diff options
| -rw-r--r-- | include/afl-fuzz.h | 21 | ||||
| -rw-r--r-- | include/cmplog.h | 17 | ||||
| -rw-r--r-- | include/debug.h | 24 | ||||
| -rw-r--r-- | include/types.h | 43 | ||||
| -rw-r--r-- | instrumentation/afl-compiler-rt.o.c | 89 | ||||
| -rw-r--r-- | instrumentation/cmplog-instructions-pass.cc | 586 | ||||
| -rw-r--r-- | src/afl-cc.c | 49 | ||||
| -rw-r--r-- | src/afl-fuzz-init.c | 38 | ||||
| -rw-r--r-- | src/afl-fuzz-one.c | 81 | ||||
| -rw-r--r-- | src/afl-fuzz-queue.c | 1 | ||||
| -rw-r--r-- | src/afl-fuzz-redqueen.c | 1437 | ||||
| -rw-r--r-- | src/afl-fuzz-state.c | 1 | ||||
| -rw-r--r-- | src/afl-fuzz.c | 60 |
13 files changed, 2055 insertions, 392 deletions
diff --git a/include/afl-fuzz.h b/include/afl-fuzz.h index 988a907d..8a2122dc 100644 --- a/include/afl-fuzz.h +++ b/include/afl-fuzz.h @@ -145,12 +145,22 @@ extern s16 interesting_16[INTERESTING_8_LEN + INTERESTING_16_LEN]; extern s32 interesting_32[INTERESTING_8_LEN + INTERESTING_16_LEN + INTERESTING_32_LEN]; +struct tainted { + + u32 pos; + u32 len; + struct tainted *next; + struct tainted *prev; + +}; + struct queue_entry { u8 *fname; /* File name for the test case */ u32 len; /* Input length */ - u8 cal_failed; /* Calibration failed? */ + u8 colorized, /* Do not run redqueen stage again */ + cal_failed; /* Calibration failed? */ bool trim_done, /* Trimmed? */ was_fuzzed, /* historical, but needed for MOpt */ passed_det, /* Deterministic stages passed? */ @@ -158,7 +168,6 @@ struct queue_entry { var_behavior, /* Variable behavior? */ favored, /* Currently favored? */ fs_redundant, /* Marked as redundant in the fs? */ - fully_colorized, /* Do not run redqueen stage again */ is_ascii, /* Is the input just ascii text? */ disabled; /* Is disabled from fuzz selection */ @@ -183,7 +192,11 @@ struct queue_entry { u8 *testcase_buf; /* The testcase buffer, if loaded. */ - struct queue_entry *next; /* Next element, if any */ + u8 * cmplog_colorinput; /* the result buf of colorization */ + struct tainted *taint; /* Taint information from CmpLog */ + + struct queue_entry *mother, /* queue entry this based on */ + *next; /* Next element, if any */ }; @@ -636,6 +649,8 @@ typedef struct afl_state { /* cmplog forkserver ids */ s32 cmplog_fsrv_ctl_fd, cmplog_fsrv_st_fd; u32 cmplog_prev_timed_out; + u32 cmplog_max_filesize; + u32 cmplog_lvl; struct afl_pass_stat *pass_stats; struct cmp_map * orig_cmp_map; diff --git a/include/cmplog.h b/include/cmplog.h index bf557785..6392c503 100644 --- a/include/cmplog.h +++ b/include/cmplog.h @@ -30,8 +30,10 @@ #include "config.h" +#define CMPLOG_LVL_MAX 3 + #define CMP_MAP_W 65536 -#define CMP_MAP_H 256 +#define CMP_MAP_H 32 #define CMP_MAP_RTN_H (CMP_MAP_H / 4) #define SHAPE_BYTES(x) (x + 1) @@ -41,13 +43,12 @@ struct cmp_header { - unsigned hits : 20; - - unsigned cnt : 20; - unsigned id : 16; - - unsigned shape : 5; // from 0 to 31 + unsigned hits : 24; + unsigned id : 24; + unsigned shape : 5; unsigned type : 1; + unsigned attribute : 4; + unsigned reserved : 6; } __attribute__((packed)); @@ -55,6 +56,8 @@ struct cmp_operands { u64 v0; u64 v1; + u64 v0_128; + u64 v1_128; }; diff --git a/include/debug.h b/include/debug.h index ef5b195b..fc1f39cb 100644 --- a/include/debug.h +++ b/include/debug.h @@ -295,8 +295,8 @@ static inline const char *colorfilter(const char *x) { \ SAYF(bSTOP RESET_G1 CURSOR_SHOW cRST cLRD \ "\n[-] PROGRAM ABORT : " cRST x); \ - SAYF(cLRD "\n Location : " cRST "%s(), %s:%d\n\n", __func__, \ - __FILE__, __LINE__); \ + SAYF(cLRD "\n Location : " cRST "%s(), %s:%u\n\n", __func__, \ + __FILE__, (u32)__LINE__); \ exit(1); \ \ } while (0) @@ -308,8 +308,8 @@ static inline const char *colorfilter(const char *x) { \ SAYF(bSTOP RESET_G1 CURSOR_SHOW cRST cLRD \ "\n[-] PROGRAM ABORT : " cRST x); \ - SAYF(cLRD "\n Stop location : " cRST "%s(), %s:%d\n\n", __func__, \ - __FILE__, __LINE__); \ + SAYF(cLRD "\n Stop location : " cRST "%s(), %s:%u\n\n", __func__, \ + __FILE__, (u32)__LINE__); \ abort(); \ \ } while (0) @@ -322,8 +322,8 @@ static inline const char *colorfilter(const char *x) { fflush(stdout); \ SAYF(bSTOP RESET_G1 CURSOR_SHOW cRST cLRD \ "\n[-] SYSTEM ERROR : " cRST x); \ - SAYF(cLRD "\n Stop location : " cRST "%s(), %s:%d\n", __func__, \ - __FILE__, __LINE__); \ + SAYF(cLRD "\n Stop location : " cRST "%s(), %s:%u\n", __func__, \ + __FILE__, (u32)__LINE__); \ SAYF(cLRD " OS message : " cRST "%s\n", strerror(errno)); \ exit(1); \ \ @@ -344,12 +344,12 @@ static inline const char *colorfilter(const char *x) { /* Show a prefixed debug output. */ -#define DEBUGF(x...) \ - do { \ - \ - SAYF(cMGN "[D] " cBRI "DEBUG: " cRST x); \ - SAYF(cRST ""); \ - \ +#define DEBUGF(x...) \ + do { \ + \ + fprintf(stderr, cMGN "[D] " cBRI "DEBUG: " cRST x); \ + fprintf(stderr, cRST ""); \ + \ } while (0) /* Error-checking versions of read() and write() that call RPFATAL() as diff --git a/include/types.h b/include/types.h index 3e3bc953..d5c31597 100644 --- a/include/types.h +++ b/include/types.h @@ -26,9 +26,11 @@ #include <stdint.h> #include <stdlib.h> -typedef uint8_t u8; -typedef uint16_t u16; -typedef uint32_t u32; +typedef uint8_t u8; +typedef uint16_t u16; +typedef uint32_t u32; +typedef unsigned __int128 uint128_t; +typedef uint128_t u128; /* Extended forkserver option values */ @@ -57,10 +59,12 @@ typedef uint32_t u32; typedef unsigned long long u64; -typedef int8_t s8; -typedef int16_t s16; -typedef int32_t s32; -typedef int64_t s64; +typedef int8_t s8; +typedef int16_t s16; +typedef int32_t s32; +typedef int64_t s64; +typedef __int128 int128_t; +typedef int128_t s128; #ifndef MIN #define MIN(a, b) \ @@ -114,6 +118,31 @@ typedef int64_t s64; \ }) +// It is impossible to define 128 bit constants, so ... +#define SWAPN(_x, _l) \ + ({ \ + \ + u128 _res = (_x), _ret; \ + char *d = (char *)&_ret, *s = (char *)&_res; \ + int i; \ + for (i = 0; i < 16; i++) \ + d[15 - i] = s[i]; \ + u32 sr = 128U - ((_l) << 3U); \ + (_ret >>= sr); \ + (u128) _ret; \ + \ + }) + +#define SWAPNN(_x, _y, _l) \ + ({ \ + \ + char *d = (char *)(_x), *s = (char *)(_y); \ + u32 i, l = (_l)-1; \ + for (i = 0; i <= l; i++) \ + d[l - i] = s[i]; \ + \ + }) + #ifdef AFL_LLVM_PASS #if defined(__linux__) || !defined(__ANDROID__) #define AFL_SR(s) (srandom(s)) diff --git a/instrumentation/afl-compiler-rt.o.c b/instrumentation/afl-compiler-rt.o.c index b735d8df..5d75af78 100644 --- a/instrumentation/afl-compiler-rt.o.c +++ b/instrumentation/afl-compiler-rt.o.c @@ -161,7 +161,7 @@ void send_forkserver_error(int error) { u32 status; if (!error || error > 0xffff) return; status = (FS_OPT_ERROR | FS_OPT_SET_ERROR(error)); - if (write(FORKSRV_FD + 1, (char *)&status, 4) != 4) return; + if (write(FORKSRV_FD + 1, (char *)&status, 4) != 4) { return; } } @@ -544,11 +544,11 @@ static void __afl_start_snapshots(void) { if (__afl_dictionary_len && __afl_dictionary) status |= FS_OPT_AUTODICT; memcpy(tmp, &status, 4); - if (write(FORKSRV_FD + 1, tmp, 4) != 4) return; + if (write(FORKSRV_FD + 1, tmp, 4) != 4) { return; } if (__afl_sharedmem_fuzzing || (__afl_dictionary_len && __afl_dictionary)) { - if (read(FORKSRV_FD, &was_killed, 4) != 4) _exit(1); + if (read(FORKSRV_FD, &was_killed, 4) != 4) { _exit(1); } if (getenv("AFL_DEBUG")) { @@ -1207,7 +1207,9 @@ void __sanitizer_cov_trace_pc_guard_init(uint32_t *start, uint32_t *stop) { ///// CmpLog instrumentation -void __cmplog_ins_hook1(uint8_t arg1, uint8_t arg2) { +void __cmplog_ins_hook1(uint8_t arg1, uint8_t arg2, uint8_t attr) { + + // fprintf(stderr, "hook1 arg0=%02x arg1=%02x attr=%u\n", arg1, arg2, attr); if (unlikely(!__afl_cmp_map)) return; @@ -1216,6 +1218,7 @@ void __cmplog_ins_hook1(uint8_t arg1, uint8_t arg2) { k &= CMP_MAP_W - 1; __afl_cmp_map->headers[k].type = CMP_TYPE_INS; + __afl_cmp_map->headers[k].attribute = attr; u32 hits = __afl_cmp_map->headers[k].hits; __afl_cmp_map->headers[k].hits = hits + 1; @@ -1230,7 +1233,7 @@ void __cmplog_ins_hook1(uint8_t arg1, uint8_t arg2) { } -void __cmplog_ins_hook2(uint16_t arg1, uint16_t arg2) { +void __cmplog_ins_hook2(uint16_t arg1, uint16_t arg2, uint8_t attr) { if (unlikely(!__afl_cmp_map)) return; @@ -1239,6 +1242,7 @@ void __cmplog_ins_hook2(uint16_t arg1, uint16_t arg2) { k &= CMP_MAP_W - 1; __afl_cmp_map->headers[k].type = CMP_TYPE_INS; + __afl_cmp_map->headers[k].attribute = attr; u32 hits = __afl_cmp_map->headers[k].hits; __afl_cmp_map->headers[k].hits = hits + 1; @@ -1251,7 +1255,9 @@ void __cmplog_ins_hook2(uint16_t arg1, uint16_t arg2) { } -void __cmplog_ins_hook4(uint32_t arg1, uint32_t arg2) { +void __cmplog_ins_hook4(uint32_t arg1, uint32_t arg2, uint8_t attr) { + + // fprintf(stderr, "hook4 arg0=%x arg1=%x attr=%u\n", arg1, arg2, attr); if (unlikely(!__afl_cmp_map)) return; @@ -1260,6 +1266,7 @@ void __cmplog_ins_hook4(uint32_t arg1, uint32_t arg2) { k &= CMP_MAP_W - 1; __afl_cmp_map->headers[k].type = CMP_TYPE_INS; + __afl_cmp_map->headers[k].attribute = attr; u32 hits = __afl_cmp_map->headers[k].hits; __afl_cmp_map->headers[k].hits = hits + 1; @@ -1272,7 +1279,9 @@ void __cmplog_ins_hook4(uint32_t arg1, uint32_t arg2) { } -void __cmplog_ins_hook8(uint64_t arg1, uint64_t arg2) { +void __cmplog_ins_hook8(uint64_t arg1, uint64_t arg2, uint8_t attr) { + + // fprintf(stderr, "hook8 arg0=%lx arg1=%lx attr=%u\n", arg1, arg2, attr); if (unlikely(!__afl_cmp_map)) return; @@ -1281,6 +1290,7 @@ void __cmplog_ins_hook8(uint64_t arg1, uint64_t arg2) { k &= CMP_MAP_W - 1; __afl_cmp_map->headers[k].type = CMP_TYPE_INS; + __afl_cmp_map->headers[k].attribute = attr; u32 hits = __afl_cmp_map->headers[k].hits; __afl_cmp_map->headers[k].hits = hits + 1; @@ -1293,16 +1303,77 @@ void __cmplog_ins_hook8(uint64_t arg1, uint64_t arg2) { } +// support for u24 to u120 via llvm _ExitInt(). size is in bytes minus 1 +void __cmplog_ins_hookN(uint128_t arg1, uint128_t arg2, uint8_t attr, + uint8_t size) { + + // fprintf(stderr, "hookN arg0=%llx:%llx arg1=%llx:%llx bytes=%u attr=%u\n", + // (u64)(arg1 >> 64), (u64)arg1, (u64)(arg2 >> 64), (u64)arg2, size + 1, + // attr); + + if (unlikely(!__afl_cmp_map)) return; + + uintptr_t k = (uintptr_t)__builtin_return_address(0); + k = (k >> 4) ^ (k << 8); + k &= CMP_MAP_W - 1; + + __afl_cmp_map->headers[k].type = CMP_TYPE_INS; + __afl_cmp_map->headers[k].attribute = attr; + + u32 hits = __afl_cmp_map->headers[k].hits; + __afl_cmp_map->headers[k].hits = hits + 1; + + __afl_cmp_map->headers[k].shape = size; + + hits &= CMP_MAP_H - 1; + __afl_cmp_map->log[k][hits].v0 = (u64)arg1; + __afl_cmp_map->log[k][hits].v1 = (u64)arg2; + + if (size > 7) { + + __afl_cmp_map->log[k][hits].v0_128 = (u64)(arg1 >> 64); + __afl_cmp_map->log[k][hits].v1_128 = (u64)(arg2 >> 64); + + } + +} + +void __cmplog_ins_hook16(uint128_t arg1, uint128_t arg2, uint8_t attr) { + + if (unlikely(!