Go调度器学习之系统调用详解

2023-05-14 17:05:04 调用 调度 详解

0. 简介

上篇博客,我们分析了Go调度器中的抢占策略,这篇,我们将分析一下,在系统调用时发生的调度行为。

1. 系统调用

下面,我们将以一个简单的文件打开的系统调用,来分析一下Go调度器在系统调用时做了什么。

1.1 场景

package main

import (
   "fmt"
   "io/ioutil"
   "os"
)

func main() {
   f, err := os.Open("./file")
   if err != nil {
      panic(err)
   }
   defer f.Close()

   content, err := ioutil.ReadAll(f)
   if err != nil {
      panic(err)
   }
   fmt.Println(string(content))
}

如上简单的代码,读取一个名为file的本地文件,然后打印其数据,我们通过汇编代码来分析一下其调用过程:

$ go build -GCflags "-N -l" -o main main.go
$ objdump -d main >> main.i

可以发现,在main.i中,从main.main函数,对于文件Open操作的调用关系为:main.main -> os.Open -> os.openFile -> os.openFileNolog -> syscall.openat -> syscall.Syscall6.abi0 -> runtime.entersyscall.abi0,而Syscall6的汇编如下:

TEXT ·Syscall6(SB),NOSPLIT,$0-80
   CALL   runtime·entersyscall(SB)
   MOVQ   a1+8(FP), DI
   MOVQ   a2+16(FP), SI
   MOVQ   a3+24(FP), DX
   MOVQ   a4+32(FP), R10
   MOVQ   a5+40(FP), R8
   MOVQ   a6+48(FP), R9
   MOVQ   trap+0(FP), AX // syscall entry
   SYSCALL
   CMPQ   AX, $0xfffffffffffff001
   JLS    ok6
   MOVQ   $-1, r1+56(FP)
   MOVQ   $0, r2+64(FP)
   NEGQ   AX
   MOVQ   AX, err+72(FP)
   CALL   runtime·exitsyscall(SB)
   RET
ok6:
   MOVQ   AX, r1+56(FP)
   MOVQ   DX, r2+64(FP)
   MOVQ   $0, err+72(FP)
   CALL   runtime·exitsyscall(SB)
   RET

1.2 陷入系统调用

可以发现,系统调用最终会进入到runtime.entersyscall函数:

func entersyscall() {
   reentersyscall(getcallerpc(), getcallersp())
}

runtime.entersyscall函数会调用runtime.reentersyscall

func reentersyscall(pc, sp uintptr) {
   _g_ := getg()

   // Disable preemption because during this function g is in Gsyscall status,
   // but can have inconsistent g->sched, do not let GC observe it.
   _g_.m.locks++

   // Entersyscall must not call any function that might split/grow the stack.
   // (See details in comment above.)
   // Catch calls that might, by replacing the stack guard with something that
   // will trip any stack check and leaving a flag to tell newstack to die.
   _g_.stackguard0 = stackPreempt
   _g_.throwsplit = true

   // Leave SP around for GC and traceback.
   save(pc, sp)  // 保存pc和sp
   _g_.syscallsp = sp
   _g_.syscallpc = pc
   casgstatus(_g_, _Grunning, _Gsyscall)
   if _g_.syscallsp < _g_.stack.lo || _g_.stack.hi < _g_.syscallsp {
      systemstack(func() {
         print("entersyscall inconsistent ", hex(_g_.syscallsp), " [", hex(_g_.stack.lo), ",", hex(_g_.stack.hi), "]\n")
         throw("entersyscall")
      })
   }

   if trace.enabled {
      systemstack(traceGoSysCall)
      // systemstack itself clobbers g.sched.{pc,sp} and we might
      // need them later when the G is genuinely blocked in a
      // syscall
      save(pc, sp)
   }

   if atomic.Load(&sched.sysmonwait) != 0 {
      systemstack(entersyscall_sysmon)
      save(pc, sp)
   }

   if _g_.m.p.ptr().runSafePointFn != 0 {
      // runSafePointFn may stack split if run on this stack
      systemstack(runSafePointFn)
      save(pc, sp)
   }

   // 一下解绑P和M
   _g_.m.syscalltick = _g_.m.p.ptr().syscalltick
   _g_.sysblocktraced = true
   pp := _g_.m.p.ptr()
   pp.m = 0
   _g_.m.oldp.set(pp)  // 存储一下旧P
   _g_.m.p = 0
   atomic.Store(&pp.status, _Psyscall)
   if sched.gcwaiting != 0 {
      systemstack(entersyscall_gcwait)
      save(pc, sp)
   }

   _g_.m.locks--
}

可以发现,runtime.reentersyscall除了做一些保障性的工作外,最重要的是做了以下三件事:

  • 保存当前goroutine的PC和栈指针SP的内容;
  • 将当前goroutine的状态置为_Gsyscall
  • 将当前P的状态置为_Psyscall,并解绑P和M,让当前M陷入内核的系统调用中,P被释放,可以被其他找工作的M找到并且执行剩下的goroutine

1.3 从系统调用恢复

func exitsyscall() {
   _g_ := getg()

   _g_.m.locks++ // see comment in entersyscall
   if getcallersp() > _g_.syscallsp {
      throw("exitsyscall: syscall frame is no longer valid")
   }

   _g_.waitsince = 0
   oldp := _g_.m.oldp.ptr()  // 拿到开始存储的旧P
   _g_.m.oldp = 0
   if exitsyscallfast(oldp) {
      if trace.enabled {
         if oldp != _g_.m.p.ptr() || _g_.m.syscalltick != _g_.m.p.ptr().syscalltick {
            systemstack(traceGoStart)
         }
      }
      // There's a cpu for us, so we can run.
      _g_.m.p.ptr().syscalltick++
      // We need to cas the status and scan before resuming...
      casgstatus(_g_, _Gsyscall, _Grunning)

      ...

      return
   }

   ...

