diff options
| author | Andrea Fioraldi <andreafioraldi@gmail.com> | 2019-11-11 14:36:06 +0100 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2019-11-11 14:36:06 +0100 |
| commit | 659db7e421b47da4b04110a141d9c20307f74ecc (patch) | |
| tree | 18f9c38cc5270adcf445a62b974712cead4a01c4 | |
| parent | cd84339bccc104a51a5da614a9f82cc4ae615cce (diff) | |
| parent | 01d55372441960c435af8f3bd6b61d1302042728 (diff) | |
| download | afl++-659db7e421b47da4b04110a141d9c20307f74ecc.tar.gz | |
Merge branch 'master' into radamsa
114 files changed, 3593 insertions, 363 deletions
diff --git a/.custom-format.py b/.custom-format.py index b7416843..8d762006 100755 --- a/.custom-format.py +++ b/.custom-format.py @@ -1,4 +1,19 @@ #!/usr/bin/env python3 +# +# american fuzzy lop++ - custom code formatter +# -------------------------------------------- +# +# Written and maintaned by Andrea Fioraldi <andreafioraldi@gmail.com> +# +# Copyright 2015, 2016, 2017 Google Inc. All rights reserved. +# Copyright 2019 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 +# import subprocess import sys diff --git a/.gitignore b/.gitignore index f7907c76..ac3a653b 100644 --- a/.gitignore +++ b/.gitignore @@ -23,6 +23,8 @@ afl-clang-fast.8 afl-cmin.8 afl-fuzz.8 afl-gcc.8 +afl-gcc-fast.8 +afl-g++-fast.8 afl-gotcpu.8 afl-plot.8 afl-showmap.8 diff --git a/.travis.yml b/.travis.yml index 9ef95bcf..4569bd9c 100644 --- a/.travis.yml +++ b/.travis.yml @@ -1,11 +1,18 @@ +dist: bionic language: c env: - - AFL_I_DONT_CARE_ABOUT_MISSING_CRASHES=1 AFL_NO_UI=1 + - AFL_I_DONT_CARE_ABOUT_MISSING_CRASHES=1 AFL_NO_UI=1 AFL_STOP_MANUALLY=1 + # - AFL_I_DONT_CARE_ABOUT_MISSING_CRASHES=1 AFL_NO_UI=1 AFL_EXIT_WHEN_DONE=1 + # TODO: test AFL_BENCH_UNTIL_CRASH once we have a target that crashes + # - AFL_I_DONT_CARE_ABOUT_MISSING_CRASHES=1 AFL_NO_UI=1 AFL_BENCH_JUST_ONE=1 + +before_install: + - sudo apt update + - sudo apt install -y libtool libtool-bin automake bison libglib2.0 build-essential clang gcc-7 gcc-7-plugin-dev libc++-7-dev script: - - make - - ./afl-gcc ./test-instr.c -o test-instr - - mkdir seeds; mkdir out - - echo "" > seeds/nil_seed - - timeout --preserve-status 5s ./afl-fuzz -i seeds -o out/ -- ./test-instr + - gcc -v + - clang -v + - make distrib + - make tests diff --git a/Dockerfile b/Dockerfile new file mode 100644 index 00000000..558968d8 --- /dev/null +++ b/Dockerfile @@ -0,0 +1,27 @@ +FROM ubuntu:eoan +MAINTAINER David Carlier <devnexen@gmail.com> +LABEL "about"="AFLplusplus docker image" +RUN apt-get update && apt-get install -y --no-install-recommends \ + automake \ + bison \ + build-essential \ + clang \ + clang-9 \ + flex \ + gcc-9 \ + gcc-9-plugin-dev \ + libc++-9-dev \ + libtool \ + libtool-bin \ + libglib2.0-dev \ + llvm-9-tools \ + python-setuptools \ + wget \ + && rm -fr /var/lib/apt/lists/* +RUN mkdir /app +WORKDIR ["/app"] +COPY . . +ENV CC=gcc-9 +ENV CXX=g++-9 +ENV LLVM_CONFIG=llvm-config-9 +RUN make clean && make distrib && make install diff --git a/Makefile b/Makefile index 7ab9ae45..a29af7cb 100644 --- a/Makefile +++ b/Makefile @@ -2,7 +2,7 @@ # american fuzzy lop - makefile # ----------------------------- # -# Written and maintained by Michal Zalewski <lcamtuf@google.com> +# Written by Michal Zalewski # # Copyright 2013, 2014, 2015, 2016, 2017 Google Inc. All rights reserved. # @@ -16,9 +16,6 @@ # For Heiko: #TEST_MMAP=1 -PROGNAME = afl -VERSION = $(shell grep '^\#define VERSION ' include/config.h | cut -d '"' -f2) - PREFIX ?= /usr/local BIN_PATH = $(PREFIX)/bin HELPER_PATH = $(PREFIX)/lib/afl @@ -26,6 +23,9 @@ DOC_PATH = $(PREFIX)/share/doc/afl MISC_PATH = $(PREFIX)/share/afl MAN_PATH = $(PREFIX)/man/man8 +PROGNAME = afl +VERSION = $(shell grep '^\#define VERSION ' ../config.h | cut -d '"' -f2) + # PROGS intentionally omit afl-as, which gets installed elsewhere. PROGS = afl-gcc afl-fuzz afl-showmap afl-tmin afl-gotcpu afl-analyze @@ -33,10 +33,9 @@ SH_PROGS = afl-plot afl-cmin afl-whatsup afl-system-config MANPAGES=$(foreach p, $(PROGS) $(SH_PROGS), $(p).8) CFLAGS ?= -O3 -funroll-loops -CFLAGS += -Wall -D_FORTIFY_SOURCE=2 -g -Wno-pointer-sign \ - -I include/ \ - -DAFL_PATH=\"$(HELPER_PATH)\" -DDOC_PATH=\"$(DOC_PATH)\" \ - -DBIN_PATH=\"$(BIN_PATH)\" -Wno-unused-function +CFLAGS += -Wall -D_FORTIFY_SOURCE=2 -g -Wno-pointer-sign -I include/ \ + -DAFL_PATH=\"$(HELPER_PATH)\" -DBIN_PATH=\"$(BIN_PATH)\" \ + -DDOC_PATH=\"$(DOC_PATH)\" -Wno-unused-function AFL_FUZZ_FILES = $(wildcard src/afl-fuzz*.c) @@ -100,16 +99,24 @@ all: test_x86 test_shm test_python27 ready $(PROGS) afl-as test_build all_done man: $(MANPAGES) -$(MAKE) -C llvm_mode + -$(MAKE) -C gcc_plugin tests: source-only @cd test ; ./test.sh +performance-tests: performance-test +test-performance: performance-test + +performance-test: source-only + @cd test ; ./test-performance.sh + + 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, libdislocator, libtokencap" + @echo "source-only: everything for source code fuzzing: llvm_mode, 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" @@ -124,6 +131,8 @@ help: ifndef AFL_NO_X86 test_x86: + @echo "[*] Checking for the default compiler cc..." + @which $(CC) >/dev/null || ( echo; echo "Oops, looks like there is no compiler '"$(CC)"' in your path."; echo; echo "Don't panic! You can restart with '"$(_)" CC=<yourCcompiler>'."; echo; exit 1 ) @echo "[*] Checking for the ability to compile x86 code..." @echo 'main() { __asm__("xorb %al, %al"); }' | $(CC) -w -x c - -o .test1 || ( echo; echo "Oops, looks like your compiler can't generate x86 code."; echo; echo "Don't panic! You can use the LLVM or QEMU mode, but see docs/INSTALL first."; echo "(To ignore this error, set AFL_NO_X86=1 and try again.)"; echo; exit 1 ) @rm -f .test1 @@ -219,6 +228,9 @@ code-format: ./.custom-format.py -i llvm_mode/*.c ./.custom-format.py -i llvm_mode/*.h ./.custom-format.py -i llvm_mode/*.cc + ./.custom-format.py -i gcc_plugin/*.c + ./.custom-format.py -i gcc_plugin/*.h + ./.custom-format.py -i gcc_plugin/*.cc ./.custom-format.py -i qemu_mode/patches/*.h ./.custom-format.py -i qemu_mode/libcompcov/*.c ./.custom-format.py -i qemu_mode/libcompcov/*.cc @@ -232,7 +244,7 @@ ifndef AFL_NO_X86 test_build: afl-gcc afl-as afl-showmap @echo "[*] Testing the CC wrapper and instrumentation output..." - unset AFL_USE_ASAN AFL_USE_MSAN AFL_CC; AFL_QUIET=1 AFL_INST_RATIO=100 AFL_PATH=. ./$(TEST_CC) $(CFLAGS) test-instr.c -o test-instr $(LDFLAGS) + @unset AFL_USE_ASAN AFL_USE_MSAN AFL_CC; AFL_INST_RATIO=100 AFL_PATH=. ./$(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 ) ./afl-showmap -m none -q -o .test-instr0 ./test-instr < /dev/null echo 1 | ./afl-showmap -m none -q -o .test-instr1 ./test-instr @rm -f test-instr @@ -259,6 +271,7 @@ clean: rm -f $(PROGS) libradamsa.so afl-as as afl-g++ afl-clang afl-clang++ *.o src/*.o *~ a.out core core.[1-9][0-9]* *.stackdump .test .test1 .test2 test-instr .test-instr0 .test-instr1 qemu_mode/qemu-3.1.1.tar.xz afl-qemu-trace afl-gcc-fast afl-gcc-pass.so afl-gcc-rt.o afl-g++-fast *.so unicorn_mode/24f55a7973278f20f0de21b904851d99d4716263.tar.gz *.8 rm -rf out_dir qemu_mode/qemu-3.1.1 unicorn_mode/unicorn *.dSYM */*.dSYM -$(MAKE) -C llvm_mode clean + -$(MAKE) -C gcc_plugin clean $(MAKE) -C libdislocator clean $(MAKE) -C libtokencap clean $(MAKE) -C qemu_mode/unsigaction clean @@ -267,6 +280,7 @@ clean: distrib: all -$(MAKE) -C llvm_mode + -$(MAKE) -C gcc_plugin $(MAKE) -C libdislocator $(MAKE) -C libtokencap cd qemu_mode && sh ./build_qemu_support.sh @@ -280,6 +294,7 @@ binary-only: all source-only: all -$(MAKE) -C llvm_mode + -$(MAKE) -C gcc_plugin $(MAKE) -C libdislocator $(MAKE) -C libtokencap @@ -296,7 +311,7 @@ source-only: all @./$* -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 <mh@mh-sec.de>, Heiko \"hexc0der\" Eissfeldt <heiko.eissfeldt@hexco.de> and Andrea Fioraldi <andreafioraldi@gmail.com>" >> $@ + @echo "afl++ was written by Michal \"lcamtuf\" Zalewski and is maintained by Marc \"van Hauser\" Heuse <mh@mh-sec.de>, Heiko \"hexcoder-\" Eissfeldt <heiko.eissfeldt@hexco.de> and Andrea Fioraldi <andreafioraldi@gmail.com>" >> $@ @echo The homepage of afl++ is: https://github.com/vanhauser-thc/AFLplusplus >> $@ @echo >> $@ @echo .SH LICENSE >> $@ @@ -308,7 +323,7 @@ install: all $(MANPAGES) 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-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 $${DESTDIR}$(HELPER_PATH); fi ifndef AFL_TRACE_PC if [ -f afl-clang-fast -a -f libLLVMInsTrim.so -a -f afl-llvm-rt.o ]; then set -e; install -m 755 afl-clang-fast $${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 else diff --git a/README.md b/README.md index f818e271..583b7df8 100644 --- a/README.md +++ b/README.md @@ -1,8 +1,8 @@ # american fuzzy lop plus plus (afl++) - Release Version: 2.54c + Release Version: 2.58c - Github Version: 2.54d + Github Version: 2.58d includes all necessary/interesting changes from Google's afl 2.56b @@ -22,7 +22,7 @@ ## The enhancements compared to the original stock afl Many improvements were made over the official afl release - which did not - get any improvements since November 2017. + get any feature improvements since November 2017. Among other changes afl++ has a more performant llvm_mode, supports llvm up to version 9, QEMU 3.1, more speed and crashfixes for QEMU, @@ -52,15 +52,17 @@ A more thorough list is available in the PATCHES file. - | Feature/Instrumentation | LLVM | GCC | QEMU | Unicorn | - | ----------------------- |:----:|:---:|:----:| -------:| - | laf-intel / CompCov | x | | x | x | - | NeverZero | x(1)| x | x | x | - | Persistent mode | x | | x | | - | Whitelist | x | | | | - | InsTrim | x | | | | + | Feature/Instrumentation | AFL-GCC | LLVM_MODE | GCC_PLUGIN | QEMU_MODE | Unicorn | + | ----------------------- |:-------:|:---------:|:----------:|:---------:|:-------:| + | laf-intel / CompCov | | x | | x | x | + | NeverZero | x | x(1) | (2) | x | x | + | Persistent mode | | x | x | x | | + | Whitelist | | x | x | | | + | InsTrim | | x | | | | + neverZero: (1) only in LLVM >= 9.0 due to a bug in llvm in previous versions + (2) gcc create non-performant code, hence it is disabled in gcc_plugin So all in all this is the best-of AFL that is currently out there :-) @@ -103,14 +105,24 @@ These build options exist: * distrib: everything (for both binary-only and source code fuzzing) * install: installs everything you have compiled with the build options above * clean: cleans everything. for qemu_mode and unicorn_mode it means it deletes all downloads as well +* tests: runs test cases to ensure that all features are still working as they should * help: shows these build options -You can also build statically linked versions of the afl++ binaries by passing the STATIC=1 argument to make: +[Unless you are on Mac OS X](https://developer.apple.com/library/archive/qa/qa1118/_index.html) you can also build statically linked versions of the +afl++ binaries by passing the STATIC=1 argument to make: ```shell $ make all STATIC=1 ``` +Note that afl++ is faster and better the newer the compilers used. +Hence gcc-9 and especially llvm-9 should be the compilers of choice. +If your distribution does not have them, you can use the Dockerfile: + +```shell +$ docker build -t aflplusplus +``` + ## 1) Challenges of guided fuzzing @@ -453,7 +465,7 @@ parsers and grammars, but isn't nearly as good as the -x mode. If a dictionary is really hard to come by, another option is to let AFL run for a while, and then use the token capture library that comes as a companion -utility with AFL. For that, see [libtokencap/README.md](libtokencap/README.md). +utility with AFL. For that, see [libtokencap/README.md](libtokencap/README.tokencap.md). ## 11) Crash triage @@ -619,7 +631,7 @@ without feedback, bug reports, or patches from: Jonathan Gray Filipe Cabecinhas Nico Weber Jodie Cunningham Andrew Griffiths Parker Thompson - Jonathan Neuschfer Tyler Nighswander + Jonathan Neuschaefer Tyler Nighswander Ben Nagy Samir Aguiar Aidan Thornton Aleksandar Nikolich Sam Hakim Laszlo Szekeres @@ -638,7 +650,7 @@ without feedback, bug reports, or patches from: Austin Seipp Daniel Komaromy Daniel Binderman Jonathan Metzman Vegard Nossum Jan Kneschke - Kurt Roeckx Marcel Bohme + Kurt Roeckx Marcel Boehme Van-Thuan Pham Abhik Roychoudhury Joshua J. Drake Toby Hutton Rene Freingruber Sergey Davidoff diff --git a/TODO b/TODO index 40828168..b1a96569 100644 --- a/TODO +++ b/TODO @@ -1,16 +1,15 @@ -Roadmap 2.55d: -============== +Roadmap 2.60: +============= afl-fuzz: - - radamsa mutator + - radamsa mutator (via dlopen()) gcc_plugin: - - needs to be rewritten - - whitelist support - - skip over uninteresting blocks - laf-intel - - neverZero + +libdislocator: + - add a wrapper for posix_memalign qemu_mode: - update to 4.x (probably this will be skipped :( ) diff --git a/afl-cmin b/afl-cmin index 88635550..e4e0fd85 100755 --- a/afl-cmin +++ b/afl-cmin @@ -3,7 +3,7 @@ # american fuzzy lop - corpus minimization tool # --------------------------------------------- # -# Written and maintained by Michal Zalewski <lcamtuf@google.com> +# Written by Michal Zalewski # # Copyright 2014, 2015 Google Inc. All rights reserved. # @@ -36,7 +36,7 @@ # array sizes. # -echo "corpus minimization tool for afl-fuzz by <lcamtuf@google.com>" +echo "corpus minimization tool for afl-fuzz by Michal Zalewski" echo ######### diff --git a/afl-plot b/afl-plot index bc86fb85..b6108a09 100755 --- a/afl-plot +++ b/afl-plot @@ -3,7 +3,7 @@ # american fuzzy lop - Advanced Persistent Graphing # ------------------------------------------------- # -# Written and maintained by Michal Zalewski <lcamtuf@google.com> +# Written by Michal Zalewski # Based on a design & prototype by Michael Rash. # # Copyright 2014, 2015 Google Inc. All rights reserved. @@ -15,7 +15,7 @@ # http://www.apache.org/licenses/LICENSE-2.0 # -echo "progress plotting utility for afl-fuzz by <lcamtuf@google.com>" +echo "progress plotting utility for afl-fuzz by Michal Zalewski" echo if [ ! "$#" = "2" ]; then diff --git a/afl-system-config b/afl-system-config index a6e0efa4..2a7df17f 100755 --- a/afl-system-config +++ b/afl-system-config @@ -48,10 +48,10 @@ if [ "$PLATFORM" = "OpenBSD" ] ; then echo echo 'System security features cannot be disabled on OpenBSD.' fi -if [ "$PLATFORM" = "FreeBSD" ] ; then +if [ "$PLATFORM" = "NetBSD" ] ; then echo echo It is recommended to enable unprivileged users to set cpu affinity -to be able to use afl-gotcpu meaningfully. +echo to be able to use afl-gotcpu meaningfully. /sbin/sysctl -w security.models.extensions.user_set_cpu_affinity=1 fi if [ "$PLATFORM" = "Darwin" ] ; then diff --git a/afl-whatsup b/afl-whatsup index 505f7eba..2666d208 100755 --- a/afl-whatsup +++ b/afl-whatsup @@ -3,7 +3,7 @@ # american fuzzy lop - status check tool # -------------------------------------- # -# Written and maintained by Michal Zalewski <lcamtuf@google.com> +# Written by Michal Zalewski # # Copyright 2015 Google Inc. All rights reserved. # @@ -17,7 +17,7 @@ # instances of afl-fuzz. # -echo "status check tool for afl-fuzz by <lcamtuf@google.com>" +echo "status check tool for afl-fuzz by Michal Zalewski" echo test "$1" = "-h" && { echo $0 diff --git a/afl-wine-trace b/afl-wine-trace index f8284d7e..65525a33 100755 --- a/afl-wine-trace +++ b/afl-wine-trace @@ -4,9 +4,10 @@ import os import sys import pefile import shutil +import subprocess if len(sys.argv) < 2: - print("[afl-wine-trace] usage: wine-cov binary [args...]\n") + print("[afl-wine-trace] usage: ./afl-wine-trace binary [args...]\n") exit(1) if os.getenv("AFL_PATH"): @@ -42,14 +43,20 @@ else: elif pe.FILE_HEADER.Machine == pefile.MACHINE_TYPE["IMAGE_FILE_MACHINE_I386"]: qemu_path += "i386" else: - print ("[wine-cov] unsuppoted architecture\n") + print ("[afl-wine-trace] unsuppoted architecture\n") exit(1) qemu_path = shutil.which(qemu_path) -if os.getenv("WINECOV_WINE_PATH"): - wine_path = os.getenv("WINECOV_WINE_PATH") +wine_path = None +if os.getenv("AFL_WINE_PATH"): + wine_path = os.getenv("AFL_WINE_PATH") else: - wine_path = "/usr/lib/wine/wine" + if not wine_path and shutil.which("wine"): + wine_path = shutil.which("wine") + if not wine_path and os.path.exists("/usr/bin/wine"): + wine_path = "/usr/bin/wine" + if not wine_path and os.path.exists("/usr/lib/wine/wine"): + wine_path = "/usr/lib/wine/wine" if pe.FILE_HEADER.Machine == pefile.MACHINE_TYPE["IMAGE_FILE_MACHINE_AMD64"] or pe.FILE_HEADER.Machine == pefile.MACHINE_TYPE["IMAGE_FILE_MACHINE_IA64"]: wine_path += "64" elif pe.FILE_HEADER.Machine == pefile.MACHINE_TYPE["IMAGE_FILE_MACHINE_I386"]: @@ -58,4 +65,11 @@ else: print ("[afl-wine-trace] unsopported architecture\n") exit(1) -os.execve(qemu_path, [qemu_path, wine_path] + sys.argv[1:], os.environ) +argv = sys.argv[1:] +for i in range(len(argv)): + if ".cur_input" in argv[i]: + argv[i] = subprocess.run([os.path.join(os.path.dirname(wine_path), "winepath"), "--windows", argv[i]], universal_newlines=True, stdout=subprocess.PIPE).stdout + break + +print("[afl-wine-trace] exec:", " ".join([qemu_path, wine_path] + argv)) +os.execve(qemu_path, [qemu_path, wine_path] + argv, os.environ) diff --git a/dictionaries/gif.dict b/dictionaries/gif.dict index 71148937..441b3b8d 100644 --- a/dictionaries/gif.dict +++ b/dictionaries/gif.dict @@ -2,7 +2,7 @@ # AFL dictionary for GIF images # ----------------------------- # -# Created by Michal Zalewski <lcamtuf@google.com> +# Created by Michal Zalewski # header_87a="87a" diff --git a/dictionaries/html_tags.dict b/dictionaries/html_tags.dict index ba946df3..2805de90 100644 --- a/dictionaries/html_tags.dict +++ b/dictionaries/html_tags.dict @@ -5,7 +5,7 @@ # A basic collection of HTML tags likely to matter to HTML parsers. Does *not* # include any attributes or attribute values. # -# Created by Michal Zalewski <lcamtuf@google.com> +# Created by Michal Zalewski # tag_a="<a>" diff --git a/dictionaries/jpeg.dict b/dictionaries/jpeg.dict index 15efede7..40282f1a 100644 --- a/dictionaries/jpeg.dict +++ b/dictionaries/jpeg.dict @@ -2,7 +2,7 @@ # AFL dictionary for JPEG images # ------------------------------ # -# Created by Michal Zalewski <lcamtuf@google.com> +# Created by Michal Zalewski # header_jfif="JFIF\x00" diff --git a/dictionaries/js.dict b/dictionaries/js.dict index 9db37bfe..7926364c 100644 --- a/dictionaries/js.dict +++ b/dictionaries/js.dict @@ -4,7 +4,7 @@ # # Contains basic reserved keywords and syntax building blocks. # -# Created by Michal Zalewski <lcamtuf@google.com> +# Created by Michal Zalewski # keyword_arguments="arguments" diff --git a/dictionaries/png.dict b/dictionaries/png.dict index ea12d19e..ad9ea328 100644 --- a/dictionaries/png.dict +++ b/dictionaries/png.dict @@ -5,7 +5,7 @@ # Just the basic, standard-originating sections; does not include vendor # extensions. # -# Created by Michal Zalewski <lcamtuf@google.com> +# Created by Michal Zalewski # header_png="\x89PNG\x0d\x0a\x1a\x0a" diff --git a/dictionaries/regexp.dict b/dictionaries/regexp.dict new file mode 100644 index 00000000..957d18e2 --- /dev/null +++ b/dictionaries/regexp.dict @@ -0,0 +1,244 @@ +# +# AFL dictionary for JS regex +# --------------------------- +# +# Contains various regular expressions. +# +# Created by Yang Guo <yangguo@chromium.org> +# +# Contributed by Dhiraj Mishra <dhiraj@inputzero.io> +# +"?" +"abc" +"()" +"[]" +"abc|def" +"abc|def|ghi" +"^xxx$" +"ab\\b\\d\\bcd" +"\\w|\\d" +"a*?" +"abc+" +"abc+?" +"xyz?" +"xyz??" +"xyz{0,1}" +"xyz{0,1}?" +"xyz{93}" +"xyz{1,32}" +"xyz{1,32}?" +"xyz{1,}" +"xyz{1,}?" +"a\\fb\\nc\\rd\\te\\vf" +"a\\nb\\bc" +"/^\d*\./" +"(?:foo)" +"(?: foo )" +"foo|(bar|baz)|quux" +"foo(?=bar)baz" +"foo(?!bar)baz" +"foo(?<=bar)baz" +"foo(?<!bar)baz" +"()" +"(?=)" +"[]" +"[x]" +"[xyz]" +"[a-zA-Z0-9]" +"[-123]" +"[^123]" +"]" +"}" +"[a-b-c]" +"[x\\dz]" +"[\\d-z]" +"[\\d-\\d]" +"[z-\\d]" +"\\cj\\cJ\\ci\\cI\\ck\\cK" +"\\c!" +"\\c_" +"\\c~" +"[\\c!]" +"[\\c_]" +"[\\c~]" +"[\\ca]" +"[\\cz]" +"[\\cA]" +"[\\cZ]" +"[\\c1]" +"\\[\\]\\{\\}\\(\\)\\%\\^\\#\\ " +"[\\[\\]\\{\\}\\(\\)\\%\\^\\#\\ ]" +"\\8" +"\\9" +"\\11" +"\\11a" +"\\011" +"\\118" +"\\111" +"\\1111" +"(x)(x)(x)\\1" +"(x)(x)(x)\\2" +"(x)(x)(x)\\3" +"(x)(x)(x)\\4" +"(x)(x)(x)\\1*" +"(x)(x)(x)\\3*" +"(x)(x)(x)\\4*" +"(x)(x)(x)(x)(x)(x)(x)(x)(x)(x)\\10" +"(x)(x)(x)(x)(x)(x)(x)(x)(x)(x)\\11" +"(a)\\1" +"(a\\1)" +"(\\1a)" +"(\\2)(\\1)" +"(?=a){0,10}a" +"(?=a){1,10}a" +"(?=a){9,10}a" +"(?!a)?a" +"\\1(a)" +"(?!(a))\\1" +"(?!