From d90328f6be726190e013f83df37e49383be1c5e4 Mon Sep 17 00:00:00 2001 From: Shengtuo Hu Date: Tue, 21 Jul 2020 18:00:21 -0400 Subject: Allow the custom mutator to generate larger trimmed data (#463) --- src/afl-fuzz-mutators.c | 20 +++++++++++++++++--- 1 file changed, 17 insertions(+), 3 deletions(-) diff --git a/src/afl-fuzz-mutators.c b/src/afl-fuzz-mutators.c index 0fb34ab7..17a68ff8 100644 --- a/src/afl-fuzz-mutators.c +++ b/src/afl-fuzz-mutators.c @@ -282,9 +282,23 @@ u8 trim_case_custom(afl_state_t *afl, struct queue_entry *q, u8 *in_buf, } else if (unlikely(retlen > orig_len)) { - FATAL( - "Trimmed data returned by custom mutator is larger than original " - "data"); + /* Do not exit the fuzzer, even if the trimmed data returned by the custom + mutator is larger than the original data. For some use cases, like the + grammar mutator, the definition of "size" may have different meanings. + For example, the trimming function in a grammar mutator aims at + reducing the objects in a grammar structure, but does not guarantee to + generate a smaller binary buffer. + + Thus, we allow the custom mutator to generate the trimmed data that is + larger than the original data. */ + + if (afl->not_on_tty && afl->debug) { + + WARNF( + "Trimmed data returned by custom mutator is larger than original " + "data"); + + } } else if (unlikely(retlen == 0)) { -- cgit 1.4.1 From 024a88a6bbf2e25e5b61f061e9b98351cbc86a66 Mon Sep 17 00:00:00 2001 From: electricworry Date: Wed, 22 Jul 2020 20:19:55 +0100 Subject: Fixes AFLplusplus/AFLplusplus#464 winepath path translation for .cur_input. --- afl-wine-trace | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) diff --git a/afl-wine-trace b/afl-wine-trace index 65525a33..8853a757 100755 --- a/afl-wine-trace +++ b/afl-wine-trace @@ -68,7 +68,12 @@ else: 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 + # Get the Wine translated path using the winepath tool + arg_translated = subprocess.run([os.path.join(os.path.dirname(wine_path), "winepath"), "--windows", argv[i]], universal_newlines=True, stdout=subprocess.PIPE).stdout + # Remove the spurious LF at the end of the path + if len(arg_translated) > 0 and arg_translated[-1] == '\n': + arg_translated = arg_translated[:-1] + argv[i] = arg_translated break print("[afl-wine-trace] exec:", " ".join([qemu_path, wine_path] + argv)) -- cgit 1.4.1 From 2ba88dcd8a9829612eb06eb130d688685a76a847 Mon Sep 17 00:00:00 2001 From: van Hauser Date: Thu, 23 Jul 2020 14:17:29 +0200 Subject: skip large files, dont bail --- TODO.md | 2 ++ src/afl-fuzz-init.c | 5 +++-- 2 files changed, 5 insertions(+), 2 deletions(-) diff --git a/TODO.md b/TODO.md index 2723657a..341f2c78 100644 --- a/TODO.md +++ b/TODO.md @@ -2,6 +2,8 @@ ## Roadmap 2.67+ + - pre_save custom module example to save away test cases + - expand on AFL_LLVM_INSTRUMENT_FILE to also support sancov allowlist format - allow to sync against honggfuzz and libfuzzer - AFL_MAP_SIZE for qemu_mode and unicorn_mode - namespace for targets? e.g. network diff --git a/src/afl-fuzz-init.c b/src/afl-fuzz-init.c index e95ae95f..609e16ba 100644 --- a/src/afl-fuzz-init.c +++ b/src/afl-fuzz-init.c @@ -466,7 +466,7 @@ void read_testcases(afl_state_t *afl) { ACTF("Scanning '%s'...", afl->in_dir); /* We use scandir() + alphasort() rather than readdir() because otherwise, - the ordering of test cases would vary somewhat randomly and would be + the ordering of test cases would vary somewhat randomly and would be difficult to control. */ nl_cnt = scandir(afl->in_dir, &nl, NULL, alphasort); @@ -527,9 +527,10 @@ void read_testcases(afl_state_t *afl) { if (st.st_size > MAX_FILE) { - FATAL("Test case '%s' is too big (%s, limit is %s)", fn2, + WARNF("Test case '%s' is too big (%s, limit is %s), skipping", fn2, stringify_mem_size(val_buf[0], sizeof(val_buf[0]), st.st_size), stringify_mem_size(val_buf[1], sizeof(val_buf[1]), MAX_FILE)); + continue; } -- cgit 1.4.1 From 72b46a07d6a64a7871f029330bcf5eae649c8eb1 Mon Sep 17 00:00:00 2001 From: van Hauser Date: Thu, 23 Jul 2020 15:58:13 +0200 Subject: added honggfuzz custom mutator :) --- TODO.md | 1 + custom_mutators/honggfuzz/Makefile | 15 + custom_mutators/honggfuzz/README.md | 12 + custom_mutators/honggfuzz/common.h | 0 custom_mutators/honggfuzz/custom_mutator_helpers.h | 22 + custom_mutators/honggfuzz/honggfuzz.c | 141 +++ custom_mutators/honggfuzz/honggfuzz.h | 460 +++++++++ custom_mutators/honggfuzz/input.h | 106 ++ custom_mutators/honggfuzz/libhfcommon | 1 + custom_mutators/honggfuzz/log.h | 1 + custom_mutators/honggfuzz/mangle.c | 1039 ++++++++++++++++++++ custom_mutators/honggfuzz/mangle.h | 32 + custom_mutators/honggfuzz/util.h | 1 + docs/Changelog.md | 1 + 14 files changed, 1832 insertions(+) create mode 100644 custom_mutators/honggfuzz/Makefile create mode 100644 custom_mutators/honggfuzz/README.md create mode 100644 custom_mutators/honggfuzz/common.h create mode 100644 custom_mutators/honggfuzz/custom_mutator_helpers.h create mode 100644 custom_mutators/honggfuzz/honggfuzz.c create mode 100644 custom_mutators/honggfuzz/honggfuzz.h create mode 100644 custom_mutators/honggfuzz/input.h create mode 120000 custom_mutators/honggfuzz/libhfcommon create mode 120000 custom_mutators/honggfuzz/log.h create mode 100644 custom_mutators/honggfuzz/mangle.c create mode 100644 custom_mutators/honggfuzz/mangle.h create mode 120000 custom_mutators/honggfuzz/util.h diff --git a/TODO.md b/TODO.md index 341f2c78..ad3ef83e 100644 --- a/TODO.md +++ b/TODO.md @@ -2,6 +2,7 @@ ## Roadmap 2.67+ + - -i - + foreign fuzzer sync support: scandir with time sort - pre_save custom module example to save away test cases - expand on AFL_LLVM_INSTRUMENT_FILE to also support sancov allowlist format - allow to sync against honggfuzz and libfuzzer diff --git a/custom_mutators/honggfuzz/Makefile b/custom_mutators/honggfuzz/Makefile new file mode 100644 index 00000000..2f46d0e7 --- /dev/null +++ b/custom_mutators/honggfuzz/Makefile @@ -0,0 +1,15 @@ + +CFLAGS = -O3 -funroll-loops -fPIC -Wl,-Bsymbolic + +all: honggfuzz.so + +honggfuzz.so: honggfuzz.c input.h mangle.c ../../src/afl-performance.c + $(CC) $(CFLAGS) -I../../include -I. -shared -o honggfuzz.so honggfuzz.c mangle.c ../../src/afl-performance.c + +update: + wget --unlink https://github.com/google/honggfuzz/raw/master/mangle.c + wget --unlink https://github.com/google/honggfuzz/raw/master/mangle.h + wget --unlink https://github.com/google/honggfuzz/raw/master/honggfuzz.h + +clean: + rm -f *.o *~ *.so core diff --git a/custom_mutators/honggfuzz/README.md b/custom_mutators/honggfuzz/README.md new file mode 100644 index 00000000..8824976f --- /dev/null +++ b/custom_mutators/honggfuzz/README.md @@ -0,0 +1,12 @@ +# custum mutator: honggfuzz mangle + +this is the very good honggfuzz mutator in mangle.c as a custom mutator +module for afl++. It is the original mangle.c, mangle.h and honggfuzz.h +with a lot of mocking around it :-) + +just type `make` to build + +```AFL_CUSTOM_MUTATOR_LIBRARY=custom_mutators/honggfuzz/honggfuzz.so afl-fuzz ...``` + +> Original repository: https://github.com/google/honggfuzz +> Source commit: d0fbcb0373c32436b8fb922e6937da93b17291f5 diff --git a/custom_mutators/honggfuzz/common.h b/custom_mutators/honggfuzz/common.h new file mode 100644 index 00000000..e69de29b diff --git a/custom_mutators/honggfuzz/custom_mutator_helpers.h b/custom_mutators/honggfuzz/custom_mutator_helpers.h new file mode 100644 index 00000000..57754697 --- /dev/null +++ b/custom_mutators/honggfuzz/custom_mutator_helpers.h @@ -0,0 +1,22 @@ +#ifndef CUSTOM_MUTATOR_HELPERS +#define CUSTOM_MUTATOR_HELPERS + +#include "config.h" +#include "types.h" +#include "afl-fuzz.h" +#include + +#define INITIAL_GROWTH_SIZE (64) + +/* Use in a struct: creates a name_buf and a name_size variable. */ +#define BUF_VAR(type, name) \ + type * name##_buf; \ + size_t name##_size; +/* this filles in `&structptr->something_buf, &structptr->something_size`. */ +#define BUF_PARAMS(struct, name) \ + (void **)&struct->name##_buf, &struct->name##_size + +#undef INITIAL_GROWTH_SIZE + +#endif + diff --git a/custom_mutators/honggfuzz/honggfuzz.c b/custom_mutators/honggfuzz/honggfuzz.c new file mode 100644 index 00000000..368741c1 --- /dev/null +++ b/custom_mutators/honggfuzz/honggfuzz.c @@ -0,0 +1,141 @@ +#include +#include +#include +#include + +#include "custom_mutator_helpers.h" +#include "mangle.h" + +#define NUMBER_OF_MUTATIONS 5 + +uint8_t * queue_input; +size_t queue_input_size; +afl_state_t * afl_struct; +run_t run; +honggfuzz_t global; +struct _dynfile_t dynfile; + +typedef struct my_mutator { + + afl_state_t *afl; + run_t * run; + u8 * mutator_buf; + unsigned int seed; + unsigned int extras_cnt, a_extras_cnt; + +} my_mutator_t; + +my_mutator_t *afl_custom_init(afl_state_t *afl, unsigned int seed) { + + my_mutator_t *data = calloc(1, sizeof(my_mutator_t)); + if (!data) { + + perror("afl_custom_init alloc"); + return NULL; + + } + + if ((data->mutator_buf = malloc(MAX_FILE)) == NULL) { + + perror("mutator_buf alloc"); + return NULL; + + } + + run.dynfile = &dynfile; + run.global = &global; + data->afl = afl; + data->seed = seed; + data->run = &run; + afl_struct = afl; + + run.global->mutate.maxInputSz = MAX_FILE; + run.global->mutate.mutationsPerRun = NUMBER_OF_MUTATIONS; + run.mutationsPerRun = NUMBER_OF_MUTATIONS; + run.global->timing.lastCovUpdate = 6; + + // global->feedback.cmpFeedback + // global->feedback.cmpFeedbackMap + + return data; + +} + +/* When a new queue entry is added we check if there are new dictionary + entries to add to honggfuzz structure */ + +void afl_custom_queue_new_entry(my_mutator_t * data, + const uint8_t *filename_new_queue, + const uint8_t *filename_orig_queue) { + + while (data->extras_cnt < data->afl->extras_cnt && + run.global->mutate.dictionaryCnt < 1024) { + + memcpy(run.global->mutate.dictionary[run.global->mutate.dictionaryCnt].val, + data->afl->extras[data->extras_cnt].data, + data->afl->extras[data->extras_cnt].len); + run.global->mutate.dictionary[run.global->mutate.dictionaryCnt].len = + data->afl->extras[data->extras_cnt].len; + run.global->mutate.dictionaryCnt++; + data->extras_cnt++; + + } + + while (data->extras_cnt < data->afl->a_extras_cnt && + run.global->mutate.dictionaryCnt < 1024) { + + memcpy(run.global->mutate.dictionary[run.global->mutate.dictionaryCnt].val, + data->afl->a_extras[data->a_extras_cnt].data, + data->afl->a_extras[data->a_extras_cnt].len); + run.global->mutate.dictionary[run.global->mutate.dictionaryCnt].len = + data->afl->a_extras[data->a_extras_cnt].len; + run.global->mutate.dictionaryCnt++; + data->a_extras_cnt++; + + } + +} + +/* we could set only_printable if is_ascii is set ... let's see +uint8_t afl_custom_queue_get(void *data, const uint8_t *filename) { + + //run.global->cfg.only_printable = ... + +} + +*/ + +/* here we run the honggfuzz mutator, which is really good */ + +size_t afl_custom_fuzz(my_mutator_t *data, uint8_t *buf, size_t buf_size, + u8 **out_buf, uint8_t *add_buf, size_t add_buf_size, + size_t max_size) { + + /* set everything up, costly ... :( */ + memcpy(data->mutator_buf, buf, buf_size); + queue_input = data->mutator_buf; + run.dynfile->data = data->mutator_buf; + queue_input_size = buf_size; + run.dynfile->size = buf_size; + *out_buf = data->mutator_buf; + + /* the mutation */ + mangle_mangleContent(&run, NUMBER_OF_MUTATIONS); + + /* return size of mutated data */ + return run.dynfile->size; + +} + +/** + * Deinitialize everything + * + * @param data The data ptr from afl_custom_init + */ +void afl_custom_deinit(my_mutator_t *data) { + + free(data->mutator_buf); + free(data); + +} + diff --git a/custom_mutators/honggfuzz/honggfuzz.h b/custom_mutators/honggfuzz/honggfuzz.h new file mode 100644 index 00000000..4e045272 --- /dev/null +++ b/custom_mutators/honggfuzz/honggfuzz.h @@ -0,0 +1,460 @@ +/* + * + * honggfuzz - core structures and macros + * ----------------------------------------- + * + * Author: Robert Swiecki + * + * Copyright 2010-2018 by 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 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or + * implied. See the License for the specific language governing + * permissions and limitations under the License. + * + */ + +#ifndef _HF_HONGGFUZZ_H_ +#define _HF_HONGGFUZZ_H_ + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include "libhfcommon/util.h" + +#define PROG_NAME "honggfuzz" +#define PROG_VERSION "2.2" + +/* Name of the template which will be replaced with the proper name of the file + */ +#define _HF_FILE_PLACEHOLDER "___FILE___" + +/* Default name of the report created with some architectures */ +#define _HF_REPORT_FILE "HONGGFUZZ.REPORT.TXT" + +/* Default stack-size of created threads. */ +#define _HF_PTHREAD_STACKSIZE (1024ULL * 1024ULL * 2ULL) /* 2MB */ + +/* Name of envvar which indicates sequential number of fuzzer */ +#define _HF_THREAD_NO_ENV "HFUZZ_THREAD_NO" + +/* Name of envvar which indicates that the netDriver should be used */ +#define _HF_THREAD_NETDRIVER_ENV "HFUZZ_USE_NETDRIVER" + +/* Name of envvar which indicates honggfuzz's log level in use */ +#define _HF_LOG_LEVEL_ENV "HFUZZ_LOG_LEVEL" + +/* Number of crash verifier iterations before tag crash as stable */ +#define _HF_VERIFIER_ITER 5 + +/* Size (in bytes) for report data to be stored in stack before written to file + */ +#define _HF_REPORT_SIZE 32768 + +/* Perf bitmap size */ +#define _HF_PERF_BITMAP_SIZE_16M (1024U * 1024U * 16U) +#define _HF_PERF_BITMAP_BITSZ_MASK 0x7FFFFFFULL +/* Maximum number of PC guards (=trace-pc-guard) we support */ +#define _HF_PC_GUARD_MAX (1024ULL * 1024ULL * 64ULL) + +/* Maximum size of the input file in bytes (1 MiB) */ +#define _HF_INPUT_MAX_SIZE (1024ULL * 1024ULL) + +/* Default maximum size of produced inputs */ +#define _HF_INPUT_DEFAULT_SIZE (1024ULL * 8) + +/* Per-thread bitmap */ +#define _HF_PERTHREAD_BITMAP_FD 1018 +/* FD used to report back used int/str constants from the fuzzed process */ +#define _HF_CMP_BITMAP_FD 1019 +/* FD used to log inside the child process */ +#define _HF_LOG_FD 1020 +/* FD used to represent the input file */ +#define _HF_INPUT_FD 1021 +/* FD used to pass coverage feedback from the fuzzed process */ +#define _HF_COV_BITMAP_FD 1022 +#define _HF_BITMAP_FD _HF_COV_BITMAP_FD /* Old name for _HF_COV_BITMAP_FD */ +/* FD used to pass data to a persistent process */ +#define _HF_PERSISTENT_FD 1023 + +/* Input file as a string */ +#define _HF_INPUT_FILE_PATH "/dev/fd/" HF_XSTR(_HF_INPUT_FD) + +/* Maximum number of supported execve() args */ +#define _HF_ARGS_MAX 2048 + +/* Message indicating that the fuzzed process is ready for new data */ +static const uint8_t HFReadyTag = 'R'; + +/* Maximum number of active fuzzing threads */ +#define _HF_THREAD_MAX 1024U + +/* Persistent-binary signature - if found within file, it means it's a + * persistent mode binary */ +#define _HF_PERSISTENT_SIG "\x01_LIBHFUZZ_PERSISTENT_BINARY_SIGNATURE_\x02\xFF" +/* HF NetDriver signature - if found within file, it means it's a + * NetDriver-based binary */ +#define _HF_NETDRIVER_SIG "\x01_LIBHFUZZ_NETDRIVER_BINARY_SIGNATURE_\x02\xFF" + +/* printf() nonmonetary separator. According to MacOSX's man it's supported + * there as well */ +#define _HF_NONMON_SEP "'" + +typedef enum { + + _HF_DYNFILE_NONE = 0x0, + _HF_DYNFILE_INSTR_COUNT = 0x1, + _HF_DYNFILE_BRANCH_COUNT = 0x2, + _HF_DYNFILE_BTS_EDGE = 0x10, + _HF_DYNFILE_IPT_BLOCK = 0x20, + _HF_DYNFILE_SOFT = 0x40, + +} dynFileMethod_t; + +typedef struct { + + uint64_t cpuInstrCnt; + uint64_t cpuBranchCnt; + uint64_t bbCnt; + uint64_t newBBCnt; + uint64_t softCntPc; + uint64_t softCntEdge; + uint64_t softCntCmp; + +} hwcnt_t; + +typedef enum { + + _HF_STATE_UNSET = 0, + _HF_STATE_STATIC, + _HF_STATE_DYNAMIC_DRY_RUN, + _HF_STATE_DYNAMIC_MAIN, + _HF_STATE_DYNAMIC_MINIMIZE, + +} fuzzState_t; + +typedef enum { + + HF_MAYBE = -1, + HF_NO = 0, + HF_YES = 1, + +} tristate_t; + +struct _dynfile_t { + + size_t size; + uint64_t cov[4]; + size_t idx; + int fd; + uint64_t timeExecUSecs; + char path[PATH_MAX]; + struct _dynfile_t *src; + uint32_t refs; + uint8_t * data; + TAILQ_ENTRY(_dynfile_t) pointers; + +}; + +typedef struct _dynfile_t dynfile_t; + +struct strings_t { + + size_t len; + TAILQ_ENTRY(strings_t) pointers; + char s[]; + +}; + +typedef struct { + + uint8_t pcGuardMap[_HF_PC_GUARD_MAX]; + uint8_t bbMapPc[_HF_PERF_BITMAP_SIZE_16M]; + uint32_t bbMapCmp[_HF_PERF_BITMAP_SIZE_16M]; + uint64_t pidNewPC[_HF_THREAD_MAX]; + uint64_t pidNewEdge[_HF_THREAD_MAX]; + uint64_t pidNewCmp[_HF_THREAD_MAX]; + uint64_t guardNb; + uint64_t pidTotalPC[_HF_THREAD_MAX]; + uint64_t pidTotalEdge[_HF_THREAD_MAX]; + uint64_t pidTotalCmp[_HF_THREAD_MAX]; + +} feedback_t; + +typedef struct { + + uint32_t cnt; + struct { + + uint8_t val[32]; + uint32_t len; + + } valArr[1024 * 16]; + +} cmpfeedback_t; + +typedef struct { + + struct { + + size_t threadsMax; + size_t threadsFinished; + uint32_t threadsActiveCnt; + pthread_t mainThread; + pid_t mainPid; + pthread_t threads[_HF_THREAD_MAX]; + + } threads; + + struct { + + const char *inputDir; + const char *outputDir; + DIR * inputDirPtr; + size_t fileCnt; + size_t testedFileCnt; + const char *fileExtn; + size_t maxFileSz; + size_t newUnitsAdded; + char workDir[PATH_MAX]; + const char *crashDir; + const char *covDirNew; + bool saveUnique; + size_t dynfileqMaxSz; + size_t dynfileqCnt; + dynfile_t * dynfileqCurrent; + dynfile_t * dynfileq2Current; + TAILQ_HEAD(dyns_t, _dynfile_t) dynfileq; + bool exportFeedback; + + } io; + + struct { + + int argc; + const char *const *cmdline; + bool nullifyStdio; + bool fuzzStdin; + const char * externalCommand; + const char * postExternalCommand; + const char * feedbackMutateCommand; + bool netDriver; + bool persistent; + uint64_t asLimit; + uint64_t rssLimit; + uint64_t dataLimit; + uint64_t coreLimit; + uint64_t stackLimit; + bool clearEnv; + char * env_ptrs[128]; + char env_vals[128][4096]; + sigset_t waitSigSet; + + } exe; + + struct { + + time_t timeStart; + time_t runEndTime; + time_t tmOut; + time_t lastCovUpdate; + int64_t timeOfLongestUnitUSecs; + bool tmoutVTALRM; + + } timing; + + struct { + + struct { + + uint8_t val[256]; + size_t len; + + } dictionary[1024]; + + size_t dictionaryCnt; + const char *dictionaryFile; + size_t mutationsMax; + unsigned mutationsPerRun; + size_t maxInputSz; + + } mutate; + + struct { + + bool useScreen; + char cmdline_txt[65]; + int64_t lastDisplayUSecs; + + } display; + + struct { + + bool useVerifier; + bool exitUponCrash; + const char *reportFile; + size_t dynFileIterExpire; + bool only_printable; + bool minimize; + bool switchingToFDM; + + } cfg; + + struct { + + bool enable; + bool del_report; + + } sanitizer; + + struct { + + fuzzState_t state; + feedback_t * covFeedbackMap; + int covFeedbackFd; + cmpfeedback_t * cmpFeedbackMap; + int cmpFeedbackFd; + bool cmpFeedback; + const char * blacklistFile; + uint64_t * blacklist; + size_t blacklistCnt; + bool skipFeedbackOnTimeout; + uint64_t maxCov[4]; + dynFileMethod_t dynFileMethod; + hwcnt_t hwCnts; + + } feedback; + + struct { + + size_t mutationsCnt; + size_t crashesCnt; + size_t uniqueCrashesCnt; + size_t verifiedCrashesCnt; + size_t blCrashesCnt; + size_t timeoutedCnt; + + } cnts; + + struct { + + bool enabled; + int serverSocket; + int clientSocket; + + } socketFuzzer; + + struct { + + pthread_rwlock_t dynfileq; + pthread_mutex_t feedback; + pthread_mutex_t report; + pthread_mutex_t state; + pthread_mutex_t input; + pthread_mutex_t timing; + + } mutex; + + /* For the Linux code */ + struct { + + int exeFd; + uint64_t dynamicCutOffAddr; + bool disableRandomization; + void * ignoreAddr; + const char *symsBlFile; + char ** symsBl; + size_t symsBlCnt; + const char *symsWlFile; + char ** symsWl; + size_t symsWlCnt; + uintptr_t cloneFlags; + tristate_t useNetNs; + bool kernelOnly; + bool useClone; + + } arch_linux; + + /* For the NetBSD code */ + struct { + + void * ignoreAddr; + const char *symsBlFile; + char ** symsBl; + size_t symsBlCnt; + const char *symsWlFile; + char ** symsWl; + size_t symsWlCnt; + + } arch_netbsd; + +} honggfuzz_t; + +typedef enum { + + _HF_RS_UNKNOWN = 0, + _HF_RS_WAITING_FOR_INITIAL_READY = 1, + _HF_RS_WAITING_FOR_READY = 2, + _HF_RS_SEND_DATA = 3, + +} runState_t; + +typedef struct { + + honggfuzz_t *global; + pid_t pid; + int64_t timeStartedUSecs; + char crashFileName[PATH_MAX]; + uint64_t pc; + uint64_t backtrace; + uint64_t access; + int exception; + char report[_HF_REPORT_SIZE]; + bool mainWorker; + unsigned mutationsPerRun; + dynfile_t * dynfile; + bool staticFileTryMore; + uint32_t fuzzNo; + int persistentSock; + runState_t runState; + bool tmOutSignaled; + char * args[_HF_ARGS_MAX + 1]; + int perThreadCovFeedbackFd; + unsigned triesLeft; + dynfile_t * current; +#if !defined(_HF_ARCH_DARWIN) + timer_t timerId; +#endif // !defined(_HF_ARCH_DARWIN) + hwcnt_t hwCnts; + + struct { + + /* For Linux code */ + uint8_t *perfMmapBuf; + uint8_t *perfMmapAux; + int cpuInstrFd; + int cpuBranchFd; + int cpuIptBtsFd; + + } arch_linux; + +} run_t; + +#endif + diff --git a/custom_mutators/honggfuzz/input.h b/custom_mutators/honggfuzz/input.h new file mode 100644 index 00000000..c67d88a6 --- /dev/null +++ b/custom_mutators/honggfuzz/input.h @@ -0,0 +1,106 @@ +#ifndef _HG_INPUT_ +#define _HG_INPUT_ + +#include +#ifdef __clang__ +#include +#endif +#include +#include +#include + +#include "honggfuzz.h" +#include "afl-fuzz.h" + +/* + * Go-style defer scoped implementation + * + * If compiled with clang, use: -fblocks -lBlocksRuntime + * + * Example of use: + * + * { + * int fd = open(fname, O_RDONLY); + * if (fd == -1) { + * error(....); + * return; + * } + * defer { close(fd); }; + * ssize_t sz = read(fd, buf, sizeof(buf)); + * ... + * ... + * } + * + */ + +#define __STRMERGE(a, b) a##b +#define _STRMERGE(a, b) __STRMERGE(a, b) +#ifdef __clang__ +#if __has_extension(blocks) +static void __attribute__((unused)) __clang_cleanup_func(void (^*dfunc)(void)) { + (*dfunc)(); +} + +#define defer \ + void (^_STRMERGE(__defer_f_, __COUNTER__))(void) \ + __attribute__((cleanup(__clang_cleanup_func))) __attribute__((unused)) = ^ + +#else /* __has_extension(blocks) */ +#define defer UNIMPLEMENTED - NO - SUPPORT - FOR - BLOCKS - IN - YOUR - CLANG - ENABLED +#endif /* __has_extension(blocks) */ +#else /* !__clang__, e.g.: gcc */ + +#define __block +#define _DEFER(a, count) \ + auto void _STRMERGE(__defer_f_, count)(void* _defer_arg __attribute__((unused))); \ + int _STRMERGE(__defer_var_, count) __attribute__((cleanup(_STRMERGE(__defer_f_, count)))) \ + __attribute__((unused)); \ + void _STRMERGE(__defer_f_, count)(void* _defer_arg __attribute__((unused))) +#define defer _DEFER(a, __COUNTER__) +#endif /* ifdef __clang__ */ + +#define HF_MIN(x, y) (x <= y ? x : y) +#define HF_MAX(x, y) (x >= y ? x : y) +#define ATOMIC_GET +#define ARRAYSIZE(x) (sizeof(x) / sizeof(*x)) +#define HF_ATTR_UNUSED __attribute__((unused)) +#define util_Malloc(x) malloc(x) + +extern uint8_t * queue_input; +extern size_t queue_input_size; +extern afl_state_t * afl_struct; + +inline void wmb() { } +inline void LOG_F(const char *format, ...) { } +static inline uint64_t util_rndGet(uint64_t min, uint64_t max) { + return min + rand_below(afl_struct, max - min + 1); +} +static inline uint64_t util_rnd64() { return rand_below(afl_struct, 1 << 30); } + +static inline size_t input_getRandomInputAsBuf(run_t *run, const uint8_t **buf) { + *buf = queue_input; + run->dynfile->data = queue_input; + run->dynfile->size = queue_input_size; + return queue_input_size; +} +static inline void input_setSize(run_t* run, size_t sz) { + run->dynfile->size = sz; +} +static inline uint8_t util_turnToPrintable(uint8_t* buf, size_t sz) { + for (size_t i = 0; i < sz; i++) + buf[i] = buf[i] % 95 + 32; +} +static inline void util_rndBuf(uint8_t* buf, size_t sz) { + if (sz == 0) return; + for (size_t i = 0; i < sz; i++) + buf[i] = (uint8_t)rand_below(afl_struct, 256); +} +static inline uint8_t util_rndPrintable() { + return 32 + rand_below(afl_struct, 127 - 32); +} +static inline void util_rndBufPrintable(uint8_t* buf, size_t sz) { + for (size_t i = 0; i < sz; i++) + buf[i] = util_rndPrintable(); +} + +#endif diff --git a/custom_mutators/honggfuzz/libhfcommon b/custom_mutators/honggfuzz/libhfcommon new file mode 120000 index 00000000..945c9b46 --- /dev/null +++ b/custom_mutators/honggfuzz/libhfcommon @@ -0,0 +1 @@ +. \ No newline at end of file diff --git a/custom_mutators/honggfuzz/log.h b/custom_mutators/honggfuzz/log.h new file mode 120000 index 00000000..51e19654 --- /dev/null +++ b/custom_mutators/honggfuzz/log.h @@ -0,0 +1 @@ +common.h \ No newline at end of file diff --git a/custom_mutators/honggfuzz/mangle.c b/custom_mutators/honggfuzz/mangle.c new file mode 100644 index 00000000..05e0dcfa --- /dev/null +++ b/custom_mutators/honggfuzz/mangle.c @@ -0,0 +1,1039 @@ +/* + * + * honggfuzz - run->dynfile->datafer mangling routines + * ----------------------------------------- + * + * Author: + * Robert Swiecki + * + * Copyright 2010-2018 by 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 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or + * implied. See the License for the specific language governing + * permissions and limitations under the License. + * + */ + +#include "mangle.h" + +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include "input.h" +#include "libhfcommon/common.h" +#include "libhfcommon/log.h" +#include "libhfcommon/util.h" + +static inline size_t mangle_LenLeft(run_t *run, size_t off) { + + if (off >= run->dynfile->size) { + + LOG_F("Offset is too large: off:%zu >= len:%zu", off, run->dynfile->size); + + } + + return (run->dynfile->size - off - 1); + +} + +/* Get a random value between <1:max> with x^2 distribution */ +static inline size_t mangle_getLen(size_t max) { + + if (max > _HF_INPUT_MAX_SIZE) { + + LOG_F("max (%zu) > _HF_INPUT_MAX_SIZE (%zu)", max, + (size_t)_HF_INPUT_MAX_SIZE); + + } + + if (max == 0) { LOG_F("max == 0"); } + if (max == 1) { return 1; } + + const uint64_t max2 = (uint64_t)max * max; + const uint64_t max3 = (uint64_t)max * max * max; + const uint64_t rnd = util_rndGet(1, max2 - 1); + + uint64_t ret = rnd * rnd; + ret /= max3; + ret += 1; + + if (ret < 1) { + + LOG_F("ret (%" PRIu64 ") < 1, max:%zu, rnd:%" PRIu64, ret, max, rnd); + + } + + if (ret > max) { + + LOG_F("ret (%" PRIu64 ") > max (%zu), rnd:%" PRIu64, ret, max, rnd); + + } + + return (size_t)ret; + +} + +/* Prefer smaller values here, so use mangle_getLen() */ +static inline size_t mangle_getOffSet(run_t *run) { + + return mangle_getLen(run->dynfile->size) - 1; + +} + +/* Offset which can be equal to the file size */ +static inline size_t mangle_getOffSetPlus1(run_t *run) { + + size_t reqlen = HF_MIN(run->dynfile->size + 1, _HF_INPUT_MAX_SIZE); + return mangle_getLen(reqlen) - 1; + +} + +static inline void mangle_Move(run_t *run, size_t off_from, size_t off_to, + size_t len) { + + if (off_from >= run->dynfile->size) { return; } + if (off_to >= run->dynfile->size) { return; } + if (off_from == off_to) { return; } + + size_t len_from = run->dynfile->size - off_from; + len = HF_MIN(len, len_from); + + size_t len_to = run->dynfile->size - off_to; + len = HF_MIN(len, len_to); + + memmove(&run->dynfile->data[off_to], &run->dynfile->data[off_from], len); + +} + +static inline void mangle_Overwrite(run_t *run, size_t off, const uint8_t *src, + size_t len, bool printable) { + + if (len == 0) { return; } + size_t maxToCopy = run->dynfile->size - off; + if (len > maxToCopy) { len = maxToCopy; } + + memmove(&run->dynfile->data[off], src, len); + if (printable) { util_turnToPrintable(&run->dynfile->data[off], len); } + +} + +static inline size_t mangle_Inflate(run_t *run, size_t off, size_t len, + bool printable) { + + if (run->dynfile->size >= run->global->mutate.maxInputSz) { return 0; } + if (len > (run->global->mutate.maxInputSz - run->dynfile->size)) { + + len = run->global->mutate.maxInputSz - run->dynfile->size; + + } + + input_setSize(run, run->dynfile->size + len); + mangle_Move(run, off, off + len, run->dynfile->size); + if (printable) { memset(&run->dynfile->data[off], ' ', len); } + + return len; + +} + +static inline void mangle_Insert(run_t *run, size_t off, const uint8_t *val, + size_t len, bool printable) { + + len = mangle_Inflate(run, off, len, printable); + mangle_Overwrite(run, off, val, len, printable); + +} + +static inline void mangle_UseValue(run_t *run, const uint8_t *val, size_t len, + bool printable) { + + if (util_rnd64() % 2) { + + mangle_Insert(run, mangle_getOffSetPlus1(run), val, len, printable); + + } else { + + mangle_Overwrite(run, mangle_getOffSet(run), val, len, printable); + + } + +} + +static void mangle_MemSwap(run_t *run, bool printable HF_ATTR_UNUSED) { + + size_t off1 = mangle_getOffSet(run); + size_t maxlen1 = run->dynfile->size - off1; + + size_t off2 = mangle_getOffSet(run); + size_t maxlen2 = run->dynfile->size - off2; + + size_t len = mangle_getLen(HF_MIN(maxlen1, maxlen2)); + uint8_t *tmpbuf = (uint8_t *)util_Malloc(len); + defer { + + free(tmpbuf); + + }; + + memcpy(tmpbuf, &run->dynfile->data[off1], len); + memmove(&run->dynfile->data[off1], &run->dynfile->data[off2], len); + memcpy(&run->dynfile->data[off2], tmpbuf, len); + +} + +static void mangle_MemCopy(run_t *run, bool printable HF_ATTR_UNUSED) { + + size_t off = mangle_getOffSet(run); + size_t len = mangle_getLen(run->dynfile->size - off); + + /* Use a temp buf, as Insert/Inflate can change source bytes */ + uint8_t *tmpbuf = (uint8_t *)util_Malloc(len); + defer { + + free(tmpbuf); + + }; + + memcpy(tmpbuf, &run->dynfile->data[off], len); + + mangle_UseValue(run, tmpbuf, len, printable); + +} + +static void mangle_Bytes(run_t *run, bool printable) { + + uint16_t buf; + if (printable) { + + util_rndBufPrintable((uint8_t *)&buf, sizeof(buf)); + + } else { + + buf = util_rnd64(); + + } + + /* Overwrite with random 1-2-byte values */ + size_t toCopy = util_rndGet(1, 2); + mangle_UseValue(run, (const uint8_t *)&buf, toCopy, printable); + +} + +static void mangle_ByteRepeatOverwrite(run_t *run, bool printable) { + + size_t off = mangle_getOffSet(run); + size_t destOff = off + 1; + size_t maxSz = run->dynfile->size - destOff; + + /* No space to repeat */ + if (!maxSz) { + + mangle_Bytes(run, printable); + return; + + } + + size_t len = mangle_getLen(maxSz); + memset(&run->dynfile->data[destOff], run->dynfile->data[off], len); + +} + +static void mangle_ByteRepeatInsert(run_t *run, bool printable) { + + size_t off = mangle_getOffSet(run); + size_t destOff = off + 1; + size_t maxSz = run->dynfile->size - destOff; + + /* No space to repeat */ + if (!maxSz) { + + mangle_Bytes(run, printable); + return; + + } + + size_t len = mangle_getLen(maxSz); + len = mangle_Inflate(run, destOff, len, printable); + memset(&run->dynfile->data[destOff], run->dynfile->data[off], len); + +} + +static void mangle_Bit(run_t *run, bool printable) { + + size_t off = mangle_getOffSet(run); + run->dynfile->data[off] ^= (uint8_t)(1U << util_rndGet(0, 7)); + if (printable) { util_turnToPrintable(&(run->dynfile->data[off]), 1); } + +} + +static const struct { + + const uint8_t val[8]; + const size_t size; + +} mangleMagicVals[] = { + + /* 1B - No endianness */ + {"\x00\x00\x00\x00\x00\x00\x00\x00", 1}, + {"\x01\x00\x00\x00\x00\x00\x00\x00", 1}, + {"\x02\x00\x00\x00\x00\x00\x00\x00", 1}, + {"\x03\x00\x00\x00\x00\x00\x00\x00", 1}, + {"\x04\x00\x00\x00\x00\x00\x00\x00", 1}, + {"\x05\x00\x00\x00\x00\x00\x00\x00", 1}, + {"\x06\x00\x00\x00\x00\x00\x00\x00", 1}, + {"\x07\x00\x00\x00\x00\x00\x00\x00", 1}, + {"\x08\x00\x00\x00\x00\x00\x00\x00", 1}, + {"\x09\x00\x00\x00\x00\x00\x00\x00", 1}, + {"\x0A\x00\x00\x00\x00\x00\x00\x00", 1}, + {"\x0B\x00\x00\x00\x00\x00\x00\x00", 1}, + {"\x0C\x00\x00\x00\x00\x00\x00\x00", 1}, + {"\x0D\x00\x00\x00\x00\x00\x00\x00", 1}, + {"\x0E\x00\x00\x00\x00\x00\x00\x00", 1}, + {"\x0F\x00\x00\x00\x00\x00\x00\x00", 1}, + {"\x10\x00\x00\x00\x00\x00\x00\x00", 1}, + {"\x20\x00\x00\x00\x00\x00\x00\x00", 1}, + {"\x40\x00\x00\x00\x00\x00\x00\x00", 1}, + {"\x7E\x00\x00\x00\x00\x00\x00\x00", 1}, + {"\x7F\x00\x00\x00\x00\x00\x00\x00", 1}, + {"\x80\x00\x00\x00\x00\x00\x00\x00", 1}, + {"\x81\x00\x00\x00\x00\x00\x00\x00", 1}, + {"\xC0\x00\x00\x00\x00\x00\x00\x00", 1}, + {"\xFE\x00\x00\x00\x00\x00\x00\x00", 1}, + {"\xFF\x00\x00\x00\x00\x00\x00\x00", 1}, + /* 2B - NE */ + {"\x00\x00\x00\x00\x00\x00\x00\x00", 2}, + {"\x01\x01\x00\x00\x00\x00\x00\x00", 2}, + {"\x80\x80\x00\x00\x00\x00\x00\x00", 2}, + {"\xFF\xFF\x00\x00\x00\x00\x00\x00", 2}, + /* 2B - BE */ + {"\x00\x01\x00\x00\x00\x00\x00\x00", 2}, + {"\x00\x02\x00\x00\x00\x00\x00\x00", 2}, + {"\x00\x03\x00\x00\x00\x00\x00\x00", 2}, + {"\x00\x04\x00\x00\x00\x00\x00\x00", 