   // Call the scheduler.
   mcall(exitsyscall0)

   // Scheduler returned, so we're allowed to run now.
   // Delete the syscallsp infORMation that we left for
   // the garbage collector during the system call.
   // Must wait until now because until gosched returns
   // we don't know for sure that the garbage collector
   // is not running.
   _g_.syscallsp = 0
   _g_.m.p.ptr().syscalltick++
   _g_.throwsplit = false
}

其中,exitsyscallfast函数有以下个分支:

  • 如果旧的P还没有被其他M占用,依旧处于_Psyscall状态,那么直接通过wirep函数获取这个P,返回true;
  • 如果旧的P被占用了,那么调用exitsyscallfast_pidle去获取空闲的P来执行,返回true;
  • 如果没有空闲的P,则返回false;
//go:nosplit
func exitsyscallfast(oldp *p) bool {
   _g_ := getg()

   // Freezetheworld sets stopwait but does not retake P's.
   if sched.stopwait == freezeStopWait {
      return false
   }

   // 如果上一个P没有被其他M占用,还处于_Psyscall状态,那么直接通过wirep函数获取此P
   // Try to re-acquire the last P.
   if oldp != nil && oldp.status == _Psyscall && atomic.Cas(&oldp.status, _Psyscall, _Pidle) {
      // There's a cpu for us, so we can run.
      wirep(oldp)
      exitsyscallfast_reacquired()
      return true
   }

   // Try to get any other idle P.
   if sched.pidle != 0 {
      var ok bool
      systemstack(func() {
         ok = exitsyscallfast_pidle()
         if ok && trace.enabled {
            if oldp != nil {
               // Wait till traceGoSysBlock event is emitted.
               // This ensures consistency of the trace (the goroutine is started after it is blocked).
               for oldp.syscalltick == _g_.m.syscalltick {
                  osyield()
               }
            }
            traceGoSysExit(0)
         }
      })
      if ok {
         return true
      }
   }
   return false
}

exitsyscallfast函数返回false后,则会调用exitsyscall0函数去处理:

func exitsyscall0(gp *g) {
   casgstatus(gp, _Gsyscall, _Grunnable)
   dropg() // 因为当前m没有找到p,所以先解开g和m
   lock(&sched.lock)
   var _p_ *p
   if schedEnabled(gp) {
      _p_ = pidleget() // 还是尝试找一下有没有空闲的p
   }
   var locked bool
   if _p_ == nil { // 如果还是没有空闲p,那么把g扔到全局队列去等待调度
      globrunqput(gp)

      // Below, we stoplockedm if gp is locked. globrunqput releases
      // ownership of gp, so we must check if gp is locked prior to
      // committing the release by unlocking sched.lock, otherwise we
      // could race with another M transitioning gp from unlocked to
      // locked.
      locked = gp.lockedm != 0
   } else if atomic.Load(&sched.sysmonwait) != 0 {
      atomic.Store(&sched.sysmonwait, 0)
      notewakeup(&sched.sysmonnote)
   }
   unlock(&sched.lock)
   if _p_ != nil { // 如果找到了空闲p,那么就去执行,这个分支永远不会返回
      acquirep(_p_)
      execute(gp, false) // Never returns.
   }
   if locked {
      // Wait until another thread schedules gp and so m again.
      //
      // N.B. lockedm must be this M, as this g was running on this M
      // before entersyscall.
      stoplockedm()
      execute(gp, false) // Never returns.
   }
   stopm() // 这里还是没有找到空闲p的条件,停止这个m,因为没有p,所以m应该要开始找工作了
   schedule() // Never returns. // 通过schedule函数进行调度
}

exitsyscall0函数还是会尝试找一个空闲的P,没有的话就把goroutine扔到全局队列,然后停止这个M,并且调用schedule函数等待调度;如果找到了空闲P,则会利用这个P去执行此goroutine

2. 小结

通过以上分析,可以发现goroutine有关系统调用的调度还是比较简单的:

  • 在发生系统调用时会将此goroutine设置为_Gsyscall状态;
  • 并将P设置为_Psyscall状态,并且解绑M和P,使得这个P可以去执行其他的goroutine,而M就陷入系统内核调用中了;
  • 当该M从内核调用中恢复到用户态时,会优先去获取原来的旧P,如果该旧P还未被其他M占用,则利用该P继续执行本goroutine
  • 如果没有获取到旧P,那么会尝试去P的空闲列表获取一个P来执行;
  • 如果空闲列表中没有获取到P,就会把goroutine扔到全局队列中,等到继续执行。

可以发现,如果系统发生着很频繁的系统调用,很可能会产生很多的M,在IO密集型的场景下,甚至会发生线程数超过10000的panic事件。而Go团队为此也进行了很多努力,下一节我们将介绍的网络轮询器将介绍,至少在网络IO密集型场景,Go SDK是怎么优化的。

以上就是Go调度器学习之系统调用详解的详细内容,更多关于Go调度器 系统调用的资料请关注其它相关文章!

相关文章