\\1(a\\1)\\1)\\1" +"\\1\\2(a(?:\\1(b\\1\\2))\\2)\\1" +"[\\0]" +"[\\11]" +"[\\11a]" +"[\\011]" +"[\\00011]" +"[\\118]" +"[\\111]" +"[\\1111]" +"\\x60" +"\\x3z" +"\\c" +"\\u0034" +"\\u003z" +"foo[z]*" +"\\u{12345}" +"\\u{12345}\\u{23456}" +"\\u{12345}{3}" +"\\u{12345}*" +"\\ud808\\udf45*" +"[\\ud808\\udf45-\\ud809\\udccc]" +"a" +"a|b" +"a\\n" +"a$" +"a\\b!" +"a\\Bb" +"a*?" +"a?" +"a??" +"a{0,1}?" +"a{1,2}?" +"a+?" +"(a)" +"(a)\\1" +"(\\1a)" +"\\1(a)" +"a\\s" +"a\\S" +"a\\D" +"a\\w" +"a\\W" +"a." +"a\\q" +"a[a]" +"a[^a]" +"a[a-z]" +"a(?:b)" +"a(?=b)" +"a(?!b)" +"\\x60" +"\\u0060" +"\\cA" +"\\q" +"\\1112" +"(a)\\1" +"(?!a)?a\\1" +"(?:(?=a))a\\1" +"a{}" +"a{,}" +"a{" +"a{z}" +"a{12z}" +"a{12," +"a{12,3b" +"{}" +"{,}" +"{" +"{z}" +"{1z}" +"{12," +"{12,3b" +"a" +"abc" +"a[bc]d" +"a|bc" +"ab|c" +"a||bc" +"(?:ab)" +"(?:ab|cde)" +"(?:ab)|cde" +"(ab)" +"(ab|cde)" +"(ab)\\1" +"(ab|cde)\\1" +"(?:ab)?" +"(?:ab)+" +"a?" +"a+" +"a??" +"a*?" +"a+?" +"(?:a?)?" +"(?:a+)?" +"(?:a?)+" +"(?:a*)+" +"(?:a+)+" +"(?:a?)*" +"(?:a*)*" +"(?:a+)*" +"a{0}" +"(?:a+){0,0}" +"a*b" +"a+b" +"a*b|c" +"a+b|c" +"(?:a{5,1000000}){3,1000000}" +"(?:ab){4,7}" +"a\\bc" +"a\\sc" +"a\\Sc" +"a(?=b)c" +"a(?=bbb|bb)c" +"a(?!bbb|bb)c" +"\xe2\x81\xa3" +"[\xe2\x81\xa3]" +"\xed\xb0\x80" +"\xed\xa0\x80" +"(\xed\xb0\x80)\x01" +"((\xed\xa0\x80))\x02" +"\xf0\x9f\x92\xa9" +"\x01" +"\x0f" +"[-\xf0\x9f\x92\xa9]+" +"[\xf0\x9f\x92\xa9-\xf4\x8f\xbf\xbf]" +"\[DataMember\((.+?)\)\]" +"/\d{1,2}\/\d{1,2}\/\d{4}/" +"a*b\+\|[0-9]\|\d{1,9}" +"(?<=)" +"(?<=a)" +"(?<!)" +"(?<!a)" +"(?<a>)" +"(?<a>.)" +"(?<a>.)\\k<a>" +"\\p{Script=Greek}" +"\\P{sc=Greek}" +"\\p{Script_Extensions=Greek}" +"\\P{scx=Greek}" +"\\p{General_Category=Decimal_Number}" +"\\P{gc=Decimal_Number}" +"\\p{gc=Nd}" +"\\P{Decimal_Number}" +"\\p{Nd}" +"\\P{Any}" +"\\p{Changes_When_NFKC_Casefolded}" diff --git a/dictionaries/sql.dict b/dictionaries/sql.dict index 58342473..efa44ba8 100644 --- a/dictionaries/sql.dict +++ b/dictionaries/sql.dict @@ -11,7 +11,7 @@ # standpoint, because they are usually not allowed in non-privileged # contexts). # -# Created by Michal Zalewski <lcamtuf@google.com> +# Created by Michal Zalewski # function_abs=" abs(1)" diff --git a/dictionaries/tiff.dict b/dictionaries/tiff.dict index 8f04b5af..720e56ce 100644 --- a/dictionaries/tiff.dict +++ b/dictionaries/tiff.dict @@ -5,7 +5,7 @@ # Just the basic, standard-originating sections; does not include vendor # extensions. # -# Created by Michal Zalewski <lcamtuf@google.com> +# Created by Michal Zalewski # header_ii="II*\x00" diff --git a/dictionaries/webp.dict b/dictionaries/webp.dict index 8a70e73b..53aa28c7 100644 --- a/dictionaries/webp.dict +++ b/dictionaries/webp.dict @@ -2,7 +2,7 @@ # AFL dictionary for WebP images # ------------------------------ # -# Created by Michal Zalewski <lcamtuf@google.com> +# Created by Michal Zalewski # header_RIFF="RIFF" diff --git a/dictionaries/xml.dict b/dictionaries/xml.dict index 8127aa28..d8375452 100644 --- a/dictionaries/xml.dict +++ b/dictionaries/xml.dict @@ -4,7 +4,7 @@ # # Several basic syntax elements and attributes, modeled on libxml2. # -# Created by Michal Zalewski <lcamtuf@google.com> +# Created by Michal Zalewski # attr_encoding=" encoding=\"1\"" diff --git a/docs/ChangeLog b/docs/ChangeLog index 7ccc8b66..7bc6dec4 100644 --- a/docs/ChangeLog +++ b/docs/ChangeLog @@ -14,11 +14,41 @@ sending a mail to <afl-users+subscribe@googlegroups.com>. ---------------------- -Version ++2.54d (dev): +Version ++2.58d (dev): ---------------------- + - afl-analyze: added AFL_SKIP_BIN_CHECK support + - better random numbers for gcc_plugin and llvm_mode (thanks to devnexen) + - afl-fuzz: CPU affinity support for DragonFly + - llvm_mode: float splitting is now configured via AFL_LLVM_LAF_SPLIT_FLOATS + - libtokencap: support for *BSD/OSX added + - libcompcov floating point splitting support for qemu and unicorn + - Dockerfile by courtesy of devnexen + - ripped regex.dictionary from Google afl PR + - qemu and unicorn download scripts now try to download until the full + download succeeded. f*ckin travis fails downloading 40% of the time! + - removed unnecessary warnings + + +-------------------------- +Version ++2.58c (release): +-------------------------- + + - reverted patch to not unlink and recreate the input file, it resulted in + performance loss of ~10% + - added test/test-performance.sh script + - (re)added gcc_plugin, fast inline instrumentation is not yet finished, + however it includes the whitelisting and persistance feature! by hexcoder- + - gcc_plugin tests added to testing framework + + +-------------------------------- +Version ++2.54d-2.57c (release): +-------------------------------- + + - we jump to 2.57 instead of 2.55 to catch up with Google's versioning - persistent mode for QEMU (see qemu_mode/README.md) - - custom mutator library is now a standard mutator, to exclusivly use it + - custom mutator library is now an additional mutator, to exclusivly use it add AFL_CUSTOM_MUTATOR_ONLY (that will trigger the previous behaviour) - new library qemu_mode/unsigaction which filters sigaction events - afl-fuzz: new command line option -I to execute a command on a new crash @@ -41,6 +71,7 @@ Version ++2.54d (dev): - updated documentation - Wine mode to run Win32 binaries with the QEMU instrumentation (-W) - CompareCoverage for ARM target in QEMU/Unicorn + - laf-intel in llvm_mode now also handles floating point comparisons -------------------------- diff --git a/docs/QuickStartGuide.txt b/docs/QuickStartGuide.txt index 9190dc98..723611e3 100644 --- a/docs/QuickStartGuide.txt +++ b/docs/QuickStartGuide.txt @@ -45,6 +45,8 @@ how to hit the ground running: 7) compile and use llvm_mode (afl-clang-fast/afl-clang-fast++) as it is way faster and has a few cool features +8) There is a basic docker build with 'docker build -t aflplusplus .' + That's it. Sit back, relax, and - time permitting - try to skim through the following files: diff --git a/docs/env_variables.txt b/docs/env_variables.txt index e763118e..427f8cca 100644 --- a/docs/env_variables.txt +++ b/docs/env_variables.txt @@ -71,11 +71,11 @@ 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++ -------------------------------------------------- +2) Settings for afl-clang-fast / afl-clang-fast++ / afl-gcc-fast / afl-g++-fast +--------------------------------------------------------------------------------- -The native LLVM instrumentation helper accepts a subset of the settings -discussed in section #1, with the exception of: +The native instrumentation helpers (llvm_mode and gcc_plugin) accept a subset +of the settings discussed in section #1, with the exception of: - AFL_AS, since this toolchain does not directly invoke GNU as. @@ -100,7 +100,7 @@ Then there are a few specific features that are only available in llvm_mode: - Setting AFL_LLVM_LAF_SPLIT_COMPARES will split all floating point and 64, 32 and 16 bit integer CMP instructions - See llvm_mode/README.laf-intel for more information. + See llvm_mode/README.laf-intel.md for more information. WHITELIST ========= @@ -109,7 +109,7 @@ Then there are a few specific features that are only available in llvm_mode: - Setting AFL_LLVM_WHITELIST with a filename will only instrument those files that match the names listed in this file. - See llvm_mode/README.whitelist for more information. + See llvm_mode/README.whitelist.md for more information. INSTRIM ======= @@ -122,7 +122,7 @@ 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 + See llvm_mode/README.instrim.md NOT_ZERO ======== @@ -133,7 +133,18 @@ Then there are a few specific features that are only available in llvm_mode: slowdown due a performance issue that is only fixed in llvm 9+. This feature increases path discovery by a little bit. - See llvm_mode/README.neverzero + See llvm_mode/README.neverzero.md + +Then there are a few specific features that are only available in the gcc_plugin: + + WHITELIST + ========= + This feature allows selective instrumentation of the source + + - Setting AFL_GCC_WHITELIST with a filename will only instrument those + files that match the names listed in this file (one filename per line). + + See gcc_plugin/README.whitelist.md for more information. 3) Settings for afl-fuzz ------------------------ @@ -257,7 +268,7 @@ The QEMU wrapper used to instrument binary-only code supports several settings: - Setting AFL_COMPCOV_LEVEL enables the CompareCoverage tracing of all cmp and sub in x86 and x86_64 and memory comparions functions (e.g. strcmp, memcmp, ...) when libcompcov is preloaded using AFL_PRELOAD. - More info at qemu_mode/libcompcov/README.compcov. + More info at qemu_mode/libcompcov/README.md. There are two levels at the moment, AFL_COMPCOV_LEVEL=1 that instruments only comparisons with immediate values / read-only memory and AFL_COMPCOV_LEVEL=2 that instruments all the comparions. Level 2 is more @@ -318,7 +329,7 @@ of decimal. 8) Settings for libdislocator.so -------------------------------- -The library honors three environmental variables: +The library honors these environmental variables: - AFL_LD_LIMIT_MB caps the size of the maximum heap usage permitted by the library, in megabytes. The default value is 1 GB. Once this is exceeded, diff --git a/docs/notes_for_asan.txt b/docs/notes_for_asan.txt index 972ca909..09ca172e 100644 --- a/docs/notes_for_asan.txt +++ b/docs/notes_for_asan.txt @@ -34,7 +34,7 @@ Note that ASAN is incompatible with -static, so be mindful of that. There is also the option of generating a corpus using a non-ASAN binary, and then feeding it to an ASAN-instrumented one to check for bugs. This is faster, and can give you somewhat comparable results. You can also try using -libdislocator (see libdislocator/README.dislocator in the parent directory) as a +libdislocator (see libdislocator/README.dislocator.md in the parent directory) as a lightweight and hassle-free (but less thorough) alternative. 2) Long version diff --git a/docs/perf_tips.txt b/docs/perf_tips.txt index 215895b6..0cac8f7b 100644 --- a/docs/perf_tips.txt +++ b/docs/perf_tips.txt @@ -50,12 +50,15 @@ Even if you don't have a lightweight harness for a particular target, remember that you can always use another, related library to generate a corpus that will be then manually fed to a more resource-hungry program later on. +Also note that reading the fuzzing input via stdin is faster than reading from +a file. + 3) Use LLVM instrumentation --------------------------- -When fuzzing slow targets, you can gain 2x performance improvement by using -the LLVM-based instrumentation mode described in llvm_mode/README.llvm. Note -that this mode requires the use of clang and will not work with GCC. +When fuzzing slow targets, you can gain 20-100% performance improvement by +using the LLVM-based instrumentation mode described in llvm_mode/README.llvm. +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 work well for certain types of self-contained libraries, and for fast targets, @@ -72,6 +75,9 @@ If you are only intested in specific parts of the code being fuzzed, you can whitelist the files that are actually relevant. This improves the speed and accuracy of afl. See llvm_mode/README.whitelist +Also use the InsTrim mode on larger binaries, this improves performance and +coverage a lot. + 4) Profile and optimize the binary ---------------------------------- @@ -161,6 +167,11 @@ and not waste CPU time. There are several OS-level factors that may affect fuzzing speed: + - If you have no risk of power loss then run your fuzzing on a tmpfs + partition. This increases the performance noticably. + Alternatively you can use AFL_TMPDIR to point to a tmpfs location to + just write the input file to a tmpfs. + - High system load. Use idle machines where possible. Kill any non-essential CPU hogs (idle browser windows, media players, complex screensavers, etc). diff --git a/docs/power_schedules.txt b/docs/power_schedules.txt index f5f66bd6..7b9d34c4 100644 --- a/docs/power_schedules.txt +++ b/docs/power_schedules.txt @@ -2,7 +2,7 @@ afl++'s power schedules based on AFLfast <a href="https://comp.nus.edu.sg/~mboehme/paper/CCS16.pdf"><img src="https://comp.nus.edu.sg/~mboehme/paper/CCS16.png" align="right" width="250"></a> Power schedules implemented by Marcel Böhme \<marcel.boehme@acm.org\>. -AFLFast is an extension of AFL which was written by Michal Zalewski \<lcamtuf@google.com\>. +AFLFast is an extension of AFL which was written by Michal Zalewski. AFLfast has helped in the success of Team Codejitsu at the finals of the DARPA Cyber Grand Challenge where their bot Galactica took **2nd place** in terms of #POVs proven (see red bar at https://www.cybergrandchallenge.com/event#results). AFLFast exposed several previously unreported CVEs that could not be exposed by AFL in 24 hours and otherwise exposed vulnerabilities significantly faster than AFL while generating orders of magnitude more unique crashes. diff --git a/docs/sister_projects.txt b/docs/sister_projects.txt index a2eb2a22..25e5560c 100644 --- a/docs/sister_projects.txt +++ b/docs/sister_projects.txt @@ -319,11 +319,13 @@ Fuzzer shell for SQLite (Richard Hipp) Support for Python mutation modules (Christian Holler) ------------------------------------------------------ +now integrated in AFL++, originally from here https://github.com/choller/afl/blob/master/docs/mozilla/python_modules.txt Support for selective instrumentation (Christian Holler) -------------------------------------------------------- +now integrated in AFL++, originally from here https://github.com/choller/afl/blob/master/docs/mozilla/partial_instrumentation.txt Kernel fuzzing (Dmitry Vyukov) diff --git a/experimental/argv_fuzzing/argv-fuzz-inl.h b/experimental/argv_fuzzing/argv-fuzz-inl.h index b042d38c..5d411046 100644 --- a/experimental/argv_fuzzing/argv-fuzz-inl.h +++ b/experimental/argv_fuzzing/argv-fuzz-inl.h @@ -2,7 +2,7 @@ american fuzzy lop - sample argv fuzzing wrapper ------------------------------------------------ - Written by Michal Zalewski <lcamtuf@google.com> + Written by Michal Zalewski Copyright 2015 Google Inc. All rights reserved. diff --git a/experimental/asan_cgroups/limit_memory.sh b/experimental/asan_cgroups/limit_memory.sh index b0c77d15..97950410 100755 --- a/experimental/asan_cgroups/limit_memory.sh +++ b/experimental/asan_cgroups/limit_memory.sh @@ -7,7 +7,7 @@ # David A. Wheeler <dwheeler@ida.org> # # Edits to bring the script in line with afl-cmin and other companion scripts -# by Michal Zalewski <lcamtuf@google.com>. All bugs are my fault. +# by Michal Zalewski. All bugs are my fault. # # Copyright 2015 Institute for Defense Analyses. # diff --git a/experimental/canvas_harness/canvas_harness.html b/experimental/canvas_harness/canvas_harness.html index 19293667..7b31d8b4 100644 --- a/experimental/canvas_harness/canvas_harness.html +++ b/experimental/canvas_harness/canvas_harness.html @@ -4,7 +4,7 @@ american fuzzy lop - <canvas> harness ------------------------------------- - Written and maintained by Michal Zalewski <lcamtuf@google.com> + Written by Michal Zalewski Copyright 2013, 2014 Google Inc. All rights reserved. diff --git a/experimental/clang_asm_normalize/as b/experimental/clang_asm_normalize/as index 4ee07e69..bd83c4ff 100755 --- a/experimental/clang_asm_normalize/as +++ b/experimental/clang_asm_normalize/as @@ -3,7 +3,7 @@ # american fuzzy lop - clang assembly normalizer # ---------------------------------------------- # -# Written and maintained by Michal Zalewski <lcamtuf@google.com> +# Written by Michal Zalewski # The idea for this wrapper comes from Ryan Govostes. # # Copyright 2013, 2014 Google Inc. All rights reserved. diff --git a/experimental/crash_triage/triage_crashes.sh b/experimental/crash_triage/triage_crashes.sh index 5894a4d6..205bc143 100755 --- a/experimental/crash_triage/triage_crashes.sh +++ b/experimental/crash_triage/triage_crashes.sh @@ -3,7 +3,7 @@ # american fuzzy lop - crash triage utility # ----------------------------------------- # -# Written and maintained by Michal Zalewski <lcamtuf@google.com> +# Written by Michal Zalewski # # Copyright 2013, 2014, 2017 Google Inc. All rights reserved. # @@ -22,7 +22,7 @@ # necessary. # -echo "crash triage utility for afl-fuzz by <lcamtuf@google.com>" +echo "crash triage utility for afl-fuzz by Michal Zalewski" echo ulimit -v 100000 2>/dev/null diff --git a/experimental/distributed_fuzzing/sync_script.sh b/experimental/distributed_fuzzing/sync_script.sh index 2d5e0635..31b0e436 100755 --- a/experimental/distributed_fuzzing/sync_script.sh +++ b/experimental/distributed_fuzzing/sync_script.sh @@ -3,7 +3,7 @@ # american fuzzy lop - fuzzer synchronization tool # ------------------------------------------------ # -# Written and maintained by Michal Zalewski <lcamtuf@google.com> +# Written by Michal Zalewski # # Copyright 2014 Google Inc. All rights reserved. # diff --git a/experimental/persistent_demo/persistent_demo.c b/experimental/persistent_demo/persistent_demo.c index d091febe..a94c8374 100644 --- a/experimental/persistent_demo/persistent_demo.c +++ b/experimental/persistent_demo/persistent_demo.c @@ -2,7 +2,7 @@ american fuzzy lop - persistent mode example -------------------------------------------- - Written and maintained by Michal Zalewski <lcamtuf@google.com> + Written by Michal Zalewski Copyright 2015 Google Inc. All rights reserved. diff --git a/experimental/post_library/post_library.so.c b/experimental/post_library/post_library.so.c index 72620ff0..048eea70 100644 --- a/experimental/post_library/post_library.so.c +++ b/experimental/post_library/post_library.so.c @@ -2,7 +2,7 @@ american fuzzy lop - postprocessor library example -------------------------------------------------- - Written and maintained by Michal Zalewski <lcamtuf@google.com> + Written by Michal Zalewski Copyright 2015 Google Inc. All rights reserved. diff --git a/experimental/post_library/post_library_png.so.c b/experimental/post_library/post_library_png.so.c index 080a6938..6ba95c1a 100644 --- a/experimental/post_library/post_library_png.so.c +++ b/experimental/post_library/post_library_png.so.c @@ -2,7 +2,7 @@ american fuzzy lop - postprocessor for PNG ------------------------------------------ - Written and maintained by Michal Zalewski <lcamtuf@google.com> + Written by Michal Zalewski Copyright 2015 Google Inc. All rights reserved. diff --git a/gcc_plugin/Makefile b/gcc_plugin/Makefile new file mode 100644 index 00000000..287b6545 --- /dev/null +++ b/gcc_plugin/Makefile @@ -0,0 +1,130 @@ +# +# american fuzzy lop - GCC plugin instrumentation +# ----------------------------------------------- +# +# Written by Austin Seipp <aseipp@pobox.com> and +# Laszlo Szekeres <lszekeres@google.com> and +# Michal Zalewski and +# Heiko Eißfeldt <heiko@hexco.de> +# +# 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 +# + +PREFIX ?= /usr/local +HELPER_PATH = $(PREFIX)/lib/afl +BIN_PATH = $(PREFIX)/bin + +CFLAGS ?= -O3 -g -funroll-loops +CFLAGS += -Wall -D_FORTIFY_SOURCE=2 -Wno-pointer-sign \ + -DAFL_PATH=\"$(HELPER_PATH)\" -DBIN_PATH=\"$(BIN_PATH)\" \ + +CXXFLAGS ?= -O3 -g -funroll-loops +CXXEFLAGS := $(CXXFLAGS) -Wall -D_FORTIFY_SOURCE=2 + +CC ?= gcc +CXX ?= g++ + +PLUGIN_FLAGS = -fPIC -fno-rtti -I"$(shell $(CC) -print-file-name=plugin)/include" + +ifeq "$(shell echo '\#include <sys/ipc.h>@\#include <sys/shm.h>@int main() { int _id = shmget(IPC_PRIVATE, 65536, IPC_CREAT | IPC_EXCL | 0600); shmctl(_id, IPC_RMID, 0); return 0;}' | tr @ '\n' | $(CC) -x c - -o .test2 2>/dev/null && echo 1 || echo 0 )" "1" + SHMAT_OK=1 +else + SHMAT_OK=0 + CFLAGS+=-DUSEMMAP=1 + LDFLAGS += -lrt +endif + +ifeq "$(TEST_MMAP)" "1" + SHMAT_OK=0 + CFLAGS+=-DUSEMMAP=1 + LDFLAGS += -lrt +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)'..." + @which $(CC) >/dev/null 2>&1 || ( echo "[-] Oops, can't find '$(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-gcc-fast: afl-gcc-fast.c | test_deps + $(CC) $(CFLAGS) $< -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) -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) ../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) + ../afl-showmap -m none -q -o .test-instr0 ./test-instr </dev/null + echo 1 | ../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/vanhauser-thc/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-gcc-fast' to compile programs." + +.NOTPARALLEL: clean + +vpath % .. +%.8: % + @echo .TH $* 8 `date "+%Y-%m-%d"` "afl++" > ../$@ + @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 >> ../$@ + @echo "afl++ was written by Michal \"lcamtuf\" Zalewski and is maintained by Marc \"van Hauser\" Heuse <mh@mh-sec.de>, Heiko \"hexcoder-\" Eissfeldt <heiko.eissfeldt@hexco.de> and Andrea Fioraldi <andreafioraldi@gmail.com>" >> ../$@ + @echo The homepage of afl++ is: https://github.com/vanhauser-thc/AFLplusplus >> ../$@ + @echo >> ../$@ + @echo .SH LICENSE >> ../$@ + @echo Apache License Version 2.0, January 2004 >> ../$@ + ln -sf afl-gcc-fast.8 ../afl-g++-fast.8 + +clean: + rm -f *.o *.so *~ a.out core core.