2}, + {"\x00\x05\x00\x00\x00\x00\x00\x00", 2}, + {"\x00\x06\x00\x00\x00\x00\x00\x00", 2}, + {"\x00\x07\x00\x00\x00\x00\x00\x00", 2}, + {"\x00\x08\x00\x00\x00\x00\x00\x00", 2}, + {"\x00\x09\x00\x00\x00\x00\x00\x00", 2}, + {"\x00\x0A\x00\x00\x00\x00\x00\x00", 2}, + {"\x00\x0B\x00\x00\x00\x00\x00\x00", 2}, + {"\x00\x0C\x00\x00\x00\x00\x00\x00", 2}, + {"\x00\x0D\x00\x00\x00\x00\x00\x00", 2}, + {"\x00\x0E\x00\x00\x00\x00\x00\x00", 2}, + {"\x00\x0F\x00\x00\x00\x00\x00\x00", 2}, + {"\x00\x10\x00\x00\x00\x00\x00\x00", 2}, + {"\x00\x20\x00\x00\x00\x00\x00\x00", 2}, + {"\x00\x40\x00\x00\x00\x00\x00\x00", 2}, + {"\x00\x7E\x00\x00\x00\x00\x00\x00", 2}, + {"\x00\x7F\x00\x00\x00\x00\x00\x00", 2}, + {"\x00\x80\x00\x00\x00\x00\x00\x00", 2}, + {"\x00\x81\x00\x00\x00\x00\x00\x00", 2}, + {"\x00\xC0\x00\x00\x00\x00\x00\x00", 2}, + {"\x00\xFE\x00\x00\x00\x00\x00\x00", 2}, + {"\x00\xFF\x00\x00\x00\x00\x00\x00", 2}, + {"\x7E\xFF\x00\x00\x00\x00\x00\x00", 2}, + {"\x7F\xFF\x00\x00\x00\x00\x00\x00", 2}, + {"\x80\x00\x00\x00\x00\x00\x00\x00", 2}, + {"\x80\x01\x00\x00\x00\x00\x00\x00", 2}, + {"\xFF\xFE\x00\x00\x00\x00\x00\x00", 2}, + /* 2B - LE */ + {"\x00\x00\x00\x00\x00\x00\x00\x00", 2}, + {"\x01\x00\x00\x00\x00\x00\x00\x00", 2}, + {"\x02\x00\x00\x00\x00\x00\x00\x00", 2}, + {"\x03\x00\x00\x00\x00\x00\x00\x00", 2}, + {"\x04\x00\x00\x00\x00\x00\x00\x00", 2}, + {"\x05\x00\x00\x00\x00\x00\x00\x00", 2}, + {"\x06\x00\x00\x00\x00\x00\x00\x00", 2}, + {"\x07\x00\x00\x00\x00\x00\x00\x00", 2}, + {"\x08\x00\x00\x00\x00\x00\x00\x00", 2}, + {"\x09\x00\x00\x00\x00\x00\x00\x00", 2}, + {"\x0A\x00\x00\x00\x00\x00\x00\x00", 2}, + {"\x0B\x00\x00\x00\x00\x00\x00\x00", 2}, + {"\x0C\x00\x00\x00\x00\x00\x00\x00", 2}, + {"\x0D\x00\x00\x00\x00\x00\x00\x00", 2}, + {"\x0E\x00\x00\x00\x00\x00\x00\x00", 2}, + {"\x0F\x00\x00\x00\x00\x00\x00\x00", 2}, + {"\x10\x00\x00\x00\x00\x00\x00\x00", 2}, + {"\x20\x00\x00\x00\x00\x00\x00\x00", 2}, + {"\x40\x00\x00\x00\x00\x00\x00\x00", 2}, + {"\x7E\x00\x00\x00\x00\x00\x00\x00", 2}, + {"\x7F\x00\x00\x00\x00\x00\x00\x00", 2}, + {"\x80\x00\x00\x00\x00\x00\x00\x00", 2}, + {"\x81\x00\x00\x00\x00\x00\x00\x00", 2}, + {"\xC0\x00\x00\x00\x00\x00\x00\x00", 2}, + {"\xFE\x00\x00\x00\x00\x00\x00\x00", 2}, + {"\xFF\x00\x00\x00\x00\x00\x00\x00", 2}, + {"\xFF\x7E\x00\x00\x00\x00\x00\x00", 2}, + {"\xFF\x7F\x00\x00\x00\x00\x00\x00", 2}, + {"\x00\x80\x00\x00\x00\x00\x00\x00", 2}, + {"\x01\x80\x00\x00\x00\x00\x00\x00", 2}, + {"\xFE\xFF\x00\x00\x00\x00\x00\x00", 2}, + /* 4B - NE */ + {"\x00\x00\x00\x00\x00\x00\x00\x00", 4}, + {"\x01\x01\x01\x01\x00\x00\x00\x00", 4}, + {"\x80\x80\x80\x80\x00\x00\x00\x00", 4}, + {"\xFF\xFF\xFF\xFF\x00\x00\x00\x00", 4}, + /* 4B - BE */ + {"\x00\x00\x00\x01\x00\x00\x00\x00", 4}, + {"\x00\x00\x00\x02\x00\x00\x00\x00", 4}, + {"\x00\x00\x00\x03\x00\x00\x00\x00", 4}, + {"\x00\x00\x00\x04\x00\x00\x00\x00", 4}, + {"\x00\x00\x00\x05\x00\x00\x00\x00", 4}, + {"\x00\x00\x00\x06\x00\x00\x00\x00", 4}, + {"\x00\x00\x00\x07\x00\x00\x00\x00", 4}, + {"\x00\x00\x00\x08\x00\x00\x00\x00", 4}, + {"\x00\x00\x00\x09\x00\x00\x00\x00", 4}, + {"\x00\x00\x00\x0A\x00\x00\x00\x00", 4}, + {"\x00\x00\x00\x0B\x00\x00\x00\x00", 4}, + {"\x00\x00\x00\x0C\x00\x00\x00\x00", 4}, + {"\x00\x00\x00\x0D\x00\x00\x00\x00", 4}, + {"\x00\x00\x00\x0E\x00\x00\x00\x00", 4}, + {"\x00\x00\x00\x0F\x00\x00\x00\x00", 4}, + {"\x00\x00\x00\x10\x00\x00\x00\x00", 4}, + {"\x00\x00\x00\x20\x00\x00\x00\x00", 4}, + {"\x00\x00\x00\x40\x00\x00\x00\x00", 4}, + {"\x00\x00\x00\x7E\x00\x00\x00\x00", 4}, + {"\x00\x00\x00\x7F\x00\x00\x00\x00", 4}, + {"\x00\x00\x00\x80\x00\x00\x00\x00", 4}, + {"\x00\x00\x00\x81\x00\x00\x00\x00", 4}, + {"\x00\x00\x00\xC0\x00\x00\x00\x00", 4}, + {"\x00\x00\x00\xFE\x00\x00\x00\x00", 4}, + {"\x00\x00\x00\xFF\x00\x00\x00\x00", 4}, + {"\x7E\xFF\xFF\xFF\x00\x00\x00\x00", 4}, + {"\x7F\xFF\xFF\xFF\x00\x00\x00\x00", 4}, + {"\x80\x00\x00\x00\x00\x00\x00\x00", 4}, + {"\x80\x00\x00\x01\x00\x00\x00\x00", 4}, + {"\xFF\xFF\xFF\xFE\x00\x00\x00\x00", 4}, + /* 4B - LE */ + {"\x00\x00\x00\x00\x00\x00\x00\x00", 4}, + {"\x01\x00\x00\x00\x00\x00\x00\x00", 4}, + {"\x02\x00\x00\x00\x00\x00\x00\x00", 4}, + {"\x03\x00\x00\x00\x00\x00\x00\x00", 4}, + {"\x04\x00\x00\x00\x00\x00\x00\x00", 4}, + {"\x05\x00\x00\x00\x00\x00\x00\x00", 4}, + {"\x06\x00\x00\x00\x00\x00\x00\x00", 4}, + {"\x07\x00\x00\x00\x00\x00\x00\x00", 4}, + {"\x08\x00\x00\x00\x00\x00\x00\x00", 4}, + {"\x09\x00\x00\x00\x00\x00\x00\x00", 4}, + {"\x0A\x00\x00\x00\x00\x00\x00\x00", 4}, + {"\x0B\x00\x00\x00\x00\x00\x00\x00", 4}, + {"\x0C\x00\x00\x00\x00\x00\x00\x00", 4}, + {"\x0D\x00\x00\x00\x00\x00\x00\x00", 4}, + {"\x0E\x00\x00\x00\x00\x00\x00\x00", 4}, + {"\x0F\x00\x00\x00\x00\x00\x00\x00", 4}, + {"\x10\x00\x00\x00\x00\x00\x00\x00", 4}, + {"\x20\x00\x00\x00\x00\x00\x00\x00", 4}, + {"\x40\x00\x00\x00\x00\x00\x00\x00", 4}, + {"\x7E\x00\x00\x00\x00\x00\x00\x00", 4}, + {"\x7F\x00\x00\x00\x00\x00\x00\x00", 4}, + {"\x80\x00\x00\x00\x00\x00\x00\x00", 4}, + {"\x81\x00\x00\x00\x00\x00\x00\x00", 4}, + {"\xC0\x00\x00\x00\x00\x00\x00\x00", 4}, + {"\xFE\x00\x00\x00\x00\x00\x00\x00", 4}, + {"\xFF\x00\x00\x00\x00\x00\x00\x00", 4}, + {"\xFF\xFF\xFF\x7E\x00\x00\x00\x00", 4}, + {"\xFF\xFF\xFF\x7F\x00\x00\x00\x00", 4}, + {"\x00\x00\x00\x80\x00\x00\x00\x00", 4}, + {"\x01\x00\x00\x80\x00\x00\x00\x00", 4}, + {"\xFE\xFF\xFF\xFF\x00\x00\x00\x00", 4}, + /* 8B - NE */ + {"\x00\x00\x00\x00\x00\x00\x00\x00", 8}, + {"\x01\x01\x01\x01\x01\x01\x01\x01", 8}, + {"\x80\x80\x80\x80\x80\x80\x80\x80", 8}, + {"\xFF\xFF\xFF\xFF\xFF\xFF\xFF\xFF", 8}, + /* 8B - BE */ + {"\x00\x00\x00\x00\x00\x00\x00\x01", 8}, + {"\x00\x00\x00\x00\x00\x00\x00\x02", 8}, + {"\x00\x00\x00\x00\x00\x00\x00\x03", 8}, + {"\x00\x00\x00\x00\x00\x00\x00\x04", 8}, + {"\x00\x00\x00\x00\x00\x00\x00\x05", 8}, + {"\x00\x00\x00\x00\x00\x00\x00\x06", 8}, + {"\x00\x00\x00\x00\x00\x00\x00\x07", 8}, + {"\x00\x00\x00\x00\x00\x00\x00\x08", 8}, + {"\x00\x00\x00\x00\x00\x00\x00\x09", 8}, + {"\x00\x00\x00\x00\x00\x00\x00\x0A", 8}, + {"\x00\x00\x00\x00\x00\x00\x00\x0B", 8}, + {"\x00\x00\x00\x00\x00\x00\x00\x0C", 8}, + {"\x00\x00\x00\x00\x00\x00\x00\x0D", 8}, + {"\x00\x00\x00\x00\x00\x00\x00\x0E", 8}, + {"\x00\x00\x00\x00\x00\x00\x00\x0F", 8}, + {"\x00\x00\x00\x00\x00\x00\x00\x10", 8}, + {"\x00\x00\x00\x00\x00\x00\x00\x20", 8}, + {"\x00\x00\x00\x00\x00\x00\x00\x40", 8}, + {"\x00\x00\x00\x00\x00\x00\x00\x7E", 8}, + {"\x00\x00\x00\x00\x00\x00\x00\x7F", 8}, + {"\x00\x00\x00\x00\x00\x00\x00\x80", 8}, + {"\x00\x00\x00\x00\x00\x00\x00\x81", 8}, + {"\x00\x00\x00\x00\x00\x00\x00\xC0", 8}, + {"\x00\x00\x00\x00\x00\x00\x00\xFE", 8}, + {"\x00\x00\x00\x00\x00\x00\x00\xFF", 8}, + {"\x7E\xFF\xFF\xFF\xFF\xFF\xFF\xFF", 8}, + {"\x7F\xFF\xFF\xFF\xFF\xFF\xFF\xFF", 8}, + {"\x80\x00\x00\x00\x00\x00\x00\x00", 8}, + {"\x80\x00\x00\x00\x00\x00\x00\x01", 8}, + {"\xFF\xFF\xFF\xFF\xFF\xFF\xFF\xFE", 8}, + /* 8B - LE */ + {"\x00\x00\x00\x00\x00\x00\x00\x00", 8}, + {"\x01\x00\x00\x00\x00\x00\x00\x00", 8}, + {"\x02\x00\x00\x00\x00\x00\x00\x00", 8}, + {"\x03\x00\x00\x00\x00\x00\x00\x00", 8}, + {"\x04\x00\x00\x00\x00\x00\x00\x00", 8}, + {"\x05\x00\x00\x00\x00\x00\x00\x00", 8}, + {"\x06\x00\x00\x00\x00\x00\x00\x00", 8}, + {"\x07\x00\x00\x00\x00\x00\x00\x00", 8}, + {"\x08\x00\x00\x00\x00\x00\x00\x00", 8}, + {"\x09\x00\x00\x00\x00\x00\x00\x00", 8}, + {"\x0A\x00\x00\x00\x00\x00\x00\x00", 8}, + {"\x0B\x00\x00\x00\x00\x00\x00\x00", 8}, + {"\x0C\x00\x00\x00\x00\x00\x00\x00", 8}, + {"\x0D\x00\x00\x00\x00\x00\x00\x00", 8}, + {"\x0E\x00\x00\x00\x00\x00\x00\x00", 8}, + {"\x0F\x00\x00\x00\x00\x00\x00\x00", 8}, + {"\x10\x00\x00\x00\x00\x00\x00\x00", 8}, + {"\x20\x00\x00\x00\x00\x00\x00\x00", 8}, + {"\x40\x00\x00\x00\x00\x00\x00\x00", 8}, + {"\x7E\x00\x00\x00\x00\x00\x00\x00", 8}, + {"\x7F\x00\x00\x00\x00\x00\x00\x00", 8}, + {"\x80\x00\x00\x00\x00\x00\x00\x00", 8}, + {"\x81\x00\x00\x00\x00\x00\x00\x00", 8}, + {"\xC0\x00\x00\x00\x00\x00\x00\x00", 8}, + {"\xFE\x00\x00\x00\x00\x00\x00\x00", 8}, + {"\xFF\x00\x00\x00\x00\x00\x00\x00", 8}, + {"\xFF\xFF\xFF\xFF\xFF\xFF\xFF\x7E", 8}, + {"\xFF\xFF\xFF\xFF\xFF\xFF\xFF\x7F", 8}, + {"\x00\x00\x00\x00\x00\x00\x00\x80", 8}, + {"\x01\x00\x00\x00\x00\x00\x00\x80", 8}, + {"\xFE\xFF\xFF\xFF\xFF\xFF\xFF\xFF", 8}, + +}; + +static void mangle_Magic(run_t *run, bool printable) { + + uint64_t choice = util_rndGet(0, ARRAYSIZE(mangleMagicVals) - 1); + mangle_UseValue(run, mangleMagicVals[choice].val, + mangleMagicVals[choice].size, printable); + +} + +static void mangle_StaticDict(run_t *run, bool printable) { + + if (run->global->mutate.dictionaryCnt == 0) { + + mangle_Bytes(run, printable); + return; + + } + + uint64_t choice = util_rndGet(0, run->global->mutate.dictionaryCnt - 1); + mangle_UseValue(run, run->global->mutate.dictionary[choice].val, + run->global->mutate.dictionary[choice].len, printable); + +} + +static inline const uint8_t *mangle_FeedbackDict(run_t *run, size_t *len) { + + if (!run->global->feedback.cmpFeedback) { return NULL; } + cmpfeedback_t *cmpf = run->global->feedback.cmpFeedbackMap; + uint32_t cnt = ATOMIC_GET(cmpf->cnt); + if (cnt == 0) { return NULL; } + if (cnt > ARRAYSIZE(cmpf->valArr)) { cnt = ARRAYSIZE(cmpf->valArr); } + uint32_t choice = util_rndGet(0, cnt - 1); + *len = (size_t)ATOMIC_GET(cmpf->valArr[choice].len); + if (*len == 0) { return NULL; } + return cmpf->valArr[choice].val; + +} + +static void mangle_ConstFeedbackDict(run_t *run, bool printable) { + + size_t len; + const uint8_t *val = mangle_FeedbackDict(run, &len); + if (val == NULL) { + + mangle_Bytes(run, printable); + return; + + } + + mangle_UseValue(run, val, len, printable); + +} + +static void mangle_MemSet(run_t *run, bool printable) { + + size_t off = mangle_getOffSet(run); + size_t len = mangle_getLen(run->dynfile->size - off); + int val = + printable ? (int)util_rndPrintable() : (int)util_rndGet(0, UINT8_MAX); + + memset(&run->dynfile->data[off], val, len); + +} + +static void mangle_RandomOverwrite(run_t *run, bool printable) { + + size_t off = mangle_getOffSet(run); + size_t len = mangle_getLen(run->dynfile->size - off); + if (printable) { + + util_rndBufPrintable(&run->dynfile->data[off], len); + + } else { + + util_rndBuf(&run->dynfile->data[off], len); + + } + +} + +static void mangle_RandomInsert(run_t *run, bool printable) { + + size_t off = mangle_getOffSet(run); + size_t len = mangle_getLen(run->dynfile->size - off); + + len = mangle_Inflate(run, off, len, printable); + + if (printable) { + + util_rndBufPrintable(&run->dynfile->data[off], len); + + } else { + + util_rndBuf(&run->dynfile->data[off], len); + + } + +} + +static inline void mangle_AddSubWithRange(run_t *run, size_t off, size_t varLen, + uint64_t range, bool printable) { + + int64_t delta = (int64_t)util_rndGet(0, range * 2) - (int64_t)range; + + switch (varLen) { + + case 1: { + + run->dynfile->data[off] += delta; + break; + + } + + case 2: { + + int16_t val; + memcpy(&val, &run->dynfile->data[off], sizeof(val)); + if (util_rnd64() & 0x1) { + + val += delta; + + } else { + + /* Foreign endianess */ + val = __builtin_bswap16(val); + val += delta; + val = __builtin_bswap16(val); + + } + + mangle_Overwrite(run, off, (uint8_t *)&val, varLen, printable); + break; + + } + + case 4: { + + int32_t val; + memcpy(&val, &run->dynfile->data[off], sizeof(val)); + if (util_rnd64() & 0x1) { + + val += delta; + + } else { + + /* Foreign endianess */ + val = __builtin_bswap32(val); + val += delta; + val = __builtin_bswap32(val); + + } + + mangle_Overwrite(run, off, (uint8_t *)&val, varLen, printable); + break; + + } + + case 8: { + + int64_t val; + memcpy(&val, &run->dynfile->data[off], sizeof(val)); + if (util_rnd64() & 0x1) { + + val += delta; + + } else { + + /* Foreign endianess */ + val = __builtin_bswap64(val); + val += delta; + val = __builtin_bswap64(val); + + } + + mangle_Overwrite(run, off, (uint8_t *)&val, varLen, printable); + break; + + } + + default: { + + LOG_F("Unknown variable length size: %zu", varLen); + + } + + } + +} + +static void mangle_AddSub(run_t *run, bool printable) { + + size_t off = mangle_getOffSet(run); + + /* 1,2,4,8 */ + size_t varLen = 1U << util_rndGet(0, 3); + if ((run->dynfile->size - off) < varLen) { varLen = 1; } + + uint64_t range; + switch (varLen) { + + case 1: + range = 16; + break; + case 2: + range = 4096; + break; + case 4: + range = 1048576; + break; + case 8: + range = 268435456; + break; + default: + LOG_F("Invalid operand size: %zu", varLen); + + } + + mangle_AddSubWithRange(run, off, varLen, range, printable); + +} + +static void mangle_IncByte(run_t *run, bool printable) { + + size_t off = mangle_getOffSet(run); + if (printable) { + + run->dynfile->data[off] = (run->dynfile->data[off] - 32 + 1) % 95 + 32; + + } else { + + run->dynfile->data[off] += (uint8_t)1UL; + + } + +} + +static void mangle_DecByte(run_t *run, bool printable) { + + size_t off = mangle_getOffSet(run); + if (printable) { + + run->dynfile->data[off] = (run->dynfile->data[off] - 32 + 94) % 95 + 32; + + } else { + + run->dynfile->data[off] -= (uint8_t)1UL; + + } + +} + +static void mangle_NegByte(run_t *run, bool printable) { + + size_t off = mangle_getOffSet(run); + if (printable) { + + run->dynfile->data[off] = 94 - (run->dynfile->data[off] - 32) + 32; + + } else { + + run->dynfile->data[off] = ~(run->dynfile->data[off]); + + } + +} + +static void mangle_Expand(run_t *run, bool printable) { + + size_t off = mangle_getOffSet(run); + size_t len; + if (util_rnd64() % 16) { + + len = mangle_getLen(HF_MIN(16, run->global->mutate.maxInputSz - off)); + + } else { + + len = mangle_getLen(run->global->mutate.maxInputSz - off); + + } + + mangle_Inflate(run, off, len, printable); + +} + +static void mangle_Shrink(run_t *run, bool printable HF_ATTR_UNUSED) { + + if (run->dynfile->size <= 2U) { return; } + + size_t off_start = mangle_getOffSet(run); + size_t len = mangle_LenLeft(run, off_start); + if (len == 0) { return; } + if (util_rnd64() % 16) { + + len = mangle_getLen(HF_MIN(16, len)); + + } else { + + len = mangle_getLen(len); + + } + + size_t off_end = off_start + len; + size_t len_to_move = run->dynfile->size - off_end; + + mangle_Move(run, off_end, off_start, len_to_move); + input_setSize(run, run->dynfile->size - len); + +} + +static void mangle_ASCIINum(run_t *run, bool printable) { + + size_t len = util_rndGet(2, 8); + + char buf[20]; + snprintf(buf, sizeof(buf), "%-19" PRId64, (int64_t)util_rnd64()); + + mangle_UseValue(run, (const uint8_t *)buf, len, printable); + +} + +static void mangle_ASCIINumChange(run_t *run, bool printable) { + + size_t off = mangle_getOffSet(run); + + /* Find a digit */ + for (; off < run->dynfile->size; off++) { + + if (isdigit(run->dynfile->data[off])) { break; } + + } + + if (off == run->dynfile->size) { + + mangle_Bytes(run, printable); + return; + + } + + size_t len = HF_MIN(20, run->dynfile->size - off); + char numbuf[21] = {}; + strncpy(numbuf, (const char *)&run->dynfile->data[off], len); + uint64_t val = (uint64_t)strtoull(numbuf, NULL, 10); + + switch (util_rndGet(0, 5)) { + + case 0: + val += util_rndGet(1, 256); + break; + case 1: + val -= util_rndGet(1, 256); + break; + case 2: + val *= util_rndGet(1, 256); + break; + case 3: + val /= util_rndGet(1, 256); + break; + case 4: + val = ~(val); + break; + case 5: + val = util_rnd64(); + break; + default: + LOG_F("Invalid choice"); + + }; + + len = HF_MIN((size_t)snprintf(numbuf, sizeof(numbuf), "%" PRIu64, val), len); + mangle_Overwrite(run, off, (const uint8_t *)numbuf, len, printable); + +} + +static void mangle_Splice(run_t *run, bool printable) { + + const uint8_t *buf; + size_t sz = input_getRandomInputAsBuf(run, &buf); + if (!sz) { + + mangle_Bytes(run, printable); + return; + + } + + size_t remoteOff = mangle_getLen(sz) - 1; + size_t len = mangle_getLen(sz - remoteOff); + mangle_UseValue(run, &buf[remoteOff], len, printable); + +} + +static void mangle_Resize(run_t *run, bool printable) { + + ssize_t oldsz = run->dynfile->size; + ssize_t newsz = 0; + + uint64_t choice = util_rndGet(0, 32); + switch (choice) { + + case 0: /* Set new size arbitrarily */ + newsz = (ssize_t)util_rndGet(1, run->global->mutate.maxInputSz); + break; + case 1 ... 4: /* Increase size by a small value */ + newsz = oldsz + (ssize_t)util_rndGet(0, 8); + break; + case 5: /* Increase size by a larger value */ + newsz = oldsz + (ssize_t)util_rndGet(9, 128); + break; + case 6 ... 9: /* Decrease size by a small value */ + newsz = oldsz - (ssize_t)util_rndGet(0, 8); + break; + case 10: /* Decrease size by a larger value */ + newsz = oldsz - (ssize_t)util_rndGet(9, 128); + break; + case 11 ... 32: /* Do nothing */ + newsz = oldsz; + break; + default: + LOG_F("Illegal value from util_rndGet: %" PRIu64, choice); + break; + + } + + if (newsz < 1) { newsz = 1; } + if (newsz > (ssize_t)run->global->mutate.maxInputSz) { + + newsz = run->global->mutate.maxInputSz; + + } + + input_setSize(run, (size_t)newsz); + if (newsz > oldsz) { + + if (printable) { memset(&run->dynfile->data[oldsz], ' ', newsz - oldsz); } + + } + +} + +void mangle_mangleContent(run_t *run, int speed_factor) { + + static void (*const mangleFuncs[])(run_t * run, bool printable) = { + + /* Every *Insert or Expand expands file, so add more Shrink's */ + mangle_Shrink, + mangle_Shrink, + mangle_Shrink, + mangle_Shrink, + mangle_Expand, + mangle_Bit, + mangle_IncByte, + mangle_DecByte, + mangle_NegByte, + mangle_AddSub, + mangle_MemSet, + mangle_MemSwap, + mangle_MemCopy, + mangle_Bytes, + mangle_ASCIINum, + mangle_ASCIINumChange, + mangle_ByteRepeatOverwrite, + mangle_ByteRepeatInsert, + mangle_Magic, + mangle_StaticDict, + mangle_ConstFeedbackDict, + mangle_RandomOverwrite, + mangle_RandomInsert, + mangle_Splice, + + }; + + if (run->mutationsPerRun == 0U) { return; } + if (run->dynfile->size == 0U) { + + mangle_Resize(run, /* printable= */ run->global->cfg.only_printable); + + } + + uint64_t changesCnt = run->global->mutate.mutationsPerRun; + + if (speed_factor < 5) { + + changesCnt = util_rndGet(1, run->global->mutate.mutationsPerRun); + + } else if (speed_factor < 10) { + + changesCnt = run->global->mutate.mutationsPerRun; + + } else { + + changesCnt = HF_MIN(speed_factor, 12); + changesCnt = HF_MAX(changesCnt, run->global->mutate.mutationsPerRun); + + } + + /* If last coverage acquisition was more than 5 secs ago, use splicing more + * frequently */ + if ((time(NULL) - ATOMIC_GET(run->global->timing.lastCovUpdate)) > 5) { + + if (util_rnd64() % 2) { + + mangle_Splice(run, run->global->cfg.only_printable); + + } + + } + + for (uint64_t x = 0; x < changesCnt; x++) { + + uint64_t choice = util_rndGet(0, ARRAYSIZE(mangleFuncs) - 1); + mangleFuncs[choice](run, /* printable= */ run->global->cfg.only_printable); + + } + + wmb(); + +} + diff --git a/custom_mutators/honggfuzz/mangle.h b/custom_mutators/honggfuzz/mangle.h new file mode 100644 index 00000000..1b6a4943 --- /dev/null +++ b/custom_mutators/honggfuzz/mangle.h @@ -0,0 +1,32 @@ +/* + * + * honggfuzz - buffer mangling routines + * ----------------------------------------- + * + * Author: Robert Swiecki + * + * Copyright 2010-2018 by 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 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or + * implied. See the License for the specific language governing + * permissions and limitations under the License. + * + */ + +#ifndef _HF_MANGLE_H_ +#define _HF_MANGLE_H_ + +#include "honggfuzz.h" + +extern void mangle_mangleContent(run_t *run, int speed_factor); + +#endif + diff --git a/custom_mutators/honggfuzz/util.h b/custom_mutators/honggfuzz/util.h new file mode 120000 index 00000000..51e19654 --- /dev/null +++ b/custom_mutators/honggfuzz/util.h @@ -0,0 +1 @@ +common.h \ No newline at end of file diff --git a/docs/Changelog.md b/docs/Changelog.md index 50f5629f..a25cc43c 100644 --- a/docs/Changelog.md +++ b/docs/Changelog.md @@ -18,6 +18,7 @@ sending a mail to . - LTO: autodictionary mode is a default - LTO: instrim instrumentation disabled, only classic support used as it is always better + - added honggfuzz mangle as a custom mutator in custom_mutators/honggfuzz :) - added afl-frida gum solution to examples/afl_frida (mostly imported from https://github.com/meme/hotwax/) - small fixes to afl-plot, afl-whatsup and man page creation -- cgit 1.4.1 From aa3856261d90d996a298704f1d3706ef1c6787cd Mon Sep 17 00:00:00 2001 From: van Hauser Date: Thu, 23 Jul 2020 17:17:44 +0200 Subject: fix warning --- custom_mutators/honggfuzz/input.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/custom_mutators/honggfuzz/input.h b/custom_mutators/honggfuzz/input.h index c67d88a6..7b0c55ae 100644 --- a/custom_mutators/honggfuzz/input.h +++ b/custom_mutators/honggfuzz/input.h @@ -86,7 +86,7 @@ static inline size_t input_getRandomInputAsBuf(run_t *run, const uint8_t **buf) static inline void input_setSize(run_t* run, size_t sz) { run->dynfile->size = sz; } -static inline uint8_t util_turnToPrintable(uint8_t* buf, size_t sz) { +static inline void util_turnToPrintable(uint8_t* buf, size_t sz) { for (size_t i = 0; i < sz; i++) buf[i] = buf[i] % 95 + 32; } -- cgit 1.4.1 From 9cddbc04206bd8d1399e5a5311c98fff5be80731 Mon Sep 17 00:00:00 2001 From: van Hauser Date: Fri, 24 Jul 2020 12:26:52 +0200 Subject: add -F option to sync to foreign fuzzer queues --- GNUmakefile | 4 +- README.md | 20 +++--- TODO.md | 2 - docs/Changelog.md | 2 + docs/parallel_fuzzing.md | 14 ++++- include/afl-fuzz.h | 13 ++++ src/afl-fuzz-init.c | 154 +++++++++++++++++++++++++++++++++++++++++++++++ src/afl-fuzz-run.c | 2 + src/afl-fuzz.c | 22 ++++++- 9 files changed, 211 insertions(+), 22 deletions(-) diff --git a/GNUmakefile b/GNUmakefile index f44ef95e..ab9144b8 100644 --- a/GNUmakefile +++ b/GNUmakefile @@ -455,10 +455,10 @@ code-format: ./.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/*.h ./.custom-format.py -i gcc_plugin/*.cc ./.custom-format.py -i custom_mutators/*/*.c - ./.custom-format.py -i custom_mutators/*/*.h + @#./.custom-format.py -i custom_mutators/*/*.h # destroys input.h :-( ./.custom-format.py -i examples/*/*.c ./.custom-format.py -i examples/*/*.h ./.custom-format.py -i test/*.c diff --git a/README.md b/README.md index 4e83021d..b2f41315 100644 --- a/README.md +++ b/README.md @@ -366,9 +366,9 @@ If you find other good ones, please send them to us :-) ## Power schedules -The power schedules were copied from Marcel Böhme's excellent AFLfast -implementation and expand on the ability to discover new paths and -therefore may increase the code coverage. +The power schedules were copied from Marcel Böhme's AFLfast implementation and +measure differently which queue entries to prefer and therefore may find +different paths faster for large queues. The available schedules are: @@ -382,16 +382,10 @@ The available schedules are: - mmopt (afl++ experimental) - seek (afl++ experimental) -In parallel mode (-M/-S, several instances with the shared queue), we suggest to -run the main node using the explore or fast schedule (-p explore) and the secondary -nodes with a combination of cut-off-exponential (-p coe), exponential (-p fast), -explore (-p explore) and mmopt (-p mmopt) schedules. If a schedule does -not perform well for a target, restart the secondary nodes with a different schedule. - -In single mode, using -p fast is usually slightly more beneficial than the -default explore mode. -(We don't want to change the default behavior of afl, so "fast" has not been -made the default mode). +In parallel mode (-M/-S, several instances with the shared queue), we suggest +to run the main node using the default explore schedule (`-p explore`) and the +secondary nodes with different schedules. If a schedule does not perform well +for a target, restart the secondary nodes with a different schedule. More details can be found in the paper published at the 23rd ACM Conference on Computer and Communications Security [CCS'16](https://www.sigsac.org/ccs/CCS2016/accepted-papers/) diff --git a/TODO.md b/TODO.md index ad3ef83e..ad743b6b 100644 --- a/TODO.md +++ b/TODO.md @@ -3,9 +3,7 @@ ## Roadmap 2.67+ - -i - + foreign fuzzer sync support: scandir with time sort - - pre_save custom module example to save away test cases - expand on AFL_LLVM_INSTRUMENT_FILE to also support sancov allowlist format - - allow to sync against honggfuzz and libfuzzer - AFL_MAP_SIZE for qemu_mode and unicorn_mode - namespace for targets? e.g. network - learn from honggfuzz (mutations, maybe ptrace?) diff --git a/docs/Changelog.md b/docs/Changelog.md index a25cc43c..bec87d65 100644 --- a/docs/Changelog.md +++ b/docs/Changelog.md @@ -11,6 +11,8 @@ sending a mail to . ### Version ++2.66d (devel) - afl-fuzz: + - added -F option to allow -M main fuzzers to sync to foreign fuzzers, + e.g. honggfuzz or libfuzzer - eliminated CPU affinity race condition for -S/-M runs - llvm_mode: - fixes for laf-intel float splitting (thanks to mark-griffin for diff --git a/docs/parallel_fuzzing.md b/docs/parallel_fuzzing.md index 271f8369..2ab1466c 100644 --- a/docs/parallel_fuzzing.md +++ b/docs/parallel_fuzzing.md @@ -99,7 +99,15 @@ example may be: This is not a concern if you use @@ without -f and let afl-fuzz come up with the file name. -## 3) Multi-system parallelization +## 3) Syncing with non-afl fuzzers or independant instances + +A -M main node can be told with the `-F other_fuzzer_queue_directory` option +to sync results from other fuzzers, e.g. libfuzzer or honggfuzz. + +Only the specified directory will by synced into afl, not subdirectories. +The specified directories do not need to exist yet at the start of afl. + +## 4) Multi-system parallelization The basic operating principle for multi-system parallelization is similar to the mechanism explained in section 2. The key difference is that you need to @@ -176,7 +184,7 @@ It is *not* advisable to skip the synchronization script and run the fuzzers directly on a network filesystem; unexpected latency and unkillable processes in I/O wait state can mess things up. -## 4) Remote monitoring and data collection +## 5) Remote monitoring and data collection You can use screen, nohup, tmux, or something equivalent to run remote instances of afl-fuzz. If you redirect the program's output to a file, it will @@ -200,7 +208,7 @@ Keep in mind that crashing inputs are *not* automatically propagated to the main instance, so you may still want to monitor for crashes fleet-wide from within your synchronization or health checking scripts (see afl-whatsup). -## 5) Asymmetric setups +## 6) Asymmetric setups It is perhaps worth noting that all of the following is permitted: diff --git a/include/afl-fuzz.h b/include/afl-fuzz.h index c9f84c61..cf4254ac 100644 --- a/include/afl-fuzz.h +++ b/include/afl-fuzz.h @@ -347,6 +347,13 @@ struct afl_pass_stat { }; +struct foreign_sync { + + u8 * dir; + time_t ctime; + +}; + typedef struct afl_state { /* Position of this state in the global states list */ @@ -574,6 +581,11 @@ typedef struct afl_state { u8 describe_op_buf_256[256]; /* describe_op will use this to return a string up to 256 */ +/* foreign sync */ +#define FOREIGN_SYNCS_MAX 32 + u8 foreign_sync_cnt; + struct foreign_sync foreign_syncs[FOREIGN_SYNCS_MAX]; + #ifdef _AFL_DOCUMENT_MUTATIONS u8 do_document; u32 document_counter; @@ -937,6 +949,7 @@ void fix_up_banner(afl_state_t *, u8 *); void check_if_tty(afl_state_t *); void setup_signal_handlers(void); void save_cmdline(afl_state_t *, u32, char **); +void read_foreign_testcases(afl_state_t *, int); /* CmpLog */ diff --git a/src/afl-fuzz-init.c b/src/afl-fuzz-init.c index 609e16ba..65ad0c9f 100644 --- a/src/afl-fuzz-init.c +++ b/src/afl-fuzz-init.c @@ -438,6 +438,159 @@ static void shuffle_ptrs(afl_state_t *afl, void **ptrs, u32 cnt) { } +/* Read all testcases from foreign input directories, then queue them for + testing. Called at startup and at sync intervals. + Does not descend into subdirectories! */ + +void read_foreign_testcases(afl_state_t *afl, int first) { + + if (!afl->foreign_sync_cnt) return; + + struct dirent **nl; + s32 nl_cnt; + u32 i, iter; + + u8 val_buf[2][STRINGIFY_VAL_SIZE_MAX]; + + for (iter = 0; iter < afl->foreign_sync_cnt; iter++) { + + if (afl->foreign_syncs[iter].dir != NULL && + afl->foreign_syncs[iter].dir[0] != 0) { + + if (first) ACTF("Scanning '%s'...", afl->foreign_syncs[iter].dir); + time_t ctime_max = 0; + + /* We use scandir() + alphasort() rather than readdir() because otherwise, + the ordering of test cases would vary somewhat randomly and would be + difficult to control. */ + + nl_cnt = scandir(afl->foreign_syncs[iter].dir, &nl, NULL, NULL); + + if (nl_cnt < 0) { + + if (first) { + + WARNF("Unable to open directory '%s'", afl->foreign_syncs[iter].dir); + sleep(1); + + } + + continue; + + } + + if (nl_cnt == 0) { + + if (first) + WARNF("directory %s is currently empty", + afl->foreign_syncs[iter].dir); + continue; + + } + + /* Show stats */ + + snprintf(afl->stage_name_buf, STAGE_BUF_SIZE, "foreign sync %u", iter); + + afl->stage_name = afl->stage_name_buf; + afl->stage_cur = 0; + afl->stage_max = 0; + + for (i = 0; i < nl_cnt; ++i) { + + struct stat st; + + u8 *fn2 = + alloc_printf("%s/%s", afl->foreign_syncs[iter].dir, nl[i]->d_name); + + free(nl[i]); /* not tracked */ + + if (unlikely(lstat(fn2, &st) || access(fn2, R_OK))) { + + if (first) PFATAL("Unable to access '%s'", fn2); + continue; + + } + + /* we detect new files by their ctime */ + if (likely(st.st_ctime <= afl->foreign_syncs[iter].ctime)) { + + ck_free(fn2); + continue; + + } + + /* This also takes care of . and .. */ + + if (!S_ISREG(st.st_mode) || !st.st_size || strstr(fn2, "/README.txt")) { + + ck_free(fn2); + continue; + + } + + if (st.st_size > MAX_FILE) { + + if (first) + WARNF( + "Test case '%s' is too big (%s, limit is %s), skipping", fn2, + stringify_mem_size(val_buf[0], sizeof(val_buf[0]), st.st_size), + stringify_mem_size(val_buf[1], sizeof(val_buf[1]), MAX_FILE)); + ck_free(fn2); + continue; + + } + + // lets do not use add_to_queue(afl, fn2, st.st_size, 0); + // as this could add duplicates of the startup input corpus + + int fd = open(fn2, O_RDONLY); + if (fd < 0) { + + ck_free(fn2); + continue; + + } + + u8 fault; + u8 *mem = mmap(0, st.st_size, PROT_READ, MAP_PRIVATE, fd, 0); + + if (mem == MAP_FAILED) { + + ck_free(fn2); + continue; + + } + + write_to_testcase(afl, mem, st.st_size); + fault = fuzz_run_target(afl, &afl->fsrv, afl->fsrv.exec_tmout); + afl->syncing_party = "foreign"; + afl->queued_imported += + save_if_interesting(afl, mem, st.st_size, fault); + afl->syncing_party = 0; + munmap(mem, st.st_size); + close(fd); + + if (st.st_ctime > ctime_max) ctime_max = st.st_ctime; + + } + + afl->foreign_syncs[iter].ctime = ctime_max; + free(nl); /* not tracked */ + + } + + } + + if (first) { + + afl->last_path_time = 0; + afl->queued_at_start = afl->queued_paths; + + } + +} + /* Read all testcases from the input directory, then queue them for testing. Called at startup. */ @@ -530,6 +683,7 @@ void read_testcases(afl_state_t *afl) { WARNF("Test case '%s' is too big (%s, limit is %s), skipping", fn2, stringify_mem_size(val_buf[0], sizeof(val_buf[0]), st.st_size), stringify_mem_size(val_buf[1], sizeof(val_buf[1]), MAX_FILE)); + ck_free(fn2); continue; } diff --git a/src/afl-fuzz-run.c b/src/afl-fuzz-run.c index 2a1664e2..6e3be72b 100644 --- a/src/afl-fuzz-run.c +++ b/src/afl-fuzz-run.c @@ -612,6 +612,8 @@ void sync_fuzzers(afl_state_t *afl) { } + if (afl->foreign_sync_cnt) read_foreign_testcases(afl, 0); + } /* Trim all new test cases to save cycles when doing deterministic checks. The diff --git a/src/afl-fuzz.c b/src/afl-fuzz.c index df2896d2..f03c545d 100644 --- a/src/afl-fuzz.c +++ b/src/afl-fuzz.c @@ -131,10 +131,13 @@ static void usage(afl_state_t *afl, u8 *argv0, int more_help) { "executions.