languages/go
languages/go
1
0
Fork 0
mirror of https://github.com/golang/go synced 2024-10-01 21:54:53 +00:00
Code Issues Packages Projects Releases Wiki Activity

runtime: move timer recheck after GC recheck

Browse source 
When rechecking for work after transitioning from a spinning to
non-spinning M, checking timers before GC isn't useful. That is, if
there is GC work available, it will run immediately and the updated
pollUntil is unused.

Move this check to just before netpoll, where pollUntil is used. While
this technically improves efficiency in the (rare) case that we find
GC work in this block, the primary motivation is simply to improve
clarity by moving the update closer to use.

For #43997

Change-Id: Ibc7fb308ac4a582875c200659c9e272121a89f3b
Reviewed-on: https://go-review.googlesource.com/c/go/+/308654
Trust: Michael Pratt <mpratt@google.com>
Trust: Michael Knyszek <mknyszek@google.com>
Run-TryBot: Michael Pratt <mpratt@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
This commit is contained in:
Michael Pratt 2021-04-08 16:57:57 -04:00
parent 381252f312
commit 4d56576ec0
1 changed files with 9 additions and 7 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

16
src/runtime/proc.go
View file

@ -2810,13 +2810,7 @@ top:
goto top
}
// Similar to above, check for timer creation or expiry concurrently with
// transitioning from spinning to non-spinning. Note that we cannot use
// checkTimers here because it calls adjusttimers which may need to allocate
// memory, and that isn't allowed when we don't have an active P.
pollUntil = checkTimersNoP(allpSnapshot, timerpMaskSnapshot, pollUntil)
// Finally, check for idle-priority GC work.
// Check for idle-priority GC work again.
_p_, gp = checkIdleGCNoP()
if _p_ != nil {
acquirep(_p_)
@ -2834,6 +2828,14 @@ top:
return gp, false
}
// Finally, check for timer creation or expiry concurrently with
// transitioning from spinning to non-spinning.
//
// Note that we cannot use checkTimers here because it calls
// adjusttimers which may need to allocate memory, and that isn't
// allowed when we don't have an active P.
pollUntil = checkTimersNoP(allpSnapshot, timerpMaskSnapshot, pollUntil)
// Poll network until next timer.
if netpollinited() && (atomic.Load(&netpollWaiters) > 0 || pollUntil != 0) && atomic.Xchg64(&sched.lastpoll, 0) != 0 {
atomic.Store64(&sched.pollUntil, uint64(pollUntil))