Closed Bug 993067 Opened 10 years ago Closed 10 years ago

crash from MOZ_ASSERT(gl())) at GrallocTextureHost.cpp:349 on Peak

Categories

(Core :: Graphics: Layers, defect)

ARM
Gonk (Firefox OS)
defect
Not set
normal

Tracking

()

RESOLVED FIXED
mozilla31

People

(Reporter: kats, Assigned: nical)

Details

Attachments

(2 files)

Attached file Backtrace of crash
Running latest m-c code on a Peak. I have export B2G_DEBUG=1 and export B2G_DUMP_PAINTING=1 enabled in my .userconfig file, and I also turned on layer dumping in the compositor by changing gDumpCompositorTree to true at [1]. However, running this results in a crash and device reboot when I unlock the homescreen on first start. The backtrace is attached.

[1] http://mxr.mozilla.org/mozilla-central/source/gfx/layers/ipc/CompositorParent.cpp?rev=6dc852777a4d#645
I see this crash on buri too - note, I only see it with the layers.dump pref enabled, with just paint-list dumping via the environment variable, things are fine (as expected - the backtrace shows it crashing while trying to readback the image data).
hm weird, not sure why I put this assertion here. IsValid() will return false if there is no gl context and we shouldn't crash for something that is just a debug helper. Let's just remove the assertion.
Assignee: nobody → nical.bugzilla
Attachment #8403091 - Flags: review?(chrislord.net)
Comment on attachment 8403091 [details] [diff] [review]
remove the assertion

Review of attachment 8403091 [details] [diff] [review]:
-----------------------------------------------------------------

Yup, seems fine to me - is this function explicitly only for debugging? If so, is that documented? I have some doubts as to whether it will work correctly or not with tiles...
Attachment #8403091 - Flags: review?(chrislord.net) → review+
(In reply to Chris Lord [:cwiiis] from comment #4)
> Comment on attachment 8403091 [details] [diff] [review]
> remove the assertion
> 
> Review of attachment 8403091 [details] [diff] [review]:
> -----------------------------------------------------------------
> 
> Yup, seems fine to me - is this function explicitly only for debugging? If
> so, is that documented? I have some doubts as to whether it will work
> correctly or not with tiles...

It should be documented (looks like it's not). This functionality is by definition always causing a read-back, so we can't afford to rely on it for anything but debugging anyways.
https://hg.mozilla.org/mozilla-central/rev/efbe86094565
Status: NEW → RESOLVED
Closed: 10 years ago
Resolution: --- → FIXED
Target Milestone: --- → mozilla31
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: