This is reincarnation of http://reviews.llvm.org/D17648 with the bug fix pointed out by Adhemerval (zatrazz).
Currently ThreadState holds both logical state (required for race-detection algorithm, user-visible)
and physical state (various caches, most notably malloc cache). Move physical state in a new
Process entity. Besides just being the right thing from abstraction point of view, this solves several
problems:
Cache everything on P level in Go. Currently we cache on a mix of goroutine and OS thread levels.
This unnecessary increases memory consumption.
Properly handle free operations in Go. Frees are issue by GC which don't have goroutine context.
As the result we could not do anything more than just clearing shadow. For example, we leaked
sync objects and heap block descriptors.
This will allow to get rid of libc malloc in Go (now we have Processor context for internal allocator cache).
This in turn will allow to get rid of dependency on libc entirely.
Potentially we can make Processor per-CPU in C++ mode instead of per-thread, which will
reduce resource consumption.
The distinction between Thread and Processor is currently used only by Go, C++ creates Processor per OS thread,
which is equivalent to the current scheme.
llvm-svn: 267678
rtl/tsan_rtl.cc
rtl/tsan_rtl_mutex.cc
rtl/tsan_rtl_report.cc
+ rtl/tsan_rtl_proc.cc
rtl/tsan_rtl_thread.cc
rtl/tsan_stack_trace.cc
rtl/tsan_stat.cc
-type tsan_go.cc ..\rtl\tsan_interface_atomic.cc ..\rtl\tsan_clock.cc ..\rtl\tsan_flags.cc ..\rtl\tsan_md5.cc ..\rtl\tsan_mutex.cc ..\rtl\tsan_report.cc ..\rtl\tsan_rtl.cc ..\rtl\tsan_rtl_mutex.cc ..\rtl\tsan_rtl_report.cc ..\rtl\tsan_rtl_thread.cc ..\rtl\tsan_stat.cc ..\rtl\tsan_suppressions.cc ..\rtl\tsan_sync.cc ..\rtl\tsan_stack_trace.cc ..\..\sanitizer_common\sanitizer_allocator.cc ..\..\sanitizer_common\sanitizer_common.cc ..\..\sanitizer_common\sanitizer_flags.cc ..\..\sanitizer_common\sanitizer_stacktrace.cc ..\..\sanitizer_common\sanitizer_libc.cc ..\..\sanitizer_common\sanitizer_printf.cc ..\..\sanitizer_common\sanitizer_suppressions.cc ..\..\sanitizer_common\sanitizer_thread_registry.cc ..\rtl\tsan_platform_windows.cc ..\..\sanitizer_common\sanitizer_win.cc ..\..\sanitizer_common\sanitizer_deadlock_detector1.cc ..\..\sanitizer_common\sanitizer_stackdepot.cc ..\..\sanitizer_common\sanitizer_persistent_allocator.cc ..\..\sanitizer_common\sanitizer_flag_parser.cc ..\..\sanitizer_common\sanitizer_symbolizer.cc > gotsan.cc
