// Copyright 2009 The Go Authors. All rights reserved. // Use of this source code is governed by a BSD-style // license that can be found in the LICENSE file. // Malloc profiling. // Patterned after tcmalloc's algorithms; shorter code. package runtime import ( "internal/abi" "runtime/internal/atomic" "runtime/internal/sys" "unsafe" ) // NOTE(rsc): Everything here could use cas if contention became an issue. var ( // profInsertLock protects changes to the start of all *bucket linked lists profInsertLock mutex // profBlockLock protects the contents of every blockRecord struct profBlockLock mutex // profMemActiveLock protects the active field of every memRecord struct profMemActiveLock mutex // profMemFutureLock is a set of locks that protect the respective elements // of the future array of every memRecord struct profMemFutureLock [len(memRecord{}.future)]mutex ) // All memory allocations are local and do not escape outside of the profiler. // The profiler is forbidden from referring to garbage-collected memory. const ( // profile types memProfile bucketType = 1 + iota blockProfile mutexProfile // size of bucket hash table buckHashSize = 179999 // maxStack is the max depth of stack to record in bucket. // Note that it's only used internally as a guard against // wildly out-of-bounds slicing of the PCs that come after // a bucket struct, and it could increase in the future. maxStack = 32 ) type bucketType int // A bucket holds per-call-stack profiling information. // The representation is a bit sleazy, inherited from C. // This struct defines the bucket header. It is followed in // memory by the stack words and then the actual record // data, either a memRecord or a blockRecord. // // Per-call-stack profiling information. // Lookup by hashing call stack into a linked-list hash table. // // None of the fields in this bucket header are modified after // creation, including its next and allnext links. // // No heap pointers. type bucket struct { _ sys.NotInHeap next *bucket allnext *bucket typ bucketType // memBucket or blockBucket (includes mutexProfile) hash uintptr size uintptr nstk uintptr } // A memRecord is the bucket data for a bucket of type memProfile, // part of the memory profile. type memRecord struct { // The following complex 3-stage scheme of stats accumulation // is required to obtain a consistent picture of mallocs and frees // for some point in time. // The problem is that mallocs come in real time, while frees // come only after a GC during concurrent sweeping. So if we would // naively count them, we would get a skew toward mallocs. // // Hence, we delay information to get consistent snapshots as // of mark termination. Allocations count toward the next mark // termination's snapshot, while sweep frees count toward the // previous mark termination's snapshot: // // MT MT MT MT // .·| .·| .·| .·| // .·˙ | .·˙ | .·˙ | .·˙ | // .·˙ | .·˙ | .·˙ | .·˙ | // .·˙ |.·˙ |.·˙ |.·˙ | // // alloc → ▲ ← free // ┠┅┅┅┅┅┅┅┅┅┅┅P // C+2 → C+1 → C // // alloc → ▲ ← free // ┠┅┅┅┅┅┅┅┅┅┅┅P // C+2 → C+1 → C // // Since we can't publish a consistent snapshot until all of // the sweep frees are accounted for, we wait until the next // mark termination ("MT" above) to publish the previous mark // termination's snapshot ("P" above). To do this, allocation // and free events are accounted to *future* heap profile // cycles ("C+n" above) and we only publish a cycle once all // of the events from that cycle must be done. Specifically: // // Mallocs are accounted to cycle C+2. // Explicit frees are accounted to cycle C+2. // GC frees (done during sweeping) are accounted to cycle C+1. // // After mark termination, we increment the global heap // profile cycle counter and accumulate the stats from cycle C // into the active profile. // active is the currently published profile. A profiling // cycle can be accumulated into active once its complete. active memRecordCycle // future records the profile events we're counting for cycles // that have not yet been published. This is ring buffer // indexed by the global heap profile cycle C and stores // cycles C, C+1, and C+2. Unlike active, these counts are // only for a single cycle; they are not cumulative across // cycles. // // We store cycle C here because there's a window between when // C becomes the active cycle and when we've flushed it to // active. future [3]memRecordCycle } // memRecordCycle type memRecordCycle struct { allocs, frees uintptr alloc_bytes, free_bytes uintptr } // add accumulates b into a. It does not zero b. func (a *memRecordCycle) add(b *memRecordCycle) { a.allocs += b.allocs a.frees += b.frees a.alloc_bytes += b.alloc_bytes a.free_bytes += b.free_bytes } // A blockRecord is the bucket data for a bucket of type blockProfile, // which is used in blocking and mutex profiles. type blockRecord struct { count float64 cycles int64 } var ( mbuckets atomic.UnsafePointer // *bucket, memory profile buckets bbuckets atomic.UnsafePointer // *bucket, blocking profile buckets xbuckets atomic.UnsafePointer // *bucket, mutex profile buckets buckhash atomic.UnsafePointer // *buckhashArray mProfCycle mProfCycleHolder ) type buckhashArray [buckHashSize]atomic.UnsafePointer // *bucket const mProfCycleWrap = uint32(len(memRecord{}.future)) * (2 << 24) // mProfCycleHolder holds the global heap profile cycle number (wrapped at // mProfCycleWrap, stored starting at bit 1), and a flag (stored at bit 0) to // indicate whether future[cycle] in all buckets has been queued to flush into // the active profile. type mProfCycleHolder struct { value atomic.Uint32 } // read returns the current cycle count. func (c *mProfCycleHolder) read() (cycle uint32) { v := c.value.Load() cycle = v >> 1 return cycle } // setFlushed sets the flushed flag. It returns the current cycle