Closed Bug 1326305 Opened 7 years ago Closed 7 years ago

Intermittent leakcheck | default process: 636891 bytes leaked (AsyncLatencyLogger, BackstagePass, CSSStyleSheet, CSSStyleSheetInner, CacheObserver, ...)

Categories

(Toolkit :: General, defect)

defect
Not set
normal

Tracking

()

RESOLVED FIXED
mozilla53
Tracking Status
firefox51 --- unaffected
firefox52 --- unaffected
firefox53 --- fixed

People

(Reporter: intermittent-bug-filer, Assigned: jaws)

References

Details

(Keywords: intermittent-failure, memory-leak)

Attachments

(1 file)

Looks like we're leaking the world here. Kats, is this the leak you were fixing on the graphics branch?
Component: General → CSS Parsing and Computation
Flags: needinfo?(bugmail)
Keywords: mlk
No, the leak on the graphics branch was specific to the webrender code, which is only on the graphics branch.
Flags: needinfo?(bugmail)
It's always happening in the toolkit/modules/tests/browser directory, and it started on the push for bug 1325464 (confirmed by a large number of retriggers before and after).

Conveniently, that push included changes in that test directory via rev cd10db6087dd. Or maybe one of the JSM changes in toolkit/modules?
https://hg.mozilla.org/integration/autoland/rev/cd10db6087ddd3a8e872c0de6ef35a886250173e

Any idea what might be going on here, Jared?
Blocks: 1325464
Component: CSS Parsing and Computation → General
Flags: needinfo?(jaws)
Product: Core → Toolkit
Version: unspecified → Trunk
I'm looking in to this now.
While working on bug 1325464 I encountered a test failure that was due to the usage of object-shorthand for constructors. This was commented on at https://hg.mozilla.org/mozilla-central/annotate/15ea2e33fcd4/browser/base/content/browser-fullScreenAndPointerLock.js#l34

I've pushed to tryserver a patch that reverts the changes from bug 1325464 in /toolkit/modules with respect to constructors: https://treeherder.mozilla.org/#/jobs?repo=try&revision=a29979633ad9577c97c1375d98e2c0276d432c8d
(In reply to Jared Wein [:jaws] (please needinfo? me) from comment #5)
> While working on bug 1325464 I encountered a test failure that was due to
> the usage of object-shorthand for constructors. This was commented on at
> https://hg.mozilla.org/mozilla-central/annotate/15ea2e33fcd4/browser/base/
> content/browser-fullScreenAndPointerLock.js#l34

That one should be rewritten to a class (actually my initial impl was a class, but es6-class was not enabled at that moment).
The above trypush didn't fix the leak. This trypush reverts all of the changes under /toolkit/modules, https://treeherder.mozilla.org/#/jobs?repo=try&revision=5c3d4d4674d4727b2da9d3403e9292fec24415b5.

It fails the eslint job but that can be ignored because this is just seeing if the leak goes away.

(In reply to Xidorn Quan [:xidorn] (UTC+10) from comment #7)
> (In reply to Jared Wein [:jaws] (please needinfo? me) from comment #5)
> > While working on bug 1325464 I encountered a test failure that was due to
> > the usage of object-shorthand for constructors. This was commented on at
> > https://hg.mozilla.org/mozilla-central/annotate/15ea2e33fcd4/browser/base/
> > content/browser-fullScreenAndPointerLock.js#l34
> 
> That one should be rewritten to a class (actually my initial impl was a
> class, but es6-class was not enabled at that moment).

Can you please file a bug for this? That could be good-first-bug material :)
Flags: needinfo?(xidorn+moz)
(In reply to Jared Wein [:jaws] (please needinfo? me) from comment #8)
> > That one should be rewritten to a class (actually my initial impl was a
> > class, but es6-class was not enabled at that moment).
> 
> Can you please file a bug for this? That could be good-first-bug material :)

Filed bug 1326581.
Flags: needinfo?(xidorn+moz)
The following try pushes reverse all the changes under second level directories:

browser/base: https://treeherder.mozilla.org/#/jobs?repo=try&revision=9324465b85578cbafcc59f967306218b54dfd77f
browser/components: https://treeherder.mozilla.org/#/jobs?repo=try&revision=15750c6100d2b7b5b4813e06fd37a1555958bb07
browser/experiments: https://treeherder.mozilla.org/#/jobs?repo=try&revision=0ef915d41a67ecc73396953f53147e3d787c2fb7
browser/extensions: https://treeherder.mozilla.org/#/jobs?repo=try&revision=dd8ddd85baf376e2e6d877cc20ca991671c33ca0
browser/modules: https://treeherder.mozilla.org/#/jobs?repo=try&revision=f12b9a4819e55d68c5dcd145f5502c0fa56a3fd2
browser/tools: https://treeherder.mozilla.org/#/jobs?repo=try&revision=dba43d79118800b62fb04d74e43e62450a3da593
storage/test: https://treeherder.mozilla.org/#/jobs?repo=try&revision=5dd00c6c0174e0871d0d64a7fe2767bd4c386cd5
toolkit/components: https://treeherder.mozilla.org/#/jobs?repo=try&revision=f4f67d511dd2172d9742eeea0ed095cf73e15a56
toolkit/content: https://treeherder.mozilla.org/#/jobs?repo=try&revision=7ca55707a5a9bf35e8b888abcc278ba3441a3449
toolkit/crashreporter: https://treeherder.mozilla.org/#/jobs?repo=try&revision=ab4e21cfdeeabb663c1561abc8e7671460eb884b
toolkit/forgetaboutsite: https://treeherder.mozilla.org/#/jobs?repo=try&revision=96d0507b4851269188716704de279067f9c92daf
toolkit/identity: https://treeherder.mozilla.org/#/jobs?repo=try&revision=bd84e41f9802d11b3a08096d2f5b22fc952635f4
toolkit/modules: https://treeherder.mozilla.org/#/jobs?repo=try&revision=5213738d2650e30df6c99e089c61c725f3adbaaa
toolkit/mozapps: https://treeherder.mozilla.org/#/jobs?repo=try&revision=25f2856ab37560c5b7363b0b2f0b64b1f06c46b3
Of the previous trypushes, https://treeherder.mozilla.org/#/jobs?repo=try&revision=c0e5dde3f9a33538b20279a476f6679404ee5dce is the only one that doesn't reproduce the leak.

I scanned through the list of files that changed in that push and ProductAddonChecker.jsm stands out as the only file that changed that required a manual merge. It looks like I screwed up the merge and this is the cause for the leaks.

I've pushed a patch to tryserver that should fix this and will request review once I have confirmation that it is fixed: https://treeherder.mozilla.org/#/jobs?repo=try&revision=46b06b6b9ec382eb66e1474545448a3d9a61d704
Comment on attachment 8823765 [details]
Bug 1326305 - Revert bad manual merge from bug 1325464 that lead to memory leaks.

https://reviewboard.mozilla.org/r/102278/#review102996
Attachment #8823765 - Flags: review+
Pushed by jwein@mozilla.com:
https://hg.mozilla.org/integration/autoland/rev/0be68b87d9db
Revert bad manual merge from bug 1325464 that lead to memory leaks. r=rhelmer
Attachment #8823765 - Flags: review?(kmaglione+bmo)
Status: ASSIGNED → RESOLVED
Closed: 7 years ago
Resolution: --- → FIXED
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: