#include "tsan_mman.h"
#include "tsan_suppressions.h"
#include "tsan_symbolize.h"
+#include "ubsan/ubsan_init.h"
+
+#ifdef __SSE3__
+// <emmintrin.h> transitively includes <stdlib.h>,
+// and it's prohibited to include std headers into tsan runtime.
+// So we do this dirty trick.
+#define _MM_MALLOC_H_INCLUDED
+#define __MM_MALLOC_H
+#include <emmintrin.h>
+typedef __m128i m128;
+#endif
volatile int __tsan_resumed = 0;
namespace __tsan {
-#ifndef TSAN_GO
+#if !defined(SANITIZER_GO) && !SANITIZER_MAC
THREADLOCAL char cur_thread_placeholder[sizeof(ThreadState)] ALIGNED(64);
#endif
static char ctx_placeholder[sizeof(Context)] ALIGNED(64);
static ThreadContextBase *CreateThreadContext(u32 tid) {
// Map thread trace when context is created.
- MapThreadTrace(GetThreadTrace(tid), TraceSize() * sizeof(Event));
- MapThreadTrace(GetThreadTraceHeader(tid), sizeof(Trace));
- new(ThreadTrace(tid)) Trace();
+ char name[50];
+ internal_snprintf(name, sizeof(name), "trace %u", tid);
+ MapThreadTrace(GetThreadTrace(tid), TraceSize() * sizeof(Event), name);
+ const uptr hdr = GetThreadTraceHeader(tid);
+ internal_snprintf(name, sizeof(name), "trace header %u", tid);
+ MapThreadTrace(hdr, sizeof(Trace), name);
+ new((void*)hdr) Trace();
+ // We are going to use only a small part of the trace with the default
+ // value of history_size. However, the constructor writes to the whole trace.
+ // Unmap the unused part.
+ uptr hdr_end = hdr + sizeof(Trace);
+ hdr_end -= sizeof(TraceHeader) * (kTraceParts - TraceParts());
+ hdr_end = RoundUp(hdr_end, GetPageSizeCached());
+ if (hdr_end < hdr + sizeof(Trace))
+ UnmapOrDie((void*)hdr_end, hdr + sizeof(Trace) - hdr_end);
void *mem = internal_alloc(MBlockThreadContex, sizeof(ThreadContext));
return new(mem) ThreadContext(tid);
}
-#ifndef TSAN_GO
+#ifndef SANITIZER_GO
static const u32 kThreadQuarantineSize = 16;
#else
static const u32 kThreadQuarantineSize = 64;
, nmissed_expected()
, thread_registry(new(thread_registry_placeholder) ThreadRegistry(
CreateThreadContext, kMaxTid, kThreadQuarantineSize, kMaxTidReuse))
+ , racy_mtx(MutexTypeRacy, StatMtxRacy)
, racy_stacks(MBlockRacyStacks)
, racy_addresses(MBlockRacyAddresses)
+ , fired_suppressions_mtx(MutexTypeFired, StatMtxFired)
, fired_suppressions(8) {
}
// , ignore_reads_and_writes()
// , ignore_interceptors()
, clock(tid, reuse_count)
-#ifndef TSAN_GO
+#ifndef SANITIZER_GO
, jmp_bufs(MBlockJmpBuf)
#endif
, tid(tid)
, stk_size(stk_size)
, tls_addr(tls_addr)
, tls_size(tls_size)
-#ifndef TSAN_GO
+#ifndef SANITIZER_GO
, last_sleep_clock(tid)
#endif
{
}
+#ifndef SANITIZER_GO
static void MemoryProfiler(Context *ctx, fd_t fd, int i) {
uptr n_threads;
uptr n_running_threads;
ctx->thread_registry->GetNumberOfThreads(&n_threads, &n_running_threads);
InternalScopedBuffer<char> buf(4096);
- internal_snprintf(buf.data(), buf.size(), "%d: nthr=%d nlive=%d\n",
- i, n_threads, n_running_threads);
- internal_write(fd, buf.data(), internal_strlen(buf.data()));
- WriteMemoryProfile(buf.data(), buf.size());
- internal_write(fd, buf.data(), internal_strlen(buf.data()));
+ WriteMemoryProfile(buf.data(), buf.size(), n_threads, n_running_threads);
+ WriteToFile(fd, buf.data(), internal_strlen(buf.data()));
}
static void BackgroundThread(void *arg) {
-#ifndef TSAN_GO
// This is a non-initialized non-user thread, nothing to see here.