count and the // previous value of the flushed flag. func (c *mProfCycleHolder) setFlushed() (cycle uint32, alreadyFlushed bool) { for { prev := c.value.Load() cycle = prev >> 1 alreadyFlushed = (prev & 0x1) != 0 next := prev | 0x1 if c.value.CompareAndSwap(prev, next) { return cycle, alreadyFlushed } } } // increment increases the cycle count by one, wrapping the value at // mProfCycleWrap. It clears the flushed flag. func (c *mProfCycleHolder) increment() { // We explicitly wrap mProfCycle rather than depending on // uint wraparound because the memRecord.future ring does not // itself wrap at a power of two. for { prev := c.value.Load() cycle := prev >> 1 cycle = (cycle + 1) % mProfCycleWrap next := cycle << 1 if c.value.CompareAndSwap(prev, next) { break } } } // newBucket allocates a bucket with the given type and number of stack entries. func newBucket(typ bucketType, nstk int) *bucket { size := unsafe.Sizeof(bucket{}) + uintptr(nstk)*unsafe.Sizeof(uintptr(0)) switch typ { default: throw("invalid profile bucket type") case memProfile: size += unsafe.Sizeof(memRecord{}) case blockProfile, mutexProfile: size += unsafe.Sizeof(blockRecord{}) } b := (*bucket)(persistentalloc(size, 0, &memstats.buckhash_sys)) b.typ = typ b.nstk = uintptr(nstk) return b } // stk returns the slice in b holding the stack. func (b *bucket) stk() []uintptr { stk := (*[maxStack]uintptr)(add(unsafe.Pointer(b), unsafe.Sizeof(*b))) if b.nstk > maxStack { // prove that slicing works; otherwise a failure requires a P throw("bad profile stack count") } return stk[:b.nstk:b.nstk] } // mp returns the memRecord associated with the memProfile bucket b. func (b *bucket) mp() *memRecord { if b.typ != memProfile { throw("bad use of bucket.mp") } data := add(unsafe.Pointer(b), unsafe.Sizeof(*b)+b.nstk*unsafe.Sizeof(uintptr(0))) return (*memRecord)(data) } // bp returns the blockRecord associated with the blockProfile bucket b. func (b *bucket) bp() *blockRecord { if b.typ != blockProfile && b.typ != mutexProfile { throw("bad use of bucket.bp") } data := add(unsafe.Pointer(b), unsafe.Sizeof(*b)+b.nstk*unsafe.Sizeof(uintptr(0))) return (*blockRecord)(data) } // Return the bucket for stk[0:nstk], allocating new bucket if needed. func stkbucket(typ bucketType, size uintptr, stk []uintptr, alloc bool) *bucket { bh := (*buckhashArray)(buckhash.Load()) if bh == nil { lock(&profInsertLock) // check again under the lock bh = (*buckhashArray)(buckhash.Load()) if bh == nil { bh = (*buckhashArray)(sysAlloc(unsafe.Sizeof(buckhashArray{}), &memstats.buckhash_sys)) if bh == nil { throw("runtime: cannot allocate memory") } buckhash.StoreNoWB(unsafe.Pointer(bh)) } unlock(&profInsertLock) } // Hash stack. var h uintptr for _, pc := range stk { h += pc h += h << 10 h ^= h >> 6 } // hash in size h += size h += h << 10 h ^= h >> 6 // finalize h += h << 3 h ^= h >> 11 i := int(h % buckHashSize) // first check optimistically, without the lock for b := (*bucket)(bh[i].Load()); b != nil; b = b.next { if b.typ == typ && b.hash == h && b.size == size && eqslice(b.stk(), stk) { return b } } if !alloc { return nil } lock(&profInsertLock) // check again under the insertion lock for b := (*bucket)(bh[i].Load()); b != nil; b = b.next { if b.typ == typ && b.hash == h && b.size == size && eqslice(b.stk(), stk) { unlock(&profInsertLock) return b } } // Create new bucket. b := newBucket(typ, len(stk)) copy(b.stk(), stk) b.hash = h b.size = size var allnext *atomic.UnsafePointer if typ == memProfile { allnext = &mbuckets } else if typ == mutexProfile { allnext = &xbuckets } else { allnext = &bbuckets } b.next = (*bucket)(bh[i].Load()) b.allnext = (*bucket)(allnext.Load()) bh[i].StoreNoWB(unsafe.Pointer(b)) allnext.StoreNoWB(unsafe.Pointer(b)) unlock(&profInsertLock) return b } func eqslice(x, y []uintptr) bool { if len(x) != len(y) { return false } for i, xi := range x { if xi != y[i] { return false } } return true } // mProf_NextCycle publishes the next heap profile cycle and creates a // fresh heap profile cycle. This operation is fast and can be done // during STW. The caller must call mProf_Flush before calling // mProf_NextCycle again. // // This is called by mark termination during STW so allocations and // frees after the world is started again count towards a new heap // profiling cycle. func mProf_NextCycle() { mProfCycle.increment() } // mProf_Flush flushes the events from the current heap profiling // cycle into the active profile. After this it is safe to start a new // heap profiling cycle with mProf_NextCycle. // // This is called by GC after mark termination starts the world. In // contrast with mProf_NextCycle, this is somewhat expensive, but safe // to do concurrently. func mProf_Flush() { cycle, alreadyFlushed := mProfCycle.setFlushed() if alreadyFlushed { return } index := cycle % uint32(len(memRecord{}.future)) lock(&profMemActiveLock) lock(&profMemFutureLock[index]) mProf_FlushLocked(index) unlock(&profMemFutureLock[index]) unlock(&profMemActiveLock) } // mProf_FlushLocked flushes the events from the heap profiling cycle at index // into the active profile. The caller must hold the lock for the active profile // (profMemActiveLock) and for the profiling cycle at index // (profMemFutureLock[index]). func mProf_FlushLocked(index uint32) { assertLockHeld(&profMemActiveLock) assertLockHeld(&profMemFutureLock[index]) head := (*bucket)(mbuckets.Load()) for b := head; b != nil; b = b.allnext { mp := b.mp() // Flush cycle C into the published profile and clear // it for reuse. mpc := &mp.future[index] mp.active.add(mpc) *mpc = memRecordCycle{} } } // mProf_PostSweep records that all sweep frees for this GC cycle have // completed. This has the effect of publishing the heap profile // snapshot as of the last mark termination without advancing the heap // profile cycle. func mProf_PostSweep() { // Flush cycle C+1 to the active profile so everything as of // the last mark termination becomes visible. *Don't* advance // the cycle, since we're still accumulating allocs in cycle // C+2, which have to become C+1 in the next mark termination // and so on. cycle := mProfCycle.read() + 1 index := cycle % uint32(len(memRecord{}.future)) lock(&profMemActiveLock) lock(&profMemFutureLock[index]) mProf_FlushLocked(index) unlock(&profMemFutureLock[index]) unlock(&profMemActiveLock) } // Called by malloc to record a profiled block. func mProf_Malloc(p unsafe.Pointer, size uintptr) { var stk [maxStack]uintptr nstk := callers(4, stk[:]) index := (mProfCycle.read() + 2) % uint32(len(memRecord{}.future)) b := stkbucket(memProfile, size, stk[:nstk], true) mp := b.mp() mpc := &mp.future[index] lock(&profMemFutureLock[index]) mpc.allocs++ mpc.alloc_bytes += size unlock(&profMemFutureLock[index]) // Setprofilebucket locks a bunch of other mutexes, so we call it outside of // the profiler locks. This reduces potential contention and chances of // deadlocks. Since the object must be alive during the call to // mProf_Malloc, it's fine to do this non-atomically. systemstack(func() { setprofilebucket(p, b) }) } // Called when freeing a profiled block. func mProf_Free(b *bucket, size uintptr) { index := (mProfCycle.read() + 1) % uint32(len(memRecord{}.future)) mp := b.mp() mpc := &mp.future[index] lock(&profMemFutureLock[index]) mpc.frees++ mpc.free_bytes += size unlock(&profMemFutureLock[index]) } var blockprofilerate uint64 // in CPU ticks // SetBlockProfileRate controls the fraction of goroutine blocking events // that are reported in the blocking profile. The profiler aims to sample // an average of one blocking event per rate nanoseconds spent blocked. // // To include every blocking event in the profile, pass rate = 1. // To turn off profiling entirely, pass rate <= 0. func SetBlockProfileRate(rate int) { var r int64 if rate <= 0 { r = 0 // disable profiling } else if rate == 1 { r = 1 // profile everything } else { // convert ns to cycles, use float64 to prevent overflow during multiplication r = int64(float64(rate) * float64(ticksPerSecond()) / (1000 * 1000 * 1000)) if r == 0 { r = 1 } } atomic.Store64(&blockprofilerate, uint64(r)) } func blockevent(cycles int64, skip int) { if cycles <= 0 { cycles = 1 } rate := int64(atomic.Load64(&blockprofilerate)) if blocksampled(cycles, rate) { saveblockevent(cycles, rate, skip+1, blockProfile) } } // blocksampled returns true for all events where cycles >= rate. Shorter // events have a cycles/rate random chance of returning true. func blocksampled(cycles, rate int64) bool { if rate <= 0 || (rate > cycles && cheaprand64()%rate > cycles) { return false } return true } func saveblockevent(cycles, rate int64, skip int, which bucketType) { gp := getg() var nstk int var stk [maxStack]uintptr if gp.m.curg == nil || gp.m.curg == gp { nstk = callers(skip, stk[:]) } else { nstk = gcallers(gp.m.curg, skip, stk[:]) } saveBlockEventStack(cycles, rate, stk[:nstk], which) } // lockTimer assists with profiling contention on runtime-internal locks. // // There are several steps between the time that an M experiences contention and // when that contention may be added to the profile. This comes from our // constraints: We need to keep the critical section of each lock small, // especially when those locks are contended. The reporting code cannot acquire // new locks until the M has released all other locks, which means no memory // allocations and encourages use of (temporary) M-local storage. // // The M will have space for storing one call stack that caused contention, and // for the magnitude of that contention. It will also have space to store the // magnitude of additional contention the M caused, since it only has space to // remember one call stack and might encounter several contention events before // it releases all of its locks and is thus able to transfer the local buffer // into the profile. // // The M will collect the call stack when it unlocks the contended lock. That // minimizes the impact on the critical section of the contended lock, and // matches the mutex profile's behavior for contention in sync.Mutex: measured // at the Unlock method. // // The profile for contention on sync.Mutex blames the caller of Unlock for the // amount of contention experienced by the callers of Lock which had to wait. // When there are several critical sections, this allows identifying which of // them is responsible. // // Matching that behavior for runtime-internal locks will require identifying // which Ms are blocked on the mutex. The semaphore-based implementation is // ready to allow that, but the futex-based implementation will require a bit // more work. Until then, we report contention on runtime-internal locks with a // call stack taken from the unlock call (like the rest of the user-space // "mutex" profile), but assign it a duration value based on how long the // previous lock call took (like the user-space "block" profile). // // Thus, reporting the call stacks of runtime-internal lock contention is // guarded by GODEBUG for now. Set GODEBUG=runtimecontentionstacks=1 to enable. // // TODO(rhysh): plumb through the delay duration, remove GODEBUG, update comment // // The M will track this by storing a pointer to the lock; lock/unlock pairs for // runtime-internal locks are always on the same M. // // Together, that demands several steps for recording contention. First, when // finally acquiring a contended lock, the M decides whether it should plan to // profile that event by storing a pointer to the lock in its "to be profiled // upon unlock" field. If that field is already set, it uses the relative // magnitudes to weight a random choice between itself and the other lock, with // the loser's time being added to the "additional contention" field. Otherwise // if the M's call stack buffer is occupied, it does the comparison against that // sample's magnitude. // // Second, having unlocked a mutex the M checks to see if it should capture the // call stack into its local buffer. Finally, when the M unlocks its last mutex, // it transfers the local buffer into the profile. As part of that step, it also // transfers any "additional contention" time to the profile. Any lock // contention that it experiences while adding samples to the profile will be // recorded later as "additional contention" and not include a call stack, to // avoid an echo. type lockTimer struct { lock *mutex timeRate int64 timeStart int64 tickStart int64 } func (lt *lockTimer) begin() { rate := int64(atomic.Load64(&mutexprofilerate)) lt.timeRate = gTrackingPeriod if rate != 0 && rate < lt.timeRate { lt.timeRate = rate } if int64(cheaprand())%lt.timeRate == 0 { lt.timeStart = nanotime() } if rate > 0 && int64(cheaprand())%rate == 0 { lt.tickStart = cputicks() } } func (lt *lockTimer) end() { gp := getg() if lt.timeStart != 0 { nowTime := nanotime() gp.m.mLockProfile.waitTime.Add((nowTime - lt.timeStart) * lt.timeRate) } if lt.tickStart != 0 { nowTick := cputicks() gp.m.mLockProfile.recordLock(nowTick-lt.tickStart, lt.lock) } } type mLockProfile struct { waitTime atomic.Int64 // total nanoseconds spent waiting in runtime.lockWithRank stack [maxStack]uintptr // stack that experienced contention in runtime.lockWithRank pending uintptr // *mutex that experienced contention (to be traceback-ed) cycles int64 // cycles attributable to "pending" (if set), otherwise to "stack" cyclesLost int64 // contention for which we weren't able to record a call stack disabled bool // attribute all time to "lost" } func (prof *mLockProfile) recordLock(cycles int64, l *mutex) { if cycles <= 0 { return } if prof.disabled { // We're experiencing contention while attempting to report contention. // Make a note of its magnitude, but don't allow it to be the sole cause // of another contention report. prof.cyclesLost += cycles return } if uintptr(unsafe.Pointer(l)) == prof.pending { // Optimization: we'd already planned to profile this same lock (though // possibly from a different unlock site). prof.cycles += cycles return } if prev := prof.cycles; prev > 0 { // We can only store one call stack for runtime-internal lock contention // on this M, and we've already got one. Decide which should stay, and // add the other to the report for runtime._LostContendedRuntimeLock. prevScore := uint64(cheaprand64()) % uint64(prev) thisScore := uint64(cheaprand64()) % uint64(cycles) if prevScore > thisScore { prof.cyclesLost += cycles return } else { prof.cyclesLost += prev } } // Saving the *mutex as a uintptr is safe because: // - lockrank_on.go does this too, which gives it regular exercise // - the lock would only move if it's stack allocated, which means it // cannot experience multi-M contention prof.pending = uintptr(unsafe.Pointer(l)) prof.cycles = cycles } // From unlock2, we might not be holding a p in this code. // //go:nowritebarrierrec func (prof *mLockProfile) recordUnlock(l *mutex) { if uintptr(unsafe.Pointer(l)) == prof.pending { prof.captureStack() } if gp := getg(); gp.m.locks == 1 && gp.m.mLockProfile.cycles != 0 { prof.store() } } func (prof *mLockProfile) captureStack() { skip := 3 // runtime.(*mLockProfile).recordUnlock runtime.unlock2 runtime.unlockWithRank if staticLockRanking { // When static lock ranking is enabled, we'll always be on the system // stack at this point. There will be a runtime.unlockWithRank.func1 // frame, and if the call to runtime.unlock took place on a user stack // then there'll also be a runtime.systemstack frame. To keep stack // traces somewhat consistent whether or not static lock ranking is // enabled, we'd like to skip those. But it's hard to tell how long // we've been on the system stack so accept an extra frame in that case, // with a leaf of "runtime.unlockWithRank runtime.unlock" instead of // "runtime.unlock". skip += 1 // runtime.unlockWithRank.func1 } prof.pending = 0 if debug.runtimeContentionStacks.Load() == 0 { prof.stack[0] = abi.FuncPCABIInternal(_LostContendedRuntimeLock) + sys.PCQuantum prof.stack[1] = 0 return } var nstk int gp := getg() sp := getcallersp() pc := getcallerpc() systemstack(func() { var u unwinder u.initAt(pc, sp, 0, gp, unwindSilentErrors|unwindJumpStack) nstk = tracebackPCs(&u, skip, prof.stack[:]) }) if nstk < len(prof.stack) { prof.stack[nstk] = 0 } } func (prof *mLockProfile) store() { // Report any contention we experience within this function as "lost"; it's // important that the act of reporting a contention event not lead to a // reportable contention event. This also means we can use prof.stack // without copying, since it won't change during this function. mp := acquirem() prof.disabled = true nstk := maxStack for i := 0; i < nstk; i++ { if pc := prof.stack[i]; pc == 0 { nstk = i break } } cycles, lost := prof.cycles, prof.cyclesLost prof.cycles, prof.cyclesLost = 0, 0 rate := int64(atomic.Load64(&mutexprofilerate)) saveBlockEventStack(cycles, rate, prof.stack[:nstk], mutexProfile) if lost > 0 { lostStk := [...]uintptr{ abi.FuncPCABIInternal(_LostContendedRuntimeLock) + sys.PCQuantum, } saveBlockEventStack(lost, rate, lostStk[:], mutexProfile) } prof.disabled = false releasem(mp) } func saveBlockEventStack(cycles, rate int64, stk []uintptr, which bucketType) { b := stkbucket(which, 0, stk, true) bp := b.bp() lock(&profBlockLock) // We want to up-scale the count and cycles according to the // probability that the event was sampled. For block profile events, // the sample probability is 1 if