978492 - crash in gfxContext::gfxContext(mozilla::gfx::DrawTarget*, mozilla::gfx::PointTyped<mozilla::gfx::UnknownUnits> const&)

Status: RESOLVED DUPLICATE of bug 952721

(Firefox for Android Graveyard :: Toolbar, defect)

Description

[Kevin Brosnan]

This bug was filed from the Socorro interface and is 
report bp-c8e4fdbb-b60b-4e0b-abc6-52f852140301.
=============================================================

Crashed scrolling on the site listed in the URL field. Nexus 5 Android 4.4.2

Comment 1

[Kevin Brosnan [Ex-Mozilla]]

This might be a regression worth checking for one.

Comment 2

[Robert Kaiser]

IMHO we should check how far this might be related with bug 952721 or not.

Comment 3

[u279076]

This is showing up explosive on Firefox 28 Beta desktop builds as well starting with Firefox 28.0b9:
https://crash-analysis.mozilla.com/rkaiser/2014-03-09/2014-03-09.firefox.28.explosiveness.html

Although, it currently sits at #168 on Desktop:
https://crash-stats.mozilla.com/report/list?signature=gfxContext%3A%3AgfxContext%28mozilla%3A%3Agfx%3A%3ADrawTarget*%2C%20mozilla%3A%3Agfx%3A%3APointTyped%3Cmozilla%3A%3Agfx%3A%3AUnknownUnits%3E%20const%26%29

One comment mentions pinching and zooming on imgur.com

Comment 4

[u279076]

Here is the pushlog from Beta 8 to Beta 9:
http://hg.mozilla.org/releases/mozilla-beta/pushloghtml?fromchange=FIREFOX_28_0b8_RELEASE&tochange=FIREFOX_28_0b9_RELEASE

Comment 5

[Benjamin Smedberg]

The desktop issue is reasonably likely to be from bug 970294, which changed the signature of gfxContext::gfxContext. What we need to see though is whether an equivalent signature from beta 8 disappeared, probably [@ gfxContext::gfxContext(mozilla::gfx::DrawTarget*)].

If we just moved it around, this probably doesn't need to block.

Comment 6

[u279076]

[@ gfxContext::gfxContext(mozilla::gfx::DrawTarget*)] hasn't disappeared completely but has dropped off. It's down 8 positions in the last three days to #31 (0.32% [-0.16%]).

Comment 7

[u279076]

Addendum, I see no crashes for [@ gfxContext::gfxContext(mozilla::gfx::DrawTarget*)] against Firefox 28.0b9 specifically so I expect to see this signature continue to drop in rank as time goes on.

Source:
https://crash-stats.mozilla.com/report/list?signature=gfxContext%3A%3AgfxContext%28mozilla%3A%3Agfx%3A%3ADrawTarget*%29&product=Firefox&query_type=contains&range_unit=weeks&process_type=any&hang_type=any&date=2014-03-10+18%3A00%3A00&range_value=1#reports

Comment 8

[Benjamin Smedberg]

ok no need to track

Comment 9

[investfund2007]

I have win7 and FF crashes fairly often in gfxContext::gfxContext(mozilla::gfx::DrawTarget*, mozilla::gfx::PointTyped<mozilla::gfx::UnknownUnits> const&). It started with version 27 and continued on version 28

Comment 10

[u279076]

(In reply to investfund2007 from comment #9)
> I have win7 and FF crashes fairly often in
> gfxContext::gfxContext(mozilla::gfx::DrawTarget*,
> mozilla::gfx::PointTyped<mozilla::gfx::UnknownUnits> const&). It started
> with version 27 and continued on version 28

It would be good to get more information from you about this crash. Specifically, what are you doing before the crash, what hardware are you using, and provide links to your crash reports from about:crashes.

Comment 11

[Rob]

Another bp-f28a55a6-841d-4b56-b9f6-943002140316 .

Comment 12

[Rob]

Nightly 31.0a1 (2014-03-18) Android:
bp-9f773aba-3573-4999-91c8-81b722140319 "Google Street View", in "Desktop Mode". 