__afl_cmp_map)) return; + + uintptr_t k = (uintptr_t)__builtin_return_address(0); + k = (k >> 4) ^ (k << 8); + k &= CMP_MAP_W - 1; + + __afl_cmp_map->headers[k].type = CMP_TYPE_INS; + __afl_cmp_map->headers[k].attribute = attr; + + u32 hits = __afl_cmp_map->headers[k].hits; + __afl_cmp_map->headers[k].hits = hits + 1; + + __afl_cmp_map->headers[k].shape = 15; + + hits &= CMP_MAP_H - 1; + __afl_cmp_map->log[k][hits].v0 = (u64)arg1; + __afl_cmp_map->log[k][hits].v1 = (u64)arg2; + __afl_cmp_map->log[k][hits].v0_128 = (u64)(arg1 >> 64); + __afl_cmp_map->log[k][hits].v1_128 = (u64)(arg2 >> 64); + +} + #if defined(__APPLE__) #pragma weak __sanitizer_cov_trace_const_cmp1 = __cmplog_ins_hook1 #pragma weak __sanitizer_cov_trace_const_cmp2 = __cmplog_ins_hook2 #pragma weak __sanitizer_cov_trace_const_cmp4 = __cmplog_ins_hook4 #pragma weak __sanitizer_cov_trace_const_cmp8 = __cmplog_ins_hook8 + #pragma weak __sanitizer_cov_trace_const_cmp16 = __cmplog_ins_hook16 #pragma weak __sanitizer_cov_trace_cmp1 = __cmplog_ins_hook1 #pragma weak __sanitizer_cov_trace_cmp2 = __cmplog_ins_hook2 #pragma weak __sanitizer_cov_trace_cmp4 = __cmplog_ins_hook4 #pragma weak __sanitizer_cov_trace_cmp8 = __cmplog_ins_hook8 + #pragma weak __sanitizer_cov_trace_cmp16 = __cmplog_ins_hook16 #else void __sanitizer_cov_trace_const_cmp1(uint8_t arg1, uint8_t arg2) __attribute__((alias("__cmplog_ins_hook1"))); @@ -1312,6 +1383,8 @@ void __sanitizer_cov_trace_const_cmp4(uint32_t arg1, uint32_t arg2) __attribute__((alias("__cmplog_ins_hook4"))); void __sanitizer_cov_trace_const_cmp8(uint64_t arg1, uint64_t arg2) __attribute__((alias("__cmplog_ins_hook8"))); +void __sanitizer_cov_trace_const_cmp16(uint128_t arg1, uint128_t arg2) + __attribute__((alias("__cmplog_ins_hook16"))); void __sanitizer_cov_trace_cmp1(uint8_t arg1, uint8_t arg2) __attribute__((alias("__cmplog_ins_hook1"))); @@ -1321,6 +1394,8 @@ void __sanitizer_cov_trace_cmp4(uint32_t arg1, uint32_t arg2) __attribute__((alias("__cmplog_ins_hook4"))); void __sanitizer_cov_trace_cmp8(uint64_t arg1, uint64_t arg2) __attribute__((alias("__cmplog_ins_hook8"))); +void __sanitizer_cov_trace_cmp16(uint128_t arg1, uint128_t arg2) + __attribute__((alias("__cmplog_ins_hook16"))); #endif /* defined(__APPLE__) */ void __sanitizer_cov_trace_switch(uint64_t val, uint64_t *cases) { diff --git a/instrumentation/cmplog-instructions-pass.cc b/instrumentation/cmplog-instructions-pass.cc index 3499ccf0..a74fb6c8 100644 --- a/instrumentation/cmplog-instructions-pass.cc +++ b/instrumentation/cmplog-instructions-pass.cc @@ -85,9 +85,25 @@ class CmpLogInstructions : public ModulePass { char CmpLogInstructions::ID = 0; +template <class Iterator> +Iterator Unique(Iterator first, Iterator last) { + + while (first != last) { + + Iterator next(first); + last = std::remove(++next, last, *first); + first = next; + + } + + return last; + +} + bool CmpLogInstructions::hookInstrs(Module &M) { std::vector<Instruction *> icomps; + std::vector<SwitchInst *> switches; LLVMContext & C = M.getContext(); Type * VoidTy = Type::getVoidTy(C); @@ -95,13 +111,15 @@ bool CmpLogInstructions::hookInstrs(Module &M) { IntegerType *Int16Ty = IntegerType::getInt16Ty(C); IntegerType *Int32Ty = IntegerType::getInt32Ty(C); IntegerType *Int64Ty = IntegerType::getInt64Ty(C); + IntegerType *Int128Ty = IntegerType::getInt128Ty(C); #if LLVM_VERSION_MAJOR < 9 Constant * #else FunctionCallee #endif - c1 = M.getOrInsertFunction("__cmplog_ins_hook1", VoidTy, Int8Ty, Int8Ty + c1 = M.getOrInsertFunction("__cmplog_ins_hook1", VoidTy, Int8Ty, Int8Ty, + Int8Ty #if LLVM_VERSION_MAJOR < 5 , NULL @@ -118,7 +136,8 @@ bool CmpLogInstructions::hookInstrs(Module &M) { #else FunctionCallee #endif - c2 = M.getOrInsertFunction("__cmplog_ins_hook2", VoidTy, Int16Ty, Int16Ty + c2 = M.getOrInsertFunction("__cmplog_ins_hook2", VoidTy, Int16Ty, Int16Ty, + Int8Ty #if LLVM_VERSION_MAJOR < 5 , NULL @@ -135,7 +154,8 @@ bool CmpLogInstructions::hookInstrs(Module &M) { #else FunctionCallee #endif - c4 = M.getOrInsertFunction("__cmplog_ins_hook4", VoidTy, Int32Ty, Int32Ty + c4 = M.getOrInsertFunction("__cmplog_ins_hook4", VoidTy, Int32Ty, Int32Ty, + Int8Ty #if LLVM_VERSION_MAJOR < 5 , NULL @@ -152,7 +172,8 @@ bool CmpLogInstructions::hookInstrs(Module &M) { #else FunctionCallee #endif - c8 = M.getOrInsertFunction("__cmplog_ins_hook8", VoidTy, Int64Ty, Int64Ty + c8 = M.getOrInsertFunction("__cmplog_ins_hook8", VoidTy, Int64Ty, Int64Ty, + Int8Ty #if LLVM_VERSION_MAJOR < 5 , NULL @@ -164,6 +185,42 @@ bool CmpLogInstructions::hookInstrs(Module &M) { FunctionCallee cmplogHookIns8 = c8; #endif +#if LLVM_VERSION_MAJOR < 9 + Constant * +#else + FunctionCallee +#endif + c16 = M.getOrInsertFunction("__cmplog_ins_hook16", VoidTy, Int128Ty, + Int128Ty, Int8Ty +#if LLVM_VERSION_MAJOR < 5 + , + NULL +#endif + ); +#if LLVM_VERSION_MAJOR < 9 + Function *cmplogHookIns16 = cast<Function>(c16); +#else + FunctionCallee cmplogHookIns16 = c16; +#endif + +#if LLVM_VERSION_MAJOR < 9 + Constant * +#else + FunctionCallee +#endif + cN = M.getOrInsertFunction("__cmplog_ins_hookN", VoidTy, Int128Ty, + Int128Ty, Int8Ty, Int8Ty +#if LLVM_VERSION_MAJOR < 5 + , + NULL +#endif + ); +#if LLVM_VERSION_MAJOR < 9 + Function *cmplogHookInsN = cast<Function>(cN); +#else + FunctionCallee cmplogHookInsN = cN; +#endif + /* iterate over all functions, bbs and instruction and add suitable calls */ for (auto &F : M) { @@ -174,35 +231,16 @@ bool CmpLogInstructions::hookInstrs(Module &M) { for (auto &IN : BB) { CmpInst *selectcmpInst = nullptr; - if ((selectcmpInst = dyn_cast<CmpInst>(&IN))) { - if (selectcmpInst->getPredicate() == CmpInst::ICMP_EQ || - selectcmpInst->getPredicate() == CmpInst::ICMP_NE || - selectcmpInst->getPredicate() == CmpInst::ICMP_UGT || - selectcmpInst->getPredicate() == CmpInst::ICMP_SGT || - selectcmpInst->getPredicate() == CmpInst::ICMP_ULT || - selectcmpInst->getPredicate() == CmpInst::ICMP_SLT || - selectcmpInst->getPredicate() == CmpInst::ICMP_UGE || - selectcmpInst->getPredicate() == CmpInst::ICMP_SGE || - selectcmpInst->getPredicate() == CmpInst::ICMP_ULE || - selectcmpInst->getPredicate() == CmpInst::ICMP_SLE || - selectcmpInst->getPredicate() == CmpInst::FCMP_OGE || - selectcmpInst->getPredicate() == CmpInst::FCMP_UGE || - selectcmpInst->getPredicate() == CmpInst::FCMP_OLE || - selectcmpInst->getPredicate() == CmpInst::FCMP_ULE || - selectcmpInst->getPredicate() == CmpInst::FCMP_OGT || - selectcmpInst->getPredicate() == CmpInst::FCMP_UGT || - selectcmpInst->getPredicate() == CmpInst::FCMP_OLT || - selectcmpInst->getPredicate() == CmpInst::FCMP_ULT || - selectcmpInst->getPredicate() == CmpInst::FCMP_UEQ || - selectcmpInst->getPredicate() == CmpInst::FCMP_OEQ || - selectcmpInst->getPredicate() == CmpInst::FCMP_UNE || - selectcmpInst->getPredicate() == CmpInst::FCMP_ONE) { - - icomps.push_back(selectcmpInst); + icomps.push_back(selectcmpInst); - } + } + + SwitchInst *switchInst = nullptr; + if ((switchInst = dyn_cast<SwitchInst>(BB.getTerminator()))) { + + if (switchInst->getNumCases() > 1) { switches.push_back(switchInst); } } @@ -212,101 +250,473 @@ bool CmpLogInstructions::hookInstrs(Module &M) { } - if (!icomps.size()) return false; - // if (!be_quiet) errs() << "Hooking " << icomps.size() << " cmp - // instructions\n"; + // unique the collected switches + switches.erase(Unique(switches.begin(), switches.end()), switches.end()); + + // Instrument switch values for cmplog + if (switches.size()) { + + if (!be_quiet) + errs() << "Hooking " << switches.size() << " switch instructions\n"; - for (auto &selectcmpInst : icomps) { + for (auto &SI : switches) { - IRBuilder<> IRB(selectcmpInst->getParent()); - IRB.SetInsertPoint(selectcmpInst); + Value * Val = SI->getCondition(); + unsigned int max_size = Val->getType()->getIntegerBitWidth(), cast_size; + unsigned char do_cast = 0; - auto op0 = selectcmpInst->getOperand(0); - auto op1 = selectcmpInst->getOperand(1); + if (!SI->getNumCases() || max_size <= 8) { - IntegerType * intTyOp0 = NULL; - IntegerType * intTyOp1 = NULL; - unsigned max_size = 0; - std::vector<Value *> args; + // if (!be_quiet) errs() << "skip trivial switch..\n"; + continue; - if (selectcmpInst->getOpcode() == Instruction::FCmp) { + } + + IRBuilder<> IRB(SI->getParent()); + IRB.SetInsertPoint(SI); + + if (max_size % 8) { + + max_size = (((max_size / 8) + 1) * 8); + do_cast = 1; + + } + + if (max_size > 128) { + + if (!be_quiet) { + + fprintf(stderr, + "Cannot handle this switch bit size: %u (truncating)\n", + max_size); + + } + + max_size = 128; + do_cast = 1; + + } + + // do we need to cast? + switch (max_size) { + + case 8: + case 16: + case 32: + case 64: + case 128: + cast_size = max_size; + break; + default: + cast_size = 128; + do_cast = 1; + + } + + Value *CompareTo = Val; + + if (do_cast) { + + ConstantInt *cint = dyn_cast<ConstantInt>(Val); + if (cint) { + + uint64_t val = cint->getZExtValue(); + // fprintf(stderr, "ConstantInt: %lu\n", val); + switch (cast_size) { + + case 8: + CompareTo = ConstantInt::get(Int8Ty, val); + break; + case 16: + CompareTo = ConstantInt::get(Int16Ty, val); + break; + case 32: + CompareTo = ConstantInt::get(Int32Ty, val); + break; + case 64: + CompareTo = ConstantInt::get(Int64Ty, val); + break; + case 128: + CompareTo = ConstantInt::get(Int128Ty, val); + break; + + } - auto ty0 = op0->getType(); - if (ty0->isHalfTy() + } else { + + CompareTo = IRB.CreateBitCast(Val, IntegerType::get(C, cast_size)); + + } + + } + + for (SwitchInst::CaseIt i = SI->case_begin(), e = SI->case_end(); i != e; + ++i) { + +#if LLVM_VERSION_MAJOR < 5 + ConstantInt *cint = i.getCaseValue(); +#else + ConstantInt *cint = i->getCaseValue(); +#endif + + if (cint) { + + std::vector<Value *> args; + args.push_back(CompareTo); + + Value *new_param = cint; + + if (do_cast) { + + uint64_t val = cint->getZExtValue(); + // fprintf(stderr, "ConstantInt: %lu\n", val); + switch (cast_size) { + + case 8: + new_param = ConstantInt::get(Int8Ty, val); + break; + case 16: + new_param = ConstantInt::get(Int16Ty, val); + break; + case 32: + new_param = ConstantInt::get(Int32Ty, val); + break; + case 64: + new_param = ConstantInt::get(Int64Ty, val); + break; + case 128: + new_param = ConstantInt::get(Int128Ty, val); + break; + + } + + } + + if (new_param) { + + args.push_back(new_param); + ConstantInt *attribute = ConstantInt::get(Int8Ty, 1); + args.push_back(attribute); + if (cast_size != max_size) { + + ConstantInt *bitsize = + ConstantInt::get(Int8Ty, (max_size / 8) - 1); + args.push_back(bitsize); + + } + + switch (cast_size) { + + case 8: + IRB.CreateCall(cmplogHookIns1, args); + break; + case 16: + IRB.CreateCall(cmplogHookIns2, args); + break; + case 32: + IRB.CreateCall(cmplogHookIns4, args); + break; + case 64: + IRB.CreateCall(cmplogHookIns8, args); + break; + case 128: + if (max_size == 128) { + + IRB.CreateCall(cmplogHookIns16, args); + + } else { + + IRB.CreateCall(cmplogHookInsN, args); + + } + + break; + + } + + } + + } + + } + + } + + } + + if (icomps.size()) { + + // if (!be_quiet) errs() << "Hooking " << icomps.size() << + // " cmp instructions\n"; + + for (auto &selectcmpInst : icomps) { + + IRBuilder<> IRB(selectcmpInst->getParent()); + IRB.SetInsertPoint(selectcmpInst); + + Value *op0 = selectcmpInst->getOperand(0); + Value *op1 = selectcmpInst->getOperand(1); + + IntegerType * intTyOp0 = NULL; + IntegerType * intTyOp1 = NULL; + unsigned max_size = 0, cast_size = 0; + unsigned char attr = 0, do_cast = 0; + std::vector<Value *> args; + + CmpInst *cmpInst = dyn_cast<CmpInst>(selectcmpInst); + + if (!cmpInst) { continue; } + + switch (cmpInst->getPredicate()) { + + case CmpInst::ICMP_NE: + case CmpInst::FCMP_UNE: + case CmpInst::FCMP_ONE: + break; + case CmpInst::ICMP_EQ: + case CmpInst::FCMP_UEQ: + case CmpInst::FCMP_OEQ: + attr += 1; + break; + case CmpInst::ICMP_UGT: + case CmpInst::ICMP_SGT: + case CmpInst::FCMP_OGT: + case CmpInst::FCMP_UGT: + attr += 2; + break; + case CmpInst::ICMP_UGE: + case CmpInst::ICMP_SGE: + case CmpInst::FCMP_OGE: + case CmpInst::FCMP_UGE: + attr += 3; + break; + case CmpInst::ICMP_ULT: + case CmpInst::ICMP_SLT: + case CmpInst::FCMP_OLT: + case CmpInst::FCMP_ULT: + attr += 4; + break; + case CmpInst::ICMP_ULE: + case CmpInst::ICMP_SLE: + case CmpInst::FCMP_OLE: + case CmpInst::FCMP_ULE: + attr += 5; + break; + default: + break; + + } + + if (selectcmpInst->getOpcode() == Instruction::FCmp) { + + auto ty0 = op0->getType(); + if (ty0->isHalfTy() #if LLVM_VERSION_MAJOR >= 11 - || ty0->isBFloatTy() + || ty0->isBFloatTy() #endif - ) - max_size = 16; - else if (ty0->isFloatTy()) - max_size = 32; - else if (ty0->isDoubleTy()) - max_size = 64; + ) + max_size = 16; + else if (ty0->isFloatTy()) + max_size = 32; + else if (ty0->isDoubleTy()) + max_size = 64; + else if (ty0->isX86_FP80Ty()) + max_size = 80; + else if (ty0->isFP128Ty() || ty0->isPPC_FP128Ty()) + max_size = 128; + + attr += 8; + do_cast = 1; - if (max_size) { + } else { - Value *V0 = IRB.CreateBitCast(op0, IntegerType::get(C, max_size)); - intTyOp0 = dyn_cast<IntegerType>(V0->getType()); - Value *V1 = IRB.CreateBitCast(op1, IntegerType::get(C, max_size)); - intTyOp1 = dyn_cast<IntegerType>(V1->getType()); + intTyOp0 = dyn_cast<IntegerType>(op0->getType()); + intTyOp1 = dyn_cast<IntegerType>(op1->getType()); if (intTyOp0 && intTyOp1) { max_size = intTyOp0->getBitWidth() > intTyOp1->getBitWidth() ? intTyOp0->getBitWidth() : intTyOp1->getBitWidth(); - args.push_back(V0); - args.push_back(V1); - } else { + } + + } + + if (!max_size) { continue; } + + // _ExtInt() with non-8th values + if (max_size % 8) { + + max_size = (((max_size / 8) + 1) * 8); + do_cast = 1; + + } + + if (max_size > 128) { + + if (!be_quiet) { - max_size = 0; + fprintf(stderr, + "Cannot handle this compare bit size: %u (truncating)\n", + max_size); } + max_size = 128; + do_cast = 1; + + } + + // do we need to cast? + switch (max_size) { + + case 8: + case 16: + case 32: + case 64: + case 128: + cast_size = max_size; + break; + default: + cast_size = 128; + do_cast = 1; + } - } else { + if (do_cast) { + + // F*cking LLVM optimized out any kind of bitcasts of ConstantInt values + // creating illegal calls. WTF. So we have to work around this. + + ConstantInt *cint = dyn_cast<ConstantInt>(op0); + if (cint) { + + uint64_t val = cint->getZExtValue(); + // fprintf(stderr, "ConstantInt: %lu\n", val); + ConstantInt *new_param = NULL; + switch (cast_size) { + + case 8: + new_param = ConstantInt::get(Int8Ty, val); + break; + case 16: + new_param = ConstantInt::get(Int16Ty, val); + break; + case 32: + new_param = ConstantInt::get(Int32Ty, val); + break; + case 64: + new_param = ConstantInt::get(Int64Ty, val); + break; + case 128: + new_param = ConstantInt::get(Int128Ty, val); + break; + + } + + if (!new_param) { continue; } + args.push_back(new_param); + + } else { + + Value *V0 = IRB.CreateBitCast(op0, IntegerType::get(C, cast_size)); + args.push_back(V0); + + } + + cint = dyn_cast<ConstantInt>(op1); + if (cint) { + + uint64_t val = cint->getZExtValue(); + ConstantInt *new_param = NULL; + switch (cast_size) { + + case 8: + new_param = ConstantInt::get(Int8Ty, val); + break; + case 16: + new_param = ConstantInt::get(Int16Ty, val); + break; + case 32: + new_param = ConstantInt::get(Int32Ty, val); + break; + case 64: + new_param = ConstantInt::get(Int64Ty, val); + break; + case 128: + new_param = ConstantInt::get(Int128Ty, val); + break; + + } + + if (!new_param) { continue; } + args.push_back(new_param); + + } else { + + Value *V1 = IRB.CreateBitCast(op1, IntegerType::get(C, cast_size)); + args.push_back(V1); - intTyOp0 = dyn_cast<IntegerType>(op0->getType()); - intTyOp1 = dyn_cast<IntegerType>(op1->getType()); + } - if (intTyOp0 && intTyOp1) { + } else { - max_size = intTyOp0->getBitWidth() > intTyOp1->getBitWidth() - ? intTyOp0->getBitWidth() - : intTyOp1->getBitWidth(); args.push_back(op0); args.push_back(op1); } - } + ConstantInt *attribute = ConstantInt::get(Int8Ty, attr); + args.push_back(attribute); + + if (cast_size != max_size) { + + ConstantInt *bitsize = ConstantInt::get(Int8Ty, (max_size / 8) - 1); + args.push_back(bitsize); + + } + + // fprintf(stderr, "_ExtInt(%u) castTo %u with attr %u didcast %u\n", + // max_size, cast_size, attr, do_cast); + + switch (cast_size) { - if (max_size < 8 || max_size > 64 || !intTyOp0 || !intTyOp1) continue; - - switch (max_size) { - - case 8: - IRB.CreateCall(cmplogHookIns1, args); - break; - case 16: - IRB.CreateCall(cmplogHookIns2, args); - break; - case 32: - IRB.CreateCall(cmplogHookIns4, args); - break; - case 64: - IRB.CreateCall(cmplogHookIns8, args); - break; - default: - break; + case 8: + IRB.CreateCall(cmplogHookIns1, args); + break; + case 16: + IRB.CreateCall(cmplogHookIns2, args); + break; + case 32: + IRB.CreateCall(cmplogHookIns4, args); + break; + case 64: + IRB.CreateCall(cmplogHookIns8, args); + break; + case 128: + if (max_size == 128) { + + IRB.CreateCall(cmplogHookIns16, args); + + } else { + + IRB.CreateCall(cmplogHookInsN, args); + + } + + break; + + } } } - return true; + if (switches.size() || icomps.size()) + return true; + else + return false; } diff --git a/src/afl-cc.c b/src/afl-cc.c index 8fb42718..02c9c7c5 100644 --- a/src/afl-cc.c +++ b/src/afl-cc.c @@ -529,9 +529,9 @@ static void edit_params(u32 argc, char **argv, char **envp) { cc_params[cc_par_cnt++] = alloc_printf( "-Wl,-mllvm=-load=%s/cmplog-routines-pass.so", obj_path); cc_params[cc_par_cnt++] = alloc_printf( - "-Wl,-mllvm=-load=%s/split-switches-pass.so", obj_path); - cc_params[cc_par_cnt++] = alloc_printf( "-Wl,-mllvm=-load=%s/cmplog-instructions-pass.so", obj_path); + cc_params[cc_par_cnt++] = alloc_printf( + "-Wl,-mllvm=-load=%s/split-switches-pass.so", obj_path); } else { @@ -541,18 +541,18 @@ static void edit_params(u32 argc, char **argv, char **envp) { cc_params[cc_par_cnt++] = alloc_printf("%s/cmplog-routines-pass.so", obj_path); - // reuse split switches from laf cc_params[cc_par_cnt++] = "-Xclang"; cc_params[cc_par_cnt++] = "-load"; cc_params[cc_par_cnt++] = "-Xclang"; cc_params[cc_par_cnt++] = - alloc_printf("%s/split-switches-pass.so", obj_path); + alloc_printf("%s/cmplog-instructions-pass.so", obj_path); + // reuse split switches from laf cc_params[cc_par_cnt++] = "-Xclang"; cc_params[cc_par_cnt++] = "-load"; cc_params[cc_par_cnt++] = "-Xclang"; cc_params[cc_par_cnt++] = - alloc_printf("%s/cmplog-instructions-pass.so", obj_path); + alloc_printf("%s/split-switches-pass.so", obj_path); } @@ -792,10 +792,8 @@ static void edit_params(u32 argc, char **argv, char **envp) { } -#if defined(USEMMAP) - #if !defined(__HAIKU__) +#if defined(USEMMAP) && !defined(__HAIKU__) cc_params[cc_par_cnt++] = "-lrt"; - #endif #endif cc_params[cc_par_cnt++] = "-D__AFL_HAVE_MANUAL_CONTROL=1"; @@ -858,6 +856,7 @@ static void edit_params(u32 argc, char **argv, char **envp) { cc_params[cc_par_cnt++] = "-D__AFL_COVERAGE_DISCARD()=__afl_coverage_discard()"; cc_params[cc_par_cnt++] = "-D__AFL_COVERAGE_ABORT()=__afl_coverage_abort()"; + cc_params[cc_par_cnt++] = "-D__AFL_FUZZ_TESTCASE_BUF=(__afl_fuzz_ptr ? __afl_fuzz_ptr : " "__afl_fuzz_alt_ptr)"; @@ -967,10 +966,8 @@ static void edit_params(u32 argc, char **argv, char **envp) { alloc_printf("-Wl,--dynamic-list=%s/dynamic_list.txt", obj_path); #endif - #if defined(USEMMAP) - #if !defined(__HAIKU__) + #if defined(USEMMAP) && !defined(__HAIKU__) cc_params[cc_par_cnt++] = "-lrt"; - #endif #endif } @@ -1278,7 +1275,6 @@ int main(int argc, char **argv, char **envp) { } - // this is a hidden option if (strncasecmp(ptr2, "llvmnative", strlen("llvmnative")) == 0 || strncasecmp(ptr2, "llvm-native", strlen("llvm-native")) == 0) { @@ -1349,29 +1345,28 @@ int main(int argc, char **argv, char **envp) { if (strncasecmp(ptr2, "ngram", strlen("ngram")) == 0) { - ptr2 += strlen("ngram"); - while (*ptr2 && (*ptr2 < '0' || *ptr2 > '9')) - ptr2++; + u8 *ptr3 = ptr2 + strlen("ngram"); + while (*ptr3 && (*ptr3 < '0' || *ptr3 > '9')) + ptr3++; - if (!*ptr2) { + if (!*ptr3) { - if ((ptr2 = getenv("AFL_LLVM_NGRAM_SIZE")) == NULL) + if ((ptr3 = getenv("AFL_LLVM_NGRAM_SIZE")) == NULL) FATAL( "you must set the NGRAM size with (e.g. for value 2) " "AFL_LLVM_INSTRUMENT=ngram-2"); } - ngram_size = atoi(ptr2); + ngram_size = atoi(ptr3); if (ngram_size < 2 || ngram_size > NGRAM_SIZE_MAX) FATAL( "NGRAM instrumentation option must be between 2 and " - "NGRAM_SIZE_MAX " - "(%u)", + "NGRAM_SIZE_MAX (%u)", NGRAM_SIZE_MAX); instrument_opt_mode |= (INSTRUMENT_OPT_NGRAM); - ptr2 = alloc_printf("%u", ngram_size); - setenv("AFL_LLVM_NGRAM_SIZE", ptr2, 1); + u8 *ptr4 = alloc_printf("%u", ngram_size); + setenv("AFL_LLVM_NGRAM_SIZE", ptr4, 1); } @@ -1507,6 +1502,7 @@ int main(int argc, char **argv, char **envp) { "((instrumentation/README.ngram.md)\n" " INSTRIM: Dominator tree (for LLVM <= 6.0) " "(instrumentation/README.instrim.md)\n\n"); + #undef NATIVE_MSG SAYF( @@ -1641,16 +1637,15 @@ int main(int argc, char **argv, char **envp) { if (have_lto) SAYF("afl-cc LTO with ld=%s %s\n", AFL_REAL_LD, AFL_CLANG_FLTO); if (have_llvm) - SAYF("afl-cc LLVM version %d using binary path \"%s\".