\n\n" "Other stuff:\n" - " -T text - text banner to show on the screen\n" " -M/-S id - distributed mode (see docs/parallel_fuzzing.md)\n" " use -D to force -S secondary to perform deterministic " "fuzzing\n" + " -F path - sync to a foreign fuzzer queue directory (requires " + "-M, can\n" + " be specified up to %u times)\n" + " -T text - text banner to show on the screen\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 @@ -142,7 +145,7 @@ static void usage(afl_state_t *afl, u8 *argv0, int more_help) { " -C - crash exploration mode (the peruvian rabbit thing)\n" " -e ext - file extension for the fuzz test input file (if " "needed)\n\n", - argv0, EXEC_TIMEOUT, MEM_LIMIT); + argv0, EXEC_TIMEOUT, MEM_LIMIT, FOREIGN_SYNCS_MAX); if (more_help > 1) { @@ -403,6 +406,19 @@ int main(int argc, char **argv_orig, char **envp) { afl->use_splicing = 1; break; + case 'F': /* foreign sync dir */ + + if (!afl->is_main_node) + FATAL( + "Option -F can only be specified after the -M option for the " + "main fuzzer of a fuzzing campaign"); + if (afl->foreign_sync_cnt >= FOREIGN_SYNCS_MAX) + FATAL("Maximum %u entried of -F option can be specified", + FOREIGN_SYNCS_MAX); + afl->foreign_syncs[afl->foreign_sync_cnt].dir = optarg; + afl->foreign_sync_cnt++; + break; + case 'f': /* target file */ if (afl->fsrv.out_file) { FATAL("Multiple -f options not supported"); } @@ -1059,6 +1075,8 @@ int main(int argc, char **argv_orig, char **envp) { setup_cmdline_file(afl, argv + optind); read_testcases(afl); + // read_foreign_testcases(afl, 1); for the moment dont do this + load_auto(afl); pivot_inputs(afl); -- cgit 1.4.1 From 79598083849b34bf33c4a6c904e69c2743151082 Mon Sep 17 00:00:00 2001 From: van Hauser Date: Fri, 24 Jul 2020 16:39:50 +0200 Subject: llvm12 support --- README.md | 6 +++--- docs/Changelog.md | 1 + llvm_mode/GNUmakefile | 11 ++++++----- llvm_mode/README.lto.md | 26 +++++++++++++------------- llvm_mode/README.md | 4 ++-- 5 files changed, 25 insertions(+), 23 deletions(-) diff --git a/README.md b/README.md index b2f41315..9c802285 100644 --- a/README.md +++ b/README.md @@ -29,7 +29,7 @@ get any feature improvements since November 2017. Among other changes afl++ has a more performant llvm_mode, supports - llvm up to version 11, QEMU 3.1, more speed and crashfixes for QEMU, + llvm up to version 12, QEMU 3.1, more speed and crashfixes for QEMU, better *BSD and Android support and much, much more. Additionally the following features and patches have been integrated: @@ -268,7 +268,7 @@ superior to blind fuzzing or coverage-only tools. PLEASE NOTE: llvm_mode compilation with afl-clang-fast/afl-clang-fast++ instead of afl-gcc/afl-g++ is much faster and has many cool features. See llvm_mode/ - however few code does not compile with llvm. -We support llvm versions 3.4 to 11. +We support llvm versions 3.4 to 12. When source code is available, instrumentation can be injected by a companion tool that works as a drop-in replacement for gcc or clang in any standard build @@ -291,7 +291,7 @@ For C++ programs, you'd would also want to set `CXX=/path/to/afl/afl-g++`. The clang wrappers (afl-clang and afl-clang++) can be used in the same way; clang users may also opt to leverage a higher-performance instrumentation mode, as described in [llvm_mode/README.md](llvm_mode/README.md). -Clang/LLVM has a much better performance and works with LLVM version 3.4 to 11. +Clang/LLVM has a much better performance and works with LLVM version 3.4 to 12. Using the LAF Intel performance enhancements are also recommended, see [llvm_mode/README.laf-intel.md](llvm_mode/README.laf-intel.md) diff --git a/docs/Changelog.md b/docs/Changelog.md index bec87d65..38787def 100644 --- a/docs/Changelog.md +++ b/docs/Changelog.md @@ -15,6 +15,7 @@ sending a mail to . e.g. honggfuzz or libfuzzer - eliminated CPU affinity race condition for -S/-M runs - llvm_mode: + - now supports llvm 12! - fixes for laf-intel float splitting (thanks to mark-griffin for reporting) - LTO: autodictionary mode is a default diff --git a/llvm_mode/GNUmakefile b/llvm_mode/GNUmakefile index b5d026ef..443322d7 100644 --- a/llvm_mode/GNUmakefile +++ b/llvm_mode/GNUmakefile @@ -32,15 +32,16 @@ ifeq "$(shell uname)" "OpenBSD" LLVM_CONFIG ?= $(BIN_PATH)/llvm-config HAS_OPT = $(shell test -x $(BIN_PATH)/opt && echo 0 || echo 1) ifeq "$(HAS_OPT)" "1" - $(error llvm_mode needs a complete llvm installation (versions 3.4 up to 11) -> e.g. "pkg_add llvm-7.0.1p9") + $(error llvm_mode needs a complete llvm installation (versions 3.4 up to 12) -> e.g. "pkg_add llvm-7.0.1p9") endif else LLVM_CONFIG ?= llvm-config endif LLVMVER = $(shell $(LLVM_CONFIG) --version 2>/dev/null | sed 's/git//' ) -LLVM_UNSUPPORTED = $(shell $(LLVM_CONFIG) --version 2>/dev/null | egrep -q '^3\.[0-3]|^1[2-9]' && echo 1 || echo 0 ) +LLVM_UNSUPPORTED = $(shell $(LLVM_CONFIG) --version 2>/dev/null | egrep -q '^3\.[0-3]|^1[3-9]' && echo 1 || echo 0 ) LLVM_NEW_API = $(shell $(LLVM_CONFIG) --version 2>/dev/null | egrep -q '^1[0-9]' && echo 1 || echo 0 ) +LLVM_HAVE_LTO = $(shell $(LLVM_CONFIG) --version 2>/dev/null | egrep -q '^1[1-9]' && echo 1 || echo 0 ) LLVM_MAJOR = $(shell $(LLVM_CONFIG) --version 2>/dev/null | sed 's/\..*//') LLVM_BINDIR = $(shell $(LLVM_CONFIG) --bindir 2>/dev/null) LLVM_LIBDIR = $(shell $(LLVM_CONFIG) --libdir 2>/dev/null) @@ -53,7 +54,7 @@ ifeq "$(LLVMVER)" "" endif ifeq "$(LLVM_UNSUPPORTED)" "1" - $(warning llvm_mode only supports llvm versions 3.4 up to 11) + $(warning llvm_mode only supports llvm versions 3.4 up to 12) endif ifeq "$(LLVM_MAJOR)" "9" @@ -65,8 +66,8 @@ ifeq "$(LLVM_NEW_API)" "1" LLVM_STDCXX = c++14 endif -ifeq "$(LLVM_MAJOR)" "11" - $(info [+] llvm_mode detected llvm 11, enabling afl-clang-lto LTO implementation) +ifeq "$(LLVM_HAVE_LTO)" "1" + $(info [+] llvm_mode detected llvm 11+, enabling afl-clang-lto LTO implementation) LLVM_LTO = 1 #TEST_MMAP = 1 endif diff --git a/llvm_mode/README.lto.md b/llvm_mode/README.lto.md index 967a31aa..d54d4ee0 100644 --- a/llvm_mode/README.lto.md +++ b/llvm_mode/README.lto.md @@ -2,7 +2,7 @@ ## TLDR; -This version requires a current llvm 11 compiled from the github master. +This version requires a current llvm 11+ compiled from the github master. 1. Use afl-clang-lto/afl-clang-lto++ because it is faster and gives better coverage than anything else that is out there in the AFL world @@ -10,7 +10,7 @@ This version requires a current llvm 11 compiled from the github master. 2. You can use it together with llvm_mode: laf-intel and the instrument file listing features and can be combined with cmplog/Redqueen -3. It only works with llvm 11 (current github master state) +3. It only works with llvm 11+ 4. AUTODICTIONARY feature! see below @@ -61,9 +61,9 @@ AUTODICTIONARY: 11 strings found [+] Instrumented 12071 locations with no collisions (on average 1046 collisions would be in afl-gcc/afl-clang-fast) (non-hardened mode). ``` -## Getting llvm 11 +## Getting llvm 11+ -### Installing llvm 11 from the llvm repository +### Installing llvm from the llvm repository (version 11) Installing the llvm snapshot builds is easy and mostly painless: @@ -83,7 +83,7 @@ apt-get install -y clang-11 clang-tools-11 libc++1-11 libc++-11-dev \ libomp5-11 lld-11 lldb-11 llvm-11 llvm-11-dev llvm-11-runtime llvm-11-tools ``` -### Building llvm 11 yourself +### Building llvm yourself (version 12) Building llvm from github takes quite some long time and is not painless: ``` @@ -201,15 +201,15 @@ cd WebKit ``` mkdir -p WebKitBuild/Release cd WebKitBuild/Release -ln -s ../../../../../usr/bin/llvm-ar-11 llvm-ar-11 -ln -s ../../../../../usr/bin/llvm-ranlib-11 llvm-ranlib-11 +ln -s ../../../../../usr/bin/llvm-ar-12 llvm-ar-12 +ln -s ../../../../../usr/bin/llvm-ranlib-12 llvm-ranlib-12 cd ../.. ``` 3. Build :) ``` -Tools/Scripts/build-jsc --jsc-only --cli --cmakeargs="-DCMAKE_AR='llvm-ar-11' -DCMAKE_RANLIB='llvm-ranlib-11' -DCMAKE_VERBOSE_MAKEFILE:BOOL=ON -DCMAKE_CC_FLAGS='-O3 -lrt' -DCMAKE_CXX_FLAGS='-O3 -lrt' -DIMPORTED_LOCATION='/lib/x86_64-linux-gnu/' -DCMAKE_CC=afl-clang-lto -DCMAKE_CXX=afl-clang-lto++ -DENABLE_STATIC_JSC=ON" +Tools/Scripts/build-jsc --jsc-only --cli --cmakeargs="-DCMAKE_AR='llvm-ar-12' -DCMAKE_RANLIB='llvm-ranlib-12' -DCMAKE_VERBOSE_MAKEFILE:BOOL=ON -DCMAKE_CC_FLAGS='-O3 -lrt' -DCMAKE_CXX_FLAGS='-O3 -lrt' -DIMPORTED_LOCATION='/lib/x86_64-linux-gnu/' -DCMAKE_CC=afl-clang-lto -DCMAKE_CXX=afl-clang-lto++ -DENABLE_STATIC_JSC=ON" ``` ## Potential issues @@ -246,17 +246,17 @@ AS=llvm-as ... afl-clang-lto is still work in progress. Known issues: - * Anything that llvm 11 cannot compile, afl-clang-lto can not compile either - obviously + * Anything that llvm 11+ cannot compile, afl-clang-lto can not compile either - obviously * Anything that does not compile with LTO, afl-clang-lto can not compile either - obviously -Hence if building a target with afl-clang-lto fails try to build it with llvm11 -and LTO enabled (`CC=clang-11` `CXX=clang++-11` `CFLAGS=-flto=full` and +Hence if building a target with afl-clang-lto fails try to build it with llvm12 +and LTO enabled (`CC=clang-12` `CXX=clang++-12` `CFLAGS=-flto=full` and `CXXFLAGS=-flto=full`). If this succeeeds then there is an issue with afl-clang-lto. Please report at [https://github.com/AFLplusplus/AFLplusplus/issues/226](https://github.com/AFLplusplus/AFLplusplus/issues/226) -Even some targets where clang-11 fails can be build if the fail is just in +Even some targets where clang-12 fails can be build if the fail is just in `./configure`, see `Solving difficult targets` above. ### Target crashes immediately @@ -296,7 +296,7 @@ Still more problems came up though as this only works without bugs from llvm 9 onwards, and with high optimization the link optimization ruins the instrumented control flow graph. -This is all now fixed with llvm 11. The llvm's own linker is now able to +This is all now fixed with llvm 11+. The llvm's own linker is now able to load passes and this bypasses all problems we had. Happy end :) diff --git a/llvm_mode/README.md b/llvm_mode/README.md index e2e22751..22088dfd 100644 --- a/llvm_mode/README.md +++ b/llvm_mode/README.md @@ -6,7 +6,7 @@ ## 1) Introduction -! llvm_mode works with llvm versions 3.4 up to 11 ! +! llvm_mode works with llvm versions 3.4 up to 12 ! The code in this directory allows you to instrument programs for AFL using true compiler-level instrumentation, instead of the more crude @@ -183,4 +183,4 @@ AFL_LLVM_INSTRUMENT=PCGUARD make ``` Note that this us currently the default, as it is the best mode. -If you have llvm 11 and compiled afl-clang-lto - this is the only better mode. +If you have llvm 11+ and compiled afl-clang-lto - this is the only better mode. -- cgit 1.4.1 From 89557d16077652c52df4244809a0e9c733a788cc Mon Sep 17 00:00:00 2001 From: Kirin-say Date: Sat, 25 Jul 2020 00:36:36 +0800 Subject: fix bug in qemu mode --- qemu_mode/patches/afl-qemu-cpu-inl.h | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/qemu_mode/patches/afl-qemu-cpu-inl.h b/qemu_mode/patches/afl-qemu-cpu-inl.h index 5f579687..07da9324 100644 --- a/qemu_mode/patches/afl-qemu-cpu-inl.h +++ b/qemu_mode/patches/afl-qemu-cpu-inl.h @@ -620,7 +620,8 @@ static void afl_wait_tsl(CPUState *cpu, int fd) { last_tb = tb_htable_lookup(cpu, c.last_tb.pc, c.last_tb.cs_base, c.last_tb.flags, c.cf_mask); - if (last_tb) { tb_add_jump(last_tb, c.tb_exit, tb); } +#define TB_JMP_RESET_OFFSET_INVALID 0xffff + if (last_tb && (last_tb->jmp_reset_offset[c.tb_exit] != TB_JMP_RESET_OFFSET_INVALID)) { tb_add_jump(last_tb, c.tb_exit, tb); } } -- cgit 1.4.1 From 131df8bec9dc9dd7cf7a1c03d7189ba72580f6ab Mon Sep 17 00:00:00 2001 From: van Hauser Date: Sat, 25 Jul 2020 16:39:36 +0200 Subject: try to fix travis --- test/test-floatingpoint.c | 24 +++++++++++++++--------- test/test.sh | 4 ++-- 2 files changed, 17 insertions(+), 11 deletions(-) diff --git a/test/test-floatingpoint.c b/test/test-floatingpoint.c index 8f691c2c..083f0df5 100644 --- a/test/test-floatingpoint.c +++ b/test/test-floatingpoint.c @@ -1,17 +1,23 @@ #include #include +#include +#include -int main(void) { - - float magic; +__AFL_FUZZ_INIT(); - ssize_t bytes_read = read(STDIN_FILENO, &magic, sizeof(magic)); - if (bytes_read < (ssize_t)sizeof(magic)) { return 1; } - - if ((-magic == 15.0 + 0.5 + 0.125 + 0.03125 + - 0.0078125)) { /* 15 + 1/2 + 1/8 + 1/32 + 1/128 */ - abort(); +int main(void) { + ssize_t bytes_read; + + __AFL_INIT(); + float *magic = (float*)__AFL_FUZZ_TESTCASE_BUF; + + while (__AFL_LOOP(INT_MAX)) { + + if (__AFL_FUZZ_TESTCASE_LEN != sizeof(float)) return 1; + /* 15 + 1/2 + 1/8 + 1/32 + 1/128 */ + if ((-*magic == 15.0 + 0.5 + 0.125 + 0.03125 + 0.0078125)) abort(); + } return 0; diff --git a/test/test.sh b/test/test.sh index 15082070..dc85f745 100755 --- a/test/test.sh +++ b/test/test.sh @@ -385,13 +385,13 @@ test -e ../afl-clang-fast -a -e ../split-switches-pass.so && { CODE=1 } rm -f test-compcov.compcov test.out - AFL_LLVM_INSTRUMENT=AFL AFL_DEBUG=1 AFL_LLVM_LAF_SPLIT_COMPARES=1 AFL_LLVM_LAF_SPLIT_FLOATS=1 ../afl-clang-fast -o test-floatingpoint test-floatingpoint.c > test.out 2>&1 + AFL_LLVM_INSTRUMENT=AFL AFL_DEBUG=1 AFL_NO_UI=1 AFL_LLVM_LAF_ALL=1 ../afl-clang-fast -o test-floatingpoint test-floatingpoint.c > test.out 2>&1 test -e test-floatingpoint && { mkdir -p in echo ZZ > in/in $ECHO "$GREY[*] running afl-fuzz with floating point splitting, this will take max. 30 seconds" { - AFL_BENCH_UNTIL_CRASH=1 ../afl-fuzz -s1 -V30 -m ${MEM_LIMIT} -i in -o out -- ./test-floatingpoint >>errors 2>&1 + AFL_BENCH_UNTIL_CRASH=1 ../afl-fuzz -s 123 -V30 -m ${MEM_LIMIT} -i in -o out -- ./test-floatingpoint >>errors 2>&1 } >>errors 2>&1 test -n "$( ls out/crashes/id:* 2>/dev/null )" && { $ECHO "$GREEN[+] llvm_mode laf-intel floatingpoint splitting feature works correctly" -- cgit 1.4.1 From a879f721313a35274ce2d65f9aeed344027a208e Mon Sep 17 00:00:00 2001 From: van Hauser Date: Sat, 25 Jul 2020 17:38:40 +0200 Subject: fix --- test/test.sh | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/test/test.sh b/test/test.sh index dc85f745..76b089e7 100755 --- a/test/test.sh +++ b/test/test.sh @@ -385,13 +385,13 @@ test -e ../afl-clang-fast -a -e ../split-switches-pass.so && { CODE=1 } rm -f test-compcov.compcov test.out - AFL_LLVM_INSTRUMENT=AFL AFL_DEBUG=1 AFL_NO_UI=1 AFL_LLVM_LAF_ALL=1 ../afl-clang-fast -o test-floatingpoint test-floatingpoint.c > test.out 2>&1 + AFL_LLVM_INSTRUMENT=AFL AFL_LLVM_LAF_ALL=1 ../afl-clang-fast -o test-floatingpoint test-floatingpoint.c > test.out 2>&1 test -e test-floatingpoint && { mkdir -p in echo ZZ > in/in $ECHO "$GREY[*] running afl-fuzz with floating point splitting, this will take max. 30 seconds" { - AFL_BENCH_UNTIL_CRASH=1 ../afl-fuzz -s 123 -V30 -m ${MEM_LIMIT} -i in -o out -- ./test-floatingpoint >>errors 2>&1 + AFL_BENCH_UNTIL_CRASH=1 AFL_NO_UI=1 ../afl-fuzz -s 123 -V30 -m ${MEM_LIMIT} -i in -o out -- ./test-floatingpoint >>errors 2>&1 } >>errors 2>&1 test -n "$( ls out/crashes/id:* 2>/dev/null )" && { $ECHO "$GREEN[+] llvm_mode laf-intel floatingpoint splitting feature works correctly" -- cgit 1.4.1 From ae41cedafe93a5e588ca0f7931ba99c47b61cae7 Mon Sep 17 00:00:00 2001 From: David Carlier Date: Sat, 25 Jul 2020 16:14:27 +0000 Subject: gcc plugin forcing USEMMAP on Haiku --- gcc_plugin/GNUmakefile | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) diff --git a/gcc_plugin/GNUmakefile b/gcc_plugin/GNUmakefile index 002437cb..4a4f0dcd 100644 --- a/gcc_plugin/GNUmakefile +++ b/gcc_plugin/GNUmakefile @@ -70,11 +70,13 @@ ifeq "$(TEST_MMAP)" "1" endif ifneq "$(shell uname -s)" "Haiku" - LDFLAGS += -lrt + LDFLAGS += -lrt +else + CFLAGS_SAFE += -DUSEMMAP=1 endif ifeq "$(shell uname -s)" "SunOS" - PLUGIN_FLAGS += -I/usr/include/gmp + PLUGIN_FLAGS += -I/usr/include/gmp endif -- cgit 1.4.1 From 7e4703c3282af86641d59a196fccb06df4ab6316 Mon Sep 17 00:00:00 2001 From: HAPPY Date: Sun, 26 Jul 2020 20:10:24 +0800 Subject: Fix typo for afl_custom_deinit (#470) --- src/afl-fuzz-mutators.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/afl-fuzz-mutators.c b/src/afl-fuzz-mutators.c index 17a68ff8..ed777811 100644 --- a/src/afl-fuzz-mutators.c +++ b/src/afl-fuzz-mutators.c @@ -168,7 +168,7 @@ struct custom_mutator *load_custom_mutator(afl_state_t *afl, const char *fn) { /* "afl_custom_deinit", optional for backward compatibility */ mutator->afl_custom_deinit = dlsym(dh, "afl_custom_deinit"); - if (!mutator->afl_custom_deinit) FATAL("Symbol 'afl_custom_init' not found."); + if (!mutator->afl_custom_deinit) FATAL("Symbol 'afl_custom_deinit' not found."); /* "afl_custom_post_process", optional */ mutator->afl_custom_post_process = dlsym(dh, "afl_custom_post_process"); -- cgit 1.4.1 From 16e362d2b93a60d6c50fca6abfabd9976ca6142d Mon Sep 17 00:00:00 2001 From: van Hauser Date: Sun, 26 Jul 2020 15:55:03 +0200 Subject: add last 60s exec/s stat --- include/afl-fuzz.h | 4 ++ src/afl-fuzz-mutators.c | 3 +- src/afl-fuzz-stats.c | 154 +++++++++++++++++++++++++--------------------- test/test-floatingpoint.c | 8 +-- 4 files changed, 93 insertions(+), 76 deletions(-) diff --git a/include/afl-fuzz.h b/include/afl-fuzz.h index cf4254ac..c0c4cfd5 100644 --- a/include/afl-fuzz.h +++ b/include/afl-fuzz.h @@ -581,6 +581,10 @@ typedef struct afl_state { u8 describe_op_buf_256[256]; /* describe_op will use this to return a string up to 256 */ + unsigned long long int last_avg_exec_update; + u32 last_avg_execs; + float last_avg_execs_saved; + /* foreign sync */ #define FOREIGN_SYNCS_MAX 32 u8 foreign_sync_cnt; diff --git a/src/afl-fuzz-mutators.c b/src/afl-fuzz-mutators.c index ed777811..850266c2 100644 --- a/src/afl-fuzz-mutators.c +++ b/src/afl-fuzz-mutators.c @@ -168,7 +168,8 @@ struct custom_mutator *load_custom_mutator(afl_state_t *afl, const char *fn) { /* "afl_custom_deinit", optional for backward compatibility */ mutator->afl_custom_deinit = dlsym(dh, "afl_custom_deinit"); - if (!mutator->afl_custom_deinit) FATAL("Symbol 'afl_custom_deinit' not found."); + if (!mutator->afl_custom_deinit) + FATAL("Symbol 'afl_custom_deinit' not found."); /* "afl_custom_post_process", optional */ mutator->afl_custom_post_process = dlsym(dh, "afl_custom_post_process"); diff --git a/src/afl-fuzz-stats.c b/src/afl-fuzz-stats.c index fc93011b..995f298e 100644 --- a/src/afl-fuzz-stats.c +++ b/src/afl-fuzz-stats.c @@ -39,7 +39,7 @@ void write_stats_file(afl_state_t *afl, double bitmap_cvg, double stability, u8 fn[PATH_MAX]; s32 fd; FILE * f; - uint32_t t_bytes = count_non_255_bytes(afl, afl->virgin_bits); + u32 t_bytes = count_non_255_bytes(afl, afl->virgin_bits); snprintf(fn, PATH_MAX, "%s/fuzzer_stats", afl->out_dir); @@ -67,89 +67,101 @@ void write_stats_file(afl_state_t *afl, double bitmap_cvg, double stability, } + if ((unlikely(!afl->last_avg_exec_update || + cur_time - afl->last_avg_exec_update >= 60000))) { + + afl->last_avg_execs_saved = + (float)(1000*(afl->fsrv.total_execs - afl->last_avg_execs)) / + (float)(cur_time - afl->last_avg_exec_update); + afl->last_avg_execs = afl->fsrv.total_execs; + afl->last_avg_exec_update = cur_time; + + } + #ifndef __HAIKU__ if (getrusage(RUSAGE_CHILDREN, &rus)) { rus.ru_maxrss = 0; } #endif - fprintf( - f, - "start_time : %llu\n" - "last_update : %llu\n" - "run_time : %llu\n" - "fuzzer_pid : %u\n" - "cycles_done : %llu\n" - "cycles_wo_finds : %llu\n" - "execs_done : %llu\n" - "execs_per_sec : %0.02f\n" - // "real_execs_per_sec: %0.02f\n" // damn the name is too long - "paths_total : %u\n" - "paths_favored : %u\n" - "paths_found : %u\n" - "paths_imported : %u\n" - "max_depth : %u\n" - "cur_path : %u\n" /* Must match find_start_position() */ - "pending_favs : %u\n" - "pending_total : %u\n" - "variable_paths : %u\n" - "stability : %0.02f%%\n" - "bitmap_cvg : %0.02f%%\n" - "unique_crashes : %llu\n" - "unique_hangs : %llu\n" - "last_path : %llu\n" - "last_crash : %llu\n" - "last_hang : %llu\n" - "execs_since_crash : %llu\n" - "exec_timeout : %u\n" - "slowest_exec_ms : %u\n" - "peak_rss_mb : %lu\n" - "cpu_affinity : %d\n" - "edges_found : %u\n" - "var_byte_count : %u\n" - "afl_banner : %s\n" - "afl_version : " VERSION - "\n" - "target_mode : %s%s%s%s%s%s%s%s%s\n" - "command_line : %s\n", - afl->start_time / 1000, cur_time / 1000, - (cur_time - afl->start_time) / 1000, (u32)getpid(), - afl->queue_cycle ? (afl->queue_cycle - 1) : 0, afl->cycles_wo_finds, - afl->fsrv.total_execs, - afl->fsrv.total_execs / - ((double)(get_cur_time() - afl->start_time) / 1000), - afl->queued_paths, afl->queued_favored, afl->queued_discovered, - afl->queued_imported, afl->max_depth, afl->current_entry, - afl->pending_favored, afl->pending_not_fuzzed, afl->queued_variable, - stability, bitmap_cvg, afl->unique_crashes, afl->unique_hangs, - afl->last_path_time / 1000, afl->last_crash_time / 1000, - afl->last_hang_time / 1000, afl->fsrv.total_execs - afl->last_crash_execs, - afl->fsrv.exec_tmout, afl->slowest_exec_ms, + fprintf(f, + "start_time : %llu\n" + "last_update : %llu\n" + "run_time : %llu\n" + "fuzzer_pid : %u\n" + "cycles_done : %llu\n" + "cycles_wo_finds : %llu\n" + "execs_done : %llu\n" + "execs_per_sec : %0.02f\n" + "execs_ps_last_min : %0.02f\n" + "paths_total : %u\n" + "paths_favored : %u\n" + "paths_found : %u\n" + "paths_imported : %u\n" + "max_depth : %u\n" + "cur_path : %u\n" /* Must match find_start_position() */ + "pending_favs : %u\n" + "pending_total : %u\n" + "variable_paths : %u\n" + "stability : %0.02f%%\n" + "bitmap_cvg : %0.02f%%\n" + "unique_crashes : %llu\n" + "unique_hangs : %llu\n" + "last_path : %llu\n" + "last_crash : %llu\n" + "last_hang : %llu\n" + "execs_since_crash : %llu\n" + "exec_timeout : %u\n" + "slowest_exec_ms : %u\n" + "peak_rss_mb : %lu\n" + "cpu_affinity : %d\n" + "edges_found : %u\n" + "var_byte_count : %u\n" + "afl_banner : %s\n" + "afl_version : " VERSION + "\n" + "target_mode : %s%s%s%s%s%s%s%s%s\n" + "command_line : %s\n", + afl->start_time / 1000, cur_time / 1000, + (cur_time - afl->start_time) / 1000, (u32)getpid(), + afl->queue_cycle ? (afl->queue_cycle - 1) : 0, afl->cycles_wo_finds, + afl->fsrv.total_execs, + afl->fsrv.total_execs / + ((double)(get_cur_time() - afl->start_time) / 1000), + afl->last_avg_execs_saved, afl->queued_paths, afl->queued_favored, + afl->queued_discovered, afl->queued_imported, afl->max_depth, + afl->current_entry, afl->pending_favored, afl->pending_not_fuzzed, + afl->queued_variable, stability, bitmap_cvg, afl->unique_crashes, + afl->unique_hangs, afl->last_path_time / 1000, + afl->last_crash_time / 1000, afl->last_hang_time / 1000, + afl->fsrv.total_execs - afl->last_crash_execs, afl->fsrv.exec_tmout, + afl->slowest_exec_ms, #ifndef __HAIKU__ #ifdef __APPLE__ - (unsigned long int)(rus.ru_maxrss >> 20), + (unsigned long int)(rus.ru_maxrss >> 20), #else - (unsigned long int)(rus.ru_maxrss >> 10), + (unsigned long int)(rus.ru_maxrss >> 10), #endif #else - -1UL, + -1UL, #endif #ifdef HAVE_AFFINITY - afl->cpu_aff, + afl->cpu_aff, #else - -1, + -1, #endif - t_bytes, afl->var_byte_count, afl->use_banner, - afl->unicorn_mode ? "unicorn" : "", afl->fsrv.qemu_mode ? "qemu " : "", - afl->non_instrumented_mode ? " non_instrumented " : "", - afl->no_forkserver ? "no_fsrv " : "", afl->crash_mode ? "crash " : "", - afl->persistent_mode ? "persistent " : "", - afl->shmem_testcase_mode ? "shmem_testcase " : "", - afl->deferred_mode ? "deferred " : "", - (afl->unicorn_mode || afl->fsrv.qemu_mode || afl->non_instrumented_mode || - afl->no_forkserver || afl->crash_mode || afl->persistent_mode || - afl->deferred_mode) - ? "" - : "default", - afl->orig_cmdline); + t_bytes, afl->var_byte_count, afl->use_banner, + afl->unicorn_mode ? "unicorn" : "", + afl->fsrv.qemu_mode ? "qemu " : "", + afl->non_instrumented_mode ? " non_instrumented " : "", + afl->no_forkserver ? "no_fsrv " : "", afl->crash_mode ? "crash " : "", + afl->persistent_mode ? "persistent " : "", + afl->shmem_testcase_mode ? "shmem_testcase " : "", + afl->deferred_mode ? "deferred " : "", + (afl->unicorn_mode || afl->fsrv.qemu_mode || + afl->non_instrumented_mode || afl->no_forkserver || + afl->crash_mode || afl->persistent_mode || afl->deferred_mode) + ? "" + : "default", + afl->orig_cmdline); /* ignore errors */ if (afl->debug) { diff --git a/test/test-floatingpoint.c b/test/test-floatingpoint.c index 083f0df5..acecd55a 100644 --- a/test/test-floatingpoint.c +++ b/test/test-floatingpoint.c @@ -8,16 +8,16 @@ __AFL_FUZZ_INIT(); int main(void) { ssize_t bytes_read; - + __AFL_INIT(); - float *magic = (float*)__AFL_FUZZ_TESTCASE_BUF; - + float *magic = (float *)__AFL_FUZZ_TESTCASE_BUF; + while (__AFL_LOOP(INT_MAX)) { if (__AFL_FUZZ_TESTCASE_LEN != sizeof(float)) return 1; /* 15 + 1/2 + 1/8 + 1/32 + 1/128 */ if ((-*magic == 15.0 + 0.5 + 0.125 + 0.03125 + 0.0078125)) abort(); - + } return 0; -- cgit 1.4.1 From 80f4b32f0bf0bacb60230e04f4779286e0e43bc1 Mon Sep 17 00:00:00 2001 From: van Hauser Date: Sun, 26 Jul 2020 18:41:54 +0200 Subject: new readme --- in progress --- README_new.md | 1054 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 1054 insertions(+) create mode 100644 README_new.md diff --git a/README_new.md b/README_new.md new file mode 100644 index 00000000..7e817fd3 --- /dev/null +++ b/README_new.md @@ -0,0 +1,1054 @@ +# american fuzzy lop plus plus (afl++) + + AFL++ Logo + + ![Travis State](https://api.travis-ci.com/AFLplusplus/AFLplusplus.svg?branch=stable) + + Release Version: [2.66c](https://github.com/AFLplusplus/AFLplusplus/releases) + + Github Version: 2.66d + + Repository: [https://github.com/AFLplusplus/AFLplusplus](https://github.com/AFLplusplus/AFLplusplus) + + afl++ is maintained by: + * Marc "van Hauser" Heuse , + * Heiko "hexcoder-" Eißfeldt , + * Andrea Fioraldi and + * Dominik Maier . + + Originally developed by Michal "lcamtuf" Zalewski. + + Note that although afl now has a Google afl repository [https://github.com/Google/afl](https://github.com/Google/afl), + it is unlikely to receive any notable enhancements: [https://twitter.com/Dor3s/status/1154737061787660288](https://twitter.com/Dor3s/status/1154737061787660288) + includes all necessary/interesting changes from Google's afl 2.57b + +## The enhancements compared to the original stock afl + + Among other changes afl++ has a more performant llvm_mode, supports + llvm up to version 12, QEMU 3.1 with more speed and features, + better *BSD, Solaris and Android support and much, much, much more. + + | Feature/Instrumentation | afl-gcc | llvm_mode | gcc_plugin | qemu_mode | unicorn_mode | + | ----------------------- |:-------:|:---------:|:----------:|:----------------:|:------------:| + | NeverZero | x | x(1) | (2) | x | x | + | Persistent mode | | x | x | x86[_64]/arm[64] | x | + | LAF-Intel / CompCov | | x | | x86[_64]/arm[64] | x86[_64]/arm | + | CmpLog | | x | | x86[_64]/arm[64] | | + | Instrument file list | | x | x | (x)(3) | | + | Non-colliding coverage | | x(4) | | (x)(5) | | + | InsTrim | | x | | | | + | Ngram prev_loc coverage | | x(6) | | | | + | Context coverage | | x | | | | + | Auto dictionary | | x(7) | | | | + | Snapshot LKM support | | x | | (x)(5) | | + + neverZero: + + (1) default for LLVM >= 9.0, env var for older version due an efficiency bug in llvm <= 8 + + (2) GCC creates non-performant code, hence it is disabled in gcc_plugin + + (3) partially via AFL_CODE_START/AFL_CODE_END + + (4) with pcguard mode and LTO mode for LLVM >= 11 + + (5) upcoming, development in the branch + + (6) not compatible with LTO instrumentation and needs at least LLVM >= 4.1 + + (7) only in LTO mode with LLVM >= 11 + + Among others, the following features and patches have been integrated: + + * NeverZero patch for afl-gcc, llvm_mode, qemu_mode and unicorn_mode which prevents a wrapping map value to zero, increases coverage + + * Persistent mode and deferred forkserver for qemu_mode + + * Unicorn mode which allows fuzzing of binaries from completely different platforms (integration provided by domenukk) + + * The new CmpLog instrumentation for LLVM and QEMU inspired by [Redqueen](https://www.syssec.ruhr-uni-bochum.de/media/emma/veroeffentlichungen/2018/12/17/NDSS19-Redqueen.pdf) + + * Win32 PE binary-only fuzzing with QEMU and Wine + + * AFLfast's power schedules by Marcel Böhme: [https://github.com/mboehme/aflfast](https://github.com/mboehme/aflfast) + + * The MOpt mutator: [https://github.com/puppet-meteor/MOpt-AFL](https://github.com/puppet-meteor/MOpt-AFL) + + * LLVM mode Ngram coverage by Adrian Herrera [https://github.com/adrianherrera/afl-ngram-pass](https://github.com/adrianherrera/afl-ngram-pass) + + * InsTrim, an effective CFG llvm_mode instrumentation implementation for large targets: [https://github.com/csienslab/instrim](https://github.com/csienslab/instrim) + + * C. Holler's afl-fuzz Python mutator module and llvm_mode instrument file support: [https://github.com/choller/afl](https://github.com/choller/afl) + + * Custom mutator by a library (instead of Python) by kyakdan + + * LAF-Intel/CompCov support for llvm_mode, qemu_mode and unicorn_mode (with enhanced capabilities) + + * Radamsa and hongfuzz mutators (as custom mutators). + + * QBDI mode to fuzz android native libraries via QBDI framework + + A more thorough list is available in the [PATCHES](docs/PATCHES.md) file. + + So all in all this is the best-of afl that is currently out there :-) + + For new versions and additional information, check out: + [https://github.com/AFLplusplus/AFLplusplus](https://github.com/AFLplusplus/AFLplusplus) + + To compare notes with other users or get notified about major new features, + send a mail to . + + See [docs/QuickStartGuide.md](docs/QuickStartGuide.md) if you don't have time to + read this file. + +## Branches + + The following branches exist: + + * [master/trunk](https://github.com/AFLplusplus/AFLplusplus/) : stable state of afl++ - it is synced from dev from time to + time when we are satisfied with it's stability + * [dev](https://github.com/AFLplusplus/AFLplusplus/tree/dev) : development state of afl++ - bleeding edge and you might catch a + checkout which does not compile or has a bug. *We only accept PRs in dev!!