1232087 - Graphics driver crash leads to permanently black tabs

Status: RESOLVED FIXED

(Core :: Graphics, defect)

Description

[Brad Lassey [:blassey] (use needinfo?)]

While scrolling a web page windows reported that my graphics driver had crashed. All other applications seem to have recovered gracefully, but the content area of my Nightly tabs remained black. Neither force-reloading nor creating new tabs resolved the black tabs. Only shutting down and restarting the browser allowed me to view content.

my about:support
Application Basics
------------------

Name: Firefox
Version: 45.0a1
Build ID: 20151211030207
Update Channel: nightly
User Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:45.0) Gecko/20100101 Firefox/45.0
Multiprocess Windows: 1/1 (default: true)
Safe Mode: false

Crash Reports for the Last 3 Days
---------------------------------

All Crash Reports

Extensions
----------

Name: Firefox Hello
Version: 0.1
Enabled: true
ID: [e-mail]

Graphics
--------

Adapter Description: Intel(R) HD Graphics 530
Adapter Description (GPU #2): NVIDIA GeForce GTX 960M
Adapter Drivers: igdumdim64 igd10iumd64 igd10iumd64 igd12umd64 igdumdim32 igd10iumd32 igd10iumd32 igd12umd32
Adapter Drivers (GPU #2): nvd3dumx,nvwgf2umx,nvwgf2umx,nvwgf2umx nvd3dum,nvwgf2um,nvwgf2um,nvwgf2um
Adapter RAM: Unknown
Adapter RAM (GPU #2): 2048
Asynchronous Pan/Zoom: wheel input enabled
Device ID: 0x191b
Device ID (GPU #2): 0x139b
Direct2D Enabled: true
DirectWrite Enabled: true (10.0.10586.0)
Driver Date: 10-1-2015
Driver Date (GPU #2): 7-22-2015
Driver Version: 20.19.15.4300
Driver Version (GPU #2): 10.18.13.5362
GPU #2 Active: false
GPU Accelerated Windows: 1/1 Direct3D 11 (OMTC)
Subsys ID: 06e41028
Subsys ID (GPU #2): 06e41028
Supports Hardware H264 Decoding: Yes
Vendor ID: 0x8086
Vendor ID (GPU #2): 0x10de
WebGL Renderer: Google Inc. -- ANGLE (Intel(R) HD Graphics 530 Direct3D11 vs_5_0 ps_5_0)
windowLayerManagerRemote: true
AzureCanvasBackend: direct2d 1.1
AzureContentBackend: direct2d 1.1
AzureFallbackCanvasBackend: cairo
AzureSkiaAccelerated: 0

Important Modified Preferences
------------------------------

accessibility.typeaheadfind.flashBar: 0
browser.cache.disk.capacity: 358400
browser.cache.disk.filesystem_reported: 1
browser.cache.disk.smart_size.first_run: false
browser.cache.frecency_experiment: 4
browser.download.importedFromSqlite: true
browser.places.smartBookmarksVersion: 7
browser.sessionstore.upgradeBackup.latestBuildID: 20151211030207
browser.startup.homepage_override.buildID: 20151211030207
browser.startup.homepage_override.mstone: 45.0a1
dom.apps.reset-permissions: true
dom.mozApps.used: true
extensions.lastAppVersion: 45.0a1
gfx.direct3d.last_used_feature_level_idx: 0
gfx.driver-init.appVersion: 42.0
gfx.driver-init.deviceID: 0x191b
gfx.driver-init.driverVersion: 20.19.15.4300
gfx.driver-init.feature-d2d: true
gfx.driver-init.feature-d3d11: true
gfx.driver-init.status: 2
media.gmp-manager.buildID: 20151211030207
media.gmp-manager.lastCheck: 1449849266
media.hardware-video-decoding.failed: false
network.cookie.prefsMigrated: true
network.predictor.cleaned-up: true
places.database.lastMaintenance: 1449767564
places.history.expiration.transient_current_max_pages: 104858
plugin.disable_full_page_plugin_for_types: application/pdf
plugin.importedState: true
privacy.sanitize.migrateClearSavedPwdsOnExit: true
privacy.sanitize.migrateFx3Prefs: true
security.sandbox.content.tempDirSuffix: {4da67282-f63a-4962-92a0-b1a2adb8331f}
services.sync.declinedEngines:
services.sync.engine.addons: false
services.sync.engine.prefs: false
services.sync.engine.prefs.modified: false
services.sync.lastPing: 1449852716
services.sync.lastSync: Sat Dec 12 2015 00:24:40 GMT-0500 (Eastern Standard Time)
services.sync.numClients: 5
storage.vacuum.last.index: 1
storage.vacuum.last.places.sqlite: 1449767564
ui.osk.debug.keyboardDisplayReason: IKPOS: Rotation sensor not found.