+type tsan_go.cc ..\rtl\tsan_interface_atomic.cc ..\rtl\tsan_clock.cc ..\rtl\tsan_flags.cc ..\rtl\tsan_md5.cc ..\rtl\tsan_mutex.cc ..\rtl\tsan_report.cc ..\rtl\tsan_rtl.cc ..\rtl\tsan_rtl_mutex.cc ..\rtl\tsan_rtl_report.cc ..\rtl\tsan_rtl_thread.cc ..\rtl\tsan_rtl_proc.cc ..\rtl\tsan_stat.cc ..\rtl\tsan_suppressions.cc ..\rtl\tsan_sync.cc ..\rtl\tsan_stack_trace.cc ..\..\sanitizer_common\sanitizer_allocator.cc ..\..\sanitizer_common\sanitizer_common.cc ..\..\sanitizer_common\sanitizer_flags.cc ..\..\sanitizer_common\sanitizer_stacktrace.cc ..\..\sanitizer_common\sanitizer_libc.cc ..\..\sanitizer_common\sanitizer_printf.cc ..\..\sanitizer_common\sanitizer_suppressions.cc ..\..\sanitizer_common\sanitizer_thread_registry.cc ..\rtl\tsan_platform_windows.cc ..\..\sanitizer_common\sanitizer_win.cc ..\..\sanitizer_common\sanitizer_deadlock_detector1.cc ..\..\sanitizer_common\sanitizer_stackdepot.cc ..\..\sanitizer_common\sanitizer_persistent_allocator.cc ..\..\sanitizer_common\sanitizer_flag_parser.cc ..\..\sanitizer_common\sanitizer_symbolizer.cc > gotsan.cc
gcc -c -o race_windows_amd64.syso gotsan.cc -I..\rtl -I..\.. -I..\..\sanitizer_common -I..\..\..\include -m64 -Wall -fno-exceptions -fno-rtti -DSANITIZER_GO -Wno-error=attributes -Wno-attributes -Wno-format -Wno-maybe-uninitialized -DSANITIZER_DEBUG=0 -O3 -fomit-frame-pointer -std=c++11
../rtl/tsan_rtl_mutex.cc
../rtl/tsan_rtl_report.cc
../rtl/tsan_rtl_thread.cc
+ ../rtl/tsan_rtl_proc.cc
../rtl/tsan_stack_trace.cc
../rtl/tsan_stat.cc
../rtl/tsan_suppressions.cc
fi
$CC $DIR/gotsan.cc -c -o $DIR/race_$SUFFIX.syso $FLAGS $CFLAGS
-$CC $OSCFLAGS test.c $DIR/race_$SUFFIX.syso -m64 -o $DIR/test $OSLDFLAGS
+$CC $OSCFLAGS test.c $DIR/race_$SUFFIX.syso -m64 -g -o $DIR/test $OSLDFLAGS
export GORACE="exitcode=0 atexit_sleep_ms=0"
if [ "$SILENT" != "1" ]; then
#include <stdio.h>
-void __tsan_init(void **thr, void (*cb)(long, void*));
+void __tsan_init(void **thr, void **proc, void (*cb)(long, void*));
void __tsan_fini();
void __tsan_map_shadow(void *addr, unsigned long size);
void __tsan_go_start(void *thr, void **chthr, void *pc);
void __tsan_go_end(void *thr);
+void __tsan_proc_create(void **pproc);
+void __tsan_proc_destroy(void *proc);
+void __tsan_proc_wire(void *proc, void *thr);
+void __tsan_proc_unwire(void *proc, void *thr);
void __tsan_read(void *thr, void *addr, void *pc);
void __tsan_write(void *thr, void *addr, void *pc);
void __tsan_func_enter(void *thr, void *pc);
void __tsan_func_exit(void *thr);
void __tsan_malloc(void *thr, void *pc, void *p, unsigned long sz);
-void __tsan_free(void *thr, void *p, unsigned long sz);
+void __tsan_free(void *proc, void *p, unsigned long sz);
void __tsan_acquire(void *thr, void *addr);
void __tsan_release(void *thr, void *addr);
void __tsan_release_merge(void *thr, void *addr);
int main(void) {
void *thr0 = 0;
- __tsan_init(&thr0, symbolize_cb);
+ void *proc0 = 0;
+ __tsan_init(&thr0, &proc0, symbolize_cb);
char *buf = (char*)((unsigned long)buf0 + (64<<10) - 1 & ~((64<<10) - 1));
