1170717 - Don't dynamically check the heapState in gray barrier

Status: RESOLVED FIXED

(Core :: JavaScript: GC, defect)

Description

[Terrence Cole [:terrence]]

Attachment: no_need_for_dynamic_check_for_isheapbusy_in_gray_barriers-v0.diff

Now that we disable barriers using heapState, the isHeapBusy check in the gray barrier should a dupe.

Comment 1

[Terrence Cole [:terrence]]

Try run is at:
https://treeherder.mozilla.org/#/jobs?repo=try&revision=590ca54a2875

I give this 50/50 on going spectacularly poorly. Will depend if I missed any obvious entry points to the gray barrier.

Comment 2

[Andrew McCreight [:mccr8]]

Comment on attachment 8614244 [details] [diff] [review]
no_need_for_dynamic_check_for_isheapbusy_in_gray_barriers-v0.diff

Review of attachment 8614244 [details] [diff] [review]:
-----------------------------------------------------------------

I don't know anything about that part of the code, sorry.

Comment 3

[Terrence Cole [:terrence]]

Comment on attachment 8614244 [details] [diff] [review]
no_need_for_dynamic_check_for_isheapbusy_in_gray_barriers-v0.diff

Yup, that went quite poorly.

Comment 4

[Terrence Cole [:terrence]]

Attachment: no_need_for_dynamic_check_for_isheapbusy_in_gray_barriers-v1.diff

New try run is green. 
https://treeherder.mozilla.org/#/jobs?repo=try&revision=dc1c41a4c6f8

I was thinking of the pre-barrier -- the read barrier did not check the heapState before. I do think it should though and try does seem to be fine with the change.

Comment 5

[Jon Coppeard (:jonco)]

Comment on attachment 8614785 [details] [diff] [review]
no_need_for_dynamic_check_for_isheapbusy_in_gray_barriers-v1.diff

Review of attachment 8614785 [details] [diff] [review]:
-----------------------------------------------------------------

LGTM.

Comment 6

[Pulsebot]

https://hg.mozilla.org/integration/mozilla-inbound/rev/335614e5c352

Comment 7

[Wes Kocher (:KWierso) (Not reading bugmail; email directly if needed)]

https://hg.mozilla.org/mozilla-central/rev/335614e5c352