\n", LLVM_MAJOR, + SAYF("afl-cc LLVM version %d using the binary path \"%s\".\n", LLVM_MAJOR, LLVM_BINDIR); #endif -#if defined(USEMMAP) +#ifdef USEMMAP #if !defined(__HAIKU__) - cc_params[cc_par_cnt++] = "-lrt"; - SAYF("Compiled with shm_open support (adds -lrt when linking).\n"); - #else SAYF("Compiled with shm_open support.\n"); + #else + SAYF("Compiled with shm_open support (adds -lrt when linking).\n"); #endif #else SAYF("Compiled with shmat support.\n"); diff --git a/src/afl-fuzz-init.c b/src/afl-fuzz-init.c index dbffa4f9..cbff6d7e 100644 --- a/src/afl-fuzz-init.c +++ b/src/afl-fuzz-init.c @@ -729,6 +729,30 @@ void read_testcases(afl_state_t *afl, u8 *directory) { add_to_queue(afl, fn2, st.st_size >= MAX_FILE ? MAX_FILE : st.st_size, passed_det); + if (unlikely(afl->shm.cmplog_mode)) { + + if (afl->cmplog_lvl == 1) { + + if (!afl->cmplog_max_filesize || + afl->cmplog_max_filesize < st.st_size) { + + afl->cmplog_max_filesize = st.st_size; + + } + + } else if (afl->cmplog_lvl == 2) { + + if (!afl->cmplog_max_filesize || + afl->cmplog_max_filesize > st.st_size) { + + afl->cmplog_max_filesize = st.st_size; + + } + + } + + } + if (unlikely(afl->schedule >= FAST && afl->schedule <= RARE)) { u64 cksum = hash64(afl->fsrv.trace_bits, afl->fsrv.map_size, HASH_CONST); @@ -756,6 +780,20 @@ void read_testcases(afl_state_t *afl, u8 *directory) { } + if (unlikely(afl->shm.cmplog_mode)) { + + if (afl->cmplog_max_filesize < 1024) { + + afl->cmplog_max_filesize = 1024; + + } else { + + afl->cmplog_max_filesize = (((afl->cmplog_max_filesize >> 10) + 1) << 10); + + } + + } + afl->last_path_time = 0; afl->queued_at_start = afl->queued_paths; diff --git a/src/afl-fuzz-one.c b/src/afl-fuzz-one.c index f9509e86..596bae22 100644 --- a/src/afl-fuzz-one.c +++ b/src/afl-fuzz-one.c @@ -165,7 +165,7 @@ static u8 could_be_arith(u32 old_val, u32 new_val, u8 blen) { /* See if one-byte adjustments to any byte could produce this result. */ - for (i = 0; i < blen; ++i) { + for (i = 0; (u8)i < blen; ++i) { u8 a = old_val >> (8 * i), b = new_val >> (8 * i); @@ -193,7 +193,7 @@ static u8 could_be_arith(u32 old_val, u32 new_val, u8 blen) { diffs = 0; - for (i = 0; i < blen / 2; ++i) { + for (i = 0; (u8)i < blen / 2; ++i) { u16 a = old_val >> (16 * i), b = new_val >> (16 * i); @@ -290,7 +290,7 @@ static u8 could_be_interest(u32 old_val, u32 new_val, u8 blen, u8 check_le) { /* See if two-byte insertions over old_val could give us new_val. */ - for (i = 0; (s32)i < blen - 1; ++i) { + for (i = 0; (u8)i < blen - 1; ++i) { for (j = 0; j < sizeof(interesting_16) / 2; ++j) { @@ -545,14 +545,31 @@ u8 fuzz_one_original(afl_state_t *afl) { else orig_perf = perf_score = calculate_score(afl, afl->queue_cur); - if (unlikely(perf_score == 0)) { goto abandon_entry; } + if (unlikely(perf_score <= 0)) { goto abandon_entry; } - if (unlikely(afl->shm.cmplog_mode && !afl->queue_cur->fully_colorized)) { + if (unlikely(afl->shm.cmplog_mode && + afl->queue_cur->colorized < afl->cmplog_lvl && + (u32)len <= afl->cmplog_max_filesize)) { - if (input_to_state_stage(afl, in_buf, out_buf, len, - afl->queue_cur->exec_cksum)) { + if (unlikely(len < 4)) { - goto abandon_entry; + afl->queue_cur->colorized = 0xff; + + } else { + + if (afl->cmplog_lvl == 3 || + (afl->cmplog_lvl == 2 && afl->queue_cur->tc_ref) || + !(afl->fsrv.total_execs % afl->queued_paths) || + get_cur_time() - afl->last_path_time > 15000) { + + if (input_to_state_stage(afl, in_buf, out_buf, len, + afl->queue_cur->exec_cksum)) { + + goto abandon_entry; + + } + + } } @@ -2796,7 +2813,7 @@ static u8 mopt_common_fuzzing(afl_state_t *afl, MOpt_globals_t MOpt_globals) { } - s32 len, temp_len; + u32 len, temp_len; u32 i; u32 j; u8 *in_buf, *out_buf, *orig_in, *ex_tmp, *eff_map = 0; @@ -2952,14 +2969,31 @@ static u8 mopt_common_fuzzing(afl_state_t *afl, MOpt_globals_t MOpt_globals) { else orig_perf = perf_score = calculate_score(afl, afl->queue_cur); - if (unlikely(perf_score == 0)) { goto abandon_entry; } + if (unlikely(perf_score <= 0)) { goto abandon_entry; } - if (unlikely(afl->shm.cmplog_mode && !afl->queue_cur->fully_colorized)) { + if (unlikely(afl->shm.cmplog_mode && + afl->queue_cur->colorized < afl->cmplog_lvl && + (u32)len <= afl->cmplog_max_filesize)) { - if (input_to_state_stage(afl, in_buf, out_buf, len, - afl->queue_cur->exec_cksum)) { + if (unlikely(len < 4)) { - goto abandon_entry; + afl->queue_cur->colorized = 0xff; + + } else { + + if (afl->cmplog_lvl == 3 || + (afl->cmplog_lvl == 2 && afl->queue_cur->tc_ref) || + !(afl->fsrv.total_execs % afl->queued_paths) || + get_cur_time() - afl->last_path_time > 15000) { + + if (input_to_state_stage(afl, in_buf, out_buf, len, + afl->queue_cur->exec_cksum)) { + + goto abandon_entry; + + } + + } } @@ -3315,7 +3349,7 @@ static u8 mopt_common_fuzzing(afl_state_t *afl, MOpt_globals_t MOpt_globals) { orig_hit_cnt = new_hit_cnt; - for (i = 0; (s32)i < len - 1; ++i) { + for (i = 0; i < len - 1; ++i) { /* Let's consult the effector map... */ @@ -3357,7 +3391,7 @@ static u8 mopt_common_fuzzing(afl_state_t *afl, MOpt_globals_t MOpt_globals) { orig_hit_cnt = new_hit_cnt; - for (i = 0; (s32)i < len - 3; ++i) { + for (i = 0; i < len - 3; ++i) { /* Let's consult the effector map... */ if (!eff_map[EFF_APOS(i)] && !eff_map[EFF_APOS(i + 1)] && @@ -3489,7 +3523,7 @@ skip_bitflip: orig_hit_cnt = new_hit_cnt; - for (i = 0; (s32)i < len - 1; ++i) { + for (i = 0; i < len - 1; ++i) { u16 orig = *(u16 *)(out_buf + i); @@ -3615,7 +3649,7 @@ skip_bitflip: orig_hit_cnt = new_hit_cnt; - for (i = 0; (s32)i < len - 3; ++i) { + for (i = 0; i < len - 3; ++i) { u32 orig = *(u32 *)(out_buf + i); @@ -3805,7 +3839,7 @@ skip_arith: orig_hit_cnt = new_hit_cnt; - for (i = 0; (s32)i < len - 1; ++i) { + for (i = 0; i < len - 1; ++i) { u16 orig = *(u16 *)(out_buf + i); @@ -3891,7 +3925,7 @@ skip_arith: orig_hit_cnt = new_hit_cnt; - for (i = 0; (s32)i < len - 3; ++i) { + for (i = 0; i < len - 3; ++i) { u32 orig = *(u32 *)(out_buf + i); @@ -4120,7 +4154,7 @@ skip_user_extras: /* See the comment in the earlier code; extras are sorted by size. */ - if ((s32)(afl->a_extras[j].len) > (s32)(len - i) || + if ((afl->a_extras[j].len) > (len - i) || !memcmp(afl->a_extras[j].data, out_buf + i, afl->a_extras[j].len) || !memchr(eff_map + EFF_APOS(i), 1, EFF_SPAN_ALEN(i, afl->a_extras[j].len))) { @@ -4837,7 +4871,7 @@ pacemaker_fuzzing: u32 copy_from, copy_to, copy_len; copy_len = choose_block_len(afl, new_len - 1); - if ((s32)copy_len > temp_len) copy_len = temp_len; + if (copy_len > temp_len) copy_len = temp_len; copy_from = rand_below(afl, new_len - copy_len + 1); copy_to = rand_below(afl, temp_len - copy_len + 1); @@ -5033,8 +5067,7 @@ pacemaker_fuzzing: the last differing byte. Bail out if the difference is just a single byte or so. */ - locate_diffs(in_buf, new_buf, MIN(len, (s32)target->len), &f_diff, - &l_diff); + locate_diffs(in_buf, new_buf, MIN(len, target->len), &f_diff, &l_diff); if (f_diff < 0 || l_diff < 2 || f_diff == l_diff) { diff --git a/src/afl-fuzz-queue.c b/src/afl-fuzz-queue.c index 66938635..aec57a6e 100644 --- a/src/afl-fuzz-queue.c +++ b/src/afl-fuzz-queue.c @@ -433,6 +433,7 @@ void add_to_queue(afl_state_t *afl, u8 *fname, u32 len, u8 passed_det) { q->passed_det = passed_det; q->trace_mini = NULL; q->testcase_buf = NULL; + q->mother = afl->queue_cur; #ifdef INTROSPECTION q->bitsmap_size = afl->bitsmap_size; diff --git a/src/afl-fuzz-redqueen.c b/src/afl-fuzz-redqueen.c index 28585afe..955a9232 100644 --- a/src/afl-fuzz-redqueen.c +++ b/src/afl-fuzz-redqueen.c @@ -28,6 +28,8 @@ #include "afl-fuzz.h" #include "cmplog.h" +//#define _DEBUG + ///// Colorization struct range { @@ -35,6 +37,8 @@ struct range { u32 start; u32 end; struct range *next; + struct range *prev; + u8 ok; }; @@ -44,6 +48,8 @@ static struct range *add_range(struct range *ranges, u32 start, u32 end) { r->start = start; r->end = end; r->next = ranges; + r->ok = 0; + if (likely(ranges)) ranges->prev = r; return r; } @@ -51,45 +57,61 @@ static struct range *add_range(struct range *ranges, u32 start, u32 end) { static struct range *pop_biggest_range(struct range **ranges) { struct range *r = *ranges; - struct range *prev = NULL; struct range *rmax = NULL; - struct range *prev_rmax = NULL; u32 max_size = 0; while (r) { - u32 s = r->end - r->start; - if (s >= max_size) { + if (!r->ok) { + + u32 s = 1 + r->end - r->start; + + if (s >= max_size) { + + max_size = s; + rmax = r; - max_size = s; - prev_rmax = prev; - rmax = r; + } } - prev = r; r = r->next; } - if (rmax) { + return rmax; - if (prev_rmax) { +} - prev_rmax->next = rmax->next; +#ifdef _DEBUG +// static int logging = 0; +static void dump(char *txt, u8 *buf, u32 len) { - } else { + u32 i; + fprintf(stderr, "DUMP %s %llx ", txt, hash64(buf, len, 0)); + for (i = 0; i < len; i++) + fprintf(stderr, "%02x", buf[i]); + fprintf(stderr, "\n"); - *ranges = rmax->next; +} - } +static void dump_file(char *path, char *name, u32 counter, u8 *buf, u32 len) { - } + char fn[4096]; + if (!path) path = "."; + snprintf(fn, sizeof(fn), "%s/%s%d", path, name, counter); + int fd = open(fn, O_RDWR | O_CREAT | O_TRUNC, 0644); + if (fd >= 0) { - return rmax; + write(fd, buf, len); + close(fd); + + } } +#endif + static u8 get_exec_checksum(afl_state_t *afl, u8 *buf, u32 len, u64 *cksum) { if (unlikely(common_fuzz_stuff(afl, buf, len))) { return 1; } @@ -99,107 +121,270 @@ static u8 get_exec_checksum(afl_state_t *afl, u8 *buf, u32 len, u64 *cksum) { } -static void xor_replace(u8 *buf, u32 len) { +/* replace everything with different values but stay in the same type */ +static void type_replace(afl_state_t *afl, u8 *buf, u32 len) { u32 i; + u8 c; for (i = 0; i < len; ++i) { - buf[i] ^= 0xff; + // wont help for UTF or non-latin charsets + do { + + switch (buf[i]) { + + case 'A' ... 'F': + c = 'A' + rand_below(afl, 1 + 'F' - 'A'); + break; + case 'a' ... 'f': + c = 'a' + rand_below(afl, 1 + 'f' - 'a'); + break; + case '0': + c = '1'; + break; + case '1': + c = '0'; + break; + case '2' ... '9': + c = '2' + rand_below(afl, 1 + '9' - '2'); + break; + case 'G' ... 'Z': + c = 'G' + rand_below(afl, 1 + 'Z' - 'G'); + break; + case 'g' ... 