Open Bug 722030 Opened 8 years ago Updated 3 years ago

Crash in nsFrame::DisplayBorderBackgroundOutline with AMD Radeon HD 6xxx series


(Core :: Layout, defect, critical)

37 Branch
Windows 7
Not set



Tracking Status
firefox37 - wontfix
firefox38 + wontfix
firefox38.0.5 + wontfix
firefox39 + wontfix
firefox40 - wontfix
firefox48 --- wontfix
firefox49 --- wontfix
firefox-esr45 --- affected
firefox50 --- wontfix
firefox51 --- wontfix
firefox52 --- wontfix
firefox53 --- wontfix


(Reporter: scoobidiver, Unassigned)


(Blocks 1 open bug)


(Keywords: crash, regression, Whiteboard: [platform-rel-AMD])

Crash Data

It's #23 top crasher in 10.0b6.
It's probably related to bug 700288, bug 722024 and bug 722025 as it happens mainly with AMD Radeon HD 6xxx serie GPUs.

As its crash signature is not moving like bug 722024 and bug 722025, it's easier to track across versions.
It seems to be a 10.0 regression and first appeared in 10.0a1/20111002 with an Intel GPU and an old driver: bp-764592be-48a7-4144-bd81-8b63d2111006.

Signature 	nsFrame::DisplayBorderBackgroundOutline(nsDisplayListBuilder*, nsDisplayListSet const&, bool) More Reports Search
UUID	0295b590-0c08-4e10-9d78-7e5532120128
Date Processed	2012-01-28 02:05:01
Uptime	11262
Last Crash	1.1 days before submission
Install Age	7.4 hours since version was first installed.
Install Time	2012-01-27 18:41:40
Product	Firefox
Version	10.0
Build ID	20120123235200
Release Channel	beta
OS	Windows NT
OS Version	6.1.7601 Service Pack 1
Build Architecture	x86
Build Architecture Info	AuthenticAMD family 20 model 1 stepping 0
Crash Address	0xffffffff91ac7e09
App Notes 	
AdapterVendorID: 1002, AdapterDeviceID: 9803, AdapterSubsysID: 00000000, AdapterDriverVersion: 8.792.0.0
D2D? D2D+
DWrite? DWrite+
D3D10 Layers? D3D10 Layers+
Processor Notes 	This dump is too long and has triggered the automatic truncation routine
EMCheckCompatibility	True

Frame 	Module 	Signature [Expand] 	Source
0 	xul.dll 	nsFrame::DisplayBorderBackgroundOutline 	layout/generic/nsFrame.cpp:1347
1 	xul.dll 	nsInlineFrame::BuildDisplayList 	layout/generic/nsInlineFrame.cpp:188
2 	xul.dll 	nsIFrame::BuildDisplayListForChild 	layout/generic/nsFrame.cpp:1995
3 	xul.dll 	nsInlineFrame::BuildDisplayList 	layout/generic/nsInlineFrame.cpp:188
4 	xul.dll 	nsIFrame::BuildDisplayListForChild 	layout/generic/nsFrame.cpp:1995
5 	xul.dll 	nsBlockFrame::BuildDisplayList 	layout/generic/nsBlockFrame.cpp:6228
6 	xul.dll 	nsIFrame::BuildDisplayListForChild 	layout/generic/nsFrame.cpp:1995
7 	xul.dll 	nsBlockFrame::BuildDisplayList 	layout/generic/nsBlockFrame.cpp:6228

More reports at:*%2C%20nsDisplayListSet%20const%26%2C%20bool%29
Depends on: 722538
Depends on: 755974
[Tracking Requested - why for this release]:

This has been spiking over the last few days on Firefox 37.0.2, it's #16 with 0.9% of all crashes in yesterday's data for that release.*%2C%20nsDisplayListSet%20const%26%2C%20bool%29#tab-sigsummary says that the main gfx adapters for this crash are:

Advanced Micro Devices, Inc. 	AMD E-350 AMD Radeon HD 6310 series	1863 	62.706 %
Advanced Micro Devices, Inc. 	2411E6FE Display driver	553 	18.613 %
Advanced Micro Devices, Inc. 	AMD Radeon HD 6310 Graphics AMD Radeon HD 6310 Gr AMD Radeon HD 6310 Graphics	236 	7.943 %
[Tracking Requested - why for this release]:

Yep, seeing this in user feedback as well. Since 37 is so close to the end of the cycle, a fix in 38 would be nice
Version: 10 Branch → 37 Branch
Flags: needinfo?(kairo)
Given that we're scheduled to release 38 in 5 days, I'm going to set this as wontfix for 37. Looks like this does affect 38 and 39 albeit in lower quantities. I have nommed for tracking and will let Sylvestre and Liz make the call on whether they want to track this for these releases.
Hoping to help Sylvestre and Liz, I went ahead and tracked for 38, 38.0.5, 39, and 40, since it's a week later.
Tyler, what did you ni? me for here? I can't see any question I need to answer.
Flags: needinfo?(kairo)
Too late for 38.0.5 but happy to take a fix for 39
Kairo, this is popping up regularly on the forums, just wanted your help tracking it down
(In reply to Tyler Downer [:Tyler] from comment #8)
> Kairo, this is popping up regularly on the forums, just wanted your help
> tracking it down

This is a small-volume crash that we had for quite some time. I don't know any specific on tracking it down, it's not even exclusive to the Radeon HD 6xxx series, it's just hit there more frequently than on others. If anything can help there, it's an engineer taking a deeper look into what's going on in those dumps/stacks.
Liz, currently 37 crashes in 39 b1, compared to 160 in 37.0.2. Around 1 in 40/41.
Marking wontfix for 39. I'm not sure this is worth tracking. It spikes intermittently but no one is working on it.

David is there anything that looks actionable here?  Or can you suggest someone who might take a look?
Flags: needinfo?(dmajor)
This is another instance of the "the AMD bug" aka bug 772330.

Normally we'd see this particular crash as nsStyleContext::DoGetStyleBorder, but it got inlined into the caller on this particular build.

Bug 1155836 has made an attempt at fixing these. I don't want to claim victory just yet, but I am hopeful. Please alert me if you see a new spike of "AuthenticAMD family 20" on any channel with buildid greater than 2015-05-08.
Flags: needinfo?(dmajor)
Not tracking the amd bug...
Crash Signature: [@ nsFrame::DisplayBorderBackgroundOutline(nsDisplayListBuilder*, nsDisplayListSet const&, bool)] → [@ nsFrame::DisplayBorderBackgroundOutline(nsDisplayListBuilder*, nsDisplayListSet const&, bool)] [@ nsFrame::DisplayBorderBackgroundOutline]
Crash volume for signature 'nsFrame::DisplayBorderBackgroundOutline':
 - nightly (version 51): 1 crash from 2016-08-01.
 - aurora  (version 50): 2 crashes from 2016-08-01.
 - beta    (version 49): 43 crashes from 2016-08-02.
 - release (version 48): 51 crashes from 2016-07-25.
 - esr     (version 45): 10 crashes from 2016-05-02.

Crash volume on the last weeks (Week N is from 08-22 to 08-28):
            W. N-1  W. N-2  W. N-3
 - nightly       0       1       0
 - aurora        1       1       0
 - beta         13       1       0
 - release      16      12       6
 - esr           2       4       1

Affected platform: Windows

Crash rank on the last 7 days:
           Browser   Content     Plugin
 - nightly
 - aurora            #1014
 - beta    #461      #514
 - release #947      #1065
 - esr
Wontfix for beta, not a concern right now
Calixte here is another for your consideration.  Maybe if it's a lower bug number/from years ago/untouched for a year, set the bar higher, for the # of crashes or crash rank that our bot will mark affected.  I'm not sure where the line is.
Clearing the ni on calixte, bugs about the bot should be reported here:
Flags: needinfo?(cdenizet)
platform-rel: --- → ?
Whiteboard: [platform-rel-AMD]
platform-rel: ? → ---
You need to log in before you can comment on or make changes to this bug.