1098655 - Huge number of crashes in igd10umd32.dll and igd10iumd32.dll

Status: RESOLVED INCOMPLETE

(Core :: Graphics, defect)

Description

[Mats Palmgren (inactive)]

Firefox 33.1 in the past 7 days has:

2,776 crashes in igd10umd32.dll:
https://crash-stats.mozilla.com/search/?signature=~igd10umd32.dll&product=Firefox&version=33.1&_facets=signature&_columns=date&_columns=signature&_columns=product&_columns=version&_columns=build_id&_columns=platform

and 3,263 crashes in igd10iumd32.dll:
https://crash-stats.mozilla.com/search/?signature=~igd10iumd32.dll&product=Firefox&version=33.1&_facets=signature&_columns=date&_columns=signature&_columns=product&_columns=version&_columns=build_id&_columns=platform

The sum of those would put it comfortably in the #2 top-crash slot
(after OOM | small) with >4% of all crashes
https://crash-stats.mozilla.com/topcrasher/products/Firefox/versions/33.1?days=7

Comment 1

[Milan Sreckovic [:milan] (needinfo for best results)]

Are these stacks helping point us to the right direction?

Comment 2

[Bas Schouten (:bas.schouten)]

I think this could possibly be the size bug we've been talking about.

Comment 3

[Jeff Muizelaar [:jrmuizel]]

Are they gone now?

Comment 4

[Jeff Muizelaar [:jrmuizel]]

Are these crashes there on trunk?

Comment 5

[Mats Palmgren (inactive)]

Hard to say; there are still some crashes in 36a1:
https://crash-stats.mozilla.com/search/?signature=~igd10umd32.dll&product=Firefox&version=36.0a1&_facets=signature&_columns=date&_columns=signature&_columns=product&_columns=version&_columns=build_id&_columns=platform

https://crash-stats.mozilla.com/search/?signature=~igd10iumd32.dll&product=Firefox&version=36.0a1&_facets=signature&_columns=date&_columns=signature&_columns=product&_columns=version&_columns=build_id&_columns=platform

but the volume is low...  I can't tell if that's b/c our code has improved
or that our Nightly users are using more modern hardware where the problem
doesn't occur.

Perhaps someone in QA can give you a better analysis based on the population
numbers/characteristics on each channel vs. the crash volume?

Comment 6

[Mats Palmgren (inactive)]

After looking through some of those reports, a couple of data points, fwiw:
bug 960692 in Nightly(36a1) BuildID: 20141113030201
bp-5dc07b41-864f-42d0-adac-91fd72141117
bug 1056627 Nightly(36a1) BuildID: 20141016030201
bp-d30e7a9a-33c7-4173-aac1-d5b2d2141112

bug 960692 appears to be the most common.

Comment 7

[u279076]

I'm closing this meta bug since all other bugs it was tracking are now closed and we're tracking numerous other Intel driver crashes in other bugs at this point. Please reopen this bug report if there's still use for this meta bug and please attach dependencies as required.