1540719 - Symbol allocation doesn't trigger a last ditch GC on failure

Status: RESOLVED FIXED

(Core :: JavaScript: GC, enhancement, P3)

Description

[Jon Coppeard (:jonco)]

Currently we don't do a last ditch GC if symbol allocation fails. The code works like this because it was written when you need to take a lock for allocating in the atoms zone, which would also be required for the last ditch GC. That is not the case any more so we can remove this restriction.

Comment 1

[Jon Coppeard (:jonco)]

Attachment: Bug 1540719 - Perform a last ditch GC if symbol allocation fails as we do for most other GC things

This removes the NoGC specifier and ensures GC things are passed in handles rather than raw pointers. The existing comments helpfully pointed out where we need to use DependentAddPtr.

Comment 2

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/mozilla-inbound/rev/87ae5e3b6962
Perform a last ditch GC if symbol allocation fails as we do for most other GC things r=sfink

Comment 3

[Noemi Erli[:noemi_erli]]

Backed out 2 changesets (Bug 1505622, Bug 1540719) for spidermonkey bustages in Allocator.cpp

Push with failures: https://treeherder.mozilla.org/#/jobs?repo=mozilla-inbound&resultStatus=testfailed%2Cbusted%2Cexception&classifiedState=unclassified&revision=0199e2beb6f94dda62383ff2f7b0d401b148d16a&selectedJob=239002452

Failure log: https://treeherder.mozilla.org/logviewer.html#/jobs?job_id=239002452&repo=mozilla-inbound&lineNumber=5890

Backout: https://hg.mozilla.org/integration/mozilla-inbound/rev/7d3896c97b076b5784b4f3831aee796a5b1e311f

Comment 4

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/mozilla-inbound/rev/54dffe309994
Perform a last ditch GC if symbol allocation fails as we do for most other GC things r=sfink

Comment 5

[Jon Coppeard (:jonco)]

Bustage was unrelated to this patch.

Comment 6

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/mozilla-inbound/rev/401de283153d
Fix rooting hazard in test code r=me

Comment 7

[Jon Coppeard (:jonco)]

(In reply to Jon Coppeard (:jonco) from comment #5)
In fact, there was more bustage that was related to this patch. Hopefully this sticks now.

Comment 8

[Cosmin Sabou [:CosminS]]

https://hg.mozilla.org/mozilla-central/rev/54dffe309994
https://hg.mozilla.org/mozilla-central/rev/401de283153d