PS: The prior BP was "Google Street View", in "Mobile Mode".

Comment 13

[u279076]

(In reply to Rob from comment #12)
> Nightly 31.0a1 (2014-03-18) Android:
> bp-9f773aba-3573-4999-91c8-81b722140319 "Google Street View", in "Desktop
> Mode". 

I can reproduce this easily on a Nexus 5 with that use case. In other words:

1. Go to maps.google.com
2. Click the 2nd button from the right (looks like a page)
3. Scroll down to the bottom of the page and click the Classic link
4. Double tap the map to zoom in
5. Continue to double tap until you crash
> Nightly will crash after zooming in a few levels

Kevin, let me know if that's helpful to you tracking down a regression window.

Comment 14

[investfund2007]

(In reply to Anthony Hughes, QA Mentor (:ashughes) from comment #10)
> (In reply to investfund2007 from comment #9)
> > I have win7 and FF crashes fairly often in
> > gfxContext::gfxContext(mozilla::gfx::DrawTarget*,
> > mozilla::gfx::PointTyped<mozilla::gfx::UnknownUnits> const&). It started
> > with version 27 and continued on version 28
> 
> It would be good to get more information from you about this crash.
> Specifically, what are you doing before the crash, what hardware are you
> using, and provide links to your crash reports from about:crashes.

https://crash-stats.mozilla.com/report/index/48acd798-573c-43cf-9425-1b2d42140319
I have notebook thinkpad t420, 2 video cards: integrated Intel and nvidia 4200. Video outputed only to external monitor.
Crash occurs when I try to watch youtube videos and switching between tabs

Comment 15

[Rob]

(In reply to Anthony Hughes, QA Mentor (:ashughes) from comment #13)
> (In reply to Rob from comment #12)
> ...
> Kevin, let me know if that's helpful to you tracking down a regression
> window.


Re: Regression Window (but not the most helpful 'window length'):

Half a year ago it worked WONDERFULLY, you (or at least I, on my HTC One (w/1080P)) could use two fingers (one at a time) and literally RUN down the street like a madman (really fast).

Now (past few weeks) it is 'dog slow', I'd be lucky to be able to turn around and chase my tail. Really clumsy, and can not run to save my life.


Trying to crash it does not work 100% of the time, sometimes (rarely) you can get a 15 min. session out of it (and close it when finished) without problems.


This may be related to the 'GL problems' mentioned in the BrowserMark Testsuite Bug 758654 .

Comment 16

[Rob]

Socorro says related Bug 941800 [mozilla::gl::SwapRAndBComponents] on Aurora bp-4f8266a3-26ed-4284-9d0c-d77f72140321 caused when using 'GSV'.

Comment 17

[Kevin Brosnan [Ex-Mozilla]]

This morphed from gfxContext::gfxContext(mozilla::gfx::DrawTarget*) to this signature during the development of Feb-28 to March-1 build.

Comment 18

[Kevin Brosnan [Ex-Mozilla]]

Anthony if there is any follow up for desktop it should happen in a new bug. But I suspect comment 17 applies to desktop as well.

Comment 19

[u279076]

(In reply to Kevin Brosnan [:kbrosnan] from comment #18)
> Anthony if there is any follow up for desktop it should happen in a new bug.
> But I suspect comment 17 applies to desktop as well.

Thanks. FWIW we were tracking bug 952721 as affecting desktop as well. This is still significant volume on Desktop but Android is the most impacted so far. Before we split that bug out I'd like to see if any resolution on Android helps this on Desktop. Considering this is in Core it's quite possible a fix to Android might improve Desktop.

Comment 20

[Robert Kaiser]

(In reply to Anthony Hughes, QA Mentor (:ashughes) from comment #19)
> Thanks. FWIW we were tracking bug 952721 as affecting desktop as well. This
> is still significant volume on Desktop but Android is the most impacted so
> far. Before we split that bug out I'd like to see if any resolution on
> Android helps this on Desktop. Considering this is in Core it's quite
> possible a fix to Android might improve Desktop.

I think we probably should have different bugs for this on desktop and mobile, unless gfx devs think otherwise.