// We don't use ScopedIgnoreInterceptors, because we want ignores to be
// enabled even when the thread function exits (e.g. during pthread thread
// shutdown code).
cur_thread()->ignore_interceptors++;
-#endif
const u64 kMs2Ns = 1000 * 1000;
fd_t mprof_fd = kInvalidFd;
if (flags()->profile_memory && flags()->profile_memory[0]) {
- InternalScopedBuffer<char> filename(4096);
- internal_snprintf(filename.data(), filename.size(), "%s.%d",
- flags()->profile_memory, (int)internal_getpid());
- uptr openrv = OpenFile(filename.data(), true);
- if (internal_iserror(openrv)) {
- Printf("ThreadSanitizer: failed to open memory profile file '%s'\n",
- &filename[0]);
+ if (internal_strcmp(flags()->profile_memory, "stdout") == 0) {
+ mprof_fd = 1;
+ } else if (internal_strcmp(flags()->profile_memory, "stderr") == 0) {
+ mprof_fd = 2;
} else {
- mprof_fd = openrv;
+ InternalScopedString filename(kMaxPathLength);
+ filename.append("%s.%d", flags()->profile_memory, (int)internal_getpid());
+ fd_t fd = OpenFile(filename.data(), WrOnly);
+ if (fd == kInvalidFd) {
+ Printf("ThreadSanitizer: failed to open memory profile file '%s'\n",
+ &filename[0]);
+ } else {
+ mprof_fd = fd;
+ }
}
}
// Flush memory if requested.
if (flags()->flush_memory_ms > 0) {
if (last_flush + flags()->flush_memory_ms * kMs2Ns < now) {
- if (flags()->verbosity > 0)
- Printf("ThreadSanitizer: periodic memory flush\n");
+ VPrintf(1, "ThreadSanitizer: periodic memory flush\n");
FlushShadowMemory();
last_flush = NanoTime();
}
}
+ // GetRSS can be expensive on huge programs, so don't do it every 100ms.
if (flags()->memory_limit_mb > 0) {
uptr rss = GetRSS();
uptr limit = uptr(flags()->memory_limit_mb) << 20;
- if (flags()->verbosity > 0) {
- Printf("ThreadSanitizer: memory flush check"
- " RSS=%llu LAST=%llu LIMIT=%llu\n",
- (u64)rss>>20, (u64)last_rss>>20, (u64)limit>>20);
- }
+ VPrintf(1, "ThreadSanitizer: memory flush check"
+ " RSS=%llu LAST=%llu LIMIT=%llu\n",
+ (u64)rss >> 20, (u64)last_rss >> 20, (u64)limit >> 20);
if (2 * rss > limit + last_rss) {
- if (flags()->verbosity > 0)
- Printf("ThreadSanitizer: flushing memory due to RSS\n");
+ VPrintf(1, "ThreadSanitizer: flushing memory due to RSS\n");
FlushShadowMemory();
rss = GetRSS();
- if (flags()->verbosity > 0)
- Printf("ThreadSanitizer: memory flushed RSS=%llu\n", (u64)rss>>20);
+ VPrintf(1, "ThreadSanitizer: memory flushed RSS=%llu\n", (u64)rss>>20);
}
last_rss = rss;
}
if (mprof_fd != kInvalidFd)
MemoryProfiler(ctx, mprof_fd, i);
-#ifndef TSAN_GO
// Flush symbolizer cache if requested.
if (flags()->flush_symbolizer_ms > 0) {
u64 last = atomic_load(&ctx->last_symbolize_time_ns,
atomic_store(&ctx->last_symbolize_time_ns, 0, memory_order_relaxed);
}
}
-#endif
}
}
ctx->background_thread = internal_start_thread(&BackgroundThread, 0);
}
+#ifndef __mips__
static void StopBackgroundThread() {
atomic_store(&ctx->stop_background_thread, 1, memory_order_relaxed);
internal_join_thread(ctx->background_thread);
ctx->background_thread = 0;
}
+#endif
+#endif
void DontNeedShadowFor(uptr addr, uptr size) {
uptr shadow_beg = MemToShadow(addr);
// Global data is not 64K aligned, but there are no adjacent mappings,
// so we can get away with unaligned mapping.
// CHECK_EQ(addr, addr & ~((64 << 10) - 1)); // windows wants 64K alignment
- MmapFixedNoReserve(MemToShadow(addr), size * kShadowMultiplier);
+ MmapFixedNoReserve(MemToShadow(addr), size * kShadowMultiplier, "shadow");
+
+ // Meta shadow is 2:1, so tread carefully.
+ static bool data_mapped = false;
+ static uptr mapped_meta_end = 0;
+ uptr meta_begin = (uptr)MemToMeta(addr);
+ uptr meta_end = (uptr)MemToMeta(addr + size);
+ meta_begin = RoundDownTo(meta_begin, 64 << 10);
+ meta_end = RoundUpTo(meta_end, 64 << 10);
+ if (!data_mapped) {
+ // First call maps data+bss.
+ data_mapped = true;
+ MmapFixedNoReserve(meta_begin, meta_end - meta_begin, "meta shadow");
+ } else {
+ // Mapping continous heap.
+ // Windows wants 64K alignment.
+ meta_begin = RoundDownTo(meta_begin, 64 << 10);
+ meta_end = RoundUpTo(meta_end, 64 << 10);
+ if (meta_end <= mapped_meta_end)
+ return;
+ if (meta_begin < mapped_meta_end)
+ meta_begin = mapped_meta_end;
+ MmapFixedNoReserve(meta_begin, meta_end - meta_begin, "meta shadow");
+ mapped_meta_end = meta_end;
+ }
+ VPrintf(2, "mapped meta shadow for (%p-%p) at (%p-%p)\n",
+ addr, addr+size, meta_begin, meta_end);
}
-void MapThreadTrace(uptr addr, uptr size) {
+void MapThreadTrace(uptr addr, uptr size, const char *name) {
DPrintf("#0: Mapping trace at %p-%p(0x%zx)\n", addr, addr + size, size);
- CHECK_GE(addr, kTraceMemBegin);
- CHECK_LE(addr + size, kTraceMemBegin + kTraceMemSize);
+ CHECK_GE(addr, kTraceMemBeg);
+ CHECK_LE(addr + size, kTraceMemEnd);
CHECK_EQ(addr, addr & ~((64 << 10) - 1)); // windows wants 64K alignment
- uptr addr1 = (uptr)MmapFixedNoReserve(addr, size);
+ uptr addr1 = (uptr)MmapFixedNoReserve(addr, size, name);
if (addr1 != addr) {
Printf("FATAL: ThreadSanitizer can not mmap thread trace (%p/%p->%p)\n",
addr, size, addr1);
}
}
+static void CheckShadowMapping() {
+ for (uptr i = 0; i < ARRAY_SIZE(UserRegions); i += 2) {
+ const uptr beg = UserRegions[i];
+ const uptr end = UserRegions[i + 1];
+ VPrintf(3, "checking shadow region %p-%p\n", beg, end);
+ for (uptr p0 = beg; p0 <= end; p0 += (end - beg) / 4) {
+ for (int x = -1; x <= 1; x++) {
+ const uptr p = p0 + x;
+ if (p < beg || p >= end)
+ continue;
+ const uptr s = MemToShadow(p);
+ const uptr m = (uptr)MemToMeta(p);
+ VPrintf(3, " checking pointer %p: shadow=%p meta=%p\n", p, s, m);
+ CHECK(IsAppMem(p));
+ CHECK(IsShadowMem(s));
+ CHECK_EQ(p & ~(kShadowCell - 1), ShadowToMem(s));
+ CHECK(IsMetaMem(m));
+ }
+ }
+ }
+}
+
void Initialize(ThreadState *thr) {
// Thread safe because done before all threads exist.
static bool is_initialized = false;
// Install tool-specific callbacks in sanitizer_common.
SetCheckFailedCallback(TsanCheckFailed);
-#ifndef TSAN_GO
+ ctx = new(ctx_placeholder) Context;
+ const char *options = GetEnv(kTsanOptionsEnv);
+ CacheBinaryName();
+ InitializeFlags(&ctx->flags, options);
+ CheckVMASize();
+#ifndef SANITIZER_GO
InitializeAllocator();
+ ReplaceSystemMalloc();
#endif
InitializeInterceptors();
- const char *env = InitializePlatform();
+ CheckShadowMapping();
+ InitializePlatform();
InitializeMutex();
InitializeDynamicAnnotations();
- ctx = new(ctx_placeholder) Context;
-#ifndef TSAN_GO
+#ifndef SANITIZER_GO
InitializeShadowMemory();
#endif
- InitializeFlags(&ctx->flags, env);
// Setup correct file descriptor for error reports.
- __sanitizer_set_report_path(flags()->log_path);
+ __sanitizer_set_report_path(common_flags()->log_path);
InitializeSuppressions();
-#ifndef TSAN_GO
+#ifndef SANITIZER_GO
InitializeLibIgnore();
- Symbolizer::Init(common_flags()->external_symbolizer_path);
- Symbolizer::Get()->AddHooks(EnterSymbolizer, ExitSymbolizer);
-#endif
+ Symbolizer::GetOrInit()->AddHooks(EnterSymbolizer, ExitSymbolizer);
+ // On MIPS, TSan initialization is run before
+ // __pthread_initialize_minimal_internal() is finished, so we can not spawn
+ // new threads.