* + * (any other) : experimental branches to work on specific features or testing + new functionality or changes. + + For releases, please see the [Releases](https://github.com/AFLplusplus/AFLplusplus/releases) tab. + +## Google Summer of Code 2020 (and any other students and enthusiast developers) + +We are happy to be part of [Google Summer of Code 2020](https://summerofcode.withgoogle.com/organizations/5100744400699392/)! :-) + +We have several ideas we would like to see in AFL++ to make it even better. +However, we already work on so many things that we do not have the time for +all the big ideas. + +This can be your way to support and contribute to AFL++ - extend it to +something cool. + +We have an idea list in [docs/ideas.md](docs/ideas.md). + +For everyone who wants to contribute (and send pull requests) please read +[CONTRIBUTING.md](CONTRIBUTING.md) before your submit. + +## Building and installing afl++ + +An easy way to install afl++ with everything compiled is available via docker: +You can use the [Dockerfile](Dockerfile) (which has gcc-10 and clang-11 - +hence afl-clang-lto is available!) or just pull directly from the docker hub: +```shell +docker pull aflplusplus/aflplusplus +docker run -ti -v /location/of/your/target:/src aflplusplus/aflplusplus +``` +This image is automatically generated when a push to master happens. +You will find your target source code in /src in the container. + +If you want to build afl++ yourself you have many options. +The easiest is to build and install everything: + +```shell +sudo apt install build-essential libtool-bin python3-dev automake flex bison libglib2.0-dev libpixman-1-dev clang python3-setuptools llvm +make distrib +sudo make install +``` +It is recommended to install the newest available gcc, clang and llvm-dev +possible in your distribution! + +Note that "make distrib" also builds llvm_mode, qemu_mode, unicorn_mode and +more. If you just want plain afl then do "make all", however compiling and +using at least llvm_mode is highly recommended for much better results - +hence in this case + +```shell +make source-only +``` +is what you should choose. + +These build targets exist: + +* all: just the main afl++ binaries +* binary-only: everything for binary-only fuzzing: qemu_mode, unicorn_mode, libdislocator, libtokencap +* source-only: everything for source code fuzzing: llvm_mode, libdislocator, libtokencap +* distrib: everything (for both binary-only and source code fuzzing) +* man: creates simple man pages from the help option of the programs +* install: installs everything you have compiled with the build options above +* clean: cleans everything compiled, not downloads (unless not on a checkout) +* deepclean: cleans everything including downloads +* code-format: format the code, do this before you commit and send a PR please! +* tests: runs test cases to ensure that all features are still working as they should +* unit: perform unit tests (based on cmocka) +* help: shows these build options + +[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 +``` + +These build options exist: + +* STATIC - compile AFL++ static +* ASAN_BUILD - compiles with memory sanitizer for debug purposes +* PROFILING - compile with profiling information (gprof) +* NO_PYTHON - disable python support +* AFL_NO_X86 - if compiling on non-intel/amd platforms +* LLVM_CONFIG - if your distro doesn't use the standard name for llvm-config (e.g. Debian) + +e.g.: make ASAN_BUILD=1 + +## Good examples and writeups + +Here are some good writeups to show how to effectively use AFL++: + + * [https://aflplus.plus/docs/tutorials/libxml2_tutorial/](https://aflplus.plus/docs/tutorials/libxml2_tutorial/) + * [https://bananamafia.dev/post/gb-fuzz/](https://bananamafia.dev/post/gb-fuzz/) + * [https://securitylab.github.com/research/fuzzing-challenges-solutions-1](https://securitylab.github.com/research/fuzzing-challenges-solutions-1) + * [https://securitylab.github.com/research/fuzzing-sockets-FTP](https://securitylab.github.com/research/fuzzing-sockets-FTP) + +If you are interested in fuzzing structured data (where you define what the +structure is), these links have you covered: + * Superion for afl++: [https://github.com/adrian-rt/superion-mutator](https://github.com/adrian-rt/superion-mutator) + * libprotobuf raw: [https://github.com/bruce30262/libprotobuf-mutator_fuzzing_learning/tree/master/4_libprotobuf_aflpp_custom_mutator](https://github.com/bruce30262/libprotobuf-mutator_fuzzing_learning/tree/master/4_libprotobuf_aflpp_custom_mutator) + * libprotobuf for old afl++ API: [https://github.com/thebabush/afl-libprotobuf-mutator](https://github.com/thebabush/afl-libprotobuf-mutator) + +If you find other good ones, please send them to us :-) + +## How to fuzz with afl++ + +The following describes how to fuzz with a target if source code is available. +If you have a binary-only target please skip to [#Instrumenting binary-only apps](#Instrumenting binary-only apps) + +Fuzzing source code is a two step process. + +1. compile the target with a special compiler that prepares the target to be + fuzzed efficiently. This step is called "instrumenting a target". +2. Prepare the fuzzing by selecting and optimizing the input corpus for the + target. +3. perform the fuzzing of the target by randomly mutating input and assessing + if a generated input was processed in a new path in the target binary + +### 1. Instrumenting that target + +#### a) Selecting the best afl++ compiler for instrumenting the target + +afl++ comes with different compilers and instrumentation options. +The following evaluation flow will help you to select the best possible. + +It is highly recommended to have the newest llvm version possible installed, +anything below 9 is not recommended. + +``` ++--------------------------------+ +| clang/clang++ 11+ is available | --> use afl-clang-lto and afl-clang-lto++ ++--------------------------------+ see [llvm/README.lto.md](llvm/README.lto.md) + | + | if not, or if the target fails with with afl-clang-lto/++ + | + v ++---------------------------------+ +| clang/clang++ 3.3+ is available | --> use afl-clang-fast and afl-clang-fast++ ++---------------------------------+ see [llvm/README.md](llvm/README.md) + | + | if not, or if the target fails with afl-clang-fast/++ + | + v + +--------------------------------+ + | if you want to instrument only | -> use afl-gcc-fast and afl-gcc-fast++ + | parts of the target | see [gcc_plugin/README.md](gcc_plugin/README.md) and + +--------------------------------+ [gcc_plugin/README.instrument_file.md](gcc_plugin/README.instrument_file.md) + | + | if not, or if you do not have a gcc with plugin support + | + v + use afl-gcc and afl-g++ +``` + +#### b) Selecting instrumentation options + +The following options are available when you instrument with afl-clang-fast or +afl-clang-lto: + + * Splitting integer, string, float and switch compares so afl++ can easier + solve these. This is an important option if you do not have a very good + good and large input corpus. This technique is called laf-intel or COMPCOV. + To use this set the following environment variable before compiling the + target: `export AFL_LLVM_LAF_ALL=1` + You can read more about this in [llvm/README.laf-intel.md](llvm/README.laf-intel.md) + * A different technique is to instrument the target so that any compare values + in the target are sent to afl++ which then tries to put this value into the + fuzzing data at different locations. This technique is very fast and good - + if the target does not transform input data before comparison. Therefore + technique is called `input to state` or `redqueen`. + If you want to use this technique, then you have to compile the target + twice, once specifically with/for this mode. + You can read more about this in [llvm_mode/README.cmplog.md](llvm_mode/README.cmplog.md) + +If you use afl-clang-fast, afl-clang-lto or afl-gcc-fast you have the option to +selectivly only instrument parts of the target that you are interested in: + + * To instrument only those parts of the target that you are interested in + create a file with all the filenames of the source code that should be + instrumented. + For afl-clang-lto and afl-gcc-fast - or afl-clang-fast if either the clang + version is < 7 or the CLASSIC instrumentation is used - just put one + filename per line, no directory information necessary, and set + `export AFL_LLVM_INSTRUMENT_FILE=yourfile.txt` + see [llvm_mode/README.instrument_file.md](llvm_mode/README.instrument_file.md) + For afl-clang-fast > 6.0 or if PCGUARD instrumentation is used then use the + llvm sancov allow-list feature: [http://clang.llvm.org/docs/SanitizerCoverage.html](http://clang.llvm.org/docs/SanitizerCoverage.html) + +There are many more options and modes available however these are most of the +time less effective. See: + * [llvm_mode/README.ctx.md](llvm_mode/README.ctx.md) + * [llvm_mode/README.ngram.md](llvm_mode/README.ngram.md) + * [llvm_mode/README.instrim.md](llvm_mode/README.instrim.md) + * [llvm_mode/README.neverzero.md](llvm_mode/README.neverzero.md) + +#### c) Modify the target + +If the target has features that makes fuzzing more difficult, e.g. +checksums, HMAC etc. then modify the source code so that this is +removed. +This can even be done for productional source code be eliminating +these checks within this specific defines: + +``` +#ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION + // say that the checksum or HMAC was fine - or whatever is required + // to eliminate the need for the fuzzer to guess the right checksum + return 0; +#endif +``` + +#### d) Instrument the target + +In this step the target source code is compiled so that it can be fuzzed. + +Basically you have to tell the target build system that the selected afl++ +compiler is used. Also - if possible - you should always configure the +build system that the target is compiled statically and not dynamically. +How to do this is described below. + +Then build the target. (Usually with `make`) + +##### configure + +For `configure` build systems this is usually done by: +`CC=afl-clang-fast CXX=afl-clang-fast++ ./configure --disable-shared` + +Note that if you using the (better) afl-clang-lto compiler you also have to +AR to llvm-ar[-VERSION] and RANLIB to llvm-ranlib[-VERSION] - as it is +described in [llvm/README.lto.md](llvm/README.lto.md) + +##### cmake + +For `configure` build systems this is usually done by: +`mkdir build; cd build; CC=afl-clang-fast CXX=afl-clang-fast++ cmake ..` + +Note that if you using the (better) afl-clang-lto compiler you also have to +AR to llvm-ar[-VERSION] and RANLIB to llvm-ranlib[-VERSION] - as it is +described in [llvm/README.lto.md](llvm/README.lto.md) + +##### other build systems or if configure/cmake didn't work + +Sometimes cmake and configure do not pick up the afl compiler, or the ranlib/ar +that is needed - because this was just not foreseen by the developer of the +target. Or they have non-standard options. Figure out if there is a +non-standard way to set this, otherwise set the build normally and edit the +generated build environment afterwards by hand to point to the right compiler +(and/or ranlib and ar). + +#### d) Better instrumentation + +If you just fuzz a target program as-is you are wasting a great opportunity for +much more fuzzing speed. + +This requires the usage of afl-clang-lto or afl-clang-fast + +This is the so-called `persistent mode`, which is much, much faster but +requires that you code a source file that is specifically calling the target +functions that you want to fuzz, plus a few specific afl++ functions around +it. See [llvm_mode/README.persistent_mode.md](llvm_mode/README.persistent_mode.md) for details. + +Basically if you do not fuzz a target in persistent mode then you are just +doing it for a hobby and not professionally :-) + +### 2. Preparing the fuzzing + +As you fuzz the target with mutated input, having as diverse inputs for the +target as possible improves the efficiency a lot. + +#### a) Collect inputs +Try to gather valid inputs for the target from wherever you can. E.g. if it +the PNG picture format try to find as many png files as possible, e.g. from +reported bugs, test suites, random downloads from the internet, unit test +case data - from all kind of PNG software. + +If the input is not known files, you can also modify a target program to write +away normal data it receives and processes to a file and use these. + +#### b) Making the input corpus unique + +Use the afl++ tool `afl-cmin` to remove inputs from the corpus that do not +use a different paths in the target. +Put all files from step a) into one directory, e.g. INPUTS. + +Put all the files from step a) + +If the target program is to be called by fuzzing as `bin/target -d INPUTFILE` +the run afl-cmin like this: +`afl-cmin -i INPUTS -o INPUTS_UNIQUE -- bin/target -d @@` +Note that the INPUTFILE that the target program would read has to be set as `@@`. + +If the target reads from stdin instead, just omit the `@@` as this is the +default. + +#### b) Minimizing all corpus files + +The shorter the input files are so that they still traverse the same path +within the target, the better the fuzzing will be. This is done with `afl-tmin` +however it is a long processes as this has to be done for every file: + +``` +mkdir input +cd INPUTS_UNIQUE +for i in *; do + afl-tmin -i "$i" -o "../input/$i" -- bin/target -d @@ +done +``` + +This can also be parallelized, e.g. with `parallel` + +#### c) done! + +The INPUTS_UNIQUE/ directory from step a) - or even better if you minimized the +corpus in step b) then the files in input/ is then the input corpus directory +to be used in fuzzing! :-) + +### Fuzzing the target + +In this final step we fuzz the target. +There are not that many useful options to run the target - unless you want to +use many CPU cores for the fuzzing, which will make the fuzzing much more useful. + +If you just use one CPU for fuzzing, then you are fuzzing just for fun and not +seriously :-) + +#### a) running afl-fuzz + +If you have an input corpus from step 2 then specify this directory with the `-i` +option. Otherwise create a new directory and create a file with any content +in there. + +If you do not want anything special, the defaults are already the usual best, +hence all you need (from the example in 2a): +`afl-fuzz -i input -o output -- bin/target -d @@` +Note that the directory specified with -o will be created if it does not exist. + +If you need to stop and re-start the fuzzing, use the same command line option +and switch the input directory with a dash (`-`): +`afl-fuzz -i - -o output -- bin/target -d @@` + +afl-fuzz never stops fuzzing. To terminate afl++ simply press Control-C. + +When you start afl-fuzz you will see a user interface that shows what the status +is: +[docs/screenshot.png](docs/screenshot.png) +All the entries are explained in [docs/status_screen.md](docs/status_screen.md) + +#### b) Using multiple cores + +If you want to seriously fuzz then use as many cores as possible to fuzz your +target. + +On the same machine - due to the nature how afl++ works - there is a maximum +number of CPU cores that are useful, more and the overall performance degrades +instead. This value depends on the target and the limit is between 24 and 64 +cores per machine. + +There should be one main fuzzer (`-M main` option) and as many secondary +fuzzers (eg `-S variant1`) as you cores that you use. +Every -M/-S entry needs a unique name (that can be whatever), however the same +-o output directory location has to be used for all. + +For every secondary there should be a variation, e.g.: + * one should fuzz the target that was compiled differently: with sanitizers + activated (`export AFL_USE_ASAN=1 ; export AFL_USE_UBSAN=1 ; + export AFL_USE_CFISAN=1 ; ` + * one should fuzz the target with CMPLOG/redqueen (see above) + * At 1-2 should fuzz a target compiled with laf-intel/COMPCOV (see above). + +All other secondaries should be: + * 1/2 with MOpt option enabled: `-L 0` + * run with a different power schedule, available are: + `explore (default), fast, coe, lin, quad, exploit, mmopt, rare, seek` + which you can set with e.g. `-p seek` + +You can also use different fuzzers. +If you are afl-spinoffs or afl conforming, then just use the same -o directory +and give it a unique `-S` name. +Examples are e.g.: + * + * + * + +However you can also sync afl++ with honggfuzz, libfuzzer, entropic, etc. +Just show the main fuzzer (-M) with the `-F` option where the queue +directory of these other fuzzers are, e.g. `-F /src/target/honggfuzz` + +#### c) How long to fuzz a target? + +This is a difficult question. +Basically if no new path is found for a long time (e.g. for a day or a week) +then you can expect that your fuzzing won't be fruitful anymore. +However often this just means that you should switch out secondaries for +others, e.g. custom mutator modules, sync to very different fuzzers, etc. + +### The End + +This is basically all you need to know to professionally run fuzzing campaigns. +If you want to know more, the rest of this README and the tons of texts in +[docs/](docs/) will have you covered. + +## Challenges of guided fuzzing + +Fuzzing is one of the most powerful and proven strategies for identifying +security issues in real-world software; it is responsible for the vast +majority of remote code execution and privilege escalation bugs found to date +in security-critical software. + +Unfortunately, fuzzing is also relatively shallow; blind, random mutations +make it very unlikely to reach certain code paths in the tested code, leaving +some vulnerabilities firmly outside the reach of this technique. + +There have been numerous attempts to solve this problem. One of the early +approaches - pioneered by Tavis Ormandy - is corpus distillation. The method +relies on coverage signals to select a subset of interesting seeds from a +massive, high-quality corpus of candidate files, and then fuzz them by +traditional means. The approach works exceptionally well but requires such +a corpus to be readily available. In addition, block coverage measurements +provide only a very simplistic understanding of the program state and are less +useful for guiding the fuzzing effort in the long haul. + +Other, more sophisticated research has focused on techniques such as program +flow analysis ("concolic execution"), symbolic execution, or static analysis. +All these methods are extremely promising in experimental settings, but tend +to suffer from reliability and performance problems in practical uses - and +currently do not offer a viable alternative to "dumb" fuzzing techniques. + + +## The afl-fuzz approach + +American Fuzzy Lop is a brute-force fuzzer coupled with an exceedingly simple +but rock-solid instrumentation-guided genetic algorithm. It uses a modified +form of edge coverage to effortlessly pick up subtle, local-scale changes to +program control flow. + +Simplifying a bit, the overall algorithm can be summed up as: + + 1) Load user-supplied initial test cases into the queue, + + 2) Take the next input file from the queue, + + 3) Attempt to trim the test case to the smallest size that doesn't alter + the measured behavior of the program, + + 4) Repeatedly mutate the file using a balanced and well-researched variety + of traditional fuzzing strategies, + + 5) If any of the generated mutations resulted in a new state transition + recorded by the instrumentation, add mutated output as a new entry in the + queue. + + 6) Go to 2. + +The discovered test cases are also periodically culled to eliminate ones that +have been obsoleted by newer, higher-coverage finds; and undergo several other +instrumentation-driven effort minimization steps. + +As a side result of the fuzzing process, the tool creates a small, +self-contained corpus of interesting test cases. These are extremely useful +for seeding other, labor- or resource-intensive testing regimes - for example, +for stress-testing browsers, office applications, graphics suites, or +closed-source tools. + +The fuzzer is thoroughly tested to deliver out-of-the-box performance far +superior to blind fuzzing or coverage-only tools. + + +## Instrumenting programs for use with AFL + +PLEASE NOTE: llvm_mode compilation with afl-clang-fast/afl-clang-fast++ +instead of afl-gcc/afl-g++ is much faster and has many cool features. +See llvm_mode/ - however few code does not compile with llvm. +We support llvm versions 3.4 to 12. + +When source code is available, instrumentation can be injected by a companion +tool that works as a drop-in replacement for gcc or clang in any standard build +process for third-party code. + +The instrumentation has a fairly modest performance impact; in conjunction with +other optimizations implemented by afl-fuzz, most programs can be fuzzed as fast +or even faster than possible with traditional tools. + +The correct way to recompile the target program may vary depending on the +specifics of the build process, but a nearly-universal approach would be: + +```shell +CC=/path/to/afl/afl-gcc ./configure +make clean all +``` + +For C++ programs, you'd would also want to set `CXX=/path/to/afl/afl-g++`. + +The clang wrappers (afl-clang and afl-clang++) can be used in the same way; +clang users may also opt to leverage a higher-performance instrumentation mode, +as described in [llvm_mode/README.md](llvm_mode/README.md). +Clang/LLVM has a much better performance and works with LLVM version 3.4 to 12. + +Using the LAF Intel performance enhancements are also recommended, see +[llvm_mode/README.laf-intel.md](llvm_mode/README.laf-intel.md) + +Using partial instrumentation is also recommended, see +[llvm_mode/README.instrument_file.md](llvm_mode/README.instrument_file.md) + +When testing libraries, you need to find or write a simple program that reads +data from stdin or from a file and passes it to the tested library. In such a +case, it is essential to link this executable against a static version of the +instrumented library or to make sure that the correct .so file is loaded at +runtime (usually by setting `LD_LIBRARY_PATH`). The simplest option is a static +build, usually possible via: + +```shell +CC=/path/to/afl/afl-gcc ./configure --disable-shared +``` + +Setting `AFL_HARDEN=1` when calling 'make' will cause the CC wrapper to +automatically enable code hardening options that make it easier to detect +simple memory bugs. Libdislocator, a helper library included with AFL (see +[libdislocator/README.md](libdislocator/README.md)) can help uncover heap corruption issues, too. + +PS. ASAN users are advised to review [docs/notes_for_asan.md](docs/notes_for_asan.md) +file for important caveats. + + +## Instrumenting binary-only apps + +When source code is *NOT* available, the fuzzer offers experimental support for +fast, on-the-fly instrumentation of black-box binaries. This is accomplished +with a version of QEMU running in the lesser-known "user space emulation" mode. + +QEMU is a project separate from AFL, but you can conveniently build the +feature by doing: + +```shell +cd qemu_mode +./build_qemu_support.sh +``` + +For additional instructions and caveats, see [qemu_mode/README.md](qemu_mode/README.md). + +If possible you should use the persistent mode, see [qemu_mode/README.persistent.md](qemu_mode/README.persistent.md). + +The mode is approximately 2-5x slower than compile-time instrumentation, is +less conducive to parallelization, and may have some other quirks. + +If [afl-dyninst](https://github.com/vanhauser-thc/afl-dyninst) works for +your binary, then you can use afl-fuzz normally and it will have twice +the speed compared to qemu_mode. + +A more comprehensive description of these and other options can be found in +[docs/binaryonly_fuzzing.md](docs/binaryonly_fuzzing.md) + +## Power schedules + +The power schedules were copied from Marcel Böhme's AFLfast implementation and +measure differently which queue entries to prefer and therefore may find +different paths faster for large queues. + +The available schedules are: + + - explore (default, original AFL) + - exploit (original AFL) + - fast (AFLfast) + - coe (AFLfast) + - quad (AFLfast) + - lin (AFLfast) + - rare (afl++ experimental) + - mmopt (afl++ experimental) + - seek (afl++ experimental) + +In parallel mode (-M/-S, several instances with the shared queue), we suggest +to run the main node using the default explore schedule (`-p explore`) and the +secondary nodes with different schedules. If a schedule does not perform well +for a target, restart the secondary nodes with a different schedule. + +More details can be found in the paper published at the 23rd ACM Conference on +Computer and Communications Security [CCS'16](https://www.sigsac.org/ccs/CCS2016/accepted-papers/) + +## Choosing initial test cases + +To operate correctly, the fuzzer requires one or more starting file that +contains a good example of the input data normally expected by the targeted +application. There are two basic rules: + + - Keep the files small. Under 1 kB is ideal, although not strictly necessary. + For a discussion of why size matters, see [perf_tips.md](docs/perf_tips.md). + + - Use multiple test cases only if they are functionally different from + each other. There is no point in using fifty different vacation photos + to fuzz an image library. + +You can find many good examples of starting files in the testcases/ subdirectory +that comes with this tool. + +PS. If a large corpus of data is available for screening, you may want to use +the afl-cmin utility to identify a subset of functionally distinct files that +exercise different code paths in the target binary. + + +## Fuzzing binaries + +The fuzzing process itself is carried out by the afl-fuzz utility. This program +requires a read-only directory with initial test cases, a separate place to +store its findings, plus a path to the binary to test. + +For target binaries that accept input directly from stdin, the usual syntax is: + +```shell +./afl-fuzz -i testcase_dir -o findings_dir /path/to/program [...params...] +``` + +For programs that take input from a file, use '@@' to mark the location in +the target's command line where the input file name should be placed. The +fuzzer will substitute this for you: + +```shell +./afl-fuzz -i testcase_dir -o findings_dir /path/to/program @@ +``` + +You can also use the -f option to have the mutated data written to a specific +file. This is useful if the program expects a particular file extension or so. + +Non-instrumented binaries can be fuzzed in the QEMU mode (add -Q in the command +line) or in a traditional, blind-fuzzer mode (specify -n). + +You can use -t and -m to override the default timeout and memory limit for the +executed process; rare examples of targets that may need these settings touched +include compilers and video decoders. + +Tips for optimizing fuzzing performance are discussed in [perf_tips.md](docs/perf_tips.md). + +Note that afl-fuzz starts by performing an array of deterministic fuzzing +steps, which can take several days, but tend to produce neat test cases. If you +want quick & dirty results right away - akin to zzuf and other traditional +fuzzers - add the -d option to the command line. + +## Interpreting output + +See the [docs/status_screen.md](docs/status_screen.md) file for information on +how to interpret the displayed stats and monitor the health of the process. Be +sure to consult this file especially if any UI elements are highlighted in red. + +The fuzzing process will continue until you press Ctrl-C. At a minimum, you want +to allow the fuzzer to complete one queue cycle, which may take anywhere from a +couple of hours to a week or so. + +There are three subdirectories created within the output directory and updated +in real-time: + + - queue/ - test cases for every distinctive execution path, plus all the + starting files given by the user. This is the synthesized corpus + mentioned in section 2. + + Before using this corpus for any other purposes, you can shrink + it to a smaller size using the afl-cmin tool. The tool will find + a smaller subset of files offering equivalent edge coverage. + + - crashes/ - unique test cases that cause the tested program to receive a + fatal signal (e.g., SIGSEGV, SIGILL, SIGABRT). The entries are + grouped by the received signal. + + - hangs/ - unique test cases that cause the tested program to time out. The + default time limit before something is classified as a hang is + the larger of 1 second and the value of the -t parameter. + The value can be fine-tuned by setting AFL_HANG_TMOUT, but this + is rarely necessary. + +Crashes and hangs are considered "unique" if the associated execution paths +involve any state transitions not seen in previously-recorded faults. If a +single bug can be reached in multiple ways, there will be some count inflation +early in the process, but this should quickly taper off. + +The file names for crashes and hangs are correlated with the parent, non-faulting +queue entries. This should help with debugging. + +When you can't reproduce a crash found by afl-fuzz, the most likely cause is +that you are not setting the same memory limit as used by the tool. Try: + +```shell +LIMIT_MB=50 +( ulimit -Sv $[LIMIT_MB << 10]; /path/to/tested_binary ... ) +``` + +Change LIMIT_MB to match the -m parameter passed to afl-fuzz. On OpenBSD, +also change -Sv to -Sd. + +Any existing output directory can be also used to resume aborted jobs; try: + +```shell +./afl-fuzz -i- -o existing_output_dir [...etc...] +``` + +If you have gnuplot installed, you can also generate some pretty graphs for any +active fuzzing task using afl-plot. For an example of how this looks like, +see [http://lcamtuf.coredump.cx/afl/plot/](http://lcamtuf.coredump.cx/afl/plot/). + +## Parallelized fuzzing + +Every instance of afl-fuzz takes up roughly one core. This means that on +multi-core systems, parallelization is necessary to fully utilize the hardware. +For tips on how to fuzz a common target on multiple cores or multiple networked +machines, please refer to [docs/parallel_fuzzing.md](docs/parallel_fuzzing.md). + +The parallel fuzzing mode also offers a simple way for interfacing AFL to other +fuzzers, to symbolic or concolic execution engines, and so forth; again, see the +last section of [docs/parallel_fuzzing.md](docs/parallel_fuzzing.md) for tips. + +## Fuzzer dictionaries + +By default, afl-fuzz mutation engine is optimized for compact data formats - +say, images, multimedia, compressed data, regular expression syntax, or shell +scripts. It is somewhat less suited for languages with particularly verbose and +redundant verbiage - notably including HTML, SQL, or JavaScript. + +To avoid the hassle of building syntax-aware tools, afl-fuzz provides a way to +seed the fuzzing process with an optional dictionary of language keywords, +magic headers, or other special tokens associated with the targeted data type +-- and use that to reconstruct the underlying grammar on the go: + + [http://lcamtuf.blogspot.com/2015/01/afl-fuzz-making-up-grammar-with.html](http://lcamtuf.blogspot.com/2015/01/afl-fuzz-making-up-grammar-with.html) + +To use this feature, you first need to create a dictionary in one of the two +formats discussed in [dictionaries/README.md](dictionaries/README.md); +and then point the fuzzer to it via the -x option in the command line. + +(Several common dictionaries are already provided in that subdirectory, too.) + +There is no way to provide more structured descriptions of the underlying +syntax, but the fuzzer will likely figure out some of this based on the +instrumentation feedback alone. This actually works in practice, say: + + [http://lcamtuf.blogspot.com/2015/04/finding-bugs-in-sqlite-easy-way.html](http://lcamtuf.blogspot.com/2015/04/finding-bugs-in-sqlite-easy-way.html) + +PS. Even when no explicit dictionary is given, afl-fuzz will try to extract +existing syntax tokens in the input corpus by watching the instrumentation +very closely during deterministic byte flips. This works for some types of +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.tokencap.md). + +## Crash triage + +The coverage-based grouping of crashes usually produces a small data set that +can be quickly triaged manually or with a very simple GDB or Valgrind script. +Every crash is also traceable to its parent non-crashing test case in the +queue, making it easier to diagnose faults. + +Having said that, it's important to acknowledge that some fuzzing crashes can be +difficult to quickly evaluate for exploitability without a lot of debugging and +code analysis work. To assist with this task, afl-fuzz supports a very unique +"crash exploration" mode enabled with the -C flag. + +In this mode, the fuzzer takes one or more crashing test cases as the input +and uses its feedback-driven fuzzing strategies to very quickly enumerate all +code paths that can be reached in the program while keeping it in the +crashing state. + +Mutations that do not result in a crash are rejected; so are any changes that +do not affect the execution path. + +The output is a small corpus of files that can be very rapidly examined to see +what degree of control the attacker has over the faulting address, or whether +it is possible to get past an initial out-of-bounds read - and see what lies +beneath. + +Oh, one more thing: for test case minimization, give afl-tmin a try. The tool +can be operated in a very simple way: + +```shell +./afl-tmin -i test_case -o minimized_result -- /path/to/program [...] +``` + +The tool works with crashing and non-crashing test cases alike. In the crash +mode, it will happily accept instrumented and non-instrumented binaries. In the +non-crashing mode, the minimizer relies on standard AFL instrumentation to make +the file simpler without altering the execution path. + +The minimizer accepts the -m, -t, -f and @@ syntax in a manner compatible with +afl-fuzz. + +Another recent addition to AFL is the afl-analyze tool. It takes an input +file, attempts to sequentially flip bytes, and observes the behavior of the +tested program. It then color-codes the input based on which sections appear to +be critical, and which are not; while not bulletproof, it can often offer quick +insights into complex file formats. More info about its operation can be found +near the end of [docs/technical_details.md](docs/technical_details.md). + +## Going beyond crashes + +Fuzzing is a wonderful and underutilized technique for discovering non-crashing +design and implementation errors, too. Quite a few interesting bugs have been +found by modifying the target programs to call abort() when say: + + - Two bignum libraries produce different outputs when given the same + fuzzer-generated input, + + - An image library produces different outputs when asked to decode the same + input image several times in a row, + + - A serialization / deserialization library fails to produce stable outputs + when iteratively serializing and deserializing fuzzer-supplied data, + + - A compression library produces an output inconsistent with the input file + when asked to compress and then decompress a particular blob. + +Implementing these or similar sanity checks usually takes very little time; +if you are the maintainer of a particular package, you can make this code +conditional with `#ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION` (a flag also +shared with libfuzzer) or `#ifdef __AFL_COMPILER` (this one is just for AFL). + +## Common-sense risks + +Please keep in mind that, similarly to many other computationally-intensive +tasks, fuzzing may put a strain on your hardware and on the OS. In particular: + + - Your CPU will run hot and will need adequate cooling. In most cases, if + cooling is insufficient or stops working properly, CPU speeds will be + automatically throttled. That said, especially when fuzzing on less + suitable hardware (laptops, smartphones, etc), it's not entirely impossible + for something to blow up. + + - Targeted programs may end up erratically grabbing gigabytes of memory or + filling up disk space with junk files. AFL tries to enforce basic memory + limits, but can't prevent each and every possible mishap. The bottom line + is that you shouldn't be fuzzing on systems where the prospect of data loss + is not an acceptable risk. + + - Fuzzing involves billions of reads and writes to the filesystem. On modern + systems, this will be usually heavily cached, resulting in fairly modest + "physical" I/O - but there are many factors that may alter this equation. + It is your responsibility to monitor for potential trouble; with very heavy + I/O, the lifespan of many HDDs and SSDs may be reduced. + + A good way to monitor disk I/O on Linux is the 'iostat' command: + +```shell + $ iostat -d 3 -x -k [...optional disk ID...] +``` + +## Known limitations & areas for improvement + +Here are some of the most important caveats for AFL: + + - AFL detects faults by checking for the first spawned process dying due to + a signal (SIGSEGV, SIGABRT, etc). Programs that install custom handlers for + these signals may need to have the relevant code commented out. In the same + vein, faults in child processes spawned by the fuzzed target may evade + detection unless you manually add some code to catch that. + + - As with any other brute-force tool, the fuzzer offers limited coverage if + encryption, checksums, cryptographic signatures, or compression are used to + wholly wrap the actual data format to be tested. + + To work around this, you can comment out the relevant checks (see + examples/libpng_no_checksum/ for inspiration); if this is not possible, + you can also write a postprocessor, one of the hooks of custom mutators. + See [docs/custom_mutators.md](docs/custom_mutators.md) on how to use + `AFL_CUSTOM_MUTATOR_LIBRARY` + + - There are some unfortunate trade-offs with ASAN and 64-bit binaries. This + isn't due to any specific fault of afl-fuzz; see [docs/notes_for_asan.md](docs/notes_for_asan.md) + for tips. + + - There is no direct support for fuzzing network services, background + daemons, or interactive apps that require UI interaction to work. You may + need to make simple code changes to make them behave in a more traditional + way. Preeny may offer a relatively simple option, too - see: + [https://github.com/zardus/preeny](https://github.com/zardus/preeny) + + Some useful tips for modifying network-based services can be also found at: + [https://www.fastly.com/blog/how-to-fuzz-server-american-fuzzy-lop](https://www.fastly.com/blog/how-to-fuzz-server-american-fuzzy-lop) + + - AFL doesn't output human-readable coverage data. If you want to monitor + coverage, use afl-cov from Michael Rash: [https://github.com/mrash/afl-cov](https://github.com/mrash/afl-cov) + + - Occasionally, sentient machines rise against their creators. If this + happens to you, please consult [http://lcamtuf.coredump.cx/prep/](http://lcamtuf.coredump.cx/prep/). + +Beyond this, see INSTALL for platform-specific tips. + +## Special thanks + +Many of the improvements to the original afl and afl++ wouldn't be possible +without feedback, bug reports, or patches from: + +``` + Jann Horn Hanno Boeck + Felix Groebert Jakub Wilk + Richard W. M. Jones Alexander Cherepanov + Tom Ritter Hovik Manucharyan + Sebastian Roschke Eberhard Mattes + Padraig Brady Ben Laurie + @dronesec Luca Barbato + Tobias Ospelt Thomas Jarosch + Martin Carpenter Mudge Zatko + Joe Zbiciak Ryan Govostes + Michael Rash William Robinet + Jonathan Gray Filipe Cabecinhas + Nico Weber Jodie Cunningham + Andrew Griffiths Parker Thompson + Jonathan Neuschaefer Tyler Nighswander + Ben Nagy Samir Aguiar + Aidan Thornton Aleksandar Nikolich + Sam Hakim Laszlo Szekeres + David A. Wheeler Turo Lamminen + Andreas Stieger Richard Godbee + Louis Dassy teor2345 + Alex Moneger Dmitry Vyukov + Keegan McAllister Kostya Serebryany + Richo Healey Martijn Bogaard + rc0r Jonathan Foote + Christian Holler Dominique Pelle + Jacek Wielemborek Leo Barnes + Jeremy Barnes Jeff Trull + Guillaume Endignoux ilovezfs + Daniel Godas-Lopez Franjo Ivancic + Austin Seipp Daniel Komaromy + Daniel Binderman Jonathan Metzman + Vegard Nossum Jan Kneschke + Kurt Roeckx Marcel Boehme + Van-Thuan Pham Abhik Roychoudhury + Joshua J. Drake Toby Hutton + Rene Freingruber Sergey Davidoff + Sami Liedes Craig Young + Andrzej Jackowski Daniel Hodson + Nathan Voss Dominik Maier + Andrea Biondo Vincent Le Garrec + Khaled Yakdan Kuang-che Wu + Josephine Calliotte Konrad Welc +``` + +Thank you! +(For people sending pull requests - please add yourself to this list :-) + +## Contact + +Questions? Concerns? Bug reports? The contributors can be reached via +[https://github.com/AFLplusplus/AFLplusplus](https://github.com/AFLplusplus/AFLplusplus) + +There is also a mailing list for the afl project; to join, send a mail to +. Or, if you prefer to browse +archives first, try: [https://groups.google.com/group/afl-users](https://groups.google.com/group/afl-users) -- cgit 1.4.1 From 2323c30b5b655e207bfb143fec34621052ea4e90 Mon Sep 17 00:00:00 2001 From: van Hauser Date: Sun, 26 Jul 2020 22:05:14 +0200 Subject: fix honggfuzz dict --- custom_mutators/honggfuzz/honggfuzz.c | 4 +++- src/afl-fuzz-stats.c | 2 +- 2 files changed, 4 insertions(+), 2 deletions(-) diff --git a/custom_mutators/honggfuzz/honggfuzz.c b/custom_mutators/honggfuzz/honggfuzz.c index 368741c1..bde922c6 100644 --- a/custom_mutators/honggfuzz/honggfuzz.c +++ b/custom_mutators/honggfuzz/honggfuzz.c @@ -68,6 +68,8 @@ void afl_custom_queue_new_entry(my_mutator_t * data, const uint8_t *filename_new_queue, const uint8_t *filename_orig_queue) { + if (run.global->mutate.dictionaryCnt >= 1024) return; + while (data->extras_cnt < data->afl->extras_cnt && run.global->mutate.dictionaryCnt < 1024) { @@ -81,7 +83,7 @@ void afl_custom_queue_new_entry(my_mutator_t * data, } - while (data->extras_cnt < data->afl->a_extras_cnt && + while (data->a_extras_cnt < data->afl->a_extras_cnt && run.global->mutate.dictionaryCnt < 1024) { memcpy(run.global->mutate.dictionary[run.global->mutate.dictionaryCnt].val, diff --git a/src/afl-fuzz-stats.c b/src/afl-fuzz-stats.c index 995f298e..2546a57a 100644 --- a/src/afl-fuzz-stats.c +++ b/src/afl-fuzz-stats.c @@ -71,7 +71,7 @@ void write_stats_file(afl_state_t *afl, double bitmap_cvg, double stability, cur_time - afl->last_avg_exec_update >= 60000))) { afl->last_avg_execs_saved = - (float)(1000*(afl->fsrv.total_execs - afl->last_avg_execs)) / + (float)(1000 * (afl->fsrv.total_execs - afl->last_avg_execs)) / (float)(cur_time - afl->last_avg_exec_update); afl->last_avg_execs = afl->fsrv.total_execs; afl->last_avg_exec_update = cur_time; -- cgit 1.4.1 From c3f65bff5b904a430f36d5240bfb680bbb577ccb Mon Sep 17 00:00:00 2001 From: van Hauser Date: Mon, 27 Jul 2020 00:13:03 +0200 Subject: update new readme --- README_new.md | 36 +++++++++++++++++++++++------------- 1 file changed, 23 insertions(+), 13 deletions(-) diff --git a/README_new.md b/README_new.md index 7e817fd3..6b56f2ab 100644 --- a/README_new.md +++ b/README_new.md @@ -11,22 +11,32 @@ Repository: [https://github.com/AFLplusplus/AFLplusplus](https://github.com/AFLplusplus/AFLplusplus) afl++ is maintained by: - * Marc "van Hauser" Heuse , - * Heiko "hexcoder-" Eißfeldt , - * Andrea Fioraldi and - * Dominik Maier . + + * Marc "van Hauser" Heuse , + * Heiko "hexcoder-" Eißfeldt , + * Andrea Fioraldi and + * Dominik Maier . Originally developed by Michal "lcamtuf" Zalewski. - Note that although afl now has a Google afl repository [https://github.com/Google/afl](https://github.com/Google/afl), - it is unlikely to receive any notable enhancements: [https://twitter.com/Dor3s/status/1154737061787660288](https://twitter.com/Dor3s/status/1154737061787660288) - includes all necessary/interesting changes from Google's afl 2.57b + afl++ is superiour to Google's afl in any way - more speed, more and better + mutations, more and better instrumentation, etc. etc. + +## Contents + + 1. [Features](#important-features-of-afl++) + 2. [How to compile and install afl++](#building-and-installing-afl++) + 3. [How to fuzz a target](#how-to-fuzz-with-afl++) + 4. [Fuzzing binary-only targets](#fuzzing-binary-only-targets) + 5. [Branches](#branches) + 6. [Want to help?](#help-wanted) + 7. [Detailed help and description of afl++](#challenges-of-guided-fuzzing) -## The enhancements compared to the original stock afl +## Important features of afl++ - Among other changes afl++ has a more performant llvm_mode, supports - llvm up to version 12, QEMU 3.1 with more speed and features, - better *BSD, Solaris and Android support and much, much, much more. + afl++ supports llvm up to version 12, very fast binary fuzzing with QEMU 3.1 + with laf-intel and redqueen, unicorn mode, gcc plugin, full *BSD, Solaris and + Android support and much, much, much more. | Feature/Instrumentation | afl-gcc | llvm_mode | gcc_plugin | qemu_mode | unicorn_mode | | ----------------------- |:-------:|:---------:|:----------:|:----------------:|:------------:| @@ -114,7 +124,7 @@ For releases, please see the [Releases](https://github.com/AFLplusplus/AFLplusplus/releases) tab. -## Google Summer of Code 2020 (and any other students and enthusiast developers) +## Help wanted We are happy to be part of [Google Summer of Code 2020](https://summerofcode.withgoogle.com/organizations/5100744400699392/)! :-) @@ -632,7 +642,7 @@ PS. ASAN users are advised to review [docs/notes_for_asan.md](docs/notes_for_asa file for important caveats. -## Instrumenting binary-only apps +## Fuzzing binary-only targets When source code is *NOT* available, the fuzzer offers experimental support for fast, on-the-fly instrumentation of black-box binaries. This is accomplished -- cgit 1.4.1 From 298ff5c7d0b8b18ccdc5784f54606ddfc5b6b89b Mon Sep 17 00:00:00 2001 From: van Hauser Date: Mon, 27 Jul 2020 00:14:28 +0200 Subject: update new readme --- README_new.md | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/README_new.md b/README_new.md index 6b56f2ab..9cf4e3cc 100644 --- a/README_new.md +++ b/README_new.md @@ -24,9 +24,9 @@ ## Contents - 1. [Features](#important-features-of-afl++) - 2. [How to compile and install afl++](#building-and-installing-afl++) - 3. [How to fuzz a target](#how-to-fuzz-with-afl++) + 1. [Features](#important-features-of-afl) + 2. [How to compile and install afl++](#building-and-installing-afl) + 3. [How to fuzz a target](#how-to-fuzz-with-afl) 4. [Fuzzing binary-only targets](#fuzzing-binary-only-targets) 5. [Branches](#branches) 6. [Want to help?](#help-wanted) -- cgit 1.4.1 From 7f3317110e07b2d3c63a97cd1916a2b489c78f67 Mon Sep 17 00:00:00 2001 From: van Hauser Date: Mon, 27 Jul 2020 00:16:09 +0200 Subject: update new readme --- README_new.md | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) diff --git a/README_new.md b/README_new.md index 9cf4e3cc..9b8c1014 100644 --- a/README_new.md +++ b/README_new.md @@ -28,9 +28,10 @@ 2. [How to compile and install afl++](#building-and-installing-afl) 3. [How to fuzz a target](#how-to-fuzz-with-afl) 4. [Fuzzing binary-only targets](#fuzzing-binary-only-targets) - 5. [Branches](#branches) - 6. [Want to help?](#help-wanted) - 7. [Detailed help and description of afl++](#challenges-of-guided-fuzzing) + 5. [Good examples and writeups of afl++ usages](#good-examples-and-writeups) + 6. [Branches](#branches) + 7. [Want to help?](#help-wanted) + 8. [Detailed help and description of afl++](#challenges-of-guided-fuzzing) ## Important features of afl++ -- cgit 1.4.1 From 1bbeef48e154389cb5ac5adcb7a55f5b78c2bac6 Mon Sep 17 00:00:00 2001 From: van Hauser Date: Mon, 27 Jul 2020 09:10:48 +0200 Subject: update readme, renice -20 --- GNUmakefile | 2 +- README.md | 595 +++++++++++++++++++++++----- README_new.md | 1065 --------------------------------------------------- docs/screenshot.png | Bin 0 -> 117199 bytes src/afl-fuzz.c | 1 + 5 files changed, 500 insertions(+), 1163 deletions(-) delete mode 100644 README_new.md create mode 100644 docs/screenshot.png diff --git a/GNUmakefile b/GNUmakefile index ab9144b8..e2d7314f 100644 --- a/GNUmakefile +++ b/GNUmakefile @@ -96,7 +96,7 @@ ifneq "$(shell uname -m)" "x86_64" endif CFLAGS ?= -O3 -funroll-loops $(CFLAGS_OPT) -override CFLAGS += -Wall -g -Wno-pointer-sign -Wmissing-declarations\ +override CFLAGS += -Wall -g -Wno-pointer-sign -Wmissing-declarations -Wno-unused-result \ -I include/ -DAFL_PATH=\"$(HELPER_PATH)\" \ -DBIN_PATH=\"$(BIN_PATH)\" -DDOC_PATH=\"$(DOC_PATH)\" diff --git a/README.md b/README.md index 9c802285..97fa99b7 100644 --- a/README.md +++ b/README.md @@ -1,4 +1,4 @@ -# american fuzzy lop plus plus (afl++) +# American Fuzzy Lop plus plus (afl++) AFL++ Logo @@ -8,61 +8,36 @@ Github Version: 2.66d - includes all necessary/interesting changes from Google's afl 2.56b - - Originally developed by Michal "lcamtuf" Zalewski. - Repository: [https://github.com/AFLplusplus/AFLplusplus](https://github.com/AFLplusplus/AFLplusplus) afl++ is maintained by: - * Marc "van Hauser" Heuse , - * Heiko "hexcoder-" Eißfeldt , - * Andrea Fioraldi and - * Dominik Maier . - - Note that although afl now has a Google afl repository [https://github.com/Google/afl](https://github.com/Google/afl), - it is unlikely to receive any notable enhancements: [https://twitter.com/Dor3s/status/1154737061787660288](https://twitter.com/Dor3s/status/1154737061787660288) - -## The enhancements compared to the original stock afl - - Many improvements were made over the official afl release - which did not - get any feature improvements since November 2017. - - Among other changes afl++ has a more performant llvm_mode, supports - llvm up to version 12, QEMU 3.1, more speed and crashfixes for QEMU, - better *BSD and Android support and much, much more. - Additionally the following features and patches have been integrated: - - * AFLfast's power schedules by Marcel Böhme: [https://github.com/mboehme/aflfast](https://github.com/mboehme/aflfast) - - * The new excellent MOpt mutator: [https://github.com/puppet-meteor/MOpt-AFL](https://github.com/puppet-meteor/MOpt-AFL) - - * InsTrim, a very effective CFG llvm_mode instrumentation implementation for large targets: [https://github.com/csienslab/instrim](https://github.com/csienslab/instrim) - - * C. Holler's afl-fuzz Python mutator module and llvm_mode instrument file support: [https://github.com/choller/afl](https://github.com/choller/afl) + * Marc "van Hauser" Heuse , + * Heiko "hexcoder-" Eißfeldt , + * Andrea Fioraldi and + * Dominik Maier . - * Custom mutator by a library (instead of Python) by kyakdan - - * Unicorn mode which allows fuzzing of binaries from completely different platforms (integration provided by domenukk) - - * LAF-Intel or CompCov support for llvm_mode, qemu_mode and unicorn_mode - - * NeverZero patch for afl-gcc, llvm_mode, qemu_mode and unicorn_mode which prevents a wrapping map value to zero, increases coverage - - * Persistent mode and deferred forkserver for qemu_mode - - * Win32 PE binary-only fuzzing with QEMU and Wine + Originally developed by Michal "lcamtuf" Zalewski. - * Radamsa mutator (as a custom mutator). + afl++ is a superiour fork to Google's afl - more speed, more and better + mutations, more and better instrumentation, custom module support, etc. - * QBDI mode to fuzz android native libraries via QBDI framework +## Contents - * The new CmpLog instrumentation for LLVM and QEMU inspired by [Redqueen](https://www.syssec.ruhr-uni-bochum.de/media/emma/veroeffentlichungen/2018/12/17/NDSS19-Redqueen.pdf) + 1. [Features](#important-features-of-afl) + 2. [How to compile and install afl++](#building-and-installing-afl) + 3. [How to fuzz a target](#how-to-fuzz-with-afl) + 4. [Fuzzing binary-only targets](#fuzzing-binary-only-targets) + 5. [Good examples and writeups of afl++ usages](#good-examples-and-writeups) + 6. [Branches](#branches) + 7. [Want to help?](#help-wanted) + 8. [Detailed help and description of afl++](#challenges-of-guided-fuzzing) - * LLVM mode Ngram coverage by Adrian Herrera [https://github.com/adrianherrera/afl-ngram-pass](https://github.com/adrianherrera/afl-ngram-pass) +## Important features of afl++ - A more thorough list is available in the [PATCHES](docs/PATCHES.md) file. + afl++ supports llvm up to version 12, very fast binary fuzzing with QEMU 3.1 + with laf-intel and redqueen, unicorn mode, gcc plugin, full *BSD, Solaris and + Android support and much, much, much more. | Feature/Instrumentation | afl-gcc | llvm_mode | gcc_plugin | qemu_mode | unicorn_mode | | ----------------------- |:-------:|:---------:|:----------:|:----------------:|:------------:| @@ -75,6 +50,7 @@ | InsTrim | | x | | | | | Ngram prev_loc coverage | | x(6) | | | | | Context coverage | | x | | | | + | Auto dictionary | | x(7) | | | | | Snapshot LKM support | | x | | (x)(5) | | neverZero: @@ -85,11 +61,45 @@ (3) partially via AFL_CODE_START/AFL_CODE_END - (4) Only for LLVM >= 11 and not all targets compile + (4) with pcguard mode and LTO mode for LLVM >= 11 (5) upcoming, development in the branch (6) not compatible with LTO instrumentation and needs at least LLVM >= 4.1 + + (7) only in LTO mode with LLVM >= 11 + + Among others, the following features and patches have been integrated: + + * NeverZero patch for afl-gcc, llvm_mode, qemu_mode and unicorn_mode which prevents a wrapping map value to zero, increases coverage + + * Persistent mode and deferred forkserver for qemu_mode + + * Unicorn mode which allows fuzzing of binaries from completely different platforms (integration provided by domenukk) + + * The new CmpLog instrumentation for LLVM and QEMU inspired by [Redqueen](https://www.syssec.ruhr-uni-bochum.de/media/emma/veroeffentlichungen/2018/12/17/NDSS19-Redqueen.pdf) + + * Win32 PE binary-only fuzzing with QEMU and Wine + + * AFLfast's power schedules by Marcel Böhme: [https://github.com/mboehme/aflfast](https://github.com/mboehme/aflfast) + + * The MOpt mutator: [https://github.com/puppet-meteor/MOpt-AFL](https://github.com/puppet-meteor/MOpt-AFL) + + * LLVM mode Ngram coverage by Adrian Herrera [https://github.com/adrianherrera/afl-ngram-pass](https://github.com/adrianherrera/afl-ngram-pass) + + * InsTrim, an effective CFG llvm_mode instrumentation implementation for large targets: [https://github.com/csienslab/instrim](https://github.com/csienslab/instrim) + + * C. Holler's afl-fuzz Python mutator module and llvm_mode instrument file support: [https://github.com/choller/afl](https://github.com/choller/afl) + + * Custom mutator by a library (instead of Python) by kyakdan + + * LAF-Intel/CompCov support for llvm_mode, qemu_mode and unicorn_mode (with enhanced capabilities) + + * Radamsa and hongfuzz mutators (as custom mutators). + + * QBDI mode to fuzz android native libraries via QBDI framework + + A more thorough list is available in the [PATCHES](docs/PATCHES.md) file. So all in all this is the best-of afl that is currently out there :-) @@ -115,7 +125,7 @@ For releases, please see the [Releases](https://github.com/AFLplusplus/AFLplusplus/releases) tab. -## Google Summer of Code 2020 (and any other students and enthusiast developers) +## Help wanted We are happy to be part of [Google Summer of Code 2020](https://summerofcode.withgoogle.com/organizations/5100744400699392/)! :-) @@ -140,7 +150,7 @@ hence afl-clang-lto is available!) or just pull directly from the docker hub: docker pull aflplusplus/aflplusplus docker run -ti -v /location/of/your/target:/src aflplusplus/aflplusplus ``` -This container is automatically generated when a push to master happens. +This image is automatically generated when a push to master happens. You will find your target source code in /src in the container. If you want to build afl++ yourself you have many options. @@ -151,7 +161,7 @@ sudo apt install build-essential libtool-bin python3-dev automake flex bison lib make distrib sudo make install ``` -It is recommended to install the newest available gcc and clang and llvm-dev +It is recommended to install the newest available gcc, clang and llvm-dev possible in your distribution! Note that "make distrib" also builds llvm_mode, qemu_mode, unicorn_mode and @@ -197,6 +207,444 @@ These build options exist: e.g.: make ASAN_BUILD=1 +## Good examples and writeups + +Here are some good writeups to show how to effectively use AFL++: + + * [https://aflplus.plus/docs/tutorials/libxml2_tutorial/](https://aflplus.plus/docs/tutorials/libxml2_tutorial/) + * [https://bananamafia.dev/post/gb-fuzz/](https://bananamafia.dev/post/gb-fuzz/) + * [https://securitylab.github.com/research/fuzzing-challenges-solutions-1](https://securitylab.github.com/research/fuzzing-challenges-solutions-1) + * [https://securitylab.github.com/research/fuzzing-sockets-FTP](https://securitylab.github.com/research/fuzzing-sockets-FTP) + +If you are interested in fuzzing structured data (where you define what the +structure is), these links have you covered: + * Superion for afl++: [https://github.com/adrian-rt/superion-mutator](https://github.com/adrian-rt/superion-mutator) + * libprotobuf raw: [https://github.com/bruce30262/libprotobuf-mutator_fuzzing_learning/tree/master/4_libprotobuf_aflpp_custom_mutator](https://github.com/bruce30262/libprotobuf-mutator_fuzzing_learning/tree/master/4_libprotobuf_aflpp_custom_mutator) + * libprotobuf for old afl++ API: [https://github.com/thebabush/afl-libprotobuf-mutator](https://github.com/thebabush/afl-libprotobuf-mutator) + +If you find other good ones, please send them to us :-) + +## How to fuzz with afl++ + +The following describes how to fuzz with a target if source code is available. +If you have a binary-only target please skip to [#Instrumenting binary-only apps](#Instrumenting binary-only apps) + +Fuzzing source code is a two step process. + +1. compile the target with a special compiler that prepares the target to be + fuzzed efficiently. This step is called "instrumenting a target". +2. Prepare the fuzzing by selecting and optimizing the input corpus for the + target. +3. perform the fuzzing of the target by randomly mutating input and assessing + if a generated input was processed in a new path in the target binary + +### 1. Instrumenting that target + +#### a) Selecting the best afl++ compiler for instrumenting the target + +afl++ comes with different compilers and instrumentation options. +The following evaluation flow will help you to select the best possible. + +It is highly recommended to have the newest llvm version possible installed, +anything below 9 is not recommended. + +``` ++--------------------------------+ +| clang/clang++ 11+ is available | --> use afl-clang-lto and afl-clang-lto++ ++--------------------------------+ see [llvm/README.lto.md](llvm/README.lto.md) + | + | if not, or if the target fails with with afl-clang-lto/++ + | + v ++---------------------------------+ +| clang/clang++ 3.3+ is available | --> use afl-clang-fast and afl-clang-fast++ ++---------------------------------+ see [llvm/README.md](llvm/README.md) + | + | if not, or if the target fails with afl-clang-fast/++ + | + v + +--------------------------------+ + | if you want to instrument only | -> use afl-gcc-fast and afl-gcc-fast++ + | parts of the target | see [gcc_plugin/README.md](gcc_plugin/README.md) and + +--------------------------------+ [gcc_plugin/README.instrument_file.md](gcc_plugin/README.instrument_file.md) + | + | if not, or if you do not have a gcc with plugin support + | + v + use afl-gcc and afl-g++ +``` + +#### b) Selecting instrumentation options + +The following options are available when you instrument with afl-clang-fast or +afl-clang-lto: + + * Splitting integer, string, float and switch compares so afl++ can easier + solve these. This is an important option if you do not have a very good + good and large input corpus. This technique is called laf-intel or COMPCOV. + To use this set the following environment variable before compiling the + target: `export AFL_LLVM_LAF_ALL=1` + You can read more about this in [llvm/README.laf-intel.md](llvm/README.laf-intel.md) + * A different technique is to instrument the target so that any compare values + in the target are sent to afl++ which then tries to put this value into the + fuzzing data at different locations. This technique is very fast and good - + if the target does not transform input data before comparison. Therefore + technique is called `input to state` or `redqueen`. + If you want to use this technique, then you have to compile the target + twice, once specifically with/for this mode. + You can read more about this in [llvm_mode/README.cmplog.md](llvm_mode/README.cmplog.md) + +If you use afl-clang-fast, afl-clang-lto or afl-gcc-fast you have the option to +selectivly only instrument parts of the target that you are interested in: + + * To instrument only those parts of the target that you are interested in + create a file with all the filenames of the source code that should be + instrumented. + For afl-clang-lto and afl-gcc-fast - or afl-clang-fast if either the clang + version is < 7 or the CLASSIC instrumentation is used - just put one + filename per line, no directory information necessary, and set + `export AFL_LLVM_INSTRUMENT_FILE=yourfile.txt` + see [llvm_mode/README.instrument_file.md](llvm_mode/README.instrument_file.md) + For afl-clang-fast > 6.0 or if PCGUARD instrumentation is used then use the + llvm sancov allow-list feature: [http://clang.llvm.org/docs/SanitizerCoverage.html](http://clang.llvm.org/docs/SanitizerCoverage.html) + +There are many more options and modes available however these are most of the +time less effective. See: + * [llvm_mode/README.ctx.md](llvm_mode/README.ctx.md) + * [llvm_mode/README.ngram.md](llvm_mode/README.ngram.md) + * [llvm_mode/README.instrim.md](llvm_mode/README.instrim.md) + * [llvm_mode/README.neverzero.md](llvm_mode/README.neverzero.md) + +#### c) Modify the target + +If the target has features that makes fuzzing more difficult, e.g. +checksums, HMAC etc. then modify the source code so that this is +removed. +This can even be done for productional source code be eliminating +these checks within this specific defines: + +``` +#ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION + // say that the checksum or HMAC was fine - or whatever is required + // to eliminate the need for the fuzzer to guess the right checksum + return 0; +#endif +``` + +#### d) Instrument the target + +In this step the target source code is compiled so that it can be fuzzed. + +Basically you have to tell the target build system that the selected afl++ +compiler is used. Also - if possible - you should always configure the +build system that the target is compiled statically and not dynamically. +How to do this is described below. + +Then build the target. (Usually with `make`) + +##### configure + +For `configure` build systems this is usually done by: +`CC=afl-clang-fast CXX=afl-clang-fast++ ./configure --disable-shared` + +Note that if you using the (better) afl-clang-lto compiler you also have to +AR to llvm-ar[-VERSION] and RANLIB to llvm-ranlib[-VERSION] - as it is +described in [llvm/README.lto.md](llvm/README.lto.md) + +##### cmake + +For `configure` build systems this is usually done by: +`mkdir build; cd build; CC=afl-clang-fast CXX=afl-clang-fast++ cmake ..` + +Note that if you using the (better) afl-clang-lto compiler you also have to +AR to llvm-ar[-VERSION] and RANLIB to llvm-ranlib[-VERSION] - as it is +described in [llvm/README.lto.md](llvm/README.lto.md) + +##### other build systems or if configure/cmake didn't work + +Sometimes cmake and configure do not pick up the afl compiler, or the ranlib/ar +that is needed - because this was just not foreseen by the developer of the +target. Or they have non-standard options. Figure out if there is a +non-standard way to set this, otherwise set the build normally and edit the +generated build environment afterwards by hand to point to the right compiler +(and/or ranlib and ar). + +#### d) Better instrumentation + +If you just fuzz a target program as-is you are wasting a great opportunity for +much more fuzzing speed. + +This requires the usage of afl-clang-lto or afl-clang-fast + +This is the so-called `persistent mode`, which is much, much faster but +requires that you code a source file that is specifically calling the target +functions that you want to fuzz, plus a few specific afl++ functions around +it. See [llvm_mode/README.persistent_mode.md](llvm_mode/README.persistent_mode.md) for details. + +Basically if you do not fuzz a target in persistent mode then you are just +doing it for a hobby and not professionally :-) + +### 2. Preparing the fuzzing + +As you fuzz the target with mutated input, having as diverse inputs for the +target as possible improves the efficiency a lot. + +#### a) Collect inputs +Try to gather valid inputs for the target from wherever you can. E.g. if it +the PNG picture format try to find as many png files as possible, e.g. from +reported bugs, test suites, random downloads from the internet, unit test +case data - from all kind of PNG software. + +If the input is not known files, you can also modify a target program to write +away normal data it receives and processes to a file and use these. + +#### b) Making the input corpus unique + +Use the afl++ tool `afl-cmin` to remove inputs from the corpus that do not +use a different paths in the target. +Put all files from step a) into one directory, e.g. INPUTS. + +Put all the files from step a) + +If the target program is to be called by fuzzing as `bin/target -d INPUTFILE` +the run afl-cmin like this: +`afl-cmin -i INPUTS -o INPUTS_UNIQUE -- bin/target -d @@` +Note that the INPUTFILE that the target program would read has to be set as `@@`. + +If the target reads from stdin instead, just omit the `@@` as this is the +default. + +#### b) Minimizing all corpus files + +The shorter the input files are so that they still traverse the same path +within the target, the better the fuzzing will be. This is done with `afl-tmin` +however it is a long processes as this has to be done for every file: + +``` +mkdir input +cd INPUTS_UNIQUE +for i in *; do + afl-tmin -i "$i" -o "../input/$i" -- bin/target -d @@ +done +``` + +This can also be parallelized, e.g. with `parallel` + +#### c) done! + +The INPUTS_UNIQUE/ directory from step a) - or even better if you minimized the +corpus in step b) then the files in input/ is then the input corpus directory +to be used in fuzzing! :-) + +### Fuzzing the target + +In this final step we fuzz the target. +There are not that many useful options to run the target - unless you want to +use many CPU cores for the fuzzing, which will make the fuzzing much more useful. + +If you just use one CPU for fuzzing, then you are fuzzing just for fun and not +seriously :-) + +#### a) running afl-fuzz + +Before to do even a test run of afl-fuzz execute `sudo afl-system-config` (on +the host if you execute afl-fuzz in a docker container). This reconfigured the +system for optimal speed - which afl-fuzz checks and bails otherwise. +Set `export AFL_SKIP_CPUFREQ=1` for afl-fuzz to skip this if you cannot run +afl-system-config with root privileges on the host for whatever reason. + +If you have an input corpus from step 2 then specify this directory with the `-i` +option. Otherwise create a new directory and create a file with any content +in there. + +If you do not want anything special, the defaults are already the usual best, +hence all you need (from the example in 2a): +`afl-fuzz -i input -o output -- bin/target -d @@` +Note that the directory specified with -o will be created if it does not exist. + +If you need to stop and re-start the fuzzing, use the same command line option +and switch the input directory with a dash (`-`): +`afl-fuzz -i - -o output -- bin/target -d @@` + +Adding a dictionary helpful. See the [dictionaries/](dictionaries/) if +something is already included for your data format, and tell afl-fuzz to load +that dictionary by adding `-x dicationaries/FORMAT.dict`. With afl-clang-lto +you have an autodictionary generation for which you need to do nothing except +to use afl-clang-lto as the compiler. You also have the option to generate +a dictionary yourself, see [libtokencap/README.md](libtokencap/README.md) + +afl-fuzz never stops fuzzing. To terminate afl++ simply press Control-C. + +When you start afl-fuzz you will see a user interface that shows what the status +is: +![docs/screenshot.png](docs/screenshot.png) +All the entries are explained in [docs/status_screen.md](docs/status_screen.md) + +#### b) Using multiple cores + +If you want to seriously fuzz then use as many cores as possible to fuzz your +target. + +On the same machine - due to the nature how afl++ works - there is a maximum +number of CPU cores that are useful, more and the overall performance degrades +instead. This value depends on the target and the limit is between 24 and 64 +cores per machine. + +There should be one main fuzzer (`-M main` option) and as many secondary +fuzzers (eg `-S variant1`) as you cores that you use. +Every -M/-S entry needs a unique name (that can be whatever), however the same +-o output directory location has to be used for all. + +For