cycles >= rate, and cycles / rate // otherwise. For mutex profile events, the sample probability is 1 / rate. // We scale the events by 1 / (probability the event was sampled). if which == blockProfile && cycles < rate { // Remove sampling bias, see discussion on http://golang.org/cl/299991. bp.count += float64(rate) / float64(cycles) bp.cycles += rate } else if which == mutexProfile { bp.count += float64(rate) bp.cycles += rate * cycles } else { bp.count++ bp.cycles += cycles } unlock(&profBlockLock) } var mutexprofilerate uint64 // fraction sampled // SetMutexProfileFraction controls the fraction of mutex contention events // that are reported in the mutex profile. On average 1/rate events are // reported. The previous rate is returned. // // To turn off profiling entirely, pass rate 0. // To just read the current rate, pass rate < 0. // (For n>1 the details of sampling may change.) func SetMutexProfileFraction(rate int) int { if rate < 0 { return int(mutexprofilerate) } old := mutexprofilerate atomic.Store64(&mutexprofilerate, uint64(rate)) return int(old) } //go:linkname mutexevent sync.event func mutexevent(cycles int64, skip int) { if cycles < 0 { cycles = 0 } rate := int64(atomic.Load64(&mutexprofilerate)) if rate > 0 && cheaprand64()%rate == 0 { saveblockevent(cycles, rate, skip+1, mutexProfile) } } // Go interface to profile data. // A StackRecord describes a single execution stack. type StackRecord struct { Stack0 [32]uintptr // stack trace for this record; ends at first 0 entry } // Stack returns the stack trace associated with the record, // a prefix of r.Stack0. func (r *StackRecord) Stack() []uintptr { for i, v := range r.Stack0 { if v == 0 { return r.Stack0[0:i] } } return r.Stack0[0:] } // MemProfileRate controls the fraction of memory allocations // that are recorded and reported in the memory profile. // The profiler aims to sample an average of // one allocation per MemProfileRate bytes allocated. // // To include every allocated block in the profile, set MemProfileRate to 1. // To turn off profiling entirely, set MemProfileRate to 0. // // The tools that process the memory profiles assume that the // profile rate is constant across the lifetime of the program // and equal to the current value. Programs that change the // memory profiling rate should do so just once, as early as // possible in the execution of the program (for example, // at the beginning of main). var MemProfileRate int = 512 * 1024 // disableMemoryProfiling is set by the linker if runtime.MemProfile // is not used and the link type guarantees nobody else could use it // elsewhere. var disableMemoryProfiling bool // A MemProfileRecord describes the live objects allocated // by a particular call sequence (stack trace). type MemProfileRecord struct { AllocBytes, FreeBytes int64 // number of bytes allocated, freed AllocObjects, FreeObjects int64 // number of objects allocated, freed Stack0 [32]uintptr // stack trace for this record; ends at first 0 entry } // InUseBytes returns the number of bytes in use (AllocBytes - FreeBytes). func (r *MemProfileRecord) InUseBytes() int64 { return r.AllocBytes - r.FreeBytes } // InUseObjects returns the number of objects in use (AllocObjects - FreeObjects). func (r *MemProfileRecord) InUseObjects() int64 { return r.AllocObjects - r.FreeObjects } // Stack returns the stack trace associated with the record, // a prefix of r.Stack0. func (r *MemProfileRecord) Stack() []uintptr { for i, v := range r.Stack0 { if v == 0 { return r.Stack0[0:i] } } return r.Stack0[0:] } // MemProfile returns a profile of memory allocated and freed per allocation // site. // // MemProfile returns n, the number of records in the current memory profile. // If len(p) >= n, MemProfile copies the profile into p and returns n, true. // If len(p) < n, MemProfile does not change p and returns n, false. // // If inuseZero is true, the profile includes allocation records // where r.AllocBytes > 0 but r.AllocBytes == r.FreeBytes. // These are sites where memory was allocated, but it has all // been released back to the runtime. // // The returned profile may be up to two garbage collection cycles old. // This is to avoid skewing the profile toward allocations; because // allocations happen in real time but frees are delayed until the garbage // collector performs sweeping, the profile only accounts for allocations // that have had a chance to be freed by the garbage collector. // // Most clients should use the runtime/pprof package or // the testing package's -test.memprofile flag instead // of calling MemProfile directly. func MemProfile(p []MemProfileRecord, inuseZero bool) (n int, ok bool) { cycle := mProfCycle.read() // If we're between mProf_NextCycle and mProf_Flush, take care // of flushing to the active profile so we only have to look // at the active profile below. index := cycle % uint32(len(memRecord{}.future)) lock(&profMemActiveLock) lock(&profMemFutureLock[index]) mProf_FlushLocked(index) unlock(&profMemFutureLock[index]) clear := true head := (*bucket)(mbuckets.Load()) for b := head; b != nil; b = b.allnext { mp := b.mp() if inuseZero || mp.active.alloc_bytes != mp.active.free_bytes { n++ } if mp.active.allocs != 0 || mp.active.frees != 0 { clear = false } } if clear { // Absolutely no data, suggesting that a garbage collection // has not yet happened. In order to allow profiling when // garbage collection is disabled from the beginning of execution, // accumulate all of the cycles, and recount buckets. n = 0 for b := head; b != nil; b = b.allnext { mp := b.mp() for c := range mp.future { lock(&profMemFutureLock[c]) mp.active.add(&mp.future[c]) mp.future[c] = memRecordCycle{} unlock(&profMemFutureLock[c]) } if inuseZero || mp.active.alloc_bytes != mp.active.free_bytes { n++ } } } if n <= len(p) { ok = true idx := 0 for b := head; b != nil; b = b.allnext { mp := b.mp() if inuseZero || mp.active.alloc_bytes != mp.active.free_bytes { record(&p[idx], b) idx++ } } } unlock(&profMemActiveLock) return } // Write b's data to r. func record(r *MemProfileRecord, b *bucket) { mp := b.mp() r.AllocBytes = int64(mp.active.alloc_bytes) r.FreeBytes = int64(mp.active.free_bytes) r.AllocObjects = int64(mp.active.allocs) r.FreeObjects = int64(mp.active.frees) if raceenabled { racewriterangepc(unsafe.Pointer(&r.Stack0[0]), unsafe.Sizeof(r.Stack0), getcallerpc(), abi.FuncPCABIInternal(MemProfile)) } if msanenabled { msanwrite(unsafe.Pointer(&r.Stack0[0]), unsafe.Sizeof(r.Stack0)) } if asanenabled { asanwrite(unsafe.Pointer(&r.Stack0[0]), unsafe.Sizeof(r.Stack0)) } copy(r.Stack0[:], b.stk()) for i := int(b.nstk); i < len(r.Stack0); i++ { r.Stack0[i] = 0 } } func iterate_memprof(fn func(*bucket, uintptr, *uintptr, uintptr, uintptr, uintptr)) { lock(&profMemActiveLock) head := (*bucket)(mbuckets.Load()) for b := head; b != nil; b = b.allnext { mp := b.mp() fn(b, b.nstk, &b.stk()[0], b.size, mp.active.allocs, mp.active.frees) } unlock(&profMemActiveLock) } // BlockProfileRecord describes blocking events originated // at a particular call sequence (stack trace). type BlockProfileRecord struct { Count int64 Cycles int64 StackRecord } // BlockProfile returns n, the number of records in the current blocking profile. // If len(p) >= n, BlockProfile copies the profile into p and returns n, true. // If len(p) < n, BlockProfile does not change p and returns n, false. // // Most clients should use the [runtime/pprof] package or // the [testing] package's -test.blockprofile flag instead // of calling BlockProfile directly. func BlockProfile(p []BlockProfileRecord) (n int, ok bool) { lock(&profBlockLock) head := (*bucket)(bbuckets.Load()) for b := head; b != nil; b = b.allnext { n++ } if n <= len(p) { ok = true for b := head; b != nil; b = b.allnext { bp := b.bp() r := &p[0] r.Count = int64(bp.count) // Prevent callers from having to worry about division by zero errors. // See discussion on http://golang.org/cl/299991. if r.Count == 0 { r.Count = 1 } r.Cycles = bp.cycles if raceenabled { racewriterangepc(unsafe.Pointer(&r.Stack0[0]), unsafe.Sizeof(r.Stack0), getcallerpc(), abi.FuncPCABIInternal(BlockProfile)) } if msanenabled { msanwrite(unsafe.Pointer(&r.Stack0[0]), unsafe.Sizeof(r.Stack0)) } if asanenabled { asanwrite(unsafe.Pointer(&r.Stack0[0]), unsafe.Sizeof(r.Stack0)) } i := copy(r.Stack0[:], b.stk()) for ; i < len(r.Stack0); i++ { r.Stack0[i] = 0 } p = p[1:] } } unlock(&profBlockLock) return } // MutexProfile returns n, the number of records in the current mutex profile. // If len(p) >= n, MutexProfile copies the profile into p and returns n, true. // Otherwise, MutexProfile does not change p, and returns n, false. // // Most clients should use the [runtime/pprof] package // instead of calling MutexProfile directly. func MutexProfile(p []BlockProfileRecord) (n int, ok bool) { lock(&profBlockLock) head := (*bucket)(xbuckets.Load()) for b := head; b != nil; b = b.allnext { n++ } if n <= len(p) { ok = true for b := head; b != nil; b = b.allnext { bp := b.bp() r := &p[0] r.Count = int64(bp.count) r.Cycles = bp.cycles i := copy(r.Stack0[:], b.stk()) for ; i < len(r.Stack0); i++ { r.Stack0[i] = 0 } p = p[1:] } } unlock(&profBlockLock) return } // ThreadCreateProfile returns n, the number of records in the thread creation profile. // If len(p) >= n, ThreadCreateProfile copies the profile into p and returns n, true. // If len(p) < n, ThreadCreateProfile does not change p and returns n, false. // // Most clients should use the runtime/pprof package instead // of calling ThreadCreateProfile directly. func ThreadCreateProfile(p []StackRecord) (n int, ok bool) { first := (*m)(atomic.Loadp(unsafe.Pointer(&allm))) for mp := first; mp != nil; mp = mp.alllink { n++ } if n <= len(p) { ok = true i := 0 for mp := first; mp != nil; mp = mp.alllink { p[i].Stack0 = mp.createstack i++ } } return } //go:linkname runtime_goroutineProfileWithLabels runtime/pprof.runtime_goroutineProfileWithLabels func runtime_goroutineProfileWithLabels(p []StackRecord, labels []unsafe.Pointer) (n int, ok bool) { return goroutineProfileWithLabels(p, labels) } // labels may be nil. If labels is non-nil, it must have the same length as p. func goroutineProfileWithLabels(p []StackRecord, labels []unsafe.Pointer) (n int, ok bool) { if labels != nil && len(labels) != len(p) { labels = nil } return goroutineProfileWithLabelsConcurrent(p, labels) } var goroutineProfile = struct { sema uint32 active bool offset atomic.Int64 records []StackRecord labels []unsafe.Pointer }{ sema: 1, } // goroutineProfileState indicates the status of a goroutine's stack for the // current in-progress goroutine profile. Goroutines' stacks are initially // "Absent" from the profile, and end up "Satisfied" by the time the profile is // complete. While a goroutine's stack is being captured, its // goroutineProfileState will be "InProgress" and it will not be able to run // until the capture completes and the state moves to "Satisfied". // // Some goroutines (the finalizer goroutine, which at various times can be // either a "system" or a "user" goroutine, and the goroutine that is // coordinating the profile, any goroutines created during the profile) move // directly to the "Satisfied" state. type goroutineProfileState uint32 const ( goroutineProfileAbsent goroutineProfileState = iota goroutineProfileInProgress goroutineProfileSatisfied ) type goroutineProfileStateHolder atomic.Uint32 func (p *goroutineProfileStateHolder) Load() goroutineProfileState { return goroutineProfileState((*atomic.Uint32)(p).Load()) } func (p *goroutineProfileStateHolder) Store(value