__tsan_map_shadow(buf, 4096);
__tsan_malloc(thr0, (char*)&barfoo + 1, buf, 10);
+ __tsan_free(proc0, buf, 10);
__tsan_free(thr0, buf, 10);
__tsan_func_enter(thr0, (char*)&main + 1);
__tsan_malloc(thr0, (char*)&barfoo + 1, buf, 10);
__tsan_go_start(thr0, &thr1, (char*)&barfoo + 1);
void *thr2 = 0;
__tsan_go_start(thr0, &thr2, (char*)&barfoo + 1);
+ __tsan_func_exit(thr0);
+ __tsan_proc_unwire(proc0, thr0);
+ __tsan_proc_wire(proc0, thr1);
__tsan_func_enter(thr1, (char*)&foobar + 1);
__tsan_func_enter(thr1, (char*)&foobar + 1);
__tsan_write(thr1, buf, (char*)&barfoo + 1);
__tsan_func_exit(thr1);
__tsan_func_exit(thr1);
__tsan_go_end(thr1);
+ void *proc1 = 0;
+ __tsan_proc_create(&proc1);
+ __tsan_proc_wire(proc1, thr2);
__tsan_func_enter(thr2, (char*)&foobar + 1);
__tsan_read(thr2, buf, (char*)&barfoo + 1);
+ __tsan_free(proc1, buf, 10);
__tsan_func_exit(thr2);
__tsan_go_end(thr2);
- __tsan_func_exit(thr0);
+ __tsan_proc_destroy(proc0);
+ __tsan_proc_destroy(proc1);
__tsan_fini();
return 0;
}
return thr;
}
-void __tsan_init(ThreadState **thrp, void (*cb)(uptr cmd, void *cb)) {
+void __tsan_init(ThreadState **thrp, Processor **procp,
+ void (*cb)(uptr cmd, void *cb)) {
go_runtime_cb = cb;
ThreadState *thr = AllocGoroutine();
main_thr = *thrp = thr;
Initialize(thr);
+ *procp = thr->proc;
inited = true;
}
MemoryResetRange(0, 0, (uptr)p, sz);
}
-void __tsan_free(ThreadState *thr, uptr p, uptr sz) {
- if (thr)
- ctx->metamap.FreeRange(thr, 0, p, sz);
+void __tsan_free(Processor *proc, uptr p, uptr sz) {
+ ctx->metamap.FreeRange(proc, p, sz);
}
void __tsan_go_start(ThreadState *parent, ThreadState **pthr, void *pc) {
ThreadState *thr = AllocGoroutine();
*pthr = thr;
int goid = ThreadCreate(parent, (uptr)pc, 0, true);
+ Processor *proc = parent->proc;
+ // Temporary borrow proc to handle goroutine start.
+ ProcUnwire(proc, parent);
+ ProcWire(proc, thr);
ThreadStart(thr, goid, 0);
+ ProcUnwire(proc, thr);
+ ProcWire(proc, parent);
}
void __tsan_go_end(ThreadState *thr) {
+ Processor *proc = thr->proc;
ThreadFinish(thr);
+ ProcUnwire(proc, thr);
internal_free(thr);
}
+void __tsan_proc_create(Processor **pproc) {
+ *pproc = ProcCreate();
+}
+
+void __tsan_proc_destroy(Processor *proc) {
+ ProcDestroy(proc);
+}
+
+void __tsan_proc_wire(Processor *proc, ThreadState *thr) {
+ ProcWire(proc, thr);
+}
+
+void __tsan_proc_unwire(Processor *proc, ThreadState *thr) {
+ ProcUnwire(proc, thr);
+}
+
void __tsan_acquire(ThreadState *thr, void *addr) {
Acquire(thr, 0, (uptr)addr);
}
MD5Hash md5_hash(const void *data, uptr size);
+struct Processor;
struct ThreadState;
class ThreadContext;
struct Context;
if (sz != 0) {
// If sz == 0, munmap will return EINVAL and don't unmap any memory.
