Closed Bug 722269 Opened 8 years ago Closed 7 years ago

Firefox crash @ mozilla::gfx::DrawTargetD2D::DrawSurface

Categories

(Core :: Graphics, defect, critical)

All
Windows 7
defect
Not set
critical

Tracking

()

RESOLVED FIXED
mozilla18
Tracking Status
firefox17 --- fixed

People

(Reporter: brunoaiss, Assigned: bas.schouten)

References

()

Details

(Keywords: crash, reproducible, Whiteboard: [qa?])

Crash Data

Attachments

(1 file)

Firefox had already been opened for about 24h.
I had, at the moment, about 200 tabs in which 30 were loaded.
I had been in the same tab group for a while (4-5 hours).
I tried to change the tag group so I went to the panorama view to select another tab.
After pressing ctrl+shift+E firefox hanged for about 2 seconds and then crashed.
In order to reproduce you need to:
 * Have, at least, 4 groups with tabs in them.
 * Work in a tab group for a while (1h should be enough)
 * Load and change url of some pages (go to google, search, read articles, view images...)
 * Have firefox open for some hours. Seems like you need more than 16h

.
Priority: -- → P4
Does this crash show up in about:crashes? We need a crash ID to investigate that further and to see if Panorama is even the culprit.
Keywords: crash
This is one:
https://crash-stats.mozilla.com/report/index/bp-b9d0a1cb-b508-4717-be39-7b3392120127
here's another:
https://crash-stats.mozilla.com/report/index/bp-57c7eb1e-037e-41ba-a566-afd012120128

I think this is also one caused by the same reason:
https://crash-stats.mozilla.com/report/index/bp-763857aa-ce04-4de1-ad5a-a83672120124

There are more in about:crashes but when I click them the page shows it does not recognize the id.
Does it happen in Safe Mode (see https://support.mozilla.org/en-US/kb/Safe%20Mode)?
Does it happen with a new profile (see https://support.mozilla.org/en-US/kb/Managing-profiles)?

It's a low volume crash with only 42 crashes in 9.0.1 over the last week.

More reports at:
https://crash-stats.mozilla.com/report/list?signature=DrawingContext%3A%3ADrawBitmap%28ID2D1Bitmap*%2C+D2D_RECT_F+const*%2C+float%2C+D2D1_BITMAP_INTERPOLATION_MODE%2C+D2D_RECT_F+const*%29
Status: UNCONFIRMED → NEW
Crash Signature: [@ DrawingContext::DrawBitmap(ID2D1Bitmap*, D2D_RECT_F const*, float, D2D1_BITMAP_INTERPOLATION_MODE, D2D_RECT_F const*)]
Component: Panorama → Canvas: 2D
Ever confirmed: true
Product: Firefox → Core
QA Contact: panorama → canvas.2d
Summary: Firefox crash after trying to open tab group → Firefox crash @ mozilla::gfx::DrawTargetD2D::DrawSurface after trying to open tab group
Target Milestone: Firefox 10 → ---
Version: 9 Branch → 10 Branch
Component: Canvas: 2D → Graphics
QA Contact: canvas.2d → thebes
(In reply to Scoobidiver from comment #3)
> Does it happen in Safe Mode (see
> https://support.mozilla.org/en-US/kb/Safe%20Mode)?
Yes
> Does it happen with a new profile (see
> https://support.mozilla.org/en-US/kb/Managing-profiles)?

Didn't test. Need me to test?
(In reply to brunoaiss from comment #4)
> Didn't test. Need me to test?
Yes. Please.
I hit this twice with Aurora 17 (2012-09-18). Both times, I was trying to load this URL with pdf.js 0.4.11:

https://wiki.postgresql.org/images/7/73/Range-types-pgopen-2012.pdf
I can reproduce in Nightly: bp-c7a5ec47-7e0f-417e-b3e6-d5af92120920.
Keywords: reproducible
Priority: P4 → --
Hardware: x86_64 → All
Version: 10 Branch → Trunk
Summary: Firefox crash @ mozilla::gfx::DrawTargetD2D::DrawSurface after trying to open tab group → Firefox crash @ mozilla::gfx::DrawTargetD2D::DrawSurface
Assignee: nobody → bas.schouten
Status: NEW → ASSIGNED
Comment on attachment 663164 [details] [diff] [review]
Make DrawTargetD2D::DrawSurface deal with more surfaces

Review of attachment 663164 [details] [diff] [review]:
-----------------------------------------------------------------

::: gfx/2d/DrawTargetD2D.cpp
@@ +295,3 @@
>      {
> +      RefPtr<DataSourceSurface> srcSurf =
> +        aSurface->GetDataSurface();

same line
Attachment #663164 - Flags: review?(joe) → review+
https://hg.mozilla.org/mozilla-central/rev/7ed5b4e7e4fc
Status: ASSIGNED → RESOLVED
Closed: 7 years ago
Resolution: --- → FIXED
Target Milestone: --- → mozilla18
Can we get this backported to Aurora? Seems low-risk enough.
Comment on attachment 663164 [details] [diff] [review]
Make DrawTargetD2D::DrawSurface deal with more surfaces

Review of attachment 663164 [details] [diff] [review]:
-----------------------------------------------------------------

This is very low risk and probably a good idea to get in Aurora now.
Attachment #663164 - Flags: approval-mozilla-aurora?
Comment on attachment 663164 [details] [diff] [review]
Make DrawTargetD2D::DrawSurface deal with more surfaces

Low risk fix for reproducible crash, let's get it on Aurora.
Attachment #663164 - Flags: approval-mozilla-aurora? → approval-mozilla-aurora+
This crash still appears in Socorro, on Firefox 17 beta 1 and the 10/14 Firefox 19 Nightly.

The stack traces are pretty much the same as those in reports listed in this bug:
https://crash-stats.mozilla.com/report/index/9b8c6dba-0c11-4062-a948-86e1a2121016
https://crash-stats.mozilla.com/report/index/7b578df6-1485-478d-99ec-d97112121016
QA Contact: ioana.budnar
I can't reproduce it with the STR in comment 6 in 17.0b1+PDF Viewer 0.4.11, Aurora and Nightly.
An overwhelming majority (96%) of the crashes with this signature in the last week are happening on Firefox 16.0.1 or earlier versions. Can we consider this verified based on the fact that crashes are trending way down and the fact that the test in comment 6 no longer reproduces?
Bas, Lukas, can one of you look over the crashes in comment 17 and let me know if they are the same crash or not?
Keywords: verifyme
Whiteboard: [qa?]
(In reply to Ioana Budnar [QA] from comment #20)
> Bas, Lukas, can one of you look over the crashes in comment 17 and let me
> know if they are the same crash or not?

It's very hard to say considering in how many ways these codepaths get hit.
(In reply to Bas Schouten (:bas.schouten) from comment #21)
> (In reply to Ioana Budnar [QA] from comment #20)
> > Bas, Lukas, can one of you look over the crashes in comment 17 and let me
> > know if they are the same crash or not?
> 
> It's very hard to say considering in how many ways these codepaths get hit.

Would you consider safe marking this bug as verified in that case? Please also see comment 19 for the details there.
You need to log in before you can comment on or make changes to this bug.