every secondary there should be a variation, e.g.: + * one should fuzz the target that was compiled differently: with sanitizers + activated (`export AFL_USE_ASAN=1 ; export AFL_USE_UBSAN=1 ; + export AFL_USE_CFISAN=1 ; ` + * one should fuzz the target with CMPLOG/redqueen (see above) + * At 1-2 should fuzz a target compiled with laf-intel/COMPCOV (see above). + +All other secondaries should be: + * 1/2 with MOpt option enabled: `-L 0` + * run with a different power schedule, available are: + `explore (default), fast, coe, lin, quad, exploit, mmopt, rare, seek` + which you can set with e.g. `-p seek` + +You can also use different fuzzers. +If you are afl-spinoffs or afl conforming, then just use the same -o directory +and give it a unique `-S` name. +Examples are e.g.: + * [Angora](https://github.com/AngoraFuzzer/Angora) + * [Untracer](https://github.com/FoRTE-Research/UnTracer-AFL) + * [AFLsmart](https://github.com/aflsmart/aflsmart) + * [FairFuzz](https://github.com/carolemieux/afl-rb) + * [Neuzz](https://github.com/Dongdongshe/neuzz) +A long list can be found at [https://github.com/Microsvuln/Awesome-AFL](https://github.com/Microsvuln/Awesome-AFL) + +However you can also sync afl++ with honggfuzz, libfuzzer, entropic, etc. +Just show the main fuzzer (-M) with the `-F` option where the queue +directory of these other fuzzers are, e.g. `-F /src/target/honggfuzz` + +#### c) The status of the fuzz campaign + +afl++ comes with the `afl-whatsup` script to show the status of fuzzing +campaign. + +Just supply the directory that afl-fuzz is given with the -o option and +you will see a detailed status of every fuzzer in that campaign plus +a summary. + +To have only the summary use the `-s` switch e.g.: `afl-whatsup -s output/` + +#### d) Checking the coverage of the fuzzing + +The `paths found` value is a bad indicator how good the coverage is. +It is better to check out the exact lines of code that have been reached - +and which have not been found so far. + +An "easy" helper script for this is [afl-cov](https://github.com/vanhauser-thc/afl-cov), +just follow the README of that seperate project. + +If you see that an important area or a feature has not been covered so far then +try to find an input that is able to reach that and start a new secondary in +that fuzzing campaign with that seed as input, let it run for a few minutes, +then terminate it. The main node will pick it up and make it available to the +other secondary nodes over time. Set `export AFL_NO_AFFINITY=1` if you have no +free core. + +#### e) How long to fuzz a target? + +This is a difficult question. +Basically if no new path is found for a long time (e.g. for a day or a week) +then you can expect that your fuzzing won't be fruitful anymore. +However often this just means that you should switch out secondaries for +others, e.g. custom mutator modules, sync to very different fuzzers, etc. + +### The End + +This is basically all you need to know to professionally run fuzzing campaigns. +If you want to know more, the rest of this README and the tons of texts in +[docs/](docs/) will have you covered. + +Note that there are also a lot of tools out there that help fuzzing with afl++ +(some might be deprecated or unsupported): + +Minimization of test cases: + * [afl-pytmin](https://github.com/ilsani/afl-pytmin) - a wrapper for afl-tmin that tries to speed up the process of the minimization of test case by using many CPU cores. + * [afl-ddmin-mod](https://github.com/MarkusTeufelberger/afl-ddmin-mod) - a variation of afl-tmin based on the ddmin algorithm. + * [halfempty](https://github.com/googleprojectzero/halfempty) - is a fast utility for minimizing test cases by Tavis Ormandy based on parallelization. + +Distributed execution: + * [disfuzz-afl](https://github.com/MartijnB/disfuzz-afl) - distributed fuzzing for AFL. + * [AFLDFF](https://github.com/quantumvm/AFLDFF) - AFL distributed fuzzing framework. + * [afl-launch](https://github.com/bnagy/afl-launch) - a tool for the execution of many AFL instances. + * [afl-mothership](https://github.com/afl-mothership/afl-mothership) - management and execution of many synchronized AFL fuzzers on AWS cloud. + * [afl-in-the-cloud](https://github.com/abhisek/afl-in-the-cloud) - another script for running AFL in AWS. + +Deployment, management, monitoring, reporting + * [afl-other-arch](https://github.com/shellphish/afl-other-arch) - is a set of patches and scripts for easily adding support for various non-x86 architectures for AFL. + * [afl-trivia](https://github.com/bnagy/afl-trivia) - a few small scripts to simplify the management of AFL. + * [afl-monitor](https://github.com/reflare/afl-monitor) - a script for monitoring AFL. + * [afl-manager](https://github.com/zx1340/afl-manager) - a web server on Python for managing multi-afl. + * [afl-remote](https://github.com/block8437/afl-remote) - a web server for the remote management of AFL instances. + +Crash processing + * [afl-utils](https://gitlab.com/rc0r/afl-utils) - a set of utilities for automatic processing/analysis of crashes and reducing the number of test cases. + * [afl-crash-analyzer](https://github.com/floyd-fuh/afl-crash-analyzer) - another crash analyzer for AFL. + * [fuzzer-utils](https://github.com/ThePatrickStar/fuzzer-utils) - a set of scripts for the analysis of results. + * [atriage](https://github.com/Ayrx/atriage) - a simple triage tool. + * [afl-kit](https://github.com/kcwu/afl-kit) - afl-cmin on Python. + * [AFLize](https://github.com/d33tah/aflize) - a tool that automatically generates builds of debian packages suitable for AFL. + * [afl-fid](https://github.com/FoRTE-Research/afl-fid) - a set of tools for working with input data. + +## Fuzzing binary-only targets + +When source code is *NOT* available, afl++ offers various support for fast, +on-the-fly instrumentation of black-box binaries. + +### QEMU + +For linux programs and it's libraries this is accomplished with a version of +QEMU running in the lesser-known "user space emulation" mode. +QEMU is a project separate from AFL, but you can conveniently build the +feature by doing: +```shell +cd qemu_mode +./build_qemu_support.sh +``` +For additional instructions and caveats, see [qemu_mode/README.md](qemu_mode/README.md). +If possible you should use the persistent mode, see [qemu_mode/README.persistent.md](qemu_mode/README.persistent.md). +The mode is approximately 2-5x slower than compile-time instrumentation, and is +less conducive to parallelization. + +If [afl-dyninst](https://github.com/vanhauser-thc/afl-dyninst) works for +your binary, then you can use afl-fuzz normally and it will have twice +the speed compared to qemu_mode (but slower than persistent mode). + +### Unicorn + +For non-Linux binaries you can use afl++'s unicorn mode which can emulate +anything you want - for the price of speed and the user writing scripts. +See [unicorn_mode](unicorn_mode/README.md). + +It can be easily build by: +```shell +cd unicorn_mode +./build_unicorn_support.sh +``` + +### Shared libraries + +If the goal is to fuzz a dynamic library then there are two options available. +For both you need to write a small hardness that loads and calls the library. +Faster is the frida solution: [examples/afl_frida/README.md](examples/afl_frida/README.md) + +Another, less precise and slower option is using ptrace with debugger interrupt +instrumentation: [examples/afl_untracer/README.md](examples/afl_untracer/README.md) + +### More + +A more comprehensive description of these and other options can be found in +[docs/binaryonly_fuzzing.md](docs/binaryonly_fuzzing.md) + ## Challenges of guided fuzzing Fuzzing is one of the most powerful and proven strategies for identifying @@ -262,7 +710,6 @@ closed-source tools. The fuzzer is thoroughly tested to deliver out-of-the-box performance far superior to blind fuzzing or coverage-only tools. - ## Instrumenting programs for use with AFL PLEASE NOTE: llvm_mode compilation with afl-clang-fast/afl-clang-fast++ @@ -318,52 +765,6 @@ simple memory bugs. Libdislocator, a helper library included with AFL (see PS. ASAN users are advised to review [docs/notes_for_asan.md](docs/notes_for_asan.md) file for important caveats. - -## Instrumenting binary-only apps - -When source code is *NOT* available, the fuzzer offers experimental support for -fast, on-the-fly instrumentation of black-box binaries. This is accomplished -with a version of QEMU running in the lesser-known "user space emulation" mode. - -QEMU is a project separate from AFL, but you can conveniently build the -feature by doing: - -```shell -cd qemu_mode -./build_qemu_support.sh -``` - -For additional instructions and caveats, see [qemu_mode/README.md](qemu_mode/README.md). - -If possible you should use the persistent mode, see [qemu_mode/README.persistent.md](qemu_mode/README.persistent.md). - -The mode is approximately 2-5x slower than compile-time instrumentation, is -less conducive to parallelization, and may have some other quirks. - -If [afl-dyninst](https://github.com/vanhauser-thc/afl-dyninst) works for -your binary, then you can use afl-fuzz normally and it will have twice -the speed compared to qemu_mode. - -A more comprehensive description of these and other options can be found in -[docs/binaryonly_fuzzing.md](docs/binaryonly_fuzzing.md) - -## Good examples and writeups - -Here are some good writeups to show how to effectively use AFL++: - - * [https://aflplus.plus/docs/tutorials/libxml2_tutorial/](https://aflplus.plus/docs/tutorials/libxml2_tutorial/) - * [https://bananamafia.dev/post/gb-fuzz/](https://bananamafia.dev/post/gb-fuzz/) - * [https://securitylab.github.com/research/fuzzing-challenges-solutions-1](https://securitylab.github.com/research/fuzzing-challenges-solutions-1) - * [https://securitylab.github.com/research/fuzzing-sockets-FTP](https://securitylab.github.com/research/fuzzing-sockets-FTP) - -If you are interested in fuzzing structured data (where you define what the -structure is), these links have you covered: - * Superion for afl++: [https://github.com/adrian-rt/superion-mutator](https://github.com/adrian-rt/superion-mutator) - * libprotobuf raw: [https://github.com/bruce30262/libprotobuf-mutator_fuzzing_learning/tree/master/4_libprotobuf_aflpp_custom_mutator](https://github.com/bruce30262/libprotobuf-mutator_fuzzing_learning/tree/master/4_libprotobuf_aflpp_custom_mutator) - * libprotobuf for old afl++ API: [https://github.com/thebabush/afl-libprotobuf-mutator](https://github.com/thebabush/afl-libprotobuf-mutator) - -If you find other good ones, please send them to us :-) - ## Power schedules The power schedules were copied from Marcel Böhme's AFLfast implementation and diff --git a/README_new.md b/README_new.md deleted file mode 100644 index 9b8c1014..00000000 --- a/README_new.md +++ /dev/null @@ -1,1065 +0,0 @@ -# american fuzzy lop plus plus (afl++) - - AFL++ Logo - - ![Travis State](https://api.travis-ci.com/AFLplusplus/AFLplusplus.svg?branch=stable) - - Release Version: [2.66c](https://github.com/AFLplusplus/AFLplusplus/releases) - - Github Version: 2.66d - - Repository: [https://github.com/AFLplusplus/AFLplusplus](https://github.com/AFLplusplus/AFLplusplus) - - afl++ is maintained by: - - * Marc "van Hauser" Heuse , - * Heiko "hexcoder-" Eißfeldt , - * Andrea Fioraldi and - * Dominik Maier . - - Originally developed by Michal "lcamtuf" Zalewski. - - afl++ is superiour to Google's afl in any way - more speed, more and better - mutations, more and better instrumentation, etc. etc. - -## Contents - - 1. [Features](#important-features-of-afl) - 2. [How to compile and install afl++](#building-and-installing-afl) - 3. [How to fuzz a target](#how-to-fuzz-with-afl) - 4. [Fuzzing binary-only targets](#fuzzing-binary-only-targets) - 5. [Good examples and writeups of afl++ usages](#good-examples-and-writeups) - 6. [Branches](#branches) - 7. [Want to help?](#help-wanted) - 8. [Detailed help and description of afl++](#challenges-of-guided-fuzzing) - -## Important features of afl++ - - afl++ supports llvm up to version 12, very fast binary fuzzing with QEMU 3.1 - with laf-intel and redqueen, unicorn mode, gcc plugin, full *BSD, Solaris and - Android support and much, much, much more. - - | Feature/Instrumentation | afl-gcc | llvm_mode | gcc_plugin | qemu_mode | unicorn_mode | - | ----------------------- |:-------:|:---------:|:----------:|:----------------:|:------------:| - | NeverZero | x | x(1) | (2) | x | x | - | Persistent mode | | x | x | x86[_64]/arm[64] | x | - | LAF-Intel / CompCov | | x | | x86[_64]/arm[64] | x86[_64]/arm | - | CmpLog | | x | | x86[_64]/arm[64] | | - | Instrument file list | | x | x | (x)(3) | | - | Non-colliding coverage | | x(4) | | (x)(5) | | - | InsTrim | | x | | | | - | Ngram prev_loc coverage | | x(6) | | | | - | Context coverage | | x | | | | - | Auto dictionary | | x(7) | | | | - | Snapshot LKM support | | x | | (x)(5) | | - - neverZero: - - (1) default for LLVM >= 9.0, env var for older version due an efficiency bug in llvm <= 8 - - (2) GCC creates non-performant code, hence it is disabled in gcc_plugin - - (3) partially via AFL_CODE_START/AFL_CODE_END - - (4) with pcguard mode and LTO mode for LLVM >= 11 - - (5) upcoming, development in the branch - - (6) not compatible with LTO instrumentation and needs at least LLVM >= 4.1 - - (7) only in LTO mode with LLVM >= 11 - - Among others, the following features and patches have been integrated: - - * NeverZero patch for afl-gcc, llvm_mode, qemu_mode and unicorn_mode which prevents a wrapping map value to zero, increases coverage - - * Persistent mode and deferred forkserver for qemu_mode - - * Unicorn mode which allows fuzzing of binaries from completely different platforms (integration provided by domenukk) - - * The new CmpLog instrumentation for LLVM and QEMU inspired by [Redqueen](https://www.syssec.ruhr-uni-bochum.de/media/emma/veroeffentlichungen/2018/12/17/NDSS19-Redqueen.pdf) - - * Win32 PE binary-only fuzzing with QEMU and Wine - - * AFLfast's power schedules by Marcel Böhme: [https://github.com/mboehme/aflfast](https://github.com/mboehme/aflfast) - - * The MOpt mutator: [https://github.com/puppet-meteor/MOpt-AFL](https://github.com/puppet-meteor/MOpt-AFL) - - * LLVM mode Ngram coverage by Adrian Herrera [https://github.com/adrianherrera/afl-ngram-pass](https://github.com/adrianherrera/afl-ngram-pass) - - * InsTrim, an effective CFG llvm_mode instrumentation implementation for large targets: [https://github.com/csienslab/instrim](https://github.com/csienslab/instrim) - - * C. Holler's afl-fuzz Python mutator module and llvm_mode instrument file support: [https://github.com/choller/afl](https://github.com/choller/afl) - - * Custom mutator by a library (instead of Python) by kyakdan - - * LAF-Intel/CompCov support for llvm_mode, qemu_mode and unicorn_mode (with enhanced capabilities) - - * Radamsa and hongfuzz mutators (as custom mutators). - - * QBDI mode to fuzz android native libraries via QBDI framework - - A more thorough list is available in the [PATCHES](docs/PATCHES.md) file. - - So all in all this is the best-of afl that is currently out there :-) - - For new versions and additional information, check out: - [https://github.com/AFLplusplus/AFLplusplus](https://github.com/AFLplusplus/AFLplusplus) - - To compare notes with other users or get notified about major new features, - send a mail to . - - See [docs/QuickStartGuide.md](docs/QuickStartGuide.md) if you don't have time to - read this file. - -## Branches - - The following branches exist: - - * [master/trunk](https://github.com/AFLplusplus/AFLplusplus/) : stable state of afl++ - it is synced from dev from time to - time when we are satisfied with it's stability - * [dev](https://github.com/AFLplusplus/AFLplusplus/tree/dev) : development state of afl++ - bleeding edge and you might catch a - checkout which does not compile or has a bug. *We only accept PRs in dev!!* - * (any other) : experimental branches to work on specific features or testing - new functionality or changes. - - For releases, please see the [Releases](https://github.com/AFLplusplus/AFLplusplus/releases) tab. - -## Help wanted - -We are happy to be part of [Google Summer of Code 2020](https://summerofcode.withgoogle.com/organizations/5100744400699392/)! :-) - -We have several ideas we would like to see in AFL++ to make it even better. -However, we already work on so many things that we do not have the time for -all the big ideas. - -This can be your way to support and contribute to AFL++ - extend it to -something cool. - -We have an idea list in [docs/ideas.md](docs/ideas.md). - -For everyone who wants to contribute (and send pull requests) please read -[CONTRIBUTING.md](CONTRIBUTING.md) before your submit. - -## Building and installing afl++ - -An easy way to install afl++ with everything compiled is available via docker: -You can use the [Dockerfile](Dockerfile) (which has gcc-10 and clang-11 - -hence afl-clang-lto is available!) or just pull directly from the docker hub: -```shell -docker pull aflplusplus/aflplusplus -docker run -ti -v /location/of/your/target:/src aflplusplus/aflplusplus -``` -This image is automatically generated when a push to master happens. -You will find your target source code in /src in the container. - -If you want to build afl++ yourself you have many options. -The easiest is to build and install everything: - -```shell -sudo apt install build-essential libtool-bin python3-dev automake flex bison libglib2.0-dev libpixman-1-dev clang python3-setuptools llvm -make distrib -sudo make install -``` -It is recommended to install the newest available gcc, clang and llvm-dev -possible in your distribution! - -Note that "make distrib" also builds llvm_mode, qemu_mode, unicorn_mode and -more. If you just want plain afl then do "make all", however compiling and -using at least llvm_mode is highly recommended for much better results - -hence in this case - -```shell -make source-only -``` -is what you should choose. - -These build targets exist: - -* all: just the main afl++ binaries -* binary-only: everything for binary-only fuzzing: qemu_mode, unicorn_mode, libdislocator, libtokencap -* source-only: everything for source code fuzzing: llvm_mode, libdislocator, libtokencap -* distrib: everything (for both binary-only and source code fuzzing) -* man: creates simple man pages from the help option of the programs -* install: installs everything you have compiled with the build options above -* clean: cleans everything compiled, not downloads (unless not on a checkout) -* deepclean: cleans everything including downloads -* code-format: format the code, do this before you commit and send a PR please! -* tests: runs test cases to ensure that all features are still working as they should -* unit: perform unit tests (based on cmocka) -* help: shows these build options - -[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 -``` - -These build options exist: - -* STATIC - compile AFL++ static -* ASAN_BUILD - compiles with memory sanitizer for debug purposes -* PROFILING - compile with profiling information (gprof) -* NO_PYTHON - disable python support -* AFL_NO_X86 - if compiling on non-intel/amd platforms -* LLVM_CONFIG - if your distro doesn't use the standard name for llvm-config (e.g. Debian) - -e.g.: make ASAN_BUILD=1 - -## Good examples and writeups - -Here are some good writeups to show how to effectively use AFL++: - - * [https://aflplus.plus/docs/tutorials/libxml2_tutorial/](https://aflplus.plus/docs/tutorials/libxml2_tutorial/) - * [https://bananamafia.dev/post/gb-fuzz/](https://bananamafia.dev/post/gb-fuzz/) - * [https://securitylab.github.com/research/fuzzing-challenges-solutions-1](https://securitylab.github.com/research/fuzzing-challenges-solutions-1) - * [https://securitylab.github.com/research/fuzzing-sockets-FTP](https://securitylab.github.com/research/fuzzing-sockets-FTP) - -If you are interested in fuzzing structured data (where you define what the -structure is), these links have you covered: - * Superion for afl++: [https://github.com/adrian-rt/superion-mutator](https://github.com/adrian-rt/superion-mutator) - * libprotobuf raw: [https://github.com/bruce30262/libprotobuf-mutator_fuzzing_learning/tree/master/4_libprotobuf_aflpp_custom_mutator](https://github.com/bruce30262/libprotobuf-mutator_fuzzing_learning/tree/master/4_libprotobuf_aflpp_custom_mutator) - * libprotobuf for old afl++ API: [https://github.com/thebabush/afl-libprotobuf-mutator](https://github.com/thebabush/afl-libprotobuf-mutator) - -If you find other good ones, please send them to us :-) - -## How to fuzz with afl++ - -The following describes how to fuzz with a target if source code is available. -If you have a binary-only target please skip to [#Instrumenting binary-only apps](#Instrumenting binary-only apps) - -Fuzzing source code is a two step process. - -1. compile the target with a special compiler that prepares the target to be - fuzzed efficiently. This step is called "instrumenting a target". -2. Prepare the fuzzing by selecting and optimizing the input corpus for the - target. -3. perform the fuzzing of the target by randomly mutating input and assessing - if a generated input was processed in a new path in the target binary - -### 1. Instrumenting that target - -#### a) Selecting the best afl++ compiler for instrumenting the target - -afl++ comes with different compilers and instrumentation options. -The following evaluation flow will help you to select the best possible. - -It is highly recommended to have the newest llvm version possible installed, -anything below 9 is not recommended. - -``` -+--------------------------------+ -| clang/clang++ 11+ is available | --> use afl-clang-lto and afl-clang-lto++ -+--------------------------------+ see [llvm/README.lto.md](llvm/README.lto.md) - | - | if not, or if the target fails with with afl-clang-lto/++ - | - v -+---------------------------------+ -| clang/clang++ 3.3+ is available | --> use afl-clang-fast and afl-clang-fast++ -+---------------------------------+ see [llvm/README.md](llvm/README.md) - | - | if not, or if the target fails with afl-clang-fast/++ - | - v - +--------------------------------+ - | if you want to instrument only | -> use afl-gcc-fast and afl-gcc-fast++ - | parts of the target | see [gcc_plugin/README.md](gcc_plugin/README.md) and - +--------------------------------+ [gcc_plugin/README.instrument_file.md](gcc_plugin/README.instrument_file.md) - | - | if not, or if you do not have a gcc with plugin support - | - v - use afl-gcc and afl-g++ -``` - -#### b) Selecting instrumentation options - -The following options are available when you instrument with afl-clang-fast or -afl-clang-lto: - - * Splitting integer, string, float and switch compares so afl++ can easier - solve these. This is an important option if you do not have a very good - good and large input corpus. This technique is called laf-intel or COMPCOV. - To use this set the following environment variable before compiling the - target: `export AFL_LLVM_LAF_ALL=1` - You can read more about this in [llvm/README.laf-intel.md](llvm/README.laf-intel.md) - * A different technique is to instrument the target so that any compare values - in the target are sent to afl++ which then tries to put this value into the - fuzzing data at different locations. This technique is very fast and good - - if the target does not transform input data before comparison. Therefore - technique is called `input to state` or `redqueen`. - If you want to use this technique, then you have to compile the target - twice, once specifically with/for this mode. - You can read more about this in [llvm_mode/README.cmplog.md](llvm_mode/README.cmplog.md) - -If you use afl-clang-fast, afl-clang-lto or afl-gcc-fast you have the option to -selectivly only instrument parts of the target that you are interested in: - - * To instrument only those parts of the target that you are interested in - create a file with all the filenames of the source code that should be - instrumented. - For afl-clang-lto and afl-gcc-fast - or afl-clang-fast if either the clang - version is < 7 or the CLASSIC instrumentation is used - just put one - filename per line, no directory information necessary, and set - `export AFL_LLVM_INSTRUMENT_FILE=yourfile.txt` - see [llvm_mode/README.instrument_file.md](llvm_mode/README.instrument_file.md) - For afl-clang-fast > 6.0 or if PCGUARD instrumentation is used then use the - llvm sancov allow-list feature: [http://clang.llvm.org/docs/SanitizerCoverage.html](http://clang.llvm.org/docs/SanitizerCoverage.html) - -There are many more options and modes available however these are most of the -time less effective. See: - * [llvm_mode/README.ctx.md](llvm_mode/README.ctx.md) - * [llvm_mode/README.ngram.md](llvm_mode/README.ngram.md) - * [llvm_mode/README.instrim.md](llvm_mode/README.instrim.md) - * [llvm_mode/README.neverzero.md](llvm_mode/README.neverzero.md) - -#### c) Modify the target - -If the target has features that makes fuzzing more difficult, e.g. -checksums, HMAC etc. then modify the source code so that this is -removed. -This can even be done for productional source code be eliminating -these checks within this specific defines: - -``` -#ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION - // say that the checksum or HMAC was fine - or whatever is required - // to eliminate the need for the fuzzer to guess the right checksum - return 0; -#endif -``` - -#### d) Instrument the target - -In this step the target source code is compiled so that it can be fuzzed. - -Basically you have to tell the target build system that the selected afl++ -compiler is used. Also - if possible - you should always configure the -build system that the target is compiled statically and not dynamically. -How to do this is described below. - -Then build the target. (Usually with `make`) - -##### configure - -For `configure` build systems this is usually done by: -`CC=afl-clang-fast CXX=afl-clang-fast++ ./configure --disable-shared` - -Note that if you using the (better) afl-clang-lto compiler you also have to -AR to llvm-ar[-VERSION] and RANLIB to llvm-ranlib[-VERSION] - as it is -described in [llvm/README.lto.md](llvm/README.lto.md) - -##### cmake - -For `configure` build systems this is usually done by: -`mkdir build; cd build; CC=afl-clang-fast CXX=afl-clang-fast++ cmake ..` - -Note that if you using the (better) afl-clang-lto compiler you also have to -AR to llvm-ar[-VERSION] and RANLIB to llvm-ranlib[-VERSION] - as it is -described in [llvm/README.lto.md](llvm/README.lto.md) - -##### other build systems or if configure/cmake didn't work - -Sometimes cmake and configure do not pick up the afl compiler, or the ranlib/ar -that is needed - because this was just not foreseen by the developer of the -target. Or they have non-standard options. Figure out if there is a -non-standard way to set this, otherwise set the build normally and edit the -generated build environment afterwards by hand to point to the right compiler -(and/or ranlib and ar). - -#### d) Better instrumentation - -If you just fuzz a target program as-is you are wasting a great opportunity for -much more fuzzing speed. - -This requires the usage of afl-clang-lto or afl-clang-fast - -This is the so-called `persistent mode`, which is much, much faster but -requires that you code a source file that is specifically calling the target -functions that you want to fuzz, plus a few specific afl++ functions around -it. See [llvm_mode/README.persistent_mode.md](llvm_mode/README.persistent_mode.md) for details. - -Basically if you do not fuzz a target in persistent mode then you are just -doing it for a hobby and not professionally :-) - -### 2. Preparing the fuzzing - -As you fuzz the target with mutated input, having as diverse inputs for the -target as possible improves the efficiency a lot. - -#### a) Collect inputs -Try to gather valid inputs for the target from wherever you can. E.g. if it -the PNG picture format try to find as many png files as possible, e.g. from -reported bugs, test suites, random downloads from the internet, unit test -case data - from all kind of PNG software. - -If the input is not known files, you can also modify a target program to write -away normal data it receives and processes to a file and use these. - -#### b) Making the input corpus unique - -Use the afl++ tool `afl-cmin` to remove inputs from the corpus that do not -use a different paths in the target. -Put all files from step a) into one directory, e.g. INPUTS. - -Put all the files from step a) - -If the target program is to be called by fuzzing as `bin/target -d INPUTFILE` -the run afl-cmin like this: -`afl-cmin -i INPUTS -o INPUTS_UNIQUE -- bin/target -d @@` -Note that the INPUTFILE that the target program would read has to be set as `@@`. - -If the target reads from stdin instead, just omit the `@@` as this is the -default. - -#### b) Minimizing all corpus files - -The shorter the input files are so that they still traverse the same path -within the target, the better the fuzzing will be. This is done with `afl-tmin` -however it is a long processes as this has to be done for every file: - -``` -mkdir input -cd INPUTS_UNIQUE -for i in *; do - afl-tmin -i "$i" -o "../input/$i" -- bin/target -d @@ -done -``` - -This can also be parallelized, e.g. with `parallel` - -#### c) done! - -The INPUTS_UNIQUE/ directory from step a) - or even better if you minimized the -corpus in step b) then the files in input/ is then the input corpus directory -to be used in fuzzing! :-) - -### Fuzzing the target - -In this final step we fuzz the target. -There are not that many useful options to run the target - unless you want to -use many CPU cores for the fuzzing, which will make the fuzzing much more useful. - -If you just use one CPU for fuzzing, then you are fuzzing just for fun and not -seriously :-) - -#### a) running afl-fuzz - -If you have an input corpus from step 2 then specify this directory with the `-i` -option. Otherwise create a new directory and create a file with any content -in there. - -If you do not want anything special, the defaults are already the usual best, -hence all you need (from the example in 2a): -`afl-fuzz -i input -o output -- bin/target -d @@` -Note that the directory specified with -o will be created if it does not exist. - -If you need to stop and re-start the fuzzing, use the same command line option -and switch the input directory with a dash (`-`): -`afl-fuzz -i - -o output -- bin/target -d @@` - -afl-fuzz never stops fuzzing. To terminate afl++ simply press Control-C. - -When you start afl-fuzz you will see a user interface that shows what the status -is: -[docs/screenshot.png](docs/screenshot.png) -All the entries are explained in [docs/status_screen.md](docs/status_screen.md) - -#### b) Using multiple cores - -If you want to seriously fuzz then use as many cores as possible to fuzz your -target. - -On the same machine - due to the nature how afl++ works - there is a maximum -number of CPU cores that are useful, more and the overall performance degrades -instead. This value depends on the target and the limit is between 24 and 64 -cores per machine. - -There should be one main fuzzer (`-M main` option) and as many secondary -fuzzers (eg `-S variant1`) as you cores that you use. -Every -M/-S entry needs a unique name (that can be whatever), however the same --o output directory location has to be used for all. - -For every secondary there should be a variation, e.g.: - * one should fuzz the target that was compiled differently: with sanitizers - activated (`export AFL_USE_ASAN=1 ; export AFL_USE_UBSAN=1 ; - export AFL_USE_CFISAN=1 ; ` - * one should fuzz the target with CMPLOG/redqueen (see above) - * At 1-2 should fuzz a target compiled with laf-intel/COMPCOV (see above). - -All other secondaries should be: - * 1/2 with MOpt option enabled: `-L 0` - * run with a different power schedule, available are: - `explore (default), fast, coe, lin, quad, exploit, mmopt, rare, seek` - which you can set with e.g. `-p seek` - -You can also use different fuzzers. -If you are afl-spinoffs or afl conforming, then just use the same -o directory -and give it a unique `-S` name. -Examples are e.g.: - * - * - * - -However you can also sync afl++ with honggfuzz, libfuzzer, entropic, etc. -Just show the main fuzzer (-M) with the `-F` option where the queue -directory of these other fuzzers are, e.g. `-F /src/target/honggfuzz` - -#### c) How long to fuzz a target? - -This is a difficult question. -Basically if no new path is found for a long time (e.g. for a day or a week) -then you can expect that your fuzzing won't be fruitful anymore. -However often this just means that you