DontNeedShadowFor((uptr)addr, sz);
- ctx->metamap.ResetRange(thr, pc, (uptr)addr, (uptr)sz);
+ ctx->metamap.ResetRange(thr->proc, (uptr)addr, (uptr)sz);
}
int res = REAL(munmap)(addr, sz);
return res;
namespace __tsan {
void DestroyThreadState() {
ThreadState *thr = cur_thread();
+ Processor *proc = thr->proc;
ThreadFinish(thr);
+ ProcUnwire(proc, thr);
+ ProcDestroy(proc);
ThreadSignalContext *sctx = thr->signal_ctx;
if (sctx) {
thr->signal_ctx = 0;
#endif
while ((tid = atomic_load(&p->tid, memory_order_acquire)) == 0)
internal_sched_yield();
+ Processor *proc = ProcCreate();
+ ProcWire(proc, thr);
ThreadStart(thr, tid, GetTid());
atomic_store(&p->tid, 0, memory_order_release);
}
CHECK_GE(ptr, jctx->heap_begin);
CHECK_LE(ptr + size, jctx->heap_begin + jctx->heap_size);
- ctx->metamap.FreeRange(thr, pc, ptr, size);
+ ctx->metamap.FreeRange(thr->proc, ptr, size);
}
void __tsan_java_move(jptr src, jptr dst, jptr size) {
allocator()->Init(common_flags()->allocator_may_return_null);
}
-void AllocatorThreadStart(ThreadState *thr) {
- allocator()->InitCache(&thr->alloc_cache);
- internal_allocator()->InitCache(&thr->internal_alloc_cache);
+void AllocatorProcStart(Processor *proc) {
+ allocator()->InitCache(&proc->alloc_cache);
+ internal_allocator()->InitCache(&proc->internal_alloc_cache);
}
-void AllocatorThreadFinish(ThreadState *thr) {
- allocator()->DestroyCache(&thr->alloc_cache);
- internal_allocator()->DestroyCache(&thr->internal_alloc_cache);
+void AllocatorProcFinish(Processor *proc) {
+ allocator()->DestroyCache(&proc->alloc_cache);
+ internal_allocator()->DestroyCache(&proc->internal_alloc_cache);
}
void AllocatorPrintStats() {
void *user_alloc(ThreadState *thr, uptr pc, uptr sz, uptr align, bool signal) {
if ((sz >= (1ull << 40)) || (align >= (1ull << 40)))
return allocator()->ReturnNullOrDie();
- void *p = allocator()->Allocate(&thr->alloc_cache, sz, align);
+ void *p = allocator()->Allocate(&thr->proc->alloc_cache, sz, align);
if (p == 0)
return 0;
if (ctx && ctx->initialized)
void user_free(ThreadState *thr, uptr pc, void *p, bool signal) {
if (ctx && ctx->initialized)
OnUserFree(thr, pc, (uptr)p, true);
- allocator()->Deallocate(&thr->alloc_cache, p);
+ allocator()->Deallocate(&thr->proc->alloc_cache, p);
if (signal)
SignalUnsafeCall(thr, pc);
}
void OnUserFree(ThreadState *thr, uptr pc, uptr p, bool write) {
CHECK_NE(p, (void*)0);
- uptr sz = ctx->metamap.FreeBlock(thr, pc, p);
+ uptr sz = ctx->metamap.FreeBlock(thr->proc, p);
DPrintf("#%d: free(%p, %zu)\n", thr->tid, p, sz);
if (write && thr->ignore_reads_and_writes == 0)
MemoryRangeFreed(thr, pc, (uptr)p, sz);
thr->nomalloc = 0; // CHECK calls internal_malloc().
CHECK(0);
}
- return InternalAlloc(sz, &thr->internal_alloc_cache);
+ return InternalAlloc(sz, &thr->proc->internal_alloc_cache);
}
void internal_free(void *p) {
thr->nomalloc = 0; // CHECK calls internal_malloc().
CHECK(0);
}
- InternalFree(p, &thr->internal_alloc_cache);
+ InternalFree(p, &thr->proc->internal_alloc_cache);
}
} // namespace __tsan
void __tsan_on_thread_idle() {
ThreadState *thr = cur_thread();
- allocator()->SwallowCache(&thr->alloc_cache);
- internal_allocator()->SwallowCache(&thr->internal_alloc_cache);
- ctx->metamap.OnThreadIdle(thr);
+ allocator()->SwallowCache(&thr->proc->alloc_cache);
+ internal_allocator()->SwallowCache(&thr->proc->internal_alloc_cache);
+ ctx->metamap.OnProcIdle(thr->proc);
}
} // extern "C"
void InitializeAllocator();
void ReplaceSystemMalloc();
-void AllocatorThreadStart(ThreadState *thr);
-void AllocatorThreadFinish(ThreadState *thr);
+void AllocatorProcStart(Processor *proc);
+void AllocatorProcFinish(Processor *proc);
void AllocatorPrintStats();
// For user allocations.
InitializeAllocator();
ReplaceSystemMalloc();
#endif
+ if (common_flags()->detect_deadlocks)
+ ctx->dd = DDetector::Create(flags());
+ Processor *proc = ProcCreate();
+ ProcWire(proc, thr);
InitializeInterceptors();
CheckShadowMapping();
InitializePlatform();
SetSandboxingCallback(StopBackgroundThread);
#endif
#endif
- if (common_flags()->detect_deadlocks)
- ctx->dd = DDetector::Create(flags());
VPrintf(1, "***** Running under ThreadSanitizer v2 (pid %d) *****\n",
(int)internal_getpid());
uptr *shadow_stack_pos;
};
+// A Processor represents a physical thread, or a P for Go.
