl-llvm-lto-instrumentlist.so", obj_path); - - } - - } - - // laf - if (getenv("LAF_SPLIT_SWITCHES") || getenv("AFL_LLVM_LAF_SPLIT_SWITCHES")) { - - if (lto_mode) { - - cc_params[cc_par_cnt++] = - alloc_printf("-Wl,-mllvm=-load=%s/split-switches-pass.so", obj_path); - - } else { - - cc_params[cc_par_cnt++] = "-Xclang"; - cc_params[cc_par_cnt++] = "-load"; - cc_params[cc_par_cnt++] = "-Xclang"; - cc_params[cc_par_cnt++] = - alloc_printf("%s/split-switches-pass.so", obj_path); - - } - - } - - if (getenv("LAF_TRANSFORM_COMPARES") || - getenv("AFL_LLVM_LAF_TRANSFORM_COMPARES")) { - - if (lto_mode) { - - cc_params[cc_par_cnt++] = alloc_printf( - "-Wl,-mllvm=-load=%s/compare-transform-pass.so", obj_path); - - } else { - - cc_params[cc_par_cnt++] = "-Xclang"; - cc_params[cc_par_cnt++] = "-load"; - cc_params[cc_par_cnt++] = "-Xclang"; - cc_params[cc_par_cnt++] = - alloc_printf("%s/compare-transform-pass.so", obj_path); - - } - - } - - if (getenv("LAF_SPLIT_COMPARES") || getenv("AFL_LLVM_LAF_SPLIT_COMPARES") || - getenv("AFL_LLVM_LAF_SPLIT_FLOATS")) { - - if (lto_mode) { - - cc_params[cc_par_cnt++] = - alloc_printf("-Wl,-mllvm=-load=%s/split-compares-pass.so", obj_path); - - } else { - - cc_params[cc_par_cnt++] = "-Xclang"; - cc_params[cc_par_cnt++] = "-load"; - cc_params[cc_par_cnt++] = "-Xclang"; - cc_params[cc_par_cnt++] = - alloc_printf("%s/split-compares-pass.so", obj_path); - - } - - } - - // /laf - - unsetenv("AFL_LD"); - unsetenv("AFL_LD_CALLER"); - if (cmplog_mode) { - - if (lto_mode) { - - cc_params[cc_par_cnt++] = - alloc_printf("-Wl,-mllvm=-load=%s/cmplog-routines-pass.so", obj_path); - cc_params[cc_par_cnt++] = - alloc_printf("-Wl,-mllvm=-load=%s/split-switches-pass.so", obj_path); - cc_params[cc_par_cnt++] = alloc_printf( - "-Wl,-mllvm=-load=%s/cmplog-instructions-pass.so", obj_path); - - } else { - - cc_params[cc_par_cnt++] = "-Xclang"; - cc_params[cc_par_cnt++] = "-load"; - cc_params[cc_par_cnt++] = "-Xclang"; - cc_params[cc_par_cnt++] = - alloc_printf("%s/cmplog-routines-pass.so", obj_path); - - // reuse split switches from laf - cc_params[cc_par_cnt++] = "-Xclang"; - cc_params[cc_par_cnt++] = "-load"; - cc_params[cc_par_cnt++] = "-Xclang"; - cc_params[cc_par_cnt++] = - alloc_printf("%s/split-switches-pass.so", obj_path); - - cc_params[cc_par_cnt++] = "-Xclang"; - cc_params[cc_par_cnt++] = "-load"; - cc_params[cc_par_cnt++] = "-Xclang"; - cc_params[cc_par_cnt++] = - alloc_printf("%s/cmplog-instructions-pass.so", obj_path); - - } - - cc_params[cc_par_cnt++] = "-fno-inline"; - - } - - if (lto_mode) { - -#if defined(AFL_CLANG_LDPATH) && LLVM_VERSION_MAJOR >= 12 - u8 *ld_ptr = strrchr(AFL_REAL_LD, '/'); - if (!ld_ptr) ld_ptr = "ld.lld"; - cc_params[cc_par_cnt++] = alloc_printf("-fuse-ld=%s", ld_ptr); - cc_params[cc_par_cnt++] = alloc_printf("--ld-path=%s", AFL_REAL_LD); -#else - cc_params[cc_par_cnt++] = alloc_printf("-fuse-ld=%s", AFL_REAL_LD); -#endif - - cc_params[cc_par_cnt++] = "-Wl,--allow-multiple-definition"; - - if (instrument_mode == INSTRUMENT_CFG) - cc_params[cc_par_cnt++] = - alloc_printf("-Wl,-mllvm=-load=%s/SanitizerCoverageLTO.so", obj_path); - else - - cc_params[cc_par_cnt++] = alloc_printf( - "-Wl,-mllvm=-load=%s/afl-llvm-lto-instrumentation.so", obj_path); - cc_params[cc_par_cnt++] = lto_flag; - - } else { - - if (instrument_mode == INSTRUMENT_PCGUARD) { - -#if LLVM_VERSION_MAJOR > 4 || \ - (LLVM_VERSION_MAJOR == 4 && \ - (LLVM_VERSION_MINOR > 0 || LLVM_VERSION_PATCH >= 1)) - cc_params[cc_par_cnt++] = - "-fsanitize-coverage=trace-pc-guard"; // edge coverage by default -#else - FATAL("pcguard instrumentation requires llvm 4.0.1+"); -#endif - - } else { - - cc_params[cc_par_cnt++] = "-Xclang"; - cc_params[cc_par_cnt++] = "-load"; - cc_params[cc_par_cnt++] = "-Xclang"; - if (instrument_mode == INSTRUMENT_CFG) - cc_params[cc_par_cnt++] = - alloc_printf("%s/libLLVMInsTrim.so", obj_path); - else - cc_params[cc_par_cnt++] = alloc_printf("%s/afl-llvm-pass.so", obj_path); - - } - - } - - // cc_params[cc_par_cnt++] = "-Qunused-arguments"; - - // in case LLVM is installed not via a package manager or "make install" - // e.g. compiled download or compiled from github then it's ./lib directory - // might not be in the search path. Add it if so. - u8 *libdir = strdup(LLVM_LIBDIR); - if (cpp_mode && strlen(libdir) && strncmp(libdir, "/usr", 4) && - strncmp(libdir, "/lib", 4)) { - - cc_params[cc_par_cnt++] = "-rpath"; - cc_params[cc_par_cnt++] = libdir; - - } else { - - free(libdir); - - } - - u32 idx; - if (lto_mode && argc > 1) { - - for (idx = 1; idx < argc; idx++) { - - if (!strncasecmp(argv[idx], "-fpic", 5)) have_pic = 1; - - } - - if (!have_pic) cc_params[cc_par_cnt++] = "-fPIC"; - - } - - /* Detect stray -v calls from ./configure scripts. */ - - while (--argc) { - - u8 *cur = *(++argv); - - if (!strcmp(cur, "-m32")) bit_mode = 32; - if (!strcmp(cur, "armv7a-linux-androideabi")) bit_mode = 32; - if (!strcmp(cur, "-m64")) bit_mode = 64; - - if (!strcmp(cur, "-x")) x_set = 1; - - if (!strcmp(cur, "-fsanitize=address") || !strcmp(cur, "-fsanitize=memory")) - asan_set = 1; - - if (strstr(cur, "FORTIFY_SOURCE")) fortify_set = 1; - - if (!strcmp(cur, "-Wl,-z,defs") || !strcmp(cur, "-Wl,--no-undefined")) - continue; - - if (lto_mode && !strncmp(cur, "-fuse-ld=", 9)) continue; - if (lto_mode && !strncmp(cur, "--ld-path=", 10)) continue; - - if (!strcmp(cur, "-E")) preprocessor_only = 1; - if (!strcmp(cur, "-shared")) shared_linking = 1; - - cc_params[cc_par_cnt++] = cur; - - } - - if (getenv("AFL_HARDEN")) { - - cc_params[cc_par_cnt++] = "-fstack-protector-all"; - - if (!fortify_set) cc_params[cc_par_cnt++] = "-D_FORTIFY_SOURCE=2"; - - } - - if (!asan_set) { - - if (getenv("AFL_USE_ASAN")) { - - if (getenv("AFL_USE_MSAN")) FATAL("ASAN and MSAN are mutually exclusive"); - - if (getenv("AFL_HARDEN")) - FATAL("ASAN and AFL_HARDEN are mutually exclusive"); - - cc_params[cc_par_cnt++] = "-U_FORTIFY_SOURCE"; - cc_params[cc_par_cnt++] = "-fsanitize=address"; - - } else if (getenv("AFL_USE_MSAN")) { - - if (getenv("AFL_USE_ASAN")) FATAL("ASAN and MSAN are mutually exclusive"); - - if (getenv("AFL_HARDEN")) - FATAL("MSAN and AFL_HARDEN are mutually exclusive"); - - cc_params[cc_par_cnt++] = "-U_FORTIFY_SOURCE"; - cc_params[cc_par_cnt++] = "-fsanitize=memory"; - - } - - } - - if (getenv("AFL_USE_UBSAN")) { - - cc_params[cc_par_cnt++] = "-fsanitize=undefined"; - cc_params[cc_par_cnt++] = "-fsanitize-undefined-trap-on-error"; - cc_params[cc_par_cnt++] = "-fno-sanitize-recover=all"; - - } - - if (getenv("AFL_USE_CFISAN")) { - - if (!lto_mode) { - - uint32_t i = 0, found = 0; - while (envp[i] != NULL && !found) - if (strncmp("-flto", envp[i++], 5) == 0) found = 1; - if (!found) cc_params[cc_par_cnt++] = "-flto"; - - } - - cc_params[cc_par_cnt++] = "-fsanitize=cfi"; - cc_params[cc_par_cnt++] = "-fvisibility=hidden"; - - } - - if (!getenv("AFL_DONT_OPTIMIZE")) { - - cc_params[cc_par_cnt++] = "-g"; - cc_params[cc_par_cnt++] = "-O3"; - cc_params[cc_par_cnt++] = "-funroll-loops"; - // if (strlen(march_opt) > 1 && march_opt[0] == '-') - // cc_params[cc_par_cnt++] = march_opt; - - } - - if (getenv("AFL_NO_BUILTIN") || getenv("AFL_LLVM_LAF_TRANSFORM_COMPARES") || - getenv("LAF_TRANSFORM_COMPARES") || lto_mode) { - - cc_params[cc_par_cnt++] = "-fno-builtin-strcmp"; - cc_params[cc_par_cnt++] = "-fno-builtin-strncmp"; - cc_params[cc_par_cnt++] = "-fno-builtin-strcasecmp"; - cc_params[cc_par_cnt++] = "-fno-builtin-strncasecmp"; - cc_params[cc_par_cnt++] = "-fno-builtin-memcmp"; - cc_params[cc_par_cnt++] = "-fno-builtin-bcmp"; - cc_params[cc_par_cnt++] = "-fno-builtin-strstr"; - cc_params[cc_par_cnt++] = "-fno-builtin-strcasestr"; - - } - -#if defined(USEMMAP) && !defined(__HAIKU__) - cc_params[cc_par_cnt++] = "-lrt"; -#endif - - cc_params[cc_par_cnt++] = "-D__AFL_HAVE_MANUAL_CONTROL=1"; - cc_params[cc_par_cnt++] = "-D__AFL_COMPILER=1"; - cc_params[cc_par_cnt++] = "-DFUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION=1"; - - /* When the user tries to use persistent or deferred forkserver modes by - appending a single line to the program, we want to reliably inject a - signature into the binary (to be picked up by afl-fuzz) and we want - to call a function from the runtime .o file. This is unnecessarily - painful for three reasons: - - 1) We need to convince the compiler not to optimize out the signature. - This is done with __attribute__((used)). - - 2) We need to convince the linker, when called with -Wl,--gc-sections, - not to do the same. This is done by forcing an assignment to a - 'volatile' pointer. - - 3) We need to declare __afl_persistent_loop() in the global namespace, - but doing this within a method in a class is hard - :: and extern "C" - are forbidden and __attribute__((alias(...))) doesn't work. Hence the - __asm__ aliasing trick. - - */ - - cc_params[cc_par_cnt++] = - "-D__AFL_FUZZ_INIT()=" - "int __afl_sharedmem_fuzzing = 1;" - "extern unsigned int *__afl_fuzz_len;" - "extern unsigned char *__afl_fuzz_ptr;" - "unsigned char __afl_fuzz_alt[1024000];" - "unsigned char *__afl_fuzz_alt_ptr = __afl_fuzz_alt;"; - cc_params[cc_par_cnt++] = - "-D__AFL_FUZZ_TESTCASE_BUF=(__afl_fuzz_ptr ? __afl_fuzz_ptr : " - "__afl_fuzz_alt_ptr)"; - cc_params[cc_par_cnt++] = - "-D__AFL_FUZZ_TESTCASE_LEN=(__afl_fuzz_ptr ? *__afl_fuzz_len : " - "(*__afl_fuzz_len = read(0, __afl_fuzz_alt_ptr, 1024000)) == 0xffffffff " - "? 0 : *__afl_fuzz_len)"; - - cc_params[cc_par_cnt++] = - "-D__AFL_LOOP(_A)=" - "({ static volatile char *_B __attribute__((used)); " - " _B = (char*)\"" PERSIST_SIG - "\"; " -#ifdef __APPLE__ - "__attribute__((visibility(\"default\"))) " - "int _L(unsigned int) __asm__(\"___afl_persistent_loop\"); " -#else - "__attribute__((visibility(\"default\"))) " - "int _L(unsigned int) __asm__(\"__afl_persistent_loop\"); " -#endif /* ^__APPLE__ */ - "_L(_A); })"; - - cc_params[cc_par_cnt++] = - "-D__AFL_INIT()=" - "do { static volatile char *_A __attribute__((used)); " - " _A = (char*)\"" DEFER_SIG - "\"; " -#ifdef __APPLE__ - "__attribute__((visibility(\"default\"))) " - "void _I(void) __asm__(\"___afl_manual_init\"); " -#else - "__attribute__((visibility(\"default\"))) " - "void _I(void) __asm__(\"__afl_manual_init\"); " -#endif /* ^__APPLE__ */ - "_I(); } while (0)"; - - if (x_set) { - - cc_params[cc_par_cnt++] = "-x"; - cc_params[cc_par_cnt++] = "none"; - - } - - if (preprocessor_only) { - - /* In the preprocessor_only case (-E), we are not actually compiling at - all but requesting the compiler to output preprocessed sources only. - We must not add the runtime in this case because the compiler will - simply output its binary content back on stdout, breaking any build - systems that rely on a separate source preprocessing step. */ - cc_params[cc_par_cnt] = NULL; - return; - - } - -#ifndef __ANDROID__ - switch (bit_mode) { - - case 0: - cc_params[cc_par_cnt++] = alloc_printf("%s/afl-llvm-rt.o", obj_path); - if (lto_mode) - cc_params[cc_par_cnt++] = - alloc_printf("%s/afl-llvm-rt-lto.o", obj_path); - break; - - case 32: - cc_params[cc_par_cnt++] = alloc_printf("%s/afl-llvm-rt-32.o", obj_path); - if (access(cc_params[cc_par_cnt - 1], R_OK)) - FATAL("-m32 is not supported by your compiler"); - if (lto_mode) { - - cc_params[cc_par_cnt++] = - alloc_printf("%s/afl-llvm-rt-lto-32.o", obj_path); - if (access(cc_params[cc_par_cnt - 1], R_OK)) - FATAL("-m32 is not supported by your compiler"); - - } - - break; - - case 64: - cc_params[cc_par_cnt++] = alloc_printf("%s/afl-llvm-rt-64.o", obj_path); - if (access(cc_params[cc_par_cnt - 1], R_OK)) - FATAL("-m64 is not supported by your compiler"); - if (lto_mode) { - - cc_params[cc_par_cnt++] = - alloc_printf("%s/afl-llvm-rt-lto-64.o", obj_path); - if (access(cc_params[cc_par_cnt - 1], R_OK)) - FATAL("-m64 is not supported by your compiler"); - - } - - break; - - } - - #ifndef __APPLE__ - if (!shared_linking) - cc_params[cc_par_cnt++] = - alloc_printf("-Wl,--dynamic-list=%s/dynamic_list.txt", obj_path); - #endif - -#endif - - cc_params[cc_par_cnt] = NULL; - -} - -/* Main entry point */ - -int main(int argc, char **argv, char **envp) { - - int i; - char *callname = "afl-clang-fast", *ptr = NULL; - - if (getenv("AFL_DEBUG")) { - - debug = 1; - if (strcmp(getenv("AFL_DEBUG"), "0") == 0) unsetenv("AFL_DEBUG"); - - } else if (getenv("AFL_QUIET")) - - be_quiet = 1; - - if (getenv("USE_TRACE_PC") || getenv("AFL_USE_TRACE_PC") || - getenv("AFL_LLVM_USE_TRACE_PC") || getenv("AFL_TRACE_PC")) { - - if (instrument_mode == 0) - instrument_mode = INSTRUMENT_PCGUARD; - else if (instrument_mode != INSTRUMENT_PCGUARD) - FATAL("you can not set AFL_LLVM_INSTRUMENT and AFL_TRACE_PC together"); - - } - - if ((getenv("AFL_LLVM_INSTRUMENT_FILE") != NULL || - getenv("AFL_LLVM_WHITELIST") || getenv("AFL_LLVM_ALLOWLIST") || - getenv("AFL_LLVM_DENYLIST") || getenv("AFL_LLVM_BLOCKLIST")) && - getenv("AFL_DONT_OPTIMIZE")) - WARNF( - "AFL_LLVM_ALLOWLIST/DENYLIST and AFL_DONT_OPTIMIZE cannot be combined " - "for file matching, only function matching!"); - - if (getenv("AFL_LLVM_INSTRIM") || getenv("INSTRIM") || - getenv("INSTRIM_LIB")) { - - if (instrument_mode == 0) - instrument_mode = INSTRUMENT_CFG; - else if (instrument_mode != INSTRUMENT_CFG) - FATAL( - "you can not set AFL_LLVM_INSTRUMENT and AFL_LLVM_INSTRIM together"); - - } - - if (getenv("AFL_LLVM_CTX")) instrument_opt_mode |= INSTRUMENT_OPT_CTX; - - if (getenv("AFL_LLVM_NGRAM_SIZE")) { - - instrument_opt_mode |= INSTRUMENT_OPT_NGRAM; - ngram_size = atoi(getenv("AFL_LLVM_NGRAM_SIZE")); - if (ngram_size < 2 || ngram_size > NGRAM_SIZE_MAX) - FATAL( - "NGRAM instrumentation mode must be between 2 and NGRAM_SIZE_MAX " - "(%u)", - NGRAM_SIZE_MAX); - - } - - if (getenv("AFL_LLVM_INSTRUMENT")) { - - u8 *ptr = strtok(getenv("AFL_LLVM_INSTRUMENT"), ":,;"); - - while (ptr) { - - if (strncasecmp(ptr, "afl", strlen("afl")) == 0 || - strncasecmp(ptr, "classic", strlen("classic")) == 0) { - - if (instrument_mode == INSTRUMENT_LTO) { - - instrument_mode = INSTRUMENT_CLASSIC; - lto_mode = 1; - - } else if (!instrument_mode || instrument_mode == INSTRUMENT_AFL) - - instrument_mode = INSTRUMENT_AFL; - else - FATAL("main instrumentation mode already set with %s", - instrument_mode_string[instrument_mode]); - - } - - if (strncasecmp(ptr, "pc-guard", strlen("pc-guard")) == 0 || - strncasecmp(ptr, "pcguard", strlen("pcguard")) == 0) { - - if (!instrument_mode || instrument_mode == INSTRUMENT_PCGUARD) - instrument_mode = INSTRUMENT_PCGUARD; - else - FATAL("main instrumentation mode already set with %s", - instrument_mode_string[instrument_mode]); - - } - - if (strncasecmp(ptr, "cfg", strlen("cfg")) == 0 || - strncasecmp(ptr, "instrim", strlen("instrim")) == 0) { - - if (instrument_mode == INSTRUMENT_LTO) { - - instrument_mode = INSTRUMENT_CFG; - lto_mode = 1; - - } else if (!instrument_mode || instrument_mode == INSTRUMENT_CFG) - - instrument_mode = INSTRUMENT_CFG; - else - FATAL("main instrumentation mode already set with %s", - instrument_mode_string[instrument_mode]); - - } - - if (strncasecmp(ptr, "lto", strlen("lto")) == 0) { - - lto_mode = 1; - if (!instrument_mode || instrument_mode == INSTRUMENT_LTO) - instrument_mode = INSTRUMENT_LTO; - else if (instrument_mode != INSTRUMENT_CFG) - FATAL("main instrumentation mode already set with %s", - instrument_mode_string[instrument_mode]); - - } - - if (strncasecmp(ptr, "ctx", strlen("ctx")) == 0) { - - instrument_opt_mode |= INSTRUMENT_OPT_CTX; - setenv("AFL_LLVM_CTX", "1", 1); - - } - - if (strncasecmp(ptr, "ngram", strlen("ngram")) == 0) { - - ptr += strlen("ngram"); - while (*ptr && (*ptr < '0' || *ptr > '9')) { - - ptr++; - - } - - if (!*ptr) { - - ptr = getenv("AFL_LLVM_NGRAM_SIZE"); - if (!ptr || !*ptr) { - - FATAL( - "you must set the NGRAM size with (e.g. for value 2) " - "AFL_LLVM_INSTRUMENT=ngram-2"); - - } - - } - - ngram_size = atoi(ptr); - if (ngram_size < 2 || ngram_size > NGRAM_SIZE_MAX) - FATAL( - "NGRAM instrumentation option must be between 2 and " - "NGRAM_SIZE_MAX " - "(%u)", - NGRAM_SIZE_MAX); - instrument_opt_mode |= (INSTRUMENT_OPT_NGRAM); - ptr = alloc_printf("%u", ngram_size); - setenv("AFL_LLVM_NGRAM_SIZE", ptr, 1); - - } - - ptr = strtok(NULL, ":,;"); - - } - - } - - if (strstr(argv[0], "afl-clang-lto") != NULL) { - - if (instrument_mode == 0 || instrument_mode == INSTRUMENT_LTO || - instrument_mode == INSTRUMENT_CFG) { - - lto_mode = 1; - callname = "afl-clang-lto"; - if (!instrument_mode) { - - instrument_mode = INSTRUMENT_CFG; - ptr = instrument_mode_string[instrument_mode]; - - } - - } else if (instrument_mode == INSTRUMENT_LTO || - - instrument_mode == INSTRUMENT_CLASSIC) { - - lto_mode = 1; - callname = "afl-clang-lto"; - - } else { - - if (!be_quiet) - WARNF("afl-clang-lto called with mode %s, using that mode instead", - instrument_mode_string[instrument_mode]); - - } - - } - - if (instrument_mode == 0) { - -#if LLVM_VERSION_MAJOR <= 6 - instrument_mode = INSTRUMENT_AFL; -#else - if (getenv("AFL_LLVM_INSTRUMENT_FILE") != NULL || - getenv("AFL_LLVM_WHITELIST") || getenv("AFL_LLVM_ALLOWLIST") || - getenv("AFL_LLVM_DENYLIST") || getenv("AFL_LLVM_BLOCKLIST")) { - - instrument_mode = INSTRUMENT_AFL; - WARNF( - "switching to classic instrumentation because " - "AFL_LLVM_ALLOWLIST/DENYLIST does not work with PCGUARD. Use " - "-fsanitize-coverage-allowlist=allowlist.txt or " - "-fsanitize-coverage-blocklist=denylist.txt if you want to use " - "PCGUARD. Requires llvm 12+. See https://clang.llvm.org/docs/ " - "SanitizerCoverage.html#partially-disabling-instrumentation"); - - } else - - instrument_mode = INSTRUMENT_PCGUARD; -#endif - - } - - if (instrument_opt_mode && lto_mode) - FATAL( - "CTX and NGRAM can not be used in LTO mode (and would make LTO " - "useless)"); - - if (!instrument_opt_mode) { - - if (lto_mode && instrument_mode == INSTRUMENT_CFG) - ptr = alloc_printf("InsTrimLTO"); - else - ptr = instrument_mode_string[instrument_mode]; - - } else if (instrument_opt_mode == INSTRUMENT_OPT_CTX) - - ptr = alloc_printf("%s + CTX", instrument_mode_string[instrument_mode]); - else if (instrument_opt_mode == INSTRUMENT_OPT_NGRAM) - ptr = alloc_printf("%s + NGRAM-%u", instrument_mode_string[instrument_mode], - ngram_size); - else - ptr = alloc_printf("%s + CTX + NGRAM-%u", - instrument_mode_string[instrument_mode], ngram_size); - -#ifndef AFL_CLANG_FLTO - if (lto_mode) - FATAL( - "instrumentation mode LTO specified but LLVM support not available " - "(requires LLVM 11 or higher)"); -#endif - - if (instrument_opt_mode && instrument_mode != INSTRUMENT_CLASSIC && - instrument_mode != INSTRUMENT_CFG) - FATAL( - "CTX and NGRAM instrumentation options can only be used with CFG " - "(recommended) and CLASSIC instrumentation modes!"); - - if (getenv("AFL_LLVM_SKIP_NEVERZERO") && getenv("AFL_LLVM_NOT_ZERO")) - FATAL( - "AFL_LLVM_NOT_ZERO and AFL_LLVM_SKIP_NEVERZERO can not be set " - "together"); - - if (instrument_mode == INSTRUMENT_PCGUARD && - (getenv("AFL_LLVM_INSTRUMENT_FILE") != NULL || - getenv("AFL_LLVM_WHITELIST") || getenv("AFL_LLVM_ALLOWLIST") || - getenv("AFL_LLVM_DENYLIST") || getenv("AFL_LLVM_BLOCKLIST"))) - FATAL( - "Instrumentation type PCGUARD does not support " - "AFL_LLVM_ALLOWLIST/DENYLIST! Use " - "-fsanitize-coverage-allowlist=allowlist.txt or " - "-fsanitize-coverage-blocklist=denylist.txt instead (requires llvm " - "12+), see " - "https://clang.llvm.org/docs/" - "SanitizerCoverage.html#partially-disabling-instrumentation"); - - if (argc < 2 || strcmp(argv[1], "-h") == 0) { - - if (!lto_mode) - printf("afl-clang-fast" VERSION " by in %s mode\n", - ptr); - else - printf("afl-clang-lto" VERSION - " by Marc \"vanHauser\" Heuse in %s mode\n", - ptr); - - SAYF( - "\n" - "%s[++] [options]\n" - "\n" - "This is a helper application for afl-fuzz. It serves as a drop-in " - "replacement\n" - "for clang, letting you recompile third-party code with the " - "required " - "runtime\n" - "instrumentation. A common use pattern would be one of the " - "following:\n\n" - - " CC=%s/afl-clang-fast ./configure\n" - " CXX=%s/afl-clang-fast++ ./configure\n\n" - - "In contrast to the traditional afl-clang tool, this version is " - "implemented as\n" - "an LLVM pass and tends to offer improved performance with slow " - "programs.\n\n" - - "Environment variables used:\n" - "AFL_CC: path to the C compiler to use\n" - "AFL_CXX: path to the C++ compiler to use\n" - "AFL_DEBUG: enable developer debugging output\n" - "AFL_DONT_OPTIMIZE: disable optimization instead of -O3\n" - "AFL_HARDEN: adds code hardening to catch memory bugs\n" - "AFL_INST_RATIO: percentage of branches to instrument\n" -#if LLVM_VERSION_MAJOR < 9 - "AFL_LLVM_NOT_ZERO: use cycling trace counters that skip zero\n" -#else - "AFL_LLVM_SKIP_NEVERZERO: do not skip zero on trace counters\n" -#endif - "AFL_LLVM_LAF_SPLIT_COMPARES: enable cascaded comparisons\n" - "AFL_LLVM_LAF_SPLIT_COMPARES_BITW: size limit (default 8)\n" - "AFL_LLVM_LAF_SPLIT_SWITCHES: casc. comp. in 'switch'\n" - " to cascaded comparisons\n" - "AFL_LLVM_LAF_SPLIT_FLOATS: transform floating point comp. to " - "cascaded comp.\n" - "AFL_LLVM_LAF_TRANSFORM_COMPARES: transform library comparison " - "function calls\n" - "AFL_LLVM_LAF_ALL: enables all LAF splits/transforms\n" - "AFL_LLVM_INSTRUMENT_ALLOW/AFL_LLVM_INSTRUMENT_DENY: enable instrument" - "allow/deny listing (selective instrumentation)\n" - "AFL_NO_BUILTIN: compile for use with libtokencap.so\n" - "AFL_PATH: path to instrumenting pass and runtime " - "(afl-llvm-rt.*o)\n" - "AFL_LLVM_DOCUMENT_IDS: document edge IDs given to which function (LTO " - "only)\n" - "AFL_QUIET: suppress verbose output\n" - "AFL_USE_ASAN: activate address sanitizer\n" - "AFL_USE_CFISAN: activate control flow sanitizer\n" - "AFL_USE_MSAN: activate memory sanitizer\n" - "AFL_USE_UBSAN: activate undefined behaviour sanitizer\n", - callname, BIN_PATH, BIN_PATH); - - SAYF( - "\nafl-clang-fast specific environment variables:\n" - "AFL_LLVM_CMPLOG: log operands of comparisons (RedQueen mutator)\n" - "AFL_LLVM_INSTRUMENT: set instrumentation mode: AFL, CFG " - "(INSTRIM), PCGUARD [DEFAULT], LTO, CTX, NGRAM-2 ... NGRAM-16\n" - " You can also use the old environment variables instead:\n" - " AFL_LLVM_USE_TRACE_PC: use LLVM trace-pc-guard instrumentation " - "[DEFAULT]\n" - " AFL_LLVM_INSTRIM: use light weight instrumentation InsTrim\n" - " AFL_LLVM_INSTRIM_LOOPHEAD: optimize loop tracing for speed (" - "option to INSTRIM)\n" - " AFL_LLVM_CTX: use context sensitive coverage\n" - " AFL_LLVM_NGRAM_SIZE: use ngram prev_loc count coverage\n"); - -#ifdef AFL_CLANG_FLTO - SAYF( - "\nafl-clang-lto specific environment variables:\n" - "AFL_LLVM_MAP_ADDR: use a fixed coverage map address (speed), e.g. " - "0x10000\n" - "AFL_LLVM_DOCUMENT_IDS: write all edge IDs and the corresponding " - "functions they are in into this file\n" - "AFL_LLVM_LTO_DONTWRITEID: don't write the highest ID used to a " - "global var\n" - "AFL_LLVM_LTO_STARTID: from which ID to start counting from for a " - "bb\n" - "AFL_REAL_LD: use this lld linker instead of the compiled in path\n" - "\nafl-clang-lto was built with linker target \"%s\" and LTO flags " - "\"%s\"\n" - "If anything fails - be sure to read README.lto.md!\n", - AFL_REAL_LD, AFL_CLANG_FLTO); -#endif - - SAYF( - "\nafl-clang-fast was built for llvm %s with the llvm binary path " - "of \"%s\".\n", - LLVM_VERSION, LLVM_BINDIR); - - SAYF("\n"); - - exit(1); - - } else if ((isatty(2) && !be_quiet) || - - getenv("AFL_DEBUG") != NULL) { - - if (!lto_mode) - - SAYF(cCYA "afl-clang-fast" VERSION cRST - " by in %s mode\n", - ptr); - - else - - SAYF(cCYA "afl-clang-lto" VERSION cRST - " by Marc \"vanHauser\" Heuse in mode %s\n", - ptr); - - } - - u8 *ptr2; - if (!be_quiet && !lto_mode && - ((ptr2 = getenv("AFL_MAP_SIZE")) || (ptr2 = getenv("AFL_MAPSIZE")))) { - - u32 map_size = atoi(ptr2); - if (map_size != MAP_SIZE) - WARNF("AFL_MAP_SIZE is not supported by afl-clang-fast"); - - } - - if (debug) { - - SAYF(cMGN "[D]" cRST " cd \"%s\";", getthecwd()); - for (i = 0; i < argc; i++) - SAYF(" \"%s\"", argv[i]); - SAYF("\n"); - - } - - check_environment_vars(envp); - - if (getenv("AFL_LLVM_LAF_ALL")) { - - setenv("AFL_LLVM_LAF_SPLIT_SWITCHES", "1", 1); - setenv("AFL_LLVM_LAF_SPLIT_COMPARES", "1", 1); - setenv("AFL_LLVM_LAF_SPLIT_FLOATS", "1", 1); - setenv("AFL_LLVM_LAF_TRANSFORM_COMPARES", "1", 1); - - } - - cmplog_mode = getenv("AFL_CMPLOG") || getenv("AFL_LLVM_CMPLOG"); - if (!be_quiet && cmplog_mode) - printf("CmpLog mode by \n"); - -#ifndef __ANDROID__ - find_obj(argv[0]); -#endif - - edit_params(argc, argv, envp); - - if (debug) { - - SAYF(cMGN "[D]" cRST " cd \"%s\";", getthecwd()); - for (i = 0; i < cc_par_cnt; i++) - SAYF(" \"%s\"", cc_params[i]); - SAYF("\n"); - - } - - execvp(cc_params[0], (char **)cc_params); - - FATAL("Oops, failed to execute '%s' - check your PATH", cc_params[0]); - - return 0; - -} - diff --git a/llvm_mode/afl-ld-lto.c b/llvm_mode/afl-ld-lto.c deleted file mode 100644 index 771e2d0d..00000000 --- a/llvm_mode/afl-ld-lto.c +++ /dev/null @@ -1,358 +0,0 @@ -/* - american fuzzy lop++ - wrapper for llvm 11+ lld - ----------------------------------------------- - - Written by Marc Heuse for afl++ - - Maintained by Marc Heuse , - Heiko Eißfeldt - Andrea Fioraldi - Dominik Maier - - Copyright 2019-2020 AFLplusplus Project. All rights reserved. - - Licensed under the Apache License, Version 2.0 (the "License"); - you may not use this file except in compliance with the License. - You may obtain a copy of the License at: - - http://www.apache.org/licenses/LICENSE-2.0 - - The sole purpose of this wrapper is to preprocess clang LTO files when - linking with lld and performing the instrumentation on the whole program. - -*/ - -#define AFL_MAIN -#define _GNU_SOURCE - -#include "config.h" -#include "types.h" -#include "debug.h" -#include "alloc-inl.h" - -#include -#include -#include -#include -#include -#include -#include - -#include -#include -#include -#include - -#include - -#define MAX_PARAM_COUNT 4096 - -static u8 **ld_params; /* Parameters passed to the real 'ld' */ - -static u8 *afl_path = AFL_PATH; -static u8 *real_ld = AFL_REAL_LD; - -static u8 be_quiet, /* Quiet mode (no stderr output) */ - debug, /* AFL_DEBUG */ - passthrough, /* AFL_LD_PASSTHROUGH - no link+optimize*/ - just_version; /* Just show version? */ - -static u32 ld_param_cnt = 1; /* Number of params to 'ld' */ - -/* Examine and modify parameters to pass to 'ld', 'llvm-link' and 'llmv-ar'. - Note that the file name is always the last parameter passed by GCC, - so we exploit this property to keep the code "simple". */ -static void edit_params(int argc, char **argv) { - - u32 i, instrim = 0, gold_pos = 0, gold_present = 0, rt_present = 0, - rt_lto_present = 0, inst_present = 0; - char *ptr; - - ld_params = ck_alloc(4096 * sizeof(u8 *)); - - ld_params[0] = (u8 *)real_ld; - - if (!passthrough) { - - for (i = 1; i < argc; i++) { - - if (strstr(argv[i], "/afl-llvm-rt-lto.o") != NULL) rt_lto_present = 1; - if (strstr(argv[i], "/afl-llvm-rt.o") != NULL) rt_present = 1; - if (strstr(argv[i], "/afl-llvm-lto-instr") != NULL) inst_present = 1; - - } - - for (i = 1; i < argc && !gold_pos; i++) { - - if (strcmp(argv[i], "-plugin") == 0) { - - if (strncmp(argv[i], "-plugin=", strlen("-plugin=")) == 0) { - - if (strcasestr(argv[i], "LLVMgold.so") != NULL) - gold_present = gold_pos = i + 1; - - } else if (i < argc && strcasestr(argv[i + 1], "LLVMgold.so") != NULL) { - - gold_present = gold_pos = i + 2; - - } - - } - - } - - if (!gold_pos) { - - for (i = 1; i + 1 < argc && !gold_pos; i++) { - - if (argv[i][0] != '-') { - - if (argv[i - 1][0] == '-') { - - switch (argv[i - 1][1]) { - - case 'b': - break; - case 'd': - break; - case 'e': - break; - case 'F': - break; - case 'f': - break; - case 'I': - break; - case 'l': - break; - case 'L': - break; - case 'm': - break; - case 'o': - break; - case 'O': - break; - case 'p': - if (index(argv[i - 1], '=') == NULL) gold_pos = i; - break; - case 'R': - break; - case 'T': - break; - case 'u': - break; - case 'y': - break; - case 'z': - break; - case '-': { - - if (strcmp(argv[i - 1], "--oformat") == 0) break; - if (strcmp(argv[i - 1], "--output") == 0) break; - if (strncmp(argv[i - 1], "--opt-remarks-", 14) == 0) break; - gold_pos = i; - break; - - } - - default: - gold_pos = i; - - } - - } else - - gold_pos = i; - - } - - } - - } - - if (!gold_pos) gold_pos = 1; - - } - - if (getenv("AFL_LLVM_INSTRIM")) - instrim = 1; - else if ((ptr = getenv("AFL_LLVM_INSTRUMENT")) && - (strcasestr(ptr, "CFG") == 0 || strcasestr(ptr, "INSTRIM") == 0)) - instrim = 1; - - if (debug) - SAYF(cMGN "[D] " cRST - "passthrough=%s instrim=%d, gold_pos=%d, gold_present=%s " - "inst_present=%s rt_present=%s rt_lto_present=%s\n", - passthrough ? "true" : "false", instrim, gold_pos, - gold_present ? "true" : "false", inst_present ? "true" : "false", - rt_present ? "true" : "false", rt_lto_present ? "true" : "false"); - - for (i = 1; i < argc; i++) { - - if (ld_param_cnt >= MAX_PARAM_COUNT) - FATAL( - "Too many command line parameters because of unpacking .a archives, " - "this would need to be done by hand ... sorry! :-("); - - if (strcmp(argv[i], "--afl") == 0) { - - if (!be_quiet) OKF("afl++ test command line flag detected, exiting."); - exit(0); - - } - - if (i == gold_pos && !passthrough) { - - ld_params[ld_param_cnt++] = alloc_printf("-L%s/../lib", LLVM_BINDIR); - - if (!gold_present) { - - ld_params[ld_param_cnt++] = "-plugin"; - ld_params[ld_param_cnt++] = - alloc_printf("%s/../lib/LLVMgold.so", LLVM_BINDIR); - - } - - ld_params[ld_param_cnt++] = "--allow-multiple-definition"; - - if (!inst_present) { - - if (instrim) - ld_params[ld_param_cnt++] = - alloc_printf("-mllvm=-load=%s/afl-llvm-lto-instrim.so", afl_path); - else - ld_params[ld_param_cnt++] = alloc_printf( - "-mllvm=-load=%s/afl-llvm-lto-instrumentation.so", afl_path); - - } - - if (!rt_present) - ld_params[ld_param_cnt++] = alloc_printf("%s/afl-llvm-rt.o", afl_path); - if (!rt_lto_present) - ld_params[ld_param_cnt++] = - alloc_printf("%s/afl-llvm-rt-lto.o", afl_path); - - } - - ld_params[ld_param_cnt++] = argv[i]; - - } - - ld_params[ld_param_cnt] = NULL; - -} - -/* Main entry point */ - -int main(int argc, char **argv) { - - s32 pid, i, status; - u8 * ptr; - char thecwd[PATH_MAX]; - - if ((ptr = getenv("AFL_LD_CALLER")) != NULL) { - - FATAL("ld loop detected! Set AFL_REAL_LD!\n"); - - } - - if (isatty(2) && !getenv("AFL_QUIET") && !getenv("AFL_DEBUG")) { - - SAYF(cCYA "afl-ld-to" VERSION cRST - " by Marc \"vanHauser\" Heuse \n"); - - } else - - be_quiet = 1; - - if (getenv("AFL_DEBUG") != NULL) debug = 1; - if (getenv("AFL_PATH") != NULL) afl_path = getenv("AFL_PATH"); - if (getenv("AFL_LD_PASSTHROUGH") != NULL) passthrough = 1; - if (getenv("AFL_REAL_LD") != NULL) real_ld = getenv("AFL_REAL_LD"); - - if (!afl_path || !*afl_path) afl_path = "/usr/local/lib/afl"; - - setenv("AFL_LD_CALLER", "1", 1); - - if (debug) { - - if (getcwd(thecwd, sizeof(thecwd)) != 0) strcpy(thecwd, "."); - - SAYF(cMGN "[D] " cRST "cd \"%s\";", thecwd); - for (i = 0; i < argc; i++) - SAYF(" \"%s\"", argv[i]); - SAYF("\n"); - - } - - if (argc < 2) { - - SAYF( - "\n" - "This is a helper application for afl-clang-lto. It is a wrapper " - "around GNU " - "llvm's 'lld',\n" - "executed by the toolchain whenever using " - "afl-clang-lto/afl-clang-lto++.\n" - "You probably don't want to run this program directly but rather pass " - "it as LD parameter to configure scripts\n\n" - - "Environment variables:\n" - " AFL_LD_PASSTHROUGH do not link+optimize == no instrumentation\n" - " AFL_REAL_LD point to the real llvm 11 lld if necessary\n" - - "\nafl-ld-to was compiled with the fixed real 'ld' of %s and the " - "binary path of %s\n\n", - real_ld, LLVM_BINDIR); - - exit(1); - - } - - edit_params(argc, argv); // here most of the magic happens :-) - - if (debug) { - - SAYF(cMGN "[D]" cRST " cd \"%s\";", thecwd); - for (i = 0; i < ld_param_cnt; i++) - SAYF(" \"%s\"", ld_params[i]); - SAYF("\n"); - - } - - if (!(pid = fork())) { - - if (strlen(real_ld) > 1) execvp(real_ld, (char **)ld_params); - execvp("ld", (char **)ld_params); // fallback - FATAL("Oops, failed to execute 'ld' - check your PATH"); - - } - - if (pid < 0) PFATAL("fork() failed"); - - if (waitpid(pid, &status, 0) <= 0) PFATAL("waitpid() failed"); - if (debug) SAYF(cMGN "[D] " cRST "linker result: %d\n", status); - - if (!just_version) { - - if (status == 0) { - - if (!be_quiet) OKF("Linker was successful"); - - } else { - - SAYF(cLRD "[-] " cRST - "Linker failed, please investigate and send a bug report. Most " - "likely an 'ld' option is incompatible with %s.\n", - AFL_CLANG_FLTO); - - } - - } - - exit(WEXITSTATUS(status)); - -} - diff --git a/llvm_mode/afl-llvm-common.cc b/llvm_mode/afl-llvm-common.cc deleted file mode 100644 index 189b4ec6..00000000 --- a/llvm_mode/afl-llvm-common.cc +++ /dev/null @@ -1,575 +0,0 @@ -#define AFL_LLVM_PASS - -#include "config.h" -#include "debug.h" - -#include -#include -#include -#include -#include - -#include -#include -#include - -#include - -#define IS_EXTERN extern -#include "afl-llvm-common.h" - -using namespace llvm; - -static std::list allowListFiles; -static std::list allowListFunctions; -static std::list denyListFiles; -static std::list denyListFunctions; - -char *getBBName(const llvm::BasicBlock *BB) { - - static char *name; - - if (!BB->getName().empty()) { - - name = strdup(BB->getName().str().c_str()); - return name; - - } - - std::string Str; - raw_string_ostream OS(Str); - -#if LLVM_VERSION_MAJOR >= 4 || \ - (LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR >= 7) - BB->printAsOperand(OS, false); -#endif - name = strdup(OS.str().c_str()); - return name; - -} - -/* Function that we never instrument or analyze */ -/* Note: this ignore check is also called in isInInstrumentList() */ -bool isIgnoreFunction(const llvm::Function *F) { - - // Starting from "LLVMFuzzer" these are functions used in libfuzzer based - // fuzzing campaign installations, e.g. oss-fuzz - - static const char *ignoreList[] = { - - "asan.", - "llvm.", - "sancov.", - "__ubsan_", - "ign.", - "__afl_", - "_fini", - "__libc_csu", - "__asan", - "__msan", - "__cmplog", - "__sancov", - "msan.", - "LLVMFuzzer", - "__decide_deferred", - "maybe_duplicate_stderr", - "discard_output", - "close_stdout", - "dup_and_close_stderr", - "maybe_close_fd_mask", - "ExecuteFilesOnyByOne" - - }; - - for (auto const &ignoreListFunc : ignoreList) { - - if (F->getName().startswith(ignoreListFunc)) { return true; } - - } - - return false; - -} - -void initInstrumentList() { - - char *allowlist = getenv("AFL_LLVM_ALLOWLIST"); - if (!allowlist) allowlist = getenv("AFL_LLVM_INSTRUMENT_FILE"); - if (!allowlist) allowlist = getenv("AFL_LLVM_WHITELIST"); - char *denylist = getenv("AFL_LLVM_DENYLIST"); - if (!denylist) denylist = getenv("AFL_LLVM_BLOCKLIST"); - - if (allowlist && denylist) - FATAL( - "You can only specify either AFL_LLVM_ALLOWLIST or AFL_LLVM_DENYLIST " - "but not both!"); - - if (allowlist) { - - std::string line; - std::ifstream fileStream; - fileStream.open(allowlist); - if (!fileStream) report_fatal_error("Unable to open AFL_LLVM_ALLOWLIST"); - getline(fileStream, line); - - while (fileStream) { - - int is_file = -1; - std::size_t npos; - std::string original_line = line; - - line.erase(std::remove_if(line.begin(), line.end(), ::isspace), - line.end()); - - // remove # and following - if ((npos = line.find("#")) != std::string::npos) - line = line.substr(0, npos); - - if (line.compare(0, 4, "fun:") == 0) { - - is_file = 0; - line = line.substr(4); - - } else if (line.compare(0, 9, "function:") == 0) { - - is_file = 0; - line = line.substr(9); - - } else if (line.compare(0, 4, "src:") == 0) { - - is_file = 1; - line = line.substr(4); - - } else if (line.compare(0, 7, "source:") == 0) { - - is_file = 1; - line = line.substr(7); - - } - - if (line.find(":") != std::string::npos) { - - FATAL("invalid line in AFL_LLVM_ALLOWLIST: %s", original_line.c_str()); - - } - - if (line.length() > 0) { - - // if the entry contains / or . it must be a file - if (is_file == -1) - if (line.find("/") != std::string::npos || - line.find(".") != std::string::npos) - is_file = 1; - // otherwise it is a function - - if (is_file == 1) - allowListFiles.push_back(line); - else - allowListFunctions.push_back(line); - getline(fileStream, line); - - } - - } - - if (debug) - SAYF(cMGN "[D] " cRST - "loaded allowlist with %zu file and %zu function entries\n", - allowListFiles.size(), allowListFunctions.size()); - - } - - if (denylist) { - - std::string line; - std::ifstream fileStream; - fileStream.open(denylist); - if (!fileStream) report_fatal_error("Unable to open AFL_LLVM_DENYLIST"); - getline(fileStream, line); - - while (fileStream) { - - int is_file = -1; - std::size_t npos; - std::string original_line = line; - - line.erase(std::remove_if(line.begin(), line.end(), ::isspace), - line.end()); - - // remove # and following - if ((npos = line.find("#")) != std::string::npos) - line = line.substr(0, npos); - - if (line.compare(0, 4, "fun:") == 0) { - - is_file = 0; - line = line.substr(4); - - } else if (line.compare(0, 9, "function:") == 0) { - - is_file = 0; - line = line.substr(9); - - } else if (line.compare(0, 4, "src:") == 0) { - - is_file = 1; - line = line.substr(4); - - } else if (line.compare(0, 7, "source:") == 0) { - - is_file = 1; - line = line.substr(7); - - } - - if (line.find(":") != std::string::npos) { - - FATAL("invalid line in AFL_LLVM_DENYLIST: %s", original_line.c_str()); - - } - - if (line.length() > 0) { - - // if the entry contains / or . it must be a file - if (is_file == -1) - if (line.find("/") != std::string::npos || - line.find(".") != std::string::npos) - is_file = 1; - // otherwise it is a function - - if (is_file == 1) - denyListFiles.push_back(line); - else - denyListFunctions.push_back(line); - getline(fileStream, line); - - } - - } - - if (debug) - SAYF(cMGN "[D] " cRST - "loaded denylist with %zu file and %zu function entries\n", - denyListFiles.size(), denyListFunctions.size()); - - } - -} - -void scanForDangerousFunctions(llvm::Module *M) { - - if (!M) return; - -#if LLVM_VERSION_MAJOR > 3 || \ - (LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR >= 9) - - for (GlobalIFunc &IF : M->ifuncs()) { - - StringRef ifunc_name = IF.getName(); - Constant *r = IF.getResolver(); - StringRef r_name = cast(r->getOperand(0))->getName(); - if (!be_quiet) - fprintf(stderr, - "Info: Found an ifunc with name %s that points to resolver " - "function %s, we will not instrument this, putting it into the " - "block list.\n", - ifunc_name.str().c_str(), r_name.str().c_str()); - denyListFunctions.push_back(r_name.str()); - - } - - GlobalVariable *GV = M->getNamedGlobal("llvm.global_ctors"); - if (GV && !GV->isDeclaration() && !GV->hasLocalLinkage()) { - - ConstantArray *InitList = dyn_cast(GV->getInitializer()); - - if (InitList) { - - for (unsigned i = 0, e = InitList->getNumOperands(); i != e; ++i) { - - if (ConstantStruct *CS = - dyn_cast(InitList->getOperand(i))) { - - if (CS->getNumOperands() >= 2) { - - if (CS->getOperand(1)->isNullValue()) - break; // Found a null terminator, stop here. - - ConstantInt *CI = dyn_cast(CS->getOperand(0)); - int Priority = CI ? CI->getSExtValue() : 0; - - Constant *FP = CS->getOperand(1); - if (ConstantExpr *CE = dyn_cast(FP)) - if (CE->isCast()) FP = CE->getOperand(0); - if (Function *F = dyn_cast(FP)) { - - if (!F->isDeclaration() && - strncmp(F->getName().str().c_str(), "__afl", 5) != 0) { - - if (!be_quiet) - fprintf(stderr, - "Info: Found constructor function %s with prio " - "%u, we will not instrument this, putting it into a " - "block list.\n", - F->getName().str().c_str(), Priority); - denyListFunctions.push_back(F->getName().str()); - - } - - } - - } - - } - - } - - } - - } - -#endif - -} - -static std::string getSourceName(llvm::Function *F) { - - // let's try to get the filename for the function - auto bb = &F->getEntryBlock(); - BasicBlock::iterator IP = bb->getFirstInsertionPt(); - IRBuilder<> IRB(&(*IP)); - DebugLoc Loc = IP->getDebugLoc(); - -#if LLVM_VERSION_MAJOR >= 4 || \ - (LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR >= 7) - if (Loc) { - - StringRef instFilename; - DILocation *cDILoc = dyn_cast(Loc.getAsMDNode()); - - if (cDILoc) { instFilename = cDILoc->getFilename(); } - - if (instFilename.str().empty()) { - - /* If the original location is empty, try using the inlined location - */ - DILocation *oDILoc = cDILoc->getInlinedAt(); - if (oDILoc) { instFilename = oDILoc->getFilename(); } - - } - - return instFilename.str(); - - } - -#else - if (!Loc.isUnknown()) { - - DILocation cDILoc(Loc.getAsMDNode(F->getContext())); - - StringRef instFilename = cDILoc.getFilename(); - - /* Continue only if we know where we actually are */ - return instFilename.str(); - - } - -#endif - - return std::string(""); - -} - -bool isInInstrumentList(llvm::Function *F) { - - bool return_default = true; - - // is this a function with code? If it is external we don't instrument it - // anyway and it can't be in the instrument file list. Or if it is it is - // ignored. - if (!F->size() || isIgnoreFunction(F)) return false; - - if (!denyListFiles.empty() || !denyListFunctions.empty()) { - - if (!denyListFunctions.empty()) { - - std::string instFunction = F->getName().str(); - - for (std::list::iterator it = denyListFunctions.begin(); - it != denyListFunctions.end(); ++it) { - - /* We don't check for filename equality here because - * filenames might actually be full paths. Instead we - * check that the actual filename ends in the filename - * specified in the list. We also allow UNIX-style pattern - * matching */ - - if (instFunction.length() >= it->length()) { - - if (fnmatch(("*" + *it).c_str(), instFunction.c_str(), 0) == 0) { - - if (debug) - SAYF(cMGN "[D] " cRST - "Function %s is in the deny function list, " - "not instrumenting ... \n", - instFunction.c_str()); - return false; - - } - - } - - } - - } - - if (!denyListFiles.empty()) { - - std::string source_file = getSourceName(F); - - if (!source_file.empty()) { - - for (std::list::iterator it = denyListFiles.begin(); - it != denyListFiles.end(); ++it) { - - /* We don't check for filename equality here because - * filenames might actually be full paths. Instead we - * check that the actual filename ends in the filename - * specified in the list. We also allow UNIX-style pattern - * matching */ - - if (source_file.length() >= it->length()) { - - if (fnmatch(("*" + *it).c_str(), source_file.c_str(), 0) == 0) { - - return false; - - } - - } - - } - - } else { - - // we could not find out the location. in this case we say it is not - // in the instrument file list - if (!be_quiet) - WARNF( - "No debug information found for function %s, will be " - "instrumented (recompile with -g -O[1-3]).", - F->getName().str().c_str()); - - } - - } - - } - - // if we do not have a instrument file list return true - if (!allowListFiles.empty() || !allowListFunctions.empty()) { - - return_default = false; - - if (!allowListFunctions.empty()) { - - std::string instFunction = F->getName().str(); - - for (std::list::iterator it = allowListFunctions.begin(); - it != allowListFunctions.end(); ++it) { - - /* We don't check for filename equality here because - * filenames might actually be full paths. Instead we - * check that the actual filename ends in the filename - * specified in the list. We also allow UNIX-style pattern - * matching */ - - if (instFunction.length() >= it->length()) { - - if (fnmatch(("*" + *it).c_str(), instFunction.c_str(), 0) == 0) { - - if (debug) - SAYF(cMGN "[D] " cRST - "Function %s is in the allow function list, " - "instrumenting ... \n", - instFunction.c_str()); - return true; - - } - - } - - } - - } - - if (!allowListFiles.empty()) { - - std::string source_file = getSourceName(F); - - if (!source_file.empty()) { - - for (std::list::iterator it = allowListFiles.begin(); - it != allowListFiles.end(); ++it) { - - /* We don't check for filename equality here because - * filenames might actually be full paths. Instead we - * check that the actual filename ends in the filename - * specified in the list. We also allow UNIX-style pattern - * matching */ - - if (source_file.length() >= it->length()) { - - if (fnmatch(("*" + *it).c_str(), source_file.c_str(), 0) == 0) { - - if (debug) - SAYF(cMGN "[D] " cRST - "Function %s is in the allowlist (%s), " - "instrumenting ... \n", - F->getName().str().c_str(), source_file.c_str()); - return true; - - } - - } - - } - - } else { - - // we could not find out the location. In this case we say it is not - // in the instrument file list - if (!be_quiet) - WARNF( - "No debug information found for function %s, will not be " - "instrumented (recompile with -g -O[1-3]).", - F->getName().str().c_str()); - return false; - - } - - } - - } - - return return_default; - -} - -// Calculate the number of average collisions that would occur if all -// location IDs would be assigned randomly (like normal afl/afl++). -// This uses the "balls in bins" algorithm. -unsigned long long int calculateCollisions(uint32_t edges) { - - double bins = MAP_SIZE; - double balls = edges; - double step1 = 1 - (1 / bins); - double step2 = pow(step1, balls); - double step3 = bins * step2; - double step4 = round(step3); - unsigned long long int empty = step4; - unsigned long long int collisions = edges - (MAP_SIZE - empty); - return collisions; - -} - diff --git a/llvm_mode/afl-llvm-common.h b/llvm_mode/afl-llvm-common.h deleted file mode 100644 index a1561d9c..00000000 --- a/llvm_mode/afl-llvm-common.h +++ /dev/null @@ -1,52 +0,0 @@ -#ifndef __AFLLLVMCOMMON_H -#define __AFLLLVMCOMMON_H - -#include -#include -#include - -#include -#include -#include -#include - -#include "llvm/Config/llvm-config.h" -#if LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR < 5 -typedef long double max_align_t; -#endif - -#include "llvm/IR/IRBuilder.h" -#include "llvm/IR/LegacyPassManager.h" -#include "llvm/IR/BasicBlock.h" -#include "llvm/IR/Module.h" -#include "llvm/Support/Debug.h" -#include "llvm/Support/MathExtras.h" -#include "llvm/Transforms/IPO/PassManagerBuilder.h" - -#if LLVM_VERSION_MAJOR > 3 || \ - (LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR > 4) - #include "llvm/IR/DebugInfo.h" - #include "llvm/IR/CFG.h" -#else - #include "llvm/DebugInfo.h" - #include "llvm/Support/CFG.h" -#endif - -char * getBBName(const llvm::BasicBlock *BB); -bool isIgnoreFunction(const llvm::Function *F); -void initInstrumentList(); -bool isInInstrumentList(llvm::Function *F); -unsigned long long int calculateCollisions(uint32_t edges); -void scanForDangerousFunctions(llvm::Module *M); - -#ifndef IS_EXTERN - #define IS_EXTERN -#endif - -IS_EXTERN int debug; -IS_EXTERN int be_quiet; - -#undef IS_EXTERN - -#endif - diff --git a/llvm_mode/afl-llvm-lto-instrumentation.so.cc b/llvm_mode/afl-llvm-lto-instrumentation.so.cc deleted file mode 100644 index 125db229..00000000 --- a/llvm_mode/afl-llvm-lto-instrumentation.so.cc +++ /dev/null @@ -1,957 +0,0 @@ -/* - american fuzzy lop++ - LLVM LTO instrumentation pass - ---------------------------------------------------- - - Written by Marc Heuse - - Copyright 2019-2020 AFLplusplus Project. All rights reserved. - - Licensed under the Apache License, Version 2.0 (the "License"); - you may not use this file except in compliance with the License. - You may obtain a copy of the License at: - - http://www.apache.org/licenses/LICENSE-2.0 - - This library is plugged into LLVM when invoking clang through afl-clang-lto. - - */ - -#define AFL_LLVM_PASS - -#include "config.h" -#include "debug.h" - -#include -#include -#include -#include -#include - -#include -#include -#include -#include - -#include "llvm/Config/llvm-config.h" -#include "llvm/ADT/Statistic.h" -#include "llvm/IR/IRBuilder.h" -#include "llvm/IR/LegacyPassManager.h" -#include "llvm/IR/BasicBlock.h" -#include "llvm/IR/Module.h" -#include "llvm/IR/DebugInfo.h" -#include "llvm/IR/CFG.h" -#include "llvm/IR/Verifier.h" -#include "llvm/Support/Debug.h" -#include "llvm/Support/raw_ostream.h" -#include "llvm/Transforms/IPO/PassManagerBuilder.h" -#include "llvm/Transforms/Utils/BasicBlockUtils.h" -#include "llvm/Analysis/LoopInfo.h" -#include "llvm/Analysis/MemorySSAUpdater.h" -#include "llvm/Analysis/ValueTracking.h" -#include "llvm/Pass.h" -#include "llvm/IR/Constants.h" - -#include "afl-llvm-common.h" - -using namespace llvm; - -namespace { - -class AFLLTOPass : public ModulePass { - - public: - static char ID; - - AFLLTOPass() : ModulePass(ID) { - - char *ptr; - - if (getenv("AFL_DEBUG")) debug = 1; - if ((ptr = getenv("AFL_LLVM_LTO_STARTID")) != NULL) - if ((afl_global_id = atoi(ptr)) < 0 || afl_global_id >= MAP_SIZE) - FATAL("AFL_LLVM_LTO_STARTID value of \"%s\" is not between 0 and %d\n", - ptr, MAP_SIZE - 1); - - skip_nozero = getenv("AFL_LLVM_SKIP_NEVERZERO"); - - } - - void getAnalysisUsage(AnalysisUsage &AU) const override { - - ModulePass::getAnalysisUsage(AU); - AU.addRequired(); - AU.addRequired(); - - } - - bool runOnModule(Module &M) override; - - protected: - int afl_global_id = 1, autodictionary = 1; - uint32_t function_minimum_size = 1; - uint32_t inst_blocks = 0, inst_funcs = 0, total_instr = 0; - uint64_t map_addr = 0x10000; - char * skip_nozero = NULL; - -}; - -} // namespace - -bool AFLLTOPass::runOnModule(Module &M) { - - LLVMContext & C = M.getContext(); - std::vector dictionary; - std::vector calls; - DenseMap valueMap; - std::vector BlockList; - char * ptr; - FILE * documentFile = NULL; - - srand((unsigned int)time(NULL)); - - unsigned long long int moduleID = - (((unsigned long long int)(rand() & 0xffffffff)) << 32) | getpid(); - - IntegerType *Int8Ty = IntegerType::getInt8Ty(C); - IntegerType *Int32Ty = IntegerType::getInt32Ty(C); - IntegerType *Int64Ty = IntegerType::getInt64Ty(C); - - /* Show a banner */ - setvbuf(stdout, NULL, _IONBF, 0); - - if ((isatty(2) && !getenv("AFL_QUIET")) || debug) { - - SAYF(cCYA "afl-llvm-lto" VERSION cRST - " by Marc \"vanHauser\" Heuse \n"); - - } else - - be_quiet = 1; - - if ((ptr = getenv("AFL_LLVM_DOCUMENT_IDS")) != NULL) { - - if ((documentFile = fopen(ptr, "a")) == NULL) - WARNF("Cannot access document file %s", ptr); - - } - - // we make this the default as the fixed map has problems with - // defered forkserver, early constructors, ifuncs and maybe more - /*if (getenv("AFL_LLVM_MAP_DYNAMIC"))*/ - map_addr = 0; - - if ((ptr = getenv("AFL_LLVM_MAP_ADDR"))) { - - uint64_t val; - if (!*ptr || !strcmp(ptr, "0") || !strcmp(ptr, "0x0")) { - - map_addr = 0; - - } else if (getenv("AFL_LLVM_MAP_DYNAMIC")) { - - FATAL( - "AFL_LLVM_MAP_ADDR and AFL_LLVM_MAP_DYNAMIC cannot be used together"); - - } else if (strncmp(ptr, "0x", 2) != 0) { - - map_addr = 0x10000; // the default - - } else { - - val = strtoull(ptr, NULL, 16); - if (val < 0x100 || val > 0xffffffff00000000) { - - FATAL( - "AFL_LLVM_MAP_ADDR must be a value between 0x100 and " - "0xffffffff00000000"); - - } - - map_addr = val; - - } - - } - - if (debug) { fprintf(stderr, "map address is 0x%lx\n", map_addr); } - - /* Get/set the globals for the SHM region. */ - - GlobalVariable *AFLMapPtr = NULL; - Value * MapPtrFixed = NULL; - - if (!map_addr) { - - AFLMapPtr = - new GlobalVariable(M, PointerType::get(Int8Ty, 0), false, - GlobalValue::ExternalLinkage, 0, "__afl_area_ptr"); - - } else { - - ConstantInt *MapAddr = ConstantInt::get(Int64Ty, map_addr); - MapPtrFixed = - ConstantExpr::getIntToPtr(MapAddr, PointerType::getUnqual(Int8Ty)); - - } - - ConstantInt *Zero = ConstantInt::get(Int8Ty, 0); - ConstantInt *One = ConstantInt::get(Int8Ty, 1); - - // This dumps all inialized global strings - might be useful in the future - /* - for (auto G=M.getGlobalList().begin(); G!=M.getGlobalList().end(); G++) { - - GlobalVariable &GV=*G; - if (!GV.getName().str().empty()) { - - fprintf(stderr, "Global Variable: %s", GV.getName().str().c_str()); - if (GV.hasInitializer()) - if (auto *Val = dyn_cast(GV.getInitializer())) - fprintf(stderr, " Value: \"%s\"", Val->getAsString().str().c_str()); - fprintf(stderr, "\n"); - - } - - } - - */ - - scanForDangerousFunctions(&M); - - /* Instrument all the things! */ - - int inst_blocks = 0; - - for (auto &F : M) { - - /*For debugging - AttributeSet X = F.getAttributes().getFnAttributes(); - fprintf(stderr, "DEBUG: Module %s Function %s attributes %u\n", - M.getName().str().c_str(), F.getName().str().c_str(), - X.getNumAttributes()); - */ - - if (F.size() < function_minimum_size) continue; - if (isIgnoreFunction(&F)) continue; - - // the instrument file list check - AttributeList Attrs = F.getAttributes(); - if (Attrs.hasAttribute(-1, StringRef("skipinstrument"))) { - - if (debug) - fprintf(stderr, - "DEBUG: Function %s is not in a source file that was specified " - "in the instrument file list\n", - F.getName().str().c_str()); - continue; - - } - - std::vector InsBlocks; - - if (autodictionary) { - - /* Some implementation notes. - * - * We try to handle 3 cases: - * - memcmp("foo", arg, 3) <- literal string - * - static char globalvar[] = "foo"; - * memcmp(globalvar, arg, 3) <- global variable - * - char localvar[] = "foo"; - * memcmp(locallvar, arg, 3) <- local variable - * - * The local variable case is the hardest. We can only detect that - * case if there is no reassignment or change in the variable. - * And it might not work across llvm version. - * What we do is hooking the initializer function for local variables - * (llvm.memcpy.p0i8.p0i8.i64) and note the string and the assigned - * variable. And if that variable is then used in a compare function - * we use that noted string. - * This seems not to work for tokens that have a size <= 4 :-( - * - * - if the compared length is smaller than the string length we - * save the full string. This is likely better for fuzzing but - * might be wrong in a few cases depending on optimizers - * - * - not using StringRef because there is a bug in the llvm 11 - * checkout I am using which sometimes points to wrong strings - * - * Over and out. Took me a full day. damn. mh/vh - */ - - for (auto &BB : F) { - - for (auto &IN : BB) { - - CallInst *callInst = nullptr; - - if ((callInst = dyn_cast(&IN))) { - - bool isStrcmp = true; - bool isMemcmp = true; - bool isStrncmp = true; - bool isStrcasecmp = true; - bool isStrncasecmp = true; - bool isIntMemcpy = true; - bool addedNull = false; - size_t optLen = 0; - - Function *Callee = callInst->getCalledFunction(); - if (!Callee) continue; - if (callInst->getCallingConv() != llvm::CallingConv::C) continue; - std::string FuncName = Callee->getName().str(); - isStrcmp &= !FuncName.compare("strcmp"); - isMemcmp &= !FuncName.compare("memcmp"); - isStrncmp &= !FuncName.compare("strncmp"); - isStrcasecmp &= !FuncName.compare("strcasecmp"); - isStrncasecmp &= !FuncName.compare("strncasecmp"); - isIntMemcpy &= !FuncName.compare("llvm.memcpy.p0i8.p0i8.i64"); - - /* we do something different here, putting this BB and the - successors in a block map */ - if (!FuncName.compare("__afl_persistent_loop")) { - - BlockList.push_back(&BB); - /* - for (succ_iterator SI = succ_begin(&BB), SE = - succ_end(&BB); SI != SE; ++SI) { - - BasicBlock *succ = *SI; - BlockList.push_back(succ); - - } - - */ - - } - - if (!isStrcmp && !isMemcmp && !isStrncmp && !isStrcasecmp && - !isStrncasecmp && !isIntMemcpy) - continue; - - /* Verify the strcmp/memcmp/strncmp/strcasecmp/strncasecmp function - * prototype */ - FunctionType *FT = Callee->getFunctionType(); - - isStrcmp &= FT->getNumParams() == 2 && - FT->getReturnType()->isIntegerTy(32) && - FT->getParamType(0) == FT->getParamType(1) && - FT->getParamType(0) == - IntegerType::getInt8PtrTy(M.getContext()); - isStrcasecmp &= FT->getNumParams() == 2 && - FT->getReturnType()->isIntegerTy(32) && - FT->getParamType(0) == FT->getParamType(1) && - FT->getParamType(0) == - IntegerType::getInt8PtrTy(M.getContext()); - isMemcmp &= FT->getNumParams() == 3 && - FT->getReturnType()->isIntegerTy(32) && - FT->getParamType(0)->isPointerTy() && - FT->getParamType(1)->isPointerTy() && - FT->getParamType(2)->isIntegerTy(); - isStrncmp &= FT->getNumParams() == 3 && - FT->getReturnType()->isIntegerTy(32) && - FT->getParamType(0) == FT->getParamType(1) && - FT->getParamType(0) == - IntegerType::getInt8PtrTy(M.getContext()) && - FT->getParamType(2)->isIntegerTy(); - isStrncasecmp &= FT->getNumParams() == 3 && - FT->getReturnType()->isIntegerTy(32) && - FT->getParamType(0) == FT->getParamType(1) && - FT->getParamType(0) == - IntegerType::getInt8PtrTy(M.getContext()) && - FT->getParamType(2)->isIntegerTy(); - - if (!isStrcmp && !isMemcmp && !isStrncmp && !isStrcasecmp && - !isStrncasecmp && !isIntMemcpy) - continue; - - /* is a str{n,}{case,}cmp/memcmp, check if we have - * str{case,}cmp(x, "const") or str{case,}cmp("const", x) - * strn{case,}cmp(x, "const", ..) or strn{case,}cmp("const", x, ..) - * memcmp(x, "const", ..) or memcmp("const", x, ..) */ - Value *Str1P = callInst->getArgOperand(0), - *Str2P = callInst->getArgOperand(1); - std::string Str1, Str2; - StringRef TmpStr; - bool HasStr1 = getConstantStringInfo(Str1P, TmpStr); - if (TmpStr.empty()) { - - HasStr1 = false; - - } else { - - HasStr1 = true; - Str1 = TmpStr.str(); - - } - - bool HasStr2 = getConstantStringInfo(Str2P, TmpStr); - if (TmpStr.empty()) { - - HasStr2 = false; - - } else { - - HasStr2 = true; - Str2 = TmpStr.str(); - - } - - if (debug) - fprintf(stderr, "F:%s %p(%s)->\"%s\"(%s) %p(%s)->\"%s\"(%s)\n", - FuncName.c_str(), Str1P, Str1P->getName().str().c_str(), - Str1.c_str(), HasStr1 == true ? "true" : "false", Str2P, - Str2P->getName().str().c_str(), Str2.c_str(), - HasStr2 == true ? "true" : "false"); - - // we handle the 2nd parameter first because of llvm memcpy - if (!HasStr2) { - - auto *Ptr = dyn_cast(Str2P); - if (Ptr && Ptr->isGEPWithNoNotionalOverIndexing()) { - - if (auto *Var = dyn_cast(Ptr->getOperand(0))) { - - if (Var->hasInitializer()) { - - if (auto *Array = dyn_cast( - Var->getInitializer())) { - - HasStr2 = true; - Str2 = Array->getAsString().str(); - - } - - } - - } - - } - - } - - // for the internal memcpy routine we only care for the second - // parameter and are not reporting anything. - if (isIntMemcpy == true) { - - if (HasStr2 == true) { - - Value * op2 = callInst->getArgOperand(2); - ConstantInt *ilen = dyn_cast(op2); - if (ilen) { - - uint64_t literalLength = Str2.size(); - uint64_t optLength = ilen->getZExtValue(); - if (literalLength + 1 == optLength) { - - Str2.append("\0", 1); // add null byte - addedNull = true; - - } - - } - - valueMap[Str1P] = new std::string(Str2); - - if (debug) - fprintf(stderr, "Saved: %s for %p\n", Str2.c_str(), Str1P); - continue; - - } - - continue; - - } - - // Neither a literal nor a global variable? - // maybe it is a local variable that we saved - if (!HasStr2) { - - std::string *strng = valueMap[Str2P]; - if (strng && !strng->empty()) { - - Str2 = *strng; - HasStr2 = true; - if (debug) - fprintf(stderr, "Filled2: %s for %p\n", strng->c_str(), - Str2P); - - } - - } - - if (!HasStr1) { - - auto Ptr = dyn_cast(Str1P); - - if (Ptr && Ptr->isGEPWithNoNotionalOverIndexing()) { - - if (auto *Var = dyn_cast(Ptr->getOperand(0))) { - - if (Var->hasInitializer()) { - - if (auto *Array = dyn_cast( - Var->getInitializer())) { - - HasStr1 = true; - Str1 = Array->getAsString().str(); - - } - - } - - } - - } - - } - - // Neither a literal nor a global variable? - // maybe it is a local variable that we saved - if (!HasStr1) { - - std::string *strng = valueMap[Str1P]; - if (strng && !strng->empty()) { - - Str1 = *strng; - HasStr1 = true; - if (debug) - fprintf(stderr, "Filled1: %s for %p\n", strng->c_str(), - Str1P); - - } - - } - - /* handle cases of one string is const, one string is variable */ - if (!(HasStr1 ^ HasStr2)) continue; - - std::string thestring; - - if (HasStr1) - thestring = Str1; - else - thestring = Str2; - - optLen = thestring.length(); - - if (isMemcmp || isStrncmp || isStrncasecmp) { - - Value * op2 = callInst->getArgOperand(2); - ConstantInt *ilen = dyn_cast(op2); - if (ilen) { - - uint64_t literalLength = optLen; - optLen = ilen->getZExtValue(); - if (literalLength + 1 == optLen) { // add null byte - thestring.append("\0", 1); - addedNull = true; - - } - - } - - } - - // add null byte if this is a string compare function and a null - // was not already added - if (!isMemcmp) { - - if (addedNull == false) { - - thestring.append("\0", 1); // add null byte - optLen++; - - } - - // ensure we do not have garbage - size_t offset = thestring.find('\0', 0); - if (offset + 1 < optLen) optLen = offset + 1; - thestring = thestring.substr(0, optLen); - - } - - if (!be_quiet) { - - std::string outstring; - fprintf(stderr, "%s: length %zu/%zu \"", FuncName.c_str(), optLen, - thestring.length()); - for (uint8_t i = 0; i < thestring.length(); i++) { - - uint8_t c = thestring[i]; - if (c <= 32 || c >= 127) - fprintf(stderr, "\\x%02x", c); - else - fprintf(stderr, "%c", c); - - } - - fprintf(stderr, "\"\n"); - - } - - // we take the longer string, even if the compare was to a - // shorter part. Note that depending on the optimizer of the - // compiler this can be wrong, but it is more likely that this - // is helping the fuzzer - if (optLen != thestring.length()) optLen = thestring.length(); - if (optLen > MAX_AUTO_EXTRA) optLen = MAX_AUTO_EXTRA; - if (optLen < MIN_AUTO_EXTRA) // too short? skip - continue; - - dictionary.push_back(thestring.substr(0, optLen)); - - } - - } - - } - - } - - for (auto &BB : F) { - - if (F.size() == 1) { - - InsBlocks.push_back(&BB); - continue; - - } - - uint32_t succ = 0; - for (succ_iterator SI = succ_begin(&BB), SE = succ_end(&BB); SI != SE; - ++SI) - if ((*SI)->size() > 0) succ++; - if (succ < 2) // no need to instrument - continue; - - if (BlockList.size()) { - - int skip = 0; - for (uint32_t k = 0; k < BlockList.size(); k++) { - - if (&BB == BlockList[k]) { - - if (debug) - fprintf(stderr, - "DEBUG: Function %s skipping BB with/after __afl_loop\n", - F.getName().str().c_str()); - skip = 1; - - } - - } - - if (skip) continue; - - } - - InsBlocks.push_back(&BB); - - } - - if (InsBlocks.size() > 0) { - - uint32_t i = InsBlocks.size(); - - do { - - --i; - BasicBlock * newBB = NULL; - BasicBlock * origBB = &(*InsBlocks[i]); - std::vector Successors; - Instruction * TI = origBB->getTerminator(); - uint32_t fs = origBB->getParent()->size(); - uint32_t countto; - - for (succ_iterator SI = succ_begin(origBB), SE = succ_end(origBB); - SI != SE; ++SI) { - - BasicBlock *succ = *SI; - Successors.push_back(succ); - - } - - if (fs == 1) { - - newBB = origBB; - countto = 1; - - } else { - - if (TI == NULL || TI->getNumSuccessors() < 2) continue; - countto = Successors.size(); - - } - - // if (Successors.size() != TI->getNumSuccessors()) - // FATAL("Different successor numbers %lu <-> %u\n", Successors.size(), - // TI->getNumSuccessors()); - - for (uint32_t j = 0; j < countto; j++) { - - if (fs != 1) newBB = llvm::SplitEdge(origBB, Successors[j]); - - if (!newBB) { - - if (!be_quiet) WARNF("Split failed!"); - continue; - - } - - if (documentFile) { - - fprintf(documentFile, "ModuleID=%llu Function=%s edgeID=%u\n", - moduleID, F.getName().str().c_str(), afl_global_id); - - } - - BasicBlock::iterator IP = newBB->getFirstInsertionPt(); - IRBuilder<> IRB(&(*IP)); - - /* Set the ID of the inserted basic block */ - - ConstantInt *CurLoc = ConstantInt::get(Int32Ty, afl_global_id++); - - /* Load SHM pointer */ - - Value *MapPtrIdx; - - if (map_addr) { - - MapPtrIdx = IRB.CreateGEP(MapPtrFixed, CurLoc); - - } else { - - LoadInst *MapPtr = IRB.CreateLoad(AFLMapPtr); - MapPtr->setMetadata(M.getMDKindID("nosanitize"), - MDNode::get(C, None)); - MapPtrIdx = IRB.CreateGEP(MapPtr, CurLoc); - - } - - /* Update bitmap */ - - LoadInst *Counter = IRB.CreateLoad(MapPtrIdx); - Counter->setMetadata(M.getMDKindID("nosanitize"), - MDNode::get(C, None)); - - Value *Incr = IRB.CreateAdd(Counter, One); - - if (skip_nozero == NULL) { - - auto cf = IRB.CreateICmpEQ(Incr, Zero); - auto carry = IRB.CreateZExt(cf, Int8Ty); - Incr = IRB.CreateAdd(Incr, carry); - - } - - IRB.CreateStore(Incr, MapPtrIdx) - ->setMetadata(M.getMDKindID("nosanitize"), MDNode::get(C, None)); - - // done :) - - inst_blocks++; - - } - - } while (i > 0); - - } - - } - - if (documentFile) fclose(documentFile); - documentFile = NULL; - - // save highest location ID to global variable - // do this after each function to fail faster - if (!be_quiet && afl_global_id > MAP_SIZE && - afl_global_id > FS_OPT_MAX_MAPSIZE) { - - uint32_t pow2map = 1, map = afl_global_id; - while ((map = map >> 1)) - pow2map++; - WARNF( - "We have %u blocks to instrument but the map size is only %u. Either " - "edit config.h and set MAP_SIZE_POW2 from %u to %u, then recompile " - "afl-fuzz and llvm_mode and then make this target - or set " - "AFL_MAP_SIZE with at least size %u when running afl-fuzz with this " - "target.", - afl_global_id, MAP_SIZE, MAP_SIZE_POW2, pow2map, afl_global_id); - - } - - if (!getenv("AFL_LLVM_LTO_DONTWRITEID") || dictionary.size() || map_addr) { - - // yes we could create our own function, insert it into ctors ... - // but this would be a pain in the butt ... so we use afl-llvm-rt-lto.o - - Function *f = M.getFunction("__afl_auto_init_globals"); - - if (!f) { - - fprintf(stderr, - "Error: init function could not be found (this should not " - "happen)\n"); - exit(-1); - - } - - BasicBlock *bb = &f->getEntryBlock(); - if (!bb) { - - fprintf(stderr, - "Error: init function does not have an EntryBlock (this should " - "not happen)\n"); - exit(-1); - - } - - BasicBlock::iterator IP = bb->getFirstInsertionPt(); - IRBuilder<> IRB(&(*IP)); - - if (map_addr) { - - GlobalVariable *AFLMapAddrFixed = new GlobalVariable( - M, Int64Ty, true, GlobalValue::ExternalLinkage, 0, "__afl_map_addr"); - ConstantInt *MapAddr = ConstantInt::get(Int64Ty, map_addr); - StoreInst * StoreMapAddr = IRB.CreateStore(MapAddr, AFLMapAddrFixed); - StoreMapAddr->setMetadata(M.getMDKindID("nosanitize"), - MDNode::get(C, None)); - - } - - if (getenv("AFL_LLVM_LTO_DONTWRITEID") == NULL) { - - uint32_t write_loc = afl_global_id; - - if (afl_global_id % 8) write_loc = (((afl_global_id + 8) >> 3) << 3); - - GlobalVariable *AFLFinalLoc = new GlobalVariable( - M, Int32Ty, true, GlobalValue::ExternalLinkage, 0, "__afl_final_loc"); - ConstantInt *const_loc = ConstantInt::get(Int32Ty, write_loc); - StoreInst * StoreFinalLoc = IRB.CreateStore(const_loc, AFLFinalLoc); - StoreFinalLoc->setMetadata(M.getMDKindID("nosanitize"), - MDNode::get(C, None)); - - } - - if (dictionary.size()) { - - size_t memlen = 0, count = 0, offset = 0; - char * ptr; - - for (auto token : dictionary) { - - memlen += token.length(); - count++; - - } - - if (!be_quiet) - printf("AUTODICTIONARY: %lu string%s found\n", count, - count == 1 ? "" : "s"); - - if (count) { - - if ((ptr = (char *)malloc(memlen + count)) == NULL) { - - fprintf(stderr, "Error: malloc for %lu bytes failed!\n", - memlen + count); - exit(-1); - - } - - count = 0; - - for (auto token : dictionary) { - - if (offset + token.length() < 0xfffff0 && count < MAX_AUTO_EXTRAS) { - - ptr[offset++] = (uint8_t)token.length(); - memcpy(ptr + offset, token.c_str(), token.length()); - offset += token.length(); - count++; - - } - - } - - GlobalVariable *AFLDictionaryLen = - new GlobalVariable(M, Int32Ty, false, GlobalValue::ExternalLinkage, - 0, "__afl_dictionary_len"); - ConstantInt *const_len = ConstantInt::get(Int32Ty, offset); - StoreInst *StoreDictLen = IRB.CreateStore(const_len, AFLDictionaryLen); - StoreDictLen->setMetadata(M.getMDKindID("nosanitize"), - MDNode::get(C, None)); - - ArrayType *ArrayTy = ArrayType::get(IntegerType::get(C, 8), offset); - GlobalVariable *AFLInternalDictionary = new GlobalVariable( - M, ArrayTy, true, GlobalValue::ExternalLinkage, - ConstantDataArray::get(C, - *(new ArrayRef((char *)ptr, offset))), - "__afl_internal_dictionary"); - AFLInternalDictionary->setInitializer(ConstantDataArray::get( - C, *(new ArrayRef((char *)ptr, offset)))); - AFLInternalDictionary->setConstant(true); - - GlobalVariable *AFLDictionary = new GlobalVariable( - M, PointerType::get(Int8Ty, 0), false, GlobalValue::ExternalLinkage, - 0, "__afl_dictionary"); - - Value *AFLDictOff = IRB.CreateGEP(AFLInternalDictionary, Zero); - Value *AFLDictPtr = - IRB.CreatePointerCast(AFLDictOff, PointerType::get(Int8Ty, 0)); - StoreInst *StoreDict = IRB.CreateStore(AFLDictPtr, AFLDictionary); - StoreDict->setMetadata(M.getMDKindID("nosanitize"), - MDNode::get(C, None)); - - } - - } - - } - - /* Say something nice. */ - - if (!be_quiet) { - - if (!inst_blocks) - WARNF("No instrumentation targets found."); - else { - - char modeline[100]; - snprintf(modeline, sizeof(modeline), "%s%s%s%s%s", - getenv("AFL_HARDEN") ? "hardened" : "non-hardened", - getenv("AFL_USE_ASAN") ? ", ASAN" : "", - getenv("AFL_USE_MSAN") ? ", MSAN" : "", - getenv("AFL_USE_CFISAN") ? ", CFISAN" : "", - getenv("AFL_USE_UBSAN") ? ", UBSAN" : ""); - OKF("Instrumented %u locations with no collisions (on average %llu " - "collisions would be in afl-gcc/afl-clang-fast) (%s mode).", - inst_blocks, calculateCollisions(inst_blocks), modeline); - - } - - } - - return true; - -} - -char AFLLTOPass::ID = 0; - -static void registerAFLLTOPass(const PassManagerBuilder &, - legacy::PassManagerBase &PM) { - - PM.add(new AFLLTOPass()); - -} - -static RegisterPass X("afl-lto", "afl++ LTO instrumentation pass", - false, false); - -static RegisterStandardPasses RegisterAFLLTOPass( - PassManagerBuilder::EP_FullLinkTimeOptimizationLast, registerAFLLTOPass); - diff --git a/llvm_mode/afl-llvm-lto-instrumentlist.so.cc b/llvm_mode/afl-llvm-lto-instrumentlist.so.cc deleted file mode 100644 index a7331444..00000000 --- a/llvm_mode/afl-llvm-lto-instrumentlist.so.cc +++ /dev/null @@ -1,147 +0,0 @@ -/* - american fuzzy lop++ - LLVM-mode instrumentation pass - --------------------------------------------------- - - Written by Laszlo Szekeres and - Michal Zalewski - - LLVM integration design comes from Laszlo Szekeres. C bits copied-and-pasted - from afl-as.c are Michal's fault. - - Copyright 2015, 2016 Google Inc. All rights reserved. - Copyright 2019-2020 AFLplusplus Project. All rights reserved. - - Licensed under the Apache License, Version 2.0 (the "License"); - you may not use this file except in compliance with the License. - You may obtain a copy of the License at: - - http://www.apache.org/licenses/LICENSE-2.0 - - This library is plugged into LLVM when invoking clang through afl-clang-fast. - It tells the compiler to add code roughly equivalent to the bits discussed - in ../afl-as.h. - - */ - -#define AFL_LLVM_PASS - -#include "config.h" -#include "debug.h" - -#include -#include -#include - -#include -#include -#include -#include -#include - -#include "llvm/IR/DebugInfo.h" -#include "llvm/IR/BasicBlock.h" -#include "llvm/IR/IRBuilder.h" -#include "llvm/IR/LegacyPassManager.h" -#include "llvm/IR/Module.h" -#include "llvm/Support/Debug.h" -#include "llvm/Transforms/IPO/PassManagerBuilder.h" -#include "llvm/IR/CFG.h" - -#include "afl-llvm-common.h" - -using namespace llvm; - -namespace { - -class AFLcheckIfInstrument : public ModulePass { - - public: - static char ID; - AFLcheckIfInstrument() : ModulePass(ID) { - - if (getenv("AFL_DEBUG")) debug = 1; - - initInstrumentList(); - - } - - bool runOnModule(Module &M) override; - - // StringRef getPassName() const override { - - // return "American Fuzzy Lop Instrumentation"; - // } - - protected: - std::list myInstrumentList; - -}; - -} // namespace - -char AFLcheckIfInstrument::ID = 0; - -bool AFLcheckIfInstrument::runOnModule(Module &M) { - - /* Show a banner */ - - setvbuf(stdout, NULL, _IONBF, 0); - - if ((isatty(2) && !getenv("AFL_QUIET")) || getenv("AFL_DEBUG") != NULL) { - - SAYF(cCYA "afl-llvm-lto-instrumentlist" VERSION cRST - " by Marc \"vanHauser\" Heuse \n"); - - } else if (getenv("AFL_QUIET")) - - be_quiet = 1; - - for (auto &F : M) { - - if (F.size() < 1) continue; - - // fprintf(stderr, "F:%s\n", F.getName().str().c_str()); - - if (isInInstrumentList(&F)) { - - if (debug) - SAYF(cMGN "[D] " cRST "function %s is in the instrument file list\n", - F.getName().str().c_str()); - - } else { - - if (debug) - SAYF(cMGN "[D] " cRST - "function %s is NOT in the instrument file list\n", - F.getName().str().c_str()); - - auto & Ctx = F.getContext(); - AttributeList Attrs = F.getAttributes(); - AttrBuilder NewAttrs; - NewAttrs.addAttribute("skipinstrument"); - F.setAttributes( - Attrs.addAttributes(Ctx, AttributeList::FunctionIndex, NewAttrs)); - - } - - } - - return true; - -} - -static void registerAFLcheckIfInstrumentpass(const PassManagerBuilder &, - legacy::PassManagerBase &PM) { - - PM.add(new AFLcheckIfInstrument()); - -} - -static RegisterStandardPasses RegisterAFLcheckIfInstrumentpass( - PassManagerBuilder::EP_ModuleOptimizerEarly, - registerAFLcheckIfInstrumentpass); - -static RegisterStandardPasses RegisterAFLcheckIfInstrumentpass0( - PassManagerBuilder::EP_EnabledOnOptLevel0, - registerAFLcheckIfInstrumentpass); - diff --git a/llvm_mode/afl-llvm-pass.so.cc b/llvm_mode/afl-llvm-pass.so.cc deleted file mode 100644 index 8c8c987a..00000000 --- a/llvm_mode/afl-llvm-pass.so.cc +++ /dev/null @@ -1,654 +0,0 @@ -/* - american fuzzy lop++ - LLVM-mode instrumentation pass - --------------------------------------------------- - - Written by Laszlo Szekeres , - Adrian Herrera , - Michal Zalewski - - LLVM integration design comes from Laszlo Szekeres. C bits copied-and-pasted - from afl-as.c are Michal's fault. - - NGRAM previous location coverage comes from Adrian Herrera. - - Copyright 2015, 2016 Google Inc. All rights reserved. - Copyright 2019-2020 AFLplusplus Project. All rights reserved. - - Licensed under the Apache License, Version 2.0 (the "License"); - you may not use this file except in compliance with the License. - You may obtain a copy of the License at: - - http://www.apache.org/licenses/LICENSE-2.0 - - This library is plugged into LLVM when invoking clang through afl-clang-fast. - It tells the compiler to add code roughly equivalent to the bits discussed - in ../afl-as.h. - - */ - -#define AFL_LLVM_PASS - -#include "config.h" -#include "debug.h" -#include -#include -#include - -#include -#include -#include -#include - -#include "llvm/Config/llvm-config.h" -#if LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR < 5 -typedef long double max_align_t; -#endif - -#include "llvm/IR/IRBuilder.h" -#include "llvm/IR/LegacyPassManager.h" -#include "llvm/IR/BasicBlock.h" -#include "llvm/IR/Module.h" -#include "llvm/Support/Debug.h" -#include "llvm/Support/MathExtras.h" -#include "llvm/Transforms/IPO/PassManagerBuilder.h" - -#if LLVM_VERSION_MAJOR > 3 || \ - (LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR > 4) - #include "llvm/IR/DebugInfo.h" - #include "llvm/IR/CFG.h" -#else - #include "llvm/DebugInfo.h" - #include "llvm/Support/CFG.h" -#endif - -#include "afl-llvm-common.h" -#include "llvm-ngram-coverage.h" - -using namespace llvm; - -namespace { - -class AFLCoverage : public ModulePass { - - public: - static char ID; - AFLCoverage() : ModulePass(ID) { - - initInstrumentList(); - - } - - bool runOnModule(Module &M) override; - - protected: - uint32_t ngram_size = 0; - uint32_t map_size = MAP_SIZE; - uint32_t function_minimum_size = 1; - char * ctx_str = NULL, *skip_nozero = NULL; - -}; - -} // namespace - -char AFLCoverage::ID = 0; - -/* needed up to 3.9.0 */ -#if LLVM_VERSION_MAJOR == 3 && \ - (LLVM_VERSION_MINOR < 9 || \ - (LLVM_VERSION_MINOR == 9 && LLVM_VERSION_PATCH < 1)) -uint64_t PowerOf2Ceil(unsigned in) { - - uint64_t in64 = in - 1; - in64 |= (in64 >> 1); - in64 |= (in64 >> 2); - in64 |= (in64 >> 4); - in64 |= (in64 >> 8); - in64 |= (in64 >> 16); - in64 |= (in64 >> 32); - return in64 + 1; - -} - -#endif - -/* #if LLVM_VERSION_STRING >= "4.0.1" */ -#if LLVM_VERSION_MAJOR > 4 || \ - (LLVM_VERSION_MAJOR == 4 && LLVM_VERSION_PATCH >= 1) - #define AFL_HAVE_VECTOR_INTRINSICS 1 -#endif -bool AFLCoverage::runOnModule(Module &M) { - - LLVMContext &C = M.getContext(); - - IntegerType *Int8Ty = IntegerType::getInt8Ty(C); - IntegerType *Int32Ty = IntegerType::getInt32Ty(C); -#ifdef AFL_HAVE_VECTOR_INTRINSICS - IntegerType *IntLocTy = - IntegerType::getIntNTy(C, sizeof(PREV_LOC_T) * CHAR_BIT); -#endif - struct timeval tv; - struct timezone tz; - u32 rand_seed; - unsigned int cur_loc = 0; - - /* Setup random() so we get Actually Random(TM) outputs from AFL_R() */ - gettimeofday(&tv, &tz); - rand_seed = tv.tv_sec ^ tv.tv_usec ^ getpid(); - AFL_SR(rand_seed); - - /* Show a banner */ - - setvbuf(stdout, NULL, _IONBF, 0); - - if (getenv("AFL_DEBUG")) debug = 1; - - if ((isatty(2) && !getenv("AFL_QUIET")) || getenv("AFL_DEBUG") != NULL) { - - SAYF(cCYA "afl-llvm-pass" VERSION cRST - " by and \n"); - - } else - - be_quiet = 1; - - /* - char *ptr; - if ((ptr = getenv("AFL_MAP_SIZE")) || (ptr = getenv("AFL_MAPSIZE"))) { - - map_size = atoi(ptr); - if (map_size < 8 || map_size > (1 << 29)) - FATAL("illegal AFL_MAP_SIZE %u, must be between 2^3 and 2^30", - map_size); if (map_size % 8) map_size = (((map_size >> 3) + 1) << 3); - - } - - */ - - /* Decide instrumentation ratio */ - - char * inst_ratio_str = getenv("AFL_INST_RATIO"); - unsigned int inst_ratio = 100; - - if (inst_ratio_str) { - - if (sscanf(inst_ratio_str, "%u", &inst_ratio) != 1 || !inst_ratio || - inst_ratio > 100) - FATAL("Bad value of AFL_INST_RATIO (must be between 1 and 100)"); - - } - -#if LLVM_VERSION_MAJOR < 9 - char *neverZero_counters_str = getenv("AFL_LLVM_NOT_ZERO"); -#endif - skip_nozero = getenv("AFL_LLVM_SKIP_NEVERZERO"); - - unsigned PrevLocSize = 0; - - char *ngram_size_str = getenv("AFL_LLVM_NGRAM_SIZE"); - if (!ngram_size_str) ngram_size_str = getenv("AFL_NGRAM_SIZE"); - ctx_str = getenv("AFL_LLVM_CTX"); - -#ifdef AFL_HAVE_VECTOR_INTRINSICS - /* Decide previous location vector size (must be a power of two) */ - VectorType *PrevLocTy = NULL; - - if (ngram_size_str) - if (sscanf(ngram_size_str, "%u", &ngram_size) != 1 || ngram_size < 2 || - ngram_size > NGRAM_SIZE_MAX) - FATAL( - "Bad value of AFL_NGRAM_SIZE (must be between 2 and NGRAM_SIZE_MAX " - "(%u))", - NGRAM_SIZE_MAX); - - if (ngram_size == 1) ngram_size = 0; - if (ngram_size) - PrevLocSize = ngram_size - 1; - else -#else - if (ngram_size_str) - #ifndef LLVM_VERSION_PATCH - FATAL( - "Sorry, NGRAM branch coverage is not supported with llvm version " - "%d.%d.%d!", - LLVM_VERSION_MAJOR, LLVM_VERSION_MINOR, 0); - #else - FATAL( - "Sorry, NGRAM branch coverage is not supported with llvm version " - "%d.%d.%d!", - LLVM_VERSION_MAJOR, LLVM_VERSION_MINOR, LLVM_VERSION_PATCH); - #endif -#endif - PrevLocSize = 1; - -#ifdef AFL_HAVE_VECTOR_INTRINSICS - int PrevLocVecSize = PowerOf2Ceil(PrevLocSize); - if (ngram_size) - PrevLocTy = VectorType::get(IntLocTy, PrevLocVecSize - #if LLVM_VERSION_MAJOR >= 12 - , - false - #endif - ); -#endif - - /* Get globals for the SHM region and the previous location. Note that - __afl_prev_loc is thread-local. */ - - GlobalVariable *AFLMapPtr = - new GlobalVariable(M, PointerType::get(Int8Ty, 0), false, - GlobalValue::ExternalLinkage, 0, "__afl_area_ptr"); - GlobalVariable *AFLPrevLoc; - GlobalVariable *AFLContext = NULL; - - if (ctx_str) -#ifdef __ANDROID__ - AFLContext = new GlobalVariable( - M, Int32Ty, false, GlobalValue::ExternalLinkage, 0, "__afl_prev_ctx"); -#else - AFLContext = new GlobalVariable( - M, Int32Ty, false, GlobalValue::ExternalLinkage, 0, "__afl_prev_ctx", 0, - GlobalVariable::GeneralDynamicTLSModel, 0, false); -#endif - -#ifdef AFL_HAVE_VECTOR_INTRINSICS - if (ngram_size) - #ifdef __ANDROID__ - AFLPrevLoc = new GlobalVariable( - M, PrevLocTy, /* isConstant */ false, GlobalValue::ExternalLinkage, - /* Initializer */ nullptr, "__afl_prev_loc"); - #else - AFLPrevLoc = new GlobalVariable( - M, PrevLocTy, /* isConstant */ false, GlobalValue::ExternalLinkage, - /* Initializer */ nullptr, "__afl_prev_loc", - /* InsertBefore */ nullptr, GlobalVariable::GeneralDynamicTLSModel, - /* AddressSpace */ 0, /* IsExternallyInitialized */ false); - #endif - else -#endif -#ifdef __ANDROID__ - AFLPrevLoc = new GlobalVariable( - M, Int32Ty, false, GlobalValue::ExternalLinkage, 0, "__afl_prev_loc"); -#else - AFLPrevLoc = new GlobalVariable( - M, Int32Ty, false, GlobalValue::ExternalLinkage, 0, "__afl_prev_loc", 0, - GlobalVariable::GeneralDynamicTLSModel, 0, false); -#endif - -#ifdef AFL_HAVE_VECTOR_INTRINSICS - /* Create the vector shuffle mask for updating the previous block history. - Note that the first element of the vector will store cur_loc, so just set - it to undef to allow the optimizer to do its thing. */ - - SmallVector PrevLocShuffle = {UndefValue::get(Int32Ty)}; - - for (unsigned I = 0; I < PrevLocSize - 1; ++I) - PrevLocShuffle.push_back(ConstantInt::get(Int32Ty, I)); - - for (int I = PrevLocSize; I < PrevLocVecSize; ++I) - PrevLocShuffle.push_back(ConstantInt::get(Int32Ty, PrevLocSize)); - - Constant *PrevLocShuffleMask = ConstantVector::get(PrevLocShuffle); -#endif - - // other constants we need - ConstantInt *Zero = ConstantInt::get(Int8Ty, 0); - ConstantInt *One = ConstantInt::get(Int8Ty, 1); - - LoadInst *PrevCtx = NULL; // CTX sensitive coverage - - /* Instrument all the things! */ - - int inst_blocks = 0; - scanForDangerousFunctions(&M); - - for (auto &F : M) { - - int has_calls = 0; - if (debug) - fprintf(stderr, "FUNCTION: %s (%zu)\n", F.getName().str().c_str(), - F.size()); - - if (!isInInstrumentList(&F)) continue; - - if (F.size() < function_minimum_size) continue; - - for (auto &BB : F) { - - BasicBlock::iterator IP = BB.getFirstInsertionPt(); - IRBuilder<> IRB(&(*IP)); - - // Context sensitive coverage - if (ctx_str && &BB == &F.getEntryBlock()) { - - // load the context ID of the previous function and write to to a local - // variable on the stack - PrevCtx = IRB.CreateLoad(AFLContext); - PrevCtx->setMetadata(M.getMDKindID("nosanitize"), MDNode::get(C, None)); - - // does the function have calls? and is any of the calls larger than one - // basic block? - for (auto &BB : F) { - - if (has_calls) break; - for (auto &IN : BB) { - - CallInst *callInst = nullptr; - if ((callInst = dyn_cast(&IN))) { - - Function *Callee = callInst->getCalledFunction(); - if (!Callee || Callee->size() < function_minimum_size) - continue; - else { - - has_calls = 1; - break; - - } - - } - - } - - } - - // if yes we store a context ID for this function in the global var - if (has_calls) { - - ConstantInt *NewCtx = ConstantInt::get(Int32Ty, AFL_R(map_size)); - StoreInst * StoreCtx = IRB.CreateStore(NewCtx, AFLContext); - StoreCtx->setMetadata(M.getMDKindID("nosanitize"), - MDNode::get(C, None)); - - } - - } - - if (AFL_R(100) >= inst_ratio) continue; - - /* Make up cur_loc */ - - // cur_loc++; - cur_loc = AFL_R(map_size); - -/* There is a problem with Ubuntu 18.04 and llvm 6.0 (see issue #63). - The inline function successors() is not inlined and also not found at runtime - :-( As I am unable to detect Ubuntu18.04 heree, the next best thing is to - disable this optional optimization for LLVM 6.0.0 and Linux */ -#if !(LLVM_VERSION_MAJOR == 6 && LLVM_VERSION_MINOR == 0) || !defined __linux__ - // only instrument if this basic block is the destination of a previous - // basic block that has multiple successors - // this gets rid of ~5-10% of instrumentations that are unnecessary - // result: a little more speed and less map pollution - int more_than_one = -1; - // fprintf(stderr, "BB %u: ", cur_loc); - for (pred_iterator PI = pred_begin(&BB), E = pred_end(&BB); PI != E; - ++PI) { - - BasicBlock *Pred = *PI; - - int count = 0; - if (more_than_one == -1) more_than_one = 0; - // fprintf(stderr, " %p=>", Pred); - - for (succ_iterator SI = succ_begin(Pred), E = succ_end(Pred); SI != E; - ++SI) { - - BasicBlock *Succ = *SI; - - // if (count > 0) - // fprintf(stderr, "|"); - if (Succ != NULL) count++; - // fprintf(stderr, "%p", Succ); - - } - - if (count > 1) more_than_one = 1; - - } - - // fprintf(stderr, " == %d\n", more_than_one); - if (F.size() > 1 && more_than_one != 1) { - - // in CTX mode we have to restore the original context for the caller - - // she might be calling other functions which need the correct CTX - if (ctx_str && has_calls) { - - Instruction *Inst = BB.getTerminator(); - if (isa(Inst) || isa(Inst)) { - - IRBuilder<> Post_IRB(Inst); - StoreInst * RestoreCtx = Post_IRB.CreateStore(PrevCtx, AFLContext); - RestoreCtx->setMetadata(M.getMDKindID("nosanitize"), - MDNode::get(C, None)); - - } - - } - - continue; - - } - -#endif - - ConstantInt *CurLoc; - -#ifdef AFL_HAVE_VECTOR_INTRINSICS - if (ngram_size) - CurLoc = ConstantInt::get(IntLocTy, cur_loc); - else -#endif - CurLoc = ConstantInt::get(Int32Ty, cur_loc); - - /* Load prev_loc */ - - LoadInst *PrevLoc = IRB.CreateLoad(AFLPrevLoc); - PrevLoc->setMetadata(M.getMDKindID("nosanitize"), MDNode::get(C, None)); - Value *PrevLocTrans; - -#ifdef AFL_HAVE_VECTOR_INTRINSICS - /* "For efficiency, we propose to hash the tuple as a key into the - hit_count map as (prev_block_trans << 1) ^ curr_block_trans, where - prev_block_trans = (block_trans_1 ^ ... ^ block_trans_(n-1)" */ - - if (ngram_size) - PrevLocTrans = - IRB.CreateZExt(IRB.CreateXorReduce(PrevLoc), IRB.getInt32Ty()); - else -#endif - PrevLocTrans = PrevLoc; - - if (ctx_str) - PrevLocTrans = - IRB.CreateZExt(IRB.CreateXor(PrevLocTrans, PrevCtx), Int32Ty); - else - PrevLocTrans = IRB.CreateZExt(PrevLocTrans, IRB.getInt32Ty()); - - /* Load SHM pointer */ - - LoadInst *MapPtr = IRB.CreateLoad(AFLMapPtr); - MapPtr->setMetadata(M.getMDKindID("nosanitize"), MDNode::get(C, None)); - - Value *MapPtrIdx; -#ifdef AFL_HAVE_VECTOR_INTRINSICS - if (ngram_size) - MapPtrIdx = IRB.CreateGEP( - MapPtr, - IRB.CreateZExt( - IRB.CreateXor(PrevLocTrans, IRB.CreateZExt(CurLoc, Int32Ty)), - Int32Ty)); - else -#endif - MapPtrIdx = IRB.CreateGEP(MapPtr, IRB.CreateXor(PrevLocTrans, CurLoc)); - - /* Update bitmap */ - - LoadInst *Counter = IRB.CreateLoad(MapPtrIdx); - Counter->setMetadata(M.getMDKindID("nosanitize"), MDNode::get(C, None)); - - Value *Incr = IRB.CreateAdd(Counter, One); - -#if LLVM_VERSION_MAJOR < 9 - if (neverZero_counters_str != - NULL) { // with llvm 9 we make this the default as the bug in llvm is - // then fixed -#else - if (!skip_nozero) { - -#endif - /* hexcoder: Realize a counter that skips zero during overflow. - * Once this counter reaches its maximum value, it next increments to 1 - * - * Instead of - * Counter + 1 -> Counter - * we inject now this - * Counter + 1 -> {Counter, OverflowFlag} - * Counter + OverflowFlag -> Counter - */ - - auto cf = IRB.CreateICmpEQ(Incr, Zero); - auto carry = IRB.CreateZExt(cf, Int8Ty); - Incr = IRB.CreateAdd(Incr, carry); - - } - - IRB.CreateStore(Incr, MapPtrIdx) - ->setMetadata(M.getMDKindID("nosanitize"), MDNode::get(C, None)); - - /* Update prev_loc history vector (by placing cur_loc at the head of the - vector and shuffle the other elements back by one) */ - - StoreInst *Store; - -#ifdef AFL_HAVE_VECTOR_INTRINSICS - if (ngram_size) { - - Value *ShuffledPrevLoc = IRB.CreateShuffleVector( - PrevLoc, UndefValue::get(PrevLocTy), PrevLocShuffleMask); - Value *UpdatedPrevLoc = IRB.CreateInsertElement( - ShuffledPrevLoc, IRB.CreateLShr(CurLoc, (uint64_t)1), (uint64_t)0); - - Store = IRB.CreateStore(UpdatedPrevLoc, AFLPrevLoc); - Store->setMetadata(M.getMDKindID("nosanitize"), MDNode::get(C, None)); - - } else - -#endif - { - - Store = IRB.CreateStore(ConstantInt::get(Int32Ty, cur_loc >> 1), - AFLPrevLoc); - - } - - // in CTX mode we have to restore the original context for the caller - - // she might be calling other functions which need the correct CTX. - // Currently this is only needed for the Ubuntu clang-6.0 bug - if (ctx_str && has_calls) { - - Instruction *Inst = BB.getTerminator(); - if (isa(Inst) || isa(Inst)) { - - IRBuilder<> Post_IRB(Inst); - StoreInst * RestoreCtx = Post_IRB.CreateStore(PrevCtx, AFLContext); - RestoreCtx->setMetadata(M.getMDKindID("nosanitize"), - MDNode::get(C, None)); - - } - - } - - inst_blocks++; - - } - - } - - /* - // This is currently disabled because we not only need to create/insert a - // function (easy), but also add it as a constructor with an ID < 5 - - if (getenv("AFL_LLVM_DONTWRITEID") == NULL) { - - // yes we could create our own function, insert it into ctors ... - // but this would be a pain in the butt ... so we use afl-llvm-rt.o - - Function *f = ... - - if (!f) { - - fprintf(stderr, - "Error: init function could not be created (this should not - happen)\n"); exit(-1); - - } - - ... constructor for f = 4 - - BasicBlock *bb = &f->getEntryBlock(); - if (!bb) { - - fprintf(stderr, - "Error: init function does not have an EntryBlock (this should - not happen)\n"); exit(-1); - - } - - BasicBlock::iterator IP = bb->getFirstInsertionPt(); - IRBuilder<> IRB(&(*IP)); - - if (map_size <= 0x800000) { - - GlobalVariable *AFLFinalLoc = new GlobalVariable( - M, Int32Ty, true, GlobalValue::ExternalLinkage, 0, - "__afl_final_loc"); - ConstantInt *const_loc = ConstantInt::get(Int32Ty, map_size); - StoreInst * StoreFinalLoc = IRB.CreateStore(const_loc, AFLFinalLoc); - StoreFinalLoc->setMetadata(M.getMDKindID("nosanitize"), - MDNode::get(C, None)); - - } - - } - - */ - - /* Say something nice. */ - - if (!be_quiet) { - - if (!inst_blocks) - WARNF("No instrumentation targets found."); - else { - - char modeline[100]; - snprintf(modeline, sizeof(modeline), "%s%s%s%s%s", - getenv("AFL_HARDEN") ? "hardened" : "non-hardened", - getenv("AFL_USE_ASAN") ? ", ASAN" : "", - getenv("AFL_USE_MSAN") ? ", MSAN" : "", - getenv("AFL_USE_CFISAN") ? ", CFISAN" : "", - getenv("AFL_USE_UBSAN") ? ", UBSAN" : ""); - OKF("Instrumented %u locations (%s mode, ratio %u%%).", inst_blocks, - modeline, inst_ratio); - - } - - } - - return true; - -} - -static void registerAFLPass(const PassManagerBuilder &, - legacy::PassManagerBase &PM) { - - PM.add(new AFLCoverage()); - -} - -static RegisterStandardPasses RegisterAFLPass( - PassManagerBuilder::EP_OptimizerLast, registerAFLPass); - -static RegisterStandardPasses RegisterAFLPass0( - PassManagerBuilder::EP_EnabledOnOptLevel0, registerAFLPass); - diff --git a/llvm_mode/afl-llvm-rt-lto.o.c b/llvm_mode/afl-llvm-rt-lto.o.c deleted file mode 100644 index e53785ff..00000000 --- a/llvm_mode/afl-llvm-rt-lto.o.c +++ /dev/null @@ -1,27 +0,0 @@ -/* - american fuzzy lop++ - LLVM instrumentation bootstrap - ----------------------------------------------------- - - Licensed under the Apache License, Version 2.0 (the "License"); - you may not use this file except in compliance with the License. - You may obtain a copy of the License at: - - http://www.apache.org/licenses/LICENSE-2.0 - -*/ - -#include -#include - -// to prevent the function from being removed -unsigned char __afl_lto_mode = 0; - -/* Proper initialization routine. */ - -__attribute__((constructor(0))) void __afl_auto_init_globals(void) { - - if (getenv("AFL_DEBUG")) fprintf(stderr, "[__afl_auto_init_globals]\n"); - __afl_lto_mode = 1; - -} - diff --git a/llvm_mode/afl-llvm-rt.o.c b/llvm_mode/afl-llvm-rt.o.c deleted file mode 100644 index bdafbe0b..00000000 --- a/llvm_mode/afl-llvm-rt.o.c +++ /dev/null @@ -1,1244 +0,0 @@ -/* - american fuzzy lop++ - LLVM instrumentation bootstrap - --------------------------------------------------- - - Written by Laszlo Szekeres and - Michal Zalewski - - LLVM integration design comes from Laszlo Szekeres. - - Copyright 2015, 2016 Google Inc. All rights reserved. - Copyright 2019-2020 AFLplusplus Project. All rights reserved. - - Licensed under the Apache License, Version 2.0 (the "License"); - you may not use this file except in compliance with the License. - You may obtain a copy of the License at: - - http://www.apache.org/licenses/LICENSE-2.0 - - This code is the rewrite of afl-as.h's main_payload. - -*/ - -#ifdef __ANDROID__ - #include "android-ashmem.h" -#endif -#include "config.h" -#include "types.h" -#include "cmplog.h" -#include "llvm-ngram-coverage.h" - -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include - -#include -#include -#include -#include - -#include "llvm/Config/llvm-config.h" - -#ifdef __linux__ - #include "snapshot-inl.h" -#endif - -/* This is a somewhat ugly hack for the experimental 'trace-pc-guard' mode. - Basically, we need to make sure that the forkserver is initialized after - the LLVM-generated runtime initialization pass, not before. */ - -#ifndef MAP_FIXED_NOREPLACE - #ifdef MAP_EXCL - #define MAP_FIXED_NOREPLACE MAP_EXCL | MAP_FIXED - #else - #define MAP_FIXED_NOREPLACE MAP_FIXED - #endif -#endif - -#define CTOR_PRIO 3 - -#include -#include - -/* Globals needed by the injected instrumentation. The __afl_area_initial region - is used for instrumentation output before __afl_map_shm() has a chance to - run. It will end up as .comm, so it shouldn't be too wasteful. */ - -#if MAP_SIZE <= 65536 - #define MAP_INITIAL_SIZE 256000 -#else - #define MAP_INITIAL_SIZE MAP_SIZE -#endif - -u8 __afl_area_initial[MAP_INITIAL_SIZE]; -u8 * __afl_area_ptr = __afl_area_initial; -u8 * __afl_dictionary; -u8 * __afl_fuzz_ptr; -u32 __afl_fuzz_len_dummy; -u32 *__afl_fuzz_len = &__afl_fuzz_len_dummy; - -u32 __afl_final_loc; -u32 __afl_map_size = MAP_SIZE; -u32 __afl_dictionary_len; -u64 __afl_map_addr; - -#ifdef __ANDROID__ -PREV_LOC_T __afl_prev_loc[NGRAM_SIZE_MAX]; -u32 __afl_prev_ctx; -u32 __afl_cmp_counter; -#else -__thread PREV_LOC_T __afl_prev_loc[NGRAM_SIZE_MAX]; -__thread u32 __afl_prev_ctx; -__thread u32 __afl_cmp_counter; -#endif - -int __afl_sharedmem_fuzzing __attribute__((weak)); - -struct cmp_map *__afl_cmp_map; - -/* Running in persistent mode? */ - -static u8 is_persistent; - -/* Are we in sancov mode? */ - -static u8 _is_sancov; - -/* Error reporting to forkserver controller */ - -void send_forkserver_error(int error) { - - u32 status; - if (!error || error > 0xffff) return; - status = (FS_OPT_ERROR | FS_OPT_SET_ERROR(error)); - if (write(FORKSRV_FD + 1, (char *)&status, 4) != 4) return; - -} - -/* SHM fuzzing setup. */ - -static void __afl_map_shm_fuzz() { - - char *id_str = getenv(SHM_FUZZ_ENV_VAR); - - if (id_str) { - - u8 *map = NULL; - -#ifdef USEMMAP - const char * shm_file_path = id_str; - int shm_fd = -1; - unsigned char *shm_base = NULL; - - /* create the shared memory segment as if it was a file */ - shm_fd = shm_open(shm_file_path, O_RDWR, 0600); - if (shm_fd == -1) { - - fprintf(stderr, "shm_open() failed for fuzz\n"); - send_forkserver_error(FS_ERROR_SHM_OPEN); - exit(1); - - } - - map = - (u8 *)mmap(0, MAX_FILE + sizeof(u32), PROT_READ, MAP_SHARED, shm_fd, 0); - -#else - u32 shm_id = atoi(id_str); - map = (u8 *)shmat(shm_id, NULL, 0); - -#endif - - /* Whooooops. */ - - if (!map || map == (void *)-1) { - - perror("Could not access fuzzign shared memory"); - exit(1); - - } - - __afl_fuzz_len = (u32 *)map; - __afl_fuzz_ptr = map + sizeof(u32); - - if (getenv("AFL_DEBUG")) { - - fprintf(stderr, "DEBUG: successfully got fuzzing shared memory\n"); - - } - - } else { - - fprintf(stderr, "Error: variable for fuzzing shared memory is not set\n"); - exit(1); - - } - -} - -/* SHM setup. */ - -static void __afl_map_shm(void) { - - // we we are not running in afl ensure the map exists - if (!__afl_area_ptr) { __afl_area_ptr = __afl_area_initial; } - - char *id_str = getenv(SHM_ENV_VAR); - - if (__afl_final_loc) { - - if (__afl_final_loc % 8) - __afl_final_loc = (((__afl_final_loc + 7) >> 3) << 3); - __afl_map_size = __afl_final_loc; - - if (__afl_final_loc > MAP_SIZE) { - - char *ptr; - u32 val = 0; - if ((ptr = getenv("AFL_MAP_SIZE")) != NULL) val = atoi(ptr); - if (val < __afl_final_loc) { - - if (__afl_final_loc > FS_OPT_MAX_MAPSIZE) { - - if (!getenv("AFL_QUIET")) - fprintf(stderr, - "Error: AFL++ tools *require* to set AFL_MAP_SIZE to %u " - "to be able to run this instrumented program!\n", - __afl_final_loc); - - if (id_str) { - - send_forkserver_error(FS_ERROR_MAP_SIZE); - exit(-1); - - } - - } else { - - if (!getenv("AFL_QUIET")) - fprintf(stderr, - "Warning: AFL++ tools will need to set AFL_MAP_SIZE to %u " - "to be able to run this instrumented program!\n", - __afl_final_loc); - - } - - } - - } - - } - - /* If we're running under AFL, attach to the appropriate region, replacing the - early-stage __afl_area_initial region that is needed to allow some really - hacky .init code to work correctly in projects such as OpenSSL. */ - - if (getenv("AFL_DEBUG")) - fprintf(stderr, - "DEBUG: id_str %s, __afl_area_ptr %p, __afl_area_initial %p, " - "__afl_map_addr 0x%llx, MAP_SIZE %u, __afl_final_loc %u, " - "max_size_forkserver %u/0x%x\n", - id_str == NULL ? "" : id_str, __afl_area_ptr, - __afl_area_initial, __afl_map_addr, MAP_SIZE, __afl_final_loc, - FS_OPT_MAX_MAPSIZE, FS_OPT_MAX_MAPSIZE); - - if (id_str) { - - if (__afl_area_ptr && __afl_area_ptr != __afl_area_initial) { - - if (__afl_map_addr) - munmap((void *)__afl_map_addr, __afl_final_loc); - else - free(__afl_area_ptr); - __afl_area_ptr = __afl_area_initial; - - } - -#ifdef USEMMAP - const char * shm_file_path = id_str; - int shm_fd = -1; - unsigned char *shm_base = NULL; - - /* create the shared memory segment as if it was a file */ - shm_fd = shm_open(shm_file_path, O_RDWR, 0600); - if (shm_fd == -1) { - - fprintf(stderr, "shm_open() failed\n"); - send_forkserver_error(FS_ERROR_SHM_OPEN); - exit(1); - - } - - /* map the shared memory segment to the address space of the process */ - if (__afl_map_addr) { - - shm_base = - mmap((void *)__afl_map_addr, __afl_map_size, PROT_READ | PROT_WRITE, - MAP_FIXED_NOREPLACE | MAP_SHARED, shm_fd, 0); - - } else { - - shm_base = mmap(0, __afl_map_size, PROT_READ | PROT_WRITE, MAP_SHARED, - shm_fd, 0); - - } - - if (shm_base == MAP_FAILED) { - - close(shm_fd); - shm_fd = -1; - - fprintf(stderr, "mmap() failed\n"); - if (__afl_map_addr) - send_forkserver_error(FS_ERROR_MAP_ADDR); - else - send_forkserver_error(FS_ERROR_MMAP); - exit(2); - - } - - __afl_area_ptr = shm_base; -#else - u32 shm_id = atoi(id_str); - - __afl_area_ptr = shmat(shm_id, (void *)__afl_map_addr, 0); - -#endif - - /* Whooooops. */ - - if (__afl_area_ptr == (void *)-1) { - - if (__afl_map_addr) - send_forkserver_error(FS_ERROR_MAP_ADDR); - else - send_forkserver_error(FS_ERROR_SHMAT); - _exit(1); - - } - - /* Write something into the bitmap so that even with low AFL_INST_RATIO, - our parent doesn't give up on us. */ - - __afl_area_ptr[0] = 1; - - } else if ((!__afl_area_ptr || __afl_area_ptr == __afl_area_initial) && - - __afl_map_addr) { - - __afl_area_ptr = - mmap((void *)__afl_map_addr, __afl_map_size, PROT_READ | PROT_WRITE, - MAP_FIXED_NOREPLACE | MAP_SHARED | MAP_ANONYMOUS, -1, 0); - - if (__afl_area_ptr == MAP_FAILED) { - - fprintf(stderr, "can not acquire mmap for address %p\n", - (void *)__afl_map_addr); - exit(1); - - } - - } else if (_is_sancov && __afl_area_ptr != __afl_area_initial) { - - free(__afl_area_ptr); - __afl_area_ptr = NULL; - if (__afl_final_loc > MAP_INITIAL_SIZE) - __afl_area_ptr = malloc(__afl_final_loc); - if (!__afl_area_ptr) __afl_area_ptr = __afl_area_initial; - - } - - id_str = getenv(CMPLOG_SHM_ENV_VAR); - - if (getenv("AFL_DEBUG")) { - - fprintf(stderr, "DEBUG: cmplog id_str %s\n", - id_str == NULL ? "" : id_str); - - } - - if (id_str) { - -#ifdef USEMMAP - const char * shm_file_path = id_str; - int shm_fd = -1; - unsigned char *shm_base = NULL; - - /* create the shared memory segment as if it was a file */ - shm_fd = shm_open(shm_file_path, O_RDWR, 0600); - if (shm_fd == -1) { - - fprintf(stderr, "shm_open() failed\n"); - exit(1); - - } - - /* map the shared memory segment to the address space of the process */ - shm_base = mmap(0, sizeof(struct cmp_map), PROT_READ | PROT_WRITE, - MAP_SHARED, shm_fd, 0); - if (shm_base == MAP_FAILED) { - - close(shm_fd); - shm_fd = -1; - - fprintf(stderr, "mmap() failed\n"); - exit(2); - - } - - __afl_cmp_map = shm_base; -#else - u32 shm_id = atoi(id_str); - - __afl_cmp_map = shmat(shm_id, NULL, 0); -#endif - - if (__afl_cmp_map == (void *)-1) _exit(1); - - } - -} - -#ifdef __linux__ -static void __afl_start_snapshots(void) { - - static u8 tmp[4] = {0, 0, 0, 0}; - s32 child_pid; - u32 status = 0; - u32 already_read_first = 0; - u32 was_killed; - - u8 child_stopped = 0; - - void (*old_sigchld_handler)(int) = 0; // = signal(SIGCHLD, SIG_DFL); - - /* Phone home and tell the parent that we're OK. If parent isn't there, - assume we're not running in forkserver mode and just execute program. */ - - status |= (FS_OPT_ENABLED | FS_OPT_SNAPSHOT); - if (__afl_sharedmem_fuzzing != 0) status |= FS_OPT_SHDMEM_FUZZ; - if (__afl_map_size <= FS_OPT_MAX_MAPSIZE) - status |= (FS_OPT_SET_MAPSIZE(__afl_map_size) | FS_OPT_MAPSIZE); - if (__afl_dictionary_len && __afl_dictionary) status |= FS_OPT_AUTODICT; - memcpy(tmp, &status, 4); - - if (write(FORKSRV_FD + 1, tmp, 4) != 4) return; - - if (__afl_sharedmem_fuzzing || (__afl_dictionary_len && __afl_dictionary)) { - - if (read(FORKSRV_FD, &was_killed, 4) != 4) _exit(1); - - if (getenv("AFL_DEBUG")) { - - fprintf(stderr, "target forkserver recv: %08x\n", was_killed); - - } - - if ((was_killed & (FS_OPT_ENABLED | FS_OPT_SHDMEM_FUZZ)) == - (FS_OPT_ENABLED | FS_OPT_SHDMEM_FUZZ)) { - - __afl_map_shm_fuzz(); - - } - - if ((was_killed & (FS_OPT_ENABLED | FS_OPT_AUTODICT)) == - (FS_OPT_ENABLED | FS_OPT_AUTODICT)) { - - // great lets pass the dictionary through the forkserver FD - u32 len = __afl_dictionary_len, offset = 0; - s32 ret; - - if (write(FORKSRV_FD + 1, &len, 4) != 4) { - - write(2, "Error: could not send dictionary len\n", - strlen("Error: could not send dictionary len\n")); - _exit(1); - - } - - while (len != 0) { - - ret = write(FORKSRV_FD + 1, __afl_dictionary + offset, len); - - if (ret < 1) { - - write(2, "Error: could not send dictionary\n", - strlen("Error: could not send dictionary\n")); - _exit(1); - - } - - len -= ret; - offset += ret; - - } - - } else { - - // uh this forkserver does not understand extended option passing - // or does not want the dictionary - if (!__afl_fuzz_ptr) already_read_first = 1; - - } - - } - - while (1) { - - int status; - - if (already_read_first) { - - already_read_first = 0; - - } else { - - /* Wait for parent by reading from the pipe. Abort if read fails. */ - if (read(FORKSRV_FD, &was_killed, 4) != 4) _exit(1); - - } - - #ifdef _AFL_DOCUMENT_MUTATIONS - if (__afl_fuzz_ptr) { - - static uint32_t counter = 0; - char fn[32]; - sprintf(fn, "%09u:forkserver", counter); - s32 fd_doc = open(fn, O_WRONLY | O_CREAT | O_TRUNC, 0600); - if (fd_doc >= 0) { - - if (write(fd_doc, __afl_fuzz_ptr, *__afl_fuzz_len) != *__afl_fuzz_len) { - - fprintf(stderr, "write of mutation file failed: %s\n", fn); - unlink(fn); - - } - - close(fd_doc); - - } - - counter++; - - } - - #endif - - /* If we stopped the child in persistent mode, but there was a race - condition and afl-fuzz already issued SIGKILL, write off the old - process. */ - - if (child_stopped && was_killed) { - - child_stopped = 0; - if (waitpid(child_pid, &status, 0) < 0) _exit(1); - - } - - if (!child_stopped) { - - /* Once woken up, create a clone of our process. */ - - child_pid = fork(); - if (child_pid < 0) _exit(1); - - /* In child process: close fds, resume execution. */ - - if (!child_pid) { - - //(void)nice(-20); // does not seem to improve - - signal(SIGCHLD, old_sigchld_handler); - - close(FORKSRV_FD); - close(FORKSRV_FD + 1); - - if (!afl_snapshot_take(AFL_SNAPSHOT_MMAP | AFL_SNAPSHOT_FDS | - AFL_SNAPSHOT_REGS | AFL_SNAPSHOT_EXIT)) { - - raise(SIGSTOP); - - } - - __afl_area_ptr[0] = 1; - memset(__afl_prev_loc, 0, NGRAM_SIZE_MAX * sizeof(PREV_LOC_T)); - - return; - - } - - } else { - - /* Special handling for persistent mode: if the child is alive but - currently stopped, simply restart it with SIGCONT. */ - - kill(child_pid, SIGCONT); - child_stopped = 0; - - } - - /* In parent process: write PID to pipe, then wait for child. */ - - if (write(FORKSRV_FD + 1, &child_pid, 4) != 4) _exit(1); - - if (waitpid(child_pid, &status, WUNTRACED) < 0) _exit(1); - - /* In persistent mode, the child stops itself with SIGSTOP to indicate - a successful run. In this case, we want to wake it up without forking - again. */ - - if (WIFSTOPPED(status)) child_stopped = 1; - - /* Relay wait status to pipe, then loop back. */ - - if (write(FORKSRV_FD + 1, &status, 4) != 4) _exit(1); - - } - -} - -#endif - -/* Fork server logic. */ - -static void __afl_start_forkserver(void) { - -#ifdef __linux__ - if (/*!is_persistent &&*/ !__afl_cmp_map && !getenv("AFL_NO_SNAPSHOT") && - afl_snapshot_init() >= 0) { - - __afl_start_snapshots(); - return; - - } - -#endif - - u8 tmp[4] = {0, 0, 0, 0}; - s32 child_pid; - u32 status = 0; - u32 already_read_first = 0; - u32 was_killed; - - u8 child_stopped = 0; - - void (*old_sigchld_handler)(int) = 0; // = signal(SIGCHLD, SIG_DFL); - - if (__afl_map_size <= FS_OPT_MAX_MAPSIZE) - status |= (FS_OPT_SET_MAPSIZE(__afl_map_size) | FS_OPT_MAPSIZE); - if (__afl_dictionary_len && __afl_dictionary) status |= FS_OPT_AUTODICT; - if (__afl_sharedmem_fuzzing != 0) status |= FS_OPT_SHDMEM_FUZZ; - if (status) status |= (FS_OPT_ENABLED); - memcpy(tmp, &status, 4); - - /* Phone home and tell the parent that we're OK. If parent isn't there, - assume we're not running in forkserver mode and just execute program. */ - - if (write(FORKSRV_FD + 1, tmp, 4) != 4) return; - - if (__afl_sharedmem_fuzzing || (__afl_dictionary_len && __afl_dictionary)) { - - if (read(FORKSRV_FD, &was_killed, 4) != 4) _exit(1); - - if (getenv("AFL_DEBUG")) { - - fprintf(stderr, "target forkserver recv: %08x\n", was_killed); - - } - - if ((was_killed & (FS_OPT_ENABLED | FS_OPT_SHDMEM_FUZZ)) == - (FS_OPT_ENABLED | FS_OPT_SHDMEM_FUZZ)) { - - __afl_map_shm_fuzz(); - - } - - if ((was_killed & (FS_OPT_ENABLED | FS_OPT_AUTODICT)) == - (FS_OPT_ENABLED | FS_OPT_AUTODICT)) { - - // great lets pass the dictionary through the forkserver FD - u32 len = __afl_dictionary_len, offset = 0; - s32 ret; - - if (write(FORKSRV_FD + 1, &len, 4) != 4) { - - write(2, "Error: could not send dictionary len\n", - strlen("Error: could not send dictionary len\n")); - _exit(1); - - } - - while (len != 0) { - - ret = write(FORKSRV_FD + 1, __afl_dictionary + offset, len); - - if (ret < 1) { - - write(2, "Error: could not send dictionary\n", - strlen("Error: could not send dictionary\n")); - _exit(1); - - } - - len -= ret; - offset += ret; - - } - - } else { - - // uh this forkserver does not understand extended option passing - // or does not want the dictionary - if (!__afl_fuzz_ptr) already_read_first = 1; - - } - - } - - while (1) { - - int status; - - /* Wait for parent by reading from the pipe. Abort if read fails. */ - - if (already_read_first) { - - already_read_first = 0; - - } else { - - if (read(FORKSRV_FD, &was_killed, 4) != 4) _exit(1); - - } - -#ifdef _AFL_DOCUMENT_MUTATIONS - if (__afl_fuzz_ptr) { - - static uint32_t counter = 0; - char fn[32]; - sprintf(fn, "%09u:forkserver", counter); - s32 fd_doc = open(fn, O_WRONLY | O_CREAT | O_TRUNC, 0600); - if (fd_doc >= 0) { - - if (write(fd_doc, __afl_fuzz_ptr, *__afl_fuzz_len) != *__afl_fuzz_len) { - - fprintf(stderr, "write of mutation file failed: %s\n", fn); - unlink(fn); - - } - - close(fd_doc); - - } - - counter++; - - } - -#endif - - /* If we stopped the child in persistent mode, but there was a race - condition and afl-fuzz already issued SIGKILL, write off the old - process. */ - - if (child_stopped && was_killed) { - - child_stopped = 0; - if (waitpid(child_pid, &status, 0) < 0) _exit(1); - - } - - if (!child_stopped) { - - /* Once woken up, create a clone of our process. */ - - child_pid = fork(); - if (child_pid < 0) _exit(1); - - /* In child process: close fds, resume execution. */ - - if (!child_pid) { - - //(void)nice(-20); - - signal(SIGCHLD, old_sigchld_handler); - - close(FORKSRV_FD); - close(FORKSRV_FD + 1); - return; - - } - - } else { - - /* Special handling for persistent mode: if the child is alive but - currently stopped, simply restart it with SIGCONT. */ - - kill(child_pid, SIGCONT); - child_stopped = 0; - - } - - /* In parent process: write PID to pipe, then wait for child. */ - - if (write(FORKSRV_FD + 1, &child_pid, 4) != 4) _exit(1); - - if (waitpid(child_pid, &status, is_persistent ? WUNTRACED : 0) < 0) - _exit(1); - - /* In persistent mode, the child stops itself with SIGSTOP to indicate - a successful run. In this case, we want to wake it up without forking - again. */ - - if (WIFSTOPPED(status)) child_stopped = 1; - - /* Relay wait status to pipe, then loop back. */ - - if (write(FORKSRV_FD + 1, &status, 4) != 4) _exit(1); - - } - -} - -/* A simplified persistent mode handler, used as explained in - * llvm_mode/README.md. */ - -int __afl_persistent_loop(unsigned int max_cnt) { - - static u8 first_pass = 1; - static u32 cycle_cnt; - - if (first_pass) { - - /* Make sure that every iteration of __AFL_LOOP() starts with a clean slate. - On subsequent calls, the parent will take care of that, but on the first - iteration, it's our job to erase any trace of whatever happened - before the loop. */ - - if (is_persistent) { - - memset(__afl_area_ptr, 0, __afl_map_size); - __afl_area_ptr[0] = 1; - memset(__afl_prev_loc, 0, NGRAM_SIZE_MAX * sizeof(PREV_LOC_T)); - - } - - cycle_cnt = max_cnt; - first_pass = 0; - return 1; - - } - - if (is_persistent) { - - if (--cycle_cnt) { - - raise(SIGSTOP); - - __afl_area_ptr[0] = 1; - memset(__afl_prev_loc, 0, NGRAM_SIZE_MAX * sizeof(PREV_LOC_T)); - - return 1; - - } else { - - /* When exiting __AFL_LOOP(), make sure that the subsequent code that - follows the loop is not traced. We do that by pivoting back to the - dummy output region. */ - - __afl_area_ptr = __afl_area_initial; - - } - - } - - return 0; - -} - -/* This one can be called from user code when deferred forkserver mode - is enabled. */ - -void __afl_manual_init(void) { - - static u8 init_done; - - if (getenv("AFL_DISABLE_LLVM_INSTRUMENTATION")) { - - init_done = 1; - is_persistent = 0; - __afl_sharedmem_fuzzing = 0; - if (__afl_area_ptr == NULL) __afl_area_ptr = __afl_area_initial; - - if (getenv("AFL_DEBUG")) - fprintf(stderr, - "DEBUG: disabled instrumentation because of " - "AFL_DISABLE_LLVM_INSTRUMENTATION\n"); - - } - - if (!init_done) { - - __afl_start_forkserver(); - init_done = 1; - - } - -} - -/* Initialization of the forkserver - latest possible */ - -__attribute__((constructor())) void __afl_auto_init(void) { - - if (getenv("AFL_DISABLE_LLVM_INSTRUMENTATION")) return; - - if (getenv(DEFER_ENV_VAR)) return; - - __afl_manual_init(); - -} - -/* Initialization of the shmem - earliest possible because of LTO fixed mem. */ - -__attribute__((constructor(CTOR_PRIO))) void __afl_auto_early(void) { - - is_persistent = !!getenv(PERSIST_ENV_VAR); - - if (getenv("AFL_DISABLE_LLVM_INSTRUMENTATION")) return; - - __afl_map_shm(); - -} - -/* preset __afl_area_ptr #2 */ - -__attribute__((constructor(1))) void __afl_auto_second(void) { - - if (getenv("AFL_DISABLE_LLVM_INSTRUMENTATION")) return; - u8 *ptr; - - if (__afl_final_loc) { - - if (__afl_area_ptr && __afl_area_ptr != __afl_area_initial) - free(__afl_area_ptr); - - if (__afl_map_addr) - ptr = (u8 *)mmap((void *)__afl_map_addr, __afl_final_loc, - PROT_READ | PROT_WRITE, - MAP_FIXED_NOREPLACE | MAP_SHARED | MAP_ANONYMOUS, -1, 0); - else - ptr = (u8 *)malloc(__afl_final_loc); - - if (ptr && (ssize_t)ptr != -1) __afl_area_ptr = ptr; - - } - -} - -/* preset __afl_area_ptr #1 - at constructor level 0 global variables have - not been set */ - -__attribute__((constructor(0))) void __afl_auto_first(void) { - - if (getenv("AFL_DISABLE_LLVM_INSTRUMENTATION")) return; - u8 *ptr; - - ptr = (u8 *)malloc(1024000); - - if (ptr && (ssize_t)ptr != -1) __afl_area_ptr = ptr; - -} - -/* The following stuff deals with supporting -fsanitize-coverage=trace-pc-guard. - It remains non-operational in the traditional, plugin-backed LLVM mode. - For more info about 'trace-pc-guard', see llvm_mode/README.md. - - The first function (__sanitizer_cov_trace_pc_guard) is called back on every - edge (as opposed to every basic block). */ - -void __sanitizer_cov_trace_pc_guard(uint32_t *guard) { - - // For stability analysis, if you want to know to which function unstable - // edge IDs belong - uncomment, recompile+install llvm_mode, recompile - // the target. libunwind and libbacktrace are better solutions. - // Set AFL_DEBUG_CHILD_OUTPUT=1 and run afl-fuzz with 2>file to capture - // the backtrace output - /* - uint32_t unstable[] = { ... unstable edge IDs }; - uint32_t idx; - char bt[1024]; - for (idx = 0; i < sizeof(unstable)/sizeof(uint32_t); i++) { - - if (unstable[idx] == __afl_area_ptr[*guard]) { - - int bt_size = backtrace(bt, 256); - if (bt_size > 0) { - - char **bt_syms = backtrace_symbols(bt, bt_size); - if (bt_syms) { - - fprintf(stderr, "DEBUG: edge=%u caller=%s\n", unstable[idx], - bt_syms[0]); - free(bt_syms); - - } - - } - - } - - } - - */ - -#if (LLVM_VERSION_MAJOR < 9) - - __afl_area_ptr[*guard]++; - -#else - - __afl_area_ptr[*guard] = - __afl_area_ptr[*guard] + 1 + (__afl_area_ptr[*guard] == 255 ? 1 : 0); - -#endif - -} - -/* Init callback. Populates instrumentation IDs. Note that we're using - ID of 0 as a special value to indicate non-instrumented bits. That may - still touch the bitmap, but in a fairly harmless way. */ - -void __sanitizer_cov_trace_pc_guard_init(uint32_t *start, uint32_t *stop) { - - u32 inst_ratio = 100; - char *x; - - _is_sancov = 1; - - if (getenv("AFL_DEBUG")) { - - fprintf(stderr, "Running __sanitizer_cov_trace_pc_guard_init: %p-%p\n", - start, stop); - - } - - if (start == stop || *start) return; - - x = getenv("AFL_INST_RATIO"); - if (x) inst_ratio = (u32)atoi(x); - - if (!inst_ratio || inst_ratio > 100) { - - fprintf(stderr, "[-] ERROR: Invalid AFL_INST_RATIO (must be 1-100).\n"); - abort(); - - } - - /* Make sure that the first element in the range is always set - we use that - to avoid duplicate calls (which can happen as an artifact of the underlying - implementation in LLVM). */ - - *(start++) = R(MAP_SIZE - 1) + 1; - - while (start < stop) { - - if (R(100) < inst_ratio) - *start = ++__afl_final_loc; - else - *start = 0; - - start++; - - } - -} - -///// CmpLog instrumentation - -void __cmplog_ins_hook1(uint8_t arg1, uint8_t arg2) { - - if (unlikely(!__afl_cmp_map)) return; - - uintptr_t k = (uintptr_t)__builtin_return_address(0); - k = (k >> 4) ^ (k << 8); - k &= CMP_MAP_W - 1; - - __afl_cmp_map->headers[k].type = CMP_TYPE_INS; - - u32 hits = __afl_cmp_map->headers[k].hits; - __afl_cmp_map->headers[k].hits = hits + 1; - // if (!__afl_cmp_map->headers[k].cnt) - // __afl_cmp_map->headers[k].cnt = __afl_cmp_counter++; - - __afl_cmp_map->headers[k].shape = 0; - - hits &= CMP_MAP_H - 1; - __afl_cmp_map->log[k][hits].v0 = arg1; - __afl_cmp_map->log[k][hits].v1 = arg2; - -} - -void __cmplog_ins_hook2(uint16_t arg1, uint16_t arg2) { - - if (unlikely(!__afl_cmp_map)) return; - - uintptr_t k = (uintptr_t)__builtin_return_address(0); - k = (k >> 4) ^ (k << 8); - k &= CMP_MAP_W - 1; - - __afl_cmp_map->headers[k].type = CMP_TYPE_INS; - - u32 hits = __afl_cmp_map->headers[k].hits; - __afl_cmp_map->headers[k].hits = hits + 1; - - __afl_cmp_map->headers[k].shape = 1; - - hits &= CMP_MAP_H - 1; - __afl_cmp_map->log[k][hits].v0 = arg1; - __afl_cmp_map->log[k][hits].v1 = arg2; - -} - -void __cmplog_ins_hook4(uint32_t arg1, uint32_t arg2) { - - if (unlikely(!__afl_cmp_map)) return; - - uintptr_t k = (uintptr_t)__builtin_return_address(0); - k = (k >> 4) ^ (k << 8); - k &= CMP_MAP_W - 1; - - __afl_cmp_map->headers[k].type = CMP_TYPE_INS; - - u32 hits = __afl_cmp_map->headers[k].hits; - __afl_cmp_map->headers[k].hits = hits + 1; - - __afl_cmp_map->headers[k].shape = 3; - - hits &= CMP_MAP_H - 1; - __afl_cmp_map->log[k][hits].v0 = arg1; - __afl_cmp_map->log[k][hits].v1 = arg2; - -} - -void __cmplog_ins_hook8(uint64_t arg1, uint64_t arg2) { - - if (unlikely(!__afl_cmp_map)) return; - - uintptr_t k = (uintptr_t)__builtin_return_address(0); - k = (k >> 4) ^ (k << 8); - k &= CMP_MAP_W - 1; - - __afl_cmp_map->headers[k].type = CMP_TYPE_INS; - - u32 hits = __afl_cmp_map->headers[k].hits; - __afl_cmp_map->headers[k].hits = hits + 1; - - __afl_cmp_map->headers[k].shape = 7; - - hits &= CMP_MAP_H - 1; - __afl_cmp_map->log[k][hits].v0 = arg1; - __afl_cmp_map->log[k][hits].v1 = arg2; - -} - -#if defined(__APPLE__) - #pragma weak __sanitizer_cov_trace_const_cmp1 = __cmplog_ins_hook1 - #pragma weak __sanitizer_cov_trace_const_cmp2 = __cmplog_ins_hook2 - #pragma weak __sanitizer_cov_trace_const_cmp4 = __cmplog_ins_hook4 - #pragma weak __sanitizer_cov_trace_const_cmp8 = __cmplog_ins_hook8 - - #pragma weak __sanitizer_cov_trace_cmp1 = __cmplog_ins_hook1 - #pragma weak __sanitizer_cov_trace_cmp2 = __cmplog_ins_hook2 - #pragma weak __sanitizer_cov_trace_cmp4 = __cmplog_ins_hook4 - #pragma weak __sanitizer_cov_trace_cmp8 = __cmplog_ins_hook8 -#else -void __sanitizer_cov_trace_const_cmp1(uint8_t arg1, uint8_t arg2) - __attribute__((alias("__cmplog_ins_hook1"))); -void __sanitizer_cov_trace_const_cmp2(uint16_t arg1, uint16_t arg2) - __attribute__((alias("__cmplog_ins_hook2"))); -void __sanitizer_cov_trace_const_cmp4(uint32_t arg1, uint32_t arg2) - __attribute__((alias("__cmplog_ins_hook4"))); -void __sanitizer_cov_trace_const_cmp8(uint64_t arg1, uint64_t arg2) - __attribute__((alias("__cmplog_ins_hook8"))); - -void __sanitizer_cov_trace_cmp1(uint8_t arg1, uint8_t arg2) - __attribute__((alias("__cmplog_ins_hook1"))); -void __sanitizer_cov_trace_cmp2(uint16_t arg1, uint16_t arg2) - __attribute__((alias("__cmplog_ins_hook2"))); -void __sanitizer_cov_trace_cmp4(uint32_t arg1, uint32_t arg2) - __attribute__((alias("__cmplog_ins_hook4"))); -void __sanitizer_cov_trace_cmp8(uint64_t arg1, uint64_t arg2) - __attribute__((alias("__cmplog_ins_hook8"))); -#endif /* defined(__APPLE__) */ - -void __sanitizer_cov_trace_switch(uint64_t val, uint64_t *cases) { - - if (unlikely(!__afl_cmp_map)) return; - - for (uint64_t i = 0; i < cases[0]; i++) { - - uintptr_t k = (uintptr_t)__builtin_return_address(0) + i; - k = (k >> 4) ^ (k << 8); - k &= CMP_MAP_W - 1; - - __afl_cmp_map->headers[k].type = CMP_TYPE_INS; - - u32 hits = __afl_cmp_map->headers[k].hits; - __afl_cmp_map->headers[k].hits = hits + 1; - - __afl_cmp_map->headers[k].shape = 7; - - hits &= CMP_MAP_H - 1; - __afl_cmp_map->log[k][hits].v0 = val; - __afl_cmp_map->log[k][hits].v1 = cases[i + 2]; - - } - -} - -// POSIX shenanigan to see if an area is mapped. -// If it is mapped as X-only, we have a problem, so maybe we should add a check -// to avoid to call it on .text addresses -static int area_is_mapped(void *ptr, size_t len) { - - char *p = ptr; - char *page = (char *)((uintptr_t)p & ~(sysconf(_SC_PAGE_SIZE) - 1)); - - int r = msync(page, (p - page) + len, MS_ASYNC); - if (r < 0) return errno != ENOMEM; - return 1; - -} - -void __cmplog_rtn_hook(u8 *ptr1, u8 *ptr2) { - - if (unlikely(!__afl_cmp_map)) return; - - if (!area_is_mapped(ptr1, 32) || !area_is_mapped(ptr2, 32)) return; - - uintptr_t k = (uintptr_t)__builtin_return_address(0); - k = (k >> 4) ^ (k << 8); - k &= CMP_MAP_W - 1; - - __afl_cmp_map->headers[k].type = CMP_TYPE_RTN; - - u32 hits = __afl_cmp_map->headers[k].hits; - __afl_cmp_map->headers[k].hits = hits + 1; - - __afl_cmp_map->headers[k].shape = 31; - - hits &= CMP_MAP_RTN_H - 1; - __builtin_memcpy(((struct cmpfn_operands *)__afl_cmp_map->log[k])[hits].v0, - ptr1, 32); - __builtin_memcpy(((struct cmpfn_operands *)__afl_cmp_map->log[k])[hits].v1, - ptr2, 32); - -} - diff --git a/llvm_mode/cmplog-instructions-pass.cc b/llvm_mode/cmplog-instructions-pass.cc deleted file mode 100644 index d5de3dbb..00000000 --- a/llvm_mode/cmplog-instructions-pass.cc +++ /dev/null @@ -1,292 +0,0 @@ -/* - american fuzzy lop++ - LLVM CmpLog instrumentation - -------------------------------------------------- - - Written by Andrea Fioraldi - - Copyright 2015, 2016 Google Inc. All rights reserved. - Copyright 2019-2020 AFLplusplus Project. All rights reserved. - - Licensed under the Apache License, Version 2.0 (the "License"); - you may not use this file except in compliance with the License. - You may obtain a copy of the License at: - - http://www.apache.org/licenses/LICENSE-2.0 - -*/ - -#include -#include -#include - -#include -#include -#include -#include -#include "llvm/Config/llvm-config.h" - -#include "llvm/ADT/Statistic.h" -#include "llvm/IR/IRBuilder.h" -#include "llvm/IR/LegacyPassManager.h" -#include "llvm/IR/Module.h" -#include "llvm/Support/Debug.h" -#include "llvm/Support/raw_ostream.h" -#include "llvm/Transforms/IPO/PassManagerBuilder.h" -#include "llvm/Transforms/Utils/BasicBlockUtils.h" -#include "llvm/Pass.h" -#include "llvm/Analysis/ValueTracking.h" - -#if LLVM_VERSION_MAJOR > 3 || \ - (LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR > 4) - #include "llvm/IR/Verifier.h" - #include "llvm/IR/DebugInfo.h" -#else - #include "llvm/Analysis/Verifier.h" - #include "llvm/DebugInfo.h" - #define nullptr 0 -#endif - -#include -#include "afl-llvm-common.h" - -using namespace llvm; - -namespace { - -class CmpLogInstructions : public ModulePass { - - public: - static char ID; - CmpLogInstructions() : ModulePass(ID) { - - initInstrumentList(); - - } - - bool runOnModule(Module &M) override; - -#if LLVM_VERSION_MAJOR < 4 - const char *getPassName() const override { - -#else - StringRef getPassName() const override { - -#endif - return "cmplog instructions"; - - } - - private: - bool hookInstrs(Module &M); - -}; - -} // namespace - -char CmpLogInstructions::ID = 0; - -bool CmpLogInstructions::hookInstrs(Module &M) { - - std::vector icomps; - LLVMContext & C = M.getContext(); - - Type * VoidTy = Type::getVoidTy(C); - IntegerType *Int8Ty = IntegerType::getInt8Ty(C); - IntegerType *Int16Ty = IntegerType::getInt16Ty(C); - IntegerType *Int32Ty = IntegerType::getInt32Ty(C); - IntegerType *Int64Ty = IntegerType::getInt64Ty(C); - -#if LLVM_VERSION_MAJOR < 9 - Constant * -#else - FunctionCallee -#endif - c1 = M.getOrInsertFunction("__cmplog_ins_hook1", VoidTy, Int8Ty, Int8Ty -#if LLVM_VERSION_MAJOR < 5 - , - NULL -#endif - ); -#if LLVM_VERSION_MAJOR < 9 - Function *cmplogHookIns1 = cast(c1); -#else - FunctionCallee cmplogHookIns1 = c1; -#endif - -#if LLVM_VERSION_MAJOR < 9 - Constant * -#else - FunctionCallee -#endif - c2 = M.getOrInsertFunction("__cmplog_ins_hook2", VoidTy, Int16Ty, Int16Ty -#if LLVM_VERSION_MAJOR < 5 - , - NULL -#endif - ); -#if LLVM_VERSION_MAJOR < 9 - Function *cmplogHookIns2 = cast(c2); -#else - FunctionCallee cmplogHookIns2 = c2; -#endif - -#if LLVM_VERSION_MAJOR < 9 - Constant * -#else - FunctionCallee -#endif - c4 = M.getOrInsertFunction("__cmplog_ins_hook4", VoidTy, Int32Ty, Int32Ty -#if LLVM_VERSION_MAJOR < 5 - , - NULL -#endif - ); -#if LLVM_VERSION_MAJOR < 9 - Function *cmplogHookIns4 = cast(c4); -#else - FunctionCallee cmplogHookIns4 = c4; -#endif - -#if LLVM_VERSION_MAJOR < 9 - Constant * -#else - FunctionCallee -#endif - c8 = M.getOrInsertFunction("__cmplog_ins_hook8", VoidTy, Int64Ty, Int64Ty -#if LLVM_VERSION_MAJOR < 5 - , - NULL -#endif - ); -#if LLVM_VERSION_MAJOR < 9 - Function *cmplogHookIns8 = cast(c8); -#else - FunctionCallee cmplogHookIns8 = c8; -#endif - - /* iterate over all functions, bbs and instruction and add suitable calls */ - for (auto &F : M) { - - if (!isInInstrumentList(&F)) continue; - - for (auto &BB : F) { - - for (auto &IN : BB) { - - CmpInst *selectcmpInst = nullptr; - - if ((selectcmpInst = dyn_cast(&IN))) { - - if (selectcmpInst->getPredicate() == CmpInst::ICMP_EQ || - selectcmpInst->getPredicate() == CmpInst::ICMP_NE || - selectcmpInst->getPredicate() == CmpInst::ICMP_UGT || - selectcmpInst->getPredicate() == CmpInst::ICMP_SGT || - selectcmpInst->getPredicate() == CmpInst::ICMP_ULT || - selectcmpInst->getPredicate() == CmpInst::ICMP_SLT || - selectcmpInst->getPredicate() == CmpInst::ICMP_UGE || - selectcmpInst->getPredicate() == CmpInst::ICMP_SGE || - selectcmpInst->getPredicate() == CmpInst::ICMP_ULE || - selectcmpInst->getPredicate() == CmpInst::ICMP_SLE) { - - auto op0 = selectcmpInst->getOperand(0); - auto op1 = selectcmpInst->getOperand(1); - - IntegerType *intTyOp0 = dyn_cast(op0->getType()); - IntegerType *intTyOp1 = dyn_cast(op1->getType()); - - /* this is probably not needed but we do it anyway */ - if (!intTyOp0 || !intTyOp1) { continue; } - - icomps.push_back(selectcmpInst); - - } - - } - - } - - } - - } - - if (!icomps.size()) return false; - if (!be_quiet) errs() << "Hooking " << icomps.size() << " cmp instructions\n"; - - for (auto &selectcmpInst : icomps) { - - IRBuilder<> IRB(selectcmpInst->getParent()); - IRB.SetInsertPoint(selectcmpInst); - - auto op0 = selectcmpInst->getOperand(0); - auto op1 = selectcmpInst->getOperand(1); - - IntegerType *intTyOp0 = dyn_cast(op0->getType()); - IntegerType *intTyOp1 = dyn_cast(op1->getType()); - - unsigned max_size = intTyOp0->getBitWidth() > intTyOp1->getBitWidth() - ? intTyOp0->getBitWidth() - : intTyOp1->getBitWidth(); - - std::vector args; - args.push_back(op0); - args.push_back(op1); - - switch (max_size) { - - case 8: - IRB.CreateCall(cmplogHookIns1, args); - break; - case 16: - IRB.CreateCall(cmplogHookIns2, args); - break; - case 32: - IRB.CreateCall(cmplogHookIns4, args); - break; - case 64: - IRB.CreateCall(cmplogHookIns8, args); - break; - default: - break; - - } - - } - - return true; - -} - -bool CmpLogInstructions::runOnModule(Module &M) { - - if (getenv("AFL_QUIET") == NULL) - llvm::errs() - << "Running cmplog-instructions-pass by andreafioraldi@gmail.com\n"; - else - be_quiet = 1; - hookInstrs(M); - verifyModule(M); - - return true; - -} - -static void registerCmpLogInstructionsPass(const PassManagerBuilder &, - legacy::PassManagerBase &PM) { - - auto p = new CmpLogInstructions(); - PM.add(p); - -} - -static RegisterStandardPasses RegisterCmpLogInstructionsPass( - PassManagerBuilder::EP_OptimizerLast, registerCmpLogInstructionsPass); - -static RegisterStandardPasses RegisterCmpLogInstructionsPass0( - PassManagerBuilder::EP_EnabledOnOptLevel0, registerCmpLogInstructionsPass); - -#if LLVM_VERSION_MAJOR >= 11 -static RegisterStandardPasses RegisterCmpLogInstructionsPassLTO( - PassManagerBuilder::EP_FullLinkTimeOptimizationLast, - registerCmpLogInstructionsPass); -#endif - diff --git a/llvm_mode/cmplog-routines-pass.cc b/llvm_mode/cmplog-routines-pass.cc deleted file mode 100644 index c44f38c4..00000000 --- a/llvm_mode/cmplog-routines-pass.cc +++ /dev/null @@ -1,212 +0,0 @@ -/* - american fuzzy lop++ - LLVM CmpLog instrumentation - -------------------------------------------------- - - Written by Andrea Fioraldi - - Copyright 2015, 2016 Google Inc. All rights reserved. - Copyright 2019-2020 AFLplusplus Project. All rights reserved. - - Licensed under the Apache License, Version 2.0 (the "License"); - you may not use this file except in compliance with the License. - You may obtain a copy of the License at: - - http://www.apache.org/licenses/LICENSE-2.0 - -*/ - -#include -#include -#include - -#include -#include -#include -#include -#include "llvm/Config/llvm-config.h" - -#include "llvm/ADT/Statistic.h" -#include "llvm/IR/IRBuilder.h" -#include "llvm/IR/LegacyPassManager.h" -#include "llvm/IR/Module.h" -#include "llvm/Support/Debug.h" -#include "llvm/Support/raw_ostream.h" -#include "llvm/Transforms/IPO/PassManagerBuilder.h" -#include "llvm/Transforms/Utils/BasicBlockUtils.h" -#include "llvm/Pass.h" -#include "llvm/Analysis/ValueTracking.h" - -#if LLVM_VERSION_MAJOR > 3 || \ - (LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR > 4) - #include "llvm/IR/Verifier.h" - #include "llvm/IR/DebugInfo.h" -#else - #include "llvm/Analysis/Verifier.h" - #include "llvm/DebugInfo.h" - #define nullptr 0 -#endif - -#include -#include "afl-llvm-common.h" - -using namespace llvm; - -namespace { - -class CmpLogRoutines : public ModulePass { - - public: - static char ID; - CmpLogRoutines() : ModulePass(ID) { - - initInstrumentList(); - - } - - bool runOnModule(Module &M) override; - -#if LLVM_VERSION_MAJOR < 4 - const char *getPassName() const override { - -#else - StringRef getPassName() const override { - -#endif - return "cmplog routines"; - - } - - private: - bool hookRtns(Module &M); - -}; - -} // namespace - -char CmpLogRoutines::ID = 0; - -bool CmpLogRoutines::hookRtns(Module &M) { - - std::vector calls; - LLVMContext & C = M.getContext(); - - Type *VoidTy = Type::getVoidTy(C); - // PointerType *VoidPtrTy = PointerType::get(VoidTy, 0); - IntegerType *Int8Ty = IntegerType::getInt8Ty(C); - PointerType *i8PtrTy = PointerType::get(Int8Ty, 0); - -#if LLVM_VERSION_MAJOR < 9 - Constant * -#else - FunctionCallee -#endif - c = M.getOrInsertFunction("__cmplog_rtn_hook", VoidTy, i8PtrTy, i8PtrTy -#if LLVM_VERSION_MAJOR < 5 - , - NULL -#endif - ); -#if LLVM_VERSION_MAJOR < 9 - Function *cmplogHookFn = cast(c); -#else - FunctionCallee cmplogHookFn = c; -#endif - - /* iterate over all functions, bbs and instruction and add suitable calls */ - for (auto &F : M) { - - if (!isInInstrumentList(&F)) continue; - - for (auto &BB : F) { - - for (auto &IN : BB) { - - CallInst *callInst = nullptr; - - if ((callInst = dyn_cast(&IN))) { - - Function *Callee = callInst->getCalledFunction(); - if (!Callee) continue; - if (callInst->getCallingConv() != llvm::CallingConv::C) continue; - - FunctionType *FT = Callee->getFunctionType(); - - bool isPtrRtn = FT->getNumParams() >= 2 && - !FT->getReturnType()->isVoidTy() && - FT->getParamType(0) == FT->getParamType(1) && - FT->getParamType(0)->isPointerTy(); - - if (!isPtrRtn) continue; - - calls.push_back(callInst); - - } - - } - - } - - } - - if (!calls.size()) return false; - if (!be_quiet) - errs() << "Hooking " << calls.size() - << " calls with pointers as arguments\n"; - - for (auto &callInst : calls) { - - Value *v1P = callInst->getArgOperand(0), *v2P = callInst->getArgOperand(1); - - IRBuilder<> IRB(callInst->getParent()); - IRB.SetInsertPoint(callInst); - - std::vector args; - Value * v1Pcasted = IRB.CreatePointerCast(v1P, i8PtrTy); - Value * v2Pcasted = IRB.CreatePointerCast(v2P, i8PtrTy); - args.push_back(v1Pcasted); - args.push_back(v2Pcasted); - - IRB.CreateCall(cmplogHookFn, args); - - // errs() << callInst->getCalledFunction()->getName() << "\n"; - - } - - return true; - -} - -bool CmpLogRoutines::runOnModule(Module &M) { - - if (getenv("AFL_QUIET") == NULL) - llvm::errs() - << "Running cmplog-routines-pass by andreafioraldi@gmail.com\n"; - else - be_quiet = 1; - hookRtns(M); - verifyModule(M); - - return true; - -} - -static void registerCmpLogRoutinesPass(const PassManagerBuilder &, - legacy::PassManagerBase &PM) { - - auto p = new CmpLogRoutines(); - PM.add(p); - -} - -static RegisterStandardPasses RegisterCmpLogRoutinesPass( - PassManagerBuilder::EP_OptimizerLast, registerCmpLogRoutinesPass); - -static RegisterStandardPasses RegisterCmpLogRoutinesPass0( - PassManagerBuilder::EP_EnabledOnOptLevel0, registerCmpLogRoutinesPass); - -#if LLVM_VERSION_MAJOR >= 11 -static RegisterStandardPasses RegisterCmpLogRoutinesPassLTO( - PassManagerBuilder::EP_FullLinkTimeOptimizationLast, - registerCmpLogRoutinesPass); -#endif - diff --git a/llvm_mode/compare-transform-pass.so.cc b/llvm_mode/compare-transform-pass.so.cc deleted file mode 100644 index acdd0f3b..00000000 --- a/llvm_mode/compare-transform-pass.so.cc +++ /dev/null @@ -1,587 +0,0 @@ -/* - * Copyright 2016 laf-intel - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -#include -#include -#include - -#include -#include -#include -#include -#include "llvm/Config/llvm-config.h" - -#include "llvm/ADT/Statistic.h" -#include "llvm/IR/IRBuilder.h" -#include "llvm/IR/LegacyPassManager.h" -#include "llvm/IR/Module.h" -#include "llvm/Support/Debug.h" -#include "llvm/Support/raw_ostream.h" -#include "llvm/Transforms/IPO/PassManagerBuilder.h" -#include "llvm/Transforms/Utils/BasicBlockUtils.h" -#include "llvm/Pass.h" -#include "llvm/Analysis/ValueTracking.h" - -#if LLVM_VERSION_MAJOR > 3 || \ - (LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR > 4) - #include "llvm/IR/Verifier.h" - #include "llvm/IR/DebugInfo.h" -#else - #include "llvm/Analysis/Verifier.h" - #include "llvm/DebugInfo.h" - #define nullptr 0 -#endif - -#include -#include "afl-llvm-common.h" - -using namespace llvm; - -namespace { - -class CompareTransform : public ModulePass { - - public: - static char ID; - CompareTransform() : ModulePass(ID) { - - initInstrumentList(); - - } - - bool runOnModule(Module &M) override; - -#if LLVM_VERSION_MAJOR < 4 - const char *getPassName() const override { - -#else - StringRef getPassName() const override { - -#endif - return "transforms compare functions"; - - } - - private: - bool transformCmps(Module &M, const bool processStrcmp, - const bool processMemcmp, const bool processStrncmp, - const bool processStrcasecmp, - const bool processStrncasecmp); - -}; - -} // namespace - -char CompareTransform::ID = 0; - -bool CompareTransform::transformCmps(Module &M, const bool processStrcmp, - const bool processMemcmp, - const bool processStrncmp, - const bool processStrcasecmp, - const bool processStrncasecmp) { - - DenseMap valueMap; - std::vector calls; - LLVMContext & C = M.getContext(); - IntegerType * Int8Ty = IntegerType::getInt8Ty(C); - IntegerType * Int32Ty = IntegerType::getInt32Ty(C); - IntegerType * Int64Ty = IntegerType::getInt64Ty(C); - -#if LLVM_VERSION_MAJOR < 9 - Constant * -#else - FunctionCallee -#endif - c = M.getOrInsertFunction("tolower", Int32Ty, Int32Ty -#if LLVM_VERSION_MAJOR < 5 - , - NULL -#endif - ); -#if LLVM_VERSION_MAJOR < 9 - Function *tolowerFn = cast(c); -#else - FunctionCallee tolowerFn = c; -#endif - - /* iterate over all functions, bbs and instruction and add suitable calls to - * strcmp/memcmp/strncmp/strcasecmp/strncasecmp */ - for (auto &F : M) { - - if (!isInInstrumentList(&F)) continue; - - for (auto &BB : F) { - - for (auto &IN : BB) { - - CallInst *callInst = nullptr; - - if ((callInst = dyn_cast(&IN))) { - - bool isStrcmp = processStrcmp; - bool isMemcmp = processMemcmp; - bool isStrncmp = processStrncmp; - bool isStrcasecmp = processStrcasecmp; - bool isStrncasecmp = processStrncasecmp; - bool isIntMemcpy = true; - - Function *Callee = callInst->getCalledFunction(); - if (!Callee) continue; - if (callInst->getCallingConv() != llvm::CallingConv::C) continue; - StringRef FuncName = Callee->getName(); - isStrcmp &= !FuncName.compare(StringRef("strcmp")); - isMemcmp &= !FuncName.compare(StringRef("memcmp")); - isStrncmp &= !FuncName.compare(StringRef("strncmp")); - isStrcasecmp &= !FuncName.compare(StringRef("strcasecmp")); - isStrncasecmp &= !FuncName.compare(StringRef("strncasecmp")); - isIntMemcpy &= !FuncName.compare("llvm.memcpy.p0i8.p0i8.i64"); - - if (!isStrcmp && !isMemcmp && !isStrncmp && !isStrcasecmp && - !isStrncasecmp && !isIntMemcpy) - continue; - - /* Verify the strcmp/memcmp/strncmp/strcasecmp/strncasecmp function - * prototype */ - FunctionType *FT = Callee->getFunctionType(); - - isStrcmp &= - FT->getNumParams() == 2 && FT->getReturnType()->isIntegerTy(32) && - FT->getParamType(0) == FT->getParamType(1) && - FT->getParamType(0) == IntegerType::getInt8PtrTy(M.getContext()); - isStrcasecmp &= - FT->getNumParams() == 2 && FT->getReturnType()->isIntegerTy(32) && - FT->getParamType(0) == FT->getParamType(1) && - FT->getParamType(0) == IntegerType::getInt8PtrTy(M.getContext()); - isMemcmp &= FT->getNumParams() == 3 && - FT->getReturnType()->isIntegerTy(32) && - FT->getParamType(0)->isPointerTy() && - FT->getParamType(1)->isPointerTy() && - FT->getParamType(2)->isIntegerTy(); - isStrncmp &= FT->getNumParams() == 3 && - FT->getReturnType()->isIntegerTy(32) && - FT->getParamType(0) == FT->getParamType(1) && - FT->getParamType(0) == - IntegerType::getInt8PtrTy(M.getContext()) && - FT->getParamType(2)->isIntegerTy(); - isStrncasecmp &= FT->getNumParams() == 3 && - FT->getReturnType()->isIntegerTy(32) && - FT->getParamType(0) == FT->getParamType(1) && - FT->getParamType(0) == - IntegerType::getInt8PtrTy(M.getContext()) && - FT->getParamType(2)->isIntegerTy(); - - if (!isStrcmp && !isMemcmp && !isStrncmp && !isStrcasecmp && - !isStrncasecmp && !isIntMemcpy) - continue; - - /* is a str{n,}{case,}cmp/memcmp, check if we have - * str{case,}cmp(x, "const") or str{case,}cmp("const", x) - * strn{case,}cmp(x, "const", ..) or strn{case,}cmp("const", x, ..) - * memcmp(x, "const", ..) or memcmp("const", x, ..) */ - Value *Str1P = callInst->getArgOperand(0), - *Str2P = callInst->getArgOperand(1); - StringRef Str1, Str2; - bool HasStr1 = getConstantStringInfo(Str1P, Str1); - bool HasStr2 = getConstantStringInfo(Str2P, Str2); - - if (isIntMemcpy && HasStr2) { - - valueMap[Str1P] = new std::string(Str2.str()); - // fprintf(stderr, "saved %s for %p\n", Str2.str().c_str(), Str1P); - continue; - - } - - // not literal? maybe global or local variable - if (!(HasStr1 || HasStr2)) { - - auto *Ptr = dyn_cast(Str2P); - if (Ptr && Ptr->isGEPWithNoNotionalOverIndexing()) { - - if (auto *Var = dyn_cast(Ptr->getOperand(0))) { - - if (Var->hasInitializer()) { - - if (auto *Array = - dyn_cast(Var->getInitializer())) { - - HasStr2 = true; - Str2 = Array->getAsString(); - valueMap[Str2P] = new std::string(Str2.str()); - fprintf(stderr, "glo2 %s\n", Str2.str().c_str()); - - } - - } - - } - - } - - if (!HasStr2) { - - auto *Ptr = dyn_cast(Str1P); - if (Ptr && Ptr->isGEPWithNoNotionalOverIndexing()) { - - if (auto *Var = dyn_cast(Ptr->getOperand(0))) { - - if (Var->hasInitializer()) { - - if (auto *Array = dyn_cast( - Var->getInitializer())) { - - HasStr1 = true; - Str1 = Array->getAsString(); - valueMap[Str1P] = new std::string(Str1.str()); - // fprintf(stderr, "glo1 %s\n", Str1.str().c_str()); - - } - - } - - } - - } - - } else if (isIntMemcpy) { - - valueMap[Str1P] = new std::string(Str2.str()); - // fprintf(stderr, "saved\n"); - - } - - } - - if (isIntMemcpy) continue; - - if (!(HasStr1 || HasStr2)) { - - // do we have a saved local variable initialization? - std::string *val = valueMap[Str1P]; - if (val && !val->empty()) { - - Str1 = StringRef(*val); - HasStr1 = true; - // fprintf(stderr, "loaded1 %s\n", Str1.str().c_str()); - - } else { - - val = valueMap[Str2P]; - if (val && !val->empty()) { - - Str2 = StringRef(*val); - HasStr2 = true; - // fprintf(stderr, "loaded2 %s\n", Str2.str().c_str()); - - } - - } - - } - - /* handle cases of one string is const, one string is variable */ - if (!(HasStr1 || HasStr2)) continue; - - if (isMemcmp || isStrncmp || isStrncasecmp) { - - /* check if third operand is a constant integer - * strlen("constStr") and sizeof() are treated as constant */ - Value * op2 = callInst->getArgOperand(2); - ConstantInt *ilen = dyn_cast(op2); - if (ilen) { - - uint64_t len = ilen->getZExtValue(); - // if len is zero this is a pointless call but allow real - // implementation to worry about that - if (!len) continue; - - if (isMemcmp) { - - // if size of compare is larger than constant string this is - // likely a bug but allow real implementation to worry about - // that - uint64_t literalLength = HasStr1 ? Str1.size() : Str2.size(); - if (literalLength + 1 < ilen->getZExtValue()) continue; - - } - - } else if (isMemcmp) - - // this *may* supply a len greater than the constant string at - // runtime so similarly we don't want to have to handle that - continue; - - } - - calls.push_back(callInst); - - } - - } - - } - - } - - if (!calls.size()) return false; - if (!be_quiet) - errs() << "Replacing " << calls.size() - << " calls to strcmp/memcmp/strncmp/strcasecmp/strncasecmp\n"; - - for (auto &callInst : calls) { - - Value *Str1P = callInst->getArgOperand(0), - *Str2P = callInst->getArgOperand(1); - StringRef Str1, Str2, ConstStr; - std::string TmpConstStr; - Value * VarStr; - bool HasStr1 = getConstantStringInfo(Str1P, Str1); - bool HasStr2 = getConstantStringInfo(Str2P, Str2); - uint64_t constStrLen, unrollLen, constSizedLen = 0; - bool isMemcmp = - !callInst->getCalledFunction()->getName().compare(StringRef("memcmp")); - bool isSizedcmp = isMemcmp || - !callInst->getCalledFunction()->getName().compare( - StringRef("strncmp")) || - !callInst->getCalledFunction()->getName().compare( - StringRef("strncasecmp")); - Value *sizedValue = isSizedcmp ? callInst->getArgOperand(2) : NULL; - bool isConstSized = sizedValue && isa(sizedValue); - bool isCaseInsensitive = !callInst->getCalledFunction()->getName().compare( - StringRef("strcasecmp")) || - !callInst->getCalledFunction()->getName().compare( - StringRef("strncasecmp")); - - if (!(HasStr1 || HasStr2)) { - - // do we have a saved local or global variable initialization? - std::string *val = valueMap[Str1P]; - if (val && !val->empty()) { - - Str1 = StringRef(*val); - HasStr1 = true; - - } else { - - val = valueMap[Str2P]; - if (val && !val->empty()) { - - Str2 = StringRef(*val); - HasStr2 = true; - - } - - } - - } - - if (isConstSized) { - - constSizedLen = dyn_cast(sizedValue)->getZExtValue(); - - } - - if (HasStr1) { - - TmpConstStr = Str1.str(); - VarStr = Str2P; - - } else { - - TmpConstStr = Str2.str(); - VarStr = Str1P; - - } - - // add null termination character implicit in c strings - TmpConstStr.append("\0", 1); - - // in the unusual case the const str has embedded null - // characters, the string comparison functions should terminate - // at the first null - if (!isMemcmp) - TmpConstStr.assign(TmpConstStr, 0, TmpConstStr.find('\0') + 1); - - constStrLen = TmpConstStr.length(); - // prefer use of StringRef (in comparison to std::string a StringRef has - // built-in runtime bounds checking, which makes debugging easier) - ConstStr = StringRef(TmpConstStr); - - if (isConstSized) - unrollLen = constSizedLen < constStrLen ? constSizedLen : constStrLen; - else - unrollLen = constStrLen; - - if (!be_quiet) - errs() << callInst->getCalledFunction()->getName() << ": unroll len " - << unrollLen - << ((isSizedcmp && !isConstSized) ? ", variable n" : "") << ": " - << ConstStr << "\n"; - - /* split before the call instruction */ - BasicBlock *bb = callInst->getParent(); - BasicBlock *end_bb = bb->splitBasicBlock(BasicBlock::iterator(callInst)); - - BasicBlock *next_lenchk_bb = NULL; - if (isSizedcmp && !isConstSized) { - - next_lenchk_bb = - BasicBlock::Create(C, "len_check", end_bb->getParent(), end_bb); - BranchInst::Create(end_bb, next_lenchk_bb); - - } - - BasicBlock *next_cmp_bb = - BasicBlock::Create(C, "cmp_added", end_bb->getParent(), end_bb); - BranchInst::Create(end_bb, next_cmp_bb); - PHINode *PN = PHINode::Create( - Int32Ty, (next_lenchk_bb ? 2 : 1) * unrollLen + 1, "cmp_phi"); - -#if LLVM_VERSION_MAJOR < 8 - TerminatorInst *term = bb->getTerminator(); -#else - Instruction *term = bb->getTerminator(); -#endif - BranchInst::Create(next_lenchk_bb ? next_lenchk_bb : next_cmp_bb, bb); - term->eraseFromParent(); - - for (uint64_t i = 0; i < unrollLen; i++) { - - BasicBlock * cur_cmp_bb = next_cmp_bb, *cur_lenchk_bb = next_lenchk_bb; - unsigned char c; - - if (cur_lenchk_bb) { - - IRBuilder<> cur_lenchk_IRB(&*(cur_lenchk_bb->getFirstInsertionPt())); - Value * icmp = cur_lenchk_IRB.CreateICmpEQ( - sizedValue, ConstantInt::get(sizedValue->getType(), i)); - cur_lenchk_IRB.CreateCondBr(icmp, end_bb, cur_cmp_bb); - cur_lenchk_bb->getTerminator()->eraseFromParent(); - - PN->addIncoming(ConstantInt::get(Int32Ty, 0), cur_lenchk_bb); - - } - - if (isCaseInsensitive) - c = (unsigned char)(tolower((int)ConstStr[i]) & 0xff); - else - c = (unsigned char)ConstStr[i]; - - IRBuilder<> cur_cmp_IRB(&*(cur_cmp_bb->getFirstInsertionPt())); - - Value *v = ConstantInt::get(Int64Ty, i); - Value *ele = cur_cmp_IRB.CreateInBoundsGEP(VarStr, v, "empty"); - Value *load = cur_cmp_IRB.CreateLoad(ele); - - if (isCaseInsensitive) { - - // load >= 'A' && load <= 'Z' ? load | 0x020 : load - load = cur_cmp_IRB.CreateZExt(load, Int32Ty); - std::vector args; - args.push_back(load); - load = cur_cmp_IRB.CreateCall(tolowerFn, args); - load = cur_cmp_IRB.CreateTrunc(load, Int8Ty); - - } - - Value *isub; - if (HasStr1) - isub = cur_cmp_IRB.CreateSub(ConstantInt::get(Int8Ty, c), load); - else - isub = cur_cmp_IRB.CreateSub(load, ConstantInt::get(Int8Ty, c)); - - Value *sext = cur_cmp_IRB.CreateSExt(isub, Int32Ty); - PN->addIncoming(sext, cur_cmp_bb); - - if (i < unrollLen - 1) { - - if (cur_lenchk_bb) { - - next_lenchk_bb = - BasicBlock::Create(C, "len_check", end_bb->getParent(), end_bb); - BranchInst::Create(end_bb, next_lenchk_bb); - - } - - next_cmp_bb = - BasicBlock::Create(C, "cmp_added", end_bb->getParent(), end_bb); - BranchInst::Create(end_bb, next_cmp_bb); - - Value *icmp = - cur_cmp_IRB.CreateICmpEQ(isub, ConstantInt::get(Int8Ty, 0)); - cur_cmp_IRB.CreateCondBr( - icmp, next_lenchk_bb ? next_lenchk_bb : next_cmp_bb, end_bb); - cur_cmp_bb->getTerminator()->eraseFromParent(); - - } else { - - // IRB.CreateBr(end_bb); - - } - - // add offset to varstr - // create load - // create signed isub - // create icmp - // create jcc - // create next_bb - - } - - /* since the call is the first instruction of the bb it is safe to - * replace it with a phi instruction */ - BasicBlock::iterator ii(callInst); - ReplaceInstWithInst(callInst->getParent()->getInstList(), ii, PN); - - } - - return true; - -} - -bool CompareTransform::runOnModule(Module &M) { - - if ((isatty(2) && getenv("AFL_QUIET") == NULL) || getenv("AFL_DEBUG") != NULL) - llvm::errs() << "Running compare-transform-pass by laf.intel@gmail.com, " - "extended by heiko@hexco.de\n"; - else - be_quiet = 1; - transformCmps(M, true, true, true, true, true); - verifyModule(M); - - return true; - -} - -static void registerCompTransPass(const PassManagerBuilder &, - legacy::PassManagerBase &PM) { - - auto p = new CompareTransform(); - PM.add(p); - -} - -static RegisterStandardPasses RegisterCompTransPass( - PassManagerBuilder::EP_OptimizerLast, registerCompTransPass); - -static RegisterStandardPasses RegisterCompTransPass0( - PassManagerBuilder::EP_EnabledOnOptLevel0, registerCompTransPass); - -#if LLVM_VERSION_MAJOR >= 11 -static RegisterStandardPasses RegisterCompTransPassLTO( - PassManagerBuilder::EP_FullLinkTimeOptimizationLast, registerCompTransPass); -#endif - diff --git a/llvm_mode/llvm-ngram-coverage.h b/llvm_mode/llvm-ngram-coverage.h deleted file mode 100644 index 12b666e9..00000000 --- a/llvm_mode/llvm-ngram-coverage.h +++ /dev/null @@ -1,18 +0,0 @@ -#ifndef AFL_NGRAM_CONFIG_H -#define AFL_NGRAM_CONFIG_H - -#include "../config.h" - -#if (MAP_SIZE_POW2 <= 16) -typedef u16 PREV_LOC_T; -#elif (MAP_SIZE_POW2 <= 32) -typedef u32 PREV_LOC_T; -#else -typedef u64 PREV_LOC_T; -#endif - -/* Maximum ngram size */ -#define NGRAM_SIZE_MAX 16U - -#endif - diff --git a/llvm_mode/split-compares-pass.so.cc b/llvm_mode/split-compares-pass.so.cc deleted file mode 100644 index 2fb90e5e..00000000 --- a/llvm_mode/split-compares-pass.so.cc +++ /dev/null @@ -1,1356 +0,0 @@ -/* - * Copyright 2016 laf-intel - * extended for floating point by Heiko Eißfeldt - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -#include -#include -#include - -#include -#include -#include -#include - -#include "llvm/Config/llvm-config.h" - -#include "llvm/Pass.h" -#include "llvm/Support/raw_ostream.h" -#include "llvm/IR/LegacyPassManager.h" -#include "llvm/Transforms/IPO/PassManagerBuilder.h" -#include "llvm/Transforms/Utils/BasicBlockUtils.h" -#include "llvm/IR/Module.h" - -#include "llvm/IR/IRBuilder.h" -#if LLVM_VERSION_MAJOR > 3 || \ - (LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR > 4) - #include "llvm/IR/Verifier.h" - #include "llvm/IR/DebugInfo.h" -#else - #include "llvm/Analysis/Verifier.h" - #include "llvm/DebugInfo.h" - #define nullptr 0 -#endif - -using namespace llvm; -#include "afl-llvm-common.h" - -namespace { - -class SplitComparesTransform : public ModulePass { - - public: - static char ID; - SplitComparesTransform() : ModulePass(ID) { - - initInstrumentList(); - - } - - bool runOnModule(Module &M) override; -#if LLVM_VERSION_MAJOR >= 4 - StringRef getPassName() const override { - -#else - const char *getPassName() const override { - -#endif - return "simplifies and splits ICMP instructions"; - - } - - private: - int enableFPSplit; - - size_t splitIntCompares(Module &M, unsigned bitw); - size_t splitFPCompares(Module &M); - bool simplifyCompares(Module &M); - bool simplifyFPCompares(Module &M); - bool simplifyIntSignedness(Module &M); - size_t nextPowerOfTwo(size_t in); - -}; - -} // namespace - -char SplitComparesTransform::ID = 0; - -/* This function splits FCMP instructions with xGE or xLE predicates into two - * FCMP instructions with predicate xGT or xLT and EQ */ -bool SplitComparesTransform::simplifyFPCompares(Module &M) { - - LLVMContext & C = M.getContext(); - std::vector fcomps; - IntegerType * Int1Ty = IntegerType::getInt1Ty(C); - - /* iterate over all functions, bbs and instruction and add - * all integer comparisons with >= and <= predicates to the icomps vector */ - for (auto &F : M) { - - if (!isInInstrumentList(&F)) continue; - - for (auto &BB : F) { - - for (auto &IN : BB) { - - CmpInst *selectcmpInst = nullptr; - - if ((selectcmpInst = dyn_cast(&IN))) { - - if (enableFPSplit && - (selectcmpInst->getPredicate() == CmpInst::FCMP_OGE || - selectcmpInst->getPredicate() == CmpInst::FCMP_UGE || - selectcmpInst->getPredicate() == CmpInst::FCMP_OLE || - selectcmpInst->getPredicate() == CmpInst::FCMP_ULE)) { - - auto op0 = selectcmpInst->getOperand(0); - auto op1 = selectcmpInst->getOperand(1); - - Type *TyOp0 = op0->getType(); - Type *TyOp1 = op1->getType(); - - /* this is probably not needed but we do it anyway */ - if (TyOp0 != TyOp1) { continue; } - - if (TyOp0->isArrayTy() || TyOp0->isVectorTy()) { continue; } - - fcomps.push_back(selectcmpInst); - - } - - } - - } - - } - - } - - if (!fcomps.size()) { return false; } - - /* transform for floating point */ - for (auto &FcmpInst : fcomps) { - - BasicBlock *bb = FcmpInst->getParent(); - - auto op0 = FcmpInst->getOperand(0); - auto op1 = FcmpInst->getOperand(1); - - /* find out what the new predicate is going to be */ - auto pred = dyn_cast(FcmpInst)->getPredicate(); - CmpInst::Predicate new_pred; - switch (pred) { - - case CmpInst::FCMP_UGE: - new_pred = CmpInst::FCMP_UGT; - break; - case CmpInst::FCMP_OGE: - new_pred = CmpInst::FCMP_OGT; - break; - case CmpInst::FCMP_ULE: - new_pred = CmpInst::FCMP_ULT; - break; - case CmpInst::FCMP_OLE: - new_pred = CmpInst::FCMP_OLT; - break; - default: // keep the compiler happy - continue; - - } - - /* split before the fcmp instruction */ - BasicBlock *end_bb = bb->splitBasicBlock(BasicBlock::iterator(FcmpInst)); - - /* the old bb now contains a unconditional jump to the new one (end_bb) - * we need to delete it later */ - - /* create the FCMP instruction with new_pred and add it to the old basic - * block bb it is now at the position where the old FcmpInst was */ - Instruction *fcmp_np; - fcmp_np = CmpInst::Create(Instruction::FCmp, new_pred, op0, op1); - bb->getInstList().insert(BasicBlock::iterator(bb->getTerminator()), - fcmp_np); - - /* create a new basic block which holds the new EQ fcmp */ - Instruction *fcmp_eq; - /* insert middle_bb before end_bb */ - BasicBlock *middle_bb = - BasicBlock::Create(C, "injected", end_bb->getParent(), end_bb); - fcmp_eq = CmpInst::Create(Instruction::FCmp, CmpInst::FCMP_OEQ, op0, op1); - middle_bb->getInstList().push_back(fcmp_eq); - /* add an unconditional branch to the end of middle_bb with destination - * end_bb */ - BranchInst::Create(end_bb, middle_bb); - - /* replace the uncond branch with a conditional one, which depends on the - * new_pred fcmp. True goes to end, false to the middle (injected) bb */ - auto term = bb->getTerminator(); - BranchInst::Create(end_bb, middle_bb, fcmp_np, bb); - term->eraseFromParent(); - - /* replace the old FcmpInst (which is the first inst in end_bb) with a PHI - * inst to wire up the loose ends */ - PHINode *PN = PHINode::Create(Int1Ty, 2, ""); - /* the first result depends on the outcome of fcmp_eq */ - PN->addIncoming(fcmp_eq, middle_bb); - /* if the source was the original bb we know that the fcmp_np yielded true - * hence we can hardcode this value */ - PN->addIncoming(ConstantInt::get(Int1Ty, 1), bb); - /* replace the old FcmpInst with our new and shiny PHI inst */ - BasicBlock::iterator ii(FcmpInst); - ReplaceInstWithInst(FcmpInst->getParent()->getInstList(), ii, PN); - - } - - return true; - -} - -/* This function splits ICMP instructions with xGE or xLE predicates into two - * ICMP instructions with predicate xGT or xLT and EQ */ -bool SplitComparesTransform::simplifyCompares(Module &M) { - - LLVMContext & C = M.getContext(); - std::vector icomps; - IntegerType * Int1Ty = IntegerType::getInt1Ty(C); - - /* iterate over all functions, bbs and instruction and add - * all integer comparisons with >= and <= predicates to the icomps vector */ - for (auto &F : M) { - - if (!isInInstrumentList(&F)) continue; - - for (auto &BB : F) { - - for (auto &IN : BB) { - - CmpInst *selectcmpInst = nullptr; - - if ((selectcmpInst = dyn_cast(&IN))) { - - if (selectcmpInst->getPredicate() == CmpInst::ICMP_UGE || - selectcmpInst->getPredicate() == CmpInst::ICMP_SGE || - selectcmpInst->getPredicate() == CmpInst::ICMP_ULE || - selectcmpInst->getPredicate() == CmpInst::ICMP_SLE) { - - auto op0 = selectcmpInst->getOperand(0); - auto op1 = selectcmpInst->getOperand(1); - - IntegerType *intTyOp0 = dyn_cast(op0->getType()); - IntegerType *intTyOp1 = dyn_cast(op1->getType()); - - /* this is probably not needed but we do it anyway */ - if (!intTyOp0 || !intTyOp1) { continue; } - - icomps.push_back(selectcmpInst); - - } - - } - - } - - } - - } - - if (!icomps.size()) { return false; } - - for (auto &IcmpInst : icomps) { - - BasicBlock *bb = IcmpInst->getParent(); - - auto op0 = IcmpInst->getOperand(0); - auto op1 = IcmpInst->getOperand(1); - - /* find out what the new predicate is going to be */ - auto pred = dyn_cast(IcmpInst)->getPredicate(); - CmpInst::Predicate new_pred; - switch (pred) { - - case CmpInst::ICMP_UGE: - new_pred = CmpInst::ICMP_UGT; - break; - case CmpInst::ICMP_SGE: - new_pred = CmpInst::ICMP_SGT; - break; - case CmpInst::ICMP_ULE: - new_pred = CmpInst::ICMP_ULT; - break; - case CmpInst::ICMP_SLE: - new_pred = CmpInst::ICMP_SLT; - break; - default: // keep the compiler happy - continue; - - } - - /* split before the icmp instruction */ - BasicBlock *end_bb = bb->splitBasicBlock(BasicBlock::iterator(IcmpInst)); - - /* the old bb now contains a unconditional jump to the new one (end_bb) - * we need to delete it later */ - - /* create the ICMP instruction with new_pred and add it to the old basic - * block bb it is now at the position where the old IcmpInst was */ - Instruction *icmp_np; - icmp_np = CmpInst::Create(Instruction::ICmp, new_pred, op0, op1); - bb->getInstList().insert(BasicBlock::iterator(bb->getTerminator()), - icmp_np); - - /* create a new basic block which holds the new EQ icmp */ - Instruction *icmp_eq; - /* insert middle_bb before end_bb */ - BasicBlock *middle_bb = - BasicBlock::Create(C, "injected", end_bb->getParent(), end_bb); - icmp_eq = CmpInst::Create(Instruction::ICmp, CmpInst::ICMP_EQ, op0, op1); - middle_bb->getInstList().push_back(icmp_eq); - /* add an unconditional branch to the end of middle_bb with destination - * end_bb */ - BranchInst::Create(end_bb, middle_bb); - - /* replace the uncond branch with a conditional one, which depends on the - * new_pred icmp. True goes to end, false to the middle (injected) bb */ - auto term = bb->getTerminator(); - BranchInst::Create(end_bb, middle_bb, icmp_np, bb); - term->eraseFromParent(); - - /* replace the old IcmpInst (which is the first inst in end_bb) with a PHI - * inst to wire up the loose ends */ - PHINode *PN = PHINode::Create(Int1Ty, 2, ""); - /* the first result depends on the outcome of icmp_eq */ - PN->addIncoming(icmp_eq, middle_bb); - /* if the source was the original bb we know that the icmp_np yielded true - * hence we can hardcode this value */ - PN->addIncoming(ConstantInt::get(Int1Ty, 1), bb); - /* replace the old IcmpInst with our new and shiny PHI inst */ - BasicBlock::iterator ii(IcmpInst); - ReplaceInstWithInst(IcmpInst->getParent()->getInstList(), ii, PN); - - } - - return true; - -} - -/* this function transforms signed compares to equivalent unsigned compares */ -bool SplitComparesTransform::simplifyIntSignedness(Module &M) { - - LLVMContext & C = M.getContext(); - std::vector icomps; - IntegerType * Int1Ty = IntegerType::getInt1Ty(C); - - /* iterate over all functions, bbs and instructions and add - * all signed compares to icomps vector */ - for (auto &F : M) { - - if (!isInInstrumentList(&F)) continue; - - for (auto &BB : F) { - - for (auto &IN : BB) { - - CmpInst *selectcmpInst = nullptr; - - if ((selectcmpInst = dyn_cast(&IN))) { - - if (selectcmpInst->getPredicate() == CmpInst::ICMP_SGT || - selectcmpInst->getPredicate() == CmpInst::ICMP_SLT) { - - auto op0 = selectcmpInst->getOperand(0); - auto op1 = selectcmpInst->getOperand(1); - - IntegerType *intTyOp0 = dyn_cast(op0->getType()); - IntegerType *intTyOp1 = dyn_cast(op1->getType()); - - /* see above */ - if (!intTyOp0 || !intTyOp1) { continue; } - - /* i think this is not possible but to lazy to look it up */ - if (intTyOp0->getBitWidth() != intTyOp1->getBitWidth()) { - - continue; - - } - - icomps.push_back(selectcmpInst); - - } - - } - - } - - } - - } - - if (!icomps.size()) { return false; } - - for (auto &IcmpInst : icomps) { - - BasicBlock *bb = IcmpInst->getParent(); - - auto op0 = IcmpInst->getOperand(0); - auto op1 = IcmpInst->getOperand(1); - - IntegerType *intTyOp0 = dyn_cast(op0->getType()); - unsigned bitw = intTyOp0->getBitWidth(); - IntegerType *IntType = IntegerType::get(C, bitw); - - /* get the new predicate */ - auto pred = dyn_cast(IcmpInst)->getPredicate(); - CmpInst::Predicate new_pred; - if (pred == CmpInst::ICMP_SGT) { - - new_pred = CmpInst::ICMP_UGT; - - } else { - - new_pred = CmpInst::ICMP_ULT; - - } - - BasicBlock *end_bb = bb->splitBasicBlock(BasicBlock::iterator(IcmpInst)); - - /* create a 1 bit compare for the sign bit. to do this shift and trunc - * the original operands so only the first bit remains.*/ - Instruction *s_op0, *t_op0, *s_op1, *t_op1, *icmp_sign_bit; - - s_op0 = BinaryOperator::Create(Instruction::LShr, op0, - ConstantInt::get(IntType, bitw - 1)); - bb->getInstList().insert(BasicBlock::iterator(bb->getTerminator()), s_op0); - t_op0 = new TruncInst(s_op0, Int1Ty); - bb->getInstList().insert(BasicBlock::iterator(bb->getTerminator()), t_op0); - - s_op1 = BinaryOperator::Create(Instruction::LShr, op1, - ConstantInt::get(IntType, bitw - 1)); - bb->getInstList().insert(BasicBlock::iterator(bb->getTerminator()), s_op1); - t_op1 = new TruncInst(s_op1, Int1Ty); - bb->getInstList().insert(BasicBlock::iterator(bb->getTerminator()), t_op1); - - /* compare of the sign bits */ - icmp_sign_bit = - CmpInst::Create(Instruction::ICmp, CmpInst::ICMP_EQ, t_op0, t_op1); - bb->getInstList().insert(BasicBlock::iterator(bb->getTerminator()), - icmp_sign_bit); - - /* create a new basic block which is executed if the signedness bit is - * different */ - Instruction *icmp_inv_sig_cmp; - BasicBlock * sign_bb = - BasicBlock::Create(C, "sign", end_bb->getParent(), end_bb); - if (pred == CmpInst::ICMP_SGT) { - - /* if we check for > and the op0 positive and op1 negative then the final - * result is true. if op0 negative and op1 pos, the cmp must result - * in false - */ - icmp_inv_sig_cmp = - CmpInst::Create(Instruction::ICmp, CmpInst::ICMP_ULT, t_op0, t_op1); - - } else { - - /* just the inverse of the above statement */ - icmp_inv_sig_cmp = - CmpInst::Create(Instruction::ICmp, CmpInst::ICMP_UGT, t_op0, t_op1); - - } - - sign_bb->getInstList().push_back(icmp_inv_sig_cmp); - BranchInst::Create(end_bb, sign_bb); - - /* create a new bb which is executed if signedness is equal */ - Instruction *icmp_usign_cmp; - BasicBlock * middle_bb = - BasicBlock::Create(C, "injected", end_bb->getParent(), end_bb); - /* we can do a normal unsigned compare now */ - icmp_usign_cmp = CmpInst::Create(Instruction::ICmp, new_pred, op0, op1); - middle_bb->getInstList().push_back(icmp_usign_cmp); - BranchInst::Create(end_bb, middle_bb); - - auto term = bb->getTerminator(); - /* if the sign is eq do a normal unsigned cmp, else we have to check the - * signedness bit */ - BranchInst::Create(middle_bb, sign_bb, icmp_sign_bit, bb); - term->eraseFromParent(); - - PHINode *PN = PHINode::Create(Int1Ty, 2, ""); - - PN->addIncoming(icmp_usign_cmp, middle_bb); - PN->addIncoming(icmp_inv_sig_cmp, sign_bb); - - BasicBlock::iterator ii(IcmpInst); - ReplaceInstWithInst(IcmpInst->getParent()->getInstList(), ii, PN); - - } - - return true; - -} - -size_t SplitComparesTransform::nextPowerOfTwo(size_t in) { - - --in; - in |= in >> 1; - in |= in >> 2; - in |= in >> 4; - // in |= in >> 8; - // in |= in >> 16; - return in + 1; - -} - -/* splits fcmps into two nested fcmps with sign compare and the rest */ -size_t SplitComparesTransform::splitFPCompares(Module &M) { - - size_t count = 0; - - LLVMContext &C = M.getContext(); - -#if LLVM_VERSION_MAJOR > 3 || \ - (LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR > 7) - const DataLayout &dl = M.getDataLayout(); - - /* define unions with floating point and (sign, exponent, mantissa) triples - */ - if (dl.isLittleEndian()) { - - } else if (dl.isBigEndian()) { - - } else { - - return count; - - } - -#endif - - std::vector fcomps; - - /* get all EQ, NE, GT, and LT fcmps. if the other two - * functions were executed only these four predicates should exist */ - for (auto &F : M) { - - if (!isInInstrumentList(&F)) continue; - - for (auto &BB : F) { - - for (auto &IN : BB) { - - CmpInst *selectcmpInst = nullptr; - - if ((selectcmpInst = dyn_cast(&IN))) { - - if (selectcmpInst->getPredicate() == CmpInst::FCMP_OEQ || - selectcmpInst->getPredicate() == CmpInst::FCMP_ONE || - selectcmpInst->getPredicate() == CmpInst::FCMP_UNE || - selectcmpInst->getPredicate() == CmpInst::FCMP_UGT || - selectcmpInst->getPredicate() == CmpInst::FCMP_OGT || - selectcmpInst->getPredicate() == CmpInst::FCMP_ULT || - selectcmpInst->getPredicate() == CmpInst::FCMP_OLT) { - - auto op0 = selectcmpInst->getOperand(0); - auto op1 = selectcmpInst->getOperand(1); - - Type *TyOp0 = op0->getType(); - Type *TyOp1 = op1->getType(); - - if (TyOp0 != TyOp1) { continue; } - - if (TyOp0->isArrayTy() || TyOp0->isVectorTy()) { continue; } - - fcomps.push_back(selectcmpInst); - - } - - } - - } - - } - - } - - if (!fcomps.size()) { return count; } - - IntegerType *Int1Ty = IntegerType::getInt1Ty(C); - - for (auto &FcmpInst : fcomps) { - - BasicBlock *bb = FcmpInst->getParent(); - - auto op0 = FcmpInst->getOperand(0); - auto op1 = FcmpInst->getOperand(1); - - unsigned op_size; - op_size = op0->getType()->getPrimitiveSizeInBits(); - - if (op_size != op1->getType()->getPrimitiveSizeInBits()) { continue; } - - const unsigned int sizeInBits = op0->getType()->getPrimitiveSizeInBits(); - const unsigned int precision = - sizeInBits == 32 - ? 24 - : sizeInBits == 64 - ? 53 - : sizeInBits == 128 ? 113 - : sizeInBits == 16 ? 11 - /* sizeInBits == 80 */ - : 65; - - const unsigned shiftR_exponent = precision - 1; - const unsigned long long mask_fraction = - (1ULL << (shiftR_exponent - 1)) | ((1ULL << (shiftR_exponent - 1)) - 1); - const unsigned long long mask_exponent = - (1ULL << (sizeInBits - precision)) - 1; - - // round up sizes to the next power of two - // this should help with integer compare splitting - size_t exTySizeBytes = ((sizeInBits - precision + 7) >> 3); - size_t frTySizeBytes = ((precision - 1ULL + 7) >> 3); - - IntegerType *IntExponentTy = - IntegerType::get(C, nextPowerOfTwo(exTySizeBytes) << 3); - IntegerType *IntFractionTy = - IntegerType::get(C, nextPowerOfTwo(frTySizeBytes) << 3); - - // errs() << "Fractions: IntFractionTy size " << - // IntFractionTy->getPrimitiveSizeInBits() << ", op_size " << op_size << - // ", mask " << mask_fraction << - // ", precision " << precision << "\n"; - - BasicBlock *end_bb = bb->splitBasicBlock(BasicBlock::iterator(FcmpInst)); - - /* create the integers from floats directly */ - Instruction *b_op0, *b_op1; - b_op0 = CastInst::Create(Instruction::BitCast, op0, - IntegerType::get(C, op_size)); - bb->getInstList().insert(BasicBlock::iterator(bb->getTerminator()), b_op0); - - b_op1 = CastInst::Create(Instruction::BitCast, op1, - IntegerType::get(C, op_size)); - bb->getInstList().insert(BasicBlock::iterator(bb->getTerminator()), b_op1); - - /* isolate signs of value of floating point type */ - - /* create a 1 bit compare for the sign bit. to do this shift and trunc - * the original operands so only the first bit remains.*/ - Instruction *s_s0, *t_s0, *s_s1, *t_s1, *icmp_sign_bit; - - s_s0 = - BinaryOperator::Create(Instruction::LShr, b_op0, - ConstantInt::get(b_op0->getType(), op_size - 1)); - bb->getInstList().insert(BasicBlock::iterator(bb->getTerminator()), s_s0); - t_s0 = new TruncInst(s_s0, Int1Ty); - bb->getInstList().insert(BasicBlock::iterator(bb->getTerminator()), t_s0); - - s_s1 = - BinaryOperator::Create(Instruction::LShr, b_op1, - ConstantInt::get(b_op1->getType(), op_size - 1)); - bb->getInstList().insert(BasicBlock::iterator(bb->getTerminator()), s_s1); - t_s1 = new TruncInst(s_s1, Int1Ty); - bb->getInstList().insert(BasicBlock::iterator(bb->getTerminator()), t_s1); - - /* compare of the sign bits */ - icmp_sign_bit = - CmpInst::Create(Instruction::ICmp, CmpInst::ICMP_EQ, t_s0, t_s1); - bb->getInstList().insert(BasicBlock::iterator(bb->getTerminator()), - icmp_sign_bit); - - /* create a new basic block which is executed if the signedness bits are - * equal */ - BasicBlock *signequal_bb = - BasicBlock::Create(C, "signequal", end_bb->getParent(), end_bb); - - BranchInst::Create(end_bb, signequal_bb); - - /* create a new bb which is executed if exponents are satisfying the compare - */ - BasicBlock *middle_bb = - BasicBlock::Create(C, "injected", end_bb->getParent(), end_bb); - - BranchInst::Create(end_bb, middle_bb); - - auto term = bb->getTerminator(); - /* if the signs are different goto end_bb else to signequal_bb */ - BranchInst::Create(signequal_bb, end_bb, icmp_sign_bit, bb); - term->eraseFromParent(); - - /* insert code for equal signs */ - - /* isolate the exponents */ - Instruction *s_e0, *m_e0, *t_e0, *s_e1, *m_e1, *t_e1; - - s_e0 = BinaryOperator::Create( - Instruction::LShr, b_op0, - ConstantInt::get(b_op0->getType(), shiftR_exponent)); - s_e1 = BinaryOperator::Create( - Instruction::LShr, b_op1, - ConstantInt::get(b_op1->getType(), shiftR_exponent)); - signequal_bb->getInstList().insert( - BasicBlock::iterator(signequal_bb->getTerminator()), s_e0); - signequal_bb->getInstList().insert( - BasicBlock::iterator(signequal_bb->getTerminator()), s_e1); - - t_e0 = new TruncInst(s_e0, IntExponentTy); - t_e1 = new TruncInst(s_e1, IntExponentTy); - signequal_bb->getInstList().insert( - BasicBlock::iterator(signequal_bb->getTerminator()), t_e0); - signequal_bb->getInstList().insert( - BasicBlock::iterator(signequal_bb->getTerminator()), t_e1); - - if (sizeInBits - precision < exTySizeBytes * 8) { - - m_e0 = BinaryOperator::Create( - Instruction::And, t_e0, - ConstantInt::get(t_e0->getType(), mask_exponent)); - m_e1 = BinaryOperator::Create( - Instruction::And, t_e1, - ConstantInt::get(t_e1->getType(), mask_exponent)); - signequal_bb->getInstList().insert( - BasicBlock::iterator(signequal_bb->getTerminator()), m_e0); - signequal_bb->getInstList().insert( - BasicBlock::iterator(signequal_bb->getTerminator()), m_e1); - - } else { - - m_e0 = t_e0; - m_e1 = t_e1; - - } - - /* compare the exponents of the operands */ - Instruction *icmp_exponents_equal; - Instruction *icmp_exponent_result; - BasicBlock * signequal2_bb = signequal_bb; - switch (FcmpInst->getPredicate()) { - - case CmpInst::FCMP_OEQ: - icmp_exponent_result = - CmpInst::Create(Instruction::ICmp, CmpInst::ICMP_EQ, m_e0, m_e1); - break; - case CmpInst::FCMP_ONE: - case CmpInst::FCMP_UNE: - icmp_exponent_result = - CmpInst::Create(Instruction::ICmp, CmpInst::ICMP_NE, m_e0, m_e1); - break; - /* compare the exponents of the operands (signs are equal) - * if exponents are equal -> proceed to mantissa comparison - * else get result depending on sign - */ - case CmpInst::FCMP_OGT: - case CmpInst::FCMP_UGT: - Instruction *icmp_exponent; - icmp_exponents_equal = - CmpInst::Create(Instruction::ICmp, CmpInst::ICMP_EQ, m_e0, m_e1); - signequal_bb->getInstList().insert( - BasicBlock::iterator(signequal_bb->getTerminator()), - icmp_exponents_equal); - - // shortcut for unequal exponents - signequal2_bb = signequal_bb->splitBasicBlock( - BasicBlock::iterator(signequal_bb->getTerminator())); - - /* if the exponents are equal goto middle_bb else to signequal2_bb */ - term = signequal_bb->getTerminator(); - BranchInst::Create(middle_bb, signequal2_bb, icmp_exponents_equal, - signequal_bb); - term->eraseFromParent(); - - icmp_exponent = - CmpInst::Create(Instruction::ICmp, CmpInst::ICMP_UGT, m_e0, m_e1); - signequal2_bb->getInstList().insert( - BasicBlock::iterator(signequal2_bb->getTerminator()), - icmp_exponent); - icmp_exponent_result = - BinaryOperator::Create(Instruction::Xor, icmp_exponent, t_s0); - break; - case CmpInst::FCMP_OLT: - case CmpInst::FCMP_ULT: - icmp_exponents_equal = - CmpInst::Create(Instruction::ICmp, CmpInst::ICMP_EQ, m_e0, m_e1); - signequal_bb->getInstList().insert( - BasicBlock::iterator(signequal_bb->getTerminator()), - icmp_exponents_equal); - - // shortcut for unequal exponents - signequal2_bb = signequal_bb->splitBasicBlock( - BasicBlock::iterator(signequal_bb->getTerminator())); - - /* if the exponents are equal goto middle_bb else to signequal2_bb */ - term = signequal_bb->getTerminator(); - BranchInst::Create(middle_bb, signequal2_bb, icmp_exponents_equal, - signequal_bb); - term->eraseFromParent(); - - icmp_exponent = - CmpInst::Create(Instruction::ICmp, CmpInst::ICMP_ULT, m_e0, m_e1); - signequal2_bb->getInstList().insert( - BasicBlock::iterator(signequal2_bb->getTerminator()), - icmp_exponent); - icmp_exponent_result = - BinaryOperator::Create(Instruction::Xor, icmp_exponent, t_s0); - break; - default: - continue; - - } - - signequal2_bb->getInstList().insert( - BasicBlock::iterator(signequal2_bb->getTerminator()), - icmp_exponent_result); - - { - - term = signequal2_bb->getTerminator(); - - switch (FcmpInst->getPredicate()) { - - case CmpInst::FCMP_OEQ: - /* if the exponents are satifying the compare do a fraction cmp in - * middle_bb */ - BranchInst::Create(middle_bb, end_bb, icmp_exponent_result, - signequal2_bb); - break; - case CmpInst::FCMP_ONE: - case CmpInst::FCMP_UNE: - /* if the exponents are satifying the compare do a fraction cmp in - * middle_bb */ - BranchInst::Create(end_bb, middle_bb, icmp_exponent_result, - signequal2_bb); - break; - case CmpInst::FCMP_OGT: - case CmpInst::FCMP_UGT: - case CmpInst::FCMP_OLT: - case CmpInst::FCMP_ULT: - BranchInst::Create(end_bb, signequal2_bb); - break; - default: - continue; - - } - - term->eraseFromParent(); - - } - - /* isolate the mantissa aka fraction */ - Instruction *t_f0, *t_f1; - bool needTrunc = IntFractionTy->getPrimitiveSizeInBits() < op_size; - - if (precision - 1 < frTySizeBytes * 8) { - - Instruction *m_f0, *m_f1; - m_f0 = BinaryOperator::Create( - Instruction::And, b_op0, - ConstantInt::get(b_op0->getType(), mask_fraction)); - m_f1 = BinaryOperator::Create( - Instruction::And, b_op1, - ConstantInt::get(b_op1->getType(), mask_fraction)); - middle_bb->getInstList().insert( - BasicBlock::iterator(middle_bb->getTerminator()), m_f0); - middle_bb->getInstList().insert( - BasicBlock::iterator(middle_bb->getTerminator()), m_f1); - - if (needTrunc) { - - t_f0 = new TruncInst(m_f0, IntFractionTy); - t_f1 = new TruncInst(m_f1, IntFractionTy); - middle_bb->getInstList().insert( - BasicBlock::iterator(middle_bb->getTerminator()), t_f0); - middle_bb->getInstList().insert( - BasicBlock::iterator(middle_bb->getTerminator()), t_f1); - - } else { - - t_f0 = m_f0; - t_f1 = m_f1; - - } - - } else { - - if (needTrunc) { - - t_f0 = new TruncInst(b_op0, IntFractionTy); - t_f1 = new TruncInst(b_op1, IntFractionTy); - middle_bb->getInstList().insert( - BasicBlock::iterator(middle_bb->getTerminator()), t_f0); - middle_bb->getInstList().insert( - BasicBlock::iterator(middle_bb->getTerminator()), t_f1); - - } else { - - t_f0 = b_op0; - t_f1 = b_op1; - - } - - } - - /* compare the fractions of the operands */ - Instruction *icmp_fraction_result; - Instruction *icmp_fraction_result2; - BasicBlock * middle2_bb = middle_bb; - PHINode * PN2 = nullptr; - switch (FcmpInst->getPredicate()) { - - case CmpInst::FCMP_OEQ: - icmp_fraction_result = - CmpInst::Create(Instruction::ICmp, CmpInst::ICMP_EQ, t_f0, t_f1); - middle2_bb->getInstList().insert( - BasicBlock::iterator(middle2_bb->getTerminator()), - icmp_fraction_result); - - break; - case CmpInst::FCMP_UNE: - case CmpInst::FCMP_ONE: - icmp_fraction_result = - CmpInst::Create(Instruction::ICmp, CmpInst::ICMP_NE, t_f0, t_f1); - middle2_bb->getInstList().insert( - BasicBlock::iterator(middle2_bb->getTerminator()), - icmp_fraction_result); - - break; - case CmpInst::FCMP_OGT: - case CmpInst::FCMP_UGT: - case CmpInst::FCMP_OLT: - case CmpInst::FCMP_ULT: { - - middle2_bb = middle_bb->splitBasicBlock( - BasicBlock::iterator(middle_bb->getTerminator())); - - BasicBlock *negative_bb = BasicBlock::Create( - C, "negative_value", middle2_bb->getParent(), middle2_bb); - BasicBlock *positive_bb = BasicBlock::Create( - C, "positive_value", negative_bb->getParent(), negative_bb); - - if (FcmpInst->getPredicate() == CmpInst::FCMP_OGT || - FcmpInst->getPredicate() == CmpInst::FCMP_UGT) { - - negative_bb->getInstList().push_back( - icmp_fraction_result = CmpInst::Create( - Instruction::ICmp, CmpInst::ICMP_ULT, t_f0, t_f1)); - positive_bb->getInstList().push_back( - icmp_fraction_result2 = CmpInst::Create( - Instruction::ICmp, CmpInst::ICMP_UGT, t_f0, t_f1)); - - } else { - - negative_bb->getInstList().push_back( - icmp_fraction_result = CmpInst::Create( - Instruction::ICmp, CmpInst::ICMP_UGT, t_f0, t_f1)); - positive_bb->getInstList().push_back( - icmp_fraction_result2 = CmpInst::Create( - Instruction::ICmp, CmpInst::ICMP_ULT, t_f0, t_f1)); - - } - - BranchInst::Create(middle2_bb, negative_bb); - BranchInst::Create(middle2_bb, positive_bb); - - term = middle_bb->getTerminator(); - BranchInst::Create(negative_bb, positive_bb, t_s0, middle_bb); - term->eraseFromParent(); - - PN2 = PHINode::Create(Int1Ty, 2, ""); - PN2->addIncoming(icmp_fraction_result, negative_bb); - PN2->addIncoming(icmp_fraction_result2, positive_bb); - middle2_bb->getInstList().insert( - BasicBlock::iterator(middle2_bb->getTerminator()), PN2); - - } break; - - default: - continue; - - } - - PHINode *PN = PHINode::Create(Int1Ty, 3, ""); - - switch (FcmpInst->getPredicate()) { - - case CmpInst::FCMP_OEQ: - /* unequal signs cannot be equal values */ - /* goto false branch */ - PN->addIncoming(ConstantInt::get(Int1Ty, 0), bb); - /* unequal exponents cannot be equal values, too */ - PN->addIncoming(ConstantInt::get(Int1Ty, 0), signequal_bb); - /* fractions comparison */ - PN->addIncoming(icmp_fraction_result, middle2_bb); - break; - case CmpInst::FCMP_ONE: - case CmpInst::FCMP_UNE: - /* unequal signs are unequal values */ - /* goto true branch */ - PN->addIncoming(ConstantInt::get(Int1Ty, 1), bb); - /* unequal exponents are unequal values, too */ - PN->addIncoming(icmp_exponent_result, signequal_bb); - /* fractions comparison */ - PN->addIncoming(icmp_fraction_result, middle2_bb); - break; - case CmpInst::FCMP_OGT: - case CmpInst::FCMP_UGT: - /* if op1 is negative goto true branch, - else go on comparing */ - PN->addIncoming(t_s1, bb); - PN->addIncoming(icmp_exponent_result, signequal2_bb); - PN->addIncoming(PN2, middle2_bb); - break; - case CmpInst::FCMP_OLT: - case CmpInst::FCMP_ULT: - /* if op0 is negative goto true branch, - else go on comparing */ - PN->addIncoming(t_s0, bb); - PN->addIncoming(icmp_exponent_result, signequal2_bb); - PN->addIncoming(PN2, middle2_bb); - break; - default: - continue; - - } - - BasicBlock::iterator ii(FcmpInst); - ReplaceInstWithInst(FcmpInst->getParent()->getInstList(), ii, PN); - ++count; - - } - - return count; - -} - -/* splits icmps of size bitw into two nested icmps with bitw/2 size each */ -size_t SplitComparesTransform::splitIntCompares(Module &M, unsigned bitw) { - - size_t count = 0; - - LLVMContext &C = M.getContext(); - - IntegerType *Int1Ty = IntegerType::getInt1Ty(C); - IntegerType *OldIntType = IntegerType::get(C, bitw); - IntegerType *NewIntType = IntegerType::get(C, bitw / 2); - - std::vector icomps; - - if (bitw % 2) { return 0; } - - /* not supported yet */ - if (bitw > 64) { return 0; } - - /* get all EQ, NE, UGT, and ULT icmps of width bitw. if the - * functions simplifyCompares() and simplifyIntSignedness() - * were executed only these four predicates should exist */ - for (auto &F : M) { - - if (!isInInstrumentList(&F)) continue; - - for (auto &BB : F) { - - for (auto &IN : BB) { - - CmpInst *selectcmpInst = nullptr; - - if ((selectcmpInst = dyn_cast(&IN))) { - - if (selectcmpInst->getPredicate() == CmpInst::ICMP_EQ || - selectcmpInst->getPredicate() == CmpInst::ICMP_NE || - selectcmpInst->getPredicate() == CmpInst::ICMP_UGT || - selectcmpInst->getPredicate() == CmpInst::ICMP_ULT) { - - auto op0 = selectcmpInst->getOperand(0); - auto op1 = selectcmpInst->getOperand(1); - - IntegerType *intTyOp0 = dyn_cast(op0->getType()); - IntegerType *intTyOp1 = dyn_cast(op1->getType()); - - if (!intTyOp0 || !intTyOp1) { continue; } - - /* check if the bitwidths are the one we are looking for */ - if (intTyOp0->getBitWidth() != bitw || - intTyOp1->getBitWidth() != bitw) { - - continue; - - } - - icomps.push_back(selectcmpInst); - - } - - } - - } - - } - - } - - if (!icomps.size()) { return 0; } - - for (auto &IcmpInst : icomps) { - - BasicBlock *bb = IcmpInst->getParent(); - - auto op0 = IcmpInst->getOperand(0); - auto op1 = IcmpInst->getOperand(1); - - auto pred = dyn_cast(IcmpInst)->getPredicate(); - - BasicBlock *end_bb = bb->splitBasicBlock(BasicBlock::iterator(IcmpInst)); - - /* create the comparison of the top halves of the original operands */ - Instruction *s_op0, *op0_high, *s_op1, *op1_high, *icmp_high; - - s_op0 = BinaryOperator::Create(Instruction::LShr, op0, - ConstantInt::get(OldIntType, bitw / 2)); - bb->getInstList().insert(BasicBlock::iterator(bb->getTerminator()), s_op0); - op0_high = new TruncInst(s_op0, NewIntType); - bb->getInstList().insert(BasicBlock::iterator(bb->getTerminator()), - op0_high); - - s_op1 = BinaryOperator::Create(Instruction::LShr, op1, - ConstantInt::get(OldIntType, bitw / 2)); - bb->getInstList().insert(BasicBlock::iterator(bb->getTerminator()), s_op1); - op1_high = new TruncInst(s_op1, NewIntType); - bb->getInstList().insert(BasicBlock::iterator(bb->getTerminator()), - op1_high); - - icmp_high = CmpInst::Create(Instruction::ICmp, pred, op0_high, op1_high); - bb->getInstList().insert(BasicBlock::iterator(bb->getTerminator()), - icmp_high); - - /* now we have to destinguish between == != and > < */ - if (pred == CmpInst::ICMP_EQ || pred == CmpInst::ICMP_NE) { - - /* transformation for == and != icmps */ - - /* create a compare for the lower half of the original operands */ - Instruction *op0_low, *op1_low, *icmp_low; - BasicBlock * cmp_low_bb = - BasicBlock::Create(C, "injected", end_bb->getParent(), end_bb); - - op0_low = new TruncInst(op0, NewIntType); - cmp_low_bb->getInstList().push_back(op0_low); - - op1_low = new TruncInst(op1, NewIntType); - cmp_low_bb->getInstList().push_back(op1_low); - - icmp_low = CmpInst::Create(Instruction::ICmp, pred, op0_low, op1_low); - cmp_low_bb->getInstList().push_back(icmp_low); - BranchInst::Create(end_bb, cmp_low_bb); - - /* dependent on the cmp of the high parts go to the end or go on with - * the comparison */ - auto term = bb->getTerminator(); - if (pred == CmpInst::ICMP_EQ) { - - BranchInst::Create(cmp_low_bb, end_bb, icmp_high, bb); - - } else { - - /* CmpInst::ICMP_NE */ - BranchInst::Create(end_bb, cmp_low_bb, icmp_high, bb); - - } - - term->eraseFromParent(); - - /* create the PHI and connect the edges accordingly */ - PHINode *PN = PHINode::Create(Int1Ty, 2, ""); - PN->addIncoming(icmp_low, cmp_low_bb); - if (pred == CmpInst::ICMP_EQ) { - - PN->addIncoming(ConstantInt::get(Int1Ty, 0), bb); - - } else { - - /* CmpInst::ICMP_NE */ - PN->addIncoming(ConstantInt::get(Int1Ty, 1), bb); - - } - - /* replace the old icmp with the new PHI */ - BasicBlock::iterator ii(IcmpInst); - ReplaceInstWithInst(IcmpInst->getParent()->getInstList(), ii, PN); - - } else { - - /* CmpInst::ICMP_UGT and CmpInst::ICMP_ULT */ - /* transformations for < and > */ - - /* create a basic block which checks for the inverse predicate. - * if this is true we can go to the end if not we have to go to the - * bb which checks the lower half of the operands */ - Instruction *icmp_inv_cmp, *op0_low, *op1_low, *icmp_low; - BasicBlock * inv_cmp_bb = - BasicBlock::Create(C, "inv_cmp", end_bb->getParent(), end_bb); - if (pred == CmpInst::ICMP_UGT) { - - icmp_inv_cmp = CmpInst::Create(Instruction::ICmp, CmpInst::ICMP_ULT, - op0_high, op1_high); - - } else { - - icmp_inv_cmp = CmpInst::Create(Instruction::ICmp, CmpInst::ICMP_UGT, - op0_high, op1_high); - - } - - inv_cmp_bb->getInstList().push_back(icmp_inv_cmp); - - auto term = bb->getTerminator(); - term->eraseFromParent(); - BranchInst::Create(end_bb, inv_cmp_bb, icmp_high, bb); - - /* create a bb which handles the cmp of the lower halves */ - BasicBlock *cmp_low_bb = - BasicBlock::Create(C, "injected", end_bb->getParent(), end_bb); - op0_low = new TruncInst(op0, NewIntType); - cmp_low_bb->getInstList().push_back(op0_low); - op1_low = new TruncInst(op1, NewIntType); - cmp_low_bb->getInstList().push_back(op1_low); - - icmp_low = CmpInst::Create(Instruction::ICmp, pred, op0_low, op1_low); - cmp_low_bb->getInstList().push_back(icmp_low); - BranchInst::Create(end_bb, cmp_low_bb); - - BranchInst::Create(end_bb, cmp_low_bb, icmp_inv_cmp, inv_cmp_bb); - - PHINode *PN = PHINode::Create(Int1Ty, 3); - PN->addIncoming(icmp_low, cmp_low_bb); - PN->addIncoming(ConstantInt::get(Int1Ty, 1), bb); - PN->addIncoming(ConstantInt::get(Int1Ty, 0), inv_cmp_bb); - - BasicBlock::iterator ii(IcmpInst); - ReplaceInstWithInst(IcmpInst->getParent()->getInstList(), ii, PN); - - } - - ++count; - - } - - return count; - -} - -bool SplitComparesTransform::runOnModule(Module &M) { - - int bitw = 64; - size_t count; - - char *bitw_env = getenv("AFL_LLVM_LAF_SPLIT_COMPARES_BITW"); - if (!bitw_env) bitw_env = getenv("LAF_SPLIT_COMPARES_BITW"); - if (bitw_env) { bitw = atoi(bitw_env); } - - enableFPSplit = getenv("AFL_LLVM_LAF_SPLIT_FLOATS") != NULL; - - if ((isatty(2) && getenv("AFL_QUIET") == NULL) || - getenv("AFL_DEBUG") != NULL) { - - errs() << "Split-compare-pass by laf.intel@gmail.com, extended by " - "heiko@hexco.de\n"; - - } else { - - be_quiet = 1; - - } - - if (enableFPSplit) { - - count = splitFPCompares(M); - - if (!be_quiet) { - - errs() << "Split-floatingpoint-compare-pass: " << count - << " FP comparisons split\n"; - - } - - simplifyFPCompares(M); - - } - - simplifyCompares(M); - - simplifyIntSignedness(M); - - switch (bitw) { - - case 64: - count = splitIntCompares(M, bitw); - if (!be_quiet) - errs() << "Split-integer-compare-pass " << bitw << "bit: " << count - << " split\n"; - - bitw >>= 1; -#if LLVM_VERSION_MAJOR > 3 || \ - (LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR > 7) - [[clang::fallthrough]]; /*FALLTHRU*/ /* FALLTHROUGH */ -#endif - case 32: - count = splitIntCompares(M, bitw); - if (!be_quiet) - errs() << "Split-integer-compare-pass " << bitw << "bit: " << count - << " split\n"; - - bitw >>= 1; -#if LLVM_VERSION_MAJOR > 3 || \ - (LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR > 7) - [[clang::fallthrough]]; /*FALLTHRU*/ /* FALLTHROUGH */ -#endif - case 16: - count = splitIntCompares(M, bitw); - if (!be_quiet) - errs() << "Split-integer-compare-pass " << bitw << "bit: " << count - << " split\n"; - - bitw >>= 1; - break; - - default: - if (!be_quiet) errs() << "NOT Running split-compare-pass \n"; - return false; - break; - - } - - verifyModule(M); - return true; - -} - -static void registerSplitComparesPass(const PassManagerBuilder &, - legacy::PassManagerBase &PM) { - - PM.add(new SplitComparesTransform()); - -} - -static RegisterStandardPasses RegisterSplitComparesPass( - PassManagerBuilder::EP_OptimizerLast, registerSplitComparesPass); - -static RegisterStandardPasses RegisterSplitComparesTransPass0( - PassManagerBuilder::EP_EnabledOnOptLevel0, registerSplitComparesPass); - -#if LLVM_VERSION_MAJOR >= 11 -static RegisterStandardPasses RegisterSplitComparesTransPassLTO( - PassManagerBuilder::EP_FullLinkTimeOptimizationLast, - registerSplitComparesPass); -#endif - diff --git a/llvm_mode/split-switches-pass.so.cc b/llvm_mode/split-switches-pass.so.cc deleted file mode 100644 index a79d4114..00000000 --- a/llvm_mode/split-switches-pass.so.cc +++ /dev/null @@ -1,447 +0,0 @@ -/* - * Copyright 2016 laf-intel - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -#include -#include -#include - -#include -#include -#include -#include - -#include "llvm/Config/llvm-config.h" - -#include "llvm/ADT/Statistic.h" -#include "llvm/IR/IRBuilder.h" -#include "llvm/IR/LegacyPassManager.h" -#include "llvm/IR/Module.h" -#include "llvm/Support/Debug.h" -#include "llvm/Support/raw_ostream.h" -#include "llvm/Transforms/IPO/PassManagerBuilder.h" -#include "llvm/Transforms/Utils/BasicBlockUtils.h" -#include "llvm/Pass.h" -#include "llvm/Analysis/ValueTracking.h" - -#include "llvm/IR/IRBuilder.h" -#if LLVM_VERSION_MAJOR > 3 || \ - (LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR > 4) - #include "llvm/IR/Verifier.h" - #include "llvm/IR/DebugInfo.h" -#else - #include "llvm/Analysis/Verifier.h" - #include "llvm/DebugInfo.h" - #define nullptr 0 -#endif - -#include -#include "afl-llvm-common.h" - -using namespace llvm; - -namespace { - -class SplitSwitchesTransform : public ModulePass { - - public: - static char ID; - SplitSwitchesTransform() : ModulePass(ID) { - - initInstrumentList(); - - } - - bool runOnModule(Module &M) override; - -#if LLVM_VERSION_MAJOR >= 4 - StringRef getPassName() const override { - -#else - const char *getPassName() const override { - -#endif - return "splits switch constructs"; - - } - - struct CaseExpr { - - ConstantInt *Val; - BasicBlock * BB; - - CaseExpr(ConstantInt *val = nullptr, BasicBlock *bb = nullptr) - : Val(val), BB(bb) { - - } - - }; - - typedef std::vector CaseVector; - - private: - bool splitSwitches(Module &M); - bool transformCmps(Module &M, const bool processStrcmp, - const bool processMemcmp); - BasicBlock *switchConvert(CaseVector Cases, std::vector bytesChecked, - BasicBlock *OrigBlock, BasicBlock *NewDefault, - Value *Val, unsigned level); - -}; - -} // namespace - -char SplitSwitchesTransform::ID = 0; - -/* switchConvert - Transform simple list of Cases into list of CaseRange's */ -BasicBlock *SplitSwitchesTransform::switchConvert( - CaseVector Cases, std::vector bytesChecked, BasicBlock *OrigBlock, - BasicBlock *NewDefault, Value *Val, unsigned level) { - - unsigned ValTypeBitWidth = Cases[0].Val->getBitWidth(); - IntegerType *ValType = - IntegerType::get(OrigBlock->getContext(), ValTypeBitWidth); - IntegerType * ByteType = IntegerType::get(OrigBlock->getContext(), 8); - unsigned BytesInValue = bytesChecked.size(); - std::vector setSizes; - std::vector > byteSets(BytesInValue, std::set()); - - assert(ValTypeBitWidth >= 8 && ValTypeBitWidth <= 64); - - /* for each of the possible cases we iterate over all bytes of the values - * build a set of possible values at each byte position in byteSets */ - for (CaseExpr &Case : Cases) { - - for (unsigned i = 0; i < BytesInValue; i++) { - - uint8_t byte = (Case.Val->getZExtValue() >> (i * 8)) & 0xFF; - byteSets[i].insert(byte); - - } - - } - - /* find the index of the first byte position that was not yet checked. then - * save the number of possible values at that byte position */ - unsigned smallestIndex = 0; - unsigned smallestSize = 257; - for (unsigned i = 0; i < byteSets.size(); i++) { - - if (bytesChecked[i]) continue; - if (byteSets[i].size() < smallestSize) { - - smallestIndex = i; - smallestSize = byteSets[i].size(); - - } - - } - - assert(bytesChecked[smallestIndex] == false); - - /* there are only smallestSize different bytes at index smallestIndex */ - - Instruction *Shift, *Trunc; - Function * F = OrigBlock->getParent(); - BasicBlock * NewNode = BasicBlock::Create(Val->getContext(), "NodeBlock", F); - Shift = BinaryOperator::Create(Instruction::LShr, Val, - ConstantInt::get(ValType, smallestIndex * 8)); - NewNode->getInstList().push_back(Shift); - - if (ValTypeBitWidth > 8) { - - Trunc = new TruncInst(Shift, ByteType); - NewNode->getInstList().push_back(Trunc); - - } else { - - /* not necessary to trunc */ - Trunc = Shift; - - } - - /* this is a trivial case, we can directly check for the byte, - * if the byte is not found go to default. if the byte was found - * mark the byte as checked. if this was the last byte to check - * we can finally execute the block belonging to this case */ - - if (smallestSize == 1) { - - uint8_t byte = *(byteSets[smallestIndex].begin()); - - /* insert instructions to check whether the value we are switching on is - * equal to byte */ - ICmpInst *Comp = - new ICmpInst(ICmpInst::ICMP_EQ, Trunc, ConstantInt::get(ByteType, byte), - "byteMatch"); - NewNode->getInstList().push_back(Comp); - - bytesChecked[smallestIndex] = true; - bool allBytesAreChecked = true; - - for (std::vector::iterator BCI = bytesChecked.begin(), - E = bytesChecked.end(); - BCI != E; ++BCI) { - - if (!*BCI) { - - allBytesAreChecked = false; - break; - - } - - } - - // if (std::all_of(bytesChecked.begin(), bytesChecked.end(), - // [](bool b) { return b; })) { - - if (allBytesAreChecked) { - - assert(Cases.size() == 1); - BranchInst::Create(Cases[0].BB, NewDefault, Comp, NewNode); - - /* we have to update the phi nodes! */ - for (BasicBlock::iterator I = Cases[0].BB->begin(); - I != Cases[0].BB->end(); ++I) { - - if (!isa(&*I)) { continue; } - PHINode *PN = cast(I); - - /* Only update the first occurrence. */ - unsigned Idx = 0, E = PN->getNumIncomingValues(); - for (; Idx != E; ++Idx) { - - if (PN->getIncomingBlock(Idx) == OrigBlock) { - - PN->setIncomingBlock(Idx, NewNode); - break; - - } - - } - - } - - } else { - - BasicBlock *BB = switchConvert(Cases, bytesChecked, OrigBlock, NewDefault, - Val, level + 1); - BranchInst::Create(BB, NewDefault, Comp, NewNode); - - } - - } - - /* there is no byte which we can directly check on, split the tree */ - else { - - std::vector byteVector; - std::copy(byteSets[smallestIndex].begin(), byteSets[smallestIndex].end(), - std::back_inserter(byteVector)); - std::sort(byteVector.begin(), byteVector.end()); - uint8_t pivot = byteVector[byteVector.size() / 2]; - - /* we already chose to divide the cases based on the value of byte at index - * smallestIndex the pivot value determines the threshold for the decicion; - * if a case value - * is smaller at this byte index move it to the LHS vector, otherwise to the - * RHS vector */ - - CaseVector LHSCases, RHSCases; - - for (CaseExpr &Case : Cases) { - - uint8_t byte = (Case.Val->getZExtValue() >> (smallestIndex * 8)) & 0xFF; - - if (byte < pivot) { - - LHSCases.push_back(Case); - - } else { - - RHSCases.push_back(Case); - - } - - } - - BasicBlock *LBB, *RBB; - LBB = switchConvert(LHSCases, bytesChecked, OrigBlock, NewDefault, Val, - level + 1); - RBB = switchConvert(RHSCases, bytesChecked, OrigBlock, NewDefault, Val, - level + 1); - - /* insert instructions to check whether the value we are switching on is - * equal to byte */ - ICmpInst *Comp = - new ICmpInst(ICmpInst::ICMP_ULT, Trunc, - ConstantInt::get(ByteType, pivot), "byteMatch"); - NewNode->getInstList().push_back(Comp); - BranchInst::Create(LBB, RBB, Comp, NewNode); - - } - - return NewNode; - -} - -bool SplitSwitchesTransform::splitSwitches(Module &M) { - -#if (LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR < 7) - LLVMContext &C = M.getContext(); -#endif - - std::vector switches; - - /* iterate over all functions, bbs and instruction and add - * all switches to switches vector for later processing */ - for (auto &F : M) { - - if (!isInInstrumentList(&F)) continue; - - for (auto &BB : F) { - - SwitchInst *switchInst = nullptr; - - if ((switchInst = dyn_cast(BB.getTerminator()))) { - - if (switchInst->getNumCases() < 1) continue; - switches.push_back(switchInst); - - } - - } - - } - - if (!switches.size()) return false; - if (!be_quiet) - errs() << "Rewriting " << switches.size() << " switch statements " - << "\n"; - - for (auto &SI : switches) { - - BasicBlock *CurBlock = SI->getParent(); - BasicBlock *OrigBlock = CurBlock; - Function * F = CurBlock->getParent(); - /* this is the value we are switching on */ - Value * Val = SI->getCondition(); - BasicBlock *Default = SI->getDefaultDest(); - unsigned bitw = Val->getType()->getIntegerBitWidth(); - - if (!be_quiet) - errs() << "switch: " << SI->getNumCases() << " cases " << bitw - << " bit\n"; - - /* If there is only the default destination or the condition checks 8 bit or - * less, don't bother with the code below. */ - if (!SI->getNumCases() || bitw <= 8) { - - if (!be_quiet) errs() << "skip trivial switch..\n"; - continue; - - } - - /* Create a new, empty default block so that the new hierarchy of - * if-then statements go to this and the PHI nodes are happy. - * if the default block is set as an unreachable we avoid creating one - * because will never be a valid target.*/ - BasicBlock *NewDefault = nullptr; - NewDefault = BasicBlock::Create(SI->getContext(), "NewDefault", F, Default); - BranchInst::Create(Default, NewDefault); - - /* Prepare cases vector. */ - CaseVector Cases; - for (SwitchInst::CaseIt i = SI->case_begin(), e = SI->case_end(); i != e; - ++i) -#if LLVM_VERSION_MAJOR < 5 - Cases.push_back(CaseExpr(i.getCaseValue(), i.getCaseSuccessor())); -#else - Cases.push_back(CaseExpr(i->getCaseValue(), i->getCaseSuccessor())); -#endif - /* bugfix thanks to pbst - * round up bytesChecked (in case getBitWidth() % 8 != 0) */ - std::vector bytesChecked((7 + Cases[0].Val->getBitWidth()) / 8, - false); - BasicBlock * SwitchBlock = - switchConvert(Cases, bytesChecked, OrigBlock, NewDefault, Val, 0); - - /* Branch to our shiny new if-then stuff... */ - BranchInst::Create(SwitchBlock, OrigBlock); - - /* We are now done with the switch instruction, delete it. */ - CurBlock->getInstList().erase(SI); - - /* we have to update the phi nodes! */ - for (BasicBlock::iterator I = Default->begin(); I != Default->end(); ++I) { - - if (!isa(&*I)) { continue; } - PHINode *PN = cast(I); - - /* Only update the first occurrence. */ - unsigned Idx = 0, E = PN->getNumIncomingValues(); - for (; Idx != E; ++Idx) { - - if (PN->getIncomingBlock(Idx) == OrigBlock) { - - PN->setIncomingBlock(Idx, NewDefault); - break; - - } - - } - - } - - } - - verifyModule(M); - return true; - -} - -bool SplitSwitchesTransform::runOnModule(Module &M) { - - if ((isatty(2) && getenv("AFL_QUIET") == NULL) || getenv("AFL_DEBUG") != NULL) - llvm::errs() << "Running split-switches-pass by laf.intel@gmail.com\n"; - else - be_quiet = 1; - splitSwitches(M); - verifyModule(M); - - return true; - -} - -static void registerSplitSwitchesTransPass(const PassManagerBuilder &, - legacy::PassManagerBase &PM) { - - auto p = new SplitSwitchesTransform(); - PM.add(p); - -} - -static RegisterStandardPasses RegisterSplitSwitchesTransPass( - PassManagerBuilder::EP_OptimizerLast, registerSplitSwitchesTransPass); - -static RegisterStandardPasses RegisterSplitSwitchesTransPass0( - PassManagerBuilder::EP_EnabledOnOptLevel0, registerSplitSwitchesTransPass); - -#if LLVM_VERSION_MAJOR >= 11 -static RegisterStandardPasses RegisterSplitSwitchesTransPassLTO( - PassManagerBuilder::EP_FullLinkTimeOptimizationLast, - registerSplitSwitchesTransPass); -#endif - diff --git a/src/afl-cc.c b/src/afl-cc.c new file mode 100644 index 00000000..e11ce40a --- /dev/null +++ b/src/afl-cc.c @@ -0,0 +1,1544 @@ +/* + american fuzzy lop++ - compiler instrumentation wrapper + ------------------------------------------------------- + + Written by Michal Zalewski, Laszlo Szekeres and Marc Heuse + + Copyright 2015, 2016 Google Inc. All rights reserved. + Copyright 2019-2020 AFLplusplus Project. All rights reserved. + + Licensed under the Apache License, Version 2.0 (the "License"); + you may not use this file except in compliance with the License. + You may obtain a copy of the License at: + + http://www.apache.org/licenses/LICENSE-2.0 + + */ + +#define AFL_MAIN + +#include "common.h" +#include "config.h" +#include "types.h" +#include "debug.h" +#include "alloc-inl.h" +#include "llvm-ngram-coverage.h" + +#include +#include +#include +#include +#include +#include +#include + +#if (LLVM_MAJOR - 0 == 0) + #undef LLVM_MAJOR +#endif +#if !defined(LLVM_MAJOR) + #define LLVM_MAJOR 0 +#endif + +static u8 * obj_path; /* Path to runtime libraries */ +static u8 **cc_params; /* Parameters passed to the real CC */ +static u32 cc_par_cnt = 1; /* Param count, including argv0 */ +static u8 llvm_fullpath[PATH_MAX]; +static u8 instrument_mode, instrument_opt_mode, ngram_size, lto_mode, + compiler_mode, plusplus_mode; +static u8 have_gcc, have_llvm, have_gcc_plugin, have_lto; +static u8 *lto_flag = AFL_CLANG_FLTO, *argvnull; +static u8 debug; +static u8 cwd[4096]; +static u8 cmplog_mode; +u8 use_stdin; /* dummy */ +// static u8 *march_opt = CFLAGS_OPT; + +enum { + + INSTURMENT_DEFAULT = 0, + INSTRUMENT_CLASSIC = 1, + INSTRUMENT_AFL = 1, + INSTRUMENT_PCGUARD = 2, + INSTRUMENT_INSTRIM = 3, + INSTRUMENT_CFG = 3, + INSTRUMENT_LTO = 4, + INSTRUMENT_OPT_CTX = 8, + INSTRUMENT_OPT_NGRAM = 16 + +}; + +char instrument_mode_string[18][18] = { + + "DEFAULT", "CLASSIC", "PCGUARD", "CFG", "LTO", "", "", "", "CTX", "", + "", "", "", "", "", "", "NGRAM", "" + +}; + +enum { + + UNSET = 0, + LTO = 1, + LLVM = 2, + GCC_PLUGIN = 3, + GCC = 4 + +}; + +char compiler_mode_string[6][12] = { + + "AUTOSELECT", "LLVM-LTO", "LLVM", "GCC_PLUGIN", + "GCC", "" + +}; + +u8 *getthecwd() { + + static u8 fail[] = ""; + if (getcwd(cwd, sizeof(cwd)) == NULL) return fail; + return cwd; + +} + +/* Try to find the runtime libraries. If that fails, abort. */ + +static u8 *find_object(u8 *obj, u8 *argv0) { + + u8 *afl_path = getenv("AFL_PATH"); + u8 *slash = NULL, *tmp; + + if (afl_path) { + +#ifdef __ANDROID__ + tmp = alloc_printf("%s/%s", afl_path, obj); +#else + tmp = alloc_printf("%s/%s", afl_path, obj); +#endif + + if (!access(tmp, R_OK)) { + + obj_path = afl_path; + return tmp; + + } + + ck_free(tmp); + + } + + if (argv0) slash = strrchr(argv0, '/'); + + if (slash) { + + u8 *dir; + + *slash = 0; + dir = ck_strdup(argv0); + *slash = '/'; + +#ifdef __ANDROID__ + tmp = alloc_printf("%s/%s", dir, obj); +#else + tmp = alloc_printf("%s/%s", dir, obj); +#endif + + if (!access(tmp, R_OK)) { + + obj_path = dir; + return tmp; + + } + + ck_free(tmp); + ck_free(dir); + + } + + tmp = alloc_printf("%s/%s", AFL_PATH, obj); +#ifdef __ANDROID__ + if (!access(tmp, R_OK)) { + +#else + if (!access(tmp, R_OK)) { + +#endif + + obj_path = AFL_PATH; + return tmp; + + } + + ck_free(tmp); + return NULL; + +} + +/* Try to find the runtime libraries. If that fails, abort. */ + +static void find_obj(u8 *argv0) { + + u8 *afl_path = getenv("AFL_PATH"); + u8 *slash, *tmp; + + if (afl_path) { + +#ifdef __ANDROID__ + tmp = alloc_printf("%s/afl-compiler-rt.so", afl_path); +#else + tmp = alloc_printf("%s/afl-compiler-rt.o", afl_path); +#endif + + if (!access(tmp, R_OK)) { + + obj_path = afl_path; + ck_free(tmp); + return; + + } + + ck_free(tmp); + + } + + slash = strrchr(argv0, '/'); + + if (slash) { + + u8 *dir; + + *slash = 0; + dir = ck_strdup(argv0); + *slash = '/'; + +#ifdef __ANDROID__ + tmp = alloc_printf("%s/afl-compiler-rt.so", dir); +#else + tmp = alloc_printf("%s/afl-compiler-rt.o", dir); +#endif + + if (!access(tmp, R_OK)) { + + obj_path = dir; + ck_free(tmp); + return; + + } + + ck_free(tmp); + ck_free(dir); + + } + +#ifdef __ANDROID__ + if (!access(AFL_PATH "/afl-compiler-rt.so", R_OK)) { + +#else + if (!access(AFL_PATH "/afl-compiler-rt.o", R_OK)) { + +#endif + + obj_path = AFL_PATH; + return; + + } + + FATAL( + "Unable to find 'afl-compiler-rt.o' or 'afl-llvm-pass.so'. Please set " + "AFL_PATH"); + +} + +/* Copy argv to cc_params, making the necessary edits. */ + +static void edit_params(u32 argc, char **argv, char **envp) { + + u8 fortify_set = 0, asan_set = 0, x_set = 0, bit_mode = 0, shared_linking = 0, + preprocessor_only = 0, have_unroll = 0, have_o = 0, have_pic = 0; + u8 *name; + + cc_params = ck_alloc((argc + 128) * sizeof(u8 *)); + + name = strrchr(argv[0], '/'); + if (!name) + name = argv[0]; + else + ++name; + + if (lto_mode) { + + if (lto_flag[0] != '-') + FATAL( + "Using afl-clang-lto is not possible because Makefile magic did not " + "identify the correct -flto flag"); + else + compiler_mode = LTO; + + } + + if (plusplus_mode) { + + u8 *alt_cxx = getenv("AFL_CXX"); + + if (!alt_cxx) { + + if (compiler_mode >= GCC_PLUGIN) { + + alt_cxx = "g++"; + + } else { + + if (USE_BINDIR) + snprintf(llvm_fullpath, sizeof(llvm_fullpath), "%s/clang++", + LLVM_BINDIR); + else + snprintf(llvm_fullpath, sizeof(llvm_fullpath), CLANGPP_BIN); + alt_cxx = llvm_fullpath; + + } + + } + + cc_params[0] = alt_cxx; + + } else { + + u8 *alt_cc = getenv("AFL_CC"); + + if (!alt_cc) { + + if (compiler_mode >= GCC_PLUGIN) { + + alt_cc = "gcc"; + + } else { + + if (USE_BINDIR) + snprintf(llvm_fullpath, sizeof(llvm_fullpath), "%s/clang", + LLVM_BINDIR); + else + snprintf(llvm_fullpath, sizeof(llvm_fullpath), CLANGPP_BIN); + alt_cc = llvm_fullpath; + + } + + } + + cc_params[0] = alt_cc; + + } + + if (compiler_mode == GCC) { + + cc_params[cc_par_cnt++] = "-B"; + cc_params[cc_par_cnt++] = obj_path; + + } + + if (compiler_mode == GCC_PLUGIN) { + + char *fplugin_arg = + alloc_printf("-fplugin=%s", find_object("afl-gcc-pass.so", argvnull)); + cc_params[cc_par_cnt++] = fplugin_arg; + + } + + if (compiler_mode == LLVM || compiler_mode == LTO) { + + cc_params[cc_par_cnt++] = "-Wno-unused-command-line-argument"; + + if (lto_mode && plusplus_mode) + cc_params[cc_par_cnt++] = "-lc++"; // needed by fuzzbench, early + + if (lto_mode) { + + if (getenv("AFL_LLVM_INSTRUMENT_FILE") != NULL || + getenv("AFL_LLVM_WHITELIST") || getenv("AFL_LLVM_ALLOWLIST") || + getenv("AFL_LLVM_DENYLIST") || getenv("AFL_LLVM_BLOCKLIST")) { + + cc_params[cc_par_cnt++] = "-Xclang"; + cc_params[cc_par_cnt++] = "-load"; + cc_params[cc_par_cnt++] = "-Xclang"; + cc_params[cc_par_cnt++] = + alloc_printf("%s/afl-llvm-lto-instrumentlist.so", obj_path); + + } + + } + + if (getenv("AFL_LLVM_DICT2FILE")) { + + cc_params[cc_par_cnt++] = "-Xclang"; + cc_params[cc_par_cnt++] = "-load"; + cc_params[cc_par_cnt++] = "-Xclang"; + cc_params[cc_par_cnt++] = + alloc_printf("%s/afl-llvm-dict2file.so", obj_path); + + } + + // laf + if (getenv("LAF_SPLIT_SWITCHES") || getenv("AFL_LLVM_LAF_SPLIT_SWITCHES")) { + + if (lto_mode) { + + cc_params[cc_par_cnt++] = alloc_printf( + "-Wl,-mllvm=-load=%s/split-switches-pass.so", obj_path); + + } else { + + cc_params[cc_par_cnt++] = "-Xclang"; + cc_params[cc_par_cnt++] = "-load"; + cc_params[cc_par_cnt++] = "-Xclang"; + cc_params[cc_par_cnt++] = + alloc_printf("%s/split-switches-pass.so", obj_path); + + } + + } + + if (getenv("LAF_TRANSFORM_COMPARES") || + getenv("AFL_LLVM_LAF_TRANSFORM_COMPARES")) { + + if (lto_mode) { + + cc_params[cc_par_cnt++] = alloc_printf( + "-Wl,-mllvm=-load=%s/compare-transform-pass.so", obj_path); + + } else { + + cc_params[cc_par_cnt++] = "-Xclang"; + cc_params[cc_par_cnt++] = "-load"; + cc_params[cc_par_cnt++] = "-Xclang"; + cc_params[cc_par_cnt++] = + alloc_printf("%s/compare-transform-pass.so", obj_path); + + } + + } + + if (getenv("LAF_SPLIT_COMPARES") || getenv("AFL_LLVM_LAF_SPLIT_COMPARES") || + getenv("AFL_LLVM_LAF_SPLIT_FLOATS")) { + + if (lto_mode) { + + cc_params[cc_par_cnt++] = alloc_printf( + "-Wl,-mllvm=-load=%s/split-compares-pass.so", obj_path); + + } else { + + cc_params[cc_par_cnt++] = "-Xclang"; + cc_params[cc_par_cnt++] = "-load"; + cc_params[cc_par_cnt++] = "-Xclang"; + cc_params[cc_par_cnt++] = + alloc_printf("%s/split-compares-pass.so", obj_path); + + } + + } + + // /laf + + unsetenv("AFL_LD"); + unsetenv("AFL_LD_CALLER"); + if (cmplog_mode) { + + if (lto_mode) { + + cc_params[cc_par_cnt++] = alloc_printf( + "-Wl,-mllvm=-load=%s/cmplog-routines-pass.so", obj_path); + cc_params[cc_par_cnt++] = alloc_printf( + "-Wl,-mllvm=-load=%s/split-switches-pass.so", obj_path); + cc_params[cc_par_cnt++] = alloc_printf( + "-Wl,-mllvm=-load=%s/cmplog-instructions-pass.so", obj_path); + + } else { + + cc_params[cc_par_cnt++] = "-Xclang"; + cc_params[cc_par_cnt++] = "-load"; + cc_params[cc_par_cnt++] = "-Xclang"; + cc_params[cc_par_cnt++] = + alloc_printf("%s/cmplog-routines-pass.so", obj_path); + + // reuse split switches from laf + cc_params[cc_par_cnt++] = "-Xclang"; + cc_params[cc_par_cnt++] = "-load"; + cc_params[cc_par_cnt++] = "-Xclang"; + cc_params[cc_par_cnt++] = + alloc_printf("%s/split-switches-pass.so", obj_path); + + cc_params[cc_par_cnt++] = "-Xclang"; + cc_params[cc_par_cnt++] = "-load"; + cc_params[cc_par_cnt++] = "-Xclang"; + cc_params[cc_par_cnt++] = + alloc_printf("%s/cmplog-instructions-pass.so", obj_path); + + } + + cc_params[cc_par_cnt++] = "-fno-inline"; + + } + + if (lto_mode) { + + u8 *ld_path = strdup(AFL_REAL_LD); + if (!*ld_path) ld_path = "ld.lld"; +#if defined(AFL_CLANG_LDPATH) && LLVM_MAJOR >= 12 + cc_params[cc_par_cnt++] = alloc_printf("--ld-path=%s", ld_path); +#else + cc_params[cc_par_cnt++] = alloc_printf("-fuse-ld=%s", ld_path); +#endif + + cc_params[cc_par_cnt++] = "-Wl,--allow-multiple-definition"; + + if (instrument_mode == INSTRUMENT_CFG) + cc_params[cc_par_cnt++] = alloc_printf( + "-Wl,-mllvm=-load=%s/SanitizerCoverageLTO.so", obj_path); + else + + cc_params[cc_par_cnt++] = alloc_printf( + "-Wl,-mllvm=-load=%s/afl-llvm-lto-instrumentation.so", obj_path); + cc_params[cc_par_cnt++] = lto_flag; + + } else { + + if (instrument_mode == INSTRUMENT_PCGUARD) { + +#if LLVM_MAJOR >= 4 + cc_params[cc_par_cnt++] = + "-fsanitize-coverage=trace-pc-guard"; // edge coverage by default +#else + FATAL("pcguard instrumentation requires llvm 4.0.1+"); +#endif + + } else { + + cc_params[cc_par_cnt++] = "-Xclang"; + cc_params[cc_par_cnt++] = "-load"; + cc_params[cc_par_cnt++] = "-Xclang"; + if (instrument_mode == INSTRUMENT_CFG) + cc_params[cc_par_cnt++] = + alloc_printf("%s/libLLVMInsTrim.so", obj_path); + else + cc_params[cc_par_cnt++] = + alloc_printf("%s/afl-llvm-pass.so", obj_path); + + } + + } + + // cc_params[cc_par_cnt++] = "-Qunused-arguments"; + + // in case LLVM is installed not via a package manager or "make install" + // e.g. compiled download or compiled from github then its ./lib directory + // might not be in the search path. Add it if so. + u8 *libdir = strdup(LLVM_LIBDIR); + if (plusplus_mode && strlen(libdir) && strncmp(libdir, "/usr", 4) && + strncmp(libdir, "/lib", 4)) { + + cc_params[cc_par_cnt++] = "-rpath"; + cc_params[cc_par_cnt++] = libdir; + + } else { + + free(libdir); + + } + + u32 idx; + if (lto_mode && argc > 1) { + + for (idx = 1; idx < argc; idx++) { + + if (!strncasecmp(argv[idx], "-fpic", 5)) have_pic = 1; + + } + + if (!have_pic) cc_params[cc_par_cnt++] = "-fPIC"; + + } + + } + + /* Detect stray -v calls from ./configure scripts. */ + + while (--argc) { + + u8 *cur = *(++argv); + + if (!strncmp(cur, "--afl", 5)) continue; + if (lto_mode && !strncmp(cur, "-fuse-ld=", 9)) continue; + if (lto_mode && !strncmp(cur, "--ld-path=", 10)) continue; + if (!strcmp(cur, "-Wl,-z,defs") || !strcmp(cur, "-Wl,--no-undefined")) + continue; + + if (!strcmp(cur, "-m32")) bit_mode = 32; + if (!strcmp(cur, "armv7a-linux-androideabi")) bit_mode = 32; + if (!strcmp(cur, "-m64")) bit_mode = 64; + + if (!strcmp(cur, "-fsanitize=address") || !strcmp(cur, "-fsanitize=memory")) + asan_set = 1; + + if (strstr(cur, "FORTIFY_SOURCE")) fortify_set = 1; + + if (!strcmp(cur, "-x")) x_set = 1; + if (!strcmp(cur, "-E")) preprocessor_only = 1; + if (!strcmp(cur, "-shared")) shared_linking = 1; + + if (!strncmp(cur, "-O", 2)) have_o = 1; + if (!strncmp(cur, "-f", 2) && strstr(cur, "unroll-loop")) have_unroll = 1; + + cc_params[cc_par_cnt++] = cur; + + } + + if (getenv("AFL_HARDEN")) { + + cc_params[cc_par_cnt++] = "-fstack-protector-all"; + + if (!fortify_set) cc_params[cc_par_cnt++] = "-D_FORTIFY_SOURCE=2"; + + } + + if (!asan_set) { + + if (getenv("AFL_USE_ASAN")) { + + if (getenv("AFL_USE_MSAN")) FATAL("ASAN and MSAN are mutually exclusive"); + + if (getenv("AFL_HARDEN")) + FATAL("ASAN and AFL_HARDEN are mutually exclusive"); + + cc_params[cc_par_cnt++] = "-U_FORTIFY_SOURCE"; + cc_params[cc_par_cnt++] = "-fsanitize=address"; + + } else if (getenv("AFL_USE_MSAN")) { + + if (getenv("AFL_USE_ASAN")) FATAL("ASAN and MSAN are mutually exclusive"); + + if (getenv("AFL_HARDEN")) + FATAL("MSAN and AFL_HARDEN are mutually exclusive"); + + cc_params[cc_par_cnt++] = "-U_FORTIFY_SOURCE"; + cc_params[cc_par_cnt++] = "-fsanitize=memory"; + + } + + } + + if (getenv("AFL_USE_UBSAN")) { + + cc_params[cc_par_cnt++] = "-fsanitize=undefined"; + cc_params[cc_par_cnt++] = "-fsanitize-undefined-trap-on-error"; + cc_params[cc_par_cnt++] = "-fno-sanitize-recover=all"; + + } + + if (getenv("AFL_USE_CFISAN")) { + + if (!lto_mode) { + + uint32_t i = 0, found = 0; + while (envp[i] != NULL && !found) + if (strncmp("-flto", envp[i++], 5) == 0) found = 1; + if (!found) cc_params[cc_par_cnt++] = "-flto"; + + } + + cc_params[cc_par_cnt++] = "-fsanitize=cfi"; + cc_params[cc_par_cnt++] = "-fvisibility=hidden"; + + } + + if (!getenv("AFL_DONT_OPTIMIZE")) { + + cc_params[cc_par_cnt++] = "-g"; + if (!have_o) cc_params[cc_par_cnt++] = "-O3"; + if (!have_unroll) cc_params[cc_par_cnt++] = "-funroll-loops"; + // if (strlen(march_opt) > 1 && march_opt[0] == '-') + // cc_params[cc_par_cnt++] = march_opt; + + } + + if (getenv("AFL_NO_BUILTIN") || getenv("AFL_LLVM_LAF_TRANSFORM_COMPARES") || + getenv("LAF_TRANSFORM_COMPARES") || lto_mode) { + + cc_params[cc_par_cnt++] = "-fno-builtin-strcmp"; + cc_params[cc_par_cnt++] = "-fno-builtin-strncmp"; + cc_params[cc_par_cnt++] = "-fno-builtin-strcasecmp"; + cc_params[cc_par_cnt++] = "-fno-builtin-strncasecmp"; + cc_params[cc_par_cnt++] = "-fno-builtin-memcmp"; + cc_params[cc_par_cnt++] = "-fno-builtin-bcmp"; + cc_params[cc_par_cnt++] = "-fno-builtin-strstr"; + cc_params[cc_par_cnt++] = "-fno-builtin-strcasestr"; + + } + +#if defined(USEMMAP) && !defined(__HAIKU__) + cc_params[cc_par_cnt++] = "-lrt"; +#endif + + cc_params[cc_par_cnt++] = "-D__AFL_HAVE_MANUAL_CONTROL=1"; + cc_params[cc_par_cnt++] = "-D__AFL_COMPILER=1"; + cc_params[cc_par_cnt++] = "-DFUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION=1"; + + /* When the user tries to use persistent or deferred forkserver modes by + appending a single line to the program, we want to reliably inject a + signature into the binary (to be picked up by afl-fuzz) and we want + to call a function from the runtime .o file. This is unnecessarily + painful for three reasons: + + 1) We need to convince the compiler not to optimize out the signature. + This is done with __attribute__((used)). + + 2) We need to convince the linker, when called with -Wl,--gc-sections, + not to do the same. This is done by forcing an assignment to a + 'volatile' pointer. + + 3) We need to declare __afl_persistent_loop() in the global namespace, + but doing this within a method in a class is hard - :: and extern "C" + are forbidden and __attribute__((alias(...))) doesn't work. Hence the + __asm__ aliasing trick. + + */ + + cc_params[cc_par_cnt++] = + "-D__AFL_FUZZ_INIT()=" + "int __afl_sharedmem_fuzzing = 1;" + "extern unsigned int *__afl_fuzz_len;" + "extern unsigned char *__afl_fuzz_ptr;" + "unsigned char __afl_fuzz_alt[1024000];" + "unsigned char *__afl_fuzz_alt_ptr = __afl_fuzz_alt;"; + cc_params[cc_par_cnt++] = + "-D__AFL_FUZZ_TESTCASE_BUF=(__afl_fuzz_ptr ? __afl_fuzz_ptr : " + "__afl_fuzz_alt_ptr)"; + cc_params[cc_par_cnt++] = + "-D__AFL_FUZZ_TESTCASE_LEN=(__afl_fuzz_ptr ? *__afl_fuzz_len : " + "(*__afl_fuzz_len = read(0, __afl_fuzz_alt_ptr, 1024000)) == 0xffffffff " + "? 0 : *__afl_fuzz_len)"; + + cc_params[cc_par_cnt++] = + "-D__AFL_LOOP(_A)=" + "({ static volatile char *_B __attribute__((used)); " + " _B = (char*)\"" PERSIST_SIG + "\"; " +#ifdef __APPLE__ + "__attribute__((visibility(\"default\"))) " + "int _L(unsigned int) __asm__(\"___afl_persistent_loop\"); " +#else + "__attribute__((visibility(\"default\"))) " + "int _L(unsigned int) __asm__(\"__afl_persistent_loop\"); " +#endif /* ^__APPLE__ */ + "_L(_A); })"; + + cc_params[cc_par_cnt++] = + "-D__AFL_INIT()=" + "do { static volatile char *_A __attribute__((used)); " + " _A = (char*)\"" DEFER_SIG + "\"; " +#ifdef __APPLE__ + "__attribute__((visibility(\"default\"))) " + "void _I(void) __asm__(\"___afl_manual_init\"); " +#else + "__attribute__((visibility(\"default\"))) " + "void _I(void) __asm__(\"__afl_manual_init\"); " +#endif /* ^__APPLE__ */ + "_I(); } while (0)"; + + if (x_set) { + + cc_params[cc_par_cnt++] = "-x"; + cc_params[cc_par_cnt++] = "none"; + + } + + if (preprocessor_only) { + + /* In the preprocessor_only case (-E), we are not actually compiling at + all but requesting the compiler to output preprocessed sources only. + We must not add the runtime in this case because the compiler will + simply output its binary content back on stdout, breaking any build + systems that rely on a separate source preprocessing step. */ + cc_params[cc_par_cnt] = NULL; + return; + + } + +#ifndef __ANDROID__ + + if (compiler_mode != GCC) { + + switch (bit_mode) { + + case 0: + cc_params[cc_par_cnt++] = + alloc_printf("%s/afl-compiler-rt.o", obj_path); + if (lto_mode) + cc_params[cc_par_cnt++] = + alloc_printf("%s/afl-llvm-rt-lto.o", obj_path); + break; + + case 32: + cc_params[cc_par_cnt++] = + alloc_printf("%s/afl-compiler-rt-32.o", obj_path); + if (access(cc_params[cc_par_cnt - 1], R_OK)) + FATAL("-m32 is not supported by your compiler"); + if (lto_mode) { + + cc_params[cc_par_cnt++] = + alloc_printf("%s/afl-llvm-rt-lto-32.o", obj_path); + if (access(cc_params[cc_par_cnt - 1], R_OK)) + FATAL("-m32 is not supported by your compiler"); + + } + + break; + + case 64: + cc_params[cc_par_cnt++] = + alloc_printf("%s/afl-compiler-rt-64.o", obj_path); + if (access(cc_params[cc_par_cnt - 1], R_OK)) + FATAL("-m64 is not supported by your compiler"); + if (lto_mode) { + + cc_params[cc_par_cnt++] = + alloc_printf("%s/afl-llvm-rt-lto-64.o", obj_path); + if (access(cc_params[cc_par_cnt - 1], R_OK)) + FATAL("-m64 is not supported by your compiler"); + + } + + break; + + } + + #ifndef __APPLE__ + if (!shared_linking) + cc_params[cc_par_cnt++] = + alloc_printf("-Wl,--dynamic-list=%s/dynamic_list.txt", obj_path); + #endif + + } + +#endif + + cc_params[cc_par_cnt] = NULL; + +} + +/* Main entry point */ + +int main(int argc, char **argv, char **envp) { + + int i; + char *callname = argv[0], *ptr = NULL; + + if (getenv("AFL_DEBUG")) { + + debug = 1; + if (strcmp(getenv("AFL_DEBUG"), "0") == 0) unsetenv("AFL_DEBUG"); + + } else if (getenv("AFL_QUIET")) + + be_quiet = 1; + + if ((ptr = strrchr(callname, '/')) != NULL) callname = ptr + 1; + argvnull = (u8 *)argv[0]; + check_environment_vars(envp); + + if ((ptr = find_object("as", argv[0])) != NULL) { + + have_gcc = 1; + ck_free(ptr); + + } + +#if (LLVM_MAJOR > 2) + + if ((ptr = find_object("SanitizerCoverageLTO.so", argv[0])) != NULL) { + + have_lto = 1; + ck_free(ptr); + + } + + if ((ptr = find_object("cmplog-routines-pass.so", argv[0])) != NULL) { + + have_llvm = 1; + ck_free(ptr); + + } + +#endif + + if ((ptr = find_object("afl-gcc-pass.so", argv[0])) != NULL) { + + have_gcc_plugin = 1; + ck_free(ptr); + + } + +#if (LLVM_MAJOR > 2) + + if (strncmp(callname, "afl-clang-fast", 14) == 0) { + + compiler_mode = LLVM; + + } else if (strncmp(callname, "afl-clang-lto", 13) == 0 || + + strncmp(callname, "afl-lto", 7) == 0) { + + compiler_mode = LTO; + + } else + +#endif + if (strncmp(callname, "afl-gcc-fast", 12) == 0 || + + strncmp(callname, "afl-g++-fast", 12) == 0) { + + compiler_mode = GCC_PLUGIN; + + } else if (strncmp(callname, "afl-gcc", 7) == 0 || + + strncmp(callname, "afl-g++", 7) == 0) { + + compiler_mode = GCC; + + } + + if ((ptr = getenv("AFL_CC_COMPILER"))) { + + if (compiler_mode) { + + WARNF( + "\"AFL_CC_COMPILER\" is set but a specific compiler was already " + "selected by command line parameter or symlink, ignoring the " + "environment variable!"); + + } else { + + if (strncasecmp(ptr, "LTO", 3) == 0) { + + compiler_mode = LTO; + + } else if (strncasecmp(ptr, "LLVM", 4) == 0) { + + compiler_mode = LLVM; + + } else if (strncasecmp(ptr, "GCC_P", 5) == 0 || + + strncasecmp(ptr, "GCC-P", 5) == 0 || + strncasecmp(ptr, "GCCP", 4) == 0) { + + compiler_mode = GCC_PLUGIN; + + } else if (strcasecmp(ptr, "GCC") == 0) { + + compiler_mode = GCC; + + } else + + FATAL("Unknown AFL_CC_COMPILER mode: %s\n", ptr); + + } + + } + + for (i = 1; i < argc; i++) { + + if (strncmp(argv[i], "--afl", 5) == 0) { + + if (compiler_mode) + WARNF( + "--afl-... compiler mode supersedes the AFL_CC_COMPILER and " + "symlink compiler selection!"); + + ptr = argv[i]; + ptr += 5; + while (*ptr == '-') + ptr++; + + if (strncasecmp(ptr, "LTO", 3) == 0) { + + compiler_mode = LTO; + + } else if (strncasecmp(ptr, "LLVM", 4) == 0) { + + compiler_mode = LLVM; + + } else if (strncasecmp(ptr, "GCC_P", 5) == 0 || + + strncasecmp(ptr, "GCC-P", 5) == 0 || + strncasecmp(ptr, "GCCP", 4) == 0) { + + compiler_mode = GCC_PLUGIN; + + } else if (strcasecmp(ptr, "GCC") == 0) { + + compiler_mode = GCC; + + } else + + FATAL("Unknown --afl-... compiler mode: %s\n", argv[i]); + + } + + } + + if (strlen(callname) > 2 && + (strncmp(callname + strlen(callname) - 2, "++", 2) == 0 || + strstr(callname, "-g++") != NULL)) + plusplus_mode = 1; + + if (getenv("USE_TRACE_PC") || getenv("AFL_USE_TRACE_PC") || + getenv("AFL_LLVM_USE_TRACE_PC") || getenv("AFL_TRACE_PC")) { + + if (instrument_mode == 0) + instrument_mode = INSTRUMENT_PCGUARD; + else if (instrument_mode != INSTRUMENT_PCGUARD) + FATAL("you can not set AFL_LLVM_INSTRUMENT and AFL_TRACE_PC together"); + + } + + if ((getenv("AFL_LLVM_INSTRUMENT_FILE") != NULL || + getenv("AFL_LLVM_WHITELIST") || getenv("AFL_LLVM_ALLOWLIST") || + getenv("AFL_LLVM_DENYLIST") || getenv("AFL_LLVM_BLOCKLIST")) && + getenv("AFL_DONT_OPTIMIZE")) + WARNF( + "AFL_LLVM_ALLOWLIST/DENYLIST and AFL_DONT_OPTIMIZE cannot be combined " + "for file matching, only function matching!"); + + if (getenv("AFL_LLVM_INSTRIM") || getenv("INSTRIM") || + getenv("INSTRIM_LIB")) { + + if (instrument_mode == 0) + instrument_mode = INSTRUMENT_CFG; + else if (instrument_mode != INSTRUMENT_CFG) + FATAL( + "you can not set AFL_LLVM_INSTRUMENT and AFL_LLVM_INSTRIM together"); + + } + + if (getenv("AFL_LLVM_CTX")) instrument_opt_mode |= INSTRUMENT_OPT_CTX; + + if (getenv("AFL_LLVM_NGRAM_SIZE")) { + + instrument_opt_mode |= INSTRUMENT_OPT_NGRAM; + ngram_size = atoi(getenv("AFL_LLVM_NGRAM_SIZE")); + if (ngram_size < 2 || ngram_size > NGRAM_SIZE_MAX) + FATAL( + "NGRAM instrumentation mode must be between 2 and NGRAM_SIZE_MAX " + "(%u)", + NGRAM_SIZE_MAX); + + } + + if (getenv("AFL_LLVM_INSTRUMENT")) { + + u8 *ptr = strtok(getenv("AFL_LLVM_INSTRUMENT"), ":,;"); + + while (ptr) { + + if (strncasecmp(ptr, "afl", strlen("afl")) == 0 || + strncasecmp(ptr, "classic", strlen("classic")) == 0) { + + if (instrument_mode == INSTRUMENT_LTO) { + + instrument_mode = INSTRUMENT_CLASSIC; + lto_mode = 1; + + } else if (!instrument_mode || instrument_mode == INSTRUMENT_AFL) + + instrument_mode = INSTRUMENT_AFL; + else + FATAL("main instrumentation mode already set with %s", + instrument_mode_string[instrument_mode]); + + } + + if (strncasecmp(ptr, "pc-guard", strlen("pc-guard")) == 0 || + strncasecmp(ptr, "pcguard", strlen("pcguard")) == 0) { + + if (!instrument_mode || instrument_mode == INSTRUMENT_PCGUARD) + instrument_mode = INSTRUMENT_PCGUARD; + else + FATAL("main instrumentation mode already set with %s", + instrument_mode_string[instrument_mode]); + + } + + if (strncasecmp(ptr, "cfg", strlen("cfg")) == 0 || + strncasecmp(ptr, "instrim", strlen("instrim")) == 0) { + + if (instrument_mode == INSTRUMENT_LTO) { + + instrument_mode = INSTRUMENT_CFG; + lto_mode = 1; + + } else if (!instrument_mode || instrument_mode == INSTRUMENT_CFG) + + instrument_mode = INSTRUMENT_CFG; + else + FATAL("main instrumentation mode already set with %s", + instrument_mode_string[instrument_mode]); + + } + + if (strncasecmp(ptr, "lto", strlen("lto")) == 0) { + + lto_mode = 1; + if (!instrument_mode || instrument_mode == INSTRUMENT_LTO) + instrument_mode = INSTRUMENT_LTO; + else if (instrument_mode != INSTRUMENT_CFG) + FATAL("main instrumentation mode already set with %s", + instrument_mode_string[instrument_mode]); + + } + + if (strncasecmp(ptr, "ctx", strlen("ctx")) == 0) { + + instrument_opt_mode |= INSTRUMENT_OPT_CTX; + setenv("AFL_LLVM_CTX", "1", 1); + + } + + if (strncasecmp(ptr, "ngram", strlen("ngram")) == 0) { + + ptr += strlen("ngram"); + while (*ptr && (*ptr < '0' || *ptr > '9')) + ptr++; + + if (!*ptr) { + + if ((ptr = getenv("AFL_LLVM_NGRAM_SIZE")) != NULL) + FATAL( + "you must set the NGRAM size with (e.g. for value 2) " + "AFL_LLVM_INSTRUMENT=ngram-2"); + + } + + ngram_size = atoi(ptr); + if (ngram_size < 2 || ngram_size > NGRAM_SIZE_MAX) + FATAL( + "NGRAM instrumentation option must be between 2 and " + "NGRAM_SIZE_MAX " + "(%u)", + NGRAM_SIZE_MAX); + instrument_opt_mode |= (INSTRUMENT_OPT_NGRAM); + ptr = alloc_printf("%u", ngram_size); + setenv("AFL_LLVM_NGRAM_SIZE", ptr, 1); + + } + + ptr = strtok(NULL, ":,;"); + + } + + } + + if (!compiler_mode) { + + // lto is not a default because outside of afl-cc RANLIB and AR have to + // be set to llvm versions so this would work + if (have_llvm) + compiler_mode = LLVM; + else if (have_gcc_plugin) + compiler_mode = GCC_PLUGIN; + else if (have_gcc) + compiler_mode = GCC; + else if (have_lto) + compiler_mode = LTO; + else + FATAL("no compiler mode available"); + + } + + if (argc < 2 || strncmp(argv[1], "-h", 2) == 0) { + + char *fp; + fp = realpath(argv[0], NULL); + + printf("afl-cc" VERSION + " by Michal Zalewski, Laszlo Szekeres, Marc Heuse\n"); + + SAYF( + "\n" + "afl-cc/afl-c++ [options]\n" + "\n" + "This is a helper application for afl-fuzz. It serves as a drop-in " + "replacement\n" + "for gcc and clang, letting you recompile third-party code with the " + "required\n" + "runtime instrumentation. A common use pattern would be one of the " + "following:\n\n" + + " CC=afl-cc CXX=afl-c++ ./configure --disable-shared\n" + " cmake -DCMAKE_C_COMPILERC=afl-cc -DCMAKE_CXX_COMPILER=afl-c++ .\n" + " CC=afl-cc CXX=afl-c++ meson\n\n"); + + SAYF( + " |---------------- FEATURES " + "---------------|\n" + "MODES: NCC PERSIST SNAP DICT LAF " + "CMPLOG SELECT\n" + " [LTO] llvm LTO: %s%s\n" + " PCGUARD DEFAULT yes yes yes yes yes yes " + " yes\n" + " CLASSIC yes yes yes yes yes yes " + " yes\n" + " [LLVM] llvm: %s%s\n" + " PCGUARD %s yes yes yes module yes yes " + "extern\n" + " CLASSIC %s no yes yes module yes yes " + "yes\n" + " - NORMAL\n" + " - CTX\n" + " - NGRAM-{2-16}\n" + " INSTRIM no yes yes module yes yes " + " yes\n" + " - NORMAL\n" + " - CTX\n" + " - NGRAM-{2-16}\n" + " [GCC_PLUGIN] gcc plugin: %s%s\n" + " CLASSIC DEFAULT no yes yes no no no " + " simple\n" + " [GCC] simple gcc: %s%s\n" + " CLASSIC DEFAULT no no no no no no " + " no\n\n", + have_lto ? "AVAILABLE" : "unavailable!", + compiler_mode == LTO ? " [SELECTED]" : "", + have_llvm ? "AVAILABLE" : "unavailable!", + compiler_mode == LLVM ? " [SELECTED]" : "", + LLVM_MAJOR > 6 ? "DEFAULT" : " ", + LLVM_MAJOR > 6 ? " " : "DEFAULT", + have_gcc_plugin ? "AVAILABLE" : "unavailable!", + compiler_mode == GCC_PLUGIN ? " [SELECTED]" : "", + have_gcc ? "AVAILABLE" : "unavailable!", + compiler_mode == GCC ? " [SELECTED]" : ""); + + SAYF( + "Modes:\n" + " To select the compiler mode use a symlink version (e.g. " + "afl-clang-fast), set\n" + " the environment variable AFL_CC_COMPILER to a mode (e.g. LLVM) or " + "use the\n" + " command line parameter --afl-MODE (e.g. --afl-llvm). If none is " + "selected,\n" + " afl-cc will select the best available (LLVM -> GCC_PLUGIN -> GCC).\n" + " The best is LTO but it often needs RANLIB and AR settings outside " + "of afl-cc.\n\n"); + + SAYF( + "Sub-Modes: (set via env AFL_LLVM_INSTRUMENT, afl-cc selects the best " + "available)\n" + " PCGUARD: Dominator tree instrumentation (best!) (README.llvm.md)\n" + " CLASSIC: decision target instrumentation (README.llvm.md)\n" + " CTX: CLASSIC + callee context (instrumentation/README.ctx.md)\n" + " NGRAM-x: CLASSIC + previous path " + "((instrumentation/README.ngram.md)\n" + " INSTRIM: Dominator tree (for LLVM <= 6.0) " + "(instrumentation/README.instrim.md)\n\n"); + + SAYF( + "Features: (see documentation links)\n" + " NCC: non-colliding coverage [automatic] (that is an amazing " + "thing!)\n" + " (instrumentation/README.lto.md)\n" + " PERSIST: persistent mode support [code] (huge speed increase!)\n" + " (instrumentation/README.persistent_mode.md)\n" + " SNAP: linux lkm snapshot module support [automatic] (speed " + "increase)\n" + " (https://github.com/AFLplusplus/AFL-Snapshot-LKM/)\n" + " DICT: dictionary in the target [yes=automatic or llvm module " + "pass]\n" + " (instrumentation/README.lto.md + " + "instrumentation/README.llvm.md)\n" + " LAF: comparison splitting [env] " + "(instrumentation/README.laf-intel.md)\n" + " CMPLOG: input2state exploration [env] " + "(instrumentation/README.cmplog.md)\n" + " SELECT: selective instrumentation (allow/deny) on filename or " + "function [env]\n" + " (instrumentation/README.instrument_list.md)\n\n"); + + if (argc < 2 || strncmp(argv[1], "-hh", 3)) { + + SAYF( + "To see all environment variables for the configuration of afl-cc " + "use \"-hh\".\n"); + + } else { + + SAYF( + "Environment variables used:\n" + " AFL_CC: path to the C compiler to use\n" + " AFL_CXX: path to the C++ compiler to use\n" + " AFL_DEBUG: enable developer debugging output\n" + " AFL_DONT_OPTIMIZE: disable optimization instead of -O3\n" + " AFL_HARDEN: adds code hardening to catch memory bugs\n" + " AFL_INST_RATIO: percentage of branches to instrument\n" +#if LLVM_MAJOR < 9 + " AFL_LLVM_NOT_ZERO: use cycling trace counters that skip zero\n" +#else + " AFL_LLVM_SKIP_NEVERZERO: do not skip zero on trace counters\n" +#endif + " AFL_LLVM_DICT2FILE: generate an afl dictionary based on found " + "comparisons\n" + " AFL_LLVM_LAF_ALL: enables all LAF splits/transforms\n" + " AFL_LLVM_LAF_SPLIT_COMPARES: enable cascaded comparisons\n" + " AFL_LLVM_LAF_SPLIT_COMPARES_BITW: size limit (default 8)\n" + " AFL_LLVM_LAF_SPLIT_SWITCHES: cascaded comparisons on switches\n" + " AFL_LLVM_LAF_SPLIT_FLOATS: cascaded comparisons on floats\n" + " AFL_LLVM_LAF_TRANSFORM_COMPARES: cascade comparisons for string " + "functions\n" + " AFL_LLVM_INSTRUMENT_ALLOW/AFL_LLVM_INSTRUMENT_DENY: enable " + "instrument allow/\n" + " deny listing (selective instrumentation)\n" + " AFL_NO_BUILTIN: no builtins for string compare functions (for " + "libtokencap.so)\n" + " AFL_PATH: path to instrumenting pass and runtime " + "(afl-compiler-rt.*o)\n" + " AFL_LLVM_DOCUMENT_IDS: document edge IDs given to which function " + "(LTO only)\n" + " AFL_QUIET: suppress verbose output\n" + " AFL_USE_ASAN: activate address sanitizer\n" + " AFL_USE_CFISAN: activate control flow sanitizer\n" + " AFL_USE_MSAN: activate memory sanitizer\n" + " AFL_USE_UBSAN: activate undefined behaviour sanitizer\n", + BIN_PATH, BIN_PATH); + + SAYF( + "\nLLVM/LTO/afl-clang-fast/afl-clang-lto specific environment " + "variables:\n" + " AFL_LLVM_CMPLOG: log operands of comparisons (RedQueen mutator)\n" + " AFL_LLVM_INSTRUMENT: set instrumentation mode: CLASSIC, INSTRIM, " + "PCGUARD, LTO, CTX, NGRAM-2 ... NGRAM-16\n" + " You can also use the old environment variables instead:\n" + " AFL_LLVM_USE_TRACE_PC: use LLVM trace-pc-guard instrumentation\n" + " AFL_LLVM_INSTRIM: use light weight instrumentation InsTrim\n" + " AFL_LLVM_INSTRIM_LOOPHEAD: optimize loop tracing for speed " + "(option to INSTRIM)\n" + " AFL_LLVM_CTX: use context sensitive coverage (for CLASSIC and " + "INSTRIM)\n" + " AFL_LLVM_NGRAM_SIZE: use ngram prev_loc count coverage (for " + "CLASSIC and INSTRIM)\n"); + +#ifdef AFL_CLANG_FLTO + SAYF( + "\nLTO/afl-clang-lto specific environment variables:\n" + "AFL_LLVM_MAP_ADDR: use a fixed coverage map address (speed), e.g. " + "0x10000\n" + "AFL_LLVM_DOCUMENT_IDS: write all edge IDs and the corresponding " + "functions they are in into this file\n" + "AFL_LLVM_LTO_DONTWRITEID: don't write the highest ID used to a " + "global var\n" + "AFL_LLVM_LTO_STARTID: from which ID to start counting from for a " + "bb\n" + "AFL_REAL_LD: use this lld linker instead of the compiled in path\n" + "\nafl-clang-lto was built with linker target \"%s\" and LTO flags " + "\"%s\"\n" + "If anything fails - be sure to read README.lto.md!\n", + AFL_REAL_LD, AFL_CLANG_FLTO); +#endif + + } + + SAYF( + "For any information on the available instrumentations and options " + "please \n" + "consult the README.md, especially section 3.1 about instrumenting " + "targets.\n\n"); + +#if (LLVM_MAJOR > 2) + if (have_lto) + SAYF("afl-cc LTO with ld=%s %s\n", AFL_REAL_LD, AFL_CLANG_FLTO); + if (have_llvm) + SAYF("afl-cc LLVM version %d with the the binary path \"%s\".\n", + LLVM_MAJOR, LLVM_BINDIR); + if (have_lto || have_llvm) SAYF("\n"); +#endif + + SAYF( + "Do not be overwhelmed :) afl-cc uses good defaults if no options are " + "selected.\n" + "Read the documentation for FEATURES though, all are good but few are " + "defaults.\n\n"); + + exit(1); + + } + + if (compiler_mode == LTO) { + + if (instrument_mode == 0 || instrument_mode == INSTRUMENT_LTO || + instrument_mode == INSTRUMENT_CFG) { + + lto_mode = 1; + if (!instrument_mode) { + + instrument_mode = INSTRUMENT_CFG; + ptr = instrument_mode_string[instrument_mode]; + + } + + } else if (instrument_mode == INSTRUMENT_LTO || + + instrument_mode == INSTRUMENT_CLASSIC) { + + lto_mode = 1; + + } else { + + if (!be_quiet) + WARNF("afl-clang-lto called with mode %s, using that mode instead", + instrument_mode_string[instrument_mode]); + + } + + } + + if (instrument_mode == 0 && compiler_mode < GCC_PLUGIN) { + +#if LLVM_MAJOR <= 6 + instrument_mode = INSTRUMENT_AFL; +#else + if (getenv("AFL_LLVM_INSTRUMENT_FILE") != NULL || + getenv("AFL_LLVM_WHITELIST") || getenv("AFL_LLVM_ALLOWLIST") || + getenv("AFL_LLVM_DENYLIST") || getenv("AFL_LLVM_BLOCKLIST")) { + + instrument_mode = INSTRUMENT_AFL; + WARNF( + "switching to classic instrumentation because " + "AFL_LLVM_ALLOWLIST/DENYLIST does not work with PCGUARD. Use " + "-fsanitize-coverage-allowlist=allowlist.txt or " + "-fsanitize-coverage-blocklist=denylist.txt if you want to use " + "PCGUARD. Requires llvm 12+. See https://clang.llvm.org/docs/ " + "SanitizerCoverage.html#partially-disabling-instrumentation"); + + } else + + instrument_mode = INSTRUMENT_PCGUARD; +#endif + + } + + if (instrument_opt_mode && compiler_mode != LLVM) + FATAL("CTX and NGRAM can only be used in LLVM mode"); + + if (!instrument_opt_mode) { + + if (lto_mode && instrument_mode == INSTRUMENT_CFG) + instrument_mode = INSTRUMENT_PCGUARD; + ptr = instrument_mode_string[instrument_mode]; + + } else { + + if (instrument_opt_mode == INSTRUMENT_OPT_CTX) + + ptr = alloc_printf("%s + CTX", instrument_mode_string[instrument_mode]); + else if (instrument_opt_mode == INSTRUMENT_OPT_NGRAM) + ptr = alloc_printf("%s + NGRAM-%u", + instrument_mode_string[instrument_mode], ngram_size); + else + ptr = alloc_printf("%s + CTX + NGRAM-%u", + instrument_mode_string[instrument_mode], ngram_size); + + } + +#ifndef AFL_CLANG_FLTO + if (lto_mode) + FATAL( + "instrumentation mode LTO specified but LLVM support not available " + "(requires LLVM 11 or higher)"); +#endif + + if (instrument_opt_mode && instrument_mode != INSTRUMENT_CLASSIC && + instrument_mode != INSTRUMENT_CFG) + FATAL( + "CTX and NGRAM instrumentation options can only be used with CFG " + "(recommended) and CLASSIC instrumentation modes!"); + + if (getenv("AFL_LLVM_SKIP_NEVERZERO") && getenv("AFL_LLVM_NOT_ZERO")) + FATAL( + "AFL_LLVM_NOT_ZERO and AFL_LLVM_SKIP_NEVERZERO can not be set " + "together"); + + if (instrument_mode == INSTRUMENT_PCGUARD && + (getenv("AFL_LLVM_INSTRUMENT_FILE") != NULL || + getenv("AFL_LLVM_WHITELIST") || getenv("AFL_LLVM_ALLOWLIST") || + getenv("AFL_LLVM_DENYLIST") || getenv("AFL_LLVM_BLOCKLIST"))) + FATAL( + "Instrumentation type PCGUARD does not support " + "AFL_LLVM_ALLOWLIST/DENYLIST! Use " + "-fsanitize-coverage-allowlist=allowlist.txt or " + "-fsanitize-coverage-blocklist=denylist.txt instead (requires llvm " + "12+), see " + "https://clang.llvm.org/docs/" + "SanitizerCoverage.html#partially-disabling-instrumentation"); + + u8 *ptr2; + + if ((ptr2 = getenv("AFL_LLVM_DICT2FILE")) != NULL && *ptr2 != '/') + FATAL("AFL_LLVM_DICT2FILE must be set to an absolute file path"); + + if ((isatty(2) && !be_quiet) || debug) { + + SAYF(cCYA + "afl-cc " VERSION cRST + " by Michal Zalewski, Laszlo Szekeres, Marc Heuse - mode: %s-%s\n", + compiler_mode_string[compiler_mode], ptr); + + } + + if (!be_quiet && !lto_mode && + ((ptr2 = getenv("AFL_MAP_SIZE")) || (ptr2 = getenv("AFL_MAPSIZE")))) { + + u32 map_size = atoi(ptr2); + if (map_size != MAP_SIZE) + WARNF("AFL_MAP_SIZE is not supported by afl-clang-fast"); + + } + + if (debug) { + + SAYF(cMGN "[D]" cRST " cd \"%s\";", getthecwd()); + for (i = 0; i < argc; i++) + SAYF(" \"%s\"", argv[i]); + SAYF("\n"); + + } + + if (getenv("AFL_LLVM_LAF_ALL")) { + + setenv("AFL_LLVM_LAF_SPLIT_SWITCHES", "1", 1); + setenv("AFL_LLVM_LAF_SPLIT_COMPARES", "1", 1); + setenv("AFL_LLVM_LAF_SPLIT_FLOATS", "1", 1); + setenv("AFL_LLVM_LAF_TRANSFORM_COMPARES", "1", 1); + + } + + cmplog_mode = getenv("AFL_CMPLOG") || getenv("AFL_LLVM_CMPLOG"); + if (!be_quiet && cmplog_mode) + printf("CmpLog mode by \n"); + +#ifndef __ANDROID__ + find_obj(argv[0]); +#endif + + edit_params(argc, argv, envp); + + if (debug) { + + SAYF(cMGN "[D]" cRST " cd \"%s\";", getthecwd()); + for (i = 0; i < cc_par_cnt; i++) + SAYF(" \"%s\"", cc_params[i]); + SAYF("\n"); + + } + + execvp(cc_params[0], (char **)cc_params); + + FATAL("Oops, failed to execute '%s' - check your PATH", cc_params[0]); + + return 0; + +} + diff --git a/src/afl-gcc.c b/src/afl-gcc.c deleted file mode 100644 index 97564aea..00000000 --- a/src/afl-gcc.c +++ /dev/null @@ -1,488 +0,0 @@ -/* - american fuzzy lop++ - wrapper for GCC and clang - ------------------------------------------------ - - Originally written by Michal Zalewski - - Now maintained by Marc Heuse , - Heiko Eißfeldt and - Andrea Fioraldi - - Copyright 2016, 2017 Google Inc. All rights reserved. - Copyright 2019-2020 AFLplusplus Project. All rights reserved. - - Licensed under the Apache License, Version 2.0 (the "License"); - you may not use this file except in compliance with the License. - You may obtain a copy of the License at: - - http://www.apache.org/licenses/LICENSE-2.0 - - This program is a drop-in replacement for GCC or clang. The most common way - of using it is to pass the path to afl-gcc or afl-clang via CC when invoking - ./configure. - - (Of course, use CXX and point it to afl-g++ / afl-clang++ for C++ code.) - - The wrapper needs to know the path to afl-as (renamed to 'as'). The default - is /usr/local/lib/afl/. A convenient way to specify alternative directories - would be to set AFL_PATH. - - If AFL_HARDEN is set, the wrapper will compile the target app with various - hardening options that may help detect memory management issues more - reliably. You can also specify AFL_USE_ASAN to enable ASAN. - - If you want to call a non-default compiler as a next step of the chain, - specify its location via AFL_CC or AFL_CXX. - - */ - -#define AFL_MAIN - -#include "config.h" -#include "types.h" -#include "debug.h" -#include "alloc-inl.h" - -#include -#include -#include -#include - -static u8 * as_path; /* Path to the AFL 'as' wrapper */ -static u8 **cc_params; /* Parameters passed to the real CC */ -static u32 cc_par_cnt = 1; /* Param count, including argv0 */ -static u8 be_quiet, /* Quiet mode */ - clang_mode; /* Invoked as afl-clang*? */ - -/* Try to find our "fake" GNU assembler in AFL_PATH or at the location derived - from argv[0]. If that fails, abort. */ - -static void find_as(u8 *argv0) { - - u8 *afl_path = getenv("AFL_PATH"); - u8 *slash, *tmp; - - if (afl_path) { - - tmp = alloc_printf("%s/as", afl_path); - - if (!access(tmp, X_OK)) { - - as_path = afl_path; - ck_free(tmp); - return; - - } - - ck_free(tmp); - - } - - slash = strrchr(argv0, '/'); - - if (slash) { - - u8 *dir; - - *slash = 0; - dir = ck_strdup(argv0); - *slash = '/'; - - tmp = alloc_printf("%s/afl-as", dir); - - if (!access(tmp, X_OK)) { - - as_path = dir; - ck_free(tmp); - return; - - } - - ck_free(tmp); - ck_free(dir); - - } - - if (!access(AFL_PATH "/as", X_OK)) { - - as_path = AFL_PATH; - return; - - } - - FATAL("Unable to find AFL wrapper binary for 'as'. Please set AFL_PATH"); - -} - -/* Copy argv to cc_params, making the necessary edits. */ - -static void edit_params(u32 argc, char **argv) { - - u8 fortify_set = 0, asan_set = 0; - u8 *name; - -#if defined(__FreeBSD__) && defined(WORD_SIZE_64) - u8 m32_set = 0; -#endif - - cc_params = ck_alloc((argc + 128) * sizeof(u8 *)); - - name = strrchr(argv[0], '/'); - if (!name) { - - name = argv[0]; - - /* This should never happen but fixes a scan-build warning */ - if (!name) { FATAL("Empty argv set"); } - - } else { - - ++name; - - } - - if (!strncmp(name, "afl-clang", 9)) { - - clang_mode = 1; - - setenv(CLANG_ENV_VAR, "1", 1); - - if (!strcmp(name, "afl-clang++")) { - - u8 *alt_cxx = getenv("AFL_CXX"); - cc_params[0] = alt_cxx && *alt_cxx ? alt_cxx : (u8 *)"clang++"; - - } else if (!strcmp(name, "afl-clang")) { - - u8 *alt_cc = getenv("AFL_CC"); - cc_params[0] = alt_cc && *alt_cc ? alt_cc : (u8 *)"clang"; - - } else { - - fprintf(stderr, "Name of the binary: %s\n", argv[0]); - FATAL("Name of the binary is not a known name, expected afl-clang(++)"); - - } - - } else { - - /* With GCJ and Eclipse installed, you can actually compile Java! The - instrumentation will work (amazingly). Alas, unhandled exceptions do - not call abort(), so afl-fuzz would need to be modified to equate - non-zero exit codes with crash conditions when working with Java - binaries. Meh. */ - -#ifdef __APPLE__ - - if (!strcmp(name, "afl-g++")) { - - cc_params[0] = getenv("AFL_CXX"); - - } else if (!strcmp(name, "afl-gcj")) { - - cc_params[0] = getenv("AFL_GCJ"); - - } else if (!strcmp(name, "afl-gcc")) { - - cc_params[0] = getenv("AFL_CC"); - - } else { - - fprintf(stderr, "Name of the binary: %s\n", argv[0]); - FATAL("Name of the binary is not a known name, expected afl-gcc/g++/gcj"); - - } - - if (!cc_params[0]) { - - SAYF("\n" cLRD "[-] " cRST - "On Apple systems, 'gcc' is usually just a wrapper for clang. " - "Please use the\n" - " 'afl-clang' utility instead of 'afl-gcc'. If you really have " - "GCC installed,\n" - " set AFL_CC or AFL_CXX to specify the correct path to that " - "compiler.\n"); - - FATAL("AFL_CC or AFL_CXX required on MacOS X"); - - } - -#else - - if (!strcmp(name, "afl-g++")) { - - u8 *alt_cxx = getenv("AFL_CXX"); - cc_params[0] = alt_cxx && *alt_cxx ? alt_cxx : (u8 *)"g++"; - - } else if (!strcmp(name, "afl-gcj")) { - - u8 *alt_cc = getenv("AFL_GCJ"); - cc_params[0] = alt_cc && *alt_cc ? alt_cc : (u8 *)"gcj"; - - } else if (!strcmp(name, "afl-gcc")) { - - u8 *alt_cc = getenv("AFL_CC"); - cc_params[0] = alt_cc && *alt_cc ? alt_cc : (u8 *)"gcc"; - - } else { - - fprintf(stderr, "Name of the binary: %s\n", argv[0]); - FATAL("Name of the binary is not a known name, expected afl-gcc/g++/gcj"); - - } - -#endif /* __APPLE__ */ - - } - - while (--argc) { - - u8 *cur = *(++argv); - - if (!strncmp(cur, "-B", 2)) { - - if (!be_quiet) { WARNF("-B is already set, overriding"); } - - if (!cur[2] && argc > 1) { - - argc--; - argv++; - - } - - continue; - - } - - if (!strcmp(cur, "-integrated-as")) { continue; } - - if (!strcmp(cur, "-pipe")) { continue; } - -#if defined(__FreeBSD__) && defined(WORD_SIZE_64) - if (!strcmp(cur, "-m32")) m32_set = 1; -#endif - - if (!strcmp(cur, "-fsanitize=address") || - !strcmp(cur, "-fsanitize=memory")) { - - asan_set = 1; - - } - - if (strstr(cur, "FORTIFY_SOURCE")) { fortify_set = 1; } - - cc_params[cc_par_cnt++] = cur; - - } - - cc_params[cc_par_cnt++] = "-B"; - cc_params[cc_par_cnt++] = as_path; - - if (clang_mode) { cc_params[cc_par_cnt++] = "-no-integrated-as"; } - - if (getenv("AFL_HARDEN")) { - - cc_params[cc_par_cnt++] = "-fstack-protector-all"; - - if (!fortify_set) { cc_params[cc_par_cnt++] = "-D_FORTIFY_SOURCE=2"; } - - } - - if (asan_set) { - - /* Pass this on to afl-as to adjust map density. */ - - setenv("AFL_USE_ASAN", "1", 1); - - } else if (getenv("AFL_USE_ASAN")) { - - if (getenv("AFL_USE_MSAN")) { - - FATAL("ASAN and MSAN are mutually exclusive"); - - } - - if (getenv("AFL_HARDEN")) { - - FATAL("ASAN and AFL_HARDEN are mutually exclusive"); - - } - - cc_params[cc_par_cnt++] = "-U_FORTIFY_SOURCE"; - cc_params[cc_par_cnt++] = "-fsanitize=address"; - - } else if (getenv("AFL_USE_MSAN")) { - - if (getenv("AFL_USE_ASAN")) { - - FATAL("ASAN and MSAN are mutually exclusive"); - - } - - if (getenv("AFL_HARDEN")) { - - FATAL("MSAN and AFL_HARDEN are mutually exclusive"); - - } - - cc_params[cc_par_cnt++] = "-U_FORTIFY_SOURCE"; - cc_params[cc_par_cnt++] = "-fsanitize=memory"; - - } - - if (getenv("AFL_USE_UBSAN")) { - - cc_params[cc_par_cnt++] = "-fsanitize=undefined"; - cc_params[cc_par_cnt++] = "-fsanitize-undefined-trap-on-error"; - cc_params[cc_par_cnt++] = "-fno-sanitize-recover=all"; - - } - -#if defined(USEMMAP) && !defined(__HAIKU__) - cc_params[cc_par_cnt++] = "-lrt"; -#endif - - if (!getenv("AFL_DONT_OPTIMIZE")) { - -#if defined(__FreeBSD__) && defined(WORD_SIZE_64) - - /* On 64-bit FreeBSD systems, clang -g -m32 is broken, but -m32 itself - works OK. This has nothing to do with us, but let's avoid triggering - that bug. */ - - if (!clang_mode || !m32_set) cc_params[cc_par_cnt++] = "-g"; - -#else - - cc_params[cc_par_cnt++] = "-g"; - -#endif - - cc_params[cc_par_cnt++] = "-O3"; - cc_params[cc_par_cnt++] = "-funroll-loops"; - - /* Two indicators that you're building for fuzzing; one of them is - AFL-specific, the other is shared with libfuzzer. */ - - cc_params[cc_par_cnt++] = "-D__AFL_COMPILER=1"; - cc_params[cc_par_cnt++] = "-DFUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION=1"; - - } - - if (getenv("AFL_NO_BUILTIN")) { - - cc_params[cc_par_cnt++] = "-fno-builtin-strcmp"; - cc_params[cc_par_cnt++] = "-fno-builtin-strncmp"; - cc_params[cc_par_cnt++] = "-fno-builtin-strcasecmp"; - cc_params[cc_par_cnt++] = "-fno-builtin-strncasecmp"; - cc_params[cc_par_cnt++] = "-fno-builtin-memcmp"; - cc_params[cc_par_cnt++] = "-fno-builtin-bcmp"; - cc_params[cc_par_cnt++] = "-fno-builtin-strstr"; - cc_params[cc_par_cnt++] = "-fno-builtin-strcasestr"; - - } - - cc_params[cc_par_cnt] = NULL; - -} - -/* Main entry point */ - -int main(int argc, char **argv) { - - char *env_info = - "Environment variables used by afl-gcc:\n" - "AFL_CC: path to the C compiler to use\n" - "AFL_CXX: path to the C++ compiler to use\n" - "AFL_GCJ: path to the java compiler to use\n" - "AFL_PATH: path to the instrumenting assembler\n" - "AFL_DONT_OPTIMIZE: disable optimization instead of -O3\n" - "AFL_NO_BUILTIN: compile for use with libtokencap.so\n" - "AFL_QUIET: suppress verbose output\n" - "AFL_CAL_FAST: speed up the initial calibration\n" - "AFL_HARDEN: adds code hardening to catch memory bugs\n" - "AFL_USE_ASAN: activate address sanitizer\n" - "AFL_USE_MSAN: activate memory sanitizer\n" - "AFL_USE_UBSAN: activate undefined behaviour sanitizer\n" - - "\nEnvironment variables used by afl-as (called by afl-gcc):\n" - "AFL_AS: path to the assembler to use\n" - "TMPDIR: set the directory for temporary files of afl-as\n" - "TEMP: fall back path to directory for temporary files\n" - "TMP: fall back path to directory for temporary files\n" - "AFL_INST_RATIO: percentage of branches to instrument\n" - "AFL_QUIET: suppress verbose output\n" - "AFL_KEEP_ASSEMBLY: leave instrumented assembly files\n" - "AFL_AS_FORCE_INSTRUMENT: force instrumentation for asm sources\n"; - - if (argc == 2 && strncmp(argv[1], "-h", 2) == 0) { - - printf("afl-cc" VERSION " by Michal Zalewski\n\n"); - printf("%s \n\n", argv[0]); - printf("afl-gcc has no command line options\n\n%s\n", env_info); - printf( - "NOTE: afl-gcc is deprecated, llvm_mode is much faster and has more " - "options\n"); - return -1; - - } - - if ((isatty(2) && !getenv("AFL_QUIET")) || getenv("AFL_DEBUG") != NULL) { - - SAYF(cCYA "afl-cc" VERSION cRST " by Michal Zalewski\n"); - SAYF(cYEL "[!] " cBRI "NOTE: " cRST - "afl-gcc is deprecated, llvm_mode is much faster and has more " - "options\n"); - - } else { - - be_quiet = 1; - - } - - if (argc < 2) { - - SAYF( - "\n" - "This is a helper application for afl-fuzz. It serves as a drop-in " - "replacement\n" - "for gcc or clang, letting you recompile third-party code with the " - "required\n" - "runtime instrumentation. A common use pattern would be one of the " - "following:\n\n" - - " CC=%s/afl-gcc ./configure\n" - " CXX=%s/afl-g++ ./configure\n\n%s" - - , - BIN_PATH, BIN_PATH, env_info); - - exit(1); - - } - - u8 *ptr; - if (!be_quiet && - ((ptr = getenv("AFL_MAP_SIZE")) || (ptr = getenv("AFL_MAPSIZE")))) { - - u32 map_size = atoi(ptr); - if (map_size != MAP_SIZE) { - - WARNF("AFL_MAP_SIZE is not supported by afl-gcc"); - - } - - } - - find_as(argv[0]); - - edit_params(argc, argv); - - execvp(cc_params[0], (char **)cc_params); - - FATAL("Oops, failed to execute '%s' - check your PATH", cc_params[0]); - - return 0; - -} - diff --git a/src/afl-ld-lto.c b/src/afl-ld-lto.c new file mode 100644 index 00000000..771e2d0d --- /dev/null +++ b/src/afl-ld-lto.c @@ -0,0 +1,358 @@ +/* + american fuzzy lop++ - wrapper for llvm 11+ lld + ----------------------------------------------- + + Written by Marc Heuse for afl++ + + Maintained by Marc Heuse , + Heiko Eißfeldt + Andrea Fioraldi + Dominik Maier + + Copyright 2019-2020 AFLplusplus Project. All rights reserved. + + Licensed under the Apache License, Version 2.0 (the "License"); + you may not use this file except in compliance with the License. + You may obtain a copy of the License at: + + http://www.apache.org/licenses/LICENSE-2.0 + + The sole purpose of this wrapper is to preprocess clang LTO files when + linking with lld and performing the instrumentation on the whole program. + +*/ + +#define AFL_MAIN +#define _GNU_SOURCE + +#include "config.h" +#include "types.h" +#include "debug.h" +#include "alloc-inl.h" + +#include +#include +#include +#include +#include +#include +#include + +#include +#include +#include +#include + +#include + +#define MAX_PARAM_COUNT 4096 + +static u8 **ld_params; /* Parameters passed to the real 'ld' */ + +static u8 *afl_path = AFL_PATH; +static u8 *real_ld = AFL_REAL_LD; + +static u8 be_quiet, /* Quiet mode (no stderr output) */ + debug, /* AFL_DEBUG */ + passthrough, /* AFL_LD_PASSTHROUGH - no link+optimize*/ + just_version; /* Just show version? */ + +static u32 ld_param_cnt = 1; /* Number of params to 'ld' */ + +/* Examine and modify parameters to pass to 'ld', 'llvm-link' and 'llmv-ar'. + Note that the file name is always the last parameter passed by GCC, + so we exploit this property to keep the code "simple". */ +static void edit_params(int argc, char **argv) { + + u32 i, instrim = 0, gold_pos = 0, gold_present = 0, rt_present = 0, + rt_lto_present = 0, inst_present = 0; + char *ptr; + + ld_params = ck_alloc(4096 * sizeof(u8 *)); + + ld_params[0] = (u8 *)real_ld; + + if (!passthrough) { + + for (i = 1; i < argc; i++) { + + if (strstr(argv[i], "/afl-llvm-rt-lto.o") != NULL) rt_lto_present = 1; + if (strstr(argv[i], "/afl-llvm-rt.o") != NULL) rt_present = 1; + if (strstr(argv[i], "/afl-llvm-lto-instr") != NULL) inst_present = 1; + + } + + for (i = 1; i < argc && !gold_pos; i++) { + + if (strcmp(argv[i], "-plugin") == 0) { + + if (strncmp(argv[i], "-plugin=", strlen("-plugin=")) == 0) { + + if (strcasestr(argv[i], "LLVMgold.so") != NULL) + gold_present = gold_pos = i + 1; + + } else if (i < argc && strcasestr(argv[i + 1], "LLVMgold.so") != NULL) { + + gold_present = gold_pos = i + 2; + + } + + } + + } + + if (!gold_pos) { + + for (i = 1; i + 1 < argc && !gold_pos; i++) { + + if (argv[i][0] != '-') { + + if (argv[i - 1][0] == '-') { + + switch (argv[i - 1][1]) { + + case 'b': + break; + case 'd': + break; + case 'e': + break; + case 'F': + break; + case 'f': + break; + case 'I': + break; + case 'l': + break; + case 'L': + break; + case 'm': + break; + case 'o': + break; + case 'O': + break; + case 'p': + if (index(argv[i - 1], '=') == NULL) gold_pos = i; + break; + case 'R': + break; + case 'T': + break; + case 'u': + break; + case 'y': + break; + case 'z': + break; + case '-': { + + if (strcmp(argv[i - 1], "--oformat") == 0) break; + if (strcmp(argv[i - 1], "--output") == 0) break; + if (strncmp(argv[i - 1], "--opt-remarks-", 14) == 0) break; + gold_pos = i; + break; + + } + + default: + gold_pos = i; + + } + + } else + + gold_pos = i; + + } + + } + + } + + if (!gold_pos) gold_pos = 1; + + } + + if (getenv("AFL_LLVM_INSTRIM")) + instrim = 1; + else if ((ptr = getenv("AFL_LLVM_INSTRUMENT")) && + (strcasestr(ptr, "CFG") == 0 || strcasestr(ptr, "INSTRIM") == 0)) + instrim = 1; + + if (debug) + SAYF(cMGN "[D] " cRST + "passthrough=%s instrim=%d, gold_pos=%d, gold_present=%s " + "inst_present=%s rt_present=%s rt_lto_present=%s\n", + passthrough ? "true" : "false", instrim, gold_pos, + gold_present ? "true" : "false", inst_present ? "true" : "false", + rt_present ? "true" : "false", rt_lto_present ? "true" : "false"); + + for (i = 1; i < argc; i++) { + + if (ld_param_cnt >= MAX_PARAM_COUNT) + FATAL( + "Too many command line parameters because of unpacking .a archives, " + "this would need to be done by hand ... sorry! :-("); + + if (strcmp(argv[i], "--afl") == 0) { + + if (!be_quiet) OKF("afl++ test command line flag detected, exiting."); + exit(0); + + } + + if (i == gold_pos && !passthrough) { + + ld_params[ld_param_cnt++] = alloc_printf("-L%s/../lib", LLVM_BINDIR); + + if (!gold_present) { + + ld_params[ld_param_cnt++] = "-plugin"; + ld_params[ld_param_cnt++] = + alloc_printf("%s/../lib/LLVMgold.so", LLVM_BINDIR); + + } + + ld_params[ld_param_cnt++] = "--allow-multiple-definition"; + + if (!inst_present) { + + if (instrim) + ld_params[ld_param_cnt++] = + alloc_printf("-mllvm=-load=%s/afl-llvm-lto-instrim.so", afl_path); + else + ld_params[ld_param_cnt++] = alloc_printf( + "-mllvm=-load=%s/afl-llvm-lto-instrumentation.so", afl_path); + + } + + if (!rt_present) + ld_params[ld_param_cnt++] = alloc_printf("%s/afl-llvm-rt.o", afl_path); + if (!rt_lto_present) + ld_params[ld_param_cnt++] = + alloc_printf("%s/afl-llvm-rt-lto.o", afl_path); + + } + + ld_params[ld_param_cnt++] = argv[i]; + + } + + ld_params[ld_param_cnt] = NULL; + +} + +/* Main entry point */ + +int main(int argc, char **argv) { + + s32 pid, i, status; + u8 * ptr; + char thecwd[PATH_MAX]; + + if ((ptr = getenv("AFL_LD_CALLER")) != NULL) { + + FATAL("ld loop detected! Set AFL_REAL_LD!\n"); + + } + + if (isatty(2) && !getenv("AFL_QUIET") && !getenv("AFL_DEBUG")) { + + SAYF(cCYA "afl-ld-to" VERSION cRST + " by Marc \"vanHauser\" Heuse \n"); + + } else + + be_quiet = 1; + + if (getenv("AFL_DEBUG") != NULL) debug = 1; + if (getenv("AFL_PATH") != NULL) afl_path = getenv("AFL_PATH"); + if (getenv("AFL_LD_PASSTHROUGH") != NULL) passthrough = 1; + if (getenv("AFL_REAL_LD") != NULL) real_ld = getenv("AFL_REAL_LD"); + + if (!afl_path || !*afl_path) afl_path = "/usr/local/lib/afl"; + + setenv("AFL_LD_CALLER", "1", 1); + + if (debug) { + + if (getcwd(thecwd, sizeof(thecwd)) != 0) strcpy(thecwd, "."); + + SAYF(cMGN "[D] " cRST "cd \"%s\";", thecwd); + for (i = 0; i < argc; i++) + SAYF(" \"%s\"", argv[i]); + SAYF("\n"); + + } + + if (argc < 2) { + + SAYF( + "\n" + "This is a helper application for afl-clang-lto. It is a wrapper " + "around GNU " + "llvm's 'lld',\n" + "executed by the toolchain whenever using " + "afl-clang-lto/afl-clang-lto++.\n" + "You probably don't want to run this program directly but rather pass " + "it as LD parameter to configure scripts\n\n" + + "Environment variables:\n" + " AFL_LD_PASSTHROUGH do not link+optimize == no instrumentation\n" + " AFL_REAL_LD point to the real llvm 11 lld if necessary\n" + + "\nafl-ld-to was compiled with the fixed real 'ld' of %s and the " + "binary path of %s\n\n", + real_ld, LLVM_BINDIR); + + exit(1); + + } + + edit_params(argc, argv); // here most of the magic happens :-) + + if (debug) { + + SAYF(cMGN "[D]" cRST " cd \"%s\";", thecwd); + for (i = 0; i < ld_param_cnt; i++) + SAYF(" \"%s\"", ld_params[i]); + SAYF("\n"); + + } + + if (!(pid = fork())) { + + if (strlen(real_ld) > 1) execvp(real_ld, (char **)ld_params); + execvp("ld", (char **)ld_params); // fallback + FATAL("Oops, failed to execute 'ld' - check your PATH"); + + } + + if (pid < 0) PFATAL("fork() failed"); + + if (waitpid(pid, &status, 0) <= 0) PFATAL("waitpid() failed"); + if (debug) SAYF(cMGN "[D] " cRST "linker result: %d\n", status); + + if (!just_version) { + + if (status == 0) { + + if (!be_quiet) OKF("Linker was successful"); + + } else { + + SAYF(cLRD "[-] " cRST + "Linker failed, please investigate and send a bug report. Most " + "likely an 'ld' option is incompatible with %s.\n", + AFL_CLANG_FLTO); + + } + + } + + exit(WEXITSTATUS(status)); + +} + -- cgit 1.4.1 From 4b3ad5f037ee9a36aa057bf55a69acca1f573922 Mon Sep 17 00:00:00 2001 From: vanhauser-thc Date: Sat, 5 Sep 2020 12:32:10 +0200 Subject: add cull queue, -i subdir traversal --- GNUmakefile | 230 ++++++++++++++++++++++++-------------------------- GNUmakefile.llvm | 7 +- README.md | 179 +++++++++++++++++++++++---------------- include/afl-fuzz.h | 6 +- src/afl-fuzz-extras.c | 113 ++++++++++++++++++++----- src/afl-fuzz-init.c | 108 +++++++++++++++++++++--- src/afl-fuzz-one.c | 31 +------ src/afl-fuzz-queue.c | 7 -- src/afl-fuzz.c | 33 +++++--- 9 files changed, 437 insertions(+), 277 deletions(-) diff --git a/GNUmakefile b/GNUmakefile index e1f6da95..0046a481 100644 --- a/GNUmakefile +++ b/GNUmakefile @@ -24,30 +24,31 @@ BIN_PATH = $(PREFIX)/bin HELPER_PATH = $(PREFIX)/lib/afl DOC_PATH = $(PREFIX)/share/doc/afl MISC_PATH = $(PREFIX)/share/afl -MAN_PATH = $(PREFIX)/share/man/man8 +MAN_PATH = $(PREFIX)/man/man8 PROGNAME = afl VERSION = $(shell grep '^$(HASH)define VERSION ' ../config.h | cut -d '"' -f2) # PROGS intentionally omit afl-as, which gets installed elsewhere. -PROGS = afl-gcc afl-g++ afl-fuzz afl-showmap afl-tmin afl-gotcpu afl-analyze +PROGS = afl-fuzz afl-showmap afl-tmin afl-gotcpu afl-analyze SH_PROGS = afl-plot afl-cmin afl-cmin.bash afl-whatsup afl-system-config MANPAGES=$(foreach p, $(PROGS) $(SH_PROGS), $(p).8) afl-as.8 +ASAN_OPTIONS=detect_leaks=0 ifeq "$(findstring android, $(shell $(CC) --version 2>/dev/null))" "" - ifeq "$(shell echo 'int main() {return 0; }' | $(CC) $(CFLAGS) -Werror -x c - -flto=full -o .test 2>/dev/null && echo 1 || echo 0 ; rm -f .test )" "1" +ifeq "$(shell echo 'int main() {return 0; }' | $(CC) $(CFLAGS) -Werror -x c - -flto=full -o .test 2>/dev/null && echo 1 || echo 0 ; rm -f .test )" "1" CFLAGS_FLTO ?= -flto=full - else - ifeq "$(shell echo 'int main() {return 0; }' | $(CC) $(CFLAGS) -Werror -x c - -flto=thin -o .test 2>/dev/null && echo 1 || echo 0 ; rm -f .test )" "1" +else + ifeq "$(shell echo 'int main() {return 0; }' | $(CC) $(CFLAGS) -Werror -x c - -flto=thin -o .test 2>/dev/null && echo 1 || echo 0 ; rm -f .test )" "1" CFLAGS_FLTO ?= -flto=thin - else - ifeq "$(shell echo 'int main() {return 0; }' | $(CC) $(CFLAGS) -Werror -x c - -flto -o .test 2>/dev/null && echo 1 || echo 0 ; rm -f .test )" "1" + else + ifeq "$(shell echo 'int main() {return 0; }' | $(CC) $(CFLAGS) -Werror -x c - -flto -o .test 2>/dev/null && echo 1 || echo 0 ; rm -f .test )" "1" CFLAGS_FLTO ?= -flto - endif endif endif endif +endif ifeq "$(shell echo 'int main() {return 0; }' | $(CC) -fno-move-loop-invariants -fdisable-tree-cunrolli -x c - -o .test 2>/dev/null && echo 1 || echo 0 ; rm -f .test )" "1" SPECIAL_PERFORMANCE += -fno-move-loop-invariants -fdisable-tree-cunrolli @@ -61,10 +62,7 @@ ifneq "$(shell uname)" "Darwin" endif endif # OS X does not like _FORTIFY_SOURCE=2 - # _FORTIFY_SOURCE=2 does not like -O0 - ifndef DEBUG - CFLAGS_OPT += -D_FORTIFY_SOURCE=2 - endif + CFLAGS_OPT += -D_FORTIFY_SOURCE=2 endif ifeq "$(shell uname)" "SunOS" @@ -206,10 +204,7 @@ else endif ifneq "$(filter Linux GNU%,$(shell uname))" "" - # _FORTIFY_SOURCE=2 does not like -O0 - ifndef DEBUG override CFLAGS += -D_FORTIFY_SOURCE=2 - endif LDFLAGS += -ldl -lrt endif @@ -223,11 +218,7 @@ ifneq "$(findstring NetBSD, $(shell uname))" "" LDFLAGS += -lpthread endif -ifeq "$(findstring clang, $(shell $(CC) --version 2>/dev/null))" "" - TEST_CC = afl-gcc -else - TEST_CC = afl-clang -endif +TEST_CC = afl-gcc COMM_HDR = include/alloc-inl.h include/config.h include/debug.h include/types.h @@ -277,28 +268,47 @@ ifdef TEST_MMAP LDFLAGS += -Wno-deprecated-declarations endif -all: test_x86 test_shm test_python ready $(PROGS) afl-as test_build all_done +.PHONY: all +all: test_x86 test_shm test_python ready $(PROGS) afl-as llvm gcc_plugin test_build all_done + +.PHONY: llvm +llvm: + -$(MAKE) -f GNUmakefile.llvm + @test -e afl-cc || { echo "[-] Compiling afl-cc failed. You seem not to have a working compiler." ; exit 1; } -man: afl-gcc all $(MANPAGES) +.PHONY: gcc_plugin +gcc_plugin: + -$(MAKE) -f GNUmakefile.gcc_plugin +.PHONY: man +man: $(MANPAGES) + +.PHONY: test +test: tests + +.PHONY: tests tests: source-only @cd test ; ./test-all.sh @rm -f test/errors +.PHONY: performance-tests performance-tests: performance-test +.PHONY: test-performance test-performance: performance-test +.PHONY: performance-test performance-test: source-only @cd test ; ./test-performance.sh # hint: make targets are also listed in the top level README.md +.PHONY: help help: @echo "HELP --- the following make targets exist:" @echo "==========================================" @echo "all: just the main afl++ binaries" @echo "binary-only: everything for binary-only fuzzing: qemu_mode, unicorn_mode, libdislocator, libtokencap" - @echo "source-only: everything for source code fuzzing: llvm_mode, gcc_plugin, libdislocator, libtokencap" + @echo "source-only: everything for source code fuzzing: gcc_plugin, libdislocator, libtokencap" @echo "distrib: everything (for both binary-only and source code fuzzing)" @echo "man: creates simple man pages from the help option of the programs" @echo "install: installs everything you have compiled with the build option above" @@ -322,8 +332,8 @@ help: @echo "==========================================" @echo e.g.: make ASAN_BUILD=1 +.PHONY: test_x86 ifndef AFL_NO_X86 - test_x86: @echo "[*] Checking for the default compiler cc..." @type $(CC) >/dev/null || ( echo; echo "Oops, looks like there is no compiler '"$(CC)"' in your path."; echo; echo "Don't panic! You can restart with '"$(_)" CC='."; echo; exit 1 ) @@ -332,148 +342,129 @@ test_x86: @echo "[*] Checking for the ability to compile x86 code..." @echo 'main() { __asm__("xorb %al, %al"); }' | $(CC) $(CFLAGS) -w -x c - -o .test1 || ( echo; echo "Oops, looks like your compiler can't generate x86 code."; echo; echo "Don't panic! You can use the LLVM or QEMU mode, but see docs/INSTALL first."; echo "(To ignore this error, set AFL_NO_X86=1 and try again.)"; echo; exit 1 ) @rm -f .test1 - else - test_x86: @echo "[!] Note: skipping x86 compilation checks (AFL_NO_X86 set)." - endif - +.PHONY: test_shm ifeq "$(SHMAT_OK)" "1" - test_shm: @echo "[+] shmat seems to be working." @rm -f .test2 - else - test_shm: @echo "[-] shmat seems not to be working, switching to mmap implementation" - endif - +.PHONY: test_python ifeq "$(PYTHON_OK)" "1" - test_python: @rm -f .test 2> /dev/null @echo "[+] $(PYTHON_VERSION) support seems to be working." - else - test_python: @echo "[-] You seem to need to install the package python3-dev, python2-dev or python-dev (and perhaps python[23]-apt), but it is optional so we continue" - endif - +.PHONY: ready ready: @echo "[+] Everything seems to be working, ready to compile." -afl-g++: afl-gcc - -afl-gcc: src/afl-gcc.c $(COMM_HDR) | test_x86 - $(CC) $(CFLAGS) $(CPPFLAGS) src/$@.c -o $@ $(LDFLAGS) - set -e; for i in afl-g++ afl-clang afl-clang++; do ln -sf afl-gcc $$i; done - afl-as: src/afl-as.c include/afl-as.h $(COMM_HDR) | test_x86 - $(CC) $(CFLAGS) $(CPPFLAGS) src/$@.c -o $@ $(LDFLAGS) - ln -sf afl-as as + $(CC) $(CFLAGS) src/$@.c -o $@ $(LDFLAGS) + @ln -sf afl-as as src/afl-performance.o : $(COMM_HDR) src/afl-performance.c include/hash.h - $(CC) $(CFLAGS) $(CPPFLAGS) -Iinclude $(SPECIAL_PERFORMANCE) -O3 -fno-unroll-loops -c src/afl-performance.c -o src/afl-performance.o + $(CC) -Iinclude $(SPECIAL_PERFORMANCE) -O3 -fno-unroll-loops -c src/afl-performance.c -o src/afl-performance.o src/afl-common.o : $(COMM_HDR) src/afl-common.c include/common.h - $(CC) $(CFLAGS) $(CFLAGS_FLTO) $(CPPFLAGS) -c src/afl-common.c -o src/afl-common.o + $(CC) $(CFLAGS) $(CFLAGS_FLTO) -c src/afl-common.c -o src/afl-common.o src/afl-forkserver.o : $(COMM_HDR) src/afl-forkserver.c include/forkserver.h - $(CC) $(CFLAGS) $(CFLAGS_FLTO) $(CPPFLAGS) -c src/afl-forkserver.c -o src/afl-forkserver.o + $(CC) $(CFLAGS) $(CFLAGS_FLTO) -c src/afl-forkserver.c -o src/afl-forkserver.o src/afl-sharedmem.o : $(COMM_HDR) src/afl-sharedmem.c include/sharedmem.h - $(CC) $(CFLAGS) $(CFLAGS_FLTO) $(CPPFLAGS) -c src/afl-sharedmem.c -o src/afl-sharedmem.o + $(CC) $(CFLAGS) $(CFLAGS_FLTO) -c src/afl-sharedmem.c -o src/afl-sharedmem.o afl-fuzz: $(COMM_HDR) include/afl-fuzz.h $(AFL_FUZZ_FILES) src/afl-common.o src/afl-sharedmem.o src/afl-forkserver.o src/afl-performance.o | test_x86 - $(CC) $(CFLAGS) $(COMPILE_STATIC) $(CFLAGS_FLTO) $(AFL_FUZZ_FILES) $(CPPFLAGS) src/afl-common.o src/afl-sharedmem.o src/afl-forkserver.o src/afl-performance.o -o $@ $(PYFLAGS) $(LDFLAGS) + $(CC) $(CFLAGS) $(COMPILE_STATIC) $(CFLAGS_FLTO) $(AFL_FUZZ_FILES) src/afl-common.o src/afl-sharedmem.o src/afl-forkserver.o src/afl-performance.o -o $@ $(PYFLAGS) $(LDFLAGS) afl-showmap: src/afl-showmap.c src/afl-common.o src/afl-sharedmem.o src/afl-forkserver.o $(COMM_HDR) | test_x86 - $(CC) $(CFLAGS) $(COMPILE_STATIC) $(CFLAGS_FLTO) $(CPPFLAGS) src/$@.c src/afl-common.o src/afl-sharedmem.o src/afl-forkserver.o -o $@ $(LDFLAGS) + $(CC) $(CFLAGS) $(COMPILE_STATIC) $(CFLAGS_FLTO) src/$@.c src/afl-common.o src/afl-sharedmem.o src/afl-forkserver.o -o $@ $(LDFLAGS) afl-tmin: src/afl-tmin.c src/afl-common.o src/afl-sharedmem.o src/afl-forkserver.o src/afl-performance.o $(COMM_HDR) | test_x86 - $(CC) $(CFLAGS) $(COMPILE_STATIC) $(CFLAGS_FLTO) $(CPPFLAGS) src/$@.c src/afl-common.o src/afl-sharedmem.o src/afl-forkserver.o src/afl-performance.o -o $@ $(LDFLAGS) + $(CC) $(CFLAGS) $(COMPILE_STATIC) $(CFLAGS_FLTO) src/$@.c src/afl-common.o src/afl-sharedmem.o src/afl-forkserver.o src/afl-performance.o -o $@ $(LDFLAGS) afl-analyze: src/afl-analyze.c src/afl-common.o src/afl-sharedmem.o src/afl-performance.o $(COMM_HDR) | test_x86 - $(CC) $(CFLAGS) $(COMPILE_STATIC) $(CFLAGS_FLTO) $(CPPFLAGS) src/$@.c src/afl-common.o src/afl-sharedmem.o src/afl-performance.o -o $@ $(LDFLAGS) + $(CC) $(CFLAGS) $(COMPILE_STATIC) $(CFLAGS_FLTO) src/$@.c src/afl-common.o src/afl-sharedmem.o src/afl-performance.o -o $@ $(LDFLAGS) afl-gotcpu: src/afl-gotcpu.c src/afl-common.o $(COMM_HDR) | test_x86 - $(CC) $(CFLAGS) $(COMPILE_STATIC) $(CFLAGS_FLTO) $(CPPFLAGS) src/$@.c src/afl-common.o -o $@ $(LDFLAGS) + $(CC) $(CFLAGS) $(COMPILE_STATIC) $(CFLAGS_FLTO) src/$@.c src/afl-common.o -o $@ $(LDFLAGS) +.PHONY: document +document: afl-fuzz-document # document all mutations and only do one run (use with only one input file!) -document: $(COMM_HDR) include/afl-fuzz.h $(AFL_FUZZ_FILES) src/afl-common.o src/afl-sharedmem.o src/afl-performance.o | test_x86 - $(CC) -D_DEBUG=\"1\" -D_AFL_DOCUMENT_MUTATIONS $(CFLAGS) $(CFLAGS_FLTO) $(AFL_FUZZ_FILES) $(CPPFLAGS) src/afl-common.o src/afl-sharedmem.o src/afl-forkserver.c src/afl-performance.o -o afl-fuzz-document $(PYFLAGS) $(LDFLAGS) +afl-fuzz-document: $(COMM_HDR) include/afl-fuzz.h $(AFL_FUZZ_FILES) src/afl-common.o src/afl-sharedmem.o src/afl-performance.o | test_x86 + $(CC) -D_DEBUG=\"1\" -D_AFL_DOCUMENT_MUTATIONS $(CFLAGS) $(CFLAGS_FLTO) $(AFL_FUZZ_FILES) src/afl-common.o src/afl-sharedmem.o src/afl-forkserver.c src/afl-performance.o -o afl-fuzz-document $(PYFLAGS) $(LDFLAGS) test/unittests/unit_maybe_alloc.o : $(COMM_HDR) include/alloc-inl.h test/unittests/unit_maybe_alloc.c $(AFL_FUZZ_FILES) - @$(CC) $(CFLAGS) $(ASAN_CFLAGS) $(CPPFLAGS) -c test/unittests/unit_maybe_alloc.c -o test/unittests/unit_maybe_alloc.o + @$(CC) $(CFLAGS) $(ASAN_CFLAGS) -c test/unittests/unit_maybe_alloc.c -o test/unittests/unit_maybe_alloc.o unit_maybe_alloc: test/unittests/unit_maybe_alloc.o - @$(CC) $(CFLAGS) $(CPPFLAGS) -Wl,--wrap=exit -Wl,--wrap=printf test/unittests/unit_maybe_alloc.o -o test/unittests/unit_maybe_alloc $(LDFLAGS) $(ASAN_LDFLAGS) -lcmocka + @$(CC) $(CFLAGS) -Wl,--wrap=exit -Wl,--wrap=printf test/unittests/unit_maybe_alloc.o -o test/unittests/unit_maybe_alloc $(LDFLAGS) $(ASAN_LDFLAGS) -lcmocka ./test/unittests/unit_maybe_alloc test/unittests/unit_hash.o : $(COMM_HDR) include/alloc-inl.h test/unittests/unit_hash.c $(AFL_FUZZ_FILES) src/afl-performance.o - @$(CC) $(CFLAGS) $(ASAN_CFLAGS) $(CPPFLAGS) -c test/unittests/unit_hash.c -o test/unittests/unit_hash.o + @$(CC) $(CFLAGS) $(ASAN_CFLAGS) -c test/unittests/unit_hash.c -o test/unittests/unit_hash.o unit_hash: test/unittests/unit_hash.o src/afl-performance.o - @$(CC) $(CFLAGS) $(CPPFLAGS) -Wl,--wrap=exit -Wl,--wrap=printf $^ -o test/unittests/unit_hash $(LDFLAGS) $(ASAN_LDFLAGS) -lcmocka + @$(CC) $(CFLAGS) -Wl,--wrap=exit -Wl,--wrap=printf $^ -o test/unittests/unit_hash $(LDFLAGS) $(ASAN_LDFLAGS) -lcmocka ./test/unittests/unit_hash test/unittests/unit_rand.o : $(COMM_HDR) include/alloc-inl.h test/unittests/unit_rand.c $(AFL_FUZZ_FILES) src/afl-performance.o - @$(CC) $(CFLAGS) $(ASAN_CFLAGS) $(CPPFLAGS) -c test/unittests/unit_rand.c -o test/unittests/unit_rand.o + @$(CC) $(CFLAGS) $(ASAN_CFLAGS) -c test/unittests/unit_rand.c -o test/unittests/unit_rand.o unit_rand: test/unittests/unit_rand.o src/afl-common.o src/afl-performance.o - @$(CC) $(CFLAGS) $(ASAN_CFLAGS) $(CPPFLAGS) -Wl,--wrap=exit -Wl,--wrap=printf $^ -o test/unittests/unit_rand $(LDFLAGS) $(ASAN_LDFLAGS) -lcmocka + @$(CC) $(CFLAGS) $(ASAN_CFLAGS) -Wl,--wrap=exit -Wl,--wrap=printf $^ -o test/unittests/unit_rand $(LDFLAGS) $(ASAN_LDFLAGS) -lcmocka ./test/unittests/unit_rand test/unittests/unit_list.o : $(COMM_HDR) include/list.h test/unittests/unit_list.c $(AFL_FUZZ_FILES) - @$(CC) $(CFLAGS) $(ASAN_CFLAGS) $(CPPFLAGS) -c test/unittests/unit_list.c -o test/unittests/unit_list.o + @$(CC) $(CFLAGS) $(ASAN_CFLAGS) -c test/unittests/unit_list.c -o test/unittests/unit_list.o unit_list: test/unittests/unit_list.o - @$(CC) $(CFLAGS) $(ASAN_CFLAGS) $(CPPFLAGS) -Wl,--wrap=exit -Wl,--wrap=printf test/unittests/unit_list.o -o test/unittests/unit_list $(LDFLAGS) $(ASAN_LDFLAGS) -lcmocka + @$(CC) $(CFLAGS) $(ASAN_CFLAGS) -Wl,--wrap=exit -Wl,--wrap=printf test/unittests/unit_list.o -o test/unittests/unit_list $(LDFLAGS) $(ASAN_LDFLAGS) -lcmocka ./test/unittests/unit_list test/unittests/unit_preallocable.o : $(COMM_HDR) include/alloc-inl.h test/unittests/unit_preallocable.c $(AFL_FUZZ_FILES) - @$(CC) $(CFLAGS) $(ASAN_CFLAGS) $(CPPFLAGS) -c test/unittests/unit_preallocable.c -o test/unittests/unit_preallocable.o + @$(CC) $(CFLAGS) $(ASAN_CFLAGS) -c test/unittests/unit_preallocable.c -o test/unittests/unit_preallocable.o unit_preallocable: test/unittests/unit_preallocable.o - @$(CC) $(CFLAGS) $(ASAN_CFLAGS) $(CPPFLAGS) -Wl,--wrap=exit -Wl,--wrap=printf test/unittests/unit_preallocable.o -o test/unittests/unit_preallocable $(LDFLAGS) $(ASAN_LDFLAGS) -lcmocka + @$(CC) $(CFLAGS) $(ASAN_CFLAGS) -Wl,--wrap=exit -Wl,--wrap=printf test/unittests/unit_preallocable.o -o test/unittests/unit_preallocable $(LDFLAGS) $(ASAN_LDFLAGS) -lcmocka ./test/unittests/unit_preallocable +.PHONY: unit_clean unit_clean: @rm -f ./test/unittests/unit_preallocable ./test/unittests/unit_list ./test/unittests/unit_maybe_alloc test/unittests/*.o +.PHONY: unit ifneq "$(shell uname)" "Darwin" - -unit: unit_maybe_alloc unit_preallocable unit_list unit_clean unit_rand unit_hash - +unit: unit_maybe_alloc unit_preallocable unit_list unit_clean unit_rand unit_hash else - unit: @echo [-] unit tests are skipped on Darwin \(lacks GNU linker feature --wrap\) - endif +.PHONY: code-format code-format: ./.custom-format.py -i src/*.c ./.custom-format.py -i include/*.h ./.custom-format.py -i libdislocator/*.c ./.custom-format.py -i libtokencap/*.c - ./.custom-format.py -i llvm_mode/*.c - ./.custom-format.py -i llvm_mode/*.h - ./.custom-format.py -i llvm_mode/*.cc - ./.custom-format.py -i gcc_plugin/*.c - @#./.custom-format.py -i gcc_plugin/*.h - ./.custom-format.py -i gcc_plugin/*.cc + ./.custom-format.py -i instrumentation/*.h + ./.custom-format.py -i instrumentation/*.cc + ./.custom-format.py -i instrumentation/*.c ./.custom-format.py -i custom_mutators/*/*.c @#./.custom-format.py -i custom_mutators/*/*.h # destroys input.h :-( ./.custom-format.py -i examples/*/*.c @@ -489,38 +480,40 @@ code-format: ./.custom-format.py -i *.c +.PHONY: test_build ifndef AFL_NO_X86 - -test_build: afl-gcc afl-as afl-showmap +test_build: afl-cc afl-as afl-showmap @echo "[*] Testing the CC wrapper and instrumentation output..." - @unset AFL_USE_ASAN AFL_USE_MSAN AFL_CC; AFL_DEBUG=1 AFL_INST_RATIO=100 AFL_AS_FORCE_INSTRUMENT=1 AFL_PATH=. ./$(TEST_CC) $(CFLAGS) test-instr.c -o test-instr $(LDFLAGS) 2>&1 | grep 'afl-as' >/dev/null || (echo "Oops, afl-as did not get called from "$(TEST_CC)". This is normally achieved by "$(CC)" honoring the -B option."; exit 1 ) + @unset AFL_USE_ASAN AFL_USE_MSAN AFL_CC; AFL_DEBUG=1 AFL_INST_RATIO=100 AFL_PATH=. ./$(TEST_CC) $(CFLAGS) test-instr.c -o test-instr $(LDFLAGS) 2>&1 | grep 'afl-as' >/dev/null || (echo "Oops, afl-as did not get called from "$(TEST_CC)". This is normally achieved by "$(CC)" honoring the -B option."; exit 1 ) ASAN_OPTIONS=detect_leaks=0 ./afl-showmap -m none -q -o .test-instr0 ./test-instr < /dev/null echo 1 | ASAN_OPTIONS=detect_leaks=0 ./afl-showmap -m none -q -o .test-instr1 ./test-instr @rm -f test-instr @cmp -s .test-instr0 .test-instr1; DR="$$?"; rm -f .test-instr0 .test-instr1; if [ "$$DR" = "0" ]; then echo; echo "Oops, the instrumentation does not seem to be behaving correctly!"; echo; echo "Please post to https://github.com/AFLplusplus/AFLplusplus/issues to troubleshoot the issue."; echo; exit 1; fi + @echo @echo "[+] All right, the instrumentation seems to be working!" - else - -test_build: afl-gcc afl-as afl-showmap +test_build: afl-cc afl-as afl-showmap @echo "[!] Note: skipping build tests (you may need to use LLVM or QEMU mode)." - endif - +.PHONY: all_done all_done: test_build - @if [ ! "`type clang 2>/dev/null`" = "" ]; then echo "[+] LLVM users: see llvm_mode/README.md for a faster alternative to afl-gcc."; fi + @test -e afl-cc && echo "[+] Main compiler 'afl-cc' successfully built!" || { echo "[-] Main compiler 'afl-cc' failed to built, set up a working build environment first!" ; exit 1 ; } + @test -e cmplog-instructions-pass.so && echo "[+] LLVM mode for 'afl-cc' successfully built!" || echo "[-] LLVM mode for 'afl-cc' failed to built, likely you either have not llvm installed or you have not set LLVM_CONFIG pointing to e.g. llvm-config-11. See instrumenation/README.llvm.md how to do this. Highly recommended!" + @test -e SanitizerCoverageLTO.so && echo "[+] LLVM LTO mode for 'afl-cc' successfully built!" || echo "[-] LLVM LTO mode for 'afl-cc' failed to built, this would need LLVM 11+, see instrumentation/README.lto.md how to build it" + @test -e afl-gcc-pass.so && echo "[+] gcc_plugin for 'afl-cc' successfully built!" || echo "[-] gcc_plugin for 'afl-cc' failed to built, unless you really need it that is fine - or read instrumentation/README.gcc_plugin.md how to build it" @echo "[+] All done! Be sure to review the README.md - it's pretty short and useful." @if [ "`uname`" = "Darwin" ]; then printf "\nWARNING: Fuzzing on MacOS X is slow because of the unusually high overhead of\nfork() on this OS. Consider using Linux or *BSD. You can also use VirtualBox\n(virtualbox.org) to put AFL inside a Linux or *BSD VM.\n\n"; fi @! tty <&1 >/dev/null || printf "\033[0;30mNOTE: If you can read this, your terminal probably uses white background.\nThis will make the UI hard to read. See docs/status_screen.md for advice.\033[0m\n" 2>/dev/null .NOTPARALLEL: clean all +.PHONY: clean clean: - rm -f $(PROGS) libradamsa.so afl-fuzz-document afl-as as afl-g++ afl-clang afl-clang++ *.o src/*.o *~ a.out core core.[1-9][0-9]* *.stackdump .test .test1 .test2 test-instr .test-instr0 .test-instr1 afl-qemu-trace afl-gcc-fast afl-gcc-pass.so afl-gcc-rt.o afl-g++-fast ld *.so *.8 test/unittests/*.o test/unittests/unit_maybe_alloc test/unittests/preallocable .afl-* + rm -f $(PROGS) libradamsa.so afl-fuzz-document afl-as as afl-g++ afl-clang afl-clang++ *.o src/*.o *~ a.out core core.[1-9][0-9]* *.stackdump .test .test1 .test2 test-instr .test-instr0 .test-instr1 afl-qemu-trace afl-gcc-fast afl-gcc-pass.so afl-g++-fast ld *.so *.8 test/unittests/*.o test/unittests/unit_maybe_alloc test/unittests/preallocable .afl-* afl-gcc afl-g++ rm -rf out_dir qemu_mode/qemu-3.1.1 *.dSYM */*.dSYM - -$(MAKE) -C llvm_mode clean - -$(MAKE) -C gcc_plugin clean + -$(MAKE) -f GNUmakefile.llvm clean + -$(MAKE) -f GNUmakefile.gcc_plugin clean $(MAKE) -C libdislocator clean $(MAKE) -C libtokencap clean $(MAKE) -C examples/afl_network_proxy clean @@ -530,20 +523,22 @@ clean: $(MAKE) -C qemu_mode/libcompcov clean rm -rf qemu_mode/qemu-3.1.1 ifeq "$(IN_REPO)" "1" - test -e unicorn_mode/unicornafl/Makefile && $(MAKE) -C unicorn_mode/unicornafl clean || true + test -d unicorn_mode/unicornafl && $(MAKE) -C unicorn_mode/unicornafl clean || true else rm -rf qemu_mode/qemu-3.1.1.tar.xz rm -rf unicorn_mode/unicornafl endif +.PHONY: deepclean deepclean: clean rm -rf qemu_mode/qemu-3.1.1.tar.xz rm -rf unicorn_mode/unicornafl git reset --hard >/dev/null 2>&1 || true +.PHONY: distrib distrib: all - -$(MAKE) -C llvm_mode - -$(MAKE) -C gcc_plugin + -$(MAKE) -f GNUmakefile.llvm + -$(MAKE) -f GNUmakefile.gcc_plugin $(MAKE) -C libdislocator $(MAKE) -C libtokencap $(MAKE) -C examples/afl_network_proxy @@ -552,6 +547,7 @@ distrib: all -cd qemu_mode && sh ./build_qemu_support.sh cd unicorn_mode && unset CFLAGS && sh ./build_unicorn_support.sh +.PHONY: binary-only binary-only: all $(MAKE) -C libdislocator $(MAKE) -C libtokencap @@ -561,9 +557,10 @@ binary-only: all -cd qemu_mode && sh ./build_qemu_support.sh cd unicorn_mode && unset CFLAGS && sh ./build_unicorn_support.sh +.PHONY: source-only source-only: all - -$(MAKE) -C llvm_mode - -$(MAKE) -C gcc_plugin + -$(MAKE) -f GNUmakefile.llvm + -$(MAKE) -f GNUmakefile.gcc_plugin $(MAKE) -C libdislocator $(MAKE) -C libtokencap @#$(MAKE) -C examples/afl_network_proxy @@ -573,8 +570,7 @@ source-only: all %.8: % @echo .TH $* 8 $(BUILD_DATE) "afl++" > $@ @echo .SH NAME >> $@ - @printf "%s" ".B $* \- " >> $@ - @./$* -h 2>&1 | head -n 1 | sed -e "s/$$(printf '\e')[^m]*m//g" >> $@ + @echo .B $* >> $@ @echo >> $@ @echo .SH SYNOPSIS >> $@ @./$* -h 2>&1 | head -n 3 | tail -n 1 | sed 's/^\.\///' >> $@ @@ -590,30 +586,28 @@ source-only: all @echo .SH LICENSE >> $@ @echo Apache License Version 2.0, January 2004 >> $@ +.PHONY: install install: all $(MANPAGES) - install -d -m 755 $${DESTDIR}$(BIN_PATH) $${DESTDIR}$(HELPER_PATH) $${DESTDIR}$(DOC_PATH) $${DESTDIR}$(MISC_PATH) - rm -f $${DESTDIR}$(BIN_PATH)/afl-plot.sh + @install -d -m 755 $${DESTDIR}$(BIN_PATH) $${DESTDIR}$(HELPER_PATH) $${DESTDIR}$(DOC_PATH) $${DESTDIR}$(MISC_PATH) + @rm -f $${DESTDIR}$(BIN_PATH)/afl-plot.sh + @rm -f $${DESTDIR}$(BIN_PATH)/afl-as install -m 755 $(PROGS) $(SH_PROGS) $${DESTDIR}$(BIN_PATH) - rm -f $${DESTDIR}$(BIN_PATH)/afl-as - if [ -f afl-qemu-trace ]; then install -m 755 afl-qemu-trace $${DESTDIR}$(BIN_PATH); fi - if [ -f afl-gcc-fast ]; then set e; install -m 755 afl-gcc-fast $${DESTDIR}$(BIN_PATH); ln -sf afl-gcc-fast $${DESTDIR}$(BIN_PATH)/afl-g++-fast; install -m 755 afl-gcc-pass.so afl-gcc-rt.o $${DESTDIR}$(HELPER_PATH); fi - if [ -f afl-clang-fast ]; then $(MAKE) -C llvm_mode install; fi - if [ -f libdislocator.so ]; then set -e; install -m 755 libdislocator.so $${DESTDIR}$(HELPER_PATH); fi - if [ -f libtokencap.so ]; then set -e; install -m 755 libtokencap.so $${DESTDIR}$(HELPER_PATH); fi - if [ -f libcompcov.so ]; then set -e; install -m 755 libcompcov.so $${DESTDIR}$(HELPER_PATH); fi - if [ -f afl-fuzz-document ]; then set -e; install -m 755 afl-fuzz-document $${DESTDIR}$(BIN_PATH); fi - if [ -f socketfuzz32.so -o -f socketfuzz64.so ]; then $(MAKE) -C examples/socket_fuzzing install; fi - if [ -f argvfuzz32.so -o -f argvfuzz64.so ]; then $(MAKE) -C examples/argv_fuzzing install; fi - if [ -f examples/afl_network_proxy/afl-network-server ]; then $(MAKE) -C examples/afl_network_proxy install; fi - if [ -f libAFLDriver.a ]; then install -m 644 libAFLDriver.a $${DESTDIR}$(HELPER_PATH); fi - if [ -f libAFLQemuDriver.a ]; then install -m 644 libAFLQemuDriver.a $${DESTDIR}$(HELPER_PATH); fi - - set -e; ln -sf afl-gcc $${DESTDIR}$(BIN_PATH)/afl-g++ - set -e; if [ -f afl-clang-fast ] ; then ln -sf afl-clang-fast $${DESTDIR}$(BIN_PATH)/afl-clang ; ln -sf afl-clang-fast $${DESTDIR}$(BIN_PATH)/afl-clang++ ; else ln -sf afl-gcc $${DESTDIR}$(BIN_PATH)/afl-clang ; ln -sf afl-gcc $${DESTDIR}$(BIN_PATH)/afl-clang++; fi - - mkdir -m 0755 -p ${DESTDIR}$(MAN_PATH) + @if [ -f afl-qemu-trace ]; then install -m 755 afl-qemu-trace $${DESTDIR}$(BIN_PATH); fi + @if [ -f libdislocator.so ]; then set -e; install -m 755 libdislocator.so $${DESTDIR}$(HELPER_PATH); fi + @if [ -f libtokencap.so ]; then set -e; install -m 755 libtokencap.so $${DESTDIR}$(HELPER_PATH); fi + @if [ -f libcompcov.so ]; then set -e; install -m 755 libcompcov.so $${DESTDIR}$(HELPER_PATH); fi + @if [ -f afl-fuzz-document ]; then set -e; install -m 755 afl-fuzz-document $${DESTDIR}$(BIN_PATH); fi + @if [ -f socketfuzz32.so -o -f socketfuzz64.so ]; then $(MAKE) -C examples/socket_fuzzing install; fi + @if [ -f argvfuzz32.so -o -f argvfuzz64.so ]; then $(MAKE) -C examples/argv_fuzzing install; fi + @if [ -f examples/afl_network_proxy/afl-network-server ]; then $(MAKE) -C examples/afl_network_proxy install; fi + @if [ -f examples/aflpp_driver/libAFLDriver.a ]; then install -m 644 examples/aflpp_driver/libAFLDriver.a $${DESTDIR}$(HELPER_PATH); fi + @if [ -f examples/aflpp_driver/libAFLQemuDriver.a ]; then install -m 644 examples/aflpp_driver/libAFLQemuDriver.a $${DESTDIR}$(HELPER_PATH); fi + -$(MAKE) -f GNUmakefile.llvm install + -$(MAKE) -f GNUmakefile.gcc_plugin install + ln -sf afl-cc $${DESTDIR}$(BIN_PATH)/afl-gcc + ln -sf afl-cc $${DESTDIR}$(BIN_PATH)/afl-g++ + @mkdir -m 0755 -p ${DESTDIR}$(MAN_PATH) install -m0644 *.8 ${DESTDIR}$(MAN_PATH) - install -m 755 afl-as $${DESTDIR}$(HELPER_PATH) ln -sf afl-as $${DESTDIR}$(HELPER_PATH)/as install -m 644 docs/*.md $${DESTDIR}$(DOC_PATH) diff --git a/GNUmakefile.llvm b/GNUmakefile.llvm index d4502319..d76e0b28 100644 --- a/GNUmakefile.llvm +++ b/GNUmakefile.llvm @@ -419,17 +419,14 @@ document: ./afl-compiler-rt.o: instrumentation/afl-compiler-rt.o.c | test_deps $(CLANG_BIN) $(CLANG_CFL) $(CFLAGS_SAFE) $(CPPFLAGS) -O3 -Wno-unused-result -fPIC -c $< -o $@ - ln -sf ./afl-compiler-rt.o ./afl-llvm-rt.o ./afl-compiler-rt-32.o: instrumentation/afl-compiler-rt.o.c | test_deps @printf "[*] Building 32-bit variant of the runtime (-m32)... " @$(CLANG_BIN) $(CLANG_CFL) $(CFLAGS_SAFE) $(CPPFLAGS) -O3 -Wno-unused-result -m32 -fPIC -c $< -o $@ 2>/dev/null; if [ "$$?" = "0" ]; then echo "success!"; else echo "failed (that's fine)"; fi - @test -e ./afl-compiler-rt-32.o && ln -sf ./afl-compiler-rt-32.o ./afl-llvm-rt-32.o ./afl-compiler-rt-64.o: instrumentation/afl-compiler-rt.o.c | test_deps @printf "[*] Building 64-bit variant of the runtime (-m64)... " @$(CLANG_BIN) $(CLANG_CFL) $(CFLAGS_SAFE) $(CPPFLAGS) -O3 -Wno-unused-result -m64 -fPIC -c $< -o $@ 2>/dev/null; if [ "$$?" = "0" ]; then echo "success!"; else echo "failed (that's fine)"; fi - @test -e ./afl-compiler-rt-64.o && ln -sf ./afl-compiler-rt-64.o ./afl-llvm-rt-64.o .PHONY: test_build test_build: $(PROGS) @@ -454,8 +451,8 @@ install: all @if [ -f ./afl-compiler-rt.o ]; then set -e; install -m 755 ./afl-compiler-rt.o $${DESTDIR}$(HELPER_PATH); fi @if [ -f ./afl-lto ]; then set -e; ln -sf afl-cc $${DESTDIR}$(BIN_PATH)/afl-lto; ln -sf afl-cc $${DESTDIR}$(BIN_PATH)/afl-lto++; ln -sf afl-cc $${DESTDIR}$(BIN_PATH)/afl-clang-lto; ln -sf afl-cc $${DESTDIR}$(BIN_PATH)/afl-clang-lto++; install -m 755 ./afl-llvm-lto-instrumentation.so ./afl-llvm-rt-lto*.o ./afl-llvm-lto-instrumentlist.so $${DESTDIR}$(HELPER_PATH); fi @if [ -f ./afl-ld-lto ]; then set -e; install -m 755 ./afl-ld-lto $${DESTDIR}$(BIN_PATH); fi - @if [ -f ./afl-compiler-rt-32.o ]; then set -e; install -m 755 ./afl-compiler-rt-32.o $${DESTDIR}$(HELPER_PATH); ln -sf afl-compiler-rt-32.o $${DESTDIR}$(HELPER_PATH)/afl-llvm-rt-32.o; fi - @if [ -f ./afl-compiler-rt-64.o ]; then set -e; install -m 755 ./afl-compiler-rt-64.o $${DESTDIR}$(HELPER_PATH); ln -sf afl-compiler-rt-64.o $${DESTDIR}$(HELPER_PATH)/afl-llvm-rt-64.o; fi + @if [ -f ./afl-compiler-rt-32.o ]; then set -e; install -m 755 ./afl-compiler-rt-32.o $${DESTDIR}$(HELPER_PATH); fi + @if [ -f ./afl-compiler-rt-64.o ]; then set -e; install -m 755 ./afl-compiler-rt-64.o $${DESTDIR}$(HELPER_PATH); fi @if [ -f ./compare-transform-pass.so ]; then set -e; install -m 755 ./*.so $${DESTDIR}$(HELPER_PATH); fi @if [ -f ./compare-transform-pass.so ]; then set -e; ln -sf afl-cc $${DESTDIR}$(BIN_PATH)/afl-clang-fast ; ln -sf ./afl-c++ $${DESTDIR}$(BIN_PATH)/afl-clang-fast++ ; ln -sf afl-cc $${DESTDIR}$(BIN_PATH)/afl-clang ; ln -sf ./afl-c++ $${DESTDIR}$(BIN_PATH)/afl-clang++ ; fi @if [ -f ./SanitizerCoverageLTO.so ]; then set -e; ln -sf afl-cc $${DESTDIR}$(BIN_PATH)/afl-clang-lto ; ln -sf ./afl-c++ $${DESTDIR}$(BIN_PATH)/afl-clang-lto++ ; fi diff --git a/README.md b/README.md index 4cad6b47..96b34260 100644 --- a/README.md +++ b/README.md @@ -4,9 +4,9 @@ ![Travis State](https://api.travis-ci.com/AFLplusplus/AFLplusplus.svg?branch=stable) - Release Version: [2.68c](https://github.com/AFLplusplus/AFLplusplus/releases) + Release Version: [2.67c](https://github.com/AFLplusplus/AFLplusplus/releases) - Github Version: 3.00a + Github Version: 2.67d Repository: [https://github.com/AFLplusplus/AFLplusplus](https://github.com/AFLplusplus/AFLplusplus) @@ -22,6 +22,26 @@ afl++ is a superior fork to Google's afl - more speed, more and better mutations, more and better instrumentation, custom module support, etc. +## Major changes in afl++ 3.0 + +With afl++ 3.0 we introduced changes that break some previous afl and afl++ +behaviours: + + * There are no llvm_mode and gcc_plugin subdirectories anymore and there is + only one compiler: afl-cc. All previous compilers now symlink to this one + compiler. All instrumentation source code is now in the `instrumentation/` + folder. + * qemu_mode got upgraded to QEMU 5.1, but to be able to build this a current + ninja build tool version and python3 setuptools are required. + qemu_mode also got new options like snapshotting, instrumenting specific + shared libraries, etc. and QEMU 5.1 supports more CPU targets so this is + worth it. + * When instrumenting targets, afl-cc will not supersede optimizations. This + allows to fuzz targets as same as they are built for debug or release. + * afl-fuzz' `-i` option now descends into subdirectories. + * afl-fuzz will skip over empty dictionaries and too large test cases instead + of failing. + ## Contents 1. [Features](#important-features-of-afl) @@ -39,7 +59,7 @@ with laf-intel and redqueen, unicorn mode, gcc plugin, full *BSD, Solaris and Android support and much, much, much more. - | Feature/Instrumentation | afl-gcc | llvm_mode | gcc_plugin | qemu_mode | unicorn_mode | + | Feature/Instrumentation | afl-gcc | llvm | gcc_plugin | qemu_mode | unicorn_mode | | -------------------------|:-------:|:---------:|:----------:|:----------------:|:------------:| | NeverZero | x86[_64]| x(1) | (2) | x | x | | Persistent Mode | | x | x | x86[_64]/arm[64] | x | @@ -47,9 +67,8 @@ | CmpLog | | x | | x86[_64]/arm[64] | | | Selective Instrumentation| | x | x | (x)(3) | | | Non-Colliding Coverage | | x(4) | | (x)(5) | | - | InsTrim | | x | | | | | Ngram prev_loc Coverage | | x(6) | | | | - | Context Coverage | | x | | | | + | Context Coverage | | x(6) | | | | | Auto Dictionary | | x(7) | | | | | Snapshot LKM Support | | x | | (x)(5) | | @@ -59,11 +78,11 @@ 4. with pcguard mode and LTO mode for LLVM >= 11 5. upcoming, development in the branch 6. not compatible with LTO instrumentation and needs at least LLVM >= 4.1 - 7. only in LTO mode with LLVM >= 11 + 7. automatic in LTO mode with LLVM >= 11, an extra pass for all LLVM version that writes to a file to use with afl-fuzz' `-x` Among others, the following features and patches have been integrated: - * NeverZero patch for afl-gcc, llvm_mode, qemu_mode and unicorn_mode which prevents a wrapping map value to zero, increases coverage + * NeverZero patch for afl-gcc, instrumentation, qemu_mode and unicorn_mode which prevents a wrapping map value to zero, increases coverage * Persistent mode, deferred forkserver and in-memory fuzzing for qemu_mode * Unicorn mode which allows fuzzing of binaries from completely different platforms (integration provided by domenukk) * The new CmpLog instrumentation for LLVM and QEMU inspired by [Redqueen](https://www.syssec.ruhr-uni-bochum.de/media/emma/veroeffentlichungen/2018/12/17/NDSS19-Redqueen.pdf) @@ -71,10 +90,9 @@ * AFLfast's power schedules by Marcel Böhme: [https://github.com/mboehme/aflfast](https://github.com/mboehme/aflfast) * The MOpt mutator: [https://github.com/puppet-meteor/MOpt-AFL](https://github.com/puppet-meteor/MOpt-AFL) * LLVM mode Ngram coverage by Adrian Herrera [https://github.com/adrianherrera/afl-ngram-pass](https://github.com/adrianherrera/afl-ngram-pass) - * InsTrim, a CFG llvm_mode instrumentation implementation: [https://github.com/csienslab/instrim](https://github.com/csienslab/instrim) * C. Holler's afl-fuzz Python mutator module: [https://github.com/choller/afl](https://github.com/choller/afl) * Custom mutator by a library (instead of Python) by kyakdan - * LAF-Intel/CompCov support for llvm_mode, qemu_mode and unicorn_mode (with enhanced capabilities) + * LAF-Intel/CompCov support for instrumentation, qemu_mode and unicorn_mode (with enhanced capabilities) * Radamsa and honggfuzz mutators (as custom mutators). * QBDI mode to fuzz android native libraries via Quarkslab's [QBDI](https://github.com/QBDI/QBDI) framework * Frida and ptrace mode to fuzz binary-only libraries, etc. @@ -88,7 +106,7 @@ send a mail to . See [docs/QuickStartGuide.md](docs/QuickStartGuide.md) if you don't have time to - read this file. + read this file - however this is not recommended! ## Branches @@ -105,13 +123,14 @@ ## Help wanted -We are happy to be part of [Google Summer of Code 2020](https://summerofcode.withgoogle.com/organizations/5100744400699392/)! :-) +We were happy to be part of [Google Summer of Code 2020](https://summerofcode.withgoogle.com/organizations/5100744400699392/) +and we will try to participate again in 2021! We have several ideas we would like to see in AFL++ to make it even better. However, we already work on so many things that we do not have the time for all the big ideas. -This can be your way to support and contribute to AFL++ - extend it to +This can be your way to support and contribute to AFL++ - extend it to do something cool. We have an idea list in [docs/ideas.md](docs/ideas.md). @@ -132,7 +151,7 @@ This image is automatically generated when a push to the stable repo happens. You will find your target source code in /src in the container. If you want to build afl++ yourself you have many options. -The easiest is to build and install everything: +The easiest choice is to build and install everything: ```shell sudo apt install build-essential libtool-bin python3-dev automake flex bison libglib2.0-dev libpixman-1-dev clang python3-setuptools llvm @@ -142,9 +161,9 @@ sudo make install It is recommended to install the newest available gcc, clang and llvm-dev possible in your distribution! -Note that "make distrib" also builds llvm_mode, qemu_mode, unicorn_mode and +Note that "make distrib" also builds instrumentation, qemu_mode, unicorn_mode and more. If you just want plain afl++ then do "make all", however compiling and -using at least llvm_mode is highly recommended for much better results - +using at least instrumentation is highly recommended for much better results - hence in this case ```shell @@ -156,7 +175,7 @@ These build targets exist: * all: just the main afl++ binaries * binary-only: everything for binary-only fuzzing: qemu_mode, unicorn_mode, libdislocator, libtokencap -* source-only: everything for source code fuzzing: llvm_mode, libdislocator, libtokencap +* source-only: everything for source code fuzzing: instrumentation, libdislocator, libtokencap * distrib: everything (for both binary-only and source code fuzzing) * man: creates simple man pages from the help option of the programs * install: installs everything you have compiled with the build options above @@ -212,18 +231,19 @@ If you have a binary-only target please skip to [#Instrumenting binary-only apps Fuzzing source code is a three-step process. -1. compile the target with a special compiler that prepares the target to be +1. Compile the target with a special compiler that prepares the target to be fuzzed efficiently. This step is called "instrumenting a target". 2. Prepare the fuzzing by selecting and optimizing the input corpus for the target. -3. perform the fuzzing of the target by randomly mutating input and assessing +3. Perform the fuzzing of the target by randomly mutating input and assessing if a generated input was processed in a new path in the target binary. ### 1. Instrumenting that target #### a) Selecting the best afl++ compiler for instrumenting the target -afl++ comes with different compilers and instrumentation options. +afl++ comes with a central compiler `afl-cc` that incorporates various different +kinds of compiler targets and and instrumentation options. The following evaluation flow will help you to select the best possible. It is highly recommended to have the newest llvm version possible installed, @@ -231,49 +251,62 @@ anything below 9 is not recommended. ``` +--------------------------------+ -| clang/clang++ 11+ is available | --> use afl-clang-lto and afl-clang-lto++ -+--------------------------------+ see [llvm/README.lto.md](llvm/README.lto.md) +| clang/clang++ 11+ is available | --> use LTO mode (afl-clang-lto/afl-clang-lto++) ++--------------------------------+ see [instrumentation/README.lto.md](instrumentation/README.lto.md) | - | if not, or if the target fails with afl-clang-lto/++ + | if not, or if the target fails with LTO afl-clang-lto/++ | v +---------------------------------+ -| clang/clang++ 3.3+ is available | --> use afl-clang-fast and afl-clang-fast++ -+---------------------------------+ see [llvm/README.md](llvm/README.md) +| clang/clang++ 3.3+ is available | --> use LLVM mode (afl-clang-fast/afl-clang-fast++) ++---------------------------------+ see [instrumentation/README.md](instrumentation/README.md) | - | if not, or if the target fails with afl-clang-fast/++ + | if not, or if the target fails with LLVM afl-clang-fast/++ | v +--------------------------------+ - | if you want to instrument only | -> use afl-gcc-fast and afl-gcc-fast++ - | parts of the target | see [gcc_plugin/README.md](gcc_plugin/README.md) and - +--------------------------------+ [gcc_plugin/README.instrument_list.md](gcc_plugin/README.instrument_list.md) + | if you want to instrument only | -> use GCC_PLUGIN mode (afl-gcc-fast/afl-g++-fast) + | parts of the target | see [instrumentation/README.gcc_plugin.md](instrumentation/README.gcc_plugin.md) and + +--------------------------------+ [instrumentation/README.instrument_list.md](instrumentation/README.instrument_list.md) | | if not, or if you do not have a gcc with plugin support | v - use afl-gcc and afl-g++ (or afl-clang and afl-clang++) + use GCC mode (afl-gcc/afl-g++) (or afl-clang/afl-clang++ for clang) ``` Clickable README links for the chosen compiler: - * [afl-clang-lto](llvm/README.lto.md) - * [afl-clang-fast](llvm/README.md) - * [afl-gcc-fast](gcc_plugin/README.md) - * afl-gcc has no README as it has no features + * [LTO mode - afl-clang-lto](instrumentation/README.lto.md) + * [LLVM mode - afl-clang-fast](instrumentation/README.md) + * [GCC_PLUGIN mode - afl-gcc-fast](instrumentation/README.gcc_plugin.md) + * GCC mode (afl-gcc) has no README as it has no own features + +You can select the mode for the afl-cc compiler by: + 1. passing --afl-MODE command line options to the compiler via CFLAGS/CXXFLAGS/CPPFLAGS + 2. use a symlink to afl-cc: afl-gcc, afl-g++, afl-clang, afl-clang++, + afl-clang-fast, afl-clang-fast++, afl-clang-lto, afl-clang-lto++, + afl-gcc-fast, afl-g++-fast + 3. using the environment variable AFL_CC_COMPILER with MODE + +MODE can be one of: LTO (afl-clang-lto*), LLVM (afl-clang-fast*), GCC_PLUGIN +(afl-g*-fast) or GCC (afl-gcc/afl-g++). + +Because no afl specific command-line options are accepted (beside the +--afl-MODE command), the compile-time tools make fairly broad use of environment +variables, which can be listed with `afl-cc -hh` or by reading [docs/env_variables.md](docs/env_variables.md). #### b) Selecting instrumentation options -The following options are available when you instrument with afl-clang-fast or -afl-clang-lto: +The following options are available when you instrument with LTO mode (afl-clang-fast/afl-clang-lto): * Splitting integer, string, float and switch comparisons so afl++ can easier solve these. This is an important option if you do not have a very good and large input corpus. This technique is called laf-intel or COMPCOV. To use this set the following environment variable before compiling the target: `export AFL_LLVM_LAF_ALL=1` - You can read more about this in [llvm/README.laf-intel.md](llvm/README.laf-intel.md) - * A different technique (and usually a better than laf-intel) is to + You can read more about this in [instrumentation/README.laf-intel.md](instrumentation/README.laf-intel.md) + * A different technique (and usually a better one than laf-intel) is to instrument the target so that any compare values in the target are sent to afl++ which then tries to put these values into the fuzzing data at different locations. This technique is very fast and good - if the target does not @@ -282,12 +315,13 @@ afl-clang-lto: If you want to use this technique, then you have to compile the target twice, once specifically with/for this mode, and pass this binary to afl-fuzz via the `-c` parameter. - Not that you can compile also just a cmplog binary and use that for both - however there will a performance penality. - You can read more about this in [llvm_mode/README.cmplog.md](llvm_mode/README.cmplog.md) + Note that you can compile also just a cmplog binary and use that for both + however there will be a performance penality. + You can read more about this in [instrumentation/README.cmplog.md](instrumentation/README.cmplog.md) -If you use afl-clang-fast, afl-clang-lto or afl-gcc-fast you have the option to -selectively only instrument parts of the target that you are interested in: +If you use LTO, LLVM or GCC_PLUGIN mode (afl-clang-fast/afl-clang-lto/afl-gcc-fast) + you have the option to selectively only instrument parts of the target that you +are interested in: * To instrument only those parts of the target that you are interested in create a file with all the filenames of the source code that should be @@ -299,29 +333,29 @@ selectively only instrument parts of the target that you are interested in: `export AFL_LLVM_DENYLIST=denylist.txt` - depending on if you want per default to instrument unless noted (DENYLIST) or not perform instrumentation unless requested (ALLOWLIST). - **NOTE:** In optimization functions might be inlined and then not match! - see [llvm_mode/README.instrument_list.md](llvm_mode/README.instrument_list.md) + **NOTE:** During optimization functions might be inlined and then would not match! + See [instrumentation/README.instrument_list.md](instrumentation/README.instrument_list.md) For afl-clang-fast > 6.0 or if PCGUARD instrumentation is used then use the llvm sancov allow-list feature: [http://clang.llvm.org/docs/SanitizerCoverage.html](http://clang.llvm.org/docs/SanitizerCoverage.html) The llvm sancov format works with the allowlist/denylist feature of afl++ - however afl++ is more flexible in the format. + however afl++'s format is more flexible. There are many more options and modes available however these are most of the time less effective. See: - * [llvm_mode/README.ctx.md](llvm_mode/README.ctx.md) - * [llvm_mode/README.ngram.md](llvm_mode/README.ngram.md) - * [llvm_mode/README.instrim.md](llvm_mode/README.instrim.md) + * [instrumentation/README.ctx.md](instrumentation/README.ctx.md) + * [instrumentation/README.ngram.md](instrumentation/README.ngram.md) + * [instrumentation/README.instrim.md](instrumentation/README.instrim.md) -afl++ employs never zero counting in its bitmap. You can read more about this +afl++ performs "never zero" counting in its bitmap. You can read more about this here: - * [llvm_mode/README.neverzero.md](llvm_mode/README.neverzero.md) + * [instrumentation/README.neverzero.md](instrumentation/README.neverzero.md) #### c) Modify the target If the target has features that make fuzzing more difficult, e.g. checksums, HMAC, etc. then modify the source code so that this is removed. -This can even be done for productional source code be eliminating +This can even be done for operational source code by eliminating these checks within this specific defines: ``` @@ -332,13 +366,15 @@ these checks within this specific defines: #endif ``` +All afl++ compilers will set this preprocessor definition automatically. + #### d) Instrument the target In this step the target source code is compiled so that it can be fuzzed. Basically you have to tell the target build system that the selected afl++ compiler is used. Also - if possible - you should always configure the -build system that the target is compiled statically and not dynamically. +build system such that the target is compiled statically and not dynamically. How to do this is described below. Then build the target. (Usually with `make`) @@ -349,20 +385,22 @@ For `configure` build systems this is usually done by: `CC=afl-clang-fast CXX=afl-clang-fast++ ./configure --disable-shared` Note that if you are using the (better) afl-clang-lto compiler you also have to -set AR to llvm-ar[-VERSION] and RANLIB to llvm-ranlib[-VERSION] - as it is -described in [llvm/README.lto.md](llvm/README.lto.md) +set AR to llvm-ar[-VERSION] and RANLIB to llvm-ranlib[-VERSION] - as is +described in [instrumentation/README.lto.md](instrumentation/README.lto.md). ##### cmake -For `configure` build systems this is usually done by: -`mkdir build; cd build; CC=afl-clang-fast CXX=afl-clang-fast++ cmake ..` - -Some cmake scripts require something like `-DCMAKE_CC=... -DCMAKE_CXX=...` -or `-DCMAKE_C_COMPILER=... DCMAKE_CPP_COMPILER=...` instead. +For `cmake` build systems this is usually done by: +`mkdir build; cmake -DCMAKE_C_COMPILERC=afl-cc -DCMAKE_CXX_COMPILER=afl-c++ ..` Note that if you are using the (better) afl-clang-lto compiler you also have to -set AR to llvm-ar[-VERSION] and RANLIB to llvm-ranlib[-VERSION] - as it is -described in [llvm/README.lto.md](llvm/README.lto.md) +set AR to llvm-ar[-VERSION] and RANLIB to llvm-ranlib[-VERSION] - as is +described in [instrumentation/README.lto.md](instrumentation/README.lto.md). + +##### meson + +For meson you have to set the afl++ compiler with the very first command! +`CC=afl-cc CXX=afl-c++ meson` ##### other build systems or if configure/cmake didn't work @@ -370,7 +408,7 @@ Sometimes cmake and configure do not pick up the afl++ compiler, or the ranlib/ar that is needed - because this was just not foreseen by the developer of the target. Or they have non-standard options. Figure out if there is a non-standard way to set this, otherwise set up the build normally and edit the -generated build environment afterwards manually to point to the right compiler +generated build environment afterwards manually to point it to the right compiler (and/or ranlib and ar). #### d) Better instrumentation @@ -383,12 +421,12 @@ This requires the usage of afl-clang-lto or afl-clang-fast. This is the so-called `persistent mode`, which is much, much faster but requires that you code a source file that is specifically calling the target functions that you want to fuzz, plus a few specific afl++ functions around -it. See [llvm_mode/README.persistent_mode.md](llvm_mode/README.persistent_mode.md) for details. +it. See [instrumentation/README.persistent_mode.md](instrumentation/README.persistent_mode.md) for details. Basically if you do not fuzz a target in persistent mode then you are just doing it for a hobby and not professionally :-) -### 2. Preparing the fuzzing +### 2. Preparing the fuzzing campaign As you fuzz the target with mutated input, having as diverse inputs for the target as possible improves the efficiency a lot. @@ -401,7 +439,7 @@ reported bugs, test suites, random downloads from the internet, unit test case data - from all kind of PNG software. If the input format is not known, you can also modify a target program to write -away normal data it receives and processes to a file and use these. +normal data it receives and processes to a file and use these. #### b) Making the input corpus unique @@ -415,7 +453,7 @@ the run afl-cmin like this: `afl-cmin -i INPUTS -o INPUTS_UNIQUE -- bin/target -d @@` Note that the INPUTFILE argument that the target program would read from has to be set as `@@`. -If the target reads from stdin instead, just omit the `@@` as this is the +If the target reads from stdin instead, just omit the `@@` as this is the default. #### c) Minimizing all corpus files @@ -432,7 +470,7 @@ for i in *; do done ``` -This can also be parallelized, e.g. with `parallel` +This step can also be parallelized, e.g. with `parallel` #### Done! @@ -456,7 +494,7 @@ before the start of afl-fuzz as this improves performance by a x2 speed increase #### a) Running afl-fuzz -Before to do even a test run of afl-fuzz execute `sudo afl-system-config` (on +Before you do even a test run of afl-fuzz execute `sudo afl-system-config` (on the host if you execute afl-fuzz in a docker container). This reconfigures the system for optimal speed - which afl-fuzz checks and bails otherwise. Set `export AFL_SKIP_CPUFREQ=1` for afl-fuzz to skip this check if you cannot @@ -588,7 +626,7 @@ then terminate it. The main node will pick it up and make it available to the other secondary nodes over time. Set `export AFL_NO_AFFINITY=1` if you have no free core. -Note that you in nearly all cases you can never reach full coverage. A lot of +Note that you in nearly all cases can never reach full coverage. A lot of functionality is usually behind options that were not activated or fuzz e.g. if you fuzz a library to convert image formats and your target is the png to tiff API then you will not touch any of the other library APIs and features. @@ -607,7 +645,7 @@ switch or honggfuzz. #### f) Improve the speed! - * Use [persistent mode](llvm_mode/README.persistent_mode.md) (x2-x20 speed increase) + * Use [persistent mode](instrumentation/README.persistent_mode.md) (x2-x20 speed increase) * If you do not use shmem persistent mode, use `AFL_TMPDIR` to point the input file on a tempfs location, see [docs/env_variables.md](docs/env_variables.md) * Linux: Use the [afl++ snapshot module](https://github.com/AFLplusplus/AFL-Snapshot-LKM) (x2 speed increase) * Linux: Improve kernel performance: modify `/etc/default/grub`, set `GRUB_CMDLINE_LINUX_DEFAULT="ibpb=off ibrs=off kpti=off l1tf=off mds=off mitigations=off no_stf_barrier noibpb noibrs nopcid nopti nospec_store_bypass_disable nospectre_v1 nospectre_v2 pcid=off pti=off spec_store_bypass_disable=off spectre_v2=off stf_barrier=off"`; then `update-grub` and `reboot` (warning: makes the system more insecure) @@ -1035,7 +1073,6 @@ without feedback, bug reports, or patches from: Andrea Biondo Vincent Le Garrec Khaled Yakdan Kuang-che Wu Josephine Calliotte Konrad Welc - Thomas Rooijakkers ``` Thank you! diff --git a/include/afl-fuzz.h b/include/afl-fuzz.h index 1a05f4f4..4281c554 100644 --- a/include/afl-fuzz.h +++ b/include/afl-fuzz.h @@ -162,8 +162,7 @@ struct queue_entry { u8 *trace_mini; /* Trace bytes, if kept */ u32 tc_ref; /* Trace bytes ref count */ - struct queue_entry *next, /* Next element, if any */ - *next_100; /* 100 elements ahead */ + struct queue_entry *next; /* Next element, if any */ }; @@ -575,8 +574,7 @@ typedef struct afl_state { struct queue_entry *queue, /* Fuzzing queue (linked list) */ *queue_cur, /* Current offset within the queue */ - *queue_top, /* Top of the list */ - *q_prev100; /* Previous 100 marker */ + *queue_top; /* Top of the list */ // growing buf struct queue_entry **queue_buf; diff --git a/src/afl-fuzz-extras.c b/src/afl-fuzz-extras.c index d6c368d1..58ce5b6f 100644 --- a/src/afl-fuzz-extras.c +++ b/src/afl-fuzz-extras.c @@ -101,7 +101,8 @@ void load_extras_file(afl_state_t *afl, u8 *fname, u32 *min_len, u32 *max_len, if (rptr < lptr || *rptr != '"') { - FATAL("Malformed name=\"value\" pair in line %u.", cur_line); + WARNF("Malformed name=\"value\" pair in line %u.", cur_line); + continue; } @@ -141,13 +142,19 @@ void load_extras_file(afl_state_t *afl, u8 *fname, u32 *min_len, u32 *max_len, if (*lptr != '"') { - FATAL("Malformed name=\"keyword\" pair in line %u.", cur_line); + WARNF("Malformed name=\"keyword\" pair in line %u.", cur_line); + continue; } ++lptr; - if (!*lptr) { FATAL("Empty keyword in line %u.", cur_line); } + if (!*lptr) { + + WARNF("Empty keyword in line %u.", cur_line); + continue; + + } /* Okay, let's allocate memory and copy data between "...", handling \xNN escaping, \\, and \". */ @@ -169,7 +176,9 @@ void load_extras_file(afl_state_t *afl, u8 *fname, u32 *min_len, u32 *max_len, case 1 ... 31: case 128 ... 255: - FATAL("Non-printable characters in line %u.", cur_line); + WARNF("Non-printable characters in line %u.", cur_line); + continue; + break; case '\\': @@ -185,7 +194,8 @@ void load_extras_file(afl_state_t *afl, u8 *fname, u32 *min_len, u32 *max_len, if (*lptr != 'x' || !isxdigit(lptr[1]) || !isxdigit(lptr[2])) { - FATAL("Invalid escaping (not \\xNN) in line %u.", cur_line); + WARNF("Invalid escaping (not \\xNN) in line %u.", cur_line); + continue; } @@ -209,10 +219,11 @@ void load_extras_file(afl_state_t *afl, u8 *fname, u32 *min_len, u32 *max_len, if (afl->extras[afl->extras_cnt].len > MAX_DICT_FILE) { - FATAL( + WARNF( "Keyword too big in line %u (%s, limit is %s)", cur_line, stringify_mem_size(val_bufs[0], sizeof(val_bufs[0]), klen), stringify_mem_size(val_bufs[1], sizeof(val_bufs[1]), MAX_DICT_FILE)); + continue; } @@ -232,14 +243,19 @@ static void extras_check_and_sort(afl_state_t *afl, u32 min_len, u32 max_len, u8 val_bufs[2][STRINGIFY_VAL_SIZE_MAX]; - if (!afl->extras_cnt) { FATAL("No usable files in '%s'", dir); } + if (!afl->extras_cnt) { + + WARNF("No usable data in '%s'", dir); + return; + + } qsort(afl->extras, afl->extras_cnt, sizeof(struct extra_data), compare_extras_len); - OKF("Loaded %u extra tokens, size range %s to %s.", afl->extras_cnt, - stringify_mem_size(val_bufs[0], sizeof(val_bufs[0]), min_len), - stringify_mem_size(val_bufs[1], sizeof(val_bufs[1]), max_len)); + ACTF("Loaded %u extra tokens, size range %s to %s.", afl->extras_cnt, + stringify_mem_size(val_bufs[0], sizeof(val_bufs[0]), min_len), + stringify_mem_size(val_bufs[1], sizeof(val_bufs[1]), max_len)); if (max_len > 32) { @@ -250,8 +266,8 @@ static void extras_check_and_sort(afl_state_t *afl, u32 min_len, u32 max_len, if (afl->extras_cnt > afl->max_det_extras) { - OKF("More than %d tokens - will use them probabilistically.", - afl->max_det_extras); + WARNF("More than %d tokens - will use them probabilistically.", + afl->max_det_extras); } @@ -320,9 +336,10 @@ void load_extras(afl_state_t *afl, u8 *dir) { if (st.st_size > MAX_DICT_FILE) { WARNF( - "Extra '%s' is very big (%s, limit is %s)", fn, + "Extra '%s' is too big (%s, limit is %s)", fn, stringify_mem_size(val_bufs[0], sizeof(val_bufs[0]), st.st_size), stringify_mem_size(val_bufs[1], sizeof(val_bufs[1]), MAX_DICT_FILE)); + continue; } @@ -370,16 +387,74 @@ static inline u8 memcmp_nocase(u8 *m1, u8 *m2, u32 len) { } -/* Adds a new extra / dict entry. Used for LTO autodict. */ +/* Removes duplicates from the loaded extras. This can happen if multiple files + are loaded */ + +void dedup_extras(afl_state_t *afl) { + + if (afl->extras_cnt < 2) return; + + u32 i, j, orig_cnt = afl->extras_cnt; + + for (i = 0; i < afl->extras_cnt - 1; i++) { + + for (j = i + 1; j < afl->extras_cnt; j++) { + + restart_dedup: + + // if the goto was used we could be at the end of the list + if (j >= afl->extras_cnt || afl->extras[i].len != afl->extras[j].len) + break; + + if (memcmp(afl->extras[i].data, afl->extras[j].data, + afl->extras[i].len) == 0) { + + ck_free(afl->extras[j].data); + if (j + 1 < afl->extras_cnt) // not at the end of the list? + memmove((char *)&afl->extras[j], (char *)&afl->extras[j + 1], + (afl->extras_cnt - j - 1) * sizeof(struct extra_data)); + afl->extras_cnt--; + goto restart_dedup; // restart if several duplicates are in a row + + } + + } + + } + + if (afl->extras_cnt != orig_cnt) + afl->extras = afl_realloc((void **)&afl->extras, + afl->extras_cnt * sizeof(struct extra_data)); + +} + +/* Adds a new extra / dict entry. */ void add_extra(afl_state_t *afl, u8 *mem, u32 len) { - u8 val_bufs[2][STRINGIFY_VAL_SIZE_MAX]; + u8 val_bufs[2][STRINGIFY_VAL_SIZE_MAX]; + u32 i, found = 0; + + for (i = 0; i < afl->extras_cnt; i++) { + + if (afl->extras[i].len == len) { + + if (memcmp(afl->extras[i].data, mem, len) == 0) return; + found = 1; + + } else { + + if (found) break; + + } + + } if (len > MAX_DICT_FILE) { - WARNF("Extra '%.*s' is very big (%s, limit is %s)", (int)len, mem, + WARNF("Extra '%.*s' is too big (%s, limit is %s)", (int)len, mem, stringify_mem_size(val_bufs[0], sizeof(val_bufs[0]), len), stringify_mem_size(val_bufs[1], sizeof(val_bufs[1]), MAX_DICT_FILE)); + return; } else if (len > 32) { @@ -405,8 +480,8 @@ void add_extra(afl_state_t *afl, u8 *mem, u32 len) { if (afl->extras_cnt == afl->max_det_extras + 1) { - OKF("More than %d tokens - will use them probabilistically.", - afl->max_det_extras); + WARNF("More than %d tokens - will use them probabilistically.", + afl->max_det_extras); } @@ -609,7 +684,7 @@ void load_auto(afl_state_t *afl) { } else { - OKF("No auto-generated dictionary tokens to reuse."); + ACTF("No auto-generated dictionary tokens to reuse."); } diff --git a/src/afl-fuzz-init.c b/src/afl-fuzz-init.c index 102f04b9..713849a1 100644 --- a/src/afl-fuzz-init.c +++ b/src/afl-fuzz-init.c @@ -611,17 +611,17 @@ void read_foreign_testcases(afl_state_t *afl, int first) { /* Read all testcases from the input directory, then queue them for testing. Called at startup. */ -void read_testcases(afl_state_t *afl) { +void read_testcases(afl_state_t *afl, u8 *directory) { struct dirent **nl; - s32 nl_cnt; + s32 nl_cnt, subdirs = 1; u32 i; - u8 * fn1; - + u8 * fn1, *dir = directory; u8 val_buf[2][STRINGIFY_VAL_SIZE_MAX]; /* Auto-detect non-in-place resumption attempts. */ +if (dir == NULL) { fn1 = alloc_printf("%s/queue", afl->in_dir); if (!access(fn1, F_OK)) { @@ -632,16 +632,18 @@ void read_testcases(afl_state_t *afl) { ck_free(fn1); } + dir = afl->in_dir; +} - ACTF("Scanning '%s'...", afl->in_dir); + ACTF("Scanning '%s'...", dir); /* We use scandir() + alphasort() rather than readdir() because otherwise, the ordering of test cases would vary somewhat randomly and would be difficult to control. */ - nl_cnt = scandir(afl->in_dir, &nl, NULL, alphasort); + nl_cnt = scandir(dir, &nl, NULL, alphasort); - if (nl_cnt < 0) { + if (nl_cnt < 0 && directory == NULL) { if (errno == ENOENT || errno == ENOTDIR) { @@ -656,7 +658,7 @@ void read_testcases(afl_state_t *afl) { } - PFATAL("Unable to open '%s'", afl->in_dir); + PFATAL("Unable to open '%s'", dir); } @@ -674,19 +676,29 @@ void read_testcases(afl_state_t *afl) { u8 dfn[PATH_MAX]; snprintf(dfn, PATH_MAX, "%s/.state/deterministic_done/%s", afl->in_dir, nl[i]->d_name); - u8 *fn2 = alloc_printf("%s/%s", afl->in_dir, nl[i]->d_name); + u8 *fn2 = alloc_printf("%s/%s", dir, nl[i]->d_name); u8 passed_det = 0; - free(nl[i]); /* not tracked */ - if (lstat(fn2, &st) || access(fn2, R_OK)) { PFATAL("Unable to access '%s'", fn2); } - /* This also takes care of . and .. */ + /* obviously we want to skip "descending" into . and .. directories, + however it is a good idea to skip also directories that start with + a dot */ + if (subdirs && S_ISDIR(st.st_mode) && nl[i]->d_name[0] != '.') { + + free(nl[i]); /* not tracked */ + read_testcases(afl, fn2); + ck_free(fn2); + continue; + + } + + free(nl[i]); if (!S_ISREG(st.st_mode) || !st.st_size || strstr(fn2, "/README.txt")) { @@ -718,7 +730,7 @@ void read_testcases(afl_state_t *afl) { free(nl); /* not tracked */ - if (!afl->queued_paths) { + if (!afl->queued_paths && directory == NULL) { SAYF("\n" cLRD "[-] " cRST "Looks like there are no valid test cases in the input directory! The " @@ -985,6 +997,76 @@ void perform_dry_run(afl_state_t *afl) { } + /* Now we remove all entries from the queue that have a duplicate trace map */ + + q = afl->queue; + struct queue_entry *p, *prev = NULL; + int duplicates = 0; + +restart_outer_cull_loop: + + while (q) { + + if (q->cal_failed || !q->exec_cksum) continue; + + restart_inner_cull_loop: + + p = q->next; + + while (p) { + + if (!p->cal_failed && p->exec_cksum == q->exec_cksum) { + + duplicates = 1; + --afl->pending_not_fuzzed; + + // We do not remove any of the memory allocated because for + // splicing the data might still be interesting. + // We only decouple them from the linked list. + // This will result in some leaks at exit, but who cares. + + // we keep the shorter file + if (p->len >= q->len) { + + q->next = p->next; + goto restart_inner_cull_loop; + + } else { + + if (prev) + prev->next = q = p; + else + afl->queue = q = p; + goto restart_outer_cull_loop; + + } + + } + + p = p->next; + + } + + prev = q; + q = q->next; + + } + + if (duplicates) { + + afl->max_depth = 0; + q = afl->queue; + while (q) { + + if (q->depth > afl->max_depth) afl->max_depth = q->depth; + q = q->next; + + } + + afl->q_prev100 = afl->queue = afl->queue_top = afl->queue; + + } + OKF("All test cases processed."); } diff --git a/src/afl-fuzz-one.c b/src/afl-fuzz-one.c index bf568c38..5737c1f5 100644 --- a/src/afl-fuzz-one.c +++ b/src/afl-fuzz-one.c @@ -1707,20 +1707,8 @@ custom_mutator_stage: } while (tid == afl->current_entry && afl->queued_paths > 1); - target = afl->queue; - - while (tid >= 100) { - - target = target->next_100; - tid -= 100; - - } - - while (tid--) { - - target = target->next; - - } + afl->splicing_with = tid; + target = afl->queue_buf[tid]; /* Make sure that the target has a reasonable length. */ @@ -4518,20 +4506,7 @@ pacemaker_fuzzing: } while (tid == afl->current_entry); afl->splicing_with = tid; - target = afl->queue; - - while (tid >= 100) { - - target = target->next_100; - tid -= 100; - - } - - while (tid--) { - - target = target->next; - - } + target = afl->queue_buf[tid]; /* Make sure that the target has a reasonable length. */ diff --git a/src/afl-fuzz-queue.c b/src/afl-fuzz-queue.c index c6d8225f..db91813b 100644 --- a/src/afl-fuzz-queue.c +++ b/src/afl-fuzz-queue.c @@ -239,13 +239,6 @@ void add_to_queue(afl_state_t *afl, u8 *fname, u32 len, u8 passed_det) { afl->cycles_wo_finds = 0; - if (!(afl->queued_paths % 100)) { - - afl->q_prev100->next_100 = q; - afl->q_prev100 = q; - - } - struct queue_entry **queue_buf = afl_realloc( AFL_BUF_PARAM(queue), afl->queued_paths * sizeof(struct queue_entry *)); if (unlikely(!queue_buf)) { PFATAL("alloc"); } diff --git a/src/afl-fuzz.c b/src/afl-fuzz.c index c12d5db5..bfaa22e8 100644 --- a/src/afl-fuzz.c +++ b/src/afl-fuzz.c @@ -119,8 +119,8 @@ static void usage(u8 *argv0, int more_help) { "etc.)\n" " -d - quick & dirty mode (skips deterministic steps)\n" " -n - fuzz without instrumentation (non-instrumented mode)\n" - " -x dict_file - optional fuzzer dictionary (see README.md, its really " - "good!)\n\n" + " -x dict_file - fuzzer dictionary (see README.md, specify up to 4 " + "times)\n\n" "Testing settings:\n" " -s seed - use a fixed seed for the RNG\n" @@ -243,11 +243,11 @@ static int stricmp(char const *a, char const *b) { int main(int argc, char **argv_orig, char **envp) { - s32 opt; + s32 opt, i; u64 prev_queued = 0; u32 sync_interval_cnt = 0, seek_to, show_help = 0, map_size = MAP_SIZE; - u8 * extras_dir = 0; - u8 mem_limit_given = 0, exit_1 = 0, debug = 0; + u8 * extras_dir[4]; + u8 mem_limit_given = 0, exit_1 = 0, debug = 0, extras_dir_cnt = 0; char **use_argv; struct timeval tv; @@ -450,8 +450,13 @@ int main(int argc, char **argv_orig, char **envp) { case 'x': /* dictionary */ - if (extras_dir) { FATAL("Multiple -x options not supported"); } - extras_dir = optarg; + if (extras_dir_cnt >= 4) { + + FATAL("More than four -x options are not supported"); + + } + + extras_dir[extras_dir_cnt++] = optarg; break; case 't': { /* timeout */ @@ -828,10 +833,6 @@ int main(int argc, char **argv_orig, char **envp) { "Eißfeldt, Andrea Fioraldi and Dominik Maier"); OKF("afl++ is open source, get it at " "https://github.com/AFLplusplus/AFLplusplus"); - OKF("Power schedules from github.com/mboehme/aflfast"); - OKF("Python Mutator and llvm_mode instrument file list from " - "github.com/choller/afl"); - OKF("MOpt Mutator from github.com/puppet-meteor/MOpt-AFL"); if (afl->sync_id && afl->is_main_node && afl->afl_env.afl_custom_mutator_only) { @@ -1139,7 +1140,15 @@ int main(int argc, char **argv_orig, char **envp) { pivot_inputs(afl); - if (extras_dir) { load_extras(afl, extras_dir); } + if (extras_dir_cnt) { + + for (i = 0; i < extras_dir_cnt; i++) + load_extras(afl, extras_dir[i]); + + dedup_extras(afl); + OKF("Loaded a total of %u extras.", afl->extras_cnt); + + } if (!afl->timeout_given) { find_timeout(afl); } -- cgit 1.4.1 From e30b2c6af6e369844c92c00a20ebdd53473a747c Mon Sep 17 00:00:00 2001 From: vanhauser-thc Date: Sat, 5 Sep 2020 13:18:28 +0200 Subject: final changes for pre-3.0 --- .gitignore | 6 ++ Android.bp | 6 +- GNUmakefile | 3 +- GNUmakefile.gcc_plugin | 1 + GNUmakefile.llvm | 8 ++- README.md | 2 +- docs/Changelog.md | 14 ++++ docs/FAQ.md | 104 +++++++++++++++--------------- docs/INSTALL.md | 19 +++--- docs/env_variables.md | 121 ++++++++++++++++++----------------- docs/ideas.md | 57 ----------------- docs/life_pro_tips.md | 4 +- docs/perf_tips.md | 8 +-- docs/sister_projects.md | 4 +- docs/status_screen.md | 2 +- examples/README.md | 2 +- examples/aflpp_driver/aflpp_driver.c | 2 +- include/afl-fuzz.h | 3 +- include/config.h | 4 +- include/envs.h | 1 + qemu_mode/patches/afl-qemu-cpu-inl.h | 2 +- src/afl-fuzz-init.c | 3 +- src/afl-fuzz-queue.c | 13 ++-- src/afl-fuzz-stats.c | 8 +-- src/afl-fuzz.c | 3 +- test/test-gcc-plugin.sh | 2 +- test/test-unittests.sh | 2 + 27 files changed, 188 insertions(+), 216 deletions(-) diff --git a/.gitignore b/.gitignore index 0527a0b2..e3adb6ef 100644 --- a/.gitignore +++ b/.gitignore @@ -51,6 +51,12 @@ afl-showmap.8 afl-system-config.8 afl-tmin.8 afl-whatsup.8 +afl-c++ +afl-cc +afl-lto +afl-lto++ +afl-lto++.8 +afl-lto.8 qemu_mode/libcompcov/compcovtest qemu_mode/qemu-* unicorn_mode/samples/*/\.test-* diff --git a/Android.bp b/Android.bp index e59129db..2c2114b2 100644 --- a/Android.bp +++ b/Android.bp @@ -101,7 +101,7 @@ cc_binary_host { ], srcs: [ - "llvm_mode/afl-clang-fast.c", + "src/afl-cc.c", ], } @@ -119,7 +119,7 @@ cc_binary_host { ], srcs: [ - "llvm_mode/afl-clang-fast.c", + "src/afl-cc.c", ], } @@ -136,6 +136,6 @@ cc_library_static { ], srcs: [ - "llvm_mode/afl-llvm-rt.o.c", + "instrumentation/afl-llvm-rt.o.c", ], } diff --git a/GNUmakefile b/GNUmakefile index 0046a481..7455483c 100644 --- a/GNUmakefile +++ b/GNUmakefile @@ -533,7 +533,7 @@ endif deepclean: clean rm -rf qemu_mode/qemu-3.1.1.tar.xz rm -rf unicorn_mode/unicornafl - git reset --hard >/dev/null 2>&1 || true + # NEVER EVER ACTIVATE THAT!!!!! git reset --hard >/dev/null 2>&1 || true .PHONY: distrib distrib: all @@ -591,6 +591,7 @@ install: all $(MANPAGES) @install -d -m 755 $${DESTDIR}$(BIN_PATH) $${DESTDIR}$(HELPER_PATH) $${DESTDIR}$(DOC_PATH) $${DESTDIR}$(MISC_PATH) @rm -f $${DESTDIR}$(BIN_PATH)/afl-plot.sh @rm -f $${DESTDIR}$(BIN_PATH)/afl-as + @rm -f $${DESTDIR}$(HELPER_PATH)/afl-llvm-rt.o $${DESTDIR}$(HELPER_PATH)/afl-llvm-rt-32.o $${DESTDIR}$(HELPER_PATH)/afl-llvm-rt-64.o $${DESTDIR}$(HELPER_PATH)/afl-gcc-rt.o install -m 755 $(PROGS) $(SH_PROGS) $${DESTDIR}$(BIN_PATH) @if [ -f afl-qemu-trace ]; then install -m 755 afl-qemu-trace $${DESTDIR}$(BIN_PATH); fi @if [ -f libdislocator.so ]; then set -e; install -m 755 libdislocator.so $${DESTDIR}$(HELPER_PATH); fi diff --git a/GNUmakefile.gcc_plugin b/GNUmakefile.gcc_plugin index aeb1ef16..b73fcfda 100644 --- a/GNUmakefile.gcc_plugin +++ b/GNUmakefile.gcc_plugin @@ -158,6 +158,7 @@ vpath % .. install: all ln -sf afl-cc $${DESTDIR}$(BIN_PATH)/afl-gcc-fast ln -sf afl-c++ $${DESTDIR}$(BIN_PATH)/afl-g++-fast + ln -sf afl-compiler-rt.o $${DESTDIR}$(HELPER_PATH)/afl-gcc-rt.o install -m 755 ./afl-gcc-pass.so $${DESTDIR}$(HELPER_PATH) install -m 644 -T instrumentation/README.gcc_plugin.md $${DESTDIR}$(DOC_PATH)/README.gcc_plugin.md diff --git a/GNUmakefile.llvm b/GNUmakefile.llvm index d76e0b28..1bb3d265 100644 --- a/GNUmakefile.llvm +++ b/GNUmakefile.llvm @@ -423,10 +423,12 @@ document: ./afl-compiler-rt-32.o: instrumentation/afl-compiler-rt.o.c | test_deps @printf "[*] Building 32-bit variant of the runtime (-m32)... " @$(CLANG_BIN) $(CLANG_CFL) $(CFLAGS_SAFE) $(CPPFLAGS) -O3 -Wno-unused-result -m32 -fPIC -c $< -o $@ 2>/dev/null; if [ "$$?" = "0" ]; then echo "success!"; else echo "failed (that's fine)"; fi + @test -e afl-compiler-rt-32.o && ln -sf afl-compiler-rt-32.o afl-llvm-rt-64.o ./afl-compiler-rt-64.o: instrumentation/afl-compiler-rt.o.c | test_deps @printf "[*] Building 64-bit variant of the runtime (-m64)... " @$(CLANG_BIN) $(CLANG_CFL) $(CFLAGS_SAFE) $(CPPFLAGS) -O3 -Wno-unused-result -m64 -fPIC -c $< -o $@ 2>/dev/null; if [ "$$?" = "0" ]; then echo "success!"; else echo "failed (that's fine)"; fi + @test -e afl-compiler-rt-64.o && ln -sf afl-compiler-rt-64.o afl-llvm-rt-64.o .PHONY: test_build test_build: $(PROGS) @@ -448,11 +450,11 @@ all_done: test_build install: all @install -d -m 755 $${DESTDIR}$(BIN_PATH) $${DESTDIR}$(HELPER_PATH) $${DESTDIR}$(DOC_PATH) $${DESTDIR}$(MISC_PATH) @if [ -f ./afl-cc ]; then set -e; install -m 755 ./afl-cc $${DESTDIR}$(BIN_PATH); ln -sf afl-cc $${DESTDIR}$(BIN_PATH)/afl-c++; fi - @if [ -f ./afl-compiler-rt.o ]; then set -e; install -m 755 ./afl-compiler-rt.o $${DESTDIR}$(HELPER_PATH); fi + @if [ -f ./afl-compiler-rt.o ]; then set -e; install -m 755 ./afl-compiler-rt.o $${DESTDIR}$(HELPER_PATH); ln -sf afl-compiler-rt.o $${DESTDIR}$(HELPER_PATH)afl-llvm-rt.o ;fi @if [ -f ./afl-lto ]; then set -e; ln -sf afl-cc $${DESTDIR}$(BIN_PATH)/afl-lto; ln -sf afl-cc $${DESTDIR}$(BIN_PATH)/afl-lto++; ln -sf afl-cc $${DESTDIR}$(BIN_PATH)/afl-clang-lto; ln -sf afl-cc $${DESTDIR}$(BIN_PATH)/afl-clang-lto++; install -m 755 ./afl-llvm-lto-instrumentation.so ./afl-llvm-rt-lto*.o ./afl-llvm-lto-instrumentlist.so $${DESTDIR}$(HELPER_PATH); fi @if [ -f ./afl-ld-lto ]; then set -e; install -m 755 ./afl-ld-lto $${DESTDIR}$(BIN_PATH); fi - @if [ -f ./afl-compiler-rt-32.o ]; then set -e; install -m 755 ./afl-compiler-rt-32.o $${DESTDIR}$(HELPER_PATH); fi - @if [ -f ./afl-compiler-rt-64.o ]; then set -e; install -m 755 ./afl-compiler-rt-64.o $${DESTDIR}$(HELPER_PATH); fi + @if [ -f ./afl-compiler-rt-32.o ]; then set -e; install -m 755 ./afl-compiler-rt-32.o $${DESTDIR}$(HELPER_PATH); ln -sf afl-compiler-rt-32.o $${DESTDIR}$(HELPER_PATH)afl-llvm-rt-32.o ;fi + @if [ -f ./afl-compiler-rt-64.o ]; then set -e; install -m 755 ./afl-compiler-rt-64.o $${DESTDIR}$(HELPER_PATH); ln -sf afl-compiler-rt-64.o $${DESTDIR}$(HELPER_PATH)afl-llvm-rt-64.o ; fi @if [ -f ./compare-transform-pass.so ]; then set -e; install -m 755 ./*.so $${DESTDIR}$(HELPER_PATH); fi @if [ -f ./compare-transform-pass.so ]; then set -e; ln -sf afl-cc $${DESTDIR}$(BIN_PATH)/afl-clang-fast ; ln -sf ./afl-c++ $${DESTDIR}$(BIN_PATH)/afl-clang-fast++ ; ln -sf afl-cc $${DESTDIR}$(BIN_PATH)/afl-clang ; ln -sf ./afl-c++ $${DESTDIR}$(BIN_PATH)/afl-clang++ ; fi @if [ -f ./SanitizerCoverageLTO.so ]; then set -e; ln -sf afl-cc $${DESTDIR}$(BIN_PATH)/afl-clang-lto ; ln -sf ./afl-c++ $${DESTDIR}$(BIN_PATH)/afl-clang-lto++ ; fi diff --git a/README.md b/README.md index 96b34260..c886489d 100644 --- a/README.md +++ b/README.md @@ -6,7 +6,7 @@ Release Version: [2.67c](https://github.com/AFLplusplus/AFLplusplus/releases) - Github Version: 2.67d + Github Version: 3.00a Repository: [https://github.com/AFLplusplus/AFLplusplus](https://github.com/AFLplusplus/AFLplusplus) diff --git a/docs/Changelog.md b/docs/Changelog.md index 6321aee4..9de03e78 100644 --- a/docs/Changelog.md +++ b/docs/Changelog.md @@ -9,6 +9,20 @@ Want to stay in the loop on major new features? Join our mailing list by sending a mail to . +### Version ++3.00a (develop) + - llvm_mode/ and gcc_plugin/ moved to instrumentation/ + - all compilers combined to afl-cc which emulates the previous ones + - afl-llvm/gcc-rt.o merged into afl-compiler-rt.o + - afl-fuzz + - reading testcases from -i now descends into subdirectories + - allow up to 4 -x command line options + - loaded extras now have a duplicate protection + - instrumentation + - new llvm pass: dict2file via AFL_LLVM_DICT2FILE, create afl-fuzz + -x dictionary of string comparisons found during compilation + - not overriding -Ox or -fno-unroll-loops anymore + + ### Version ++2.68c (release) - added the GSoC excellent afl++ grammar mutator by Shengtuo to our custom_mutators/ (see custom_mutators/README.md) - or get it here: diff --git a/docs/FAQ.md b/docs/FAQ.md index 064638f4..24942492 100644 --- a/docs/FAQ.md +++ b/docs/FAQ.md @@ -4,11 +4,11 @@ * [What is the difference between afl and afl++?](#what-is-the-difference-between-afl-and-afl) * [How to improve the fuzzing speed?](#how-to-improve-the-fuzzing-speed) - * [How do I fuzz a network service?](#how-do-i-fuzz-a-network-service) - * [How do I fuzz a GUI program?](#how-do-i-fuzz-a-gui-program) + * [How do I fuzz a network service?](#how-to-fuzz-a-network-service) + * [How do I fuzz a GUI program?](#how-to-fuzz-a-gui-program) * [What is an edge?](#what-is-an-edge) * [Why is my