'z': + c = 'g' + rand_below(afl, 1 + 'z' - 'g'); + break; + case '!' ... '*': + c = '!' + rand_below(afl, 1 + '*' - '!'); + break; + case ',' ... '.': + c = ',' + rand_below(afl, 1 + '.' - ','); + break; + case ':' ... '@': + c = ':' + rand_below(afl, 1 + '@' - ':'); + break; + case '[' ... '`': + c = '[' + rand_below(afl, 1 + '`' - '['); + break; + case '{' ... '~': + c = '{' + rand_below(afl, 1 + '~' - '{'); + break; + case '+': + c = '/'; + break; + case '/': + c = '+'; + break; + case ' ': + c = '\t'; + break; + case '\t': + c = ' '; + break; + /* + case '\r': + case '\n': + // nothing ... + break; + */ + default: + c = (buf[i] ^ 0xff); + + } + + } while (c == buf[i]); + + buf[i] = c; } } -static u8 colorization(afl_state_t *afl, u8 *buf, u32 len, u64 exec_cksum) { +static u8 colorization(afl_state_t *afl, u8 *buf, u32 len, u64 exec_cksum, + struct tainted **taints) { - struct range *ranges = add_range(NULL, 0, len); - u8 * backup = ck_alloc_nozero(len); + struct range * ranges = add_range(NULL, 0, len - 1), *rng; + struct tainted *taint = NULL; + u8 * backup = ck_alloc_nozero(len); + u8 * changed = ck_alloc_nozero(len); u64 orig_hit_cnt, new_hit_cnt; orig_hit_cnt = afl->queued_paths + afl->unique_crashes; afl->stage_name = "colorization"; afl->stage_short = "colorization"; - afl->stage_max = 1000; + afl->stage_max = (len << 1); - struct range *rng = NULL; afl->stage_cur = 0; + memcpy(backup, buf, len); + memcpy(changed, buf, len); + type_replace(afl, changed, len); + while ((rng = pop_biggest_range(&ranges)) != NULL && afl->stage_cur < afl->stage_max) { - u32 s = rng->end - rng->start; + u32 s = 1 + rng->end - rng->start; + + memcpy(buf + rng->start, changed + rng->start, s); - if (s != 0) { + u64 cksum; + u64 start_us = get_cur_time_us(); + if (unlikely(get_exec_checksum(afl, buf, len, &cksum))) { - /* Range not empty */ + goto checksum_fail; - memcpy(backup, buf + rng->start, s); - xor_replace(buf + rng->start, s); + } + + u64 stop_us = get_cur_time_us(); + + /* Discard if the mutations change the path or if it is too decremental + in speed - how could the same path have a much different speed + though ...*/ + if (cksum != exec_cksum || + (unlikely(stop_us - start_us > 3 * afl->queue_cur->exec_us) && + likely(!afl->fixed_seed))) { + + memcpy(buf + rng->start, backup + rng->start, s); - u64 cksum; - u64 start_us = get_cur_time_us(); - if (unlikely(get_exec_checksum(afl, buf, len, &cksum))) { + if (s > 1) { // to not add 0 size ranges - goto checksum_fail; + ranges = add_range(ranges, rng->start, rng->start - 1 + s / 2); + ranges = add_range(ranges, rng->start + s / 2, rng->end); } - u64 stop_us = get_cur_time_us(); + if (ranges == rng) { + + ranges = rng->next; + if (ranges) { ranges->prev = NULL; } + + } else if (rng->next) { + + rng->prev->next = rng->next; + rng->next->prev = rng->prev; - /* Discard if the mutations change the paths or if it is too decremental - in speed */ - if (cksum != exec_cksum || - ((stop_us - start_us > 2 * afl->queue_cur->exec_us) && - likely(!afl->fixed_seed))) { + } else { - ranges = add_range(ranges, rng->start, rng->start + s / 2); - ranges = add_range(ranges, rng->start + s / 2 + 1, rng->end); - memcpy(buf + rng->start, backup, s); + if (rng->prev) { rng->prev->next = NULL; } } + free(rng); + + } else { + + rng->ok = 1; + } - ck_free(rng); - rng = NULL; ++afl->stage_cur; } - if (afl->stage_cur < afl->stage_max) { afl->queue_cur->fully_colorized = 1; } + rng = ranges; + while (rng) { - new_hit_cnt = afl->queued_paths + afl->unique_crashes; - afl->stage_finds[STAGE_COLORIZATION] += new_hit_cnt - orig_hit_cnt; - afl->stage_cycles[STAGE_COLORIZATION] += afl->stage_cur; - ck_free(backup); + rng = rng->next; - ck_free(rng); - rng = NULL; + } - while (ranges) { + u32 i = 1; + u32 positions = 0; + while (i) { + restart: + i = 0; + struct range *r = NULL; + u32 pos = (u32)-1; rng = ranges; - ranges = rng->next; - ck_free(rng); - rng = NULL; - } + while (rng) { - return 0; + if (rng->ok == 1 && rng->start < pos) { -checksum_fail: - if (rng) { ck_free(rng); } - ck_free(backup); + if (taint && taint->pos + taint->len == rng->start) { + + taint->len += (1 + rng->end - rng->start); + positions += (1 + rng->end - rng->start); + rng->ok = 2; + goto restart; + + } else { + + r = rng; + pos = rng->start; + + } + + } + + rng = rng->next; + + } + + if (r) { + + struct tainted *t = ck_alloc_nozero(sizeof(struct tainted)); + t->pos = r->start; + t->len = 1 + r->end - r->start; + positions += (1 + r->end - r->start); + if (likely(taint)) { taint->prev = t; } + t->next = taint; + t->prev = NULL; + taint = t; + r->ok = 2; + i = 1; + } + + } + + *taints = taint; + + /* temporary: clean ranges */ while (ranges) { rng = ranges; ranges = rng->next; ck_free(rng); - rng = NULL; } + new_hit_cnt = afl->queued_paths + afl->unique_crashes; + +#ifdef _DEBUG + /* + char fn[4096]; + snprintf(fn, sizeof(fn), "%s/introspection_color.txt", afl->out_dir); + FILE *f = fopen(fn, "a"); + if (f) { + + */ + FILE *f = stderr; + fprintf(f, + "Colorization: fname=%s len=%u result=%u execs=%u found=%llu " + "taint=%u\n", + afl->queue_cur->fname, len, afl->queue_cur->colorized, afl->stage_cur, + new_hit_cnt - orig_hit_cnt, positions); +/* + fclose(f); + + } + +*/ +#endif + + afl->stage_finds[STAGE_COLORIZATION] += new_hit_cnt - orig_hit_cnt; + afl->stage_cycles[STAGE_COLORIZATION] += afl->stage_cur; + ck_free(backup); + ck_free(changed); + + return 0; + +checksum_fail: + ck_free(backup); + ck_free(changed); + return 1; } @@ -212,12 +397,19 @@ static u8 its_fuzz(afl_state_t *afl, u8 *buf, u32 len, u8 *status) { orig_hit_cnt = afl->queued_paths + afl->unique_crashes; +#ifdef _DEBUG + dump("DATA", buf, len); +#endif + if (unlikely(common_fuzz_stuff(afl, buf, len))) { return 1; } new_hit_cnt = afl->queued_paths + afl->unique_crashes; if (unlikely(new_hit_cnt != orig_hit_cnt)) { +#ifdef _DEBUG + fprintf(stderr, "NEW FIND\n"); +#endif *status = 1; } else { @@ -278,11 +470,33 @@ static int strntoull(const char *str, size_t sz, char **end, int base, } static u8 cmp_extend_encoding(afl_state_t *afl, struct cmp_header *h, - u64 pattern, u64 repl, u64 o_pattern, u32 idx, - u8 *orig_buf, u8 *buf, u32 len, u8 do_reverse, - u8 *status) { - - if (!buf) { FATAL("BUG: buf was NULL. Please report this.\n"); } + u64 pattern, u64 repl, u64 o_pattern, + u64 changed_val, u8 attr, u32 idx, u32 taint_len, + u8 *orig_buf, u8 *buf, u8 *cbuf, u32 len, + u8 do_reverse, u8 lvl, u8 *status) { + + // (void)(changed_val); // TODO + // we can use the information in changed_val to see if there is a + // computable i2s transformation. + // if (pattern != o_pattern && repl != changed_val) { + + // u64 in_diff = pattern - o_pattern, out_diff = repl - changed_val; + // if (in_diff != out_diff) { + + // switch(in_diff) { + + // detect uppercase <-> lowercase, base64, hex encoding, etc.: + // repl = reverse_transform(TYPE, pattern); + // } + // } + // } + // not 100% but would have a chance to be detected + + // fprintf(stderr, + // "Encode: %llx->%llx into %llx(<-%llx) at pos=%u " + // "taint_len=%u shape=%u attr=%u\n", + // o_pattern, pattern, repl, changed_val, idx, taint_len, + // h->shape + 1, attr); u64 *buf_64 = (u64 *)&buf[idx]; u32 *buf_32 = (u32 *)&buf[idx]; @@ -293,76 +507,215 @@ static u8 cmp_extend_encoding(afl_state_t *afl, struct cmp_header *h, u16 *o_buf_16 = (u16 *)&orig_buf[idx]; u8 * o_buf_8 = &orig_buf[idx]; - u32 its_len = len - idx; - // *status = 0; + u32 its_len = MIN(len - idx, taint_len); u8 * endptr; u8 use_num = 0, use_unum = 0; unsigned long long unum; long long num; - if (afl->queue_cur->is_ascii) { + // reverse atoi()/strnu?toll() is expensive, so we only to it in lvl == 3 + if (lvl & 4) { - endptr = buf_8; - if (strntoll(buf_8, len - idx, (char **)&endptr, 0, &num)) { + if (afl->queue_cur->is_ascii) { - if (!strntoull(buf_8, len - idx, (char **)&endptr, 0, &unum)) - use_unum = 1; + endptr = buf_8; + if (strntoll(buf_8, len - idx, (char **)&endptr, 0, &num)) { - } else + if (!strntoull(buf_8, len - idx, (char **)&endptr, 0, &unum)) + use_unum = 1; - use_num = 1; + } else - } + use_num = 1; - if (use_num && (u64)num == pattern) { + } - size_t old_len = endptr - buf_8; - size_t num_len = snprintf(NULL, 0, "%lld", num); +#ifdef _DEBUG + if (idx == 0) + fprintf(stderr, "ASCII is=%u use_num=%u use_unum=%u idx=%u %llx==%llx\n", + afl->queue_cur->is_ascii, use_num, use_unum, idx, num, pattern); +#endif - u8 *new_buf = afl_realloc((void **)&afl->out_scratch_buf, len + num_len); - if (unlikely(!new_buf)) { PFATAL("alloc"); } - memcpy(new_buf, buf, idx); + // num is likely not pattern as atoi("AAA") will be zero... + if (use_num && ((u64)num == pattern || !num)) { - snprintf(new_buf + idx, num_len, "%lld", num); - memcpy(new_buf + idx + num_len, buf_8 + old_len, len - idx - old_len); + u8 tmp_buf[32]; + size_t num_len = snprintf(tmp_buf, sizeof(tmp_buf), "%lld", repl); + size_t old_len = endptr - buf_8; - if (unlikely(its_fuzz(afl, new_buf, len, status))) { return 1; } + u8 *new_buf = afl_realloc((void **)&afl->out_scratch_buf, len + num_len); + if (unlikely(!new_buf)) { PFATAL("alloc"); } - } else if (use_unum && unum == pattern) { + memcpy(new_buf, buf, idx); + memcpy(new_buf + idx, tmp_buf, num_len); + memcpy(new_buf + idx + num_len, buf_8 + old_len, len - idx - old_len); - size_t old_len = endptr - buf_8; - size_t num_len = snprintf(NULL, 0, "%llu", unum); + if (new_buf[idx + num_len] >= '0' && new_buf[idx + num_len] <= '9') { + + new_buf[idx + num_len] = ' '; + + } - u8 *new_buf = afl_realloc((void **)&afl->out_scratch_buf, len + num_len); - if (unlikely(!new_buf)) { PFATAL("alloc"); } - memcpy(new_buf, buf, idx); + if (unlikely(its_fuzz(afl, new_buf, len, status))) { return 1; } - snprintf(new_buf + idx, num_len, "%llu", unum); - memcpy(new_buf + idx + num_len, buf_8 + old_len, len - idx - old_len); + } else if (use_unum && (unum == pattern || !unum)) { - if (unlikely(its_fuzz(afl, new_buf, len, status))) { return 1; } + u8 tmp_buf[32]; + size_t num_len = snprintf(tmp_buf, sizeof(tmp_buf), "%llu", repl); + size_t old_len = endptr - buf_8; + + u8 *new_buf = afl_realloc((void **)&afl->out_scratch_buf, len + num_len); + if (unlikely(!new_buf)) { PFATAL("alloc"); } + + memcpy(new_buf, buf, idx); + memcpy(new_buf + idx, tmp_buf, num_len); + memcpy(new_buf + idx + num_len, buf_8 + old_len, len - idx - old_len); + + if (new_buf[idx + num_len] >= '0' && new_buf[idx + num_len] <= '9') { + + new_buf[idx + num_len] = ' '; + + } + + if (unlikely(its_fuzz(afl, new_buf, len, status))) { return 1; } + + } } - if (SHAPE_BYTES(h->shape) >= 8 && *status != 1) { + // we only allow this for ascii2integer (above) + if (unlikely(pattern == o_pattern)) { return 0; } - if (its_len >= 8 && *buf_64 == pattern && *o_buf_64 == o_pattern) { + if ((lvl & 1) || ((lvl & 2) && (attr >= 8 && attr <= 15)) || attr >= 16) { - *buf_64 = repl; - if (unlikely(its_fuzz(afl, buf, len, status))) { return 1; } - *buf_64 = pattern; + if (SHAPE_BYTES(h->shape) >= 8 && *status != 1) { + + // if (its_len >= 8 && (attr == 0 || attr >= 8)) + // fprintf(stderr, + // "TestU64: %u>=4 %x==%llx" + // " %x==%llx (idx=%u attr=%u) <= %llx<-%llx\n", + // its_len, *buf_32, pattern, *o_buf_32, o_pattern, idx, attr, + // repl, changed_val); + + // if this is an fcmp (attr & 8 == 8) then do not compare the patterns - + // due to a bug in llvm dynamic float bitcasts do not work :( + // the value 16 means this is a +- 1.0 test case + if (its_len >= 8 && + ((*buf_64 == pattern && *o_buf_64 == o_pattern) || attr >= 16)) { + + u64 tmp_64 = *buf_64; + *buf_64 = repl; + if (unlikely(its_fuzz(afl, buf, len, status))) { return 1; } + if (*status == 1) { memcpy(cbuf + idx, buf_64, 8); } + *buf_64 = tmp_64; + + // fprintf(stderr, "Status=%u\n", *status); + + } + + // reverse encoding + if (do_reverse && *status != 1) { + + if (unlikely(cmp_extend_encoding(afl, h, SWAP64(pattern), SWAP64(repl), + SWAP64(o_pattern), SWAP64(changed_val), + attr, idx, taint_len, orig_buf, buf, + cbuf, len, 0, lvl, status))) { + + return 1; + + } + + } } - // reverse encoding - if (do_reverse && *status != 1) { + if (SHAPE_BYTES(h->shape) >= 4 && *status != 1) { - if (unlikely(cmp_extend_encoding(afl, h, SWAP64(pattern), SWAP64(repl), - SWAP64(o_pattern), idx, orig_buf, buf, - len, 0, status))) { + // if (its_len >= 4 && (attr <= 1 || attr >= 8)) + // fprintf(stderr, + // "TestU32: %u>=4 %x==%llx" + // " %x==%llx (idx=%u attr=%u) <= %llx<-%llx\n", + // its_len, *buf_32, pattern, *o_buf_32, o_pattern, idx, attr, + // repl, changed_val); - return 1; + if (its_len >= 4 && + ((*buf_32 == (u32)pattern && *o_buf_32 == (u32)o_pattern) || + attr >= 16)) { + + u32 tmp_32 = *buf_32; + *buf_32 = (u32)repl; + if (unlikely(its_fuzz(afl, buf, len, status))) { return 1; } + if (*status == 1) { memcpy(cbuf + idx, buf_32, 4); } + *buf_32 = tmp_32; + + // fprintf(stderr, "Status=%u\n", *status); + + } + + // reverse encoding + if (do_reverse && *status != 1) { + + if (unlikely(cmp_extend_encoding(afl, h, SWAP32(pattern), SWAP32(repl), + SWAP32(o_pattern), SWAP32(changed_val), + attr, idx, taint_len, orig_buf, buf, + cbuf, len, 0, lvl, status))) { + + return 1; + + } + + } + + } + + if (SHAPE_BYTES(h->shape) >= 2 && *status != 1) { + + if (its_len >= 2 && + ((*buf_16 == (u16)pattern && *o_buf_16 == (u16)o_pattern) || + attr >= 16)) { + + u16 tmp_16 = *buf_16; + *buf_16 = (u16)repl; + if (unlikely(its_fuzz(afl, buf, len, status))) { return 1; } + if (*status == 1) { memcpy(cbuf + idx, buf_16, 2); } + *buf_16 = tmp_16; + + } + + // reverse encoding + if (do_reverse && *status != 1) { + + if (unlikely(cmp_extend_encoding(afl, h, SWAP16(pattern), SWAP16(repl), + SWAP16(o_pattern), SWAP16(changed_val), + attr, idx, taint_len, orig_buf, buf, + cbuf, len, 0, lvl, status))) { + + return 1; + + } + + } + + } + + if (*status != 1) { // u8 + + // if (its_len >= 1 && (attr <= 1 || attr >= 8)) + // fprintf(stderr, + // "TestU8: %u>=1 %x==%x %x==%x (idx=%u attr=%u) <= %x<-%x\n", + // its_len, *buf_8, pattern, *o_buf_8, o_pattern, idx, attr, + // repl, changed_val); + + if (its_len >= 1 && + ((*buf_8 == (u8)pattern && *o_buf_8 == (u8)o_pattern) || + attr >= 16)) { + + u8 tmp_8 = *buf_8; + *buf_8 = (u8)repl; + if (unlikely(its_fuzz(afl, buf, len, status))) { return 1; } + if (*status == 1) { cbuf[idx] = *buf_8; } + *buf_8 = tmp_8; } @@ -370,49 +723,205 @@ static u8 cmp_extend_encoding(afl_state_t *afl, struct cmp_header *h, } - if (SHAPE_BYTES(h->shape) >= 4 && *status != 1) { + // here we add and subract 1 from the value, but only if it is not an + // == or != comparison + // Bits: 1 = Equal, 2 = Greater, 3 = Lesser, 4 = Float - if (its_len >= 4 && *buf_32 == (u32)pattern && - *o_buf_32 == (u32)o_pattern) { + if (lvl < 4) { return 0; } - *buf_32 = (u32)repl; - if (unlikely(its_fuzz(afl, buf, len, status))) { return 1; } - *buf_32 = pattern; + if (attr >= 8 && attr < 16) { // lesser/greater integer comparison + + u64 repl_new; + if (SHAPE_BYTES(h->shape) == 4 && its_len >= 4) { + + float *f = (float *)&repl; + float g = *f; + g += 1.0; + u32 *r = (u32 *)&g; + repl_new = (u32)*r; + + } else if (SHAPE_BYTES(h->shape) == 8 && its_len >= 8) { + + double *f = (double *)&repl; + double g = *f; + g += 1.0; + + u64 *r = (u64 *)&g; + repl_new = *r; + + } else { + + return 0; } - // reverse encoding - if (do_reverse && *status != 1) { + changed_val = repl_new; + + if (unlikely(cmp_extend_encoding(afl, h, pattern, repl_new, o_pattern, + changed_val, 16, idx, taint_len, orig_buf, + buf, cbuf, len, 1, lvl, status))) { + + return 1; + + } + + if (SHAPE_BYTES(h->shape) == 4) { + + float *f = (float *)&repl; + float g = *f; + g -= 1.0; + u32 *r = (u32 *)&g; + repl_new = (u32)*r; + + } else if (SHAPE_BYTES(h->shape) == 8) { + + double *f = (double *)&repl; + double g = *f; + g -= 1.0; + u64 *r = (u64 *)&g; + repl_new = *r; + + } else { + + return 0; + + } + + changed_val = repl_new; + + if (unlikely(cmp_extend_encoding(afl, h, pattern, repl_new, o_pattern, + changed_val, 16, idx, taint_len, orig_buf, + buf, cbuf, len, 1, lvl, status))) { + + return 1; + + } + + // transform double to float, llvm likes to do that internally ... + if (SHAPE_BYTES(h->shape) == 8 && its_len >= 4) { + + double *f = (double *)&repl; + float g = (float)*f; + repl_new = 0; +#if (__BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__) + memcpy((char *)&repl_new, (char *)&g, 4); +#else + memcpy(((char *)&repl_new) + 4, (char *)&g, 4); +#endif + changed_val = repl_new; + h->shape = 3; // modify shape - if (unlikely(cmp_extend_encoding(afl, h, SWAP32(pattern), SWAP32(repl), - SWAP32(o_pattern), idx, orig_buf, buf, - len, 0, status))) { + // fprintf(stderr, "DOUBLE2FLOAT %llx\n", repl_new); + if (unlikely(cmp_extend_encoding( + afl, h, pattern, repl_new, o_pattern, changed_val, 16, idx, + taint_len, orig_buf, buf, cbuf, len, 1, lvl, status))) { + + h->shape = 7; return 1; } + h->shape = 7; // recover shape + + } + + } else if (attr > 1 && attr < 8) { // lesser/greater integer comparison + + u64 repl_new; + + repl_new = repl + 1; + changed_val = repl_new; + if (unlikely(cmp_extend_encoding(afl, h, pattern, repl_new, o_pattern, + changed_val, 32, idx, taint_len, orig_buf, + buf, cbuf, len, 1, lvl, status))) { + + return 1; + + } + + repl_new = repl - 1; + changed_val = repl_new; + if (unlikely(cmp_extend_encoding(afl, h, pattern, repl_new, o_pattern, + changed_val, 32, idx, taint_len, orig_buf, + buf, cbuf, len, 1, lvl, status))) { + + return 1; + } } - if (SHAPE_BYTES(h->shape) >= 2 && *status != 1) { + return 0; - if (its_len >= 2 && *buf_16 == (u16)pattern && - *o_buf_16 == (u16)o_pattern) { +} - *buf_16 = (u16)repl; +static u8 cmp_extend_encoding128(afl_state_t *afl, struct cmp_header *h, + u128 pattern, u128 repl, u128 o_pattern, + u128 changed_val, u8 attr, u32 idx, + u32 taint_len, u8 *orig_buf, u8 *buf, u8 *cbuf, + u32 len, u8 do_reverse, u8 lvl, u8 *status) { + + u128 *buf_128 = (u128 *)&buf[idx]; + u64 * buf0 = (u64 *)&buf[idx]; + u64 * buf1 = (u64 *)(buf + idx + 8); + u128 *o_buf_128 = (u128 *)&orig_buf[idx]; + u32 its_len = MIN(len - idx, taint_len); + u64 v10 = (u64)repl; + u64 v11 = (u64)(repl >> 64); + + // if this is an fcmp (attr & 8 == 8) then do not compare the patterns - + // due to a bug in llvm dynamic float bitcasts do not work :( + // the value 16 means this is a +- 1.0 test case + if (its_len >= 16) { + +#ifdef _DEBUG + fprintf(stderr, "TestU128: %u>=16 (idx=%u attr=%u) (%u)\n", its_len, idx, + attr, do_reverse); + u64 v00 = (u64)pattern; + u64 v01 = pattern >> 64; + u64 ov00 = (u64)o_pattern; + u64 ov01 = o_pattern >> 64; + u64 ov10 = (u64)changed_val; + u64 ov11 = changed_val >> 64; + u64 b00 = (u64)*buf_128; + u64 b01 = *buf_128 >> 64; + u64 ob00 = (u64)*o_buf_128; + u64 ob01 = *o_buf_128 >> 64; + fprintf(stderr, + "TestU128: %llx:%llx==%llx:%llx" + " %llx:%llx==%llx:%llx <= %llx:%llx<-%llx:%llx\n", + b01, b00, v01, v00, ob01, ob00, ov01, ov00, v11, v10, ov11, ov10); +#endif + + if (*buf_128 == pattern && *o_buf_128 == o_pattern) { + + u128 tmp_128 = *buf_128; + // *buf_128 = repl; <- this crashes +#if (__BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__) + *buf0 = v10; + *buf1 = v11; +#else + *buf1 = v10; + *buf0 = v11; +#endif if (unlikely(its_fuzz(afl, buf, len, status))) { return 1; } - *buf_16 = (u16)pattern; + if (*status == 1) { memcpy(cbuf + idx, buf_128, 16); } + *buf_128 = tmp_128; + +#ifdef _DEBUG + fprintf(stderr, "Status=%u\n", *status); +#endif } // reverse encoding if (do_reverse && *status != 1) { - if (unlikely(cmp_extend_encoding(afl, h, SWAP16(pattern), SWAP16(repl), - SWAP16(o_pattern), idx, orig_buf, buf, - len, 0, status))) { + if (unlikely(cmp_extend_encoding128( + afl, h, SWAPN(pattern, 128), SWAPN(repl, 128), + SWAPN(o_pattern, 128), SWAPN(changed_val, 128), attr, idx, + taint_len, orig_buf, buf, cbuf, len, 0, lvl, status))) { return 1; @@ -422,14 +931,82 @@ static u8 cmp_extend_encoding(afl_state_t *afl, struct cmp_header *h, } - /* avoid CodeQL warning on unsigned overflow */ - if (/* SHAPE_BYTES(h->shape) >= 1 && */ *status != 1) { + return 0; - if (its_len >= 1 && *buf_8 == (u8)pattern && *o_buf_8 == (u8)o_pattern) { +} - *buf_8 = (u8)repl; +// uh a pointer read from (long double*) reads 12 bytes, not 10 ... +// so lets make this complicated. +static u8 cmp_extend_encoding_ld(afl_state_t *afl, struct cmp_header *h, + u8 *pattern, u8 *repl, u8 *o_pattern, + u8 *changed_val, u8 attr, u32 idx, + u32 taint_len, u8 *orig_buf, u8 *buf, u8 *cbuf, + u32 len, u8 do_reverse, u8 lvl, u8 *status) { + + u8 *buf_ld = &buf[idx], *o_buf_ld = &orig_buf[idx], backup[10]; + u32 its_len = MIN(len - idx, taint_len); + + if (its_len >= 10) { + +#ifdef _DEBUG + fprintf(stderr, "TestUld: %u>=10 (len=%u idx=%u attr=%u) (%u)\n", its_len, + len, idx, attr, do_reverse); + fprintf(stderr, "TestUld: "); + u32 i; + for (i = 0; i < 10; i++) + fprintf(stderr, "%02x", pattern[i]); + fprintf(stderr, "=="); + for (i = 0; i < 10; i++) + fprintf(stderr, "%02x", buf_ld[i]); + fprintf(stderr, " "); + for (i = 0; i < 10; i++) + fprintf(stderr, "%02x", o_pattern[i]); + fprintf(stderr, "=="); + for (i = 0; i < 10; i++) + fprintf(stderr, "%02x", o_buf_ld[i]); + fprintf(stderr, " <= "); + for (i = 0; i < 10; i++) + fprintf(stderr, "%02x", repl[i]); + fprintf(stderr, "=="); + for (i = 0; i < 10; i++) + fprintf(stderr, "%02x", changed_val[i]); + fprintf(stderr, "\n"); +#endif + + if (!memcmp(pattern, buf_ld, 10) && !memcmp(o_pattern, o_buf_ld, 10)) { + + // if this is an fcmp (attr & 8 == 8) then do not compare the patterns - + // due to a bug in llvm dynamic float bitcasts do not work :( + // the value 16 means this is a +- 1.0 test case + + memcpy(backup, buf_ld, 10); + memcpy(buf_ld, repl, 10); if (unlikely(its_fuzz(afl, buf, len, status))) { return 1; } - *buf_8 = (u8)pattern; + if (*status == 1) { memcpy(cbuf + idx, repl, 10); } + memcpy(buf_ld, backup, 10); + +#ifdef _DEBUG + fprintf(stderr, "Status=%u\n", *status); +#endif + + } + + } + + // reverse encoding + if (do_reverse && *status != 1) { + + u8 sp[10], sr[10], osp[10], osr[10]; + SWAPNN(sp, pattern, 10); + SWAPNN(sr, repl, 10); + SWAPNN(osp, o_pattern, 10); + SWAPNN(osr, changed_val, 10); + + if (unlikely(cmp_extend_encoding_ld(afl, h, sp, sr, osp, osr, attr, idx, + taint_len, orig_buf, buf, cbuf, len, 0, + lvl, status))) { + + return 1; } @@ -445,10 +1022,6 @@ static void try_to_add_to_dict(afl_state_t *afl, u64 v, u8 shape) { u32 k; u8 cons_ff = 0, cons_0 = 0; - - if (shape > sizeof(v)) - FATAL("shape is greater than %zu, please report!", sizeof(v)); - for (k = 0; k < shape; ++k) { if (b[k] == 0) { @@ -457,7 +1030,7 @@ static void try_to_add_to_dict(afl_state_t *afl, u64 v, u8 shape) { } else if (b[k] == 0xff) { - ++cons_ff; + ++cons_0; } else { @@ -493,28 +1066,126 @@ static void try_to_add_to_dict(afl_state_t *afl, u64 v, u8 shape) { } -static u8 cmp_fuzz(afl_state_t *afl, u32 key, u8 *orig_buf, u8 *buf, u32 len) { +static void try_to_add_to_dict128(afl_state_t *afl, u128 v) { - struct cmp_header *h = &afl->shm.cmp_map->headers[key]; - u32 i, j, idx; + u8 *b = (u8 *)&v; - u32 loggeds = h->hits; + u32 k; + u8 cons_ff = 0, cons_0 = 0; + for (k = 0; k < 16; ++k) { + + if (b[k] == 0) { + + ++cons_0; + + } else if (b[k] == 0xff) { + + ++cons_0; + + } else { + + cons_0 = cons_ff = 0; + + } + + // too many uninteresting values? try adding 2 64-bit values + if (cons_0 > 6 || cons_ff > 6) { + + u64 v64 = (u64)v; + try_to_add_to_dict(afl, v64, 8); + v64 = (u64)(v >> 64); + try_to_add_to_dict(afl, v64, 8); + + return; + + } + + } + + maybe_add_auto(afl, (u8 *)&v, 16); + u128 rev = SWAPN(v, 128); + maybe_add_auto(afl, (u8 *)&rev, 16); + +} + +static void try_to_add_to_dictN(afl_state_t *afl, u128 v, u8 size) { + + u8 *b = (u8 *)&v; + + u32 k; + u8 cons_ff = 0, cons_0 = 0; +#if (__BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__) + for (k = 0; k < size; ++k) { + +#else + for (k = 16 - size; k < 16; ++k) { + +#endif + if (b[k] == 0) { + + ++cons_0; + + } else if (b[k] == 0xff) { + + ++cons_0; + + } else { + + cons_0 = cons_ff = 0; + + } + + } + + maybe_add_auto(afl, (u8 *)&v, size); + u128 rev = SWAPN(v, size); + maybe_add_auto(afl, (u8 *)&rev, size); + +} + +static u8 cmp_fuzz(afl_state_t *afl, u32 key, u8 *orig_buf, u8 *buf, u8 *cbuf, + u32 len, u32 lvl, struct tainted *taint) { + + struct cmp_header *h = &afl->shm.cmp_map->headers[key]; + struct tainted * t; + u32 i, j, idx, taint_len; + u32 have_taint = 1, is_128 = 0, is_n = 0, is_ld = 0; + u32 loggeds = h->hits; if (h->hits > CMP_MAP_H) { loggeds = CMP_MAP_H; } u8 status = 0; - // opt not in the paper - u32 fails; - u8 found_one = 0; + u8 found_one = 0; /* loop cmps are useless, detect and ignore them */ - u64 s_v0, s_v1; - u8 s_v0_fixed = 1, s_v1_fixed = 1; - u8 s_v0_inc = 1, s_v1_inc = 1; - u8 s_v0_dec = 1, s_v1_dec = 1; + u128 s128_v0 = 0, s128_v1 = 0, orig_s128_v0 = 0, orig_s128_v1 = 0; + long double ld0, ld1, o_ld0, o_ld1; + u64 s_v0, s_v1; + u8 s_v0_fixed = 1, s_v1_fixed = 1; + u8 s_v0_inc = 1, s_v1_inc = 1; + u8 s_v0_dec = 1, s_v1_dec = 1; - for (i = 0; i < loggeds; ++i) { + switch (SHAPE_BYTES(h->shape)) { + + case 1: + case 2: + case 4: + case 8: + break; + case 16: + is_128 = 1; + break; + case 10: + if (h->attribute & 8) { is_ld = 1; } + // fall through + default: + is_n = 1; - fails = 0; + } + + // FCmp not in if level 1 only + if ((h->attribute & 8) && lvl < 2) return 0; + + for (i = 0; i < loggeds; ++i) { struct cmp_operands *o = &afl->shm.cmp_map->log[key][i]; @@ -551,55 +1222,242 @@ static u8 cmp_fuzz(afl_state_t *afl, u32 key, u8 *orig_buf, u8 *buf, u32 len) { } - for (idx = 0; idx < len && fails < 8; ++idx) { +#ifdef _DEBUG + fprintf(stderr, "Handling: %llx->%llx vs %llx->%llx attr=%u shape=%u\n", + orig_o->v0, o->v0, orig_o->v1, o->v1, h->attribute, + SHAPE_BYTES(h->shape)); +#endif + + if (taint) { + + t = taint; + + while (t->next) { + + t = t->next; + + } + + } else { + + have_taint = 0; + t = NULL; + + } + + if (unlikely(is_128 || is_n)) { + + s128_v0 = ((u128)o->v0) + (((u128)o->v0_128) << 64); + s128_v1 = ((u128)o->v1) + (((u128)o->v1_128) << 64); + orig_s128_v0 = ((u128)orig_o->v0) + (((u128)orig_o->v0_128) << 64); + orig_s128_v1 = ((u128)orig_o->v1) + (((u128)orig_o->v1_128) << 64); + + if (is_ld) { + +#if (__BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__) + memcpy((char *)&ld0, (char *)&s128_v0, sizeof(long double)); + memcpy((char *)&ld1, (char *)&s128_v1, sizeof(long double)); + memcpy((char *)&o_ld0, (char *)&orig_s128_v0, sizeof(long double)); + memcpy((char *)&o_ld1, (char *)&orig_s128_v1, sizeof(long double)); +#else + memcpy((char *)&ld0, (char *)(&s128_v0) + 6, sizeof(long double)); + memcpy((char *)&ld1, (char *)(&s128_v1) + 6, sizeof(long double)); + memcpy((char *)&o_ld0, (char *)(&orig_s128_v0) + 6, + sizeof(long double)); + memcpy((char *)&o_ld1, (char *)(&orig_s128_v1) + 6, + sizeof(long double)); +#endif + + } + + } + + for (idx = 0; idx < len; ++idx) { + + if (have_taint) { + + if (!t || idx < t->pos) { + + continue; + + } else { + + taint_len = t->pos + t->len - idx; + + if (idx == t->pos + t->len - 1) { t = t->prev; } + + } + + } else { + + taint_len = len - idx; + + } status = 0; - if (unlikely(cmp_extend_encoding(afl, h, o->v0, o->v1, orig_o->v0, idx, - orig_buf, buf, len, 1, &status))) { - return 1; + if (is_ld) { // long double special case + + if (ld0 != o_ld0 && o_ld1 != o_ld0) { + + if (unlikely(cmp_extend_encoding_ld( + afl, h, (u8 *)&ld0, (u8 *)&ld1, (u8 *)&o_ld0, (u8 *)&o_ld1, + h->attribute, idx, taint_len, orig_buf, buf, cbuf, len, 1, + lvl, &status))) { + + return 1; + + } + + } + + if (status == 1) { + + found_one = 1; + break; + + } + + if (ld1 != o_ld1 && o_ld0 != o_ld1) { + + if (unlikely(cmp_extend_encoding_ld( + afl, h, (u8 *)&ld1, (u8 *)&ld0, (u8 *)&o_ld1, (u8 *)&o_ld0, + h->attribute, idx, taint_len, orig_buf, buf, cbuf, len, 1, + lvl, &status))) { + + return 1; + + } + + } + + if (status == 1) { + + found_one = 1; + break; + + } + + } + + if (is_128) { // u128 special case + + if (s128_v0 != orig_s128_v0 && orig_s128_v0 != orig_s128_v1) { + + if (unlikely(cmp_extend_encoding128( + afl, h, s128_v0, s128_v1, orig_s128_v0, orig_s128_v1, + h->attribute, idx, taint_len, orig_buf, buf, cbuf, len, 1, + lvl, &status))) { + + return 1; + + } + + } + + if (status == 1) { + + found_one = 1; + break; + + } + + if (s128_v1 != orig_s128_v1 && orig_s128_v1 != orig_s128_v0) { + + if (unlikely(cmp_extend_encoding128( + afl, h, s128_v1, s128_v0, orig_s128_v1, orig_s128_v0, + h->attribute, idx, taint_len, orig_buf, buf, cbuf, len, 1, + lvl, &status))) { + + return 1; + + } + + } + + if (status == 1) { + + found_one = 1; + break; + + } } - if (status == 2) { + // even for u128 and long double do cmp_extend_encoding() because + // if we got here their own special trials failed and it might just be + // a cast from e.g. u64 to u128 from the input data. - ++fails; + if ((o->v0 != orig_o->v0 || lvl >= 4) && orig_o->v0 != orig_o->v1) { - } else if (status == 1) { + if (unlikely(cmp_extend_encoding( + afl, h, o->v0, o->v1, orig_o->v0, orig_o->v1, h->attribute, idx, + taint_len, orig_buf, buf, cbuf, len, 1, lvl, &status))) { + return 1; + + } + + } + + if (status == 1) { + + found_one = 1; break; } status = 0; - if (unlikely(cmp_extend_encoding(afl, h, o->v1, o->v0, orig_o->v1, idx, - orig_buf, buf, len, 1, &status))) { + if ((o->v1 != orig_o->v1 || lvl >= 4) && orig_o->v0 != orig_o->v1) { - return 1; + if (unlikely(cmp_extend_encoding( + afl, h, o->v1, o->v0, orig_o->v1, orig_o->v0, h->attribute, idx, + taint_len, orig_buf, buf, cbuf, len, 1, lvl, &status))) { - } + return 1; - if (status == 2) { + } - ++fails; + } - } else if (status == 1) { + if (status == 1) { + found_one = 1; break; } } - if (status == 1) { found_one = 1; } +#ifdef _DEBUG + fprintf(stderr, + "END: %llx->%llx vs %llx->%llx attr=%u i=%u found=%u is128=%u " + "isN=%u size=%u\n", + orig_o->v0, o->v0, orig_o->v1, o->v1, h->attribute, i, found_one, + is_128, is_n, SHAPE_BYTES(h->shape)); +#endif // If failed, add to dictionary - if (fails == 8) { + if (!found_one) { if (afl->pass_stats[key].total == 0) { - try_to_add_to_dict(afl, o->v0, SHAPE_BYTES(h->shape)); - try_to_add_to_dict(afl, o->v1, SHAPE_BYTES(h->shape)); + if (unlikely(is_128)) { + + try_to_add_to_dict128(afl, s128_v0); + try_to_add_to_dict128(afl, s128_v1); + + } else if (unlikely(is_n)) { + + try_to_add_to_dictN(afl, s128_v0, SHAPE_BYTES(h->shape)); + try_to_add_to_dictN(afl, s128_v1, SHAPE_BYTES(h->shape)); + + } else { + + try_to_add_to_dict(afl, o->v0, SHAPE_BYTES(h->shape)); + try_to_add_to_dict(afl, o->v1, SHAPE_BYTES(h->shape)); + + } } @@ -630,20 +1488,19 @@ static u8 cmp_fuzz(afl_state_t *afl, u32 key, u8 *orig_buf, u8 *buf, u32 len) { } static u8 rtn_extend_encoding(afl_state_t *afl, u8 *pattern, u8 *repl, - u8 *o_pattern, u32 idx, u8 *orig_buf, u8 *buf, - u32 len, u8 *status) { + u8 *o_pattern, u32 idx, u32 taint_len, + u8 *orig_buf, u8 *buf, u8 *cbuf, u32 len, + u8 *status) { u32 i; u32 its_len = MIN((u32)32, len - idx); - + its_len = MIN(its_len, taint_len); u8 save[32]; memcpy(save, &buf[idx], its_len); - *status = 0; - for (i = 0; i < its_len; ++i) { - if (pattern[i] != buf[idx + i] || o_pattern[i] != orig_buf[idx + i] || + if ((pattern[i] != buf[idx + i] && o_pattern[i] != orig_buf[idx + i]) || *status == 1) { break; @@ -654,6 +1511,8 @@ static u8 rtn_extend_encoding(afl_state_t *afl, u8 *pattern, u8 *repl, if (unlikely(its_fuzz(afl, buf, len, status))) { return 1; } + if (*status == 1) { memcpy(cbuf + idx, &buf[idx], i); } + } memcpy(&buf[idx], save, i); @@ -661,23 +1520,21 @@ static u8 rtn_extend_encoding(afl_state_t *afl, u8 *pattern, u8 *repl, } -static u8 rtn_fuzz(afl_state_t *afl, u32 key, u8 *orig_buf, u8 *buf, u32 len) { +static u8 rtn_fuzz(afl_state_t *afl, u32 key, u8 *orig_buf, u8 *buf, u8 *cbuf, + u32 len, struct tainted *taint) { + struct tainted * t; struct cmp_header *h = &afl->shm.cmp_map->headers[key]; - u32 i, j, idx; + u32 i, j, idx, have_taint = 1, taint_len; u32 loggeds = h->hits; if (h->hits > CMP_MAP_RTN_H) { loggeds = CMP_MAP_RTN_H; } u8 status = 