+// It is used to store internal resources like allocate cache, and does not
+// participate in race-detection logic (invisible to end user).
+// In C++ it is tied to an OS thread just like ThreadState, however ideally
+// it should be tied to a CPU (this way we will have fewer allocator caches).
+// In Go it is tied to a P, so there are significantly fewer Processor's than
+// ThreadState's (which are tied to Gs).
+// A ThreadState must be wired with a Processor to handle events.
+struct Processor {
+ ThreadState *thr; // currently wired thread, or nullptr
+#ifndef SANITIZER_GO
+ AllocatorCache alloc_cache;
+ InternalAllocatorCache internal_alloc_cache;
+#endif
+ DenseSlabAllocCache block_cache;
+ DenseSlabAllocCache sync_cache;
+ DenseSlabAllocCache clock_cache;
+ DDPhysicalThread *dd_pt;
+};
+
// This struct is stored in TLS.
struct ThreadState {
FastState fast_state;
MutexSet mset;
ThreadClock clock;
#ifndef SANITIZER_GO
- AllocatorCache alloc_cache;
- InternalAllocatorCache internal_alloc_cache;
Vector<JmpBuf> jmp_bufs;
int ignore_interceptors;
#endif
#if SANITIZER_DEBUG && !SANITIZER_GO
InternalDeadlockDetector internal_deadlock_detector;
#endif
- DDPhysicalThread *dd_pt;
DDLogicalThread *dd_lt;
+ // Current wired Processor, or nullptr. Required to handle any events.
+ Processor *proc;
+
atomic_uintptr_t in_signal_handler;
ThreadSignalContext *signal_ctx;
- DenseSlabAllocCache block_cache;
- DenseSlabAllocCache sync_cache;
- DenseSlabAllocCache clock_cache;
-
#ifndef SANITIZER_GO
u32 last_sleep_stack_id;
ThreadClock last_sleep_clock;
int ThreadCount(ThreadState *thr);
void ProcessPendingSignals(ThreadState *thr);
+Processor *ProcCreate();
+void ProcDestroy(Processor *proc);
+void ProcWire(Processor *proc, ThreadState *thr);
+void ProcUnwire(Processor *proc, ThreadState *thr);
+
void MutexCreate(ThreadState *thr, uptr pc, uptr addr,
bool rw, bool recursive, bool linker_init);
void MutexDestroy(ThreadState *thr, uptr pc, uptr addr);
Callback(ThreadState *thr, uptr pc)
: thr(thr)
, pc(pc) {
- DDCallback::pt = thr->dd_pt;
+ DDCallback::pt = thr->proc->dd_pt;
DDCallback::lt = thr->dd_lt;
}
u64 mid = s->GetId();
u32 last_lock = s->last_lock;
if (!unlock_locked)
- s->Reset(thr); // must not reset it before the report is printed
+ s->Reset(thr->proc); // must not reset it before the report is printed
s->mtx.Unlock();
if (unlock_locked) {
ThreadRegistryLock l(ctx->thread_registry);
if (unlock_locked) {
SyncVar *s = ctx->metamap.GetIfExistsAndLock(addr);
if (s != 0) {
- s->Reset(thr);
+ s->Reset(thr->proc);
s->mtx.Unlock();
}
}
if (thr->ignore_sync)
return;
thr->clock.set(thr->fast_state.epoch());
- thr->clock.acquire(&thr->clock_cache, c);
+ thr->clock.acquire(&thr->proc->clock_cache, c);
StatInc(thr, StatSyncAcquire);
}
return;
thr->clock.set(thr->fast_state.epoch());
thr->fast_synch_epoch = thr->fast_state.epoch();
- thr->clock.release(&thr->clock_cache, c);
+ thr->clock.release(&thr->proc->clock_cache, c);
StatInc(thr, StatSyncRelease);
}
return;
thr->clock.set(thr->fast_state.epoch());
thr->fast_synch_epoch = thr->fast_state.epoch();
- thr->clock.ReleaseStore(&thr->clock_cache, c);
+ thr->clock.ReleaseStore(&thr->proc->clock_cache, c);
StatInc(thr, StatSyncRelease);
}
return;
thr->clock.set(thr->fast_state.epoch());
thr->fast_synch_epoch = thr->fast_state.epoch();
- thr->clock.acq_rel(&thr->clock_cache, c);
+ thr->clock.acq_rel(&thr->proc->clock_cache, c);
StatInc(thr, StatSyncAcquire);
StatInc(thr, StatSyncRelease);
}
void ThreadContext::OnJoined(void *arg) {
ThreadState *caller_thr = static_cast<ThreadState *>(arg);
AcquireImpl(caller_thr, 0, &sync);
- sync.Reset(&caller_thr->clock_cache);
+ sync.Reset(&caller_thr->proc->clock_cache);
}
struct OnCreatedArgs {
void ThreadContext::OnDetached(void *arg) {
ThreadState *thr1 = static_cast<ThreadState*>(arg);
- sync.Reset(&thr1->clock_cache);
+ sync.Reset(&thr1->proc->clock_cache);
}
struct OnStartedArgs {
thr->shadow_stack_pos = thr->shadow_stack;
thr->shadow_stack_end = thr->shadow_stack + kInitStackSize;
#endif
-#ifndef SANITIZER_GO
- AllocatorThreadStart(thr);
-#endif
- if (common_flags()->detect_deadlocks) {
- thr->dd_pt = ctx->dd->CreatePhysicalThread();
+ if (common_flags()->detect_deadlocks)
thr->dd_lt = ctx->dd->CreateLogicalThread(unique_id);
- }
thr->fast_state.SetHistorySize(flags()->history_size);
// Commit switch to the new part of the trace.
// TraceAddEvent will reset stack0/mset0 in the new part for us.
thr->fast_synch_epoch = epoch0;
AcquireImpl(thr, 0, &sync);
StatInc(thr, StatSyncAcquire);
- sync.Reset(&thr->clock_cache);
+ sync.Reset(&thr->proc->clock_cache);
thr->is_inited = true;
DPrintf("#%d: ThreadStart epoch=%zu stk_addr=%zx stk_size=%zx "
"tls_addr=%zx tls_size=%zx\n",
}
epoch1 = thr->fast_state.epoch();
- if (common_flags()->detect_deadlocks) {
- ctx->dd->DestroyPhysicalThread(thr->dd_pt);
+ if (common_flags()->detect_deadlocks)
ctx->dd->DestroyLogicalThread(thr->dd_lt);
- }
- ctx->clock_alloc.FlushCache(&thr->clock_cache);
- ctx->metamap.OnThreadIdle(thr);
-#ifndef SANITIZER_GO
- AllocatorThreadFinish(thr);
-#endif
thr->~ThreadState();
#if TSAN_COLLECT_STATS
StatAggregate(ctx->stat, thr->stat);
DDMutexInit(thr, pc, this);
}
-void SyncVar::Reset(ThreadState *thr) {
+void SyncVar::Reset(Processor *proc) {
uid = 0;
creation_stack_id = 0;
owner_tid = kInvalidTid;
is_broken = 0;
is_linker_init = 0;
- if (thr == 0) {
+ if (proc == 0) {
CHECK_EQ(clock.size(), 0);
CHECK_EQ(read_clock.size(), 0);
} else {
- clock.Reset(&thr->clock_cache);
- read_clock.Reset(&thr->clock_cache);
+ clock.Reset(&proc->clock_cache);
+ read_clock.Reset(&proc->clock_cache);
}
}
}
void MetaMap::AllocBlock(ThreadState *thr, uptr pc, uptr p, uptr sz) {
- u32 idx = block_alloc_.Alloc(&thr->block_cache);
+ u32 idx = block_alloc_.Alloc(&thr->proc->block_cache);
MBlock *b = block_alloc_.Map(idx);
b->siz = sz;
b->tid = thr->tid;
*meta = idx | kFlagBlock;
}
-uptr MetaMap::FreeBlock(ThreadState *thr, uptr pc, uptr p) {
+uptr MetaMap::FreeBlock(Processor *proc, uptr p) {
MBlock* b = GetBlock(p);
if (b == 0)
return 0;
uptr sz = RoundUpTo(b->siz, kMetaShadowCell);
- FreeRange(thr, pc, p, sz);
+ FreeRange(proc, p, sz);
return sz;
}
-bool MetaMap::FreeRange(ThreadState *thr, uptr pc, uptr p, uptr sz) {
+bool MetaMap::FreeRange(Processor *proc, uptr p, uptr sz) {
