1204223 - [e10s] With Nvidia GeForce driver 355.98, when computer wakes up from sleep, all Firefox tabs have the black screen of death

Status: RESOLVED WORKSFORME

(Core :: Graphics, defect)

Description

[Josh Tumath]

User Agent: Mozilla/5.0 (Windows NT 10.0; WOW64; rv:42.0) Gecko/20100101 Firefox/42.0
Build ID: 20150911004112

Steps to reproduce:

Using a Windows 10 PC with an Nvidia GeForce GTX 960 graphics card and the latest driver (355.82)...

1. Open Firefox and open a few tabs with some pages
2. Leave the computer for a while to go into sleep mode
3. Return after a while and resume from sleep
# A notification appears saying "nvidia driver 355.82 stopped responding and has recovered"


Actual results:

When going back onto Firefox, every tab has the "white screen of death" (a spinner icon on a white background). However, the chrome still works fine, and when opening a new tab and visiting a page, that works too.

Comment 1

[Josh Tumath]

Firefox might not necessarily be the cause of the GPU driver failure (I'm still trying to determine that - it's hard to reproduce), but either way it seems that the e10s process that the tabs are running in does not recover well from the driver failure.

Comment 2

[Loic]

Could you type about:support in the location bar and paste the section "graphics", please.

Comment 3

[Josh Tumath]

Adapter Description	NVIDIA GeForce GTX 960
Adapter Drivers	nvd3dumx,nvwgf2umx,nvwgf2umx,nvwgf2umx nvd3dum,nvwgf2um,nvwgf2um,nvwgf2um
Adapter RAM	4095
Asynchronous Pan/Zoom	none
Device ID	0x1401
Direct2D Enabled	true
DirectWrite Enabled	true (10.0.10240.16430)
Driver Date	8-25-2015
Driver Version	10.18.13.5582
GPU #2 Active	false
GPU Accelerated Windows	1/1 Direct3D 11 (OMTC)
Subsys ID	140110de
Supports Hardware H264 Decoding	true
Vendor ID	0x10de
WebGL Renderer	Google Inc. -- ANGLE (NVIDIA GeForce GTX 960 Direct3D11 vs_5_0 ps_5_0)
windowLayerManagerRemote	true
AzureCanvasBackend	direct2d 1.1
AzureContentBackend	direct2d 1.1
AzureFallbackCanvasBackend	cairo
AzureSkiaAccelerated	0

Comment 4

[Loic]

Could you test with e10s disabled?

Comment 5

[Josh Tumath]

(In reply to Loic from comment #4)
> Could you test with e10s disabled?

So far I have not been able to reproduce it with e10s disabled, but I'll keep trying.

Comment 6

[Loic]

Ok, let's flag the bug about e10s.

Comment 7

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

Possibly a duplicate, but definitely related to bug 1183391.  Josh, the about:support in comment 3, was that after the error occured, or in a fresh session?  If in a fresh session, could you reproduce the problem then append the about:support?

Comment 8

[Josh Tumath]

(In reply to Milan Sreckovic [:milan] from comment #7)
> Possibly a duplicate, but definitely related to bug 1183391.  Josh, the
> about:support in comment 3, was that after the error occured, or in a fresh
> session?  If in a fresh session, could you reproduce the problem then append
> the about:support?

It was from a fresh session. But I haven't specifically tried to reproduce this bug from hibernation, so I will try that tomorrow with and without e10s enabled.

Comment 9

[Josh Tumath]

I was not able to reproduce this on Dev Edition, but I could reproduce the driver failure when hibernating while running Nightly. However, the tabs did recover just fine and there were no graphical errors that I saw. Also, I could not reproduce it a second time while running Nightly.

At the moment, it's still too hard to tell if the driver failure is caused by Firefox or by a fault with the driver itself. But either way, the tab throbber of death is no longer appearing on each page, nor are there any graphical issues in the chrome process, so maybe something has been fixed and this bug should be INVALID for now and I'll reopen it later if I can reproduce it again?

Comment 10

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

We have been trying to recover better from the driver resets, though it's probably too much to hope for that this is fully resolved.  I agree that we should make it invalid or worksforme until it comes back.

Comment 11

[Josh Tumath]

I've started to experience this bug again, but this time with a black screen of death on each tab (but the chrome process looks ok). Should we reopen this bug or should I post a new bug?

Comment 12

[u279076]

(In reply to Josh Tumath from comment #11)
> I've started to experience this bug again, but this time with a black screen
> of death on each tab (but the chrome process looks ok). Should we reopen
> this bug or should I post a new bug?

Does that happen when you're resuming from sleep? If so then yes, please reopen this bug report.

Thanks

Comment 13

[Josh Tumath]

(In reply to Anthony Hughes, QA Mentor (:ashughes) from comment #12)
> Does that happen when you're resuming from sleep? If so then yes, please
> reopen this bug report.

Done. Just to reiterate: after resuming from sleep, sometimes the graphics driver will restart, and all of the Firefox tabs appear black (but the main thread containing the chrome seems to be fine).

Comment 14

[u279076]

Thanks Josh. You originally noted this issue with NVIDIA v355.82 (the latest WHQL driver at the time). I'm wondering if you could try one of the earlier Windows 10 WHQL drivers from NVIDIA to confirm if this is a regression in their driver. In the meantime I'm nominating this to track for shipping e10s support.

Comment 15

[George Wright (:gw280) (needinfo me!)]

I have a GTX 970 machine running Windows 10. I'll see if I can reproduce this.

Comment 16

[NVD]

http://www.geforce.com/whats-new/articles/geforce-358-50-whql-driver-released

Please test 358.50 WHQL driver.

Comment 17

[Josh Tumath]

(In reply to NVD from comment #16)
> http://www.geforce.com/whats-new/articles/geforce-358-50-whql-driver-released
> 
> Please test 358.50 WHQL driver.

Haven't experienced any graphics failures yet with the new driver, but I'll let you know if it happens in the next few days.

Comment 18

[George Wright (:gw280) (needinfo me!)]

I can't reproduce this using my Win10/GTX670 machine, running the NVIDIA drivers 355.98. This might be an issue with Maxwell chipsets, maybe?

Application Basics
------------------

Name: Firefox
Version: 44.0a1
Build ID: 20151007030205
Update Channel: nightly
User Agent: Mozilla/5.0 (Windows NT 10.0; WOW64; rv:44.0) Gecko/20100101 Firefox/44.0
Multiprocess Windows: 1/1 (default: true)
Safe Mode: false

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

All Crash Reports

Extensions
----------

Name: Adobe Acrobat - Create PDF
Version: 1.1
Enabled: false
ID: [e-mail]

Graphics
--------

Adapter Description: NVIDIA GeForce GTX 670
Adapter Drivers: nvd3dumx,nvwgf2umx,nvwgf2umx,nvwgf2umx nvd3dum,nvwgf2um,nvwgf2um,nvwgf2um
Adapter RAM: 2048
Asynchronous Pan/Zoom: wheel input enabled
Device ID: 0x1189
Direct2D Enabled: true
DirectWrite Enabled: true (10.0.10240.16430)
Driver Date: 9-13-2015
Driver Version: 10.18.13.5598
GPU #2 Active: false
GPU Accelerated Windows: 1/1 Direct3D 11 (OMTC)
Subsys ID: 26783842
Supports Hardware H264 Decoding: Yes
Vendor ID: 0x10de
WebGL Renderer: Google Inc. -- ANGLE (NVIDIA GeForce GTX 670 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.disk.smart_size.use_old_max: false
browser.cache.frecency_experiment: 2
browser.download.importedFromSqlite: true
browser.download.manager.alertOnEXEOpen: true
browser.places.smartBookmarksVersion: 7
browser.sessionstore.upgradeBackup.latestBuildID: 20151007030205
browser.startup.homepage_override.buildID: 20151007030205
browser.startup.homepage_override.mstone: 44.0a1
dom.apps.reset-permissions: true
dom.mozApps.used: true
extensions.lastAppVersion: 44.0a1
font.internaluseonly.changed: true
gfx.direct3d.last_used_feature_level_idx: 0
gfx.driver-init.appVersion: 41.0.1
gfx.driver-init.deviceID: 0x1189
gfx.driver-init.driverVersion: 10.18.13.5598
gfx.driver-init.feature-d2d: true
gfx.driver-init.feature-d3d11: true
gfx.driver-init.status: 2
media.gmp-eme-adobe.abi: x86-msvc-x64
media.gmp-eme-adobe.lastUpdate: 1444312717
media.gmp-eme-adobe.version: 14
media.gmp-gmpopenh264.lastUpdate: 1442004313
media.gmp-gmpopenh264.version: 1.4
media.gmp-manager.buildID: 20151007030205
media.gmp-manager.lastCheck: 1444312715
media.hardware-video-decoding.failed: false
network.auth.allow-subresource-auth: 2
network.cookie.prefsMigrated: true
network.predictor.cleaned-up: true
places.database.lastMaintenance: 1443991518
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: {fe274628-a5da-4674-8e4e-d1ceee213586}
storage.vacuum.last.index: 1
storage.vacuum.last.places.sqlite: 1442118011

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

JavaScript
----------

Incremental GC: true

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

Activated: false
Prevent Accessibility: 0

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

NSPR
Expected minimum version: 4.10.9
Version in use: 4.10.9

NSS
Expected minimum version: 3.20 Basic ECC
Version in use: 3.20 Basic ECC

NSSSMIME
Expected minimum version: 3.20 Basic ECC
Version in use: 3.20 Basic ECC

NSSSSL
Expected minimum version: 3.20 Basic ECC
Version in use: 3.20 Basic ECC

NSSUTIL
Expected minimum version: 3.20
Version in use: 3.20

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

Comment 19

[George Wright (:gw280) (needinfo me!)]

(In reply to George Wright (:gw280) (:gwright) from comment #15)
> I have a GTX 970 machine running Windows 10. I'll see if I can reproduce
> this.

When I said 970 I actually meant 670. Oops.

Comment 20

[Josh Tumath]

(In reply to George Wright (:gw280) (:gwright) from comment #18)
> I can't reproduce this using my Win10/GTX670 machine, running the NVIDIA
> drivers 355.98. This might be an issue with Maxwell chipsets, maybe?

It probably is. But I haven't experienced any issues with the new driver. Maybe this is finally fixed now (famous last words).

Comment 21

[George Wright (:gw280) (needinfo me!)]

OK, we'll keep it on the radar for a little while and if you don't see any further issues we'll resolve this as WORKSFORME.

Comment 22

[Tracy Walker [:tracy]]

Josh, please don't hesitate to reopen if you see this again.