+#ifndef __mips__
StartBackgroundThread();
SetSandboxingCallback(StopBackgroundThread);
- if (flags()->detect_deadlocks)
+#endif
+#endif
+ if (common_flags()->detect_deadlocks)
ctx->dd = DDetector::Create(flags());
- if (ctx->flags.verbosity)
- Printf("***** Running under ThreadSanitizer v2 (pid %d) *****\n",
- (int)internal_getpid());
+ VPrintf(1, "***** Running under ThreadSanitizer v2 (pid %d) *****\n",
+ (int)internal_getpid());
// Initialize thread 0.
int tid = ThreadCreate(thr, 0, 0, true);
CHECK_EQ(tid, 0);
ThreadStart(thr, tid, internal_getpid());
+#if TSAN_CONTAINS_UBSAN
+ __ubsan::InitAsPlugin();
+#endif
ctx->initialized = true;
if (flags()->stop_on_start) {
}
int Finalize(ThreadState *thr) {
- Context *ctx = __tsan::ctx;
bool failed = false;
if (flags()->atexit_sleep_ms > 0 && ThreadCount(thr) > 1)
CommonSanitizerReportMutex.Unlock();
ctx->report_mtx.Unlock();
-#ifndef TSAN_GO
- if (ctx->flags.verbosity)
- AllocatorPrintStats();
+#ifndef SANITIZER_GO
+ if (Verbosity()) AllocatorPrintStats();
#endif
ThreadFinalize(thr);
if (ctx->nreported) {
failed = true;
-#ifndef TSAN_GO
+#ifndef SANITIZER_GO
Printf("ThreadSanitizer: reported %d warnings\n", ctx->nreported);
#else
Printf("Found %d data race(s)\n", ctx->nreported);
ctx->nmissed_expected);
}
- if (flags()->print_suppressions)
+ if (common_flags()->print_suppressions)
PrintMatchedSuppressions();
-#ifndef TSAN_GO
+#ifndef SANITIZER_GO
if (flags()->print_benign)
PrintMatchedBenignRaces();
#endif
failed = OnFinalize(failed);
+#if TSAN_COLLECT_STATS
StatAggregate(ctx->stat, thr->stat);
StatOutput(ctx->stat);
- return failed ? flags()->exitcode : 0;
+#endif
+
+ return failed ? common_flags()->exitcode : 0;
}
-#ifndef TSAN_GO
+#ifndef SANITIZER_GO
void ForkBefore(ThreadState *thr, uptr pc) {
ctx->thread_registry->Lock();
ctx->report_mtx.Lock();
VPrintf(1, "ThreadSanitizer: forked new process with pid %d,"
" parent had %d threads\n", (int)internal_getpid(), (int)nthread);
if (nthread == 1) {
- internal_start_thread(&BackgroundThread, 0);
+ StartBackgroundThread();
} else {
// We've just forked a multi-threaded process. We cannot reasonably function
// after that (some mutexes may be locked before fork). So just enable
}
#endif
+#ifdef SANITIZER_GO
+NOINLINE
+void GrowShadowStack(ThreadState *thr) {
+ const int sz = thr->shadow_stack_end - thr->shadow_stack;
+ const int newsz = 2 * sz;
+ uptr *newstack = (uptr*)internal_alloc(MBlockShadowStack,
+ newsz * sizeof(uptr));
+ internal_memcpy(newstack, thr->shadow_stack, sz * sizeof(uptr));
+ internal_free(thr->shadow_stack);
+ thr->shadow_stack = newstack;
+ thr->shadow_stack_pos = newstack + sz;
+ thr->shadow_stack_end = newstack + newsz;
+}
+#endif
+
u32 CurrentStackId(ThreadState *thr, uptr pc) {
- if (thr->shadow_stack_pos == 0) // May happen during bootstrap.
+ if (!thr->is_inited) // May happen during bootstrap.
return 0;
- if (pc) {
+ if (pc != 0) {
+#ifndef SANITIZER_GO
+ DCHECK_LT(thr->shadow_stack_pos, thr->shadow_stack_end);
+#else
+ if (thr->shadow_stack_pos == thr->shadow_stack_end)
+ GrowShadowStack(thr);
+#endif
thr->shadow_stack_pos[0] = pc;
thr->shadow_stack_pos++;
}
- u32 id = StackDepotPut(thr->shadow_stack,
- thr->shadow_stack_pos - thr->shadow_stack);
- if (pc)
+ u32 id = StackDepotPut(
+ StackTrace(thr->shadow_stack, thr->shadow_stack_pos - thr->shadow_stack));
+ if (pc != 0)
thr->shadow_stack_pos--;
return id;
}
unsigned trace = (thr->fast_state.epoch() / kTracePartSize) % TraceParts();
TraceHeader *hdr = &thr_trace->headers[trace];
hdr->epoch0 = thr->fast_state.epoch();
- hdr->stack0.ObtainCurrent(thr, 0);
+ ObtainCurrentStack(thr, 0, &hdr->stack0);
hdr->mset0 = thr->mset;
thr->nomalloc--;
}
return TraceSize() / kTracePartSize;
}
-#ifndef TSAN_GO
+#ifndef SANITIZER_GO
extern "C" void __tsan_trace_switch() {
TraceSwitch(cur_thread());
}
*s = 0;
}
-static inline void HandleRace(ThreadState *thr, u64 *shadow_mem,
+ALWAYS_INLINE
+void HandleRace(ThreadState *thr, u64 *shadow_mem,
Shadow cur, Shadow old) {
thr->racy_state[0] = cur.raw();
thr->racy_state[1] = old.raw();
thr->racy_shadow_addr = shadow_mem;
-#ifndef TSAN_GO
+#ifndef SANITIZER_GO
HACKY_CALL(__tsan_report_race);
#else
ReportRace(thr);
#endif
}
-static inline bool OldIsInSameSynchEpoch(Shadow old, ThreadState *thr) {
- return old.epoch() >= thr->fast_synch_epoch;
-}
-
static inline bool HappensBefore(Shadow old, ThreadState *thr) {
return thr->clock.get(old.TidWithIgnore()) >= old.epoch();
}
-ALWAYS_INLINE USED
-void MemoryAccessImpl(ThreadState *thr, uptr addr,
+ALWAYS_INLINE
+void MemoryAccessImpl1(ThreadState *thr, uptr addr,
int kAccessSizeLog, bool kAccessIsWrite, bool kIsAtomic,
u64 *shadow_mem, Shadow cur) {
StatInc(thr, StatMop);
// it's just not worth it (performance- and complexity-wise).
Shadow old(0);
- if (kShadowCnt == 1) {
- int idx = 0;
-#include "tsan_update_shadow_word_inl.h"
- } else if (kShadowCnt == 2) {
- int idx = 0;
-#include "tsan_update_shadow_word_inl.h"
- idx = 1;
-#include "tsan_update_shadow_word_inl.h"
- } else if (kShadowCnt == 4) {
- int idx = 0;
-#include "tsan_update_shadow_word_inl.h"
- idx = 1;
-#include "tsan_update_shadow_word_inl.h"
- idx = 2;
-#include "tsan_update_shadow_word_inl.h"
- idx = 3;
-#include "tsan_update_shadow_word_inl.h"
- } else if (kShadowCnt == 8) {
- int idx = 0;
-#include "tsan_update_shadow_word_inl.h"
- idx = 1;
-#include "tsan_update_shadow_word_inl.h"
- idx = 2;
-#include "tsan_update_shadow_word_inl.h"
- idx = 3;
+
+ // It release mode we manually unroll the loop,
+ // because empirically gcc generates better code this way.
+ // However, we can't afford unrolling in debug mode, because the function
+ // consumes almost 4K of stack. Gtest gives only 4K of stack to death test
+ // threads, which is not enough for the unrolled loop.
+#if SANITIZER_DEBUG
+ for (int idx = 0; idx < 4; idx++) {
#include "tsan_update_shadow_word_inl.h"
- idx = 4;
+ }
+#else
+ int idx = 0;
#include "tsan_update_shadow_word_inl.h"
- idx = 5;
+ idx = 1;
#include "tsan_update_shadow_word_inl.h"
- idx = 6;
+ idx = 2;
#include "tsan_update_shadow_word_inl.h"
- idx = 7;
+ idx = 3;
#include "tsan_update_shadow_word_inl.h"
- } else {
- CHECK(false);
- }
+#endif
// we did not find any races and had already stored
// the current access info, so we are done
while (size) {
int size1 = 1;
int kAccessSizeLog = kSizeLog1;
- if (size >= 8 && (addr & ~7) == ((addr + 8) & ~7)) {
+ if (size >= 8 && (addr & ~7) == ((addr + 7) & ~7)) {
size1 = 8;
kAccessSizeLog = kSizeLog8;
- } else if (size >= 4 && (addr & ~7) == ((addr + 4) & ~7)) {
+ } else if (size >= 4 && (addr & ~7) == ((addr + 3) & ~7)) {
size1 = 4;
kAccessSizeLog = kSizeLog4;
- } else if (size >= 2 && (addr & ~7) == ((addr + 2) & ~7)) {
+ } else if (size >= 2 && (addr & ~7) == ((addr + 1) & ~7)) {
size1 = 2;
kAccessSizeLog = kSizeLog2;
}
}
}
+ALWAYS_INLINE
+bool ContainsSameAccessSlow(u64 *s, u64 a, u64 sync_epoch, bool is_write) {
+ Shadow cur(a);
+ for (uptr i = 0; i < kShadowCnt; i++) {
+ Shadow old(LoadShadow(&s[i]));
+ if (Shadow::Addr0AndSizeAreEqual(cur, old) &&
+ old.TidWithIgnore() == cur.TidWithIgnore() &&
+ old.epoch() > sync_epoch &&
+ old.IsAtomic() == cur.IsAtomic() &&
+ old.IsRead() <= cur.IsRead())
+ return true;
+ }
+ return false;
+}
+
+#if defined(__SSE3__)
+#define SHUF(v0, v1, i0, i1, i2, i3) _mm_castps_si128(_mm_shuffle_ps( \
+ _mm_castsi128_ps(v0), _mm_castsi128_ps(v1), \
+ (i0)*1 + (i1)*4 + (i2)*16 + (i3)*64))
+ALWAYS_INLINE
+bool ContainsSameAccessFast(u64 *s, u64 a, u64 sync_epoch, bool is_write) {
+ // This is an optimized version of ContainsSameAccessSlow.
+ // load current access into access[0:63]
+ const m128 access = _mm_cvtsi64_si128(a);
+ // duplicate high part of access in addr0:
+ // addr0[0:31] = access[32:63]
+ // addr0[32:63] = access[32:63]
+ // addr0[64:95] = access[32:63]
+ // addr0[96:127] = access[32:63]
+ const m128 addr0 = SHUF(access, access, 1, 1, 1, 1);
+ // load 4 shadow slots
+ const m128 shadow0 = _mm_load_si128((__m128i*)s);
+ const m128 shadow1 = _mm_load_si128((__m128i*)s + 1);
+ // load high parts of 4 shadow slots into addr_vect:
+ // addr_vect[0:31] = shadow0[32:63]
+ // addr_vect[32:63] = shadow0[96:127]
+ // addr_vect[64:95] = shadow1[32:63]
+ // addr_vect[96:127] = shadow1[96:127]
+ m128 addr_vect = SHUF(shadow0, shadow1, 1, 3, 1, 3);
+ if (!is_write) {
+ // set IsRead bit in addr_vect
+ const m128 rw_mask1 = _mm_cvtsi64_si128(1<<15);
+ const m128 rw_mask = SHUF(rw_mask1, rw_mask1, 0, 0, 0, 0);
+ addr_vect = _mm_or_si128(addr_vect, rw_mask);
+ }
+ // addr0 == addr_vect?
+ const m128 addr_res = _mm_cmpeq_epi32(addr0, addr_vect);
+ // epoch1[0:63] = sync_epoch
+ const m128 epoch1 = _mm_cvtsi64_si128(sync_epoch);
+ // epoch[0:31] = sync_epoch[0:31]
+ // epoch[32:63] = sync_epoch[0:31]
+ // epoch[64:95] = sync_epoch[0:31]
+ // epoch[96:127] = sync_epoch[0:31]
+ const m128 epoch = SHUF(epoch1, epoch1, 0, 0, 0, 0);
+ // load low parts of shadow cell epochs into epoch_vect:
+ // epoch_vect[0:31] = shadow0[0:31]
+ // epoch_vect[32:63] = shadow0[64:95]
+ // epoch_vect[64:95] = shadow1[0:31]
+ // epoch_vect[96:127] = shadow1[64:95]
+ const m128 epoch_vect = SHUF(shadow0, shadow1, 0, 2, 0, 2);
+ // epoch_vect >= sync_epoch?
+ const m128 epoch_res = _mm_cmpgt_epi32(epoch_vect, epoch);
+ // addr_res & epoch_res
+ const m128 res = _mm_and_si128(addr_res, epoch_res);
+ // mask[0] = res[7]
+ // mask[1] = res[15]
+ // ...
+ // mask[15] = res[127]
+ const int mask = _mm_movemask_epi8(res);
+ return mask != 0;
+}
+#endif
+
+ALWAYS_INLINE
+bool ContainsSameAccess(u64 *s, u64 a, u64 sync_epoch, bool is_write) {
+#if defined(__SSE3__)
+ bool res = ContainsSameAccessFast(s, a, sync_epoch, is_write);
+ // NOTE: this check can fail if the shadow is concurrently mutated
+ // by other threads. But it still can be useful if you modify
+ // ContainsSameAccessFast and want to ensure that it's not completely broken.
+ // DCHECK_EQ(res, ContainsSameAccessSlow(s, a, sync_epoch, is_write));
+ return res;
+#else
+ return ContainsSameAccessSlow(s, a, sync_epoch, is_write);
+#endif
+}
+
ALWAYS_INLINE USED
void MemoryAccess(ThreadState *thr, uptr pc, uptr addr,
int kAccessSizeLog, bool kAccessIsWrite, bool kIsAtomic) {
(int)(1 << kAccessSizeLog), kAccessIsWrite, shadow_mem,
(uptr)shadow_mem[0], (uptr)shadow_mem[1],
(uptr)shadow_mem[2], (uptr)shadow_mem[3]);
-#if TSAN_DEBUG
+#if SANITIZER_DEBUG
if (!IsAppMem(addr)) {
Printf("Access to non app mem %zx\n", addr);
DCHECK(IsAppMem(addr));
}
#endif
- if (*shadow_mem == kShadowRodata) {
+ if (kCppMode && *shadow_mem == kShadowRodata) {
// Access to .rodata section, no races here.