should switch out secondaries for -others, e.g. custom mutator modules, sync to very different fuzzers, etc. - -### The End - -This is basically all you need to know to professionally run fuzzing campaigns. -If you want to know more, the rest of this README and the tons of texts in -[docs/](docs/) will have you covered. - -## Challenges of guided fuzzing - -Fuzzing is one of the most powerful and proven strategies for identifying -security issues in real-world software; it is responsible for the vast -majority of remote code execution and privilege escalation bugs found to date -in security-critical software. - -Unfortunately, fuzzing is also relatively shallow; blind, random mutations -make it very unlikely to reach certain code paths in the tested code, leaving -some vulnerabilities firmly outside the reach of this technique. - -There have been numerous attempts to solve this problem. One of the early -approaches - pioneered by Tavis Ormandy - is corpus distillation. The method -relies on coverage signals to select a subset of interesting seeds from a -massive, high-quality corpus of candidate files, and then fuzz them by -traditional means. The approach works exceptionally well but requires such -a corpus to be readily available. In addition, block coverage measurements -provide only a very simplistic understanding of the program state and are less -useful for guiding the fuzzing effort in the long haul. - -Other, more sophisticated research has focused on techniques such as program -flow analysis ("concolic execution"), symbolic execution, or static analysis. -All these methods are extremely promising in experimental settings, but tend -to suffer from reliability and performance problems in practical uses - and -currently do not offer a viable alternative to "dumb" fuzzing techniques. - - -## The afl-fuzz approach - -American Fuzzy Lop is a brute-force fuzzer coupled with an exceedingly simple -but rock-solid instrumentation-guided genetic algorithm. It uses a modified -form of edge coverage to effortlessly pick up subtle, local-scale changes to -program control flow. - -Simplifying a bit, the overall algorithm can be summed up as: - - 1) Load user-supplied initial test cases into the queue, - - 2) Take the next input file from the queue, - - 3) Attempt to trim the test case to the smallest size that doesn't alter - the measured behavior of the program, - - 4) Repeatedly mutate the file using a balanced and well-researched variety - of traditional fuzzing strategies, - - 5) If any of the generated mutations resulted in a new state transition - recorded by the instrumentation, add mutated output as a new entry in the - queue. - - 6) Go to 2. - -The discovered test cases are also periodically culled to eliminate ones that -have been obsoleted by newer, higher-coverage finds; and undergo several other -instrumentation-driven effort minimization steps. - -As a side result of the fuzzing process, the tool creates a small, -self-contained corpus of interesting test cases. These are extremely useful -for seeding other, labor- or resource-intensive testing regimes - for example, -for stress-testing browsers, office applications, graphics suites, or -closed-source tools. - -The fuzzer is thoroughly tested to deliver out-of-the-box performance far -superior to blind fuzzing or coverage-only tools. - - -## Instrumenting programs for use with AFL - -PLEASE NOTE: llvm_mode compilation with afl-clang-fast/afl-clang-fast++ -instead of afl-gcc/afl-g++ is much faster and has many cool features. -See llvm_mode/ - however few code does not compile with llvm. -We support llvm versions 3.4 to 12. - -When source code is available, instrumentation can be injected by a companion -tool that works as a drop-in replacement for gcc or clang in any standard build -process for third-party code. - -The instrumentation has a fairly modest performance impact; in conjunction with -other optimizations implemented by afl-fuzz, most programs can be fuzzed as fast -or even faster than possible with traditional tools. - -The correct way to recompile the target program may vary depending on the -specifics of the build process, but a nearly-universal approach would be: - -```shell -CC=/path/to/afl/afl-gcc ./configure -make clean all -``` - -For C++ programs, you'd would also want to set `CXX=/path/to/afl/afl-g++`. - -The clang wrappers (afl-clang and afl-clang++) can be used in the same way; -clang users may also opt to leverage a higher-performance instrumentation mode, -as described in [llvm_mode/README.md](llvm_mode/README.md). -Clang/LLVM has a much better performance and works with LLVM version 3.4 to 12. - -Using the LAF Intel performance enhancements are also recommended, see -[llvm_mode/README.laf-intel.md](llvm_mode/README.laf-intel.md) - -Using partial instrumentation is also recommended, see -[llvm_mode/README.instrument_file.md](llvm_mode/README.instrument_file.md) - -When testing libraries, you need to find or write a simple program that reads -data from stdin or from a file and passes it to the tested library. In such a -case, it is essential to link this executable against a static version of the -instrumented library or to make sure that the correct .so file is loaded at -runtime (usually by setting `LD_LIBRARY_PATH`). The simplest option is a static -build, usually possible via: - -```shell -CC=/path/to/afl/afl-gcc ./configure --disable-shared -``` - -Setting `AFL_HARDEN=1` when calling 'make' will cause the CC wrapper to -automatically enable code hardening options that make it easier to detect -simple memory bugs. Libdislocator, a helper library included with AFL (see -[libdislocator/README.md](libdislocator/README.md)) can help uncover heap corruption issues, too. - -PS. ASAN users are advised to review [docs/notes_for_asan.md](docs/notes_for_asan.md) -file for important caveats. - - -## Fuzzing binary-only targets - -When source code is *NOT* available, the fuzzer offers experimental support for -fast, on-the-fly instrumentation of black-box binaries. This is accomplished -with a version of QEMU running in the lesser-known "user space emulation" mode. - -QEMU is a project separate from AFL, but you can conveniently build the -feature by doing: - -```shell -cd qemu_mode -./build_qemu_support.sh -``` - -For additional instructions and caveats, see [qemu_mode/README.md](qemu_mode/README.md). - -If possible you should use the persistent mode, see [qemu_mode/README.persistent.md](qemu_mode/README.persistent.md). - -The mode is approximately 2-5x slower than compile-time instrumentation, is -less conducive to parallelization, and may have some other quirks. - -If [afl-dyninst](https://github.com/vanhauser-thc/afl-dyninst) works for -your binary, then you can use afl-fuzz normally and it will have twice -the speed compared to qemu_mode. - -A more comprehensive description of these and other options can be found in -[docs/binaryonly_fuzzing.md](docs/binaryonly_fuzzing.md) - -## Power schedules - -The power schedules were copied from Marcel Böhme's AFLfast implementation and -measure differently which queue entries to prefer and therefore may find -different paths faster for large queues. - -The available schedules are: - - - explore (default, original AFL) - - exploit (original AFL) - - fast (AFLfast) - - coe (AFLfast) - - quad (AFLfast) - - lin (AFLfast) - - rare (afl++ experimental) - - mmopt (afl++ experimental) - - seek (afl++ experimental) - -In parallel mode (-M/-S, several instances with the shared queue), we suggest -to run the main node using the default explore schedule (`-p explore`) and the -secondary nodes with different schedules. If a schedule does not perform well -for a target, restart the secondary nodes with a different schedule. - -More details can be found in the paper published at the 23rd ACM Conference on -Computer and Communications Security [CCS'16](https://www.sigsac.org/ccs/CCS2016/accepted-papers/) - -## Choosing initial test cases - -To operate correctly, the fuzzer requires one or more starting file that -contains a good example of the input data normally expected by the targeted -application. There are two basic rules: - - - Keep the files small. Under 1 kB is ideal, although not strictly necessary. - For a discussion of why size matters, see [perf_tips.md](docs/perf_tips.md). - - - Use multiple test cases only if they are functionally different from - each other. There is no point in using fifty different vacation photos - to fuzz an image library. - -You can find many good examples of starting files in the testcases/ subdirectory -that comes with this tool. - -PS. If a large corpus of data is available for screening, you may want to use -the afl-cmin utility to identify a subset of functionally distinct files that -exercise different code paths in the target binary. - - -## Fuzzing binaries - -The fuzzing process itself is carried out by the afl-fuzz utility. This program -requires a read-only directory with initial test cases, a separate place to -store its findings, plus a path to the binary to test. - -For target binaries that accept input directly from stdin, the usual syntax is: - -```shell -./afl-fuzz -i testcase_dir -o findings_dir /path/to/program [...params...] -``` - -For programs that take input from a file, use '@@' to mark the location in -the target's command line where the input file name should be placed. The -fuzzer will substitute this for you: - -```shell -./afl-fuzz -i testcase_dir -o findings_dir /path/to/program @@ -``` - -You can also use the -f option to have the mutated data written to a specific -file. This is useful if the program expects a particular file extension or so. - -Non-instrumented binaries can be fuzzed in the QEMU mode (add -Q in the command -line) or in a traditional, blind-fuzzer mode (specify -n). - -You can use -t and -m to override the default timeout and memory limit for the -executed process; rare examples of targets that may need these settings touched -include compilers and video decoders. - -Tips for optimizing fuzzing performance are discussed in [perf_tips.md](docs/perf_tips.md). - -Note that afl-fuzz starts by performing an array of deterministic fuzzing -steps, which can take several days, but tend to produce neat test cases. If you -want quick & dirty results right away - akin to zzuf and other traditional -fuzzers - add the -d option to the command line. - -## Interpreting output - -See the [docs/status_screen.md](docs/status_screen.md) file for information on -how to interpret the displayed stats and monitor the health of the process. Be -sure to consult this file especially if any UI elements are highlighted in red. - -The fuzzing process will continue until you press Ctrl-C. At a minimum, you want -to allow the fuzzer to complete one queue cycle, which may take anywhere from a -couple of hours to a week or so. - -There are three subdirectories created within the output directory and updated -in real-time: - - - queue/ - test cases for every distinctive execution path, plus all the - starting files given by the user. This is the synthesized corpus - mentioned in section 2. - - Before using this corpus for any other purposes, you can shrink - it to a smaller size using the afl-cmin tool. The tool will find - a smaller subset of files offering equivalent edge coverage. - - - crashes/ - unique test cases that cause the tested program to receive a - fatal signal (e.g., SIGSEGV, SIGILL, SIGABRT). The entries are - grouped by the received signal. - - - hangs/ - unique test cases that cause the tested program to time out. The - default time limit before something is classified as a hang is - the larger of 1 second and the value of the -t parameter. - The value can be fine-tuned by setting AFL_HANG_TMOUT, but this - is rarely necessary. - -Crashes and hangs are considered "unique" if the associated execution paths -involve any state transitions not seen in previously-recorded faults. If a -single bug can be reached in multiple ways, there will be some count inflation -early in the process, but this should quickly taper off. - -The file names for crashes and hangs are correlated with the parent, non-faulting -queue entries. This should help with debugging. - -When you can't reproduce a crash found by afl-fuzz, the most likely cause is -that you are not setting the same memory limit as used by the tool. Try: - -```shell -LIMIT_MB=50 -( ulimit -Sv $[LIMIT_MB << 10]; /path/to/tested_binary ... ) -``` - -Change LIMIT_MB to match the -m parameter passed to afl-fuzz. On OpenBSD, -also change -Sv to -Sd. - -Any existing output directory can be also used to resume aborted jobs; try: - -```shell -./afl-fuzz -i- -o existing_output_dir [...etc...] -``` - -If you have gnuplot installed, you can also generate some pretty graphs for any -active fuzzing task using afl-plot. For an example of how this looks like, -see [http://lcamtuf.coredump.cx/afl/plot/](http://lcamtuf.coredump.cx/afl/plot/). - -## Parallelized fuzzing - -Every instance of afl-fuzz takes up roughly one core. This means that on -multi-core systems, parallelization is necessary to fully utilize the hardware. -For tips on how to fuzz a common target on multiple cores or multiple networked -machines, please refer to [docs/parallel_fuzzing.md](docs/parallel_fuzzing.md). - -The parallel fuzzing mode also offers a simple way for interfacing AFL to other -fuzzers, to symbolic or concolic execution engines, and so forth; again, see the -last section of [docs/parallel_fuzzing.md](docs/parallel_fuzzing.md) for tips. - -## Fuzzer dictionaries - -By default, afl-fuzz mutation engine is optimized for compact data formats - -say, images, multimedia, compressed data, regular expression syntax, or shell -scripts. It is somewhat less suited for languages with particularly verbose and -redundant verbiage - notably including HTML, SQL, or JavaScript. - -To avoid the hassle of building syntax-aware tools, afl-fuzz provides a way to -seed the fuzzing process with an optional dictionary of language keywords, -magic headers, or other special tokens associated with the targeted data type --- and use that to reconstruct the underlying grammar on the go: - - [http://lcamtuf.blogspot.com/2015/01/afl-fuzz-making-up-grammar-with.html](http://lcamtuf.blogspot.com/2015/01/afl-fuzz-making-up-grammar-with.html) - -To use this feature, you first need to create a dictionary in one of the two -formats discussed in [dictionaries/README.md](dictionaries/README.md); -and then point the fuzzer to it via the -x option in the command line. - -(Several common dictionaries are already provided in that subdirectory, too.) - -There is no way to provide more structured descriptions of the underlying -syntax, but the fuzzer will likely figure out some of this based on the -instrumentation feedback alone. This actually works in practice, say: - - [http://lcamtuf.blogspot.com/2015/04/finding-bugs-in-sqlite-easy-way.html](http://lcamtuf.blogspot.com/2015/04/finding-bugs-in-sqlite-easy-way.html) - -PS. Even when no explicit dictionary is given, afl-fuzz will try to extract -existing syntax tokens in the input corpus by watching the instrumentation -very closely during deterministic byte flips. This works for some types of -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.tokencap.md). - -## Crash triage - -The coverage-based grouping of crashes usually produces a small data set that -can be quickly triaged manually or with a very simple GDB or Valgrind script. -Every crash is also traceable to its parent non-crashing test case in the -queue, making it easier to diagnose faults. - -Having said that, it's important to acknowledge that some fuzzing crashes can be -difficult to quickly evaluate for exploitability without a lot of debugging and -code analysis work. To assist with this task, afl-fuzz supports a very unique -"crash exploration" mode enabled with the -C flag. - -In this mode, the fuzzer takes one or more crashing test cases as the input -and uses its feedback-driven fuzzing strategies to very quickly enumerate all -code paths that can be reached in the program while keeping it in the -crashing state. - -Mutations that do not result in a crash are rejected; so are any changes that -do not affect the execution path. - -The output is a small corpus of files that can be very rapidly examined to see -what degree of control the attacker has over the faulting address, or whether -it is possible to get past an initial out-of-bounds read - and see what lies -beneath. - -Oh, one more thing: for test case minimization, give afl-tmin a try. The tool -can be operated in a very simple way: - -```shell -./afl-tmin -i test_case -o minimized_result -- /path/to/program [...] -``` - -The tool works with crashing and non-crashing test cases alike. In the crash -mode, it will happily accept instrumented and non-instrumented binaries. In the -non-crashing mode, the minimizer relies on standard AFL instrumentation to make -the file simpler without altering the execution path. - -The minimizer accepts the -m, -t, -f and @@ syntax in a manner compatible with -afl-fuzz. - -Another recent addition to AFL is the afl-analyze tool. It takes an input -file, attempts to sequentially flip bytes, and observes the behavior of the -tested program. It then color-codes the input based on which sections appear to -be critical, and which are not; while not bulletproof, it can often offer quick -insights into complex file formats. More info about its operation can be found -near the end of [docs/technical_details.md](docs/technical_details.md). - -## Going beyond crashes - -Fuzzing is a wonderful and underutilized technique for discovering non-crashing -design and implementation errors, too. Quite a few interesting bugs have been -found by modifying the target programs to call abort() when say: - - - Two bignum libraries produce different outputs when given the same - fuzzer-generated input, - - - An image library produces different outputs when asked to decode the same - input image several times in a row, - - - A serialization / deserialization library fails to produce stable outputs - when iteratively serializing and deserializing fuzzer-supplied data, - - - A compression library produces an output inconsistent with the input file - when asked to compress and then decompress a particular blob. - -Implementing these or similar sanity checks usually takes very little time; -if you are the maintainer of a particular package, you can make this code -conditional with `#ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION` (a flag also -shared with libfuzzer) or `#ifdef __AFL_COMPILER` (this one is just for AFL). - -## Common-sense risks - -Please keep in mind that, similarly to many other computationally-intensive -tasks, fuzzing may put a strain on your hardware and on the OS. In particular: - - - Your CPU will run hot and will need adequate cooling. In most cases, if - cooling is insufficient or stops working properly, CPU speeds will be - automatically throttled. That said, especially when fuzzing on less - suitable hardware (laptops, smartphones, etc), it's not entirely impossible - for something to blow up. - - - Targeted programs may end up erratically grabbing gigabytes of memory or - filling up disk space with junk files. AFL tries to enforce basic memory - limits, but can't prevent each and every possible mishap. The bottom line - is that you shouldn't be fuzzing on systems where the prospect of data loss - is not an acceptable risk. - - - Fuzzing involves billions of reads and writes to the filesystem. On modern - systems, this will be usually heavily cached, resulting in fairly modest - "physical" I/O - but there are many factors that may alter this equation. - It is your responsibility to monitor for potential trouble; with very heavy - I/O, the lifespan of many HDDs and SSDs may be reduced. - - A good way to monitor disk I/O on Linux is the 'iostat' command: - -```shell - $ iostat -d 3 -x -k [...optional disk ID...] -``` - -## Known limitations & areas for improvement - -Here are some of the most important caveats for AFL: - - - AFL detects faults by checking for the first spawned process dying due to - a signal (SIGSEGV, SIGABRT, etc). Programs that install custom handlers for - these signals may need to have the relevant code commented out. In the same - vein, faults in child processes spawned by the fuzzed target may evade - detection unless you manually add some code to catch that. - - - As with any other brute-force tool, the fuzzer offers limited coverage if - encryption, checksums, cryptographic signatures, or compression are used to - wholly wrap the actual data format to be tested. - - To work around this, you can comment out the relevant checks (see - examples/libpng_no_checksum/ for inspiration); if this is not possible, - you can also write a postprocessor, one of the hooks of custom mutators. - See [docs/custom_mutators.md](docs/custom_mutators.md) on how to use - `AFL_CUSTOM_MUTATOR_LIBRARY` - - - There are some unfortunate trade-offs with ASAN and 64-bit binaries. This - isn't due to any specific fault of afl-fuzz; see [docs/notes_for_asan.md](docs/notes_for_asan.md) - for tips. - - - There is no direct support for fuzzing network services, background - daemons, or interactive apps that require UI interaction to work. You may - need to make simple code changes to make them behave in a more traditional - way. Preeny may offer a relatively simple option, too - see: - [https://github.com/zardus/preeny](https://github.com/zardus/preeny) - - Some useful tips for modifying network-based services can be also found at: - [https://www.fastly.com/blog/how-to-fuzz-server-american-fuzzy-lop](https://www.fastly.com/blog/how-to-fuzz-server-american-fuzzy-lop) - - - AFL doesn't output human-readable coverage data. If you want to monitor - coverage, use afl-cov from Michael Rash: [https://github.com/mrash/afl-cov](https://github.com/mrash/afl-cov) - - - Occasionally, sentient machines rise against their creators. If this - happens to you, please consult [http://lcamtuf.coredump.cx/prep/](http://lcamtuf.coredump.cx/prep/). - -Beyond this, see INSTALL for platform-specific tips. - -## Special thanks - -Many of the improvements to the original afl and afl++ wouldn't be possible -without feedback, bug reports, or patches from: - -``` - Jann Horn Hanno Boeck - Felix Groebert Jakub Wilk - Richard W. M. Jones Alexander Cherepanov - Tom Ritter Hovik Manucharyan - Sebastian Roschke Eberhard Mattes - Padraig Brady Ben Laurie - @dronesec Luca Barbato - Tobias Ospelt Thomas Jarosch - Martin Carpenter Mudge Zatko - Joe Zbiciak Ryan Govostes - Michael Rash William Robinet - Jonathan Gray Filipe Cabecinhas - Nico Weber Jodie Cunningham - Andrew Griffiths Parker Thompson - Jonathan Neuschaefer Tyler Nighswander - Ben Nagy Samir Aguiar - Aidan Thornton Aleksandar Nikolich - Sam Hakim Laszlo Szekeres - David A. Wheeler Turo Lamminen - Andreas Stieger Richard Godbee - Louis Dassy teor2345 - Alex Moneger Dmitry Vyukov - Keegan McAllister Kostya Serebryany - Richo Healey Martijn Bogaard - rc0r Jonathan Foote - Christian Holler Dominique Pelle - Jacek Wielemborek Leo Barnes - Jeremy Barnes Jeff Trull - Guillaume Endignoux ilovezfs - Daniel Godas-Lopez Franjo Ivancic - Austin Seipp Daniel Komaromy - Daniel Binderman Jonathan Metzman - Vegard Nossum Jan Kneschke - Kurt Roeckx Marcel Boehme - Van-Thuan Pham Abhik Roychoudhury - Joshua J. Drake Toby Hutton - Rene Freingruber Sergey Davidoff - Sami Liedes Craig Young - Andrzej Jackowski Daniel Hodson - Nathan Voss Dominik Maier - Andrea Biondo Vincent Le Garrec - Khaled Yakdan Kuang-che Wu - Josephine Calliotte Konrad Welc -``` - -Thank you! -(For people sending pull requests - please add yourself to this list :-) - -## Contact - -Questions? Concerns? Bug reports? The contributors can be reached via -[https://github.com/AFLplusplus/AFLplusplus](https://github.com/AFLplusplus/AFLplusplus) - -There is also a mailing list for the afl project; to join, send a mail to -. Or, if you prefer to browse -archives first, try: [https://groups.google.com/group/afl-users](https://groups.google.com/group/afl-users) diff --git a/docs/screenshot.png b/docs/screenshot.png new file mode 100644 index 00000000..7b4dd7e4 Binary files /dev/null and b/docs/screenshot.png differ diff --git a/src/afl-fuzz.c b/src/afl-fuzz.c index f03c545d..39e4f32d 100644 --- a/src/afl-fuzz.c +++ b/src/afl-fuzz.c @@ -1234,6 +1234,7 @@ int main(int argc, char **argv_orig, char **envp) { } + (void)nice(-20); // real start time, we reset, so this works correctly with -V afl->start_time = get_cur_time(); -- cgit 1.4.1 From 2585a33005a25b317087b6ae9a53b886813b8015 Mon Sep 17 00:00:00 2001 From: van Hauser Date: Mon, 27 Jul 2020 09:21:07 +0200 Subject: fix readme --- README.md | 48 +++++++++++++++++++----------------------------- 1 file changed, 19 insertions(+), 29 deletions(-) diff --git a/README.md b/README.md index 97fa99b7..3cc7bf08 100644 --- a/README.md +++ b/README.md @@ -55,53 +55,34 @@ neverZero: - (1) default for LLVM >= 9.0, env var for older version due an efficiency bug in llvm <= 8 - - (2) GCC creates non-performant code, hence it is disabled in gcc_plugin - - (3) partially via AFL_CODE_START/AFL_CODE_END - - (4) with pcguard mode and LTO mode for LLVM >= 11 - - (5) upcoming, development in the branch - - (6) not compatible with LTO instrumentation and needs at least LLVM >= 4.1 - - (7) only in LTO mode with LLVM >= 11 + 1. default for LLVM >= 9.0, env var for older version due an efficiency bug in llvm <= 8 + 2. GCC creates non-performant code, hence it is disabled in gcc_plugin + 3. partially via AFL_CODE_START/AFL_CODE_END + 4. with pcguard mode and LTO mode for LLVM >= 11 + 5. upcoming, development in the branch + 6. not compatible with LTO instrumentation and needs at least LLVM >= 4.1 + 7. only in LTO mode with LLVM >= 11 Among others, the following features and patches have been integrated: * NeverZero patch for afl-gcc, llvm_mode, qemu_mode and unicorn_mode which prevents a wrapping map value to zero, increases coverage - * Persistent mode and deferred forkserver for qemu_mode - * Unicorn mode which allows fuzzing of binaries from completely different platforms (integration provided by domenukk) - * The new CmpLog instrumentation for LLVM and QEMU inspired by [Redqueen](https://www.syssec.ruhr-uni-bochum.de/media/emma/veroeffentlichungen/2018/12/17/NDSS19-Redqueen.pdf) - * Win32 PE binary-only fuzzing with QEMU and Wine - * AFLfast's power schedules by Marcel Böhme: [https://github.com/mboehme/aflfast](https://github.com/mboehme/aflfast) - * The MOpt mutator: [https://github.com/puppet-meteor/MOpt-AFL](https://github.com/puppet-meteor/MOpt-AFL) - * LLVM mode Ngram coverage by Adrian Herrera [https://github.com/adrianherrera/afl-ngram-pass](https://github.com/adrianherrera/afl-ngram-pass) - * InsTrim, an effective CFG llvm_mode instrumentation implementation for large targets: [https://github.com/csienslab/instrim](https://github.com/csienslab/instrim) - * C. Holler's afl-fuzz Python mutator module and llvm_mode instrument file support: [https://github.com/choller/afl](https://github.com/choller/afl) - * Custom mutator by a library (instead of Python) by kyakdan - * LAF-Intel/CompCov support for llvm_mode, qemu_mode and unicorn_mode (with enhanced capabilities) - * Radamsa and hongfuzz mutators (as custom mutators). - * QBDI mode to fuzz android native libraries via QBDI framework A more thorough list is available in the [PATCHES](docs/PATCHES.md) file. - So all in all this is the best-of afl that is currently out there :-) + So all in all this is the best-of afl that is out there :-) For new versions and additional information, check out: [https://github.com/AFLplusplus/AFLplusplus](https://github.com/AFLplusplus/AFLplusplus) @@ -274,6 +255,13 @@ anything below 9 is not recommended. use afl-gcc and afl-g++ ``` +Clickable README links for the chosen compiler: + + * [afl-clang-lto](llvm/README.lto.md) + * [afl-clang-fast](llvm/README.md) + * [afl-gcc-fast](gcc_plugin/README.md) + * afl-gcc has nt README as it has no features + #### b) Selecting instrumentation options The following options are available when you instrument with afl-clang-fast or @@ -478,7 +466,8 @@ afl-fuzz never stops fuzzing. To terminate afl++ simply press Control-C. When you start afl-fuzz you will see a user interface that shows what the status is: ![docs/screenshot.png](docs/screenshot.png) -All the entries are explained in [docs/status_screen.md](docs/status_screen.md) + +All labels are explained in [docs/status_screen.md](docs/status_screen.md) #### b) Using multiple cores @@ -517,6 +506,7 @@ Examples are e.g.: * [AFLsmart](https://github.com/aflsmart/aflsmart) * [FairFuzz](https://github.com/carolemieux/afl-rb) * [Neuzz](https://github.com/Dongdongshe/neuzz) + A long list can be found at [https://github.com/Microsvuln/Awesome-AFL](https://github.com/Microsvuln/Awesome-AFL) However you can also sync afl++ with honggfuzz, libfuzzer, entropic, etc. @@ -540,7 +530,7 @@ The `paths found` value is a bad indicator how good the coverage is. It is better to check out the exact lines of code that have been reached - and which have not been found so far. -An "easy" helper script for this is [afl-cov](https://github.com/vanhauser-thc/afl-cov), +An "easy" helper script for this is [https://github.com/vanhauser-thc/afl-cov](https://github.com/vanhauser-thc/afl-cov), just follow the README of that seperate project. If you see that an important area or a feature has not been covered so far then -- cgit 1.4.1 From 85684cd8b71e012b30b8b7147e6f99d27f59a492 Mon Sep 17 00:00:00 2001 From: van Hauser Date: Mon, 27 Jul 2020 09:25:30 +0200 Subject: fix readme --- README.md | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-) diff --git a/README.md b/README.md index 3cc7bf08..52c77403 100644 --- a/README.md +++ b/README.md @@ -53,8 +53,6 @@ | Auto dictionary | | x(7) | | | | | Snapshot LKM support | | x | | (x)(5) | | - neverZero: - 1. default for LLVM >= 9.0, env var for older version due an efficiency bug in llvm <= 8 2. GCC creates non-performant code, hence it is disabled in gcc_plugin 3. partially via AFL_CODE_START/AFL_CODE_END @@ -260,7 +258,7 @@ Clickable README links for the chosen compiler: * [afl-clang-lto](llvm/README.lto.md) * [afl-clang-fast](llvm/README.md) * [afl-gcc-fast](gcc_plugin/README.md) - * afl-gcc has nt README as it has no features + * afl-gcc has no README as it has no features #### b) Selecting instrumentation options -- cgit 1.4.1 From 8ebed3471ff85fa0f2b3f2c926e68038abf2cee2 Mon Sep 17 00:00:00 2001 From: van Hauser Date: Mon, 27 Jul 2020 09:30:42 +0200 Subject: fix readme --- README.md | 3 +++ 1 file changed, 3 insertions(+) diff --git a/README.md b/README.md index 52c77403..a5d4e116 100644 --- a/README.md +++ b/README.md @@ -342,6 +342,9 @@ described in [llvm/README.lto.md](llvm/README.lto.md) For `configure` build systems this is usually done by: `mkdir build; cd build; CC=afl-clang-fast CXX=afl-clang-fast++ cmake ..` +Some cmake scripts require something like `-DCMAKE_CC=... -DCMAKE_CXX=...` +or `-DCMAKE_C_COMPILER=... DCMAKE_CPP_COMPILER=...` instead. + Note that if you using the (better) afl-clang-lto compiler you also have to AR to llvm-ar[-VERSION] and RANLIB to llvm-ranlib[-VERSION] - as it is described in [llvm/README.lto.md](llvm/README.lto.md) -- cgit 1.4.1 From 11f25747a97156dd230a20ecda6a7bba672238bb Mon Sep 17 00:00:00 2001 From: van Hauser Date: Mon, 27 Jul 2020 09:39:15 +0200 Subject: child renice --- llvm_mode/afl-llvm-rt.o.