1419779 - Crash in js::jit::ReleaseProcessExecutableMemory

Status: RESOLVED INVALID

(Core :: JavaScript Engine: JIT, defect, P2)

Description

[Andrei Oprea [:andreio]]

This is happening while running code coverage tests on talos, being investigated in bug 1414593. Bisect attributes it to a JS change in Activity Stream.

https://treeherder.mozilla.org/#/jobs?repo=try&author=[e-mail]&fromchange=0534a2c0d4502267f81357322891c18aee8a694a&filter-searchStr=ccov%20g5&tochange=62b0395210633288d697f6a69eb9e07f954a2bf5&selectedJob=143633100

Comment 1

[Andrei Oprea [:andreio]]

Link to stack trace https://treeherder.mozilla.org/logviewer.html#?job_id=143633100&repo=try&lineNumber=18263

Comment 2

[Jan de Mooij [:jandem]]

<jandem> andreio: it's an assertion failure, could be a shutdown leak
<jandem> andreio: similar to bug 1411322
<firebot> bugzil.la/1411322 — FIXED, apehrson⊙mozilla.com — Assertion failure: pages_.empty(), at /builds/worker/workspace/build/src/js/src/jit/ProcessExecutabl
<jandem> it's the same assert

Comment 3

[Jason Orendorff [:jorendorff]]

Jan, can you take this?

Comment 4

[Jan de Mooij [:jandem]]

(In reply to Jason Orendorff [:jorendorff] from comment #3)
> Jan, can you take this?

See comment 2. I don't think this is a bug in the JS engine but (similar to bug 1411322) a leak elsewhere that trigers the "we freed all executable pages" assert.

Comment 5

[Andrei Oprea [:andreio]]

The leak is being tracked in bug 1414593. I wasn't sure about the info in the stack trace but it sounds like this bug is invalid then?