CrashKill/2014-11-03

General

Previous Action Items

• [mschifer] Talk to Milan on how we best can get our priorities to the gfx team
• [KaiRo] Follow up on the stability list on future of this meeting

Socorro

• bug 1093123 Crash counts for 2014-11-02 are much lower than they should be

Desktop

Beta

• bug 1089413 #1 top crash. startup crash. may be fixed. needs verifying for 34 later in the week.
• bug 1093186 crash in MarkInternal<JSString> may need a security bug filed as well
• bug 719114 (unactionable) @ScanShape is showing up in Beta at fairly high volume.

Aurora

• bug 1091903 - crash in mozilla::gfx::SourceSurfaceD2D1::SourceSurfaceD2D1
  • Seems to be fixed on trunk (unless it just has changed signature) but not fixed on Aurora

Trunk

• bug 1093076 - mozilla::plugins::PPluginModule::Bridge(mozilla::dom::PContentParent*, mozilla::plugins::PPluginModuleParent*)
  • Climbing up in the list of top crashers #11, also showing in the list of explosive reports. e10s related.
• bug 865314 (also 1089638) backout should reduce the spike in crashes related to the signatures with "PL_DHashTableOperate(PLDHashTable*, void const*, PLDHashOperator)"

Notes

• KaiRo ran a report to list all GFX adapters seen in a day in crash reports: https://crash-analysis.mozilla.com/rkaiser/2014-10-31/2014-10-31.gfxdata.html
  • Across all products and versions; desktop tends to have hex PCI IDs, mobile has names; 0x0 is unknown, may be from old versions where we didn't report yet
  • Did only do a one-off run for that day, can re-run easily for any day if needed; restricting to products or such would need quite a bit more work.

Action Items

• [KaiRo] Follow up on the stability list on future of this meeting
• [KaiRo] Look at Trello for prioritization of GFX stability issues