DescriptionRevert of Revert "Revert of [oilpan]: Change marking to do precise roots first and conservative second. (http… (patchset #2 of https://codereview.chromium.org/413133006/)
Reason for revert:
This breaks strongification of ephemerons when there are pointers on the stack. See:
https://codereview.chromium.org/461413002/
for details.
Original issue's description:
> Revert "Revert of [oilpan]: Change marking to do precise roots first and conservative second. (https://codereview.chromium.org/405403003/)"
>
> This reverts commit 765af3a45b15c4e5f65810c8ed8e20d44da04034.
>
> AFAICT the CrossThreadPointerToOrphanedPage test was a bit flaky in the sense that the compiler was free to reuse the slot where the stackPtrValue was located after the RELEASE_ASSERT. However the test assumed the stackPtrValue was on the stack (to mimic a rogue integer value finding a dead object) when doing the conservative GC further down.
> After my change to Heap::CollectGarbage it seemed the android compiler no longer kept the value around and hence the test started failing. I have now moved the RELEASE_ASSERT down to after the point where we do the conservative GC which makes the test pass on android again.
>
> I uploaded the unfixed diff first and then a second diff (moved RELEASE_ASSERT) with the fix.
>
> R=ager@chromium.org, erik.corry@gmail.com, haraken@chromium.org, oilpan-reviews@chromium.org, sigbjornf@opera.com, tkent@chromium.org, zerny@chromium.org
> BUG=
>
> Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=178845
NOTRY=true
NOTREECHECKS=true
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=180160
Patch Set 1 #
Created: 6 years, 4 months ago
(Patch set is too large to download)
Messages
Total messages: 10 (0 generated)
|