[1-9][0-9]* test-instr .test-instr0 .test-instr1 + rm -f $(PROGS) ../afl-g++-fast ../afl-g*-fast.8 diff --git a/gcc_plugin/README.gcc.md b/gcc_plugin/README.gcc.md new file mode 100644 index 00000000..676ef427 --- /dev/null +++ b/gcc_plugin/README.gcc.md @@ -0,0 +1,158 @@ +=========================================== +GCC-based instrumentation for afl-fuzz +====================================== + + (See ../docs/README.md for the general instruction manual.) + (See ../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-<VERSION>-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. 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 +../docs/env_variables.txt). 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 <hexcoder-@github.com>. + +## 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 "master" 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 ../experimental/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/README.whitelist.md b/gcc_plugin/README.whitelist.md new file mode 100644 index 00000000..8ad2068d --- /dev/null +++ b/gcc_plugin/README.whitelist.md @@ -0,0 +1,73 @@ +======================================== +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_WHITELIST 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 whitelist (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 whitelist file containing: + +``` +feature_a/a1.cpp +feature_a/a2.cpp +``` + +However if the whitelist 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 whitelist file is then set to AFL_GCC_WHITELIST when you compile +your program. For each file that didn't match the whitelist, 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/afl-gcc-fast.c b/gcc_plugin/afl-gcc-fast.c new file mode 100644 index 00000000..057b44cc --- /dev/null +++ b/gcc_plugin/afl-gcc-fast.c @@ -0,0 +1,348 @@ +/* + american fuzzy lop - GCC wrapper for GCC plugin + ------------------------------------------------ + + Written by Austin Seipp <aseipp@pobox.com> and + Laszlo Szekeres <lszekeres@google.com> 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 "../include/debug.h" +#include "../include/alloc-inl.h" + +#include <stdio.h> +#include <unistd.h> +#include <stdlib.h> +#include <string.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 */ + +/* 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 : (u8*)"g++"; + + } else { + + u8* alt_cc = getenv("AFL_CC"); + cc_params[0] = alt_cc ? alt_cc : (u8*)"gcc"; + + } + + 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_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"; + + } + +#ifdef USEMMAP + 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) { + + if (argc < 2 || strcmp(argv[1], "-h") == 0) { + + printf( + cCYA + "afl-gcc-fast" VERSION cRST + " initially by <aseipp@pobox.com>, 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" + + "You can specify custom next-stage toolchain via AFL_CC and AFL_CXX. " + "Setting\n" + "AFL_HARDEN enables hardening optimizations in the compiled code.\n\n", + BIN_PATH, BIN_PATH); + + exit(1); + + } else if (isatty(2) && !getenv("AFL_QUIET")) { + + SAYF(cCYA "afl-gcc-fast" VERSION cRST + " initially by <aseipp@pobox.com>, maintainer: hexcoder-\n"); + + } + + 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<cc_par_cnt; i++) printf("%s\n", cc_params[i]); + + } + + */ + execvp(cc_params[0], (char**)cc_params); + + FATAL("Oops, failed to execute '%s' - check your PATH", cc_params[0]); + + return 0; + +} + diff --git a/gcc_plugin/afl-gcc-pass.so.cc b/gcc_plugin/afl-gcc-pass.so.cc new file mode 100644 index 00000000..0fa07774 --- /dev/null +++ b/gcc_plugin/afl-gcc-pass.so.cc @@ -0,0 +1,578 @@ +// +// There are some TODOs in this file: +// - fix instrumentation via external call +// - fix inline instrumentation +// - implement whitelist feature +// - dont instrument blocks that are uninteresting +// - implement neverZero +// + +/* + american fuzzy lop - GCC instrumentation pass + --------------------------------------------- + + Written by Austin Seipp <aseipp@pobox.com> with bits from + Emese Revfy <re.emese@gmail.com> + + Fixed by Heiko Eißfeldt 2019 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 <http://www.gnu.org/licenses/>. + + */ + +#define BUILD_INLINE_INST + +#include "../config.h" +#include "../include/debug.h" + +/* clear helper AFL types pulls in, which intervene with gcc-plugin geaders from + * GCC-8 */ +#ifdef likely +#undef likely +#endif +#ifdef unlikely +#undef unlikely +#endif + +#include <stdio.h> +#include <stdlib.h> +#include <unistd.h> + +#include <list> +#include <string> +#include <fstream> + +#include <gcc-plugin.h> +#include <plugin-version.h> +#include <diagnostic.h> +#include <tree.h> +#include <tree-ssa.h> +#include <tree-pass.h> +#include <tree-ssa-alias.h> +#include <basic-block.h> +#include <gimple-expr.h> +#include <gimple.h> +#include <gimple-iterator.h> +#include <gimple-ssa.h> +#include <version.h> +#include <toplev.h> +#include <intl.h> +#include <context.h> +#include <stringpool.h> +#include <cgraph.h> +#include <cfgloop.h> + +/* -------------------------------------------------------------------------- */ +/* -- AFL instrumentation pass ---------------------------------------------- */ + +static int be_quiet = 0; +static unsigned int inst_ratio = 100; +static bool inst_ext = true; +static std::list<std::string> myWhitelist; + +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; + + /* 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 */ + 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 <var_decl, prev_loc.0_1, prev_loc, NULL, NULL> + 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 <bit_xor_expr, _2, prev_loc.0_1, 47231, NULL> + 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 */ + + tree one = build_int_cst(unsigned_char_type_node, 1); + // tree zero = build_int_cst(unsigned_char_type_node, 0); + + // gimple_assign <addr_expr, p_6, &map[_2], NULL, NULL> + 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 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 <mem_ref, _3, *p_6, NULL, NULL> + 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); + + // gimple_assign <plus_expr, _4, _3, 1, NULL> + 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 <ssa_name, *p_6, _4, NULL, NULL> + // tree map_ptr3 = create_tmp_var_raw(map_type, "map_ptr3"); + g = gimple_build_assign(map_ptr_g, INDIRECT_REF, tmp2); + gimple_seq_add_stmt(&seq, g); // *map_ptr3 = tmp2 + update_stmt(g); + + /* Set prev_loc to cur_loc >> 1 */ + + // gimple_assign <integer_cst, prev_loc, 23615, NULL, NULL> + 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 (!myWhitelist.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 "<internal>". + */ + const char *fname = DECL_SOURCE_FILE(fun->decl); + + if (0 != strncmp("<internal>", fname, 10) && + 0 != strncmp("<built-in>", 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<std::string>::iterator it = myWhitelist.begin(); + it != myWhitelist.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 whitelisted, so we skip instrumentation. */ + if (!instrumentBlock) { + + if (!be_quiet) { + if (!instFilename.empty()) + SAYF(cYEL "[!] " cBRI "Not in whitelist, 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 = "20191015", + .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!")); + + } + + /* Show a banner */ + if (isatty(2) && !getenv("AFL_QUIET")) { + + SAYF(G_(cCYA "afl-gcc-pass" VERSION cRST + " initially by <aseipp@pobox.com>, 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 *instWhiteListFilename = getenv("AFL_GCC_WHITELIST"); + if (instWhiteListFilename) { + + std::string line; + std::ifstream fileStream; + fileStream.open(instWhiteListFilename); + if (!fileStream) fatal_error(0, "Unable to open AFL_GCC_WHITELIST"); + getline(fileStream, line); + while (fileStream) { + + myWhitelist.push_back(line); + getline(fileStream, line); + + } + + } else if (!be_quiet && getenv("AFL_LLVM_WHITELIST")) + + SAYF(cYEL "[-] " cRST + "AFL_LLVM_WHITELIST environment variable detected - did you mean " + "AFL_GCC_WHITELIST?\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 new file mode 100644 index 00000000..5b70a247 --- /dev/null +++ b/gcc_plugin/afl-gcc-rt.o.c @@ -0,0 +1,304 @@ +/* + american fuzzy lop - GCC plugin instrumentation bootstrap + --------------------------------------------------------- + + Written by Austin Seipp <aseipp@pobox.com> and + Laszlo Szekeres <lszekeres@google.com> 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 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 <stdlib.h> +#include <signal.h> +#include <unistd.h> +#include <string.h> +#include <assert.h> + +#include <sys/mman.h> +#include <sys/shm.h> +#include <sys/wait.h> +#include <sys/types.h> + +#include <sys/mman.h> +#include <fcntl.h> + +/* 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; +#else +__thread u32 __afl_prev_loc; +#endif + +/* Trace a basic block with some ID */ +void __afl_trace(u32 x) { + + u32 l = __afl_prev_loc; + +#if 0 /* enable for neverZero feature. By default disabled since too inefficient :-( */ + /* @Marc: avoid conditional jumps here */ + __afl_area_ptr[l ^ x] += 1 + (__afl_area_ptr[l ^ x] == (u8)~0); +#else + ++__afl_area_ptr[l ^ 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) { + + printf("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; + + printf("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) { + + static u8 tmp[4]; + 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 (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.llvm. */ + +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/include/afl-as.h b/include/afl-as.h index 048866db..f0263312 100644 --- a/include/afl-as.h +++ b/include/afl-as.h @@ -2,7 +2,7 @@ american fuzzy lop++ - injectable parts --------------------------------------- - Originally written by Michal Zalewski <lcamtuf@google.com> + Originally written by Michal Zalewski Now maintained by by Marc Heuse <mh@mh-sec.de>, Heiko Eißfeldt <heiko.eissfeldt@hexco.de> and diff --git a/include/afl-fuzz.h b/include/afl-fuzz.h index a1a4ed50..342205df 100644 --- a/include/afl-fuzz.h +++ b/include/afl-fuzz.h @@ -2,7 +2,7 @@ american fuzzy lop++ - fuzzer header ------------------------------------ - Originally written by Michal Zalewski <lcamtuf@google.com> + Originally written by Michal Zalewski Now maintained by by Marc Heuse <mh@mh-sec.de>, Heiko Eißfeldt <heiko.eissfeldt@hexco.de> and @@ -72,17 +72,21 @@ #include <sys/file.h> #if defined(__APPLE__) || defined(__FreeBSD__) || defined(__OpenBSD__) || \ - defined(__NetBSD__) + defined(__NetBSD__) || defined(__DragonFly__) #include <sys/sysctl.h> #endif /* __APPLE__ || __FreeBSD__ || __OpenBSD__ */ /* For systems that have sched_setaffinity; right now just Linux, but one can hope... */ -#if defined(__linux__) || defined(__FreeBSD__) || defined(__NetBSD__) +#if defined(__linux__) || defined(__FreeBSD__) || defined(__NetBSD__) || \ + defined(__DragonFly__) #define HAVE_AFFINITY 1 +#if defined(__FreeBSD__) || defined(__DragonFly__) +#include <sys/param.h> #if defined(__FreeBSD__) #include <sys/cpuset.h> +#endif #include <sys/user.h> #include <pthread.h> #include <pthread_np.h> @@ -252,7 +256,7 @@ extern u8 *in_dir, /* Input directory with test cases */ *file_extension, /* File extension */ *orig_cmdline, /* Original command line */ *doc_path, /* Path to documentation dir */ - *infoexec, /* Command to execute on a new crash */ + *infoexec, /* Command to execute on a new crash */ *out_file; /* File to fuzz, if any */ extern u32 exec_tmout; /* Configurable exec timeout (ms) */ @@ -315,7 +319,8 @@ extern u8 skip_deterministic, /* Skip deterministic stages? */ deferred_mode, /* Deferred forkserver mode? */ fixed_seed, /* do not reseed */ fast_cal, /* Try to calibrate faster? */ - uses_asan; /* Target uses ASAN? */ + uses_asan, /* Target uses ASAN? */ + disable_trim; /* Never trim in fuzz_one */ extern s32 out_fd, /* Persistent fd for out_file */ #ifndef HAVE_ARC4RANDOM diff --git a/include/alloc-inl.h b/include/alloc-inl.h index b0815ab1..f5bb7246 100644 --- a/include/alloc-inl.h +++ b/include/alloc-inl.h @@ -2,7 +2,7 @@ american fuzzy lop++ - error-checking, memory-zeroing alloc routines -------------------------------------------------------------------- - Originally written by Michal Zalewski <lcamtuf@google.com> + Originally written by Michal Zalewski Now maintained by by Marc Heuse <mh@mh-sec.de>, Heiko Eißfeldt <heiko.eissfeldt@hexco.de> and diff --git a/include/android-ashmem.h b/include/android-ashmem.h index 6c7a98db..f4d31739 100644 --- a/include/android-ashmem.h +++ b/include/android-ashmem.h @@ -2,7 +2,7 @@ american fuzzy lop++ - android shared memory compatibility layer ---------------------------------------------------------------- - Originally written by Michal Zalewski <lcamtuf@google.com> + Originally written by Michal Zalewski Now maintained by by Marc Heuse <mh@mh-sec.de>, Heiko Eißfeldt <heiko.eissfeldt@hexco.de> and diff --git a/include/common.h b/include/common.h index e13bf0a0..7c3d4517 100644 --- a/include/common.h +++ b/include/common.h @@ -2,7 +2,7 @@ american fuzzy lop++ - common routines header --------------------------------------------- - Originally written by Michal Zalewski <lcamtuf@google.com> + Originally written by Michal Zalewski Now maintained by by Marc Heuse <mh@mh-sec.de>, Heiko Eißfeldt <heiko.eissfeldt@hexco.de> and diff --git a/include/config.h b/include/config.h index f2732ad4..144d3810 100644 --- a/include/config.h +++ b/include/config.h @@ -2,7 +2,7 @@ american fuzzy lop++ - vaguely configurable bits ------------------------------------------------ - Originally written by Michal Zalewski <lcamtuf@google.com> + Originally written by Michal Zalewski Now maintained by by Marc Heuse <mh@mh-sec.de>, Heiko Eißfeldt <heiko.eissfeldt@hexco.de> and @@ -26,7 +26,7 @@ /* Version string: */ -#define VERSION "++2.54d" // c = release, d = volatile github dev +#define VERSION "++2.58d" // c = release, d = volatile github dev /****************************************************** * * diff --git a/include/debug.h b/include/debug.h index cccfc284..ed6c29e9 100644 --- a/include/debug.h +++ b/include/debug.h @@ -2,7 +2,7 @@ american fuzzy lop++ - debug / error handling macros ---------------------------------------------------- - Originally written by Michal Zalewski <lcamtuf@google.com> + Originally written by Michal Zalewski Now maintained by by Marc Heuse <mh@mh-sec.de>, Heiko Eißfeldt <heiko.eissfeldt@hexco.de> and diff --git a/include/forkserver.h b/include/forkserver.h index 9a099888..9cabe58e 100644 --- a/include/forkserver.h +++ b/include/forkserver.h @@ -2,7 +2,7 @@ american fuzzy lop++ - forkserver header ---------------------------------------- - Originally written by Michal Zalewski <lcamtuf@google.com> + Originally written by Michal Zalewski Forkserver design by Jann Horn <jannhorn@googlemail.com> diff --git a/include/hash.h b/include/hash.h index 7085df32..1a8ac252 100644 --- a/include/hash.h +++ b/include/hash.h @@ -12,7 +12,7 @@ Austin's original code is public domain. - Other code written and maintained by Michal Zalewski <lcamtuf@google.com> + Other code written by Michal Zalewski Copyright 2016 Google Inc. All rights reserved. diff --git a/include/sharedmem.h b/include/sharedmem.h index 18e4ee9f..cec6c025 100644 --- a/include/sharedmem.h +++ b/include/sharedmem.h @@ -2,7 +2,7 @@ american fuzzy lop++ - shared memory related header --------------------------------------------------- - Originally written by Michal Zalewski <lcamtuf@google.com> + Originally written by Michal Zalewski Forkserver design by Jann Horn <jannhorn@googlemail.com> diff --git a/include/types.h b/include/types.h index 07fc7e91..3f34db66 100644 --- a/include/types.h +++ b/include/types.h @@ -2,7 +2,7 @@ american fuzzy lop++ - type definitions and minor macros -------------------------------------------------------- - Originally written by Michal Zalewski <lcamtuf@google.com> + Originally written by Michal Zalewski Now maintained by by Marc Heuse <mh@mh-sec.de>, Heiko Eißfeldt <heiko.eissfeldt@hexco.de> and @@ -79,9 +79,21 @@ typedef int64_t s64; }) #ifdef AFL_LLVM_PASS +#if defined(__linux__) +#define AFL_SR(s) (srandom(s)) #define AFL_R(x) (random() % (x)) #else +#define AFL_SR(s) +#define AFL_R(x) (arc4random_uniform(x)) +#endif +#else +#if defined(__linux__) +#define SR(s) (srandom(s)) #define R(x) (random() % (x)) +#else +#define SR(s) +#define R(x) (arc4random_uniform(x)) +#endif #endif /* ^AFL_LLVM_PASS */ #define STRINGIFY_INTERNAL(x) #x diff --git a/libdislocator/Makefile b/libdislocator/Makefile index cbaa05ea..216d2862 100644 --- a/libdislocator/Makefile +++ b/libdislocator/Makefile @@ -2,7 +2,7 @@ # american fuzzy lop - libdislocator # ---------------------------------- # -# Written by Michal Zalewski <lcamtuf@google.com> +# Written by Michal Zalewski # # Copyright 2016 Google Inc. All rights reserved. # @@ -21,6 +21,10 @@ VERSION = $(shell grep '^\#define VERSION ' ../config.h | cut -d '"' -f2) CFLAGS ?= -O3 -funroll-loops -I ../include/ CFLAGS += -Wall -D_FORTIFY_SOURCE=2 -g -Wno-pointer-sign +ifdef USEHUGEPAGE + CFLAGS += -DUSEHUGEPAGE +endif + all: libdislocator.so libdislocator.so: libdislocator.so.c ../config.h @@ -34,5 +38,5 @@ clean: install: all install -m 755 ../libdislocator.so $${DESTDIR}$(HELPER_PATH) - install -m 644 README.dislocator $${DESTDIR}$(HELPER_PATH) + install -m 644 README.dislocator.md $${DESTDIR}$(HELPER_PATH) diff --git a/libdislocator/README.md b/libdislocator/README.dislocator.md index 5d5a1464..77626901 100644 --- a/libdislocator/README.md +++ b/libdislocator/README.dislocator.md @@ -25,6 +25,8 @@ heap-related security bugs in several ways: - It checks for calloc() overflows and can cause soft or hard failures of alloc requests past a configurable memory limit (AFL_LD_LIMIT_MB, AFL_LD_HARD_FAIL). + - Optionally, in platforms supporting it, huge pages can be used by passing + USEHUGEPAGE=1 to make. Basically, it is inspired by some of the non-default options available for the OpenBSD allocator - see malloc.conf(5) on that platform for reference. It is diff --git a/libdislocator/libdislocator.so.c b/libdislocator/libdislocator.so.c index 8834a1fc..0268cc52 100644 --- a/libdislocator/libdislocator.so.c +++ b/libdislocator/libdislocator.so.c @@ -3,7 +3,7 @@ american fuzzy lop - dislocator, an abusive allocator ----------------------------------------------------- - Written and maintained by Michal Zalewski <lcamtuf@google.com> + Written by Michal Zalewski Copyright 2016 Google Inc. All rights reserved. @@ -14,7 +14,7 @@ http://www.apache.org/licenses/LICENSE-2.0 This is a companion library that can be used as a drop-in replacement - for the libc allocator in the fuzzed binaries. See README.dislocator for + for the libc allocator in the fuzzed binaries. See README.dislocator.md for more info. */ @@ -23,8 +23,13 @@ #include <stdlib.h> #include <string.h> #include <limits.h> +#include <errno.h> #include <sys/mman.h> +#ifdef __APPLE__ +#include <mach/vm_statistics.h> +#endif + #include "config.h" #include "types.h" @@ -36,6 +41,8 @@ #define MAP_ANONYMOUS MAP_ANON #endif /* !MAP_ANONYMOUS */ +#define SUPER_PAGE_SIZE 1<<21 + /* Error / message handling: */ #define DEBUGF(_x...) \ @@ -88,6 +95,10 @@ static u8 alloc_verbose, /* Additional debug messages */ hard_fail, /* abort() when max_mem exceeded? */ no_calloc_over; /* abort() on calloc() overflows? */ +#if defined __OpenBSD__ || defined __APPLE__ +#define __thread +#warning no thread support available +#endif static __thread size_t total_mem; /* Currently allocated mem */ static __thread u32 call_depth; /* To avoid recursion via fprintf() */ @@ -100,6 +111,8 @@ static __thread u32 call_depth; /* To avoid recursion via fprintf() */ static void* __dislocator_alloc(size_t len) { void* ret; + size_t tlen; + int flags, fd, sp; if (total_mem + len > max_mem || total_mem + len < total_mem) { @@ -111,13 +124,44 @@ static void* __dislocator_alloc(size_t len) { } + tlen = (1 + PG_COUNT(len + 8)) * PAGE_SIZE; + flags = MAP_PRIVATE | MAP_ANONYMOUS; + fd = -1; +#if defined(USEHUGEPAGE) + sp = (len >= SUPER_PAGE_SIZE && !