| // Copyright 2023 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. |
| |
| //go:build goexperiment.exectracer2 |
| |
| // Trace goroutine and P status management. |
| |
| package runtime |
| |
| import "internal/runtime/atomic" |
| |
| // traceGoStatus is the status of a goroutine. |
| // |
| // They correspond directly to the various goroutine |
| // statuses. |
| type traceGoStatus uint8 |
| |
| const ( |
| traceGoBad traceGoStatus = iota |
| traceGoRunnable |
| traceGoRunning |
| traceGoSyscall |
| traceGoWaiting |
| ) |
| |
| // traceProcStatus is the status of a P. |
| // |
| // They mostly correspond to the various P statuses. |
| type traceProcStatus uint8 |
| |
| const ( |
| traceProcBad traceProcStatus = iota |
| traceProcRunning |
| traceProcIdle |
| traceProcSyscall |
| |
| // traceProcSyscallAbandoned is a special case of |
| // traceProcSyscall. It's used in the very specific case |
| // where the first a P is mentioned in a generation is |
| // part of a ProcSteal event. If that's the first time |
| // it's mentioned, then there's no GoSyscallBegin to |
| // connect the P stealing back to at that point. This |
| // special state indicates this to the parser, so it |
| // doesn't try to find a GoSyscallEndBlocked that |
| // corresponds with the ProcSteal. |
| traceProcSyscallAbandoned |
| ) |
| |
| // writeGoStatus emits a GoStatus event as well as any active ranges on the goroutine. |
| func (w traceWriter) writeGoStatus(goid uint64, mid int64, status traceGoStatus, markAssist bool) traceWriter { |
| // The status should never be bad. Some invariant must have been violated. |
| if status == traceGoBad { |
| print("runtime: goid=", goid, "\n") |
| throw("attempted to trace a bad status for a goroutine") |
| } |
| |
| // Trace the status. |
| w = w.event(traceEvGoStatus, traceArg(goid), traceArg(uint64(mid)), traceArg(status)) |
| |
| // Trace any special ranges that are in-progress. |
| if markAssist { |
| w = w.event(traceEvGCMarkAssistActive, traceArg(goid)) |
| } |
| return w |
| } |
| |
| // writeProcStatusForP emits a ProcStatus event for the provided p based on its status. |
| // |
| // The caller must fully own pp and it must be prevented from transitioning (e.g. this can be |
| // called by a forEachP callback or from a STW). |
| func (w traceWriter) writeProcStatusForP(pp *p, inSTW bool) traceWriter { |
| if !pp.trace.acquireStatus(w.gen) { |
| return w |
| } |
| var status traceProcStatus |
| switch pp.status { |
| case _Pidle, _Pgcstop: |
| status = traceProcIdle |
| if pp.status == _Pgcstop && inSTW { |
| // N.B. a P that is running and currently has the world stopped will be |
| // in _Pgcstop, but we model it as running in the tracer. |
| status = traceProcRunning |
| } |
| case _Prunning: |
| status = traceProcRunning |
| // There's a short window wherein the goroutine may have entered _Gsyscall |
| // but it still owns the P (it's not in _Psyscall yet). The goroutine entering |
| // _Gsyscall is the tracer's signal that the P its bound to is also in a syscall, |
| // so we need to emit a status that matches. See #64318. |
| if w.mp.p.ptr() == pp && w.mp.curg != nil && readgstatus(w.mp.curg)&^_Gscan == _Gsyscall { |
| status = traceProcSyscall |
| } |
| case _Psyscall: |
| status = traceProcSyscall |
| default: |
| throw("attempt to trace invalid or unsupported P status") |
| } |
| w = w.writeProcStatus(uint64(pp.id), status, pp.trace.inSweep) |
| return w |
| } |
| |
| // writeProcStatus emits a ProcStatus event with all the provided information. |
| // |
| // The caller must have taken ownership of a P's status writing, and the P must be |
| // prevented from transitioning. |
| func (w traceWriter) writeProcStatus(pid uint64, status traceProcStatus, inSweep bool) traceWriter { |
| // The status should never be bad. Some invariant must have been violated. |
| if status == traceProcBad { |
| print("runtime: pid=", pid, "\n") |
| throw("attempted to trace a bad status for a proc") |
