As noted in bug 2034619 - for Windows x86 debug, this appears to have started on this large WPT-sync push near the end of April: https://hg-edge.mozilla.org/integration/autoland/pushloghtml?changeset=c27a951b8452b5a86378ea76d5aa4eaab9079ec7 Here's the first-bad log (from that push): https://treeherder.mozilla.org/logviewer?job_id=562049682&repo=autoland&task=dLBREYYxT42-SFzdsPkXaA.0&lineNumber=29414 And the last-good log (from the previous push): https://treeherder.mozilla.org/logviewer?job_id=562049669&repo=autoland&task=XdyuMYZqSxWaVTzWPlQm8Q.0 Looking for differences between those logs, one important-seeming one is that the last-good log had 34 instances of this line: ``` Firefox didn't exit cleanly, not processing leak logs ``` ...whereas the first bad log only has 1 instance of that line. So this might really have been a leak that we've been hitting for quite a while (maybe even on all platforms) but for some reason we're flagging the Firefox exit as not-being-clean and hence we skip leak-checking.
Bug 2038448 Comment 9 Edit History
Note: The actual edited comment in the bug view page will always show the original commenter’s name and original timestamp.
As noted in bug 2034619 - for Windows x86 debug, this appears to have started on this large WPT-sync push near the end of April: https://hg-edge.mozilla.org/integration/autoland/pushloghtml?changeset=c27a951b8452b5a86378ea76d5aa4eaab9079ec7 Here's the first-bad log (from that push): https://treeherder.mozilla.org/logviewer?job_id=562049682&repo=autoland&task=dLBREYYxT42-SFzdsPkXaA.0&lineNumber=29414 And the last-good log (from the previous push): https://treeherder.mozilla.org/logviewer?job_id=562049669&repo=autoland&task=XdyuMYZqSxWaVTzWPlQm8Q.0 Looking for differences between those logs, one important-seeming one is that the last-good log had 34 instances of this line: ``` Firefox didn't exit cleanly, not processing leak logs ``` ...whereas the first bad log only has 1 instance of that line. So this might really have been a leak that we've been hitting for quite a while (maybe even on all platforms) but for some reason we're flagging the Firefox exit as not-being-clean and hence we skip leak-checking. Not sure why that WPT push in particular made us start exiting cleanly more often (I guess), but that's probably a good thing in any case.
As noted in bug 2034619 - for Windows x86 debug, this appears to have started on this large WPT-sync push near the end of April: https://hg-edge.mozilla.org/integration/autoland/pushloghtml?changeset=c27a951b8452b5a86378ea76d5aa4eaab9079ec7 Here's the first-bad log (from that push): https://treeherder.mozilla.org/logviewer?job_id=562049682&repo=autoland&task=dLBREYYxT42-SFzdsPkXaA.0&lineNumber=29414 And the last-good log (from the previous push): https://treeherder.mozilla.org/logviewer?job_id=562049669&repo=autoland&task=XdyuMYZqSxWaVTzWPlQm8Q.0 Looking for differences between those logs, one important-seeming one is that the last-good log had 34 instances of this line: ``` Firefox didn't exit cleanly, not processing leak logs ``` ...whereas the first bad log only has 1 instance of that line. So this might really have been a leak that we've been hitting for quite a while (maybe even on all platforms) but for some reason we're flagging the Firefox exit as not-being-clean and hence we skip leak-checking. Not sure why that WPT-sync push in particular made us start exiting cleanly more often (I guess), but that's probably a good thing in any case.