(len % SUPER_PAGE_SIZE)); + +#if defined(__APPLE__) + if (sp) fd = VM_FLAGS_SUPERPAGE_SIZE_2MB; +#elif defined(__linux__) + if (sp) flags |= MAP_HUGETLB; +#elif defined(__FreeBSD__) + if (sp) flags |= MAP_ALIGNED_SUPER; +#endif +#else + (void)sp; +#endif + /* We will also store buffer length and a canary below the actual buffer, so let's add 8 bytes for that. */ - ret = mmap(NULL, (1 + PG_COUNT(len + 8)) * PAGE_SIZE, PROT_READ | PROT_WRITE, - MAP_PRIVATE | MAP_ANONYMOUS, -1, 0); + ret = mmap(NULL, tlen, PROT_READ | PROT_WRITE, + flags, fd, 0); +#if defined(USEHUGEPAGE) + /* We try one more time with regular call */ + if (ret == MAP_FAILED) { +#if defined(__APPLE__) + fd = -1; +#elif defined(__linux__) + flags &= -MAP_HUGETLB; +#elif defined(__FreeBSD__) + flags &= -MAP_ALIGNED_SUPER; +#endif + ret = mmap(NULL, tlen, PROT_READ | PROT_WRITE, + flags, fd, 0); + } +#endif - if (ret == (void*)-1) { + if (ret == MAP_FAILED) { if (hard_fail) FATAL("mmap() failed on alloc (OOM?)"); @@ -184,6 +228,10 @@ void* calloc(size_t elem_len, size_t elem_cnt) { } +/* TODO: add a wrapper for posix_memalign, otherwise apps who use it, + will fail when freeing the memory. +*/ + /* The wrapper for malloc(). Roughly the same, also clobbers the returned memory (unlike calloc(), malloc() is not guaranteed to return zeroed memory). */ @@ -256,13 +304,61 @@ void* realloc(void* ptr, size_t len) { } +/* posix_memalign we mainly check the proper alignment argument + if the requested size fits within the alignment we do + a normal request */ + +int posix_memalign(void** ptr, size_t align, size_t len) { + if (*ptr == NULL) + return EINVAL; + if ((align % 2) || (align % sizeof(void *))) + return EINVAL; + if (len == 0) { + *ptr = NULL; + return 0; + } + if (align >= 4 * sizeof(size_t)) len += align -1; + + *ptr = malloc(len); + + DEBUGF("posix_memalign(%p %zu, %zu)", ptr, align, len); + + return 0; +} + +/* just the non-posix fashion */ + +void *memalign(size_t align, size_t len) { + void* ret = NULL; + + if (posix_memalign(&ret, align, len)) { + DEBUGF("memalign(%zu, %zu) failed", align, len); + } + + return ret; +} + +/* sort of C11 alias of memalign only more severe, alignment-wise */ + +void *aligned_alloc(size_t align, size_t len) { + void *ret = NULL; + + if ((len % align)) return NULL; + + if (posix_memalign(&ret, align, len)) { + DEBUGF("aligned_alloc(%zu, %zu) failed", align, len); + } + + return ret; +} + __attribute__((constructor)) void __dislocator_init(void) { - u8* tmp = getenv("AFL_LD_LIMIT_MB"); + u8* tmp = (u8 *)getenv("AFL_LD_LIMIT_MB"); if (tmp) { - max_mem = atoi(tmp) * 1024 * 1024; + max_mem = atoi((char *)tmp) * 1024 * 1024; if (!max_mem) FATAL("Bad value for AFL_LD_LIMIT_MB"); } diff --git a/libtokencap/Makefile b/libtokencap/Makefile index 3fd01b2c..6e1319d8 100644 --- a/libtokencap/Makefile +++ b/libtokencap/Makefile @@ -2,7 +2,7 @@ # american fuzzy lop - libtokencap # -------------------------------- # -# Written by Michal Zalewski <lcamtuf@google.com> +# Written by Michal Zalewski # # Copyright 2016 Google Inc. All rights reserved. # @@ -24,6 +24,18 @@ CFLAGS += -Wall -D_FORTIFY_SOURCE=2 -g -Wno-pointer-sign ifeq "$(shell uname)" "Linux" TARGETS = libtokencap.so endif +ifeq "$(shell uname)" "Darwin" + TARGETS = libtokencap.so +endif +ifeq "$(shell uname)" "FreeBSD" + TARGETS = libtokencap.so +endif +ifeq "$(shell uname)" "OpenBSD" + TARGETS = libtokencap.so +endif +ifeq "$(shell uname)" "NetBSD" + TARGETS = libtokencap.so +endif all: $(TARGETS) libtokencap.so: libtokencap.so.c ../config.h @@ -37,5 +49,5 @@ clean: install: all install -m 755 ../libtokencap.so $${DESTDIR}$(HELPER_PATH) - install -m 644 README.tokencap $${DESTDIR}$(HELPER_PATH) + install -m 644 README.tokencap.md $${DESTDIR}$(HELPER_PATH) diff --git a/libtokencap/README.md b/libtokencap/README.tokencap.md index baf69da1..8aae38bf 100644 --- a/libtokencap/README.md +++ b/libtokencap/README.tokencap.md @@ -2,7 +2,7 @@ (See ../docs/README for the general instruction manual.) -This Linux-only companion library allows you to instrument `strcmp()`, `memcmp()`, +This companion library allows you to instrument `strcmp()`, `memcmp()`, and related functions to automatically extract syntax tokens passed to any of these libcalls. The resulting list of tokens may be then given as a starting dictionary to afl-fuzz (the -x option) to improve coverage on subsequent @@ -55,9 +55,10 @@ If you don't get any results, the target library is probably not using strcmp() and memcmp() to parse input; or you haven't compiled it with -fno-builtin; or the whole thing isn't dynamically linked, and LD_PRELOAD is having no effect. -PS. The library is Linux-only because there is probably no particularly portable -and non-invasive way to distinguish between read-only and read-write memory -mappings. The `__tokencap_load_mappings()` function is the only thing that would -need to be changed for other OSes. Porting to platforms with /proc/<pid>/maps -(e.g., FreeBSD) should be trivial. +Portability hints: There is probably no particularly portable and non-invasive +way to distinguish between read-only and read-write memory mappings. +The `__tokencap_load_mappings()` function is the only thing that would +need to be changed for other OSes. + +Current supported OSes are: Linux, Darwin, FreeBSD (thanks to @devnexen) diff --git a/libtokencap/libtokencap.so.c b/libtokencap/libtokencap.so.c index 17b6190c..467be05b 100644 --- a/libtokencap/libtokencap.so.c +++ b/libtokencap/libtokencap.so.c @@ -3,7 +3,7 @@ american fuzzy lop - extract tokens passed to strcmp / memcmp ------------------------------------------------------------- - Written and maintained by Michal Zalewski <lcamtuf@google.com> + Written by Michal Zalewski Copyright 2016 Google Inc. All rights reserved. @@ -15,20 +15,32 @@ This Linux-only companion library allows you to instrument strcmp(), memcmp(), and related functions to automatically extract tokens. - See README.tokencap for more info. + See README.tokencap.md for more info. */ #include <stdio.h> #include <string.h> #include <ctype.h> +#include <unistd.h> +#include <fcntl.h> #include "../types.h" #include "../config.h" -#ifndef __linux__ -#error "Sorry, this library is Linux-specific for now!" -#endif /* !__linux__ */ +#if !defined __linux__ && !defined __APPLE__ && !defined __FreeBSD__ && !defined __OpenBSD__ && !defined __NetBSD__ +# error "Sorry, this library is unsupported in this platform for now!" +#endif /* !__linux__ && !__APPLE__ && ! __FreeBSD__ && ! __OpenBSD__ && !__NetBSD__*/ + +#if defined __APPLE__ +# include <mach/vm_map.h> +# include <mach/mach_init.h> +#elif defined __FreeBSD__ || defined __OpenBSD__ || defined __NetBSD__ +# include <sys/types.h> +# include <sys/sysctl.h> +# include <sys/user.h> +# include <sys/mman.h> +#endif /* Mapping data and such */ @@ -38,7 +50,8 @@ static struct mapping { void *st, *en; } __tokencap_ro[MAX_MAPPINGS]; static u32 __tokencap_ro_cnt; static u8 __tokencap_ro_loaded; -static FILE* __tokencap_out_file; +static int __tokencap_out_file = -1; +static pid_t __tokencap_pid = -1; /* Identify read-only regions in memory. Only parameters that fall into these ranges are worth dumping when passed to strcmp() and so on. Read-write @@ -46,6 +59,8 @@ static FILE* __tokencap_out_file; static void __tokencap_load_mappings(void) { +#if defined __linux__ + u8 buf[MAX_LINE]; FILE* f = fopen("/proc/self/maps", "r"); @@ -70,6 +85,106 @@ static void __tokencap_load_mappings(void) { fclose(f); +#elif defined __APPLE__ + + struct vm_region_submap_info_64 region; + mach_msg_type_number_t cnt = VM_REGION_SUBMAP_INFO_COUNT_64; + vm_address_t base = 0; + vm_size_t size = 0; + natural_t depth = 0; + + __tokencap_ro_loaded = 1; + + while (1) { + + if (vm_region_recurse_64(mach_task_self(), &base, &size, &depth, + (vm_region_info_64_t)®ion, &cnt) != KERN_SUCCESS) break; + + if (region.is_submap) { + depth++; + } else { + /* We only care of main map addresses and the read only kinds */ + if ((region.protection & VM_PROT_READ) && !(region.protection & VM_PROT_WRITE)) { + __tokencap_ro[__tokencap_ro_cnt].st = (void *)base; + __tokencap_ro[__tokencap_ro_cnt].en = (void *)(base + size); + + if (++__tokencap_ro_cnt == MAX_MAPPINGS) break; + } + } + } + +#elif defined __FreeBSD__ || defined __OpenBSD__ || defined __NetBSD__ + +#if defined __FreeBSD__ + int mib[] = {CTL_KERN, KERN_PROC, KERN_PROC_VMMAP, __tokencap_pid}; +#elif defined __OpenBSD__ + int mib[] = {CTL_KERN, KERN_PROC_VMMAP, __tokencap_pid}; +#elif defined __NetBSD__ + int mib[] = {CTL_VM, VM_PROC, VM_PROC_MAP, __tokencap_pid, sizeof(struct kinfo_vmentry)}; +#endif + char *buf, *low, *high; + size_t miblen = sizeof(mib)/sizeof(mib[0]); + size_t len; + + if (sysctl(mib, miblen, NULL, &len, NULL, 0) == -1) return; + +#if defined __FreeBSD__ || defined __NetBSD__ + len = len * 4 / 3; +#elif defined __OpenBSD__ + len -= len % sizeof(struct kinfo_vmentry); +#endif + + buf = mmap(NULL, len, PROT_READ | PROT_WRITE, MAP_SHARED | MAP_ANON, -1, 0); + if (buf == MAP_FAILED) return; + + if (sysctl(mib, miblen, buf, &len, NULL, 0) == -1) { + + munmap(buf, len); + return; + + } + + low = buf; + high = low + len; + + __tokencap_ro_loaded = 1; + + while (low < high) { + struct kinfo_vmentry *region = (struct kinfo_vmentry *)low; + +#if defined __FreeBSD__ || defined __NetBSD__ + +#if defined __FreeBSD__ + size_t size = region->kve_structsize; + + if (size == 0) break; +#elif defined __NetBSD__ + size_t size = sizeof (*region); +#endif + + /* We go through the whole mapping of the process and track read-only addresses */ + if ((region->kve_protection & KVME_PROT_READ) && + !(region->kve_protection & KVME_PROT_WRITE)) { + +#elif defined __OpenBSD__ + + size_t size = sizeof (*region); + + /* We go through the whole mapping of the process and track read-only addresses */ + if ((region->kve_protection & KVE_PROT_READ) && + !(region->kve_protection & KVE_PROT_WRITE)) { +#endif + __tokencap_ro[__tokencap_ro_cnt].st = (void *)region->kve_start; + __tokencap_ro[__tokencap_ro_cnt].en = (void *)region->kve_end; + + if (++__tokencap_ro_cnt == MAX_MAPPINGS) break; + } + + low += size; + } + + munmap(buf, len); +#endif } /* Check an address against the list of read-only mappings. */ @@ -96,7 +211,7 @@ static void __tokencap_dump(const u8* ptr, size_t len, u8 is_text) { u32 i; u32 pos = 0; - if (len < MIN_AUTO_EXTRA || len > MAX_AUTO_EXTRA || !__tokencap_out_file) + if (len < MIN_AUTO_EXTRA || len > MAX_AUTO_EXTRA || __tokencap_out_file == -1) return; for (i = 0; i < len; i++) { @@ -122,7 +237,9 @@ static void __tokencap_dump(const u8* ptr, size_t len, u8 is_text) { buf[pos] = 0; - fprintf(__tokencap_out_file, "\"%s\"\n", buf); + int wrt_ok = ( 1 == write(__tokencap_out_file, "\"", 1)); + wrt_ok &= (pos == write(__tokencap_out_file, buf, pos)); + wrt_ok &= (2 == write(__tokencap_out_file, "\"\n", 2)); } @@ -138,7 +255,7 @@ int strcmp(const char* str1, const char* str2) { while (1) { - unsigned char c1 = *str1, c2 = *str2; + const unsigned char c1 = *str1, c2 = *str2; if (c1 != c2) return (c1 > c2) ? 1 : -1; if (!c1) return 0; @@ -180,7 +297,7 @@ int strcasecmp(const char* str1, const char* str2) { while (1) { - unsigned char c1 = tolower(*str1), c2 = tolower(*str2); + const unsigned char c1 = tolower(*str1), c2 = tolower(*str2); if (c1 != c2) return (c1 > c2) ? 1 : -1; if (!c1) return 0; @@ -200,7 +317,7 @@ int strncasecmp(const char* str1, const char* str2, size_t len) { while (len--) { - unsigned char c1 = tolower(*str1), c2 = tolower(*str2); + const unsigned char c1 = tolower(*str1), c2 = tolower(*str2); if (!c1) return 0; if (c1 != c2) return (c1 > c2) ? 1 : -1; @@ -220,12 +337,15 @@ int memcmp(const void* mem1, const void* mem2, size_t len) { if (__tokencap_is_ro(mem1)) __tokencap_dump(mem1, len, 0); if (__tokencap_is_ro(mem2)) __tokencap_dump(mem2, len, 0); + const char *strmem1 = (const char *)mem1; + const char *strmem2 = (const char *)mem2; + while (len--) { - unsigned char c1 = *(const char*)mem1, c2 = *(const char*)mem2; + const unsigned char c1 = *strmem1, c2 = *strmem2; if (c1 != c2) return (c1 > c2) ? 1 : -1; - mem1++; - mem2++; + strmem1++; + strmem2++; } @@ -233,6 +353,28 @@ int memcmp(const void* mem1, const void* mem2, size_t len) { } +#undef bcmp + +int bcmp(const void* mem1, const void* mem2, size_t len) { + + if (__tokencap_is_ro(mem1)) __tokencap_dump(mem1, len, 0); + if (__tokencap_is_ro(mem2)) __tokencap_dump(mem2, len, 0); + + const char *strmem1 = (const char *)mem1; + const char *strmem2 = (const char *)mem2; + + while (len--) { + + int diff = *strmem1 ^ *strmem2; + if (diff != 0) return 1; + strmem1++; + strmem2++; + + } + + return 0; +} + #undef strstr char* strstr(const char* haystack, const char* needle) { @@ -288,8 +430,14 @@ char* strcasestr(const char* haystack, const char* needle) { __attribute__((constructor)) void __tokencap_init(void) { u8* fn = getenv("AFL_TOKEN_FILE"); - if (fn) __tokencap_out_file = fopen(fn, "a"); - if (!__tokencap_out_file) __tokencap_out_file = stderr; + if (fn) __tokencap_out_file = open(fn, O_RDWR | O_CREAT | O_APPEND, 0655); + if (__tokencap_out_file == -1) __tokencap_out_file = STDERR_FILENO; + __tokencap_pid = getpid(); + +} +/* closing as best as we can the tokens file */ +__attribute__((destructor)) void __tokencap_shutdown(void) { + if (__tokencap_out_file != STDERR_FILENO) close(__tokencap_out_file); } diff --git a/llvm_mode/LLVMInsTrim.so.cc b/llvm_mode/LLVMInsTrim.so.cc index 4b5597e2..89738812 100644 --- a/llvm_mode/LLVMInsTrim.so.cc +++ b/llvm_mode/LLVMInsTrim.so.cc @@ -158,6 +158,7 @@ struct InsTrim : public ModulePass { bool instrumentBlock = false; DebugLoc Loc; StringRef instFilename; + unsigned int instLine = 0; for (auto &BB : F) { @@ -171,7 +172,7 @@ struct InsTrim : public ModulePass { DILocation *cDILoc = dyn_cast<DILocation>(Loc.getAsMDNode()); - unsigned int instLine = cDILoc->getLine(); + instLine = cDILoc->getLine(); instFilename = cDILoc->getFilename(); if (instFilename.str().empty()) { @@ -217,11 +218,13 @@ struct InsTrim : public ModulePass { * not whitelisted, so we skip instrumentation. */ if (!instrumentBlock) { - if (!instFilename.str().empty()) - SAYF(cYEL "[!] " cBRI "Not in whitelist, skipping %s ...\n", - instFilename.str().c_str()); - else - SAYF(cYEL "[!] " cBRI "No filename information found, skipping it"); + if (!be_quiet) { + if (!instFilename.str().empty()) + SAYF(cYEL "[!] " cBRI "Not in whitelist, skipping %s line %u...\n", + instFilename.str().c_str(), instLine); + else + SAYF(cYEL "[!] " cBRI "No filename information found, skipping it"); + } continue; } diff --git a/llvm_mode/Makefile b/llvm_mode/Makefile index 55bfab59..033babac 100644 --- a/llvm_mode/Makefile +++ b/llvm_mode/Makefile @@ -3,7 +3,7 @@ # ----------------------------------------- # # Written by Laszlo Szekeres <lszekeres@google.com> and -# Michal Zalewski <lcamtuf@google.com> +# Michal Zalewski # # LLVM integration design comes from Laszlo Szekeres. # @@ -52,7 +52,7 @@ endif CFLAGS ?= -O3 -funroll-loops CFLAGS += -Wall -D_FORTIFY_SOURCE=2 -g -Wno-pointer-sign -I ../include/ \ -DAFL_PATH=\"$(HELPER_PATH)\" -DBIN_PATH=\"$(BIN_PATH)\" \ - -DVERSION=\"$(VERSION)\" + -DLLVM_BINDIR=\"$(LLVM_BINDIR)\" -DVERSION=\"$(VERSION)\" ifdef AFL_TRACE_PC CFLAGS += -DUSE_TRACE_PC=1 endif @@ -220,7 +220,7 @@ vpath % .. @../$* -h 2>&1 | tail -n +4 >> ../$@ @echo >> ../$@ @echo .SH AUTHOR >> ../$@ - @echo "afl++ was written by Michal \"lcamtuf\" Zalewski and is maintained by Marc \"van Hauser\" Heuse <mh@mh-sec.de>, Heiko \"hexc0der\" Eissfeldt <heiko.eissfeldt@hexco.de> and Andrea Fioraldi <andreafioraldi@gmail.com>" >> ../$@ + @echo "afl++ was written by Michal \"lcamtuf\" Zalewski and is maintained by Marc \"van Hauser\" Heuse <mh@mh-sec.de>, Heiko \"hexcoder-\" Eissfeldt <heiko.eissfeldt@hexco.de> and Andrea Fioraldi <andreafioraldi@gmail.com>" >> ../$@ @echo The homepage of afl++ is: https://github.com/vanhauser-thc/AFLplusplus >> ../$@ @echo >> ../$@ @echo .SH LICENSE >> ../$@ diff --git a/llvm_mode/README.laf-intel.md b/llvm_mode/README.laf-intel.md index c787744b..462c7bac 100644 --- a/llvm_mode/README.laf-intel.md +++ b/llvm_mode/README.laf-intel.md @@ -35,4 +35,5 @@ 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_COMPARES` setting. +It is activated with the `AFL_LLVM_LAF_SPLIT_FLOATS` setting, available only +when `AFL_LLVM_LAF_SPLIT_COMPARES` is set. diff --git a/llvm_mode/README.whitelist.md b/llvm_mode/README.whitelist.md index 6a5770c2..72fb5d09 100644 --- a/llvm_mode/README.whitelist.md +++ b/llvm_mode/README.whitelist.md @@ -64,7 +64,7 @@ a2.cpp but it might lead to files being unwantedly instrumented if the same filename exists somewhere else in the project directories. -The created whitelist file is then set to AFL_INST_WHITELIST when you compile +The created whitelist file is then set to AFL_LLVM_WHITELIST when you compile your program. For each file that didn't match the whitelist, 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/llvm_mode/afl-clang-fast.c b/llvm_mode/afl-clang-fast.c index 142d6331..b2243492 100644 --- a/llvm_mode/afl-clang-fast.c +++ b/llvm_mode/afl-clang-fast.c @@ -3,7 +3,7 @@ ------------------------------------------------ Written by Laszlo Szekeres <lszekeres@google.com> and - Michal Zalewski <lcamtuf@google.com> + Michal Zalewski LLVM integration design comes from Laszlo Szekeres. @@ -32,11 +32,13 @@ #include <unistd.h> #include <stdlib.h> #include <string.h> +#include <limits.h> #include <assert.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]; /* Try to find the runtime libraries. If that fails, abort. */ @@ -104,6 +106,7 @@ static void find_obj(u8* argv0) { static void edit_params(u32 argc, char** argv) { u8 fortify_set = 0, asan_set = 0, x_set = 0, maybe_linking = 1, bit_mode = 0; + u8 has_llvm_config = 0; u8* name; cc_params = ck_alloc((argc + 128) * sizeof(u8*)); @@ -112,23 +115,29 @@ static void edit_params(u32 argc, char** argv) { if (!name) name = argv[0]; else - name++; + ++name; + + has_llvm_config = (strlen(LLVM_BINDIR) > 0); if (!strcmp(name, "afl-clang-fast++")) { u8* alt_cxx = getenv("AFL_CXX"); - cc_params[0] = alt_cxx ? alt_cxx : (u8*)"clang++"; + if (has_llvm_config) snprintf(llvm_fullpath, sizeof(llvm_fullpath), "%s/clang++", LLVM_BINDIR); + else sprintf(llvm_fullpath, "clang++"); + cc_params[0] = alt_cxx ? alt_cxx : (u8*)llvm_fullpath; } else { u8* alt_cc = getenv("AFL_CC"); - cc_params[0] = alt_cc ? alt_cc : (u8*)"clang"; + if (has_llvm_config) snprintf(llvm_fullpath, sizeof(llvm_fullpath), "%s/clang", LLVM_BINDIR); + else sprintf(llvm_fullpath, "clang"); + cc_params[0] = alt_cc ? alt_cc : (u8*)llvm_fullpath; } /* There are three ways to compile with afl-clang-fast. In the traditional mode, we use afl-llvm-pass.so, then there is libLLVMInsTrim.so which is - much faster but has less coverage. Finally tere is the experimental + much faster but has less coverage. Finally there is the experimental '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 @@ -273,6 +282,9 @@ static void edit_params(u32 argc, char** argv) { 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"; } diff --git a/llvm_mode/afl-llvm-pass.so.cc b/llvm_mode/afl-llvm-pass.so.cc index 3ca5ccc4..0c68136b 100644 --- a/llvm_mode/afl-llvm-pass.so.cc +++ b/llvm_mode/afl-llvm-pass.so.cc @@ -3,7 +3,7 @@ --------------------------------------------------- Written by Laszlo Szekeres <lszekeres@google.com> and - Michal Zalewski <lcamtuf@google.com> + Michal Zalewski LLVM integration design comes from Laszlo Szekeres. C bits copied-and-pasted from afl-as.c are Michal's fault. @@ -34,6 +34,7 @@ #include <list> #include <string> #include <fstream> +#include <sys/time.h> #include "llvm/IR/DebugInfo.h" #include "llvm/IR/BasicBlock.h" @@ -95,8 +96,16 @@ bool AFLCoverage::runOnModule(Module &M) { IntegerType *Int8Ty = IntegerType::getInt8Ty(C); IntegerType *Int32Ty = IntegerType::getInt32Ty(C); + 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 */ char be_quiet = 0; @@ -181,6 +190,8 @@ bool AFLCoverage::runOnModule(Module &M) { } + (void)instLine; + /* Continue only if we know where we actually are */ if (!instFilename.str().empty()) { diff --git a/llvm_mode/afl-llvm-rt.o.c b/llvm_mode/afl-llvm-rt.o.c index 20b34336..5740fe42 100644 --- a/llvm_mode/afl-llvm-rt.o.c +++ b/llvm_mode/afl-llvm-rt.o.c @@ -3,7 +3,7 @@ --------------------------------------------------- Written by Laszlo Szekeres <lszekeres@google.com> and - Michal Zalewski <lcamtuf@google.com> + Michal Zalewski LLVM integration design comes from Laszlo Szekeres. diff --git a/llvm_mode/compare-transform-pass.so.cc b/llvm_mode/compare-transform-pass.so.cc index e1b6e671..0ccce875 100644 --- a/llvm_mode/compare-transform-pass.so.cc +++ b/llvm_mode/compare-transform-pass.so.cc @@ -234,6 +234,10 @@ bool CompareTransform::transformCmps(Module &M, const bool processStrcmp, ConstantInt *ilen = dyn_cast<ConstantInt>(op2); sizedLen = ilen->getZExtValue(); + } else { + + sizedLen = 0; + } if (HasStr1) { diff --git a/llvm_mode/split-compares-pass.so.cc b/llvm_mode/split-compares-pass.so.cc index f1a0f94e..60420f77 100644 --- a/llvm_mode/split-compares-pass.so.cc +++ b/llvm_mode/split-compares-pass.so.cc @@ -50,6 +50,8 @@ class SplitComparesTransform : public ModulePass { } private: + int enableFPSplit; + size_t splitIntCompares(Module &M, unsigned bitw); size_t splitFPCompares(Module &M); bool simplifyCompares(Module &M); @@ -101,10 +103,11 @@ bool SplitComparesTransform::simplifyCompares(Module &M) { } - if (selectcmpInst->getPredicate() == CmpInst::FCMP_OGE || + if (enableFPSplit && ( + selectcmpInst->getPredicate() == CmpInst::FCMP_OGE || selectcmpInst->getPredicate() == CmpInst::FCMP_UGE || selectcmpInst->getPredicate() == CmpInst::FCMP_OLE || - selectcmpInst->getPredicate() == CmpInst::FCMP_ULE) { + selectcmpInst->getPredicate() == CmpInst::FCMP_ULE)) { auto op0 = selectcmpInst->getOperand(0); auto op1 = selectcmpInst->getOperand(1); @@ -115,6 +118,8 @@ bool SplitComparesTransform::simplifyCompares(Module &M) { /* this is probably not needed but we do it anyway */ if (TyOp0 != TyOp1) { continue; } + if (TyOp0->isArrayTy() || TyOp0->isVectorTy()) { continue; } + fcomps.push_back(selectcmpInst); } @@ -476,6 +481,8 @@ size_t SplitComparesTransform::splitFPCompares(Module &M) { if (TyOp0 != TyOp1) { continue; } + if (TyOp0->isArrayTy() || TyOp0->isVectorTy()) { continue; } + fcomps.push_back(selectcmpInst); } @@ -1039,6 +1046,8 @@ bool SplitComparesTransform::runOnModule(Module &M) { char *bitw_env = getenv("LAF_SPLIT_COMPARES_BITW"); if (!bitw_env) bitw_env = getenv("AFL_LLVM_LAF_SPLIT_COMPARES_BITW"); if (bitw_env) { bitw = atoi(bitw_env); } + + enableFPSplit = getenv("AFL_LLVM_LAF_SPLIT_FLOATS") != NULL; simplifyCompares(M); @@ -1048,8 +1057,9 @@ bool SplitComparesTransform::runOnModule(Module &M) { errs() << "Split-compare-pass by laf.intel@gmail.com, extended by " "heiko@hexco.de\n"; - errs() << "Split-floatingpoint-compare-pass: " << splitFPCompares(M) - << " FP comparisons splitted\n"; + if (enableFPSplit) + errs() << "Split-floatingpoint-compare-pass: " << splitFPCompares(M) + << " FP comparisons splitted\n"; switch (bitw) { diff --git a/qemu_mode/README.md b/qemu_mode/README.md index 81904cf1..c88c1e41 100644 --- a/qemu_mode/README.md +++ b/qemu_mode/README.md @@ -100,8 +100,12 @@ afl-fuzz). AFL_COMPCOV_LEVEL=1 is to instrument comparisons with only immediate values / read-only memory. AFL_COMPCOV_LEVEL=2 instruments all comparison instructions and memory comparison functions when libcompcov -is preloaded. Comparison instructions are currently instrumented only -on the x86 and x86_64 targets. +is preloaded. +AFL_COMPCOV_LEVEL=3 has the same effects of AFL_COMPCOV_LEVEL=2 but enables also +the instrumentation of the floating-point comparisons on x86 and x86_64 (experimental). + +Integer comparison instructions are currently instrumented only +on the x86, x86_64 and ARM targets. Highly recommended. diff --git a/qemu_mode/build_qemu_support.sh b/qemu_mode/build_qemu_support.sh index 55d72e0d..c3983aa1 100755 --- a/qemu_mode/build_qemu_support.sh +++ b/qemu_mode/build_qemu_support.sh @@ -4,7 +4,7 @@ # -------------------------------------- # # Originally written by Andrew Griffiths <agriffiths@google.com> and -# Michal Zalewski <lcamtuf@google.com> +# Michal Zalewski # # TCG instrumentation and block chaining support by Andrea Biondo # <andrea.biondo965@gmail.com> @@ -100,7 +100,10 @@ if [ ! "$CKSUM" = "$QEMU_SHA384" ]; then echo "[*] Downloading QEMU ${VERSION} from the web..." rm -f "$ARCHIVE" - wget -O "$ARCHIVE" -- "$QEMU_URL" || exit 1 + OK= + while [ -z "$OK" ]; do + wget -c -O "$ARCHIVE" -- "$QEMU_URL" && OK=1 + done CKSUM=`sha384sum -- "$ARCHIVE" 2>/dev/null | cut -d' ' -f1` @@ -150,6 +153,9 @@ patch -p1 <../patches/translate-all.diff || exit 1 patch -p1 <../patches/tcg.diff || exit 1 patch -p1 <../patches/i386-translate.diff || exit 1 patch -p1 <../patches/arm-translate.diff || exit 1 +patch -p1 <../patches/i386-ops_sse.diff || exit 1 +patch -p1 <../patches/i386-fpu_helper.diff || exit 1 +patch -p1 <../patches/softfloat.diff || exit 1 echo "[+] Patching done." diff --git a/qemu_mode/patches/afl-qemu-common.h b/qemu_mode/patches/afl-qemu-common.h index c86b5b45..d4024353 100644 --- a/qemu_mode/patches/afl-qemu-common.h +++ b/qemu_mode/patches/afl-qemu-common.h @@ -3,7 +3,7 @@ ------------------------------------------------------------------- Originally written by Andrew Griffiths <agriffiths@google.com> and - Michal Zalewski <lcamtuf@google.com> + Michal Zalewski TCG instrumentation and block chaining support by Andrea Biondo <andrea.biondo965@gmail.com> @@ -31,6 +31,9 @@ */ +#ifndef __AFL_QEMU_COMMON +#define __AFL_QEMU_COMMON + #include "../../config.h" #ifndef CPU_NB_REGS @@ -81,3 +84,30 @@ void tcg_gen_afl_compcov_log_call(void *func, target_ulong cur_loc, void tcg_gen_afl_maybe_log_call(target_ulong cur_loc); +void afl_float_compcov_log_32(target_ulong cur_loc, float32 arg1, float32 arg2, + void *status); +void afl_float_compcov_log_64(target_ulong cur_loc, float64 arg1, float64 arg2, + void *status); +void afl_float_compcov_log_80(target_ulong cur_loc, floatx80 arg1, + floatx80 arg2); + +/* Check if an address is valid in the current mapping */ + +static inline int is_valid_addr(target_ulong addr) { + + int l, flags; + target_ulong page; + void * p; + + page = addr & TARGET_PAGE_MASK; + l = (page + TARGET_PAGE_SIZE) - addr; + + flags = page_get_flags(page); + if (!(flags & PAGE_VALID) || !(flags & PAGE_READ)) return 0; + + return 1; + +} + +#endif + diff --git a/qemu_mode/patches/afl-qemu-cpu-inl.h b/qemu_mode/patches/afl-qemu-cpu-inl.h index 262a7dab..f4c474d8 100644 --- a/qemu_mode/patches/afl-qemu-cpu-inl.h +++ b/qemu_mode/patches/afl-qemu-cpu-inl.h @@ -3,7 +3,7 @@ ------------------------------------------------------------------- Originally written by Andrew Griffiths <agriffiths@google.com> and - Michal Zalewski <lcamtuf@google.com> + Michal Zalewski TCG instrumentation and block chaining support by Andrea Biondo <andrea.biondo965@gmail.com> @@ -451,24 +451,6 @@ static void afl_request_tsl(target_ulong pc, target_ulong cb, uint32_t flags, } -/* Check if an address is valid in the current mapping */ - -static inline int is_valid_addr(target_ulong addr) { - - int l, flags; - target_ulong page; - void * p; - - page = addr & TARGET_PAGE_MASK; - l = (page + TARGET_PAGE_SIZE) - addr; - - flags = page_get_flags(page); - if (!(flags & PAGE_VALID) || !(flags & PAGE_READ)) return 0; - - return 1; - -} - /* This is the other side of the same channel. Since timeouts are handled by afl-fuzz simply killing the child, we can just wait until the pipe breaks. */ diff --git a/qemu_mode/patches/afl-qemu-cpu-translate-inl.h b/qemu_mode/patches/afl-qemu-cpu-translate-inl.h index b6d3da1f..62858724 100644 --- a/qemu_mode/patches/afl-qemu-cpu-translate-inl.h +++ b/qemu_mode/patches/afl-qemu-cpu-translate-inl.h @@ -3,7 +3,7 @@ ------------------------------------------------------------------- Originally written by Andrew Griffiths <agriffiths@google.com> and - Michal Zalewski <lcamtuf@google.com> + Michal Zalewski TCG instrumentation and block chaining support by Andrea Biondo <andrea.biondo965@gmail.com> diff --git a/qemu_mode/patches/afl-qemu-floats.h b/qemu_mode/patches/afl-qemu-floats.h new file mode 100644 index 00000000..0b2ac2ae --- /dev/null +++ b/qemu_mode/patches/afl-qemu-floats.h @@ -0,0 +1,217 @@ +/* + american fuzzy lop++ - high-performance binary-only instrumentation + ------------------------------------------------------------------- + + Originally written by Andrew Griffiths <agriffiths@google.com> and + Michal Zalewski + + TCG instrumentation and block chaining support by Andrea Biondo + <andrea.biondo965@gmail.com> + + QEMU 3.1.1 port, TCG thread-safety, CompareCoverage and NeverZero + counters by Andrea Fioraldi <andreafioraldi@gmail.com> + + Copyright 2015, 2016, 2017 Google Inc. All rights reserved. + Copyright 2019 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 a shim patched into the separately-distributed source + code of QEMU 3.1.0. It leverages the built-in QEMU tracing functionality + to implement AFL-style instrumentation and to take care of the remaining + parts of the AFL fork server logic. + + The resulting QEMU binary is essentially a standalone instrumentation + tool; for an example of how to leverage it for other purposes, you can + have a look at afl-showmap.c. + + */ + +#include "tcg.h" +#include "afl-qemu-common.h" + +union afl_float32 { + float32 f; + struct { + u64 sign : 1; + u64 exp : 7; + u64 frac : 24; + }; +}; + + +union afl_float64 { + float64 f; + struct { + u64 sign : 1; + u64 exp : 11; + u64 frac : 52; + }; +}; + + +// TODO 16 and 128 bits floats +// TODO figure out why float*_unpack_canonical does not work + +void afl_float_compcov_log_32(target_ulong cur_loc, float32 arg1, float32 arg2, + void* status) { + + cur_loc = (cur_loc >> 4) ^ (cur_loc << 8); + cur_loc &= MAP_SIZE - 7; + + if (cur_loc >= afl_inst_rms) return; + + //float_status* s = (float_status*)status; + //FloatParts a = float32_unpack_canonical(arg1, s); + //FloatParts b = float32_unpack_canonical(arg2, s); + union afl_float32 a = { .f = arg1 }; + union afl_float32 b = { .f = arg2 }; + + // if (is_nan(a.cls) || is_nan(b.cls)) return; + + register uintptr_t idx = cur_loc; + + if (a.sign != b.sign) return; + INC_AFL_AREA(idx); + if (a.exp != b.exp) return; + INC_AFL_AREA(idx + 1); + + if ((a.frac & 0xff0000) == (b.frac & 0xff0000)) { + + INC_AFL_AREA(idx + 2); + if ((a.frac & 0xff00) == (b.frac & 0xff00)) { INC_AFL_AREA(idx + 3); } + + } + +} + +void afl_float_compcov_log_64(target_ulong cur_loc, float64 arg1, float64 arg2, + void* status) { + + cur_loc = (cur_loc >> 4) ^ (cur_loc << 8); + cur_loc &= MAP_SIZE - 7; + + if (cur_loc >= afl_inst_rms) return; + + //float_status* s = (float_status*)status; + //FloatParts a = float64_unpack_canonical(arg1, s); + //FloatParts b = float64_unpack_canonical(arg2, s); + union afl_float64 a = { .f = arg1 }; + union afl_float64 b = { .f = arg2 }; + + // if (is_nan(a.cls) || is_nan(b.cls)) return; + + register uintptr_t idx = cur_loc; + + if (a.sign == b.sign) INC_AFL_AREA(idx); + if ((a.exp & 0xff00) == (b.exp & 0xff00)) { + + INC_AFL_AREA(idx + 1); + if ((a.exp & 0xff) == (b.exp & 0xff)) INC_AFL_AREA(idx + 2); + + } + + if ((a.frac & 0xff000000000000) == (b.frac & 0xff000000000000)) { + + INC_AFL_AREA(idx + 3); + if ((a.frac & 0xff0000000000) == (b.frac & 0xff0000000000)) { + + INC_AFL_AREA(idx + 4); + if ((a.frac & 0xff00000000) == (b.frac & 0xff00000000)) { + + INC_AFL_AREA(idx + 5); + if ((a.frac & 0xff000000) == (b.frac & 0xff000000)) { + + INC_AFL_AREA(idx + 6); + if ((a.frac & 0xff0000) == (b.frac & 0xff0000)) { + + INC_AFL_AREA(idx + 7); + if ((a.frac & 0xff00) == (b.frac & 0xff00)) INC_AFL_AREA(idx + 8); + + } + + } + + } + + } + + } + +} + +void afl_float_compcov_log_80(target_ulong cur_loc, floatx80 arg1, + floatx80 arg2) { + + cur_loc = (cur_loc >> 4) ^ (cur_loc << 8); + cur_loc &= MAP_SIZE - 7; + + if (cur_loc >= afl_inst_rms) return; + + if (floatx80_invalid_encoding(arg1) || floatx80_invalid_encoding(arg2)) + return; + + flag a_sign = extractFloatx80Sign(arg1); + flag b_sign = extractFloatx80Sign(arg2); + + /*if (((extractFloatx80Exp(arg1) == 0x7fff) && + (extractFloatx80Frac(arg1) << 1)) || + ((extractFloatx80Exp(arg2) == 0x7fff) && + (extractFloatx80Frac(arg2) << 1))) + return;*/ + + register uintptr_t idx = cur_loc; + + if (a_sign == b_sign) INC_AFL_AREA(idx); + + if ((arg1.high & 0x7f00) == (arg2.high & 0x7f00)) { + + INC_AFL_AREA(idx + 1); + if ((arg1.high & 0xff) == (arg2.high & 0xff)) INC_AFL_AREA(idx + 2); + + } + + if ((arg1.low & 0xff00000000000000) == (arg2.low & 0xff00000000000000)) { + + INC_AFL_AREA(idx + 3); + if ((arg1.low & 0xff000000000000) == (arg2.low & 0xff000000000000)) { + + INC_AFL_AREA(idx + 4); + if ((arg1.low & 0xff0000000000) == (arg2.low & 0xff0000000000)) { + + INC_AFL_AREA(idx + 5); + if ((arg1.low & 0xff00000000) == (arg2.low & 0xff00000000)) { + + INC_AFL_AREA(idx + 6); + if ((arg1.low & 0xff000000) == (arg2.low & 0xff000000)) { + + INC_AFL_AREA(idx + 7); + if ((arg1.low & 0xff0000) == (arg2.low & 0xff0000)) { + + INC_AFL_AREA(idx + 8); + if ((arg1.low & 0xff00) == (arg2.low & 0xff00)) { + + INC_AFL_AREA(idx + 9); + //if ((arg1.low & 0xff) == (arg2.low & 0xff)) + // INC_AFL_AREA(idx + 10); + + } + + } + + } + + } + + } + + } + + } + +} + diff --git a/qemu_mode/patches/afl-qemu-tcg-inl.h b/qemu_mode/patches/afl-qemu-tcg-inl.h index d7a25695..e3de09d8 100644 --- a/qemu_mode/patches/afl-qemu-tcg-inl.h +++ b/qemu_mode/patches/afl-qemu-tcg-inl.h @@ -3,7 +3,7 @@ ------------------------------------------------------------------- Originally written by Andrew Griffiths <agriffiths@google.com> and - Michal Zalewski <lcamtuf@google.com> + Michal Zalewski TCG instrumentation and block chaining support by Andrea Biondo <andrea.biondo965@gmail.com> diff --git a/qemu_mode/patches/afl-qemu-translate-inl.h b/qemu_mode/patches/afl-qemu-translate-inl.h index 5f61d7c9..881dbc8d 100644 --- a/qemu_mode/patches/afl-qemu-translate-inl.h +++ b/qemu_mode/patches/afl-qemu-translate-inl.h @@ -3,7 +3,7 @@ ------------------------------------------------------------------- Originally written by Andrew Griffiths <agriffiths@google.com> and - Michal Zalewski <lcamtuf@google.com> + Michal Zalewski TCG instrumentation and block chaining support by Andrea Biondo <andrea.biondo965@gmail.com> diff --git a/qemu_mode/patches/i386-fpu_helper.diff b/qemu_mode/patches/i386-fpu_helper.diff new file mode 100644 index 00000000..3bd09d9c --- /dev/null +++ b/qemu_mode/patches/i386-fpu_helper.diff @@ -0,0 +1,54 @@ +diff --git a/target/i386/fpu_helper.c b/target/i386/fpu_helper.c +index ea5a0c48..89901315 100644 +--- a/target/i386/fpu_helper.c ++++ b/target/i386/fpu_helper.c +@@ -384,10 +384,16 @@ void helper_fxchg_ST0_STN(CPUX86State *env, int st_index) + + static const int fcom_ccval[4] = {0x0100, 0x4000, 0x0000, 0x4500}; + ++#include "../patches/afl-qemu-common.h" ++ + void helper_fcom_ST0_FT0(CPUX86State *env) + { + int ret; + ++ if (afl_compcov_level > 2 && env->eip < afl_end_code && ++ env->eip >= afl_start_code) ++ afl_float_compcov_log_80(env->eip, ST0, FT0); ++ + ret = floatx80_compare(ST0, FT0, &env->fp_status); + env->fpus = (env->fpus & ~0x4500) | fcom_ccval[ret + 1]; + } +@@ -396,6 +402,10 @@ void helper_fucom_ST0_FT0(CPUX86State *env) + { + int ret; + ++ if (afl_compcov_level > 2 && env->eip < afl_end_code && ++ env->eip >= afl_start_code) ++ afl_float_compcov_log_80(env->eip, ST0, FT0); ++ + ret = floatx80_compare_quiet(ST0, FT0, &env->fp_status); + env->fpus = (env->fpus & ~0x4500) | fcom_ccval[ret + 1]; + } +@@ -407,6 +417,10 @@ void helper_fcomi_ST0_FT0(CPUX86State *env) + int eflags; + int ret; + ++ if (afl_compcov_level > 2 && env->eip < afl_end_code && ++ env->eip >= afl_start_code) ++ afl_float_compcov_log_80(env->eip, ST0, FT0); ++ + ret = floatx80_compare(ST0, FT0, &env->fp_status); + eflags = cpu_cc_compute_all(env, CC_OP); + eflags = (eflags & ~(CC_Z | CC_P | CC_C)) | fcomi_ccval[ret + 1]; +@@ -418,6 +432,10 @@ void helper_fucomi_ST0_FT0(CPUX86State *env) + int eflags; + int ret; + ++ if (afl_compcov_level > 2 && env->eip < afl_end_code && ++ env->eip >= afl_start_code) ++ afl_float_compcov_log_80(env->eip, ST0, FT0); ++ + ret = floatx80_compare_quiet(ST0, FT0, &env->fp_status); + eflags = cpu_cc_compute_all(env, CC_OP); + eflags = (eflags & ~(CC_Z | CC_P | CC_C)) | fcomi_ccval[ret + 1]; diff --git a/qemu_mode/patches/i386-ops_sse.diff b/qemu_mode/patches/i386-ops_sse.diff new file mode 100644 index 00000000..d2779ea8 --- /dev/null +++ b/qemu_mode/patches/i386-ops_sse.diff @@ -0,0 +1,61 @@ +diff --git a/target/i386/ops_sse.h b/target/i386/ops_sse.h +index ed059897..a5296caa 100644 +--- a/target/i386/ops_sse.h ++++ b/target/i386/ops_sse.h +@@ -997,6 +997,8 @@ SSE_HELPER_CMP(cmpord, FPU_CMPORD) + + static const int comis_eflags[4] = {CC_C, CC_Z, 0, CC_Z | CC_P | CC_C}; + ++#include "../patches/afl-qemu-common.h" ++ + void helper_ucomiss(CPUX86State *env, Reg *d, Reg *s) + { + int ret; +@@ -1004,6 +1006,11 @@ void helper_ucomiss(CPUX86State *env, Reg *d, Reg *s) + + s0 = d->ZMM_S(0); + s1 = s->ZMM_S(0); ++ ++ if (afl_compcov_level > 2 && env->eip < afl_end_code && ++ env->eip >= afl_start_code) ++ afl_float_compcov_log_32(env->eip, s0, s1, &env->sse_status); ++ + ret = float32_compare_quiet(s0, s1, &env->sse_status); + CC_SRC = comis_eflags[ret + 1]; + } +@@ -1015,6 +1022,11 @@ void helper_comiss(CPUX86State *env, Reg *d, Reg *s) + + s0 = d->ZMM_S(0); + s1 = s->ZMM_S(0); ++ ++ if (afl_compcov_level > 2 && env->eip < afl_end_code && ++ env->eip >= afl_start_code) ++ afl_float_compcov_log_32(env->eip, s0, s1, &env->sse_status); ++ + ret = float32_compare(s0, s1, &env->sse_status); + CC_SRC = comis_eflags[ret + 1]; + } +@@ -1026,6 +1038,11 @@ void helper_ucomisd(CPUX86State *env, Reg *d, Reg *s) + + d0 = d->ZMM_D(0); + d1 = s->ZMM_D(0); ++ ++ if (afl_compcov_level > 2 && env->eip < afl_end_code && ++ env->eip >= afl_start_code) ++ afl_float_compcov_log_64(env->eip, d0, d1, &env->sse_status); ++ + ret = float64_compare_quiet(d0, d1, &env->sse_status); + CC_SRC = comis_eflags[ret + 1]; + } +@@ -1037,6 +1054,11 @@ void helper_comisd(CPUX86State *env, Reg *d, Reg *s) + + d0 = d->ZMM_D(0); + d1 = s->ZMM_D(0); ++ ++ if (afl_compcov_level > 2 && env->eip < afl_end_code && ++ env->eip >= afl_start_code) ++ afl_float_compcov_log_64(env->eip, d0, d1, &env->sse_status); ++ + ret = float64_compare(d0, d1, &env->sse_status); + CC_SRC = comis_eflags[ret + 1]; + } diff --git a/qemu_mode/patches/softfloat.diff b/qemu_mode/patches/softfloat.diff new file mode 100644 index 00000000..86ffb97f --- /dev/null +++ b/qemu_mode/patches/softfloat.diff @@ -0,0 +1,10 @@ +diff --git a/fpu/softfloat.c b/fpu/softfloat.c +index e1eef954..2f8d0d62 100644 +--- a/fpu/softfloat.c ++++ b/fpu/softfloat.c +@@ -7205,3 +7205,5 @@ float128 float128_scalbn(float128 a, int n, float_status *status) + , status); + + } ++ ++#include "../../patches/afl-qemu-floats.h" diff --git a/src/afl-analyze.c b/src/afl-analyze.c index 2d13621b..ee281af8 100644 --- a/src/afl-analyze.c +++ b/src/afl-analyze.c @@ -2,7 +2,7 @@ american fuzzy lop++ - file format analyzer ------------------------------------------- - Originally written by Michal Zalewski <lcamtuf@google.com> + Originally written by Michal Zalewski Now maintained by by Marc Heuse <mh@mh-sec.de>, Heiko Eißfeldt <heiko.eissfeldt@hexco.de> and @@ -841,7 +841,7 @@ int main(int argc, char** argv) { doc_path = access(DOC_PATH, F_OK) ? "docs" : DOC_PATH; - SAYF(cCYA "afl-analyze" VERSION cRST " by <lcamtuf@google.com>\n"); + SAYF(cCYA "afl-analyze" VERSION cRST " by Michal Zalewski\n"); while ((opt = getopt(argc, argv, "+i:f:m:t:eQUWh")) > 0) @@ -987,7 +987,7 @@ int main(int argc, char** argv) { if (child_timed_out) FATAL("Target binary times out (adjusting -t may help)."); - if (!anything_set()) FATAL("No instrumentation detected."); + if (getenv("AFL_SKIP_BIN_CHECK") == NULL && !anything_set()) FATAL("No instrumentation detected."); analyze(use_argv); diff --git a/src/afl-as.c b/src/afl-as.c index b5a5ed58..312ae0a7 100644 --- a/src/afl-as.c +++ b/src/afl-as.c @@ -2,7 +2,7 @@ american fuzzy lop++ - wrapper for GNU as ----------------------------------------- - Originally written by Michal Zalewski <lcamtuf@google.com> + Originally written by Michal Zalewski Now maintained by by Marc Heuse <mh@mh-sec.de>, Heiko Eißfeldt <heiko.eissfeldt@hexco.de> and @@ -499,9 +499,9 @@ int main(int argc, char** argv) { clang_mode = !!getenv(CLANG_ENV_VAR); - if (isatty(2) && !getenv("AFL_QUIET")) { + if (!getenv("AFL_QUIET")) { - SAYF(cCYA "afl-as" VERSION cRST " by <lcamtuf@google.com>\n"); + SAYF(cCYA "afl-as" VERSION cRST " by Michal Zalewski\n"); } else diff --git a/src/afl-common.c b/src/afl-common.c index e753e797..ec010c2d 100644 --- a/src/afl-common.c +++ b/src/afl-common.c @@ -2,7 +2,7 @@ american fuzzy lop++ - common routines -------------------------------------- - Originally written by Michal Zalewski <lcamtuf@google.com> + Originally written by Michal Zalewski Now maintained by by Marc Heuse <mh@mh-sec.de>, Heiko Eißfeldt <heiko.eissfeldt@hexco.de> and diff --git a/src/afl-forkserver.c b/src/afl-forkserver.c index d9f67da5..b11ee5ce 100644 --- a/src/afl-forkserver.c +++ b/src/afl-forkserver.c @@ -2,7 +2,7 @@ american fuzzy lop++ - forkserver code -------------------------------------- - Originally written by Michal Zalewski <lcamtuf@google.com> + Originally written by Michal Zalewski Forkserver design by Jann Horn <jannhorn@googlemail.com> diff --git a/src/afl-fuzz-bitmap.c b/src/afl-fuzz-bitmap.c index 746fc982..22876626 100644 --- a/src/afl-fuzz-bitmap.c +++ b/src/afl-fuzz-bitmap.c @@ -2,7 +2,7 @@ american fuzzy lop++ - bitmap related routines ---------------------------------------------- - Originally written by Michal Zalewski <lcamtuf@google.com> + Originally written by Michal Zalewski Now maintained by by Marc Heuse <mh@mh-sec.de>, Heiko Eißfeldt <heiko.eissfeldt@hexco.de> and @@ -683,10 +683,12 @@ u8 save_if_interesting(char** argv, void* mem, u32 len, u8 fault) { #endif /* ^!SIMPLE_FILES */ ++unique_crashes; - - if (infoexec) // if the user wants to be informed on new crashes - do that + + if (infoexec) // if the user wants to be informed on new crashes - do + // that if (system(infoexec) == -1) - hnb += 0; // we dont care if system errors, but we dont want a compiler warning either + hnb += 0; // we dont care if system errors, but we dont want a + // compiler warning either last_crash_time = get_cur_time(); last_crash_execs = total_execs; diff --git a/src/afl-fuzz-extras.c b/src/afl-fuzz-extras.c index 1a0e2eff..23b2c235 100644 --- a/src/afl-fuzz-extras.c +++ b/src/afl-fuzz-extras.c @@ -2,7 +2,7 @@ american fuzzy lop++ - extras relates routines ---------------------------------------------- - Originally written by Michal Zalewski <lcamtuf@google.com> + Originally written by Michal Zalewski Now maintained by by Marc Heuse <mh@mh-sec.de>, Heiko Eißfeldt <heiko.eissfeldt@hexco.de> and diff --git a/src/afl-fuzz-globals.c b/src/afl-fuzz-globals.c index da134807..50b6b802 100644 --- a/src/afl-fuzz-globals.c +++ b/src/afl-fuzz-globals.c @@ -2,7 +2,7 @@ american fuzzy lop++ - globals declarations ------------------------------------------- - Originally written by Michal Zalewski <lcamtuf@google.com> + Originally written by Michal Zalewski Now maintained by by Marc Heuse <mh@mh-sec.de>, Heiko Eißfeldt <heiko.eissfeldt@hexco.de> and @@ -74,8 +74,8 @@ u8 *in_dir, /* Input directory with test cases */ *file_extension, /* File extension */ *orig_cmdline; /* Original command line */ u8 *doc_path, /* Path to documentation dir */ - *infoexec, /* Command to execute on a new crash */ - *out_file; /* File to fuzz, if any */ + *infoexec, /* Command to execute on a new crash */ + *out_file; /* File to fuzz, if any */ u32 exec_tmout = EXEC_TIMEOUT; /* Configurable exec timeout (ms) */ u32 hang_tmout = EXEC_TIMEOUT; /* Timeout used for hang det (ms) */ @@ -125,7 +125,8 @@ u8 skip_deterministic, /* Skip deterministic stages? */ deferred_mode, /* Deferred forkserver mode? */ fixed_seed, /* do not reseed */ fast_cal, /* Try to calibrate faster? */ - uses_asan; /* Target uses ASAN? */ + uses_asan, /* Target uses ASAN? */ + disable_trim; /* Never trim in fuzz_one */ s32 out_fd, /* Persistent fd for out_file */ #ifndef HAVE_ARC4RANDOM diff --git a/src/afl-fuzz-init.c b/src/afl-fuzz-init.c index 127f5217..82ba6f8a 100644 --- a/src/afl-fuzz-init.c +++ b/src/afl-fuzz-init.c @@ -2,7 +2,7 @@ american fuzzy lop++ - initialization related routines ------------------------------------------------------ - Originally written by Michal Zalewski <lcamtuf@google.com> + Originally written by Michal Zalewski Now maintained by by Marc Heuse <mh@mh-sec.de>, Heiko Eißfeldt <heiko.eissfeldt@hexco.de> and @@ -32,7 +32,7 @@ void bind_to_free_cpu(void) { -#if defined(__linux__) || defined(__FreeBSD__) +#if defined(__linux__) || defined(__FreeBSD__) || defined(__DragonFly__) cpu_set_t c; #elif defined(__NetBSD__) cpuset_t* c; @@ -117,7 +117,7 @@ void bind_to_free_cpu(void) { } closedir(d); -#elif defined(__FreeBSD__) +#elif defined(__FreeBSD__) || defined(__DragonFly__) struct kinfo_proc* procs; size_t nprocs; size_t proccount; @@ -138,7 +138,13 @@ void bind_to_free_cpu(void) { for (i = 0; i < proccount; i++) { - if (procs[i].ki_oncpu < sizeof(cpu_used)) cpu_used[procs[i].ki_oncpu] = 1; +#if defined(__FreeBSD__) + if (procs[i].ki_oncpu < sizeof(cpu_used) && procs[i].ki_pctcpu > 10) + cpu_used[procs[i].ki_oncpu] = 1; +#elif defined(__DragonFly__) + if (procs[i].kp_lwp.kl_cpuid < sizeof(cpu_used) && procs[i].kp_lwp.kl_pctcpu > 10) + cpu_used[procs[i].kp_lwp.kl_cpuid] = 1; +#endif } @@ -166,7 +172,8 @@ void bind_to_free_cpu(void) { for (i = 0; i < proccount; i++) { - if (procs[i].p_cpuid < sizeof(cpu_used)) cpu_used[procs[i].p_cpuid] = 1; + if (procs[i].p_cpuid < sizeof(cpu_used) && procs[i].p_pctcpu > 0) + cpu_used[procs[i].p_cpuid] = 1; } @@ -198,7 +205,7 @@ void bind_to_free_cpu(void) { cpu_aff = i; -#if defined(__linux__) || defined(__FreeBSD__) +#if defined(__linux__) || defined(__FreeBSD__) || defined(__DragonFly__) CPU_ZERO(&c); CPU_SET(i, &c); #elif defined(__NetBSD__) @@ -210,7 +217,7 @@ void bind_to_free_cpu(void) { #if defined(__linux__) if (sched_setaffinity(0, sizeof(c), &c)) PFATAL("sched_setaffinity failed"); -#elif defined(__FreeBSD__) +#elif defined(__FreeBSD__) || defined(__DragonFly__) if (pthread_setaffinity_np(pthread_self(), sizeof(c), &c)) PFATAL("pthread_setaffinity failed"); #elif defined(__NetBSD__) @@ -727,7 +734,7 @@ void pivot_inputs(void) { use_name += 6; else use_name = rsl; - nfn = alloc_printf("%s/queue/id:%06u,orig:%s", out_dir, id, use_name); + nfn = alloc_printf("%s/queue/id:%06u,time:0,orig:%s", out_dir, id, use_name); #else @@ -869,7 +876,7 @@ double get_runnable_processes(void) { static double res; #if defined(__APPLE__) || defined(__FreeBSD__) || defined(__OpenBSD__) || \ - defined(__NetBSD__) + defined(__NetBSD__) || defined(__DragonFly__) /* I don't see any portable sysctl or so that would quickly give us the number of runnable processes; the 1-minute load average can be a @@ -1603,7 +1610,7 @@ void check_cpu_governor(void) { void get_core_count(void) { -#if defined(__APPLE__) || defined(__FreeBSD__) || defined(__OpenBSD__) +#if defined(__APPLE__) || defined(__FreeBSD__) || defined(__OpenBSD__) || defined(__DragonFly__) size_t s = sizeof(cpu_core_count); @@ -1649,7 +1656,7 @@ void get_core_count(void) { cur_runnable = (u32)get_runnable_processes(); -#if defined(__APPLE__) || defined(__FreeBSD__) || defined(__OpenBSD__) +#if defined(__APPLE__) || defined(__FreeBSD__) || defined(__OpenBSD__) || defined(__DragonFly__) /* Add ourselves, since the 1-minute average doesn't include that yet. */ diff --git a/src/afl-fuzz-misc.c b/src/afl-fuzz-misc.c index a7372b7d..f45642f4 100644 --- a/src/afl-fuzz-misc.c +++ b/src/afl-fuzz-misc.c @@ -2,7 +2,7 @@ american fuzzy lop++ - misc stuffs from Mordor ---------------------------------------------- - Originally written by Michal Zalewski <lcamtuf@google.com> + Originally written by Michal Zalewski Now maintained by by Marc Heuse <mh@mh-sec.de>, Heiko Eißfeldt <heiko.eissfeldt@hexco.de> and diff --git a/src/afl-fuzz-one.c b/src/afl-fuzz-one.c index c02dbeb7..6ab0266d 100644 --- a/src/afl-fuzz-one.c +++ b/src/afl-fuzz-one.c @@ -2,7 +2,7 @@ american fuzzy lop++ - fuzze_one routines in different flavours --------------------------------------------------------------- - Originally written by Michal Zalewski <lcamtuf@google.com> + Originally written by Michal Zalewski Now maintained by by Marc Heuse <mh@mh-sec.de>, Heiko Eißfeldt <heiko.eissfeldt@hexco.de> and @@ -449,7 +449,7 @@ u8 fuzz_one_original(char** argv) { * TRIMMING * ************/ - if (!dumb_mode && !queue_cur->trim_done && !custom_mutator) { + if (!dumb_mode && !queue_cur->trim_done && !custom_mutator && !disable_trim) { u8 res = trim_case(argv, queue_cur, in_buf); diff --git a/src/afl-fuzz-python.c b/src/afl-fuzz-python.c index 1a28f603..1a8b7f9d 100644 --- a/src/afl-fuzz-python.c +++ b/src/afl-fuzz-python.c @@ -2,7 +2,7 @@ american fuzzy lop++ - python extension routines ------------------------------------------------ - Originally written by Michal Zalewski <lcamtuf@google.com> + Originally written by Michal Zalewski Now maintained by by Marc Heuse <mh@mh-sec.de>, Heiko Eißfeldt <heiko.eissfeldt@hexco.de> and diff --git a/src/afl-fuzz-queue.c b/src/afl-fuzz-queue.c index 905fd931..9f036186 100644 --- a/src/afl-fuzz-queue.c +++ b/src/afl-fuzz-queue.c @@ -2,7 +2,7 @@ american fuzzy lop++ - queue relates routines --------------------------------------------- - Originally written by Michal Zalewski <lcamtuf@google.com> + Originally written by Michal Zalewski Now maintained by by Marc Heuse <mh@mh-sec.de>, Heiko Eißfeldt <heiko.eissfeldt@hexco.de> and diff --git a/src/afl-fuzz-run.c b/src/afl-fuzz-run.c index 46d12423..e12b06eb 100644 --- a/src/afl-fuzz-run.c +++ b/src/afl-fuzz-run.c @@ -2,7 +2,7 @@ american fuzzy lop++ - target execution related routines -------------------------------------------------------- - Originally written by Michal Zalewski <lcamtuf@google.com> + Originally written by Michal Zalewski Now maintained by by Marc Heuse <mh@mh-sec.de>, Heiko Eißfeldt <heiko.eissfeldt@hexco.de> and @@ -288,10 +288,9 @@ void write_to_testcase(void* mem, u32 len) { if (out_file) { - // unlink(out_file); /* Ignore errors. - // */ + unlink(out_file); /* Ignore errors. */ - fd = open(out_file, O_WRONLY | O_CREAT | O_TRUNC, 0600); + fd = open(out_file, O_WRONLY | O_CREAT | O_EXCL, 0600); if (fd < 0) PFATAL("Unable to create '%s'", out_file); @@ -331,10 +330,9 @@ void write_with_gap(void* mem, u32 len, u32 skip_at, u32 skip_len) { if (out_file) { - // unlink(out_file); /* Ignore errors. - // */ + unlink(out_file); /* Ignore errors. */ - fd = open(out_file, O_WRONLY | O_CREAT | O_TRUNC, 0600); + fd = open(out_file, O_WRONLY | O_CREAT | O_EXCL, 0600); if (fd < 0) PFATAL("Unable to create '%s'", out_file); diff --git a/src/afl-fuzz-stats.c b/src/afl-fuzz-stats.c index 7f749511..851cfb1c 100644 --- a/src/afl-fuzz-stats.c +++ b/src/afl-fuzz-stats.c @@ -2,7 +2,7 @@ american fuzzy lop++ - stats related routines --------------------------------------------- - Originally written by Michal Zalewski <lcamtuf@google.com> + Originally written by Michal Zalewski Now maintained by by Marc Heuse <mh@mh-sec.de>, Heiko Eißfeldt <heiko.eissfeldt@hexco.de> and @@ -334,9 +334,9 @@ void show_stats(void) { /* Lord, forgive me this. */ - SAYF(SET_G1 bSTG bLT bH bSTOP cCYA + SAYF(SET_G1 bSTG bLT bH bSTOP cCYA " process timing " bSTG bH30 bH5 bH bHB bH bSTOP cCYA - " overall results " bSTG bH2 bH2 bRT "\n"); + " overall results " bSTG bH2 bH2 bRT "\n"); if (dumb_mode) { @@ -413,9 +413,9 @@ void show_stats(void) { " uniq hangs : " cRST "%-6s" bSTG bV "\n", DTD(cur_ms, last_hang_time), tmp); - SAYF(bVR bH bSTOP cCYA + SAYF(bVR bH bSTOP cCYA " cycle progress " bSTG bH10 bH5 bH2 bH2 bHB bH bSTOP cCYA - " map coverage " bSTG bH bHT bH20 bH2 bVL "\n"); + " map coverage " bSTG bH bHT bH20 bH2 bVL "\n"); /* This gets funny because we want to print several variable-length variables together, but then cram them into a fixed-width field - so we need to @@ -443,9 +443,9 @@ void show_stats(void) { SAYF(bSTOP " count coverage : " cRST "%-21s" bSTG bV "\n", tmp); - SAYF(bVR bH bSTOP cCYA + SAYF(bVR bH bSTOP cCYA " stage progress " bSTG bH10 bH5 bH2 bH2 bX bH bSTOP cCYA - " findings in depth " bSTG bH10 bH5 bH2 bH2 bVL "\n"); + " findings in depth " bSTG bH10 bH5 bH2 bH2 bVL "\n"); sprintf(tmp, "%s (%0.02f%%)", DI(queued_favored), ((double)queued_favored) * 100 / queued_paths); @@ -514,7 +514,7 @@ void show_stats(void) { /* Aaaalmost there... hold on! */ - SAYF(bVR bH cCYA bSTOP + SAYF(bVR bH cCYA bSTOP " fuzzing strategy yields " bSTG bH10 bHT bH10 bH5 bHB bH bSTOP cCYA " path geometry " bSTG bH5 bH2 bVL "\n"); @@ -634,13 +634,13 @@ void show_stats(void) { sprintf(tmp, "%s/%s", DI(stage_finds[STAGE_CUSTOM_MUTATOR]), DI(stage_cycles[STAGE_CUSTOM_MUTATOR])); SAYF(bV bSTOP " custom mut. : " cRST "%-36s " bSTG bVR bH20 bH2 bH bRB - "\n" bLB bH30 bH20 bH2 bH bRB bSTOP cRST RESET_G1, + "\n" bLB bH30 bH20 bH2 bH bRB bSTOP cRST RESET_G1, tmp); } else { SAYF(bV bSTOP " trim : " cRST "%-36s " bSTG bVR bH20 bH2 bH bRB - "\n" bLB bH30 bH20 bH2 bRB bSTOP cRST RESET_G1, + "\n" bLB bH30 bH20 bH2 bRB bSTOP cRST RESET_G1, tmp); } diff --git a/src/afl-fuzz.c b/src/afl-fuzz.c index a9a576fe..eed4992c 100644 --- a/src/afl-fuzz.c +++ b/src/afl-fuzz.c @@ -2,7 +2,7 @@ american fuzzy lop - fuzzer code -------------------------------- - Originally written by Michal Zalewski <lcamtuf@google.com> + Originally written by Michal Zalewski Now maintained by by Marc Heuse <mh@mh-sec.de>, Heiko Eißfeldt <heiko.eissfeldt@hexco.de> and @@ -126,12 +126,15 @@ static void usage(u8* argv0) { " -V seconds - fuzz for a maximum total time of seconds then " "terminate\n" " -E execs - fuzz for a maximum number of total executions then " - "terminate\n\n" + "terminate\n" + " Note: -V/-E are not precise, they are checked after a queue entry " + "is done\n which can be many minutes/execs later\n\n" "Other stuff:\n" " -T text - text banner to show on the screen\n" " -M / -S id - distributed mode (see parallel_fuzzing.txt)\n" - " -I command - execute this command/script when a new crash is found\n" + " -I command - execute this command/script when a new crash is " + "found\n" " -B bitmap.txt - mutate a specific test case, use the out/fuzz_bitmap " "file\n" " -C - crash exploration mode (the peruvian rabbit thing)\n" @@ -180,7 +183,7 @@ int main(int argc, char** argv) { SAYF(cCYA "afl-fuzz" VERSION cRST - " based on afl by <lcamtuf@google.com> and a big online community\n"); + " based on afl by Michal Zalewski and a big online community\n"); doc_path = access(DOC_PATH, F_OK) ? "docs" : DOC_PATH; @@ -192,10 +195,7 @@ int main(int argc, char** argv) { switch (opt) { - case 'I': - - infoexec = optarg; - break; + case 'I': infoexec = optarg; break; case 's': { @@ -645,10 +645,13 @@ int main(int argc, char** argv) { } + if (getenv("AFL_DISABLE_TRIM")) + disable_trim = 1; + if (getenv("AFL_NO_UI") && getenv("AFL_FORCE_UI")) FATAL("AFL_NO_UI and AFL_FORCE_UI are mutually exclusive"); - if (strchr(argv[optind], '/') == NULL) + if (strchr(argv[optind], '/') == NULL && !unicorn_mode) WARNF(cLRD "Target binary called without a prefixed path, make sure you are " "fuzzing the right binary: " cRST "%s", diff --git a/src/afl-gcc.c b/src/afl-gcc.c index 8982ca97..e0706a5f 100644 --- a/src/afl-gcc.c +++ b/src/afl-gcc.c @@ -2,7 +2,7 @@ american fuzzy lop++ - wrapper for GCC and clang ------------------------------------------------ - Originally written by Michal Zalewski <lcamtuf@google.com> + Originally written by Michal Zalewski Now maintained by by Marc Heuse <mh@mh-sec.de>, Heiko Eißfeldt <heiko.eissfeldt@hexco.de> and @@ -131,7 +131,7 @@ static void edit_params(u32 argc, char** argv) { if (!name) name = argv[0]; else - name++; + ++name; if (!strncmp(name, "afl-clang", 9)) { @@ -320,6 +320,7 @@ static void edit_params(u32 argc, char** argv) { 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"; @@ -335,7 +336,7 @@ int main(int argc, char** argv) { if (argc == 2 && strcmp(argv[1], "-h") == 0) { - printf("afl-cc" VERSION " by <lcamtuf@google.com>\n\n"); + printf("afl-cc" VERSION " by Michal Zalewski\n\n"); printf("%s \n\n", argv[0]); printf("afl-gcc has no command line options\n"); printf( @@ -347,7 +348,7 @@ int main(int argc, char** argv) { if (isatty(2) && !getenv("AFL_QUIET")) { - SAYF(cCYA "afl-cc" VERSION cRST " by <lcamtuf@google.com>\n"); + 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"); diff --git a/src/afl-gotcpu.c b/src/afl-gotcpu.c index bdb727de..fe5d035f 100644 --- a/src/afl-gotcpu.c +++ b/src/afl-gotcpu.c @@ -2,7 +2,7 @@ american fuzzy lop - free CPU gizmo ----------------------------------- - Originally written by Michal Zalewski <lcamtuf@google.com> + Originally written by Michal Zalewski Now maintained by by Marc Heuse <mh@mh-sec.de>, Heiko Eißfeldt <heiko.eissfeldt@hexco.de> and @@ -52,18 +52,24 @@ #include "types.h" #include "debug.h" -#if defined(__linux__) || defined(__FreeBSD__) || defined(__NetBSD__) +#if defined(__linux__) || defined(__FreeBSD__) || defined(__NetBSD__) || defined(__APPLE__) || defined(__DragonFly__) #define HAVE_AFFINITY 1 -#if defined(__FreeBSD__) +#if defined(__FreeBSD__) || defined(__DragonFly__) #include <pthread.h> #include <pthread_np.h> +#if defined(__FreeBSD__) #include <sys/cpuset.h> +#endif #define cpu_set_t cpuset_t #elif defined(__NetBSD__) #include <pthread.h> #include <sched.h> +#elif defined(__APPLE__) +#include <pthread.h> +#include <mach/thread_act.h> +#include <mach/thread_policy.h> #endif -#endif /* __linux__ || __FreeBSD__ || __NetBSD__ */ +#endif /* __linux__ || __FreeBSD__ || __NetBSD__ || __APPLE__ */ /* Get unix time in microseconds. */ @@ -138,7 +144,7 @@ int main(int argc, char** argv) { if (argc > 1) { - printf("afl-gotcpu" VERSION " by <lcamtuf@google.com>\n"); + printf("afl-gotcpu" VERSION " by Michal Zalewski\n"); printf("\n%s \n\n", argv[0]); printf("afl-gotcpu does not have command line options\n"); printf("afl-gotcpu prints out which CPUs are available\n"); @@ -150,7 +156,7 @@ int main(int argc, char** argv) { u32 cpu_cnt = sysconf(_SC_NPROCESSORS_ONLN), idle_cpus = 0, maybe_cpus = 0, i; - SAYF(cCYA "afl-gotcpu" VERSION cRST " by <lcamtuf@google.com>\n"); + SAYF(cCYA "afl-gotcpu" VERSION cRST " by Michal Zalewski\n"); ACTF("Measuring per-core preemption rate (this will take %0.02f sec)...", ((double)CTEST_CORE_TRG_MS) / 1000); @@ -164,7 +170,7 @@ int main(int argc, char** argv) { if (!fr) { u32 util_perc; -#if defined(__linux__) || defined(__FreeBSD__) +#if defined(__linux__) || defined(__FreeBSD__) || defined(__DragonFly__) cpu_set_t c; CPU_ZERO(&c); @@ -176,9 +182,15 @@ int main(int argc, char** argv) { if (c == NULL) PFATAL("cpuset_create failed"); cpuset_set(i, c); +#elif defined(__APPLE__) + thread_affinity_policy_data_t c = { i }; + thread_port_t native_thread = pthread_mach_thread_np(pthread_self()); + if (thread_policy_set(native_thread, THREAD_AFFINITY_POLICY, + (thread_policy_t)&c, 1) != KERN_SUCCESS) + PFATAL("thread_policy_set failed"); #endif -#if defined(__FreeBSD__) +#if defined(__FreeBSD__) || defined(__DragonFly__) if (pthread_setaffinity_np(pthread_self(), sizeof(c), &c)) PFATAL("pthread_setaffinity_np failed"); #endif @@ -265,7 +277,7 @@ int main(int argc, char** argv) { u32 util_perc; - SAYF(cCYA "afl-gotcpu" VERSION cRST " by <lcamtuf@google.com>\n"); + SAYF(cCYA "afl-gotcpu" VERSION cRST " by Michal Zalewski\n"); /* Run a busy loop for CTEST_TARGET_MS. */ diff --git a/src/afl-sharedmem.c b/src/afl-sharedmem.c index 0bd1ff2f..e11221a4 100644 --- a/src/afl-sharedmem.c +++ b/src/afl-sharedmem.c @@ -2,7 +2,7 @@ american fuzzy lop++ - shared memory related code ------------------------------------------------- - Originally written by Michal Zalewski <lcamtuf@google.com> + Originally written by Michal Zalewski Forkserver design by Jann Horn <jannhorn@googlemail.com> @@ -153,8 +153,6 @@ void setup_shm(unsigned char dumb_mode) { shm_str = alloc_printf("%d", shm_id); - setenv(SHM_ENV_VAR, shm_str, 1); - /* If somebody is asking us to fuzz instrumented binaries in dumb mode, we don't want them to detect instrumentation, since we won't be sending fork server commands. This should be replaced with better auto-detection diff --git a/src/afl-showmap.c b/src/afl-showmap.c index 31f12856..b54ac2b0 100644 --- a/src/afl-showmap.c +++ b/src/afl-showmap.c @@ -2,7 +2,7 @@ american fuzzy lop++ - map display utility ------------------------------------------ - Originally written by Michal Zalewski <lcamtuf@google.com> + Originally written by Michal Zalewski Forkserver design by Jann Horn <jannhorn@googlemail.com> @@ -395,7 +395,7 @@ static void setup_signal_handlers(void) { static void show_banner(void) { - SAYF(cCYA "afl-showmap" VERSION cRST " by <lcamtuf@google.com>\n"); + SAYF(cCYA "afl-showmap" VERSION cRST " by Michal Zalewski\n"); } diff --git a/src/afl-tmin.c b/src/afl-tmin.c index 7f354727..a72e1dda 100644 --- a/src/afl-tmin.c +++ b/src/afl-tmin.c @@ -2,7 +2,7 @@ american fuzzy lop++ - test case minimizer ------------------------------------------ - Originally written by Michal Zalewski <lcamtuf@google.com> + Originally written by Michal Zalewski Forkserver design by Jann Horn <jannhorn@googlemail.com> @@ -1032,7 +1032,7 @@ int main(int argc, char** argv) { doc_path = access(DOC_PATH, F_OK) ? "docs" : DOC_PATH; - SAYF(cCYA "afl-tmin" VERSION cRST " by <lcamtuf@google.com>\n"); + SAYF(cCYA "afl-tmin" VERSION cRST " by Michal Zalewski\n"); while ((opt = getopt(argc, argv, "+i:o:f:m:t:B:xeQUWh")) > 0) diff --git a/test-instr.c b/test-instr.c index 71838462..161bdb8e 100644 --- a/test-instr.c +++ b/test-instr.c @@ -2,7 +2,7 @@ american fuzzy lop - a trivial program to test the build -------------------------------------------------------- - Written and maintained by Michal Zalewski <lcamtuf@google.com> + Written by Michal Zalewski Copyright 2014 Google Inc. All rights reserved. @@ -23,15 +23,20 @@ int main(int argc, char** argv) { char buff[8]; char* buf = buff; - if (argc > 1) + // we support command line parameter and stdin + if (argc > 1) { + buf = argv[1]; - else if (read(0, buf, sizeof(buf)) < 1) { + + } else if (read(0, buf, sizeof(buf)) < 1) { printf("Hum?