| } |
| |
| // Trace the status. |
| w = w.event(traceEvProcStatus, traceArg(pid), traceArg(status)) |
| |
| // Trace any special ranges that are in-progress. |
| if inSweep { |
| w = w.event(traceEvGCSweepActive, traceArg(pid)) |
| } |
| return w |
| } |
| |
| // goStatusToTraceGoStatus translates the internal status to tracGoStatus. |
| // |
| // status must not be _Gdead or any status whose name has the suffix "_unused." |
| func goStatusToTraceGoStatus(status uint32, wr waitReason) traceGoStatus { |
| // N.B. Ignore the _Gscan bit. We don't model it in the tracer. |
| var tgs traceGoStatus |
| switch status &^ _Gscan { |
| case _Grunnable: |
| tgs = traceGoRunnable |
| case _Grunning, _Gcopystack: |
| tgs = traceGoRunning |
| case _Gsyscall: |
| tgs = traceGoSyscall |
| case _Gwaiting, _Gpreempted: |
| // There are a number of cases where a G might end up in |
| // _Gwaiting but it's actually running in a non-preemptive |
| // state but needs to present itself as preempted to the |
| // garbage collector. In these cases, we're not going to |
| // emit an event, and we want these goroutines to appear in |
| // the final trace as if they're running, not blocked. |
| tgs = traceGoWaiting |
| if status == _Gwaiting && |
| wr == waitReasonStoppingTheWorld || |
| wr == waitReasonGCMarkTermination || |
| wr == waitReasonGarbageCollection || |
| wr == waitReasonTraceProcStatus || |
| wr == waitReasonPageTraceFlush || |
| wr == waitReasonGCWorkerActive { |
| tgs = traceGoRunning |
| } |
| case _Gdead: |
| throw("tried to trace dead goroutine") |
| default: |
| throw("tried to trace goroutine with invalid or unsupported status") |
| } |
| return tgs |
| } |
| |
| // traceSchedResourceState is shared state for scheduling resources (i.e. fields common to |
| // both Gs and Ps). |
| type traceSchedResourceState struct { |
| // statusTraced indicates whether a status event was traced for this resource |
| // a particular generation. |
| // |
| // There are 3 of these because when transitioning across generations, traceAdvance |
| // needs to be able to reliably observe whether a status was traced for the previous |
| // generation, while we need to clear the value for the next generation. |
| statusTraced [3]atomic.Uint32 |
| |
| // seq is the sequence counter for this scheduling resource's events. |
| // The purpose of the sequence counter is to establish a partial order between |
| // events that don't obviously happen serially (same M) in the stream ofevents. |
| // |
| // There are two of these so that we can reset the counter on each generation. |
| // This saves space in the resulting trace by keeping the counter small and allows |
| // GoStatus and GoCreate events to omit a sequence number (implicitly 0). |
| seq [2]uint64 |
| } |
| |
| // acquireStatus acquires the right to emit a Status event for the scheduling resource. |
| func (r *traceSchedResourceState) acquireStatus(gen uintptr) bool { |
| if !r.statusTraced[gen%3].CompareAndSwap(0, 1) { |
| return false |
| } |
| r.readyNextGen(gen) |
| return true |
| } |
| |
| // readyNextGen readies r for the generation following gen. |
| func (r *traceSchedResourceState) readyNextGen(gen uintptr) { |
| nextGen := traceNextGen(gen) |
| r.seq[nextGen%2] = 0 |
| r.statusTraced[nextGen%3].Store(0) |
| } |
| |
| // statusWasTraced returns true if the sched resource's status was already acquired for tracing. |
| func (r *traceSchedResourceState) statusWasTraced(gen uintptr) bool { |
| return r.statusTraced[gen%3].Load() != 0 |
| } |
| |
| // setStatusTraced indicates that the resource's status was already traced, for example |
| // when a goroutine is created. |
| func (r *traceSchedResourceState) setStatusTraced(gen uintptr) { |
| r.statusTraced[gen%3].Store(1) |
| } |
| |
| // nextSeq returns the next sequence number for the resource. |
| func (r *traceSchedResourceState) nextSeq(gen uintptr) traceArg { |
| r.seq[gen%2]++ |
| return traceArg(r.seq[gen%2]) |
| } |