This bug was filed from the Socorro interface and is report bp-5eec4421-0b39-4a74-9120-eaecc0170901. ============================================================= Seen while looking at crash stats: http://bit.ly/2emUlle. Crashes started using 20170831100258. 31 crashes/12 installs. Possible regression from Bug 1395063?

As mentioned in the commit message in bug 1395063, it changed the crash signatures, so instead of having crashes in moz_abort that come from page_commit, you now have crashes in page_commit directly.

IOW, it's equivalent to: https://crash-stats.mozilla.com/signature/?_sort=-date&signature=moz_abort%20%7C%20pages_commit&date=%3E%3D2017-08-25T01%3A05%3A00.000Z&date=%3C2017-09-01T01%3A05%3A00.000Z

See also bug 1395805. glandium, is this likely to be generic heap corruption?

More like OOM. (VirtualAlloc failing)

I looked at a few of the crashes. Some of them have very low "Available Page File" values, e.g.: https://crash-stats.mozilla.com/report/index/6daaf648-4acf-43d5-936c-2b3f50170910 https://crash-stats.mozilla.com/report/index/23e134f3-2d5e-4578-ac15-8024a0170909 https://crash-stats.mozilla.com/report/index/8eb5b355-d078-49ae-a208-d0f4a0170909 But some of them have plenty of virtual memory, physical memory, and page file: https://crash-stats.mozilla.com/report/index/0faa0af6-89f3-4350-a7d8-4a4c80170910 https://crash-stats.mozilla.com/report/index/59fce851-8070-4a1c-96b2-4828d0170909

I'll file a bug to get something added to the Socorro prefix list so these different overcommit crashes don't end up in a single bucket, like they were before.

(In reply to Andrew McCreight [:mccr8] from comment #6) > I'll file a bug to get something added to the Socorro prefix list so these > different overcommit crashes don't end up in a single bucket, like they were > before. Well, I guess that doesn't make too much sense. I didn't realize the old signature also wasn't going very far on the prefix list. I'll just dupe this over to the old bug for this signature, and add the signature.