bool has_something = false;
u32 *meta = MemToMeta(p);
u32 *end = MemToMeta(p + sz);
has_something = true;
while (idx != 0) {
if (idx & kFlagBlock) {
- block_alloc_.Free(&thr->block_cache, idx & ~kFlagMask);
+ block_alloc_.Free(&proc->block_cache, idx & ~kFlagMask);
break;
} else if (idx & kFlagSync) {
DCHECK(idx & kFlagSync);
SyncVar *s = sync_alloc_.Map(idx & ~kFlagMask);
u32 next = s->next;
- s->Reset(thr);
- sync_alloc_.Free(&thr->sync_cache, idx & ~kFlagMask);
+ s->Reset(proc);
+ sync_alloc_.Free(&proc->sync_cache, idx & ~kFlagMask);
idx = next;
} else {
CHECK(0);
// which can be huge. The function probes pages one-by-one until it finds a page
// without meta objects, at this point it stops freeing meta objects. Because
// thread stacks grow top-down, we do the same starting from end as well.
-void MetaMap::ResetRange(ThreadState *thr, uptr pc, uptr p, uptr sz) {
+void MetaMap::ResetRange(Processor *proc, uptr p, uptr sz) {
const uptr kMetaRatio = kMetaShadowCell / kMetaShadowSize;
const uptr kPageSize = GetPageSizeCached() * kMetaRatio;
if (sz <= 4 * kPageSize) {
// If the range is small, just do the normal free procedure.
- FreeRange(thr, pc, p, sz);
+ FreeRange(proc, p, sz);
return;
}
// First, round both ends of the range to page size.
uptr diff = RoundUp(p, kPageSize) - p;
if (diff != 0) {
- FreeRange(thr, pc, p, diff);
+ FreeRange(proc, p, diff);
p += diff;
sz -= diff;
}
diff = p + sz - RoundDown(p + sz, kPageSize);
if (diff != 0) {
- FreeRange(thr, pc, p + sz - diff, diff);
+ FreeRange(proc, p + sz - diff, diff);
sz -= diff;
}
// Now we must have a non-empty page-aligned range.
const uptr sz0 = sz;
// Probe start of the range.
while (sz > 0) {
- bool has_something = FreeRange(thr, pc, p, kPageSize);
+ bool has_something = FreeRange(proc, p, kPageSize);
p += kPageSize;
sz -= kPageSize;
if (!has_something)
}
// Probe end of the range.
while (sz > 0) {
- bool has_something = FreeRange(thr, pc, p - kPageSize, kPageSize);
+ bool has_something = FreeRange(proc, p - kPageSize, kPageSize);
sz -= kPageSize;
if (!has_something)
break;
SyncVar * s = sync_alloc_.Map(idx & ~kFlagMask);
if (s->addr == addr) {
if (myidx != 0) {
- mys->Reset(thr);
- sync_alloc_.Free(&thr->sync_cache, myidx);
+ mys->Reset(thr->proc);
+ sync_alloc_.Free(&thr->proc->sync_cache, myidx);
}
if (write_lock)
s->mtx.Lock();
if (myidx == 0) {
const u64 uid = atomic_fetch_add(&uid_gen_, 1, memory_order_relaxed);
- myidx = sync_alloc_.Alloc(&thr->sync_cache);
+ myidx = sync_alloc_.Alloc(&thr->proc->sync_cache);
mys = sync_alloc_.Map(myidx);
mys->Init(thr, pc, addr, uid);
}
}
}
-void MetaMap::OnThreadIdle(ThreadState *thr) {
- block_alloc_.FlushCache(&thr->block_cache);
- sync_alloc_.FlushCache(&thr->sync_cache);
+void MetaMap::OnProcIdle(Processor *proc) {
+ block_alloc_.FlushCache(&proc->block_cache);
+ sync_alloc_.FlushCache(&proc->sync_cache);
}
} // namespace __tsan
SyncClock clock;
void Init(ThreadState *thr, uptr pc, uptr addr, u64 uid);
- void Reset(ThreadState *thr);
+ void Reset(Processor *proc);
u64 GetId() const {
// 47 lsb is addr, then 14 bits is low part of uid, then 3 zero bits.