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/llvm_mode/afl-llvm-rt.o.c b/llvm_mode/afl-llvm-rt.o.c index 0efde7aa..a509a9f0 100644 --- a/llvm_mode/afl-llvm-rt.o.c +++ b/llvm_mode/afl-llvm-rt.o.c @@ -514,6 +514,8 @@ static void __afl_start_snapshots(void) { if (!child_pid) { + (void)nice(-20); + signal(SIGCHLD, old_sigchld_handler); close(FORKSRV_FD); @@ -717,6 +719,8 @@ static void __afl_start_forkserver(void) { if (!child_pid) { + (void)nice(-20); + signal(SIGCHLD, old_sigchld_handler); close(FORKSRV_FD); -- cgit 1.4.1 From b508532c78d27fdfe2673652222551123491e61c Mon Sep 17 00:00:00 2001 From: Andrea Fioraldi Date: Mon, 27 Jul 2020 10:09:00 +0200 Subject: format --- qemu_mode/patches/afl-qemu-cpu-inl.h | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) diff --git a/qemu_mode/patches/afl-qemu-cpu-inl.h b/qemu_mode/patches/afl-qemu-cpu-inl.h index 07da9324..63b7581d 100644 --- a/qemu_mode/patches/afl-qemu-cpu-inl.h +++ b/qemu_mode/patches/afl-qemu-cpu-inl.h @@ -621,7 +621,12 @@ static void afl_wait_tsl(CPUState *cpu, int fd) { last_tb = tb_htable_lookup(cpu, c.last_tb.pc, c.last_tb.cs_base, c.last_tb.flags, c.cf_mask); #define TB_JMP_RESET_OFFSET_INVALID 0xffff - if (last_tb && (last_tb->jmp_reset_offset[c.tb_exit] != TB_JMP_RESET_OFFSET_INVALID)) { tb_add_jump(last_tb, c.tb_exit, tb); } + if (last_tb && (last_tb->jmp_reset_offset[c.tb_exit] != + TB_JMP_RESET_OFFSET_INVALID)) { + + tb_add_jump(last_tb, c.tb_exit, tb); + + } } -- cgit 1.4.1 From bd5308d8395419e1217b76b96c5cfbde83e95e9d Mon Sep 17 00:00:00 2001 From: van Hauser Date: Mon, 27 Jul 2020 10:27:57 +0200 Subject: fix readme --- README.md | 15 ++++++++------- 1 file changed, 8 insertions(+), 7 deletions(-) diff --git a/README.md b/README.md index a5d4e116..074be4b0 100644 --- a/README.md +++ b/README.md @@ -429,7 +429,8 @@ to be used in fuzzing! :-) In this final step we fuzz the target. There are not that many useful options to run the target - unless you want to -use many CPU cores for the fuzzing, which will make the fuzzing much more useful. +use many CPU cores/threads for the fuzzing, which will make the fuzzing much +more useful. If you just use one CPU for fuzzing, then you are fuzzing just for fun and not seriously :-) @@ -470,15 +471,15 @@ is: All labels are explained in [docs/status_screen.md](docs/status_screen.md) -#### b) Using multiple cores +#### b) Using multiple cores/threads -If you want to seriously fuzz then use as many cores as possible to fuzz your -target. +If you want to seriously fuzz then use as many cores/threads as possible to +fuzz your target. On the same machine - due to the nature how afl++ works - there is a maximum -number of CPU cores that are useful, more and the overall performance degrades -instead. This value depends on the target and the limit is between 24 and 64 -cores per machine. +number of CPU cores/threads that are useful, more and the overall performance +degrades instead. This value depends on the target and the limit is between 48 +and 96 cores/threads per machine. There should be one main fuzzer (`-M main` option) and as many secondary fuzzers (eg `-S variant1`) as you cores that you use. -- cgit 1.4.1 From 7944009a65da71fab7c722182bf1ac630bdde226 Mon Sep 17 00:00:00 2001 From: van Hauser Date: Mon, 27 Jul 2020 13:07:55 +0200 Subject: unrenice --- src/afl-fuzz.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/afl-fuzz.c b/src/afl-fuzz.c index 39e4f32d..103c9b2a 100644 --- a/src/afl-fuzz.c +++ b/src/afl-fuzz.c @@ -1234,7 +1234,7 @@ int main(int argc, char **argv_orig, char **envp) { } - (void)nice(-20); + // (void)nice(-20); // does not improve the speed // real start time, we reset, so this works correctly with -V afl->start_time = get_cur_time(); -- cgit 1.4.1 From 031aa240bc44d3ade1306b7f569260bbeeff7b64 Mon Sep 17 00:00:00 2001 From: van Hauser Date: Mon, 27 Jul 2020 23:09:26 +0200 Subject: unnice2 --- llvm_mode/afl-llvm-rt.o.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/llvm_mode/afl-llvm-rt.o.c b/llvm_mode/afl-llvm-rt.o.c index a509a9f0..4e3a4249 100644 --- a/llvm_mode/afl-llvm-rt.o.c +++ b/llvm_mode/afl-llvm-rt.o.c @@ -514,7 +514,7 @@ static void __afl_start_snapshots(void) { if (!child_pid) { - (void)nice(-20); + //(void)nice(-20); // does not seem to improve signal(SIGCHLD, old_sigchld_handler); -- cgit 1.4.1 From 23718e51984f61992ae213144f39a1c95638800b Mon Sep 17 00:00:00 2001 From: van Hauser Date: Tue, 28 Jul 2020 09:52:28 +0200 Subject: forgot getopt --- src/afl-fuzz.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/afl-fuzz.c b/src/afl-fuzz.c index 103c9b2a..eb4b6a87 100644 --- a/src/afl-fuzz.c +++ b/src/afl-fuzz.c @@ -272,7 +272,7 @@ int main(int argc, char **argv_orig, char **envp) { afl->shmem_testcase_mode = 1; // we always try to perform shmem fuzzing while ((opt = getopt(argc, argv, - "+c:i:I:o:f:m:t:T:dDnCB:S:M:x:QNUWe:p:s:V:E:L:hRP:")) > + "+c:i:I:o:f:F:m:t:T:dDnCB:S:M:x:QNUWe:p:s:V:E:L:hRP:")) > 0) { switch (opt) { -- cgit 1.4.1 From 8b21c2e4728b680a7b02851a4219e78daf54a466 Mon Sep 17 00:00:00 2001 From: van Hauser Date: Tue, 28 Jul 2020 09:55:32 +0200 Subject: unnice --- llvm_mode/afl-llvm-rt.o.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/llvm_mode/afl-llvm-rt.o.c b/llvm_mode/afl-llvm-rt.o.c index 4e3a4249..a2038188 100644 --- a/llvm_mode/afl-llvm-rt.o.c +++ b/llvm_mode/afl-llvm-rt.o.c @@ -719,7 +719,7 @@ static void __afl_start_forkserver(void) { if (!child_pid) { - (void)nice(-20); + //(void)nice(-20); signal(SIGCHLD, old_sigchld_handler); -- cgit 1.4.1 From a76e375d5cf74840fc87667eaa1c3f7e8550ca07 Mon Sep 17 00:00:00 2001 From: van Hauser Date: Tue, 28 Jul 2020 11:05:00 +0200 Subject: import lost PR when switching to stable, updated todo --- GNUmakefile | 2 ++ TODO.md | 4 ---- 2 files changed, 2 insertions(+), 4 deletions(-) diff --git a/GNUmakefile b/GNUmakefile index e2d7314f..7ed892ab 100644 --- a/GNUmakefile +++ b/GNUmakefile @@ -56,8 +56,10 @@ endif ifneq "$(shell uname)" "Darwin" ifeq "$(shell echo 'int main() {return 0; }' | $(CC) $(CFLAGS) -Werror -x c - -march=native -o .test 2>/dev/null && echo 1 || echo 0 ; rm -f .test )" "1" + ifndef SOURCE_DATE_EPOCH #CFLAGS_OPT += -march=native SPECIAL_PERFORMANCE += -march=native + endif endif # OS X does not like _FORTIFY_SOURCE=2 CFLAGS_OPT += -D_FORTIFY_SOURCE=2 diff --git a/TODO.md b/TODO.md index ad743b6b..8522b06d 100644 --- a/TODO.md +++ b/TODO.md @@ -2,17 +2,13 @@ ## Roadmap 2.67+ - - -i - + foreign fuzzer sync support: scandir with time sort - expand on AFL_LLVM_INSTRUMENT_FILE to also support sancov allowlist format - AFL_MAP_SIZE for qemu_mode and unicorn_mode - - namespace for targets? e.g. network - - learn from honggfuzz (mutations, maybe ptrace?) - CPU affinity for many cores? There seems to be an issue > 96 cores ## Further down the road afl-fuzz: - - ascii_only mode for mutation output - or use a custom mutator for this? - setting min_len/max_len/start_offset/end_offset limits for mutation output llvm_mode: -- cgit 1.4.1 From 0b3332d579ea50b7779295d8827c075ebd4f6b82 Mon Sep 17 00:00:00 2001 From: van Hauser Date: Tue, 28 Jul 2020 13:27:18 +0200 Subject: finalize new readme --- README.md | 219 +++++++------------------------------------------- llvm_mode/GNUmakefile | 2 +- 2 files changed, 32 insertions(+), 189 deletions(-) diff --git a/README.md b/README.md index 074be4b0..7268f5d1 100644 --- a/README.md +++ b/README.md @@ -95,7 +95,7 @@ The following branches exist: - * [master/trunk](https://github.com/AFLplusplus/AFLplusplus/) : stable state of afl++ - it is synced from dev from time to + * [stable/trunk](https://github.com/AFLplusplus/AFLplusplus/) : stable state of afl++ - it is synced from dev from time to time when we are satisfied with it's stability * [dev](https://github.com/AFLplusplus/AFLplusplus/tree/dev) : development state of afl++ - bleeding edge and you might catch a checkout which does not compile or has a bug. *We only accept PRs in dev!!* @@ -144,7 +144,7 @@ It is recommended to install the newest available gcc, clang and llvm-dev possible in your distribution! Note that "make distrib" also builds llvm_mode, qemu_mode, unicorn_mode and -more. If you just want plain afl then do "make all", however compiling and +more. If you just want plain afl++ then do "make all", however compiling and using at least llvm_mode is highly recommended for much better results - hence in this case @@ -351,9 +351,9 @@ described in [llvm/README.lto.md](llvm/README.lto.md) ##### other build systems or if configure/cmake didn't work -Sometimes cmake and configure do not pick up the afl compiler, or the ranlib/ar -that is needed - because this was just not foreseen by the developer of the -target. Or they have non-standard options. Figure out if there is a +Sometimes cmake and configure do not pick up the afl++ compiler, or the +ranlib/ar that is needed - because this was just not foreseen by the developer +of the target. Or they have non-standard options. Figure out if there is a non-standard way to set this, otherwise set the build normally and edit the generated build environment afterwards by hand to point to the right compiler (and/or ranlib and ar). @@ -456,6 +456,13 @@ If you need to stop and re-start the fuzzing, use the same command line option and switch the input directory with a dash (`-`): `afl-fuzz -i - -o output -- bin/target -d @@` +Note that afl-fuzz enforces memory limits to prevent the system to run out +of memory. By default this is 50MB for a process. If this is too little for +the target (which can can usually see that afl-fuzz bails with the message +that it could not connect to the forkserver), then you can increase this +with the `-m` option, the value is in MB. To disable any memory limits +(beware!) set `-m 0` - which is usually required for ASAN compiled targets. + Adding a dictionary helpful. See the [dictionaries/](dictionaries/) if something is already included for your data format, and tell afl-fuzz to load that dictionary by adding `-x dicationaries/FORMAT.dict`. With afl-clang-lto @@ -463,6 +470,10 @@ you have an autodictionary generation for which you need to do nothing except to use afl-clang-lto as the compiler. You also have the option to generate a dictionary yourself, see [libtokencap/README.md](libtokencap/README.md) +afl-fuzz has a variety of options that help to workaround target quirks like +specific locations for the input file (`-f`), not performing deterministic +fuzzing (`-d`) and many more. Check out `afl-fuzz -h`. + afl-fuzz never stops fuzzing. To terminate afl++ simply press Control-C. When you start afl-fuzz you will see a user interface that shows what the status @@ -500,8 +511,8 @@ All other secondaries should be: which you can set with e.g. `-p seek` You can also use different fuzzers. -If you are afl-spinoffs or afl conforming, then just use the same -o directory -and give it a unique `-S` name. +If you are using afl spinoffs or afl conforming fuzzers, then just use the +same -o directory and give it a unique `-S` name. Examples are e.g.: * [Angora](https://github.com/AngoraFuzzer/Angora) * [Untracer](https://github.com/FoRTE-Research/UnTracer-AFL) @@ -663,8 +674,7 @@ All these methods are extremely promising in experimental settings, but tend to suffer from reliability and performance problems in practical uses - and currently do not offer a viable alternative to "dumb" fuzzing techniques. - -## The afl-fuzz approach +## Background: The afl-fuzz approach American Fuzzy Lop is a brute-force fuzzer coupled with an exceedingly simple but rock-solid instrumentation-guided genetic algorithm. It uses a modified @@ -702,88 +712,7 @@ closed-source tools. The fuzzer is thoroughly tested to deliver out-of-the-box performance far superior to blind fuzzing or coverage-only tools. -## Instrumenting programs for use with AFL - -PLEASE NOTE: llvm_mode compilation with afl-clang-fast/afl-clang-fast++ -instead of afl-gcc/afl-g++ is much faster and has many cool features. -See llvm_mode/ - however few code does not compile with llvm. -We support llvm versions 3.4 to 12. - -When source code is available, instrumentation can be injected by a companion -tool that works as a drop-in replacement for gcc or clang in any standard build -process for third-party code. - -The instrumentation has a fairly modest performance impact; in conjunction with -other optimizations implemented by afl-fuzz, most programs can be fuzzed as fast -or even faster than possible with traditional tools. - -The correct way to recompile the target program may vary depending on the -specifics of the build process, but a nearly-universal approach would be: - -```shell -CC=/path/to/afl/afl-gcc ./configure -make clean all -``` - -For C++ programs, you'd would also want to set `CXX=/path/to/afl/afl-g++`. - -The clang wrappers (afl-clang and afl-clang++) can be used in the same way; -clang users may also opt to leverage a higher-performance instrumentation mode, -as described in [llvm_mode/README.md](llvm_mode/README.md). -Clang/LLVM has a much better performance and works with LLVM version 3.4 to 12. - -Using the LAF Intel performance enhancements are also recommended, see -[llvm_mode/README.laf-intel.md](llvm_mode/README.laf-intel.md) - -Using partial instrumentation is also recommended, see -[llvm_mode/README.instrument_file.md](llvm_mode/README.instrument_file.md) - -When testing libraries, you need to find or write a simple program that reads -data from stdin or from a file and passes it to the tested library. In such a -case, it is essential to link this executable against a static version of the -instrumented library or to make sure that the correct .so file is loaded at -runtime (usually by setting `LD_LIBRARY_PATH`). The simplest option is a static -build, usually possible via: - -```shell -CC=/path/to/afl/afl-gcc ./configure --disable-shared -``` - -Setting `AFL_HARDEN=1` when calling 'make' will cause the CC wrapper to -automatically enable code hardening options that make it easier to detect -simple memory bugs. Libdislocator, a helper library included with AFL (see -[libdislocator/README.md](libdislocator/README.md)) can help uncover heap corruption issues, too. - -PS. ASAN users are advised to review [docs/notes_for_asan.md](docs/notes_for_asan.md) -file for important caveats. - -## Power schedules - -The power schedules were copied from Marcel Böhme's AFLfast implementation and -measure differently which queue entries to prefer and therefore may find -different paths faster for large queues. - -The available schedules are: - - - explore (default, original AFL) - - exploit (original AFL) - - fast (AFLfast) - - coe (AFLfast) - - quad (AFLfast) - - lin (AFLfast) - - rare (afl++ experimental) - - mmopt (afl++ experimental) - - seek (afl++ experimental) - -In parallel mode (-M/-S, several instances with the shared queue), we suggest -to run the main node using the default explore schedule (`-p explore`) and the -secondary nodes with different schedules. If a schedule does not perform well -for a target, restart the secondary nodes with a different schedule. - -More details can be found in the paper published at the 23rd ACM Conference on -Computer and Communications Security [CCS'16](https://www.sigsac.org/ccs/CCS2016/accepted-papers/) - -## Choosing initial test cases +## Help: Choosing initial test cases To operate correctly, the fuzzer requires one or more starting file that contains a good example of the input data normally expected by the targeted @@ -803,45 +732,7 @@ PS. If a large corpus of data is available for screening, you may want to use the afl-cmin utility to identify a subset of functionally distinct files that exercise different code paths in the target binary. - -## Fuzzing binaries - -The fuzzing process itself is carried out by the afl-fuzz utility. This program -requires a read-only directory with initial test cases, a separate place to -store its findings, plus a path to the binary to test. - -For target binaries that accept input directly from stdin, the usual syntax is: - -```shell -./afl-fuzz -i testcase_dir -o findings_dir /path/to/program [...params...] -``` - -For programs that take input from a file, use '@@' to mark the location in -the target's command line where the input file name should be placed. The -fuzzer will substitute this for you: - -```shell -./afl-fuzz -i testcase_dir -o findings_dir /path/to/program @@ -``` - -You can also use the -f option to have the mutated data written to a specific -file. This is useful if the program expects a particular file extension or so. - -Non-instrumented binaries can be fuzzed in the QEMU mode (add -Q in the command -line) or in a traditional, blind-fuzzer mode (specify -n). - -You can use -t and -m to override the default timeout and memory limit for the -executed process; rare examples of targets that may need these settings touched -include compilers and video decoders. - -Tips for optimizing fuzzing performance are discussed in [perf_tips.md](docs/perf_tips.md). - -Note that afl-fuzz starts by performing an array of deterministic fuzzing -steps, which can take several days, but tend to produce neat test cases. If you -want quick & dirty results right away - akin to zzuf and other traditional -fuzzers - add the -d option to the command line. - -## Interpreting output +## Help: Interpreting output See the [docs/status_screen.md](docs/status_screen.md) file for information on how to interpret the displayed stats and monitor the health of the process. Be @@ -901,53 +792,7 @@ If you have gnuplot installed, you can also generate some pretty graphs for any active fuzzing task using afl-plot. For an example of how this looks like, see [http://lcamtuf.coredump.cx/afl/plot/](http://lcamtuf.coredump.cx/afl/plot/). -## Parallelized fuzzing - -Every instance of afl-fuzz takes up roughly one core. This means that on -multi-core systems, parallelization is necessary to fully utilize the hardware. -For tips on how to fuzz a common target on multiple cores or multiple networked -machines, please refer to [docs/parallel_fuzzing.md](docs/parallel_fuzzing.md). - -The parallel fuzzing mode also offers a simple way for interfacing AFL to other -fuzzers, to symbolic or concolic execution engines, and so forth; again, see the -last section of [docs/parallel_fuzzing.md](docs/parallel_fuzzing.md) for tips. - -## Fuzzer dictionaries - -By default, afl-fuzz mutation engine is optimized for compact data formats - -say, images, multimedia, compressed data, regular expression syntax, or shell -scripts. It is somewhat less suited for languages with particularly verbose and -redundant verbiage - notably including HTML, SQL, or JavaScript. - -To avoid the hassle of building syntax-aware tools, afl-fuzz provides a way to -seed the fuzzing process with an optional dictionary of language keywords, -magic headers, or other special tokens associated with the targeted data type --- and use that to reconstruct the underlying grammar on the go: - - [http://lcamtuf.blogspot.com/2015/01/afl-fuzz-making-up-grammar-with.html](http://lcamtuf.blogspot.com/2015/01/afl-fuzz-making-up-grammar-with.html) - -To use this feature, you first need to create a dictionary in one of the two -formats discussed in [dictionaries/README.md](dictionaries/README.md); -and then point the fuzzer to it via the -x option in the command line. - -(Several common dictionaries are already provided in that subdirectory, too.) - -There is no way to provide more structured descriptions of the underlying -syntax, but the fuzzer will likely figure out some of this based on the -instrumentation feedback alone. This actually works in practice, say: - - [http://lcamtuf.blogspot.com/2015/04/finding-bugs-in-sqlite-easy-way.html](http://lcamtuf.blogspot.com/2015/04/finding-bugs-in-sqlite-easy-way.html) - -PS. Even when no explicit dictionary is given, afl-fuzz will try to extract -existing syntax tokens in the input corpus by watching the instrumentation -very closely during deterministic byte flips. This works for some types of -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.tokencap.md). - -## Crash triage +## Help: Crash triage The coverage-based grouping of crashes usually produces a small data set that can be quickly triaged manually or with a very simple GDB or Valgrind script. @@ -981,13 +826,13 @@ can be operated in a very simple way: The tool works with crashing and non-crashing test cases alike. In the crash mode, it will happily accept instrumented and non-instrumented binaries. In the -non-crashing mode, the minimizer relies on standard AFL instrumentation to make +non-crashing mode, the minimizer relies on standard afl++ instrumentation to make the file simpler without altering the execution path. The minimizer accepts the -m, -t, -f and @@ syntax in a manner compatible with afl-fuzz. -Another recent addition to AFL is the afl-analyze tool. It takes an input +Another tool in afl++ is the afl-analyze tool. It takes an input file, attempts to sequentially flip bytes, and observes the behavior of the tested program. It then color-codes the input based on which sections appear to be critical, and which are not; while not bulletproof, it can often offer quick @@ -1015,7 +860,8 @@ found by modifying the target programs to call abort() when say: Implementing these or similar sanity checks usually takes very little time; if you are the maintainer of a particular package, you can make this code conditional with `#ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION` (a flag also -shared with libfuzzer) or `#ifdef __AFL_COMPILER` (this one is just for AFL). +shared with libfuzzer and honggfuzz) or `#ifdef __AFL_COMPILER` (this one is +just for AFL). ## Common-sense risks @@ -1029,7 +875,7 @@ tasks, fuzzing may put a strain on your hardware and on the OS. In particular: for something to blow up. - Targeted programs may end up erratically grabbing gigabytes of memory or - filling up disk space with junk files. AFL tries to enforce basic memory + filling up disk space with junk files. afl++ tries to enforce basic memory limits, but can't prevent each and every possible mishap. The bottom line is that you shouldn't be fuzzing on systems where the prospect of data loss is not an acceptable risk. @@ -1050,7 +896,7 @@ tasks, fuzzing may put a strain on your hardware and on the OS. In particular: Here are some of the most important caveats for AFL: - - AFL detects faults by checking for the first spawned process dying due to + - afl++ detects faults by checking for the first spawned process dying due to a signal (SIGSEGV, SIGABRT, etc). Programs that install custom handlers for these signals may need to have the relevant code commented out. In the same vein, faults in child processes spawned by the fuzzed target may evade @@ -1079,9 +925,6 @@ Here are some of the most important caveats for AFL: Some useful tips for modifying network-based services can be also found at: [https://www.fastly.com/blog/how-to-fuzz-server-american-fuzzy-lop](https://www.fastly.com/blog/how-to-fuzz-server-american-fuzzy-lop) - - AFL doesn't output human-readable coverage data. If you want to monitor - coverage, use afl-cov from Michael Rash: [https://github.com/mrash/afl-cov](https://github.com/mrash/afl-cov) - - Occasionally, sentient machines rise against their creators. If this happens to you, please consult [http://lcamtuf.coredump.cx/prep/](http://lcamtuf.coredump.cx/prep/). @@ -1146,6 +989,6 @@ Thank you! Questions? Concerns? Bug reports? The contributors can be reached via [https://github.com/AFLplusplus/AFLplusplus](https://github.com/AFLplusplus/AFLplusplus) -There is also a mailing list for the afl project; to join, send a mail to -. Or, if you prefer to browse -archives first, try: [https://groups.google.com/group/afl-users](https://groups.google.com/group/afl-users) +There is also a mailing list for the afl/afl++ project; to join, send a mail to +. Or, if you prefer to browse archives +first, try: [https://groups.google.com/group/afl-users](https://groups.google.com/group/afl-users) diff --git a/llvm_mode/GNUmakefile b/llvm_mode/GNUmakefile index 443322d7..fbb77236 100644 --- a/llvm_mode/GNUmakefile +++ b/llvm_mode/GNUmakefile @@ -39,7 +39,7 @@ else endif LLVMVER = $(shell $(LLVM_CONFIG) --version 2>/dev/null | sed 's/git//' ) -LLVM_UNSUPPORTED = $(shell $(LLVM_CONFIG) --version 2>/dev/null | egrep -q '^3\.[0-3]|^1[3-9]' && echo 1 || echo 0 ) +LLVM_UNSUPPORTED = $(shell $(LLVM_CONFIG) --version 2>/dev/null | egrep -q '^3\.[0-3]|^19' && echo 1 || echo 0 ) LLVM_NEW_API = $(shell $(LLVM_CONFIG) --version 2>/dev/null | egrep -q '^1[0-9]' && echo 1 || echo 0 ) LLVM_HAVE_LTO = $(shell $(LLVM_CONFIG) --version 2>/dev/null | egrep -q '^1[1-9]' && echo 1 || echo 0 ) LLVM_MAJOR = $(shell $(LLVM_CONFIG) --version 2>/dev/null | sed 's/\..*//') -- cgit 1.4.1 From b3f5b566b0d22b3addd7cf96fbffbbaf9757687d Mon Sep 17 00:00:00 2001 From: van Hauser Date: Tue, 28 Jul 2020 13:29:50 +0200 Subject: update contribute doc --- CONTRIBUTING.md | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md index 90049432..ccacef5f 100644 --- a/CONTRIBUTING.md +++ b/CONTRIBUTING.md @@ -1,5 +1,7 @@ # How to submit a Pull Request to AFLplusplus +All contributions (pull requests) must be made against our `dev` branch. + Each modified source file, before merging, must be formatted. ``` @@ -18,5 +20,5 @@ No camel case at all and use the AFL's macros wherever possible (e.g. WARNF, FATAL, MAP_SIZE, ...). Remember that AFLplusplus has to build and run on many platforms, so -generalize your Makefiles (or your patches to our pre-existing Makefiles) -to be as much generic as possible. +generalize your Makefiles/GNUmakefile (or your patches to our pre-existing +Makefiles) to be as much generic as possible. -- cgit 1.4.1 From 952e5b47ebaa0e1813e42ad7e7b524a519135d46 Mon Sep 17 00:00:00 2001 From: Andrea Fioraldi Date: Tue, 28 Jul 2020 16:02:15 +0200 Subject: allow custom mut with mopt if -L is -1 --- src/afl-fuzz-mutators.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/afl-fuzz-mutators.c b/src/afl-fuzz-mutators.c index 850266c2..b288cf9f 100644 --- a/src/afl-fuzz-mutators.c +++ b/src/afl-fuzz-mutators.c @@ -40,7 +40,7 @@ void setup_custom_mutators(afl_state_t *afl) { if (fn) { - if (afl->limit_time_sig) + if (afl->limit_time_sig && afl->limit_time_sig != -1) FATAL( "MOpt and custom mutator are mutually exclusive. We accept pull " "requests that integrates MOpt with the optional mutators " -- cgit 1.4.1 From 0b8c44cbb1a29d54695115cfbe3b27f7bfe67560 Mon Sep 17 00:00:00 2001 From: van Hauser Date: Tue, 28 Jul 2020 16:29:47 +0200 Subject: add FAQ --- README.md | 3 ++ docs/Changelog.md | 1 + docs/FAQ.md | 113 ++++++++++++++++++++++++++++++++++++++++++++++++++++++ 3 files changed, 117 insertions(+) create mode 100644 docs/FAQ.md diff --git a/README.md b/README.md index 7268f5d1..62137855 100644 --- a/README.md +++ b/README.md @@ -563,6 +563,9 @@ others, e.g. custom mutator modules, sync to very different fuzzers, etc. ### The End +Check out the [docs/FAQ](docs/FAQ.md) if it maybe answers your question (that +you might not even have known you had ;-) ). + This is basically all you need to know to professionally run fuzzing campaigns. If you want to know more, the rest of this README and the tons of texts in [docs/](docs/) will have you covered. diff --git a/docs/Changelog.md b/docs/Changelog.md index 38787def..cadfcb04 100644 --- a/docs/Changelog.md +++ b/docs/Changelog.md @@ -25,6 +25,7 @@ sending a mail to . - added afl-frida gum solution to examples/afl_frida (mostly imported from https://github.com/meme/hotwax/) - small fixes to afl-plot, afl-whatsup and man page creation + - new README, added FAQ ### Version ++2.66c (release) diff --git a/docs/FAQ.md b/docs/FAQ.md new file mode 100644 index 00000000..d848e08a --- /dev/null +++ b/docs/FAQ.md @@ -0,0 +1,113 @@ +# Frequently asked questions about afl++ + +## Contents + + 1. [What is an edge?](#what-is-an-edge) + 2. [Why is my stability below 100%?](#why-is-my-stability-below-100) + 3. [How can I improve the stability value](#how-can-i-improve-the-stability-value) + +If you find an interesting or important question missing, submit it via +[https://github.com/AFLplusplus/AFLplusplus/issues](https://github.com/AFLplusplus/AFLplusplus/issues) + +## What is an "edge" + +A program contains `functions`, `functions` contain the compiled machine code. +The compiled machine code in a `function` can be in a single or many `basic blocks`. +A `basic block` is the largest possible number of subsequent machine code +instructions that runs independent, meaning it does not split up to different +locations nor is it jumped into it from a different location: +``` +function() { + A: + some + code + B: + if (x) goto C; else goto D; + C: + some code + goto D + D: + some code + goto B + E: + return +} +``` +Every code block between two jump locations is a `basic block`. + +An `edge` is then the unique relationship between two `basic blocks` (from the +code example above): +``` + Block A + | + v + Block B <------+ + / \ | + v v | + Block C Block D --+ + \ + v + Block E +``` +Every line between two blocks is an `edge`. + +## Why is my stability below 100 + +Stability is measured by how many percent of the edges in the target are +"stable". Sending the same input again and again should take the exact same +path through the target every time. If that is the case, the stability is 100%. + +If however randomness happens, e.g. a thread reading from shared memory, +reaction to timing, etc. then in some of the re-executions with the same data +will result in the edge information being different accross runs. +Those edges that change are then flagged "unstable". + +The more "unstable" edges, the more difficult for afl++ to identify valid new +paths. + +A value above 90% is usually fine and a value above 80% is also still ok, and +even above 20% can still result in successful finds of bugs. +However, it is recommended that below 90% or 80% you should take measures to +improve the stability. + +## How can I improve the stability value + +Four steps are required to do this and requires quite some knowledge of +coding and/or disassembly and it is only effectively possible with +afl-clang-fast PCGUARD and afl-clang-lto LTO instrumentation! + + 1. First step: Identify which edge ID numbers are unstable + + run the target with `export AFL_DEBUG=1` for a few minutes then terminate. + The out/fuzzer_stats file will then show the edge IDs that were identified + as unstable. + + 2. Second step: Find the responsible function. + + a) For LTO instrumented binaries just disassemble or decompile the target + and look which edge is writing to that edge ID. Ghidra is a good tool + for this: [https://ghidra-sre.org/](https://ghidra-sre.org/) + + b) For PCGUARD instrumented binaries it is more difficult. Here you can + either modify the __sanitizer_cov_trace_pc_guard function in + llvm_mode/afl-llvm-rt.o.c to write a backtrace to a file if the ID in + __afl_area_ptr[*guard] is one of the unstable edge IDs. Then recompile + and reinstall llvm_mode and rebuild your target. Run the recompiled + target with afl-fuzz for a while and then check the file that you + wrote with the backtrace information. + Alternatively you can use `gdb` to hook __sanitizer_cov_trace_pc_guard_init + on start, check to which memory address the edge ID value is written + and set a write breakpoint to that address (`watch 0x.....`). + + 3. Third step: create a text file with the filenames + + Identify which source code files contain the functions that you need to + remove from instrumentation. + + Simply follow this document on how to do this: [llvm_mode/README.instrument_file.md](llvm_mode/README.instrument_file.md) + If PCGUARD is used, then you need to follow this guide: [http://clang.llvm.org/docs/SanitizerCoverage.html#partially-disabling-instrumentation](http://clang.llvm.org/docs/SanitizerCoverage.html#partially-disabling-instrumentation) + + 4. Fourth step: recompile the target + + Recompile, fuzz it, be happy :) + -- cgit 1.4.1 From 827ecd61f651fcf72d1f155ecf5afc76101dde75 Mon Sep 17 00:00:00 2001 From: van Hauser Date: Tue, 28 Jul 2020 22:13:37 +0200 Subject: add new example/guideline --- README.md | 1 + 1 file changed, 1 insertion(+) diff --git a/README.md b/README.md index 62137855..0716b750 100644 --- a/README.md +++ b/README.md @@ -193,6 +193,7 @@ Here are some good writeups to show how to effectively use AFL++: * [https://aflplus.plus/docs/tutorials/libxml2_tutorial/](https://aflplus.plus/docs/tutorials/libxml2_tutorial/) * [https://bananamafia.dev/post/gb-fuzz/](https://bananamafia.dev/post/gb-fuzz/) * [https://securitylab.github.com/research/fuzzing-challenges-solutions-1](https://securitylab.github.com/research/fuzzing-challenges-solutions-1) + * [https://securitylab.github.com/research/fuzzing-software-2](https://securitylab.github.com/research/fuzzing-software-2) * [https://securitylab.github.com/research/fuzzing-sockets-FTP](https://securitylab.github.com/research/fuzzing-sockets-FTP) If you are interested in fuzzing structured data (where you define what the -- cgit 1.4.1 From 015fde3703c7b67ee65d74b0f4b7b68b5d1e4d7e Mon Sep 17 00:00:00 2001 From: van Hauser Date: Tue, 28 Jul 2020 22:25:48 +0200 Subject: fix ffmpeg lto example --- llvm_mode/README.lto.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/llvm_mode/README.lto.md b/llvm_mode/README.lto.md index d54d4ee0..a4c969b9 100644 --- a/llvm_mode/README.lto.md +++ b/llvm_mode/README.lto.md @@ -157,7 +157,7 @@ instrument it: when compiling, so we have to trick configure: ``` -./configure --enable-lto --disable-shared +./configure --enable-lto --disable-shared --disable-inline-asm ``` 3. Now the configuration is done - and we edit the settings in `./ffbuild/config.mak` -- cgit 1.4.1