// Measurements show that it can be 10-20% of all memory accesses.
StatInc(thr, StatMop);
}
FastState fast_state = thr->fast_state;
- if (fast_state.GetIgnoreBit())
+ if (fast_state.GetIgnoreBit()) {
+ StatInc(thr, StatMop);
+ StatInc(thr, kAccessIsWrite ? StatMopWrite : StatMopRead);
+ StatInc(thr, (StatType)(StatMop1 + kAccessSizeLog));
+ StatInc(thr, StatMopIgnored);
return;
- fast_state.IncrementEpoch();
- thr->fast_state = fast_state;
+ }
+
Shadow cur(fast_state);
cur.SetAddr0AndSizeLog(addr & 7, kAccessSizeLog);
cur.SetWrite(kAccessIsWrite);
cur.SetAtomic(kIsAtomic);
- // We must not store to the trace if we do not store to the shadow.
- // That is, this call must be moved somewhere below.
- TraceAddEvent(thr, fast_state, EventTypeMop, pc);
+ if (LIKELY(ContainsSameAccess(shadow_mem, cur.raw(),
+ thr->fast_synch_epoch, kAccessIsWrite))) {
+ StatInc(thr, StatMop);
+ StatInc(thr, kAccessIsWrite ? StatMopWrite : StatMopRead);
+ StatInc(thr, (StatType)(StatMop1 + kAccessSizeLog));
+ StatInc(thr, StatMopSame);
+ return;
+ }
+
+ if (kCollectHistory) {
+ fast_state.IncrementEpoch();
+ thr->fast_state = fast_state;
+ TraceAddEvent(thr, fast_state, EventTypeMop, pc);
+ cur.IncrementEpoch();
+ }
+
+ MemoryAccessImpl1(thr, addr, kAccessSizeLog, kAccessIsWrite, kIsAtomic,
+ shadow_mem, cur);
+}
+
+// Called by MemoryAccessRange in tsan_rtl_thread.cc
+ALWAYS_INLINE USED
+void MemoryAccessImpl(ThreadState *thr, uptr addr,
+ int kAccessSizeLog, bool kAccessIsWrite, bool kIsAtomic,
+ u64 *shadow_mem, Shadow cur) {
+ if (LIKELY(ContainsSameAccess(shadow_mem, cur.raw(),
+ thr->fast_synch_epoch, kAccessIsWrite))) {
+ StatInc(thr, StatMop);
+ StatInc(thr, kAccessIsWrite ? StatMopWrite : StatMopRead);
+ StatInc(thr, (StatType)(StatMop1 + kAccessSizeLog));
+ StatInc(thr, StatMopSame);
+ return;
+ }
- MemoryAccessImpl(thr, addr, kAccessSizeLog, kAccessIsWrite, kIsAtomic,
+ MemoryAccessImpl1(thr, addr, kAccessSizeLog, kAccessIsWrite, kIsAtomic,
shadow_mem, cur);
}
}
} else {
// The region is big, reset only beginning and end.
- const uptr kPageSize = 4096;
+ const uptr kPageSize = GetPageSizeCached();
u64 *begin = (u64*)MemToShadow(addr);
u64 *end = begin + size / kShadowCell * kShadowCnt;
u64 *p = begin;
thr->is_freeing = true;
MemoryAccessRange(thr, pc, addr, size, true);
thr->is_freeing = false;
- thr->fast_state.IncrementEpoch();
- TraceAddEvent(thr, thr->fast_state, EventTypeMop, pc);
+ if (kCollectHistory) {
+ thr->fast_state.IncrementEpoch();
+ TraceAddEvent(thr, thr->fast_state, EventTypeMop, pc);
+ }
Shadow s(thr->fast_state);
s.ClearIgnoreBit();
s.MarkAsFreed();
}
void MemoryRangeImitateWrite(ThreadState *thr, uptr pc, uptr addr, uptr size) {
- thr->fast_state.IncrementEpoch();
- TraceAddEvent(thr, thr->fast_state, EventTypeMop, pc);
+ if (kCollectHistory) {
+ thr->fast_state.IncrementEpoch();
+ TraceAddEvent(thr, thr->fast_state, EventTypeMop, pc);
+ }
Shadow s(thr->fast_state);
s.ClearIgnoreBit();
s.SetWrite(true);
void FuncEntry(ThreadState *thr, uptr pc) {
StatInc(thr, StatFuncEnter);
DPrintf2("#%d: FuncEntry %p\n", (int)thr->fast_state.tid(), (void*)pc);
- thr->fast_state.IncrementEpoch();
- TraceAddEvent(thr, thr->fast_state, EventTypeFuncEnter, pc);
+ if (kCollectHistory) {
+ thr->fast_state.IncrementEpoch();
+ TraceAddEvent(thr, thr->fast_state, EventTypeFuncEnter, pc);
+ }
// Shadow stack maintenance can be replaced with
// stack unwinding during trace switch (which presumably must be faster).