Important Locked Preferences
----------------------------

JavaScript
----------

Incremental GC: true

Accessibility
-------------

Activated: true
Prevent Accessibility: 0

Library Versions
----------------

NSPR
Expected minimum version: 4.11
Version in use: 4.11

NSS
Expected minimum version: 3.21 Basic ECC
Version in use: 3.21 Basic ECC

NSSSMIME
Expected minimum version: 3.21 Basic ECC
Version in use: 3.21 Basic ECC

NSSSSL
Expected minimum version: 3.21 Basic ECC
Version in use: 3.21 Basic ECC

NSSUTIL
Expected minimum version: 3.21
Version in use: 3.21

Experimental Features
---------------------

Comment 1

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

Resizing the window helps?  Is this a SurfaceBook?

Comment 2

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

While in IRCCloud, Brad ran into this, and getting my message back fixed it, but here's what he had with the errors:

17:43 blassey: (#0) Assert	[D2D1.1] 4CreateBitmap failure Size(529,907) Code: 0x8899000c format 0
17:43 blassey: (#1181) Assert	[D2D1.1] 4CreateBitmap failure Size(16,16) Code: 0x8899000c format 0
17:43 blassey: (#1182) Assert	[D2D1.1] 4CreateBitmap failure Size(32,32) Code: 0x8899000c format 0
17:43 blassey: (#1183) Assert	[D2D1.1] 3CreateBitmap failure Size(682,293) Code: 0x8899000c format 0
17:43 blassey: (#1184) Assert	Failed to create DrawTarget, Type: 7 Size: Size(682,293)
17:43 blassey: (#1185) Assert	Failed 2 buffer db=0x0000000000000000 dw=0x0000000000000000 for 0, 0, 682, 293

Comment 3

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

Not a SurfaceBook, dell xps 9550

Comment 4

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

Seems rather related to bug 1232042

Comment 5

[Brad Lassey [:blassey] (use needinfo?)]

2 differences between the first and second black screen. I didn't see a driver crash notification for the second (though could have missed it) and in the second the chrome was also blank (thought it was white, not black).

Comment 6

[Brad Lassey [:blassey] (use needinfo?)]

just saw this again. Got the driver crash notification and my content area was black and my chrome area was white. Interestingly an ad started rendering in the middle of the content ares). Moving the window off screen and back on fixed it. If we can detect these driver crashes, perhaps we should just force a full window invalidation.

Comment 7

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

It is a D2DERR_RECREATE_TARGET error that we're getting, so that may work.  Bas, David, could we send some message/save some information when these types of messages happen so that we "eventually" invalidate everything?  There is still the "why did the driver crash" part, but we do know that there was a Windows 10 update which seems to have made that worse for us, as well as worse for Edge (anecdotally, at least.)

Comment 8

[David Anderson [:dvander] - inactive, e-mail if emergency]

(In reply to Milan Sreckovic [:milan] from comment #7)
> It is a D2DERR_RECREATE_TARGET error that we're getting, so that may work. 
> Bas, David, could we send some message/save some information when these
> types of messages happen so that we "eventually" invalidate everything? 
> There is still the "why did the driver crash" part, but we do know that
> there was a Windows 10 update which seems to have made that worse for us, as
> well as worse for Edge (anecdotally, at least.)

This is probably bug 1188019. Bas mentioned that we should get a WM_PAINT, but I don't see any evidence for that outside of DirectComposition. I'll do some testing to confirm.

Brad, does this happen with Direct2D disabled in about:config? (set "gfx.direct2d.use1_1" to false)

If we're TDRing on scroll that is bad either way.

Comment 9

[Bas Schouten (:bas.schouten)]

My suggestion right now as the most reliable way of attempting to fix these would be to simply schedule a full redraw somewhere close to (let's say 1-2 seconds after) detecting a driver crash. This should be harmless in the case of a false positive and potentially improve the situation significantly for other usecases.

Comment 10

[Brad Lassey [:blassey] (use needinfo?)]

I haven't seen this since I disabled Direct2D

Comment 11

[Brad Lassey [:blassey] (use needinfo?)]

(In reply to Brad Lassey [:blassey] (use needinfo?) from comment #10)
> I haven't seen this since I disabled Direct2D

Spoke too soon. Just saw the graphics driver crash (and the resulting black content/white chrome) with Direct2D disabled.

Comment 12

[Bob Owen (:bobowen)]

Attachment: bug1232087bobowenAboutSupport.txt

I've started seeing something very similar on my laptop on Windows 10.

I'll get the OS crash notification and then that tab will display just black.
Refreshing or navigating to most other sites still gives black.
HTML5 and flash video still display if triggered.
Google maps also works.
Resizing, moving window of screen and tab dragging do not fix it.
The tab drag thumbnail is rendered correctly though.

If I open a new tab this will sometimes work, so it looks like the problem is content process related.
Pages rendered in parent process work.

Attached my about:support output.

Comment 13

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

Bug 1188019 landed on 12/18, so it didn't completely eliminate this.

Comment 14

[Mark Hammond [:markh] [:mhammond]]

Same problem here (although the tabs are all white for me). I regularly see driver crashes similar to bug 911729 on Nightly and Firefox is the only app that doesn't recover. Last time this happened I had a locally built Firefox running and on stdout I saw:

> Crash Annotation GraphicsCriticalError: |[0][GFX1-]: Detected rendering device reset on refresh (t=4710.6) [GFX1-]: Detected rendering device reset on refresh
> Crash Annotation GraphicsCriticalError: |[0][GFX1-]: Detected rendering device reset on refresh (t=4710.6) |[1][GFX1-]: Detected rendering device reset on refresh (t=4710.6) [GFX1-]: Detected rendering device reset on refresh

Comment 15

[David Anderson [:dvander] - inactive, e-mail if emergency]

Mark, are you using E10s? If so, bug 1245765 should fix this.

Comment 16

[David Anderson [:dvander] - inactive, e-mail if emergency]

Could anyone with this problem try again now that bug 1245765 has landed?

Comment 17

[Brad Lassey [:blassey] (use needinfo?)]

(In reply to David Anderson [:dvander] from comment #16)
> Could anyone with this problem try again now that bug 1245765 has landed?

Unless I'm mistaken, there aren't STR to confirm with.

Comment 18

[Mark Hammond [:markh] [:mhammond]]

(In reply to David Anderson [:dvander] from comment #16)
> Could anyone with this problem try again now that bug 1245765 has landed?

Next time I get a driver crash I'll let you know, but I haven't seen anywhere near as many recently. I suspect Firefox caused the crashes - I was always interacting with it at the time - but that's difficult to prove. I suspect there may have been some other recent change that makes the crash less likely.

Comment 19

[Mark Hammond [:markh] [:mhammond]]

I just got a driver crash on 48.0a1 2016-03-10. After the crash the tabs all stayed black and didn't update even after switching to different tabs. Then after about 5 seconds I was told the content process crashed, with about:crashes showing bp-9c1debb8-98fc-42c2-b3da-af93d2160315 and bp-7e169334-a570-4bc3-bebd-359522160315, the latter of which looks like bug 1255711.

Comment 20

[Mark Hammond [:markh] [:mhammond]]

I just got a driver crash on 48.0a1 (2016-03-28) and Firefox recovered perfectly \o/

Comment 21

[David Anderson [:dvander] - inactive, e-mail if emergency]

Great! I think we should be good now, with bug 1255711 and friends fixed - if it happens again please file a new bug.