goroutineProfileState) { (*atomic.Uint32)(p).Store(uint32(value)) } func (p *goroutineProfileStateHolder) CompareAndSwap(old, new goroutineProfileState) bool { return (*atomic.Uint32)(p).CompareAndSwap(uint32(old), uint32(new)) } func goroutineProfileWithLabelsConcurrent(p []StackRecord, labels []unsafe.Pointer) (n int, ok bool) { semacquire(&goroutineProfile.sema) ourg := getg() stw := stopTheWorld(stwGoroutineProfile) // Using gcount while the world is stopped should give us a consistent view // of the number of live goroutines, minus the number of goroutines that are // alive and permanently marked as "system". But to make this count agree // with what we'd get from isSystemGoroutine, we need special handling for // goroutines that can vary between user and system to ensure that the count // doesn't change during the collection. So, check the finalizer goroutine // in particular. n = int(gcount()) if fingStatus.Load()&fingRunningFinalizer != 0 { n++ } if n > len(p) { // There's not enough space in p to store the whole profile, so (per the // contract of runtime.GoroutineProfile) we're not allowed to write to p // at all and must return n, false. startTheWorld(stw) semrelease(&goroutineProfile.sema) return n, false } // Save current goroutine. sp := getcallersp() pc := getcallerpc() systemstack(func() { saveg(pc, sp, ourg, &p[0]) }) if labels != nil { labels[0] = ourg.labels } ourg.goroutineProfiled.Store(goroutineProfileSatisfied) goroutineProfile.offset.Store(1) // Prepare for all other goroutines to enter the profile. Aside from ourg, // every goroutine struct in the allgs list has its goroutineProfiled field // cleared. Any goroutine created from this point on (while // goroutineProfile.active is set) will start with its goroutineProfiled // field set to goroutineProfileSatisfied. goroutineProfile.active = true goroutineProfile.records = p goroutineProfile.labels = labels // The finalizer goroutine needs special handling because it can vary over // time between being a user goroutine (eligible for this profile) and a // system goroutine (to be excluded). Pick one before restarting the world. if fing != nil { fing.goroutineProfiled.Store(goroutineProfileSatisfied) if readgstatus(fing) != _Gdead && !isSystemGoroutine(fing, false) { doRecordGoroutineProfile(fing) } } startTheWorld(stw) // Visit each goroutine that existed as of the startTheWorld call above. // // New goroutines may not be in this list, but we didn't want to know about // them anyway. If they do appear in this list (via reusing a dead goroutine // struct, or racing to launch between the world restarting and us getting // the list), they will already have their goroutineProfiled field set to // goroutineProfileSatisfied before their state transitions out of _Gdead. // // Any goroutine that the scheduler tries to execute concurrently with this // call will start by adding itself to the profile (before the act of // executing can cause any changes in its stack). forEachGRace(func(gp1 *g) { tryRecordGoroutineProfile(gp1, Gosched) }) stw = stopTheWorld(stwGoroutineProfileCleanup) endOffset := goroutineProfile.offset.Swap(0) goroutineProfile.active = false goroutineProfile.records = nil goroutineProfile.labels = nil startTheWorld(stw) // Restore the invariant that every goroutine struct in allgs has its // goroutineProfiled field cleared. forEachGRace(func(gp1 *g) { gp1.goroutineProfiled.Store(goroutineProfileAbsent) }) if raceenabled { raceacquire(unsafe.Pointer(&labelSync)) } if n != int(endOffset) { // It's a big surprise that the number of goroutines changed while we // were collecting the profile. But probably better to return a // truncated profile than to crash the whole process. // // For instance, needm moves a goroutine out of the _Gdead state and so // might be able to change the goroutine count without interacting with // the scheduler. For code like that, the race windows are small and the // combination of features is uncommon, so it's hard to be (and remain) // sure we've caught them all. } semrelease(&goroutineProfile.sema) return n, true } // tryRecordGoroutineProfileWB asserts that write barriers are allowed and calls // tryRecordGoroutineProfile. // //go:yeswritebarrierrec func tryRecordGoroutineProfileWB(gp1 *g) { if getg().m.p.ptr() == nil { throw("no P available, write barriers are forbidden") } tryRecordGoroutineProfile(gp1, osyield) } // tryRecordGoroutineProfile ensures that gp1 has the appropriate representation // in the current goroutine profile: either that it should not be profiled, or // that a snapshot of its call stack and labels are now in the profile. func tryRecordGoroutineProfile(gp1 *g, yield func()) { if readgstatus(gp1) == _Gdead { // Dead goroutines should not appear in the profile. Goroutines that // start while profile collection is active will get goroutineProfiled // set to goroutineProfileSatisfied before transitioning out of _Gdead, // so here we check _Gdead first. return } if isSystemGoroutine(gp1, true) { // System goroutines should not appear in the profile. (The finalizer // goroutine is marked as "already profiled".) return } for { prev := gp1.goroutineProfiled.Load() if prev == goroutineProfileSatisfied { // This goroutine is already in the profile (or is new since the // start of collection, so shouldn't appear in the profile). break } if prev == goroutineProfileInProgress { // Something else is adding gp1 to the goroutine profile right now. // Give that a moment to finish. yield() continue } // While we have gp1.goroutineProfiled set to // goroutineProfileInProgress, gp1 may appear _Grunnable but will not // actually be able to run. Disable preemption for ourselves, to make // sure we finish profiling gp1 right away instead of leaving it stuck // in this limbo. mp := acquirem() if