\n"); - exit(1); + return 1; } + // we support three input cases (plus a 4th if stdin is used but there is no + // input) if (buf[0] == '0') printf("Looks like a zero to me!\n"); else if (buf[0] == '1') @@ -39,7 +44,7 @@ int main(int argc, char** argv) { else printf("Neither one or zero? How quaint!\n"); - exit(0); + return 0; } diff --git a/test/test-performance.sh b/test/test-performance.sh new file mode 100755 index 00000000..87eea665 --- /dev/null +++ b/test/test-performance.sh @@ -0,0 +1,194 @@ +#!/bin/bash + +# if you want a specific performance file (e.g. to compare features to another) +# you can set the AFL_PERFORMANCE_FILE environment variable: +FILE=$AFL_PERFORMANCE_FILE +# otherwise we use ~/.afl_performance +test -z "$FILE" && FILE=~/.afl_performance + +test -e $FILE || { + echo Warning: This script measure the performance of afl++ and saves the result for future comparisons into $FILE + echo Press ENTER to continue or CONTROL-C to abort + read IN +} + +export AFL_QUIET=1 +unset AFL_EXIT_WHEN_DONE +unset AFL_SKIP_CPUFREQ +unset AFL_DEBUG +unset AFL_HARDEN +unset AFL_USE_ASAN +unset AFL_USE_MSAN +unset AFL_CC +unset AFL_PRELOAD +unset AFL_GCC_WHITELIST +unset AFL_LLVM_WHITELIST +unset AFL_LLVM_INSTRIM +unset AFL_LLVM_LAF_SPLIT_SWITCHES +unset AFL_LLVM_LAF_TRANSFORM_COMPARES +unset AFL_LLVM_LAF_SPLIT_COMPARES + +# on OpenBSD we need to work with llvm from /usr/local/bin +test -e /usr/local/bin/opt && { + export PATH=/usr/local/bin:${PATH} +} +# on MacOS X we prefer afl-clang over afl-gcc, because +# afl-gcc does not work there +test `uname -s` = 'Darwin' -o `uname -s` = 'FreeBSD' && { + AFL_GCC=afl-clang +} || { + AFL_GCC=afl-gcc +} + +ECHO="printf %b\\n" +$ECHO \\101 2>&1 | grep -qE '^A' || { + ECHO= + test -e /bin/printf && { + ECHO="/bin/printf %b\\n" + $ECHO '\\101' 2>&1 | grep -qE '^A' || ECHO= + } +} +test -z "$ECHO" && { printf Error: printf command does not support octal character codes ; exit 1 ; } + +GREY="\\033[1;90m" +BLUE="\\033[1;94m" +GREEN="\\033[0;32m" +RED="\\033[0;31m" +YELLOW="\\033[1;93m" +RESET="\\033[0m" + +MEM_LIMIT=150 + +>> $FILE || { echo Error: can not write to $FILE ; exit 1 ; } + +echo Warning: this script is setting performance parameters with afl-system-config +sleep 1 +afl-system-config > /dev/null 2>&1 +echo Performance settings applied. +echo + +$ECHO "${RESET}${GREY}[*] starting afl++ performance test framework ..." + +$ECHO "$BLUE[*] Testing: ${AFL_GCC}" +GCC=x +test -e ../${AFL_GCC} -a -e ../afl-fuzz && { + ../${AFL_GCC} -o test-instr.plain ../test-instr.c > /dev/null 2>&1 + test -e test-instr.plain && { + $ECHO "$GREEN[+] ${AFL_GCC} compilation succeeded" + mkdir -p in + echo 0 > in/in + $ECHO "$GREY[*] running afl-fuzz for ${AFL_GCC} for 30 seconds" + { + ../afl-fuzz -V 30 -s 123 -m ${MEM_LIMIT} -i in -o out-gcc -- ./test-instr.plain + } >>errors 2>&1 + test -n "$( ls out-gcc/queue/id:000002* 2> /dev/null )" && { + GCC=`grep execs_done out-gcc/fuzzer_stats | awk '{print$3}'` + } || { + echo CUT---------------------------------------------------------------- + cat errors + echo CUT---------------------------------------------------------------- + $ECHO "$RED[!] afl-fuzz is not working correctly with ${AFL_GCC}" + } + rm -rf in out-gcc errors test-instr.plain + } || $ECHO "$RED[!] ${AFL_GCC} instrumentation failed" +} || $ECHO "$YELLOW[-] afl is not compiled, cannot test" + +$ECHO "$BLUE[*] Testing: llvm_mode" +LLVM=x +test -e ../afl-clang-fast -a -e ../afl-fuzz && { + ../afl-clang-fast -o test-instr.llvm ../test-instr.c > /dev/null 2>&1 + test -e test-instr.llvm && { + $ECHO "$GREEN[+] llvm_mode compilation succeeded" + mkdir -p in + echo 0 > in/in + $ECHO "$GREY[*] running afl-fuzz for llvm_mode for 30 seconds" + { + ../afl-fuzz -V 30 -s 123 -m ${MEM_LIMIT} -i in -o out-llvm -- ./test-instr.llvm + } >>errors 2>&1 + test -n "$( ls out-llvm/queue/id:000002* 2> /dev/null )" && { + LLVM=`grep execs_done out-llvm/fuzzer_stats | awk '{print$3}'` + } || { + echo CUT---------------------------------------------------------------- + cat errors + echo CUT---------------------------------------------------------------- + $ECHO "$RED[!] afl-fuzz is not working correctly with llvm_mode" + } + rm -rf in out-llvm errors test-instr.llvm + } || $ECHO "$RED[!] llvm_mode instrumentation failed" +} || $ECHO "$YELLOW[-] llvm_mode is not compiled, cannot test" + +$ECHO "$BLUE[*] Testing: qemu_mode" +QEMU=x +test -e ../afl-qemu-trace -a -e ../afl-fuzz && { + cc -o test-instr.qemu ../test-instr.c > /dev/null 2>&1 + test -e test-instr.qemu && { + $ECHO "$GREEN[+] native compilation with cc succeeded" + mkdir -p in + echo 0 > in/in + $ECHO "$GREY[*] running afl-fuzz for qemu_mode for 30 seconds" + { + ../afl-fuzz -Q -V 30 -s 123 -m ${MEM_LIMIT} -i in -o out-qemu -- ./test-instr.qemu + } >>errors 2>&1 + test -n "$( ls out-qemu/queue/id:000002* 2> /dev/null )" && { + QEMU=`grep execs_done out-qemu/fuzzer_stats | awk '{print$3}'` + } || { + echo CUT---------------------------------------------------------------- + cat errors + echo CUT---------------------------------------------------------------- + $ECHO "$RED[!] afl-fuzz is not working correctly with qemu_mode" + } + rm -rf in out-qemu errors test-instr.qemu + } || $ECHO "$RED[!] qemu_mode instrumentation failed" +} || $ECHO "$YELLOW[-] qemu_mode is not compiled, cannot test" + +LOW_GCC= +HIGH_GCC= +LAST_GCC= +LOW_LLVM= +HIGH_LLVM= +LAST_LLVM= +LOW_QEMU= +HIGH_QEMU= +LAST_QEMU= + +test -s $FILE && { + while read LINE; do + G=`echo $LINE | awk '{print$1}'` + L=`echo $LINE | awk '{print$2}'` + Q=`echo $LINE | awk '{print$3}'` + test "$G" = x && G= + test "$L" = x && L= + test "$Q" = x && Q= + test -n "$G" && LAST_GCC=$G + test -n "$L" && LAST_LLVM=$L + test -n "$Q" && LAST_QEMU=$Q + test -n "$G" -a -z "$LOW_GCC" && LOW_GCC=$G || { + test -n "$G" -a "$G" -lt "$LOW_GCC" 2> /dev/null && LOW_GCC=$G + } + test -n "$L" -a -z "$LOW_LLVM" && LOW_LLVM=$L || { + test -n "$L" -a "$L" -lt "$LOW_LLVM" 2> /dev/null && LOW_LLVM=$L + } + test -n "$Q" -a -z "$LOW_QEMU" && LOW_QEMU=$Q || { + test -n "$Q" -a "$Q" -lt "$LOW_QEMU" 2> /dev/null && LOW_QEMU=$Q + } + test -n "$G" -a -z "$HIGH_GCC" && HIGH_GCC=$G || { + test -n "$G" -a "$G" -gt "$HIGH_GCC" 2> /dev/null && HIGH_GCC=$G + } + test -n "$L" -a -z "$HIGH_LLVM" && HIGH_LLVM=$L || { + test -n "$L" -a "$L" -gt "$HIGH_LLVM" 2> /dev/null && HIGH_LLVM=$L + } + test -n "$Q" -a -z "$HIGH_QEMU" && HIGH_QEMU=$Q || { + test -n "$Q" -a "$Q" -gt "$HIGH_QEMU" 2> /dev/null && HIGH_QEMU=$Q + } + done < $FILE + $ECHO "$YELLOW[!] Reading saved data from $FILE completed, please compare the results:" + $ECHO "$BLUE[!] afl-cc: lowest=$LOW_GCC highest=$HIGH_GCC last=$LAST_GCC current=$GCC" + $ECHO "$BLUE[!] llvm_mode: lowest=$LOW_LLVM highest=$HIGH_LLVM last=$LAST_LLVM current=$LLVM" + $ECHO "$BLUE[!] qemu_mode: lowest=$LOW_QEMU highest=$HIGH_QEMU last=$LAST_QEMU current=$QEMU" +} || { + $ECHO "$YELLOW[!] First run, just saving data" + $ECHO "$BLUE[!] afl-gcc=$GCC llvm_mode=$LLVM qemu_mode=$QEMU" +} +echo "$GCC $LLVM $QEMU" >> $FILE +$ECHO "$GREY[*] done." +$ECHO "$RESET" diff --git a/test/test.sh b/test/test.sh index 44236b63..781313a7 100755 --- a/test/test.sh +++ b/test/test.sh @@ -24,8 +24,11 @@ $ECHO \\101 2>&1 | grep -qE '^A' || { } test -z "$ECHO" && { printf Error: printf command does not support octal character codes ; exit 1 ; } +CODE=0 + export AFL_EXIT_WHEN_DONE=1 export AFL_SKIP_CPUFREQ=1 +export AFL_I_DONT_CARE_ABOUT_MISSING_CRASHES=1 unset AFL_QUIET unset AFL_DEBUG unset AFL_HARDEN @@ -33,6 +36,7 @@ unset AFL_USE_ASAN unset AFL_USE_MSAN unset AFL_CC unset AFL_PRELOAD +unset AFL_GCC_WHITELIST unset AFL_LLVM_WHITELIST unset AFL_LLVM_INSTRIM unset AFL_LLVM_LAF_SPLIT_SWITCHES @@ -46,10 +50,11 @@ test -e /usr/local/bin/opt && { # on MacOS X we prefer afl-clang over afl-gcc, because # afl-gcc does not work there test `uname -s` = 'Darwin' -o `uname -s` = 'FreeBSD' && { -AFL_GCC=afl-clang + AFL_GCC=afl-clang } || { -AFL_GCC=afl-gcc + AFL_GCC=afl-gcc } + GREY="\\033[1;90m" BLUE="\\033[1;94m" GREEN="\\033[0;32m" @@ -59,12 +64,14 @@ RESET="\\033[0m" MEM_LIMIT=150 +export PATH=$PATH:/sbin:/bin:/usr/sbin:/usr/bin:/usr/local/sbin:/usr/local/bin + $ECHO "${RESET}${GREY}[*] starting afl++ test framework ..." $ECHO "$BLUE[*] Testing: ${AFL_GCC}, afl-showmap and afl-fuzz" test -e ../${AFL_GCC} -a -e ../afl-showmap -a -e ../afl-fuzz && { ../${AFL_GCC} -o test-instr.plain ../test-instr.c > /dev/null 2>&1 - AFL_HARDEN=1 ../${AFL_GCC} -o test-instr.harden ../test-instr.c > /dev/null 2>&1 + AFL_HARDEN=1 ../${AFL_GCC} -o test-compcov.harden test-compcov.c > /dev/null 2>&1 test -e test-instr.plain && { $ECHO "$GREEN[+] ${AFL_GCC} compilation succeeded" echo 0 | ../afl-showmap -m ${MEM_LIMIT} -o test-instr.plain.0 -r -- ./test-instr.plain > /dev/null 2>&1 @@ -72,20 +79,30 @@ test -e ../${AFL_GCC} -a -e ../afl-showmap -a -e ../afl-fuzz && { test -e test-instr.plain.0 -a -e test-instr.plain.1 && { diff -q test-instr.plain.0 test-instr.plain.1 > /dev/null 2>&1 && { $ECHO "$RED[!] ${AFL_GCC} instrumentation should be different on different input but is not" + CODE=1 } || $ECHO "$GREEN[+] ${AFL_GCC} instrumentation present and working correctly" - } || $ECHO "$RED[!] ${AFL_GCC} instrumentation failed" + } || { + $ECHO "$RED[!] ${AFL_GCC} instrumentation failed" + CODE=1 + } rm -f test-instr.plain.0 test-instr.plain.1 } || $ECHO "$RED[!] ${AFL_GCC} failed" - test -e test-instr.harden && { - grep -qa fstack-protector-all test-instr.harden > /dev/null 2>&1 && { + test -e test-compcov.harden && { + grep -Eqa 'stack_chk_fail|fstack-protector-all|fortified' test-compcov.harden > /dev/null 2>&1 && { $ECHO "$GREEN[+] ${AFL_GCC} hardened mode succeeded and is working" - } || $ECHO "$RED[!] ${AFL_GCC} hardened mode is not hardened" - rm -f test-instr.harden - } || $ECHO "$RED[!] ${AFL_GCC} hardened mode compilation failed" + } || { + $ECHO "$RED[!] ${AFL_GCC} hardened mode is not hardened" + CODE=1 + } + rm -f test-compcov.harden + } || { + $ECHO "$RED[!] ${AFL_GCC} hardened mode compilation failed" + CODE=1 + } # now we want to be sure that afl-fuzz is working # make sure core_pattern is set to core on linux (test "$(uname -s)" = "Linux" && test "$(sysctl kernel.core_pattern)" != "kernel.core_pattern = core" && { - $ECHO "$RED[!] we cannot run afl-fuzz with enabled core dumps. Run 'sudo sh afl-system-config'.$RESET" + $ECHO "$YELLOW[!] we should not run afl-fuzz with enabled core dumps. Run 'sudo sh afl-system-config'.$RESET" true }) || # make sure crash reporter is disabled on Mac OS X @@ -97,12 +114,18 @@ test -e ../${AFL_GCC} -a -e ../afl-showmap -a -e ../afl-fuzz && { echo 0 > in/in $ECHO "$GREY[*] running afl-fuzz for ${AFL_GCC}, this will take approx 10 seconds" { - ../afl-fuzz -V10 -m ${MEM_LIMIT} -i in -o out -- ./test-instr.plain > /dev/null 2>&1 - } > /dev/null 2>&1 + ../afl-fuzz -V10 -m ${MEM_LIMIT} -i in -o out -- ./test-instr.plain >>errors 2>&1 + } >>errors 2>&1 test -n "$( ls out/queue/id:000002* 2> /dev/null )" && { $ECHO "$GREEN[+] afl-fuzz is working correctly with ${AFL_GCC}" - } || $ECHO "$RED[!] afl-fuzz is not working correctly with ${AFL_GCC}" - rm -rf in out + } || { + echo CUT------------------------------------------------------------------CUT + cat errors + echo CUT------------------------------------------------------------------CUT + $ECHO "$RED[!] afl-fuzz is not working correctly with ${AFL_GCC}" + CODE=1 + } + rm -rf in out errors } rm -f test-instr.plain } || $ECHO "$YELLOW[-] afl is not compiled, cannot test" @@ -110,7 +133,11 @@ test -e ../${AFL_GCC} -a -e ../afl-showmap -a -e ../afl-fuzz && { $ECHO "$BLUE[*] Testing: llvm_mode" test -e ../afl-clang-fast && { # on FreeBSD need to set AFL_CC - export AFL_CC=`llvm-config --bindir`/clang + if which clang >/dev/null; then + export AFL_CC=`which clang` + else + export AFL_CC=`llvm-config --bindir`/clang + fi ../afl-clang-fast -o test-instr.plain ../test-instr.c > /dev/null 2>&1 AFL_HARDEN=1 ../afl-clang-fast -o test-compcov.harden test-compcov.c > /dev/null 2>&1 test -e test-instr.plain && { @@ -120,36 +147,56 @@ test -e ../afl-clang-fast && { test -e test-instr.plain.0 -a -e test-instr.plain.1 && { diff -q test-instr.plain.0 test-instr.plain.1 > /dev/null 2>&1 && { $ECHO "$RED[!] llvm_mode instrumentation should be different on different input but is not" + CODE=1 } || $ECHO "$GREEN[+] llvm_mode instrumentation present and working correctly" - } || $ECHO "$RED[!] llvm_mode instrumentation failed" + } || { + $ECHO "$RED[!] llvm_mode instrumentation failed" + CODE=1 + } rm -f test-instr.plain.0 test-instr.plain.1 - } || $ECHO "$RED[!] llvm_mode failed" + } || { + $ECHO "$RED[!] llvm_mode failed" + CODE=1 + } test -e test-compcov.harden && { grep -Eqa 'stack_chk_fail|fstack-protector-all|fortified' test-compcov.harden > /dev/null 2>&1 && { $ECHO "$GREEN[+] llvm_mode hardened mode succeeded and is working" - } || $ECHO "$RED[!] llvm_mode hardened mode is not hardened" + } || { + $ECHO "$RED[!] llvm_mode hardened mode is not hardened" + CODE=1 + } rm -f test-compcov.harden - } || $ECHO "$RED[!] llvm_mode hardened mode compilation failed" + } || { + $ECHO "$RED[!] llvm_mode hardened mode compilation failed" + CODE=1 + } # now we want to be sure that afl-fuzz is working (test "$(uname -s)" = "Linux" && test "$(sysctl kernel.core_pattern)" != "kernel.core_pattern = core" && { - $ECHO "$RED[!] we cannot run afl-fuzz with enabled core dumps. Run 'sudo sh afl-system-config'.$RESET" + $ECHO "$YELLOW[!] we should not run afl-fuzz with enabled core dumps. Run 'sudo sh afl-system-config'.$RESET" true }) || # make sure crash reporter is disabled on Mac OS X (test "$(uname -s)" = "Darwin" && test $(launchctl list 2>/dev/null | grep -q '\.ReportCrash$') && { $ECHO "$RED[!] we cannot run afl-fuzz with enabled crash reporter. Run 'sudo sh afl-system-config'.$RESET" + CODE=1 true }) || { mkdir -p in echo 0 > in/in $ECHO "$GREY[*] running afl-fuzz for llvm_mode, this will take approx 10 seconds" { - ../afl-fuzz -V10 -m ${MEM_LIMIT} -i in -o out -- ./test-instr.plain > /dev/null 2>&1 - } > /dev/null 2>&1 + ../afl-fuzz -V10 -m ${MEM_LIMIT} -i in -o out -- ./test-instr.plain >>errors 2>&1 + } >>errors 2>&1 test -n "$( ls out/queue/id:000002* 2> /dev/null )" && { $ECHO "$GREEN[+] afl-fuzz is working correctly with llvm_mode" - } || $ECHO "$RED[!] afl-fuzz is not working correctly with llvm_mode" - rm -rf in out + } || { + echo CUT------------------------------------------------------------------CUT + cat errors + echo CUT------------------------------------------------------------------CUT + $ECHO "$RED[!] afl-fuzz is not working correctly with llvm_mode" + CODE=1 + } + rm -rf in out errors } rm -f test-instr.plain @@ -158,49 +205,176 @@ test -e ../afl-clang-fast && { test -e test-compcov.instrim && { grep -Eq " [1-3] location" test.out && { $ECHO "$GREEN[+] llvm_mode InsTrim feature works correctly" - } || $ECHO "$RED[!] llvm_mode InsTrim feature failed" - } || $ECHO "$RED[!] llvm_mode InsTrim feature compilation failed" + } || { + $ECHO "$RED[!] llvm_mode InsTrim feature failed" + CODE=1 + } + } || { + $ECHO "$RED[!] llvm_mode InsTrim feature compilation failed" + CODE=1 + } rm -f test-compcov.instrim test.out AFL_LLVM_LAF_SPLIT_SWITCHES=1 AFL_LLVM_LAF_TRANSFORM_COMPARES=1 AFL_LLVM_LAF_SPLIT_COMPARES=1 ../afl-clang-fast -o test-compcov.compcov test-compcov.c > /dev/null 2> test.out test -e test-compcov.compcov && { grep -Eq " [3-9][0-9] location" test.out && { $ECHO "$GREEN[+] llvm_mode laf-intel/compcov feature works correctly" - } || $ECHO "$RED[!] llvm_mode laf-intel/compcov feature failed" - } || $ECHO "$RED[!] llvm_mode laf-intel/compcov feature compilation failed" + } || { + $ECHO "$RED[!] llvm_mode laf-intel/compcov feature failed" + CODE=1 + } + } || { + $ECHO "$RED[!] llvm_mode laf-intel/compcov feature compilation failed" + CODE=1 + } rm -f test-compcov.compcov test.out echo foobar.c > whitelist.txt AFL_LLVM_WHITELIST=whitelist.txt ../afl-clang-fast -o test-compcov test-compcov.c > test.out 2>&1 test -e test-compcov && { grep -q "No instrumentation targets found" test.out && { $ECHO "$GREEN[+] llvm_mode whitelist feature works correctly" - } || $ECHO "$RED[!] llvm_mode whitelist feature failed" - } || $ECHO "$RED[!] llvm_mode whitelist feature compilation failed" + } || { + $ECHO "$RED[!] llvm_mode whitelist feature failed" + CODE=1 + } + } || { + $ECHO "$RED[!] llvm_mode whitelist feature compilation failed" + CODE=1 + } rm -f test-compcov test.out whitelist.txt ../afl-clang-fast -o test-persistent ../experimental/persistent_demo/persistent_demo.c > /dev/null 2>&1 test -e test-persistent && { echo foo | ../afl-showmap -o /dev/null -q -r ./test-persistent && { $ECHO "$GREEN[+] llvm_mode persistent mode feature works correctly" - } || $ECHO "$RED[!] llvm_mode persistent mode feature failed to work" - } || $ECHO "$RED[!] llvm_mode persistent mode feature compilation failed" + } || { + $ECHO "$RED[!] llvm_mode persistent mode feature failed to work" + CODE=1 + } + } || { + $ECHO "$RED[!] llvm_mode persistent mode feature compilation failed" + CODE=1 + } rm -f test-persistent } || $ECHO "$YELLOW[-] llvm_mode not compiled, cannot test" +$ECHO "$BLUE[*] Testing: gcc_plugin" +export AFL_CC=`which gcc` +test -e ../afl-gcc-fast && { + ../afl-gcc-fast -o test-instr.plain.gccpi ../test-instr.c > /dev/null 2>&1 + AFL_HARDEN=1 ../afl-gcc-fast -o test-compcov.harden.gccpi test-compcov.c > /dev/null 2>&1 + test -e test-instr.plain.gccpi && { + $ECHO "$GREEN[+] gcc_plugin compilation succeeded" + echo 0 | ../afl-showmap -m ${MEM_LIMIT} -o test-instr.plain.0 -r -- ./test-instr.plain.gccpi > /dev/null 2>&1 + ../afl-showmap -m ${MEM_LIMIT} -o test-instr.plain.1 -r -- ./test-instr.plain.gccpi < /dev/null > /dev/null 2>&1 + test -e test-instr.plain.0 -a -e test-instr.plain.1 && { + diff -q test-instr.plain.0 test-instr.plain.1 > /dev/null 2>&1 && { + $ECHO "$RED[!] gcc_plugin instrumentation should be different on different input but is not" + CODE=1 + } || { + $ECHO "$GREEN[+] gcc_plugin instrumentation present and working correctly" + } + } || { + $ECHO "$RED[!] gcc_plugin instrumentation failed" + CODE=1 + } + rm -f test-instr.plain.0 test-instr.plain.1 + } || { + $ECHO "$RED[!] gcc_plugin failed" + CODE=1 + } + + test -e test-compcov.harden.gccpi && { + grep -Eqa 'stack_chk_fail|fstack-protector-all|fortified' test-compcov.harden.gccpi > /dev/null 2>&1 && { + $ECHO "$GREEN[+] gcc_plugin hardened mode succeeded and is working" + } || { + $ECHO "$RED[!] gcc_plugin hardened mode is not hardened" + CODE=1 + } + rm -f test-compcov.harden.gccpi + } || { + $ECHO "$RED[!] gcc_plugin hardened mode compilation failed" + CODE=1 + } + # now we want to be sure that afl-fuzz is working + (test "$(uname -s)" = "Linux" && test "$(sysctl kernel.core_pattern)" != "kernel.core_pattern = core" && { + $ECHO "$YELLOW[!] we should not run afl-fuzz with enabled core dumps. Run 'sudo sh afl-system-config'.$RESET" + true + }) || + # make sure crash reporter is disabled on Mac OS X + (test "$(uname -s)" = "Darwin" && test $(launchctl list 2>/dev/null | grep -q '\.ReportCrash$') && { + $ECHO "$RED[!] we cannot run afl-fuzz with enabled crash reporter. Run 'sudo sh afl-system-config'.