MetaMap();
void AllocBlock(ThreadState *thr, uptr pc, uptr p, uptr sz);
- uptr FreeBlock(ThreadState *thr, uptr pc, uptr p);
- bool FreeRange(ThreadState *thr, uptr pc, uptr p, uptr sz);
- void ResetRange(ThreadState *thr, uptr pc, uptr p, uptr sz);
+ uptr FreeBlock(Processor *proc, uptr p);
+ bool FreeRange(Processor *proc, uptr p, uptr sz);
+ void ResetRange(Processor *proc, uptr p, uptr sz);
MBlock* GetBlock(uptr p);
SyncVar* GetOrCreateAndLock(ThreadState *thr, uptr pc,
void MoveMemory(uptr src, uptr dst, uptr sz);
- void OnThreadIdle(ThreadState *thr);
+ void OnProcIdle(Processor *proc);
private:
static const u32 kFlagMask = 3u << 30;
EXPECT_NE(mb, (MBlock*)0);
EXPECT_EQ(mb->siz, 1 * sizeof(u64));
EXPECT_EQ(mb->tid, thr->tid);
- uptr sz = m->FreeBlock(thr, 0, (uptr)&block[0]);
+ uptr sz = m->FreeBlock(thr->proc, (uptr)&block[0]);
EXPECT_EQ(sz, 1 * sizeof(u64));
mb = m->GetBlock((uptr)&block[0]);
EXPECT_EQ(mb, (MBlock*)0);
EXPECT_EQ(mb1->siz, 1 * sizeof(u64));
MBlock *mb2 = m->GetBlock((uptr)&block[1]);
EXPECT_EQ(mb2->siz, 3 * sizeof(u64));
- m->FreeRange(thr, 0, (uptr)&block[0], 4 * sizeof(u64));
+ m->FreeRange(thr->proc, (uptr)&block[0], 4 * sizeof(u64));
mb1 = m->GetBlock((uptr)&block[0]);
EXPECT_EQ(mb1, (MBlock*)0);
mb2 = m->GetBlock((uptr)&block[1]);
EXPECT_NE(s2, (SyncVar*)0);
EXPECT_EQ(s2->addr, (uptr)&block[1]);
s2->mtx.ReadUnlock();
- m->FreeBlock(thr, 0, (uptr)&block[0]);
+ m->FreeBlock(thr->proc, (uptr)&block[0]);
s1 = m->GetIfExistsAndLock((uptr)&block[0]);
EXPECT_EQ(s1, (SyncVar*)0);
s2 = m->GetIfExistsAndLock((uptr)&block[1]);
EXPECT_EQ(s2, (SyncVar*)0);
- m->OnThreadIdle(thr);
+ m->OnProcIdle(thr->proc);
}
TEST(MetaMap, MoveMemory) {
EXPECT_NE(s2, (SyncVar*)0);
EXPECT_EQ(s2->addr, (uptr)&block2[1]);
s2->mtx.Unlock();
- m->FreeRange(thr, 0, (uptr)&block2[0], 4 * sizeof(u64));
+ m->FreeRange(thr->proc, (uptr)&block2[0], 4 * sizeof(u64));
}
TEST(MetaMap, ResetSync) {
u64 block[1] = {}; // fake malloc block
m->AllocBlock(thr, 0, (uptr)&block[0], 1 * sizeof(u64));
SyncVar *s = m->GetOrCreateAndLock(thr, 0, (uptr)&block[0], true);
- s->Reset(thr);
+ s->Reset(thr->proc);
s->mtx.Unlock();
- uptr sz = m->FreeBlock(thr, 0, (uptr)&block[0]);
+ uptr sz = m->FreeBlock(thr->proc, (uptr)&block[0]);
EXPECT_EQ(sz, 1 * sizeof(u64));
}