stability below 100%?](#why-is-my-stability-below-100) - * [How can I improve the stability value?](#how-can-i-improve-the-stability-value) + * [How can I improve the stability value](#how-can-i-improve-the-stability-value) If you find an interesting or important question missing, submit it via [https://github.com/AFLplusplus/AFLplusplus/issues](https://github.com/AFLplusplus/AFLplusplus/issues) @@ -18,52 +18,51 @@ If you find an interesting or important question missing, submit it via American Fuzzy Lop (AFL) was developed by Michał "lcamtuf" Zalewski starting in 2013/2014, and when he left Google end of 2017 he stopped developing it. -At the end of 2019 the Google fuzzing team took over maintenance of AFL, however -it is only accepting PRs from the community and is not developing enhancements +At the end of 2019 the Google fuzzing team took over maintance of AFL, however +it is only accepting PR from the community and is not developing enhancements anymore. -In the second quarter of 2019, 1 1/2 year later when no further development of -AFL had happened and it became clear there would none be coming, afl++ -was born, where initially community patches were collected and applied -for bug fixes and enhancements. Then from various AFL spin-offs - mostly academic +In the second quarter of 2019, 1 1/2 years after no further development of +AFL had happened and it became clear there would be none coming, afl++ +was born, where initially first community patches were collected and applied +for bugs and enhancements. Then from various AFL spin-offs - mostly academic research - features were integrated. This already resulted in a much advanced AFL. Until the end of 2019 the afl++ team had grown to four active developers which -then implemented their own research and features, making it now by far the most +then implemented their own research and feature, making it now by far the most flexible and feature rich guided fuzzer available as open source. And in independent fuzzing benchmarks it is one of the best fuzzers available, e.g. [Fuzzbench Report](https://www.fuzzbench.com/reports/2020-08-03/index.html) -## How to improve the fuzzing speed? +## How to improve the fuzzing speed - 1. Use [llvm_mode](docs/llvm_mode/README.md): afl-clang-lto (llvm >= 11) or afl-clang-fast (llvm >= 9 recommended) - 2. Use [persistent mode](llvm_mode/README.persistent_mode.md) (x2-x20 speed increase) + 1. use [instrumentation](docs/README.llvm.md): afl-clang-lto (llvm >= 11) or afl-clang-fast (llvm >= 9 recommended) + 2. Use [persistent mode](instrumentation/README.persistent_mode.md) (x2-x20 speed increase) 3. Use the [afl++ snapshot module](https://github.com/AFLplusplus/AFL-Snapshot-LKM) (x2 speed increase) - 4. If you do not use shmem persistent mode, use `AFL_TMPDIR` to put the input file directory on a tempfs location, see [docs/env_variables.md](docs/env_variables.md) - 5. Improve Linux kernel performance: modify `/etc/default/grub`, set `GRUB_CMDLINE_LINUX_DEFAULT="ibpb=off ibrs=off kpti=off l1tf=off mds=off mitigations=off no_stf_barrier noibpb noibrs nopcid nopti nospec_store_bypass_disable nospectre_v1 nospectre_v2 pcid=off pti=off spec_store_bypass_disable=off spectre_v2=off stf_barrier=off"`; then `update-grub` and `reboot` (warning: makes the system less secure) + 4. If you do not use shmem persistent mode, use `AFL_TMPDIR` to point the input file on a tempfs location, see [docs/env_variables.md](docs/env_variables.md) + 5. Improve kernel performance: modify `/etc/default/grub`, set `GRUB_CMDLINE_LINUX_DEFAULT="ibpb=off ibrs=off kpti=off l1tf=off mds=off mitigations=off no_stf_barrier noibpb noibrs nopcid nopti nospec_store_bypass_disable nospectre_v1 nospectre_v2 pcid=off pti=off spec_store_bypass_disable=off spectre_v2=off stf_barrier=off"`; then `update-grub` and `reboot` (warning: makes the system more insecure) 6. Running on an `ext2` filesystem with `noatime` mount option will be a bit faster than on any other journaling filesystem 7. Use your cores! [README.md:3.b) Using multiple cores/threads](../README.md#b-using-multiple-coresthreads) ## How do I fuzz a network service? -The short answer is - you cannot, at least not "out of the box". +The short answer is - you cannot, at least "out of the box". -Using a network channel is inadequate for several reasons: -- it has a slow-down of x10-20 on the fuzzing speed -- it does not scale to fuzzing multiple instances easily, -- instead of one initial data packet often a back-and-forth interplay of packets is needed for stateful protocols (which is totally unsupported by most coverage aware fuzzers). +Using network has a slow-down of x10-20 on the fuzzing speed, does not scale, +and finally usually it is more than one initial data packet but a back-and-forth +which is totally unsupported by most coverage aware fuzzers. The established method to fuzz network services is to modify the source code to read from a file or stdin (fd 0) (or even faster via shared memory, combine -this with persistent mode [llvm_mode/README.persistent_mode.md](llvm_mode/README.persistent_mode.md) +this with persistent mode [instrumentation/README.persistent_mode.md](instrumentation/README.persistent_mode.md) and you have a performance gain of x10 instead of a performance loss of over -x10 - that is a x100 difference!). +x10 - that is a x100 difference! If modifying the source is not an option (e.g. because you only have a binary and perform binary fuzzing) you can also use a shared library with AFL_PRELOAD -to emulate the network. This is also much faster than the real network would be. -See [examples/socket_fuzzing/](../examples/socket_fuzzing/). +to emulate the network. This is also much faster than network would be. +See [examples/socket_fuzzing/](../examples/socket_fuzzing/) There is an outdated afl++ branch that implements networking if you are desperate though: [https://github.com/AFLplusplus/AFLplusplus/tree/networking](https://github.com/AFLplusplus/AFLplusplus/tree/networking) - @@ -74,7 +73,7 @@ which allows you to define network state with different type of data packets. If the GUI program can read the fuzz data from a file (via the command line, a fixed location or via an environment variable) without needing any user -interaction then it would be suitable for fuzzing. +interaction then then yes. Otherwise it is not possible without modifying the source code - which is a very good idea anyway as the GUI functionality is a huge CPU/time overhead @@ -83,13 +82,13 @@ for the fuzzing. So create a new `main()` that just reads the test case and calls the functionality for processing the input that the GUI program is using. -## What is an "edge"? +## What is an "edge" A program contains `functions`, `functions` contain the compiled machine code. The compiled machine code in a `function` can be in a single or many `basic blocks`. A `basic block` is the largest possible number of subsequent machine code -instructions that has exactly one entrypoint (which can be be entered by multiple other basic blocks) -and runs linearly without branching or jumping to other addresses (except at the end). +instructions that runs independent, meaning it does not split up to different +locations nor is it jumped into it from a different location: ``` function() { A: @@ -99,7 +98,7 @@ function() { if (x) goto C; else goto D; C: some code - goto E + goto D D: some code goto B @@ -109,7 +108,7 @@ function() { ``` Every code block between two jump locations is a `basic block`. -An `edge` is then the unique relationship between two directly connected `basic blocks` (from the +An `edge` is then the unique relationship between two `basic blocks` (from the code example above): ``` Block A @@ -124,9 +123,8 @@ code example above): Block E ``` Every line between two blocks is an `edge`. -Note that a few basic block loop to itself, this too would be an edge. -## Why is my stability below 100%? +## Why is my stability below 100% Stability is measured by how many percent of the edges in the target are "stable". Sending the same input again and again should take the exact same @@ -134,37 +132,37 @@ path through the target every time. If that is the case, the stability is 100%. If however randomness happens, e.g. a thread reading other external data, reaction to timing, etc. then in some of the re-executions with the same data -the edge coverage result will be different accross runs. +the result in the edge information will be different accross runs. Those edges that change are then flagged "unstable". The more "unstable" edges, the more difficult for afl++ to identify valid new paths. A value above 90% is usually fine and a value above 80% is also still ok, and -even a value above 20% can still result in successful finds of bugs. -However, it is recommended that for values below 90% or 80% you should take -countermeasures to improve stability. +even above 20% can still result in successful finds of bugs. +However, it is recommended that below 90% or 80% you should take measures to +improve the stability. -## How can I improve the stability value? +## How can I improve the stability value -For fuzzing a 100% stable target that covers all edges is the best case. +For fuzzing a 100% stable target that covers all edges is the best. A 90% stable target that covers all edges is however better than a 100% stable target that ignores 10% of the edges. With instability you basically have a partial coverage loss on an edge, with -ignored functions you have a full loss on that edges. +ignore you have a full loss on that edge. There are functions that are unstable, but also provide value to coverage, eg init functions that use fuzz data as input for example. -If however a function that has nothing to do with the input data is the -source of instability, e.g. checking jitter, or is a hash map function etc. -then it should not be instrumented. +If however it is a function that has nothing to do with the input data is the +source, e.g. checking jitter, or is a hash map function etc. then it should +not be instrumented. -To be able to exclude these functions (based on AFL++'s measured stability) -the following process will allow to identify functions with variable edges. +To be able to make this decision the following process will allow you to +identify the functions with variable edges so you can make this decision. -Four steps are required to do this and it also requires quite some knowledge -of coding and/or disassembly and is effectively possible only with +Four steps are required to do this and requires quite some knowledge of +coding and/or disassembly and it is only effectively possible with afl-clang-fast PCGUARD and afl-clang-lto LTO instrumentation. 1. First step: Identify which edge ID numbers are unstable @@ -173,7 +171,7 @@ afl-clang-fast PCGUARD and afl-clang-lto LTO instrumentation. The out/fuzzer_stats file will then show the edge IDs that were identified as unstable. - 2. Second step: Find the responsible function(s). + 2. Second step: Find the responsible function. a) For LTO instrumented binaries this can be documented during compile time, just set `export AFL_LLVM_DOCUMENT_IDS=/path/to/a/file`. @@ -182,10 +180,10 @@ afl-clang-fast PCGUARD and afl-clang-lto LTO instrumentation. b) For PCGUARD instrumented binaries it is much more difficult. Here you can either modify the __sanitizer_cov_trace_pc_guard function in - llvm_mode/afl-llvm-rt.o.c to write a backtrace to a file if the ID in + instrumentation/afl-llvm-rt.o.c to write a backtrace to a file if the ID in __afl_area_ptr[*guard] is one of the unstable edge IDs. (Example code is already there). - Then recompile and reinstall llvm_mode and rebuild your target. + Then recompile and reinstall instrumentation and rebuild your target. Run the recompiled target with afl-fuzz for a while and then check the file that you wrote with the backtrace information. Alternatively you can use `gdb` to hook __sanitizer_cov_trace_pc_guard_init @@ -193,20 +191,20 @@ afl-clang-fast PCGUARD and afl-clang-lto LTO instrumentation. and set a write breakpoint to that address (`watch 0x.....`). c) in all other instrumentation types this is not possible. So just - recompile with the two mentioned above. This is just for + recompile with the the two mentioned above. This is just for identifying the functions that have unstable edges. 3. Third step: create a text file with the filenames/functions Identify which source code files contain the functions that you need to remove from instrumentation, or just specify the functions you want to - skip for instrumentation. Note that optimization might inline functions! + skip instrumenting. Note that optimization might inline functions! - Simply follow this document on how to do this: [llvm_mode/README.instrument_list.md](llvm_mode/README.instrument_list.md) + Simply follow this document on how to do this: [instrumentation/README.instrument_list.md](instrumentation/README.instrument_list.md) If PCGUARD is used, then you need to follow this guide (needs llvm 12+!): [http://clang.llvm.org/docs/SanitizerCoverage.html#partially-disabling-instrumentation](http://clang.llvm.org/docs/SanitizerCoverage.html#partially-disabling-instrumentation) - Only exclude those functions from instrumentation that provide no value + Only deny those functions from instrumentation that provide no value for coverage - that is if it does not process any fuzz data directly or indirectly (e.g. hash maps, thread management etc.). If however a function directly or indirectly handles fuzz data then you diff --git a/docs/INSTALL.md b/docs/INSTALL.md index 766f24d7..fb7b5642 100644 --- a/docs/INSTALL.md +++ b/docs/INSTALL.md @@ -24,7 +24,7 @@ There are no special dependencies to speak of; you will need GNU make and a working compiler (gcc or clang). Some of the optional scripts bundled with the program may depend on bash, gdb, and similar basic tools. -If you are using clang, please review llvm_mode/README.md; the LLVM +If you are using clang, please review README.llvm.md; the LLVM integration mode can offer substantial performance gains compared to the traditional approach. @@ -52,10 +52,10 @@ sudo gmake install Keep in mind that if you are using csh as your shell, the syntax of some of the shell commands given in the README.md and other docs will be different. -The `llvm_mode` requires a dynamically linked, fully-operational installation of +The `llvm` requires a dynamically linked, fully-operational installation of clang. At least on FreeBSD, the clang binaries are static and do not include some of the essential tools, so if you want to make it work, you may need to -follow the instructions in llvm_mode/README.md. +follow the instructions in README.llvm.md. Beyond that, everything should work as advertised. @@ -97,27 +97,24 @@ and definitely don't look POSIX-compliant. This means two things: User emulation mode of QEMU does not appear to be supported on MacOS X, so black-box instrumentation mode (`-Q`) will not work. -The llvm_mode requires a fully-operational installation of clang. The one that +The llvm instrumentation requires a fully-operational installation of clang. The one that comes with Xcode is missing some of the essential headers and helper tools. -See llvm_mode/README.md for advice on how to build the compiler from scratch. +See README.llvm.md for advice on how to build the compiler from scratch. ## 4. Linux or *BSD on non-x86 systems Standard build will fail on non-x86 systems, but you should be able to leverage two other options: - - The LLVM mode (see llvm_mode/README.md), which does not rely on + - The LLVM mode (see README.llvm.md), which does not rely on x86-specific assembly shims. It's fast and robust, but requires a complete installation of clang. - The QEMU mode (see qemu_mode/README.md), which can be also used for fuzzing cross-platform binaries. It's slower and more fragile, but can be used even when you don't have the source for the tested app. -If you're not sure what you need, you need the LLVM mode. To get it, try: - -```bash -AFL_NO_X86=1 gmake && gmake -C llvm_mode -``` +If you're not sure what you need, you need the LLVM mode, which is built by +default. ...and compile your target program with afl-clang-fast or afl-clang-fast++ instead of the traditional afl-gcc or afl-clang wrappers. diff --git a/docs/env_variables.md b/docs/env_variables.md index c47d10e8..9d289f6d 100644 --- a/docs/env_variables.md +++ b/docs/env_variables.md @@ -5,13 +5,25 @@ users or for some types of custom fuzzing setups. See README.md for the general instruction manual. -## 1) Settings for afl-gcc, afl-clang, and afl-as - and gcc_plugin afl-gcc-fast +## 1) Settings for all compilers -Because they can't directly accept command-line options, the compile-time -tools make fairly broad use of environmental variables: +Starting with afl++ 3.0 there is only one compiler: afl-cc +To select the different instrumentation modes this can be done by + 1. passing --afl-MODE command line options to the compiler + 2. use a symlink to afl-cc: afl-gcc, afl-g++, afl-clang, afl-clang++, + afl-clang-fast, afl-clang-fast++, afl-clang-lto, afl-clang-lto++, + afl-gcc-fast, afl-g++-fast + 3. using the environment variable AFL_CC_COMPILER with MODE - - Most afl tools do not print any output if stdout/stderr are redirected. - If you want to save the output in a file then set the AFL_DEBUG +MODE can one of LTO (afl-clang-lto*), LLVM (afl-clang-fast*), GCC_PLUGIN +(afl-g*-fast) or GCC (afl-gcc/afl-g++). + +Because beside the --afl-MODE command no afl specific command-line options +are accepted, the compile-time tools make fairly broad use of environmental +variables: + + - Most afl tools do not print any ouput if stout/stderr are redirected. + If you want to have the output into a file then set the AFL_DEBUG environment variable. This is sadly necessary for various build processes which fail otherwise. @@ -24,6 +36,8 @@ tools make fairly broad use of environmental variables: will cause problems in programs built with -Werror, simply because -O3 enables more thorough code analysis and can spew out additional warnings. To disable optimizations, set AFL_DONT_OPTIMIZE. + However if -O... and/or -fno-unroll-loops are set, these are not + overriden. - Setting AFL_USE_ASAN automatically enables ASAN, provided that your compiler supports that. Note that fuzzing with ASAN is mildly challenging @@ -44,7 +58,7 @@ tools make fairly broad use of environmental variables: you instrument hand-written assembly when compiling clang code by plugging a normalizer into the chain. (There is no equivalent feature for GCC.) - - Setting AFL_INST_RATIO to a percentage between 0% and 100% controls the + - Setting AFL_INST_RATIO to a percentage between 0 and 100% controls the probability of instrumenting every branch. This is (very rarely) useful when dealing with exceptionally complex programs that saturate the output bitmap. Examples include v8, ffmpeg, and perl. @@ -55,19 +69,16 @@ tools make fairly broad use of environmental variables: Setting AFL_INST_RATIO to 0 is a valid choice. This will instrument only the transitions between function entry points, but not individual branches. + Note that this is an outdated variable. A few instances (e.g. afl-gcc) + still support these, but state-of-the-art (e.g. LLVM LTO and LLVM PCGUARD) + do not need this. + - AFL_NO_BUILTIN causes the compiler to generate code suitable for use with libtokencap.so (but perhaps running a bit slower than without the flag). - TMPDIR is used by afl-as for temporary files; if this variable is not set, the tool defaults to /tmp. - - Setting AFL_KEEP_ASSEMBLY prevents afl-as from deleting instrumented - assembly files. Useful for troubleshooting problems or understanding how - the tool works. To get them in a predictable place, try something like: - - mkdir assembly_here - TMPDIR=$PWD/assembly_here AFL_KEEP_ASSEMBLY=1 make clean all - - If you are a weird person that wants to compile and instrument asm text files then use the AFL_AS_FORCE_INSTRUMENT variable: AFL_AS_FORCE_INSTRUMENT=1 afl-gcc foo.s -o foo @@ -78,19 +89,24 @@ tools make fairly broad use of environmental variables: - Setting AFL_CAL_FAST will speed up the initial calibration, if the application is very slow -## 2) Settings for afl-clang-fast / afl-clang-fast++ / afl-gcc-fast / afl-g++-fast +## 2) Settings for LLVM and LTO: afl-clang-fast / afl-clang-fast++ / afl-clang-lto / afl-clang-lto++ -The native instrumentation helpers (llvm_mode and gcc_plugin) accept a subset +The native instrumentation helpers (instrumentation and gcc_plugin) accept a subset of the settings discussed in section #1, with the exception of: + - LLVM modes support `AFL_LLVM_DICT2FILE=/absolute/path/file.txt` which will + write all constant string comparisons to this file to be used with + afl-fuzz' `-x` option. + - AFL_AS, since this toolchain does not directly invoke GNU as. - TMPDIR and AFL_KEEP_ASSEMBLY, since no temporary assembly files are created. - - AFL_INST_RATIO, as we by default use collision free instrumentation. + - AFL_INST_RATIO, as we by default collision free instrumentation is used. + Not all passes support this option though as it is an outdated feature. -Then there are a few specific features that are only available in llvm_mode: +Then there are a few specific features that are only available in instrumentation: ### Select the instrumentation mode @@ -121,7 +137,7 @@ Then there are a few specific features that are only available in llvm_mode: None of the following options are necessary to be used and are rather for manual use (which only ever the author of this LTO implementation will use). - These are used if several seperated instrumentations are performed which + These are used if several seperated instrumentation are performed which are then later combined. - AFL_LLVM_DOCUMENT_IDS=file will document to a file which edge ID was given @@ -136,7 +152,7 @@ Then there are a few specific features that are only available in llvm_mode: - AFL_LLVM_LTO_DONTWRITEID prevents that the highest location ID written into the instrumentation is set in a global variable - See llvm_mode/README.LTO.md for more information. + See instrumentation/README.LTO.md for more information. ### INSTRIM @@ -154,7 +170,7 @@ Then there are a few specific features that are only available in llvm_mode: afl-fuzz will only be able to see the path the loop took, but not how many times it was called (unless it is a complex loop). - See llvm_mode/README.instrim.md + See instrumentation/README.instrim.md ### NGRAM @@ -165,7 +181,7 @@ Then there are a few specific features that are only available in llvm_mode: config.h to at least 18 and maybe up to 20 for this as otherwise too many map collisions occur. - See llvm_mode/README.ctx.md + See instrumentation/README.ctx.md ### CTX @@ -176,7 +192,7 @@ Then there are a few specific features that are only available in llvm_mode: config.h to at least 18 and maybe up to 20 for this as otherwise too many map collisions occur. - See llvm_mode/README.ngram.md + See instrumentation/README.ngram.md ### LAF-INTEL @@ -196,17 +212,17 @@ Then there are a few specific features that are only available in llvm_mode: - Setting AFL_LLVM_LAF_ALL sets all of the above - See llvm_mode/README.laf-intel.md for more information. + See instrumentation/README.laf-intel.md for more information. ### INSTRUMENT LIST (selectively instrument files and functions) - This feature allows selective instrumentation of the source + This feature allows selectively instrumentation of the source - Setting AFL_LLVM_ALLOWLIST or AFL_LLVM_DENYLIST with a filenames and/or function will only instrument (or skip) those files that match the names listed in the specified file. - See llvm_mode/README.instrument_list.md for more information. + See instrumentation/README.instrument_list.md for more information. ### NOT_ZERO @@ -220,27 +236,34 @@ Then there are a few specific features that are only available in llvm_mode: test. If the target performs only few loops then this will give a small performance boost. - See llvm_mode/README.neverzero.md + See instrumentation/README.neverzero.md ### CMPLOG - Setting AFL_LLVM_CMPLOG=1 during compilation will tell afl-clang-fast to - produce a CmpLog binary. See llvm_mode/README.cmplog.md + produce a CmpLog binary. See instrumentation/README.cmplog.md - See llvm_mode/README.neverzero.md + See instrumentation/README.neverzero.md -Then there are a few specific features that are only available in the gcc_plugin: +## 3) Settings for GCC / GCC_PLUGIN modes -### INSTRUMENT_FILE +Then there are a few specific features that are only available in GCC and +GCC_PLUGIN mode. - This feature allows selective instrumentation of the source + - Setting AFL_KEEP_ASSEMBLY prevents afl-as from deleting instrumented + assembly files. Useful for troubleshooting problems or understanding how + the tool works. (GCC mode only) + To get them in a predictable place, try something like: - - Setting AFL_GCC_INSTRUMENT_FILE with a filename will only instrument those - files that match the names listed in this file (one filename per line). + mkdir assembly_here + TMPDIR=$PWD/assembly_here AFL_KEEP_ASSEMBLY=1 make clean all + - Setting AFL_GCC_INSTRUMENT_FILE with a filename will only instrument those + files that match the names listed in this file (one filename per line). See gcc_plugin/README.instrument_list.md for more information. + (GCC_PLUGIN mode only) -## 3) Settings for afl-fuzz +## 4) Settings for afl-fuzz The main fuzzer binary accepts several options that disable a couple of sanity checks or alter some of the more exotic semantics of the tool: @@ -278,14 +301,6 @@ checks or alter some of the more exotic semantics of the tool: don't want AFL to spend too much time classifying that stuff and just rapidly put all timeouts in that bin. - - Setting AFL_FORKSRV_INIT_TMOUT allows yout to specify a different timeout - to wait for the forkserver to spin up. The default is the `-t` value times - `FORK_WAIT_MULT` from `config.h` (usually 10), so for a `-t 100`, the - default would wait `1000` milis. Setting a different time here is useful - if the target has a very slow startup time, for example when doing - full-system fuzzing or emulation, but you don't want the actual runs - to wait too long for timeouts. - - AFL_NO_ARITH causes AFL to skip most of the deterministic arithmetics. This can be useful to speed up the fuzzing of text-based file formats. @@ -377,22 +392,12 @@ checks or alter some of the more exotic semantics of the tool: Note that this setting inhibits some of the user-friendly diagnostics normally done when starting up the forkserver and causes a pretty significant performance drop. - - - Setting AFL_MAX_DET_EXTRAS changes the count of dictionary entries/extras - (default 200), after which the entries will be used probabilistically. - So, if the dict/extras file (`-x`) contains more tokens than this threshold, - not all of the tokens will be used in each fuzzing step, every time. - Instead, there is a chance that the entry will be skipped during fuzzing. - This makes sure that the fuzzer doesn't spend all its time only inserting - the extras, but will still do other mutations. However, it decreases the - likelihood for each token to be inserted, before the next queue entry is fuzzed. - Either way, all tokens will be used eventually, in a longer fuzzing campaign. - Outdated environment variables that are that not supported anymore: AFL_DEFER_FORKSRV AFL_PERSISTENT -## 4) Settings for afl-qemu-trace +## 5) Settings for afl-qemu-trace The QEMU wrapper used to instrument binary-only code supports several settings: @@ -446,7 +451,7 @@ The QEMU wrapper used to instrument binary-only code supports several settings: stack pointer in which QEMU can find the return address when `start addr` is hitted. -## 5) Settings for afl-cmin +## 6) Settings for afl-cmin The corpus minimization script offers very little customization: @@ -472,12 +477,12 @@ to match when minimizing crashes. This will make minimization less useful, but may prevent the tool from "jumping" from one crashing condition to another in very buggy software. You probably want to combine it with the -e flag. -## 7) Settings for afl-analyze +## 8) Settings for afl-analyze You can set AFL_ANALYZE_HEX to get file offsets printed as hexadecimal instead of decimal. -## 8) Settings for libdislocator +## 9) Settings for libdislocator The library honors these environmental variables: @@ -499,12 +504,12 @@ The library honors these environmental variables: - AFL_ALIGNED_ALLOC=1 will force the alignment of the allocation size to max_align_t to be compliant with the C standard. -## 9) Settings for libtokencap +## 10) Settings for libtokencap This library accepts AFL_TOKEN_FILE to indicate the location to which the discovered tokens should be written. -## 10) Third-party variables set by afl-fuzz & other tools +## 11) Third-party variables set by afl-fuzz & other tools Several variables are not directly interpreted by afl-fuzz, but are set to optimal values if not already present in the environment: diff --git a/docs/ideas.md b/docs/ideas.md index 65e2e8e6..a5d40963 100644 --- a/docs/ideas.md +++ b/docs/ideas.md @@ -3,49 +3,6 @@ In the following, we describe a variety of ideas that could be implemented for future AFL++ versions. -For GSOC2020 interested students please see -[https://github.com/AFLplusplus/AFLplusplus/issues/208](https://github.com/AFLplusplus/AFLplusplus/issues/208) - -## Flexible Grammar Mutator (currently in development) - -Currently, AFL++'s mutation does not have deeper knowledge about the fuzzed -binary, apart from feedback, even though the developer may have insights -about the target. - -A developer may choose to provide dictionaries and implement own mutations -in python or C, but an easy mutator that behaves according to a given grammar, -does not exist. - -State-of-the-art research on grammar fuzzing has some problems in their -implementations like code quality, scalability, or ease of use and other -common issues of the academic code. - -We aim to develop a pluggable grammar mutator for afl++ that combines -various results. - -Mentor: andreafioraldi - -## perf-fuzz Linux Kernel Module - -Expand on [snapshot LKM](https://github.com/AFLplusplus/AFL-Snapshot-LKM) -To make it thread safe, can snapshot several processes at once and increase -overall performance. - -Mentor: any - -## QEMU 5-based Instrumentation - -First tests to use QEMU 4 for binary-only AFL++ showed that caching behavior -changed, which vastly decreases fuzzing speeds. - -In this task test if QEMU 5 performs better and port the afl++ QEMU 3.1 -patches to QEMU 5. - -Understanding the current instrumentation and fixing the current caching -issues will be needed. - -Mentor: andreafioraldi - ## WASM Instrumentation Currently, AFL++ can be used for source code fuzzing and traditional binaries. @@ -66,20 +23,6 @@ Either improve a single mutator thorugh learning of many different bugs Mentor: domenukk -## Reengineer `afl-fuzz` as Thread Safe, Embeddable Library (currently in development) - -Right now, afl-fuzz is single threaded, cannot safely be embedded in tools, -and not multi-threaded. It makes use of a large number of globals, must always -be the parent process and exec child processes. -Instead, afl-fuzz could be refactored to contain no global state and globals. -This allows for different use cases that could be implemented during this -project. -Note that in the mean time a lot has happened here already, but e.g. making -it all work and implement multithreading in afl-fuzz ... there is still quite -some work to do. - -Mentor: hexcoder- or vanhauser-thc - ## Collision-free Binary-Only Maps AFL++ supports collison-free maps using an LTO (link-time-optimization) pass. diff --git a/docs/life_pro_tips.md b/docs/life_pro_tips.md index a5bd7286..0004c297 100644 --- a/docs/life_pro_tips.md +++ b/docs/life_pro_tips.md @@ -30,10 +30,10 @@ Check out the `fuzzer_stats` file in the AFL output dir or try `afl-whatsup`. It could be important - consult docs/status_screen.md right away! ## Know your target? Convert it to persistent mode for a huge performance gain! -Consult section #5 in llvm_mode/README.md for tips. +Consult section #5 in README.llvm.md for tips. ## Using clang? -Check out llvm_mode/ for a faster alternative to afl-gcc! +Check out instrumentation/ for a faster alternative to afl-gcc! ## Did you know that AFL can fuzz closed-source or cross-platform binaries? Check out qemu_mode/README.md and unicorn_mode/README.md for more. diff --git a/docs/perf_tips.md b/docs/perf_tips.md index 731dc238..fbcb4d8d 100644 --- a/docs/perf_tips.md +++ b/docs/perf_tips.md @@ -51,7 +51,7 @@ a file. ## 3. Use LLVM instrumentation When fuzzing slow targets, you can gain 20-100% performance improvement by -using the LLVM-based instrumentation mode described in [the llvm_mode README](../llvm_mode/README.md). +using the LLVM-based instrumentation mode described in [the instrumentation README](../instrumentation/README.llvm.md). Note that this mode requires the use of clang and will not work with GCC. The LLVM mode also offers a "persistent", in-process fuzzing mode that can @@ -62,12 +62,12 @@ modes require you to edit the source code of the fuzzed program, but the changes often amount to just strategically placing a single line or two. If there are important data comparisons performed (e.g. `strcmp(ptr, MAGIC_HDR)`) -then using laf-intel (see llvm_mode/README.laf-intel.md) will help `afl-fuzz` a lot +then using laf-intel (see instrumentation/README.laf-intel.md) will help `afl-fuzz` a lot to get to the important parts in the code. If you are only interested in specific parts of the code being fuzzed, you can instrument_files the files that are actually relevant. This improves the speed and -accuracy of afl. See llvm_mode/README.instrument_list.md +accuracy of afl. See instrumentation/README.instrument_list.md Also use the InsTrim mode on larger binaries, this improves performance and coverage a lot. @@ -110,7 +110,7 @@ e.g.: https://launchpad.net/libeatmydata In programs that are slow due to unavoidable initialization overhead, you may -want to try the LLVM deferred forkserver mode (see llvm_mode/README.md), +want to try the LLVM deferred forkserver mode (see README.llvm.md), which can give you speed gains up to 10x, as mentioned above. Last but not least, if you are using ASAN and the performance is unacceptable, diff --git a/docs/sister_projects.md b/docs/sister_projects.md index a501ecbd..640e59f7 100644 --- a/docs/sister_projects.md +++ b/docs/sister_projects.md @@ -52,7 +52,7 @@ options. Provides an evolutionary instrumentation-guided fuzzing harness that allows some programs to be fuzzed without the fork / execve overhead. (Similar functionality is now available as the "persistent" feature described in -[the llvm_mode readme](../llvm_mode/README.md)) +[the llvm_mode readme](../instrumentation/README.llvm.md)) http://llvm.org/docs/LibFuzzer.html @@ -245,7 +245,7 @@ https://code.google.com/p/address-sanitizer/wiki/AsanCoverage#Coverage_counters ### AFL JS (Han Choongwoo) One-off optimizations to speed up the fuzzing of JavaScriptCore (now likely -superseded by LLVM deferred forkserver init - see llvm_mode/README.md). +superseded by LLVM deferred forkserver init - see README.llvm.md). https://github.com/tunz/afl-fuzz-js diff --git a/docs/status_screen.md b/docs/status_screen.md index b89468ce..2eeb8f3f 100644 --- a/docs/status_screen.md +++ b/docs/status_screen.md @@ -324,7 +324,7 @@ there are several things to look at: - Multiple threads executing at once in semi-random order. This is harmless when the 'stability' metric stays over 90% or so, but can become an issue if not. Here's what to try: - * Use afl-clang-fast from [llvm_mode](../llvm_mode/) - it uses a thread-local tracking + * Use afl-clang-fast from [instrumentation](../instrumentation/) - it uses a thread-local tracking model that is less prone to concurrency issues, * See if the target can be compiled or run without threads. Common `./configure` options include `--without-threads`, `--disable-pthreads`, or diff --git a/examples/README.md b/examples/README.md index d28aadbe..46a92c6e 100644 --- a/examples/README.md +++ b/examples/README.md @@ -47,7 +47,7 @@ Here's a quick overview of the stuff you can find in this directory: Note that the minimize_corpus.sh tool has graduated from the examples/ directory and is now available as ../afl-cmin. The LLVM mode has likewise -graduated to ../llvm_mode/*. +graduated to ../instrumentation/*. Most of the tools in this directory are meant chiefly as examples that need to be tweaked for your specific needs. They come with some basic documentation, diff --git a/examples/aflpp_driver/aflpp_driver.c b/examples/aflpp_driver/aflpp_driver.c index ff5446e9..82e55fc4 100644 --- a/examples/aflpp_driver/aflpp_driver.c +++ b/examples/aflpp_driver/aflpp_driver.c @@ -27,7 +27,7 @@ EOF # Build your target with -fsanitize-coverage=trace-pc-guard using fresh clang. clang -g -fsanitize-coverage=trace-pc-guard test_fuzzer.cc -c # Build afl-llvm-rt.o.c from the AFL distribution. -clang -c -w $AFL_HOME/llvm_mode/afl-llvm-rt.o.c +clang -c -w $AFL_HOME/instrumentation/afl-llvm-rt.o.c # Build this file, link it with afl-llvm-rt.o.o and the target code. clang++ afl_driver.cpp test_fuzzer.o afl-llvm-rt.o.o # Run AFL: diff --git a/include/afl-fuzz.h b/include/afl-fuzz.h index 4281c554..9e469864 100644 --- a/include/afl-fuzz.h +++ b/include/afl-fuzz.h @@ -935,6 +935,7 @@ u8 has_new_bits(afl_state_t *, u8 *); void load_extras_file(afl_state_t *, u8 *, u32 *, u32 *, u32); void load_extras(afl_state_t *, u8 *); +void dedup_extras(afl_state_t *); void add_extra(afl_state_t *afl, u8 *mem, u32 len); void maybe_add_auto(afl_state_t *, u8 *, u32); void save_auto(afl_state_t *); @@ -972,7 +973,7 @@ u8 fuzz_one(afl_state_t *); void bind_to_free_cpu(afl_state_t *); #endif void setup_post(afl_state_t *); -void read_testcases(afl_state_t *); +void read_testcases(afl_state_t *, u8 *); void perform_dry_run(afl_state_t *); void pivot_inputs(afl_state_t *); u32 find_start_position(afl_state_t *); diff --git a/include/config.h b/include/config.h index 77407d50..8cc70075 100644 --- a/include/config.h +++ b/include/config.h @@ -28,7 +28,7 @@ /* Version string: */ // c = release, d = volatile github dev, e = experimental branch -#define VERSION "++2.68c" +#define VERSION "++3.00a" /****************************************************** * * @@ -195,7 +195,7 @@ steps; past this point, the "extras/user" step will be still carried out, but with proportionally lower odds: */ -#define MAX_DET_EXTRAS 200 +#define MAX_DET_EXTRAS 256 /* Maximum number of auto-extracted dictionary tokens to actually use in fuzzing (first value), and to keep in memory as candidates. The latter should be much diff --git a/include/envs.h b/include/envs.h index 2dc1dbbf..d9968fcd 100644 --- a/include/envs.h +++ b/include/envs.h @@ -69,6 +69,7 @@ static char *afl_environment_variables[] = { "AFL_LLVM_CMPLOG", "AFL_LLVM_INSTRIM", "AFL_LLVM_CTX", + "AFL_LLVM_DICT2FILE", "AFL_LLVM_DOCUMENT_IDS", "AFL_LLVM_INSTRUMENT", "AFL_LLVM_INSTRIM_LOOPHEAD", diff --git a/qemu_mode/patches/afl-qemu-cpu-inl.h b/qemu_mode/patches/afl-qemu-cpu-inl.h index 63b7581d..0e38f38b 100644 --- a/qemu_mode/patches/afl-qemu-cpu-inl.h +++ b/qemu_mode/patches/afl-qemu-cpu-inl.h @@ -466,7 +466,7 @@ void afl_forkserver(CPUState *cpu) { } /* A simplified persistent mode handler, used as explained in - * llvm_mode/README.md. */ + * instrumentation/README.llvm.md */ void afl_persistent_loop(void) { diff --git a/src/afl-fuzz-init.c b/src/afl-fuzz-init.c index 713849a1..1351d274 100644 --- a/src/afl-fuzz-init.c +++ b/src/afl-fuzz-init.c @@ -626,6 +626,7 @@ if (dir == NULL) { if (!access(fn1, F_OK)) { afl->in_dir = fn1; + subdirs = 0; } else { @@ -1063,7 +1064,7 @@ restart_outer_cull_loop: } - afl->q_prev100 = afl->queue = afl->queue_top = afl->queue; + afl->queue = afl->queue_top = afl->queue; } diff --git a/src/afl-fuzz-queue.c b/src/afl-fuzz-queue.c index db91813b..af52aa45 100644 --- a/src/afl-fuzz-queue.c +++ b/src/afl-fuzz-queue.c @@ -138,8 +138,7 @@ static u8 check_if_text(struct queue_entry *q) { } // non-overlong 2-byte - if (len - offset > 1 && - ((0xC2 <= buf[offset + 0] && buf[offset + 0] <= 0xDF) && + if (len - offset > 1 && ((0xC2 <= buf[offset + 0] && buf[offset + 0] <= 0xDF) && (0x80 <= buf[offset + 1] && buf[offset + 1] <= 0xBF))) { offset += 2; @@ -230,7 +229,7 @@ void add_to_queue(afl_state_t *afl, u8 *fname, u32 len, u8 passed_det) { } else { - afl->q_prev100 = afl->queue = afl->queue_top = q; + afl->queue = afl->queue_top = q; } @@ -274,15 +273,15 @@ void add_to_queue(afl_state_t *afl, u8 *fname, u32 len, u8 passed_det) { void destroy_queue(afl_state_t *afl) { - struct queue_entry *q = afl->queue, *n; + struct queue_entry *q; + u32 i; - while (q) { + for (i = 0; i < afl->queued_paths; i++) { - n = q->next; + q = afl->queue_buf[i]; ck_free(q->fname); ck_free(q->trace_mini); ck_free(q); - q = n; } diff --git a/src/afl-fuzz-stats.c b/src/afl-fuzz-stats.c index 51eed14b..c60c65aa 100644 --- a/src/afl-fuzz-stats.c +++ b/src/afl-fuzz-stats.c @@ -982,10 +982,9 @@ void show_stats(afl_state_t *afl) { void show_init_stats(afl_state_t *afl) { struct queue_entry *q = afl->queue; - u32 min_bits = 0, max_bits = 0; + u32 min_bits = 0, max_bits = 0, max_len = 0, count = 0; u64 min_us = 0, max_us = 0; u64 avg_us = 0; - u32 max_len = 0; u8 val_bufs[4][STRINGIFY_VAL_SIZE_MAX]; #define IB(i) val_bufs[(i)], sizeof(val_bufs[(i)]) @@ -1006,6 +1005,7 @@ void show_init_stats(afl_state_t *afl) { if (q->len > max_len) { max_len = q->len; } + ++count; q = q->next; } @@ -1072,10 +1072,10 @@ void show_init_stats(afl_state_t *afl) { OKF("Here are some useful stats:\n\n" cGRA " Test case count : " cRST - "%u favored, %u variable, %u total\n" cGRA " Bitmap range : " cRST + "%u favored, %u variable, %u ignored, %u total\n" cGRA " Bitmap range : " cRST "%u to %u bits (average: %0.02f bits)\n" cGRA " Exec timing : " cRST "%s to %s us (average: %s us)\n", - afl->queued_favored, afl->queued_variable, afl->queued_paths, min_bits, + afl->queued_favored, afl->queued_variable, afl->queued_paths - count, afl->queued_paths, min_bits, max_bits, ((double)afl->total_bitmap_size) / (afl->total_bitmap_entries ? afl->total_bitmap_entries : 1), diff --git a/src/afl-fuzz.c b/src/afl-fuzz.c index bfaa22e8..73ca6aaa 100644 --- a/src/afl-fuzz.c +++ b/src/afl-fuzz.c @@ -1133,8 +1133,9 @@ int main(int argc, char **argv_orig, char **envp) { setup_cmdline_file(afl, argv + optind); - read_testcases(afl); + read_testcases(afl, NULL); // read_foreign_testcases(afl, 1); for the moment dont do this + OKF("Loaded a total of %u seeds.", afl->queued_paths); load_auto(afl); diff --git a/test/test-gcc-plugin.sh b/test/test-gcc-plugin.sh index 2ed10a72..8b8cbd8e 100755 --- a/test/test-gcc-plugin.sh +++ b/test/test-gcc-plugin.sh @@ -3,7 +3,7 @@ . ./test-pre.sh $ECHO "$BLUE[*] Testing: gcc_plugin" -test -e ../afl-gcc-fast -a -e ../afl-gcc-rt.o && { +test -e ../afl-gcc-fast -a -e ../afl-compiler-rt.o && { SAVE_AFL_CC=${AFL_CC} export AFL_CC=`command -v gcc` ../afl-gcc-fast -o test-instr.plain.gccpi ../test-instr.c > /dev/null 2>&1 diff --git a/test/test-unittests.sh b/test/test-unittests.sh index f540b5f8..58c2eea9 100755 --- a/test/test-unittests.sh +++ b/test/test-unittests.sh @@ -7,3 +7,5 @@ unset AFL_CC make -C .. unit || CODE=1 INCOMPLETE=1 : . ./test-post.sh + +rm -rf unittests/unit_hash unittests/unit_rand -- cgit 1.4.1 From 2f90f2faba92c0ef5e081ff74b54fb07eb1faaa9 Mon Sep 17 00:00:00 2001 From: vanhauser-thc Date: Sat, 5 Sep 2020 13:19:19 +0200 Subject: code-format --- src/afl-fuzz-init.c | 23 +++++++++++++---------- src/afl-fuzz-queue.c | 3 ++- src/afl-fuzz-stats.c | 16 +++++++++------- 3 files changed, 24 insertions(+), 18 deletions(-) diff --git a/src/afl-fuzz-init.c b/src/afl-fuzz-init.c index 1351d274..c834e5db 100644 --- a/src/afl-fuzz-init.c +++ b/src/afl-fuzz-init.c @@ -617,24 +617,27 @@ void read_testcases(afl_state_t *afl, u8 *directory) { s32 nl_cnt, subdirs = 1; u32 i; u8 * fn1, *dir = directory; - u8 val_buf[2][STRINGIFY_VAL_SIZE_MAX]; + u8 val_buf[2][STRINGIFY_VAL_SIZE_MAX]; /* Auto-detect non-in-place resumption attempts. */ -if (dir == NULL) { - fn1 = alloc_printf("%s/queue", afl->in_dir); - if (!access(fn1, F_OK)) { + if (dir == NULL) { - afl->in_dir = fn1; - subdirs = 0; + fn1 = alloc_printf("%s/queue", afl->in_dir); + if (!access(fn1, F_OK)) { - } else { + afl->in_dir = fn1; + subdirs = 0; + + } else { - ck_free(fn1); + ck_free(fn1); + + } + + dir = afl->in_dir; } - dir = afl->in_dir; -} ACTF("Scanning '%s'...", dir); diff --git a/src/afl-fuzz-queue.c b/src/afl-fuzz-queue.c index af52aa45..8c7bfc55 100644 --- a/src/afl-fuzz-queue.c +++ b/src/afl-fuzz-queue.c @@ -138,7 +138,8 @@ static u8 check_if_text(struct queue_entry *q) { } // non-overlong 2-byte - if (len - offset > 1 && ((0xC2 <= buf[offset + 0] && buf[offset + 0] <= 0xDF) && + if (len - offset > 1 && + ((0xC2 <= buf[offset + 0] && buf[offset + 0] <= 0xDF) && (0x80 <= buf[offset + 1] && buf[offset + 1] <= 0xBF))) { offset += 2; diff --git a/src/afl-fuzz-stats.c b/src/afl-fuzz-stats.c index c60c65aa..dfc0cd97 100644 --- a/src/afl-fuzz-stats.c +++ b/src/afl-fuzz-stats.c @@ -35,12 +35,12 @@ void write_setup_file(afl_state_t *afl, u32 argc, char **argv) { u8 fn[PATH_MAX]; snprintf(fn, PATH_MAX, "%s/fuzzer_setup", afl->out_dir); FILE *f = create_ffile(fn); - u32 i; + u32 i; fprintf(f, "# environment variables:\n"); - u32 s_afl_env = (u32) - sizeof(afl_environment_variables) / sizeof(afl_environment_variables[0]) - - 1U; + u32 s_afl_env = (u32)sizeof(afl_environment_variables) / + sizeof(afl_environment_variables[0]) - + 1U; for (i = 0; i < s_afl_env; ++i) { @@ -75,6 +75,7 @@ void write_setup_file(afl_state_t *afl, u32 argc, char **argv) { } } + fprintf(f, "\n"); fclose(f); @@ -1072,11 +1073,12 @@ void show_init_stats(afl_state_t *afl) { OKF("Here are some useful stats:\n\n" cGRA " Test case count : " cRST - "%u favored, %u variable, %u ignored, %u total\n" cGRA " Bitmap range : " cRST + "%u favored, %u variable, %u ignored, %u total\n" cGRA + " Bitmap range : " cRST "%u to %u bits (average: %0.02f bits)\n" cGRA " Exec timing : " cRST "%s to %s us (average: %s us)\n", - afl->queued_favored, afl->queued_variable, afl->queued_paths - count, afl->queued_paths, min_bits, - max_bits, + afl->queued_favored, afl->queued_variable, afl->queued_paths - count, + afl->queued_paths, min_bits, max_bits, ((double)afl->total_bitmap_size) / (afl->total_bitmap_entries ? afl->total_bitmap_entries : 1), stringify_int(IB(0), min_us), stringify_int(IB(1), max_us), -- cgit 1.4.1