$RESET" + CODE=1 + true + }) || { + mkdir -p in + echo 0 > in/in + $ECHO "$GREY[*] running afl-fuzz for gcc_plugin, this will take approx 10 seconds" + { + ../afl-fuzz -V10 -m ${MEM_LIMIT} -i in -o out -- ./test-instr.plain.gccpi >>errors 2>&1 + } >>errors 2>&1 + test -n "$( ls out/queue/id:000002* 2> /dev/null )" && { + $ECHO "$GREEN[+] afl-fuzz is working correctly with gcc_plugin" + } || { + echo CUT------------------------------------------------------------------CUT + cat errors + echo CUT------------------------------------------------------------------CUT + $ECHO "$RED[!] afl-fuzz is not working correctly with gcc_plugin" + CODE=1 + } + rm -rf in out errors + } + rm -f test-instr.plain.gccpi + + # now for the special gcc_plugin things + echo foobar.c > whitelist.txt + AFL_GCC_WHITELIST=whitelist.txt ../afl-gcc-fast -o test-compcov test-compcov.c > /dev/null 2>&1 + test -e test-compcov && { + echo 1 | ../afl-showmap -m ${MEM_LIMIT} -o - -r -- ./test-compcov 2>&1 | grep -q "Captured 1 tuples" && { + $ECHO "$GREEN[+] gcc_plugin whitelist feature works correctly" + } || { + $ECHO "$RED[!] gcc_plugin whitelist feature failed" + CODE=1 + } + } || { + $ECHO "$RED[!] gcc_plugin whitelist feature compilation failed" + CODE=1 + } + rm -f test-compcov test.out whitelist.txt + ../afl-gcc-fast -o test-persistent ../experimental/persistent_demo/persistent_demo.c > /dev/null 2>&1 + test -e test-persistent && { + echo foo | ../afl-showmap -o /dev/null -q -r ./test-persistent && { + $ECHO "$GREEN[+] gcc_plugin persistent mode feature works correctly" + } || { + $ECHO "$RED[!] gcc_plugin persistent mode feature failed to work" + CODE=1 + } + } || { + $ECHO "$RED[!] gcc_plugin persistent mode feature compilation failed" + CODE=1 + } + rm -f test-persistent +} || $ECHO "$YELLOW[-] gcc_plugin not compiled, cannot test" + $ECHO "$BLUE[*] Testing: shared library extensions" -gcc -o test-compcov test-compcov.c > /dev/null 2>&1 +cc -o test-compcov test-compcov.c > /dev/null 2>&1 test -e ../libtokencap.so && { - AFL_TOKEN_FILE=token.out LD_PRELOAD=../libtokencap.so ./test-compcov foobar > /dev/null 2>&1 + AFL_TOKEN_FILE=token.out LD_PRELOAD=../libtokencap.so DYLD_INSERT_LIBRARIES=../libtokencap.so DYLD_FORCE_FLAT_NAMESPACE=1 ./test-compcov foobar > /dev/null 2>&1 grep -q BUGMENOT token.out > /dev/null 2>&1 && { $ECHO "$GREEN[+] libtokencap did successfully capture tokens" - } || $ECHO "$RED[!] libtokencap did not capture tokens" + } || { + $ECHO "$RED[!] libtokencap did not capture tokens" + CODE=1 + } rm -f token.out } || $ECHO "$YELLOW[-] libtokencap is not compiled, cannot test" test -e ../libdislocator.so && { { ulimit -c 1 - LD_PRELOAD=../libdislocator.so ./test-compcov BUFFEROVERFLOW > test.out 2> /dev/null + # DYLD_INSERT_LIBRARIES and DYLD_FORCE_FLAT_NAMESPACE is used on Darwin/MacOSX + LD_PRELOAD=../libdislocator.so DYLD_INSERT_LIBRARIES=../libdislocator.so DYLD_FORCE_FLAT_NAMESPACE=1 ./test-compcov BUFFEROVERFLOW > test.out 2> /dev/null } > /dev/null 2>&1 grep -q BUFFEROVERFLOW test.out > /dev/null 2>&1 && { $ECHO "$RED[!] libdislocator did not detect the memory corruption" + CODE=1 } || $ECHO "$GREEN[+] libdislocator did successfully detect the memory corruption" rm -f test.out core test-compcov.core core.test-compcov } || $ECHO "$YELLOW[-] libdislocator is not compiled, cannot test" @@ -216,40 +390,75 @@ test -e ../afl-qemu-trace && { echo 0 > in/in $ECHO "$GREY[*] running afl-fuzz for qemu_mode, this will take approx 10 seconds" { - ../afl-fuzz -V10 -Q -i in -o out -- ./test-instr > /dev/null 2>&1 - } > /dev/null 2>&1 + ../afl-fuzz -V10 -Q -i in -o out -- ./test-instr >>errors 2>&1 + } >>errors 2>&1 test -n "$( ls out/queue/id:000002* 2> /dev/null )" && { $ECHO "$GREEN[+] afl-fuzz is working correctly with qemu_mode" - } || $ECHO "$RED[!] afl-fuzz is not working correctly with qemu_mode" + RUNTIME=`grep execs_done out/fuzzer_stats | awk '{print$3}'` + } || { + echo CUT------------------------------------------------------------------CUT + cat errors + echo CUT------------------------------------------------------------------CUT + $ECHO "$RED[!] afl-fuzz is not working correctly with qemu_mode" + CODE=1 + } + rm -f errors test -e ../libcompcov.so && { $ECHO "$GREY[*] running afl-fuzz for qemu_mode libcompcov, this will take approx 10 seconds" { export AFL_PRELOAD=../libcompcov.so export AFL_COMPCOV_LEVEL=2 - ../afl-fuzz -V10 -Q -i in -o out -- ./test-compcov > /dev/null 2>&1 - } > /dev/null 2>&1 + ../afl-fuzz -V10 -Q -i in -o out -- ./test-compcov >>errors 2>&1 + } >>errors 2>&1 test -n "$( ls out/queue/id:000002* 2> /dev/null )" && { $ECHO "$GREEN[+] afl-fuzz is working correctly with qemu_mode libcompcov" - } || $ECHO "$RED[!] afl-fuzz is not working correctly with qemu_mode libcompcov" + } || { + echo CUT------------------------------------------------------------------CUT + cat errors + echo CUT------------------------------------------------------------------CUT + $ECHO "$RED[!] afl-fuzz is not working correctly with qemu_mode libcompcov" + CODE=1 + } } || $ECHO "$YELLOW[-] we cannot test qemu_mode libcompcov because it is not present" - rm -rf in out + rm -f errors + + $ECHO "$GREY[*] running afl-fuzz for persistent qemu_mode, this will take approx 10 seconds" + { + export AFL_QEMU_PERSISTENT_ADDR=0x$(nm test-instr | grep "T main" | awk '{ print $1 }') + export AFL_QEMU_PERSISTENT_GPR=1 + ../afl-fuzz -V10 -Q -i in -o out -- ./test-instr > /dev/null 2>&1 + } >>errors 2>&1 + test -n "$( ls out/queue/id:000002* 2> /dev/null )" && { + $ECHO "$GREEN[+] afl-fuzz is working correctly with persistent qemu_mode" + RUNTIMEP=`grep execs_done out/fuzzer_stats | awk '{print$3}'` + test -n "$RUNTIME" -a -n "$RUNTIMEP" && { + SLOW=`expr $RUNTIME '*' 103` # persistent mode should be at least 3% faster - minimum! + FAST=`expr $RUNTIMEP '*' 100` + test "$SLOW" -lt "$FAST" && { + $ECHO "$GREEN[+] persistent qemu_mode was noticeable faster than standard qemu_mode" + } || { + $ECHO "$YELLOW[?] persistent qemu_mode was not noticeable faster than standard qemu_mode" + } + } || { + $ECHO "$YELLOW[?] we got no data on executions performed? weird!" + } + } || { + echo CUT------------------------------------------------------------------CUT + cat errors + echo CUT------------------------------------------------------------------CUT + $ECHO "$RED[!] afl-fuzz is not working correctly with persistent qemu_mode" + CODE=1 + exit 1 + } + $ECHO "$YELLOW[?] we need a test case for qemu_mode unsigaction library" + rm -rf in out errors } - } || $ECHO "$RED[-] gcc compilation of test targets failed - what is going on??" + } || { + $ECHO "$RED[-] gcc compilation of test targets failed - what is going on??" + CODE=1 + } - $ECHO "$YELLOW[?] we need a test case for qemu_mode persistent mode" - $ECHO "$YELLOW[?] we need a test case for qemu_mode unsigaction library" - # This works but there are already problems with persistent (e.g. stability) - #$ECHO "$GREY[*] running afl-fuzz for persistent qemu_mode, this will take approx 10 seconds" - #{ - # export AFL_QEMU_PERSISTENT_ADDR=0x$(nm test-instr | grep "T main" | awk '{ print $1 }') - # export AFL_QEMU_PERSISTENT_GPR=1 - # ../afl-fuzz -V10 -Q -i in -o out -- ./test-instr > /dev/null 2>&1 - #} > /dev/null 2>&1 - #test -n "$( ls out/queue/id:000002* 2> /dev/null )" && { - # $ECHO "$GREEN[+] afl-fuzz is working correctly with persistent qemu_mode" - #} || $ECHO "$RED[!] afl-fuzz is not working correctly with persistent qemu_mode" - rm -f test-instr test-compcov } || $ECHO "$YELLOW[-] qemu_mode is not compiled, cannot test" @@ -257,29 +466,54 @@ $ECHO "$BLUE[*] Testing: unicorn_mode" test -d ../unicorn_mode/unicorn && { test -e ../unicorn_mode/samples/simple/simple_target.bin -a -e ../unicorn_mode/samples/compcov_x64/compcov_target.bin && { { + # travis workaround + PY=`which python2.7` + test "$PY" = "/opt/pyenv/shims/python2.7" -a -x /usr/bin/python2.7 && PY=/usr/bin/python2.7 mkdir -p in echo 0 > in/in - $ECHO "$GREY[*] running afl-fuzz for unicorn_mode, this will take approx 15 seconds" + $ECHO "$GREY[*] Using python binary $PY" + $ECHO "$GREY[*] running afl-fuzz for unicorn_mode, this will take approx 25 seconds" { - ../afl-fuzz -V15 -U -i in -o out -d -- python ../unicorn_mode/samples/simple/simple_test_harness.py @@ > /dev/null 2>&1 - } > /dev/null 2>&1 + ../afl-fuzz -V25 -U -i in -o out -d -- "$PY" ../unicorn_mode/samples/simple/simple_test_harness.py @@ >>errors 2>&1 + } >>errors 2>&1 test -n "$( ls out/queue/id:000002* 2> /dev/null )" && { $ECHO "$GREEN[+] afl-fuzz is working correctly with unicorn_mode" - } || $ECHO "$RED[!] afl-fuzz is not working correctly with unicorn_mode" + } || { + echo CUT------------------------------------------------------------------CUT + cat errors + echo CUT------------------------------------------------------------------CUT + $ECHO "$RED[!] afl-fuzz is not working correctly with unicorn_mode" + CODE=1 + } + rm -f errors - $ECHO "$GREY[*] running afl-fuzz for unicorn_mode compcov, this will take approx 15 seconds" + printf '\x01\x01' > in/in + # This seed is close to the first byte of the comparison. + # If CompCov works, a new tuple will appear in the map => new input in queue + $ECHO "$GREY[*] running afl-fuzz for unicorn_mode compcov, this will take approx 35 seconds" { export AFL_COMPCOV_LEVEL=2 - ../afl-fuzz -V15 -U -i in -o out -d -- python ../unicorn_mode/samples/compcov_x64/compcov_test_harness.py @@ > /dev/null 2>&1 - } > /dev/null 2>&1 + ../afl-fuzz -V35 -U -i in -o out -d -- "$PY" ../unicorn_mode/samples/compcov_x64/compcov_test_harness.py @@ >>errors 2>&1 + } >>errors 2>&1 test -n "$( ls out/queue/id:000001* 2> /dev/null )" && { $ECHO "$GREEN[+] afl-fuzz is working correctly with unicorn_mode compcov" - } || $ECHO "$RED[!] afl-fuzz is not working correctly with unicorn_mode compcov" - rm -rf in out + } || { + echo CUT------------------------------------------------------------------CUT + cat errors + echo CUT------------------------------------------------------------------CUT + $ECHO "$RED[!] afl-fuzz is not working correctly with unicorn_mode compcov" + CODE=1 + } + rm -rf in out errors } - } || $ECHO "$RED[-] missing sample binaries in unicorn_mode/samples/ - what is going on??" + } || { + $ECHO "$RED[-] missing sample binaries in unicorn_mode/samples/ - what is going on??" + CODE=1 + } } || $ECHO "$YELLOW[-] unicorn_mode is not compiled, cannot test" $ECHO "$GREY[*] all test cases completed.$RESET" - +test "$CODE" = "0" && $ECHO "$GREEN[+] all tests were successful :-)$RESET" +test "$CODE" = "0" || $ECHO "$RED[-] failure in tests :-($RESET" +exit $CODE diff --git a/unicorn_mode/README.md b/unicorn_mode/README.md index ea3e3c9b..8f381b59 100644 --- a/unicorn_mode/README.md +++ b/unicorn_mode/README.md @@ -99,7 +99,7 @@ The options that enables Unicorn CompareCoverage are the same used for QEMU. AFL_COMPCOV_LEVEL=1 is to instrument comparisons with only immediate values. QEMU_COMPCOV_LEVEL=2 instruments all comparison instructions. Comparison instructions are currently instrumented only -on the x86 and x86_64 targets. +for the x86, x86_64 and ARM targets. ## 4) Gotchas, feedback, bugs diff --git a/unicorn_mode/build_unicorn_support.sh b/unicorn_mode/build_unicorn_support.sh index 1575f66c..589ab852 100755 --- a/unicorn_mode/build_unicorn_support.sh +++ b/unicorn_mode/build_unicorn_support.sh @@ -6,7 +6,7 @@ # Originally written by Nathan Voss <njvoss99@gmail.com> # # Adapted from code by Andrew Griffiths <agriffiths@google.com> and -# Michal Zalewski <lcamtuf@google.com> +# Michal Zalewski # # Adapted for AFLplusplus by Dominik Maier <mail@dmnk.co> # @@ -43,9 +43,11 @@ echo echo "[*] Performing basic sanity checks..." -if [ ! "`uname -s`" = "Linux" ]; then +PLT=`uname -s` - echo "[-] Error: Unicorn instrumentation is supported only on Linux." +if [ ! "$PLT" = "Linux" ] && [ ! "$PLT" = "Darwin" ] && [ ! "$PLT" = "FreeBSD" ] && [ ! "$PLT" = "NetBSD" ] && [ ! "$PLT" = "OpenBSD" ]; then + + echo "[-] Error: Unicorn instrumentation is unsupported on $PLT." exit 1 fi @@ -64,7 +66,43 @@ if [ ! -f "../afl-showmap" ]; then fi -for i in wget python automake autoconf sha384sum; do +if [ "$PLT" = "Linux" ]; then + CKSUMCMD='sha384sum --' + PYTHONBIN=python2 + MAKECMD=make + CORES=`nproc` + TARCMD=tar + EASY_INSTALL=easy_install +fi + +if [ "$PLT" = "Darwin" ]; then + CKSUMCMD="shasum -a 384" + PYTHONBIN=python2.7 + MAKECMD=make + CORES=`sysctl hw.ncpu | cut -d' ' -f2` + TARCMD=tar + EASY_INSTALL=easy_install-2.7 +fi + +if [ "$PLT" = "FreeBSD" ]; then + CKSUMCMD="sha384 -q" + PYTHONBIN=python2.7 + MAKECMD=gmake + CORES=`sysctl hw.ncpu | cut -d' ' -f2` + TARCMD=gtar + EASY_INSTALL=easy_install-2.7 +fi + +if [ "$PLT" = "NetBSD" ] || [ "$PLT" = "OpenBSD" ]; then + CKSUMCMD="cksum -a sha384 -q" + PYTHONBIN=python2.7 + MAKECMD=gmake + CORES=`sysctl hw.ncpu | cut -d' ' -f2` + TARCMD=gtar + EASY_INSTALL=easy_install-2.7 +fi + +for i in wget $PYTHONBIN automake autoconf $MAKECMD $TARCMD; do T=`which "$i" 2>/dev/null` @@ -77,10 +115,10 @@ for i in wget python automake autoconf sha384sum; do done -if ! which easy_install > /dev/null; then +if ! which $EASY_INSTALL > /dev/null; then # work around for unusual installs - if [ '!' -e /usr/lib/python2.7/dist-packages/easy_install.py ]; then + if [ '!' -e /usr/lib/python2.7/dist-packages/easy_install.py ] && [ '!' -e /usr/local/lib/python2.7/dist-packages/easy_install.py ] && [ '!' -e /usr/pkg/lib/python2.7/dist-packages/easy_install.py ]; then echo "[-] Error: Python setup-tools not found. Run 'sudo apt-get install python-setuptools'." exit 1 @@ -100,15 +138,18 @@ echo "[+] All checks passed!" ARCHIVE="`basename -- "$UNICORN_URL"`" -CKSUM=`sha384sum -- "$ARCHIVE" 2>/dev/null | cut -d' ' -f1` +CKSUM=`$CKSUMCMD "$ARCHIVE" 2>/dev/null | cut -d' ' -f1` if [ ! "$CKSUM" = "$UNICORN_SHA384" ]; then echo "[*] Downloading Unicorn v1.0.1 from the web..." rm -f "$ARCHIVE" - wget -O "$ARCHIVE" -- "$UNICORN_URL" || exit 1 + OK= + while [ -z "$OK" ]; do + wget -c -O "$ARCHIVE" -- "$UNICORN_URL" && OK=1 + done - CKSUM=`sha384sum -- "$ARCHIVE" 2>/dev/null | cut -d' ' -f1` + CKSUM=`$CKSUMCMD "$ARCHIVE" 2>/dev/null | cut -d' ' -f1` fi @@ -127,7 +168,7 @@ echo "[*] Uncompressing archive (this will take a while)..." rm -rf "unicorn" || exit 1 mkdir "unicorn" || exit 1 -tar xzf "$ARCHIVE" -C ./unicorn --strip-components=1 || exit 1 +$TARCMD xzf "$ARCHIVE" -C ./unicorn --strip-components=1 || exit 1 echo "[+] Unpacking successful." @@ -149,7 +190,7 @@ echo "[+] Configuration complete." echo "[*] Attempting to build Unicorn (fingers crossed!)..." -UNICORN_QEMU_FLAGS='--python=python2' make -j `nproc` || exit 1 +UNICORN_QEMU_FLAGS="--python=$PYTHONBIN" $MAKECMD -j$CORES || exit 1 echo "[+] Build process successful!" @@ -157,10 +198,10 @@ echo "[*] Installing Unicorn python bindings..." cd bindings/python || exit 1 if [ -z "$VIRTUAL_ENV" ]; then echo "[*] Info: Installing python unicorn using --user" - python setup.py install --user || exit 1 + $PYTHONBIN setup.py install --user || exit 1 else echo "[*] Info: Installing python unicorn to virtualenv: $VIRTUAL_ENV" - python setup.py install || exit 1 + $PYTHONBIN setup.py install || exit 1 fi export LIBUNICORN_PATH='$(pwd)' # in theory, this allows to switch between afl-unicorn and unicorn so files. @@ -175,7 +216,7 @@ cd ../samples/simple || exit 1 # Run afl-showmap on the sample application. If anything comes out then it must have worked! unset AFL_INST_RATIO -echo 0 | ../../../afl-showmap -U -m none -q -o .test-instr0 -- python simple_test_harness.py ./sample_inputs/sample1.bin || exit 1 +echo 0 | ../../../afl-showmap -U -m none -q -o .test-instr0 -- $PYTHONBIN simple_test_harness.py ./sample_inputs/sample1.bin || exit 1 if [ -s .test-instr0 ] then diff --git a/unicorn_mode/patches/afl-unicorn-common.h b/unicorn_mode/patches/afl-unicorn-common.h index fd88e21b..66d03803 100644 --- a/unicorn_mode/patches/afl-unicorn-common.h +++ b/unicorn_mode/patches/afl-unicorn-common.h @@ -3,7 +3,7 @@ ---------------------------------------------- Originally written by Andrew Griffiths <agriffiths@google.com> and - Michal Zalewski <lcamtuf@google.com> + Michal Zalewski Adapted for afl-unicorn by Dominik Maier <mail@dmnk.co> diff --git a/unicorn_mode/patches/afl-unicorn-cpu-inl.h b/unicorn_mode/patches/afl-unicorn-cpu-inl.h index 082d6d68..72092e29 100644 --- a/unicorn_mode/patches/afl-unicorn-cpu-inl.h +++ b/unicorn_mode/patches/afl-unicorn-cpu-inl.h @@ -3,7 +3,7 @@ ---------------------------------------------- Originally written by Andrew Griffiths <agriffiths@google.com> and - Michal Zalewski <lcamtuf@google.com> + Michal Zalewski Adapted for afl-unicorn by Dominik Maier <mail@dmnk.co> diff --git a/unicorn_mode/patches/afl-unicorn-cpu-translate-inl.h b/unicorn_mode/patches/afl-unicorn-cpu-translate-inl.h index 7c84058f..70472a72 100644 --- a/unicorn_mode/patches/afl-unicorn-cpu-translate-inl.h +++ b/unicorn_mode/patches/afl-unicorn-cpu-translate-inl.h @@ -3,7 +3,7 @@ ---------------------------------------------- Originally written by Andrew Griffiths <agriffiths@google.com> and - Michal Zalewski <lcamtuf@google.com> + Michal Zalewski Adapted for afl-unicorn by Dominik Maier <mail@dmnk.co> diff --git a/unicorn_mode/patches/afl-unicorn-tcg-op-inl.h b/unicorn_mode/patches/afl-unicorn-tcg-op-inl.h index d21bbcc7..8f4a8748 100644 --- a/unicorn_mode/patches/afl-unicorn-tcg-op-inl.h +++ b/unicorn_mode/patches/afl-unicorn-tcg-op-inl.h @@ -3,7 +3,7 @@ ---------------------------------------------- Originally written by Andrew Griffiths <agriffiths@google.com> and - Michal Zalewski <lcamtuf@google.com> + Michal Zalewski Adapted for afl-unicorn by Dominik Maier <mail@dmnk.co> diff --git a/unicorn_mode/patches/afl-unicorn-tcg-runtime-inl.h b/unicorn_mode/patches/afl-unicorn-tcg-runtime-inl.h index 95e68302..3603fae0 100644 --- a/unicorn_mode/patches/afl-unicorn-tcg-runtime-inl.h +++ b/unicorn_mode/patches/afl-unicorn-tcg-runtime-inl.h @@ -3,7 +3,7 @@ ---------------------------------------------- Originally written by Andrew Griffiths <agriffiths@google.com> and - Michal Zalewski <lcamtuf@google.com> + Michal Zalewski Adapted for afl-unicorn by Dominik Maier <mail@dmnk.co> @@ -104,3 +104,71 @@ void HELPER(afl_compcov_log_64)(void* uc_ptr, uint64_t cur_loc, uint64_t arg1, } +/* // Little endian CompCov +void HELPER(afl_compcov_log_16)(void* uc_ptr, uint64_t cur_loc, uint64_t arg1, + uint64_t arg2) { + + u8* afl_area_ptr = ((struct uc_struct*)uc_ptr)->afl_area_ptr; + + if ((arg1 & 0xff00) == (arg2 & 0xff00)) { INC_AFL_AREA(cur_loc); } + +} + +void HELPER(afl_compcov_log_32)(void* uc_ptr, uint64_t cur_loc, uint64_t arg1, + uint64_t arg2) { + + u8* afl_area_ptr = ((struct uc_struct*)uc_ptr)->afl_area_ptr; + + if ((arg1 & 0xff000000) == (arg2 & 0xff000000)) { + + INC_AFL_AREA(cur_loc + 2); + if ((arg1 & 0xff0000) == (arg2 & 0xff0000)) { + + INC_AFL_AREA(cur_loc + 1); + if ((arg1 & 0xff00) == (arg2 & 0xff00)) { INC_AFL_AREA(cur_loc); } + + } + + } + +} + +void HELPER(afl_compcov_log_64)(void* uc_ptr, uint64_t cur_loc, uint64_t arg1, + uint64_t arg2) { + + u8* afl_area_ptr = ((struct uc_struct*)uc_ptr)->afl_area_ptr; + + if ((arg1 & 0xff00000000000000) == (arg2 & 0xff00000000000000)) { + + INC_AFL_AREA(cur_loc + 6); + if ((arg1 & 0xff000000000000) == (arg2 & 0xff000000000000)) { + + INC_AFL_AREA(cur_loc + 5); + if ((arg1 & 0xff0000000000) == (arg2 & 0xff0000000000)) { + + INC_AFL_AREA(cur_loc + 4); + if ((arg1 & 0xff00000000) == (arg2 & 0xff00000000)) { + + INC_AFL_AREA(cur_loc + 3); + if ((arg1 & 0xff000000) == (arg2 & 0xff000000)) { + + INC_AFL_AREA(cur_loc + 2); + if ((arg1 & 0xff0000) == (arg2 & 0xff0000)) { + + INC_AFL_AREA(cur_loc + 1); + if ((arg1 & 0xff00) == (arg2 & 0xff00)) { INC_AFL_AREA(cur_loc); } + + } + + } + + } + + } + + } + + } + +} +*/ diff --git a/unicorn_mode/samples/compcov_x64/compcov_target.bin b/unicorn_mode/samples/compcov_x64/compcov_target.bin index 091bf1db..2874860b 100644 --- a/unicorn_mode/samples/compcov_x64/compcov_target.bin +++ b/unicorn_mode/samples/compcov_x64/compcov_target.bin Binary files differdiff --git a/unicorn_mode/samples/compcov_x64/compcov_target.c b/unicorn_mode/samples/compcov_x64/compcov_target.c index eb1205b1..0c863b25 100644 --- a/unicorn_mode/samples/compcov_x64/compcov_target.c +++ b/unicorn_mode/samples/compcov_x64/compcov_target.c @@ -16,11 +16,9 @@ int main(void) { unsigned int *data_buf = (unsigned int *) DATA_ADDRESS; - if (data_buf[0] == 0xabadcafe) { - // Cause an 'invalid read' crash if data[0..3] == '\x01\x02\x03\x04' + if (((unsigned short*)data_buf)[0] == 0x0100) { unsigned char invalid_read = *(unsigned char *) 0x00000000; - } else if (data_buf[1] == data_buf[2] + 0x4141) { - // Cause an 'invalid read' crash if (0x10 < data[0] < 0x20) and data[1] > data[2] + } else if (data_buf[1] == data_buf[2] + 0xfffe) { unsigned char invalid_read = *(unsigned char *) 0x00000000; } diff --git a/unicorn_mode/samples/compcov_x64/compcov_target.elf b/unicorn_mode/samples/compcov_x64/compcov_target.elf index 7015fb46..0f1ad916 100755 --- a/unicorn_mode/samples/compcov_x64/compcov_target.elf +++ b/unicorn_mode/samples/compcov_x64/compcov_target.elf Binary files differ |