0; - // opt not in the paper - // u32 fails = 0; u8 found_one = 0; for (i = 0; i < loggeds; ++i) { - u32 fails = 0; - struct cmpfn_operands *o = &((struct cmpfn_operands *)afl->shm.cmp_map->log[key])[i]; @@ -696,50 +1553,84 @@ static u8 rtn_fuzz(afl_state_t *afl, u32 key, u8 *orig_buf, u8 *buf, u32 len) { } - for (idx = 0; idx < len && fails < 8; ++idx) { + if (taint) { - if (unlikely(rtn_extend_encoding(afl, o->v0, o->v1, orig_o->v0, idx, - orig_buf, buf, len, &status))) { + t = taint; + while (t->next) { - return 1; + t = t->next; } - if (status == 2) { + } else { - ++fails; + have_taint = 0; + t = NULL; - } else if (status == 1) { + } - break; + for (idx = 0; idx < len; ++idx) { + + if (have_taint) { + + if (!t || idx < t->pos) { + + continue; + + } else { + + taint_len = t->pos + t->len - idx; + + if (idx == t->pos + t->len - 1) { t = t->prev; } + + } + + } else { + + taint_len = len - idx; } - if (unlikely(rtn_extend_encoding(afl, o->v1, o->v0, orig_o->v1, idx, - orig_buf, buf, len, &status))) { + status = 0; + + if (unlikely(rtn_extend_encoding(afl, o->v0, o->v1, orig_o->v0, idx, + taint_len, orig_buf, buf, cbuf, len, + &status))) { return 1; } - if (status == 2) { + if (status == 1) { + + found_one = 1; + break; + + } + + status = 0; + + if (unlikely(rtn_extend_encoding(afl, o->v1, o->v0, orig_o->v1, idx, + taint_len, orig_buf, buf, cbuf, len, + &status))) { - ++fails; + return 1; - } else if (status == 1) { + } + if (status == 1) { + + found_one = 1; break; } } - if (status == 1) { found_one = 1; } - // If failed, add to dictionary - if (fails == 8) { + if (!found_one) { - if (afl->pass_stats[key].total == 0) { + if (unlikely(!afl->pass_stats[key].total)) { maybe_add_auto(afl, o->v0, SHAPE_BYTES(h->shape)); maybe_add_auto(afl, o->v1, SHAPE_BYTES(h->shape)); @@ -791,7 +1682,44 @@ u8 input_to_state_stage(afl_state_t *afl, u8 *orig_buf, u8 *buf, u32 len, memcpy(afl->orig_cmp_map, afl->shm.cmp_map, sizeof(struct cmp_map)); - if (unlikely(colorization(afl, buf, len, exec_cksum))) { return 1; } + struct tainted *taint = NULL; + + if (!afl->queue_cur->taint || !afl->queue_cur->cmplog_colorinput) { + + if (unlikely(colorization(afl, buf, len, exec_cksum, &taint))) { return 1; } + + // no taint? still try, create a dummy to prevent again colorization + if (!taint) { + + taint = ck_alloc(sizeof(struct tainted)); + taint->len = len; + + } + + } else { + + buf = afl->queue_cur->cmplog_colorinput; + taint = afl->queue_cur->taint; + // reget the cmplog information + if (unlikely(common_fuzz_cmplog_stuff(afl, buf, len))) { return 1; } + + } + +#ifdef _DEBUG + dump("ORIG", orig_buf, len); + dump("NEW ", buf, len); +#endif + + struct tainted *t = taint; + + while (t) { + +#ifdef _DEBUG + fprintf(stderr, "T: pos=%u len=%u\n", t->pos, t->len); +#endif + t = t->next; + + } // do it manually, forkserver clear only afl->fsrv.trace_bits memset(afl->shm.cmp_map->headers, 0, sizeof(afl->shm.cmp_map->headers)); @@ -807,15 +1735,38 @@ u8 input_to_state_stage(afl_state_t *afl, u8 *orig_buf, u8 *buf, u32 len, afl->stage_max = 0; afl->stage_cur = 0; + u32 lvl; + u32 cmplog_done = afl->queue_cur->colorized; + u32 cmplog_lvl = afl->cmplog_lvl; + if (!cmplog_done) { + + lvl = 1; + + } else { + + lvl = 0; + + } + + if (cmplog_lvl >= 2 && cmplog_done < 2) { lvl += 2; } + if (cmplog_lvl >= 3 && cmplog_done < 3) { lvl += 4; } + + u8 *cbuf = afl_realloc((void **)&afl->in_scratch_buf, len + 128); + memcpy(cbuf, orig_buf, len); + u8 *virgin_backup = afl_realloc((void **)&afl->ex_buf, afl->shm.map_size); + memcpy(virgin_backup, afl->virgin_bits, afl->shm.map_size); + u32 k; for (k = 0; k < CMP_MAP_W; ++k) { if (!afl->shm.cmp_map->headers[k].hits) { continue; } - if (afl->pass_stats[k].total && - (rand_below(afl, afl->pass_stats[k].total) >= - afl->pass_stats[k].faileds || - afl->pass_stats[k].total == 0xff)) { + if (afl->pass_stats[k].faileds == 0xff || + afl->pass_stats[k].total == 0xff) { + +#ifdef _DEBUG + fprintf(stderr, "DISABLED %u\n", k); +#endif afl->shm.cmp_map->headers[k].hits = 0; // ignore this cmp @@ -841,11 +1792,19 @@ u8 input_to_state_stage(afl_state_t *afl, u8 *orig_buf, u8 *buf, u32 len, if (afl->shm.cmp_map->headers[k].type == CMP_TYPE_INS) { - if (unlikely(cmp_fuzz(afl, k, orig_buf, buf, len))) { goto exit_its; } + if (unlikely(cmp_fuzz(afl, k, orig_buf, buf, cbuf, len, lvl, taint))) { + + goto exit_its; + + } } else { - if (unlikely(rtn_fuzz(afl, k, orig_buf, buf, len))) { goto exit_its; } + if (unlikely(rtn_fuzz(afl, k, orig_buf, buf, cbuf, len, taint))) { + + goto exit_its; + + } } @@ -854,12 +1813,86 @@ u8 input_to_state_stage(afl_state_t *afl, u8 *orig_buf, u8 *buf, u32 len, r = 0; exit_its: + + afl->queue_cur->colorized = afl->cmplog_lvl; + if (afl->cmplog_lvl == CMPLOG_LVL_MAX) { + + ck_free(afl->queue_cur->cmplog_colorinput); + t = taint; + while (taint) { + + t = taint->next; + ck_free(taint); + taint = t; + + } + + afl->queue_cur->taint = NULL; + + } else { + + if (!afl->queue_cur->taint) { afl->queue_cur->taint = taint; } + + if (!afl->queue_cur->cmplog_colorinput) { + + afl->queue_cur->cmplog_colorinput = ck_alloc_nozero(len); + memcpy(afl->queue_cur->cmplog_colorinput, buf, len); + memcpy(buf, orig_buf, len); + + } + + } + + // copy the current virgin bits so we can recover the information + u8 *virgin_save = afl_realloc((void **)&afl->eff_buf, afl->shm.map_size); + memcpy(virgin_save, afl->virgin_bits, afl->shm.map_size); + // reset virgin bits to the backup previous to redqueen + memcpy(afl->virgin_bits, virgin_backup, afl->shm.map_size); + + u8 status = 0; + its_fuzz(afl, cbuf, len, &status); + + // now combine with the saved virgin bits +#ifdef WORD_SIZE_64 + u64 *v = (u64 *)afl->virgin_bits; + u64 *s = (u64 *)virgin_save; + u32 i; + for (i = 0; i < (afl->shm.map_size >> 3); i++) { + + v[i] &= s[i]; + + } + +#else + u32 *v = (u64 *)afl->virgin_bits; + u32 *s = (u64 *)virgin_save; + u32 i; + for (i = 0; i < (afl->shm.map_size >> 2); i++) { + + v[i] &= s[i]; + + } + +#endif + +#ifdef _DEBUG + dump("COMB", cbuf, len); + if (status == 1) { + + fprintf(stderr, "NEW COMBINED\n"); + + } else { + + fprintf(stderr, "NO new combined\n"); + + } + +#endif + new_hit_cnt = afl->queued_paths + afl->unique_crashes; afl->stage_finds[STAGE_ITS] += new_hit_cnt - orig_hit_cnt; afl->stage_cycles[STAGE_ITS] += afl->fsrv.total_execs - orig_execs; - memcpy(buf, orig_buf, len); - return r; } diff --git a/src/afl-fuzz-state.c b/src/afl-fuzz-state.c index 60c9684c..8423a3d1 100644 --- a/src/afl-fuzz-state.c +++ b/src/afl-fuzz-state.c @@ -102,6 +102,7 @@ void afl_state_init(afl_state_t *afl, uint32_t map_size) { afl->stats_update_freq = 1; afl->stats_avg_exec = 0; afl->skip_deterministic = 1; + afl->cmplog_lvl = 1; #ifndef NO_SPLICING afl->use_splicing = 1; #endif diff --git a/src/afl-fuzz.c b/src/afl-fuzz.c index bb2674f0..1e914ca6 100644 --- a/src/afl-fuzz.c +++ b/src/afl-fuzz.c @@ -77,13 +77,8 @@ static void at_exit() { } int kill_signal = SIGKILL; - /* AFL_KILL_SIGNAL should already be a valid int at this point */ - if (getenv("AFL_KILL_SIGNAL")) { - - kill_signal = atoi(getenv("AFL_KILL_SIGNAL")); - - } + if ((ptr = getenv("AFL_KILL_SIGNAL"))) { kill_signal = atoi(ptr); } if (pid1 > 0) { kill(pid1, kill_signal); } if (pid2 > 0) { kill(pid2, kill_signal); } @@ -103,13 +98,14 @@ static void usage(u8 *argv0, int more_help) { "Execution control settings:\n" " -p schedule - power schedules compute a seed's performance score:\n" - " <fast(default), rare, exploit, seek, mmopt, coe, " - "explore,\n" - " lin, quad> -- see docs/power_schedules.md\n" + " fast(default), explore, exploit, seek, rare, mmopt, " + "coe, lin\n" + " quad -- see docs/power_schedules.md\n" " -f file - location read by the fuzzed program (default: stdin " "or @@)\n" " -t msec - timeout for each run (auto-scaled, 50-%u ms)\n" - " -m megs - memory limit for child process (%u MB, 0 = no limit)\n" + " -m megs - memory limit for child process (%u MB, 0 = no limit " + "[default])\n" " -Q - use binary-only instrumentation (QEMU mode)\n" " -U - use unicorn-based instrumentation (Unicorn mode)\n" " -W - use qemu-based instrumentation with Wine (Wine " @@ -125,7 +121,9 @@ static void usage(u8 *argv0, int more_help) { " See docs/README.MOpt.md\n" " -c program - enable CmpLog by specifying a binary compiled for " "it.\n" - " if using QEMU, just use -c 0.\n\n" + " if using QEMU, just use -c 0.\n" + " -l cmplog_level - set the complexity/intensivity of CmpLog.\n" + " Values: 1 (default), 2 (intensive) and 3 (heavy)\n\n" "Fuzzing behavior settings:\n" " -Z - sequential queue selection instead of weighted " @@ -337,7 +335,6 @@ int main(int argc, char **argv_orig, char **envp) { if (get_afl_env("AFL_DEBUG")) { debug = afl->debug = 1; } - // map_size = get_map_size(); afl_state_init(afl, map_size); afl->debug = debug; afl_fsrv_init(&afl->fsrv); @@ -358,7 +355,8 @@ int main(int argc, char **argv_orig, char **envp) { while ((opt = getopt( argc, argv, - "+b:c:i:I:o:f:F:m:t:T:dDnCB:S:M:x:QNUWe:p:s:V:E:L:hRP:Z")) > 0) { + "+b:B:c:CdDe:E:hi:I:f:F:l:L:m:M:nNo:p:P:RQs:S:t:T:UV:Wx:Z")) > + 0) { switch (opt) { @@ -787,6 +785,26 @@ int main(int argc, char **argv_orig, char **envp) { } break; + case 'l': { + + afl->cmplog_lvl = atoi(optarg); + if (afl->cmplog_lvl < 1 || afl->cmplog_lvl > CMPLOG_LVL_MAX) { + + FATAL( + "Bad complog level value, accepted values are 1 (default), 2 and " + "%u.", + CMPLOG_LVL_MAX); + + } + + if (afl->cmplog_lvl == CMPLOG_LVL_MAX) { + + afl->cmplog_max_filesize = MAX_FILE; + + } + + } break; + case 'L': { /* MOpt mode */ if (afl->limit_time_sig) { FATAL("Multiple -L options not supported"); } @@ -1635,6 +1653,14 @@ int main(int argc, char **argv_orig, char **envp) { if (afl->use_splicing) { ++afl->cycles_wo_finds; + + if (unlikely(afl->shm.cmplog_mode && + afl->cmplog_max_filesize < MAX_FILE)) { + + afl->cmplog_max_filesize <<= 4; + + } + switch (afl->expand_havoc) { case 0: @@ -1652,6 +1678,7 @@ int main(int argc, char **argv_orig, char **envp) { } afl->expand_havoc = 2; + if (afl->cmplog_lvl < 2) afl->cmplog_lvl = 2; break; case 2: // if (!have_p) afl->schedule = EXPLOIT; @@ -1665,11 +1692,14 @@ int main(int argc, char **argv_orig, char **envp) { afl->expand_havoc = 4; break; case 4: - // if not in sync mode, enable deterministic mode? - // if (!afl->sync_id) afl->skip_deterministic = 0; afl->expand_havoc = 5; + if (afl->cmplog_lvl < 3) afl->cmplog_lvl = 3; break; case 5: + // if not in sync mode, enable deterministic mode? + if (!afl->sync_id) afl->skip_deterministic = 0; + afl->expand_havoc = 6; + case 6: // nothing else currently break; |