DCHECK_GE(thr->shadow_stack_pos, thr->shadow_stack);
-#ifndef TSAN_GO
+#ifndef SANITIZER_GO
DCHECK_LT(thr->shadow_stack_pos, thr->shadow_stack_end);
#else
- if (thr->shadow_stack_pos == thr->shadow_stack_end) {
- const int sz = thr->shadow_stack_end - thr->shadow_stack;
- const int newsz = 2 * sz;
- uptr *newstack = (uptr*)internal_alloc(MBlockShadowStack,
- newsz * sizeof(uptr));
- internal_memcpy(newstack, thr->shadow_stack, sz * sizeof(uptr));
- internal_free(thr->shadow_stack);
- thr->shadow_stack = newstack;
- thr->shadow_stack_pos = newstack + sz;
- thr->shadow_stack_end = newstack + newsz;
- }
+ if (thr->shadow_stack_pos == thr->shadow_stack_end)
+ GrowShadowStack(thr);
#endif
thr->shadow_stack_pos[0] = pc;
thr->shadow_stack_pos++;
void FuncExit(ThreadState *thr) {
StatInc(thr, StatFuncExit);
DPrintf2("#%d: FuncExit\n", (int)thr->fast_state.tid());
- thr->fast_state.IncrementEpoch();
- TraceAddEvent(thr, thr->fast_state, EventTypeFuncExit, 0);
+ if (kCollectHistory) {
+ thr->fast_state.IncrementEpoch();
+ TraceAddEvent(thr, thr->fast_state, EventTypeFuncExit, 0);
+ }
DCHECK_GT(thr->shadow_stack_pos, thr->shadow_stack);
-#ifndef TSAN_GO
+#ifndef SANITIZER_GO
DCHECK_LT(thr->shadow_stack_pos, thr->shadow_stack_end);
#endif
thr->shadow_stack_pos--;
thr->ignore_reads_and_writes++;
CHECK_GT(thr->ignore_reads_and_writes, 0);
thr->fast_state.SetIgnoreBit();
-#ifndef TSAN_GO
+#ifndef SANITIZER_GO
if (!ctx->after_multithreaded_fork)
thr->mop_ignore_set.Add(CurrentStackId(thr, pc));
#endif
CHECK_GE(thr->ignore_reads_and_writes, 0);
if (thr->ignore_reads_and_writes == 0) {
thr->fast_state.ClearIgnoreBit();
-#ifndef TSAN_GO
+#ifndef SANITIZER_GO
thr->mop_ignore_set.Reset();
#endif
}
DPrintf("#%d: ThreadIgnoreSyncBegin\n", thr->tid);
thr->ignore_sync++;
CHECK_GT(thr->ignore_sync, 0);
-#ifndef TSAN_GO
+#ifndef SANITIZER_GO
if (!ctx->after_multithreaded_fork)
thr->sync_ignore_set.Add(CurrentStackId(thr, pc));
#endif
DPrintf("#%d: ThreadIgnoreSyncEnd\n", thr->tid);
thr->ignore_sync--;
CHECK_GE(thr->ignore_sync, 0);
-#ifndef TSAN_GO
+#ifndef SANITIZER_GO
if (thr->ignore_sync == 0)
thr->sync_ignore_set.Reset();
#endif
return hash[0] == other.hash[0] && hash[1] == other.hash[1];
}
-#if TSAN_DEBUG
+#if SANITIZER_DEBUG
void build_consistency_debug() {}
#else
void build_consistency_release() {}
void build_consistency_nostats() {}
#endif
-#if TSAN_SHADOW_COUNT == 1
-void build_consistency_shadow1() {}
-#elif TSAN_SHADOW_COUNT == 2
-void build_consistency_shadow2() {}
-#elif TSAN_SHADOW_COUNT == 4
-void build_consistency_shadow4() {}
-#else
-void build_consistency_shadow8() {}
-#endif
-
} // namespace __tsan
-#ifndef TSAN_GO
+#ifndef SANITIZER_GO
// Must be included in this file to make sure everything is inlined.
#include "tsan_interface_inl.h"
#endif