Closed Bug 1370786 Opened 8 years ago Closed 8 years ago

Permaorange ASan devtools LeakSanitizer | leak at lul::LUL::NotifyAfterMap, read_procmaps, SamplerThread::SamplerThread when Gecko 55 merges to beta on 2017-06-12

Categories

(DevTools :: General, enhancement)

55 Branch
enhancement
Not set
normal

Tracking

(firefox-esr45 unaffected, firefox-esr52 unaffected, firefox53 unaffected, firefox54 unaffected, firefox55blocking fixed)

RESOLVED FIXED
Firefox 55
Tracking Status
firefox-esr45 --- unaffected
firefox-esr52 --- unaffected
firefox53 --- unaffected
firefox54 --- unaffected
firefox55 blocking fixed

People

(Reporter: philor, Assigned: tromey)

Details

Attachments

(1 file)

This is new bustage from something in the range https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=2c6289f56812&tochange=5801aa478de1 since I pushed a simulation of trunk as beta to try yesterday and didn't hit this, but now we have permaorange https://treeherder.mozilla.org/logviewer.html#?job_id=105058608&repo=try [Tracking Requested - why for this release]: merge bustage, closed tree, delayed b1
This is a blocker for 55.0b1
Just FYI, the `TypeError: DebuggerServer.removeContentServerScript is not a function` error message that is directly above the LeakSanitizer error is being fixed in Bug 1370925. But I'm assuming it's not related to this particular issue.
One hypothesis for the leak is that AddSecMap can early-return without deleting the passed-in SecMap. https://dxr.mozilla.org/mozilla-central/rev/5801aa478de12a62b2b2982659e787fcc4268d67/tools/profiler/lul/LulMain.cpp#541-543 Ownership here would be clearer with UniquePtr. I'm not sure how to test this though ... ?
Comment on attachment 8875369 [details] Bug 1370786 - use UniquePtr for SecMap in LUL; Could you give some feedback on this? And maybe tell me, if you know, how to reproduce the failure here?
Attachment #8875369 - Flags: feedback?(nfroyd)
Comment on attachment 8875369 [details] Bug 1370786 - use UniquePtr for SecMap in LUL; This patch seems good; it might be a little better to store UniquePtr inside mSecMaps, but I understand that would be a more far-reaching change. As for testing this, you can replicate philor's try-push by grabbing his patches, I guess? If you ask nicely, philor might even simulate beta with your patch added on top.
Attachment #8875369 - Flags: feedback?(nfroyd) → feedback+
From irc: <philor> tromey: oh, I see. https://hg.mozilla.org/try/rev/a3e416c66e714e2e63e2afb35529f6d698b60eb9 and https://hg.mozilla.org/try/rev/e45e86a84f88c54475eb733a116c51a49010d1df and try: -b o -p linux64-asan -u mochitest-dt -t none
Assignee: nobody → ttromey
Here's a try run for the new patch: https://treeherder.mozilla.org/#/jobs?repo=try&revision=88a983f98e35ff8cd44f2628c29837fac7a53c14 This run included :philor's patches.
Comment on attachment 8875369 [details] Bug 1370786 - use UniquePtr for SecMap in LUL; https://reviewboard.mozilla.org/r/146794/#review151590 Above and beyond! \o/
Attachment #8875369 - Flags: review?(nfroyd) → review+
Pushed by ttromey@mozilla.com: https://hg.mozilla.org/integration/autoland/rev/bf13425e15f7 use UniquePtr for SecMap in LUL; r=froydnj
Status: NEW → RESOLVED
Closed: 8 years ago
Resolution: --- → FIXED
Target Milestone: --- → Firefox 55
Product: Firefox → DevTools
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: