877629 - Opening http://elijahmanor.github.io/talks/angry-birds-javascript/index.html crashes the latest Nightly

Status: RESOLVED WORKSFORME

(Core :: Graphics, defect)

Description

[Saurabh Anand [:sawrubh]]

Attachment: about:support

STR :

Try opening http://elijahmanor.github.io/talks/angry-birds-javascript/index.html and then probably navigating quickly with the arrow keys.

Expected :
Smooth animations and movement

Actual :
Nightly crashes.

Comment 1

[Saurabh Anand [:sawrubh]]

This works smoothly on stable Firefox, Chrome and Chrome Canary.

Comment 2

[Saurabh Anand [:sawrubh]]

I'm able to reproduce it on a new profile.

Comment 3

[Saurabh Anand [:sawrubh]]

Here's the link to my crash stat : https://crash-stats.mozilla.com/report/index/bp-e28a897c-05d8-4fe3-8a51-536fd2130530

It stops crashing when I turn off Hardware Acceleration.

Comment 4

[Scoobidiver (away)]

No crash for me on Intel GMA4500HD.

Comment 5

[Robert Kaiser]

(In reply to Saurabh Anand [:sawrubh] from comment #1)
> This works smoothly on stable Firefox, Chrome and Chrome Canary.

Which version of Firefox is crashing and which version isn't?

Comment 6

[Saurabh Anand [:sawrubh]]

It's working perfectly fine in Firefox 21.0 but it's crashing in Firefox Nightly 24.0a1 (2013-06-05).

Comment 7

[Tracy Walker [:tracy]]

Is this a dupe of 839805? both have same crash signature.

Comment 8

[Robert Kaiser]

(In reply to Tracy Walker [:tracy] from comment #7)
> Is this a dupe of 839805? both have same crash signature.

We'd need to look if the whole crash stack (or a relevant part of it) is the same, and/or if the pattern of what is being done to graphics is the same on both bug 839805 and this one.
There could of course be a connection to the more general bug 805406 as well.

The bug here and bug 839805 both have actual cases/websites that cause crashes, it would be pretty helpful if we could find systems that can reproduce those and then get graphics developers access to such machines so they can debug and hopefully solve or work around the issue(s).

Comment 9

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

Bas, could you take a quick look at this and bug 839805?

Comment 10

[Bas Schouten (:bas.schouten)]

Yes.

Comment 11

[Olli Pettay [:smaug][[e-mail]]]

This or one of the variants (Bug 839805, Bug 805406) is by far the most common crasher
(excluding corrupt dumps).

Comment 12

[Scoobidiver (away)]

(In reply to Olli Pettay [:smaug] from comment #11)
> This or one of the variants (Bug 839805, Bug 805406) is by far the most
> common crasher
We only mark as topcrash the meta bug.

Comment 13

[Robert Kaiser]

(In reply to Olli Pettay [:smaug] from comment #11)
> This or one of the variants (Bug 839805, Bug 805406) is by far the most
> common crasher
> (excluding corrupt dumps).

"most common" in what? And what criteria do you use for assigning topcrasher here?


(In reply to Scoobidiver from comment #12)
> (In reply to Olli Pettay [:smaug] from comment #11)
> > This or one of the variants (Bug 839805, Bug 805406) is by far the most
> > common crasher
> We only mark as topcrash the meta bug.

I can't even decipher what your statement should mean here. Meta bugs are *not* supposed to be marked as topcrash unless it really really makes sense.

Comment 14

[Olli Pettay [:smaug][[e-mail]]]

(In reply to Robert Kaiser (:[e-mail]) from comment #13)
> (In reply to Olli Pettay [:smaug] from comment #11)
> > This or one of the variants (Bug 839805, Bug 805406) is by far the most
> > common crasher
> > (excluding corrupt dumps).
> 
> "most common" in what? And what criteria do you use for assigning topcrasher
> here?

I was looking the crash-stats' topcrashers.
gfxContext::PushClipsToDT(mozilla::gfx::DrawTarget*) is #2 (6.63%) and #1 is corrupted dump.
#3 is 3.77%.

Comment 15

[Scoobidiver (away)]

(In reply to Olli Pettay [:smaug] from comment #14)
> gfxContext::PushClipsToDT(mozilla::gfx::DrawTarget*) is #2 (6.63%) and #1 is
> corrupted dump.
How do you know the ratio of crashes with the STR of comment 0 vs. crashes with this signature? You can mark it as a top crasher whether it fixes 14% (22947/3119) of crashes with this signature.

(In reply to Robert Kaiser (:[e-mail]) from comment #13)
> I can't even decipher what your statement should mean here. Meta bugs are
> *not* supposed to be marked as topcrash unless it really really makes sense.
Meta bugs depend on bugs with STR or testcase but crash stats don't give you the ratio so you can't mark depending bugs as topcrash, only the meta one. But you can track those depending bugs for a version.

Comment 16

[Bas Schouten (:bas.schouten)]

I suspect that this will be fixed by the patch in bug 877700. I can't verify that right now and I can't push the patch there, I've asked someone else to and we should check if that fixes this bug as well.

Comment 17

[Robert Kaiser]

(In reply to Bas Schouten (:bas.schouten) from comment #16)
> I suspect that this will be fixed by the patch in bug 877700. I can't verify
> that right now and I can't push the patch there, I've asked someone else to
> and we should check if that fixes this bug as well.

Hmm, could you point out in that bug then how it is supposed to move forward? Looks like it has been dormant for almost 2 months now. :(

Comment 18

[Scoobidiver (away)]

:sawrubh, do you still crash in the latest Nightly after the landing of bug 877700's patch?

Comment 19

[Saurabh Anand [:sawrubh]]

I don't see a crash now.