gp1.goroutineProfiled.CompareAndSwap(goroutineProfileAbsent, goroutineProfileInProgress) { doRecordGoroutineProfile(gp1) gp1.goroutineProfiled.Store(goroutineProfileSatisfied) } releasem(mp) } } // doRecordGoroutineProfile writes gp1's call stack and labels to an in-progress // goroutine profile. Preemption is disabled. // // This may be called via tryRecordGoroutineProfile in two ways: by the // goroutine that is coordinating the goroutine profile (running on its own // stack), or from the scheduler in preparation to execute gp1 (running on the // system stack). func doRecordGoroutineProfile(gp1 *g) { if readgstatus(gp1) == _Grunning { print("doRecordGoroutineProfile gp1=", gp1.goid, "\n") throw("cannot read stack of running goroutine") } offset := int(goroutineProfile.offset.Add(1)) - 1 if offset >= len(goroutineProfile.records) { // Should be impossible, but better to return a truncated profile than // to crash the entire process at this point. Instead, deal with it in // goroutineProfileWithLabelsConcurrent where we have more context. return } // saveg calls gentraceback, which may call cgo traceback functions. When // called from the scheduler, this is on the system stack already so // traceback.go:cgoContextPCs will avoid calling back into the scheduler. // // When called from the goroutine coordinating the profile, we still have // set gp1.goroutineProfiled to goroutineProfileInProgress and so are still // preventing it from being truly _Grunnable. So we'll use the system stack // to avoid schedule delays. systemstack(func() { saveg(^uintptr(0), ^uintptr(0), gp1, &goroutineProfile.records[offset]) }) if goroutineProfile.labels != nil { goroutineProfile.labels[offset] = gp1.labels } } func goroutineProfileWithLabelsSync(p []StackRecord, labels []unsafe.Pointer) (n int, ok bool) { gp := getg() isOK := func(gp1 *g) bool { // Checking isSystemGoroutine here makes GoroutineProfile // consistent with both NumGoroutine and Stack. return gp1 != gp && readgstatus(gp1) != _Gdead && !isSystemGoroutine(gp1, false) } stw := stopTheWorld(stwGoroutineProfile) // World is stopped, no locking required. n = 1 forEachGRace(func(gp1 *g) { if isOK(gp1) { n++ } }) if n <= len(p) { ok = true r, lbl := p, labels // Save current goroutine. sp := getcallersp() pc := getcallerpc() systemstack(func() { saveg(pc, sp, gp, &r[0]) }) r = r[1:] // If we have a place to put our goroutine labelmap, insert it there. if labels != nil { lbl[0] = gp.labels lbl = lbl[1:] } // Save other goroutines. forEachGRace(func(gp1 *g) { if !isOK(gp1) { return } if len(r) == 0 { // Should be impossible, but better to return a // truncated profile than to crash the entire process. return } // saveg calls gentraceback, which may call cgo traceback functions. // The world is stopped, so it cannot use cgocall (which will be // blocked at exitsyscall). Do it on the system stack so it won't // call into the schedular (see traceback.go:cgoContextPCs). systemstack(func() { saveg(^uintptr(0), ^uintptr(0), gp1, &r[0]) }) if labels != nil { lbl[0] = gp1.labels lbl = lbl[1:] } r = r[1:] }) } if raceenabled { raceacquire(unsafe.Pointer(&labelSync)) } startTheWorld(stw) return n, ok } // GoroutineProfile returns n, the number of records in the active goroutine stack profile. // If len(p) >= n, GoroutineProfile copies the profile into p and returns n, true. // If len(p) < n, GoroutineProfile does not change p and returns n, false. // // Most clients should use the [runtime/pprof] package instead // of calling GoroutineProfile directly. func GoroutineProfile(p []StackRecord) (n int, ok bool) { return goroutineProfileWithLabels(p, nil) } func saveg(pc, sp uintptr, gp *g, r *StackRecord) { var u unwinder u.initAt(pc, sp, 0, gp, unwindSilentErrors) n := tracebackPCs(&u, 0, r.Stack0[:]) if n < len(r.Stack0) { r.Stack0[n] = 0 } } // Stack formats a stack trace of the calling goroutine into buf // and returns the number of bytes written to buf. // If all is true, Stack formats stack traces of all other goroutines // into buf after the trace for the current goroutine. func Stack(buf []byte, all bool) int { var stw worldStop if all { stw = stopTheWorld(stwAllGoroutinesStack) } n := 0 if len(buf) > 0 { gp := getg() sp := getcallersp() pc := getcallerpc() systemstack(func() { g0 := getg() // Force traceback=1 to override GOTRACEBACK setting, // so that Stack's results are consistent. // GOTRACEBACK is only about crash dumps. g0.m.traceback = 1 g0.writebuf = buf[0:0:len(buf)] goroutineheader(gp) traceback(pc, sp, 0, gp) if all { tracebackothers(gp) } g0.m.traceback = 0 n = len(g0.writebuf) g0.writebuf = nil }) } if all { startTheWorld(stw) } return n } // Tracing of alloc/free/gc. var tracelock mutex func tracealloc(p unsafe.Pointer, size uintptr, typ *_type) { lock(&tracelock) gp := getg() gp.m.traceback = 2 if typ == nil { print("tracealloc(", p, ", ", hex(size), ")\n") } else { print("tracealloc(", p, ", ", hex(size), ", ", toRType(typ).string(), ")\n") } if gp.m.curg == nil || gp == gp.m.curg { goroutineheader(gp) pc := getcallerpc() sp := getcallersp() systemstack(func() { traceback(pc, sp, 0, gp) }) } else { goroutineheader(gp.m.curg) traceback(^uintptr(0), ^uintptr(0), 0, gp.m.curg) } print("\n") gp.m.traceback = 0 unlock(&tracelock) } func tracefree(p unsafe.Pointer, size uintptr) { lock(&tracelock) gp := getg() gp.m.traceback = 2 print("tracefree(", p, ", ", hex(size), ")\n") goroutineheader(gp) pc := getcallerpc() sp := getcallersp() systemstack(func() { traceback(pc, sp, 0, gp) }) print("\n") gp.m.traceback = 0 unlock(&tracelock) } func tracegc() { lock(&tracelock) gp := getg() gp.m.traceback = 2 print("tracegc()\n") // running on m->g0 stack; show all non-g0 goroutines tracebackothers(gp) print("end tracegc\n") print("\n") gp.m.traceback = 0 unlock(&tracelock) }