]> Cypherpunks repositories - gostls13.git/commit
runtime: recheck GC trigger before actually starting GC
authorAustin Clements <austin@google.com>
Fri, 31 Jul 2015 17:52:17 +0000 (13:52 -0400)
committerAustin Clements <austin@google.com>
Tue, 4 Aug 2015 18:54:32 +0000 (18:54 +0000)
commit88e945fd231b40a41955f2b8505f680a83520fbf
tree5cc724fa73c719545bfc9bb9b04036c42aaf00b3
parentd5f5e658aef1103944b514a10f987690fe9ec0d3
runtime: recheck GC trigger before actually starting GC

Currently allocation checks the GC trigger speculatively during
allocation and then triggers the GC without rechecking. As a result,
it's possible for G 1 and G 2 to detect the trigger simultaneously,
both enter startGC, G 1 actually starts GC while G 2 gets preempted
until after the whole GC cycle, then G 2 immediately starts another GC
cycle even though the heap is now well under the trigger.

Fix this by re-checking the GC trigger non-speculatively just before
actually kicking off a new GC cycle.

This contributes to #11911 because when this happens, we definitely
don't finish the background sweep before starting the next GC cycle,
which can significantly delay the start of concurrent scan.

Change-Id: I560ab79ba5684ba435084410a9765d28f5745976
Reviewed-on: https://go-review.googlesource.com/13025
Reviewed-by: Russ Cox <rsc@golang.org>
Reviewed-by: Rick Hudson <rlh@golang.org>
src/runtime/mgc.go