1095781 - Getting Basic (OMTC) on Windows 8.1 instead of WARP

Status: RESOLVED WORKSFORME

(Core :: Graphics, defect)

Description

[TheRave]

Attachment: glitches.JPG

User Agent: Mozilla/5.0 (Windows NT 6.3; rv:36.0) Gecko/20100101 Firefox/36.0
Build ID: 20141107030202

Steps to reproduce:

Start with clean profile in Nightly 20141107, E10S is activated automatically.

https://www.soeren-hentzschel.at/mozilla/firefox/2014/11/07/mozilla-aktiviert-multiprozess-architektur-in-firefox-nightly/

Adapter Description	AMD Mobility Radeon X1600
Adapter Drivers	atiumdagB atidxx32 atiumdva
Adapter RAM	Unknown
Device ID	0x71c5
DirectWrite Enabled	false (6.3.9600.17111)
Driver Date	6-8-2014
Driver Version	14.200.0.1
GPU #2 Active	false
GPU Accelerated Windows	1/1 Direct3D 9 (OMTC)
Subsys ID	10b21043
Vendor ID	0x1002
WebGL Renderer	Google Inc. -- ANGLE (AMD Mobility Radeon X1600 Direct3D9Ex vs_3_0 ps_3_0)
windowLayerManagerRemote	true
AzureCanvasBackend	skia
AzureContentBackend	cairo
AzureFallbackCanvasBackend	cairo
AzureSkiaAccelerated	0


Actual results:

-page loads normally at first look until scrolling down or up
-at the point where the scrolling starts the glitches occurs in the new showed page areas
-empty spaces & jumping of content during scrolling


Expected results:

clean rendering

Comment 1

[TheRave]

Attachment: more below.JPG

Comment 2

[TheRave]

Attachment: at end.JPG

Comment 3

[Loic]

I'm not able to reproduce, surely related to specific GPUs.

Comment 4

[Ronald Tilby]

I'm seeing this issue also.  Here's my About:support info
Application Basics
------------------

Name: Firefox
Version: 36.0a1
User Agent: Mozilla/5.0 (Windows NT 6.0; rv:36.0) Gecko/20100101 Firefox/36.0
Multiprocess Windows: 1/1

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

All Crash Reports

Extensions
----------

Name: Microsoft .NET Framework Assistant
Version: 0.0.0
Enabled: false
ID: {20a82645-c095-46ed-80e3-08825760534b}

Name: Norton Toolbar
Version: 2014.7.9.8
Enabled: false
ID: {2D3F3651-74B9-4795-BDEC-6DA2F431CB62}

Graphics
--------

Adapter Description: Mobile Intel(R) 965 Express Chipset Family
Adapter Drivers: igdumdx32 igd10umd32
Adapter RAM: Unknown
Device ID: 0x2a02
DirectWrite Enabled: false (7.0.6002.23200)
Driver Date: 1-7-2010
Driver Version: 7.15.10.2039
GPU #2 Active: false
GPU Accelerated Windows: 1/1 Direct3D 9 (OMTC)
Subsys ID: 00000000
Vendor ID: 0x8086
WebGL Renderer: Google Inc. -- ANGLE (Mobile Intel(R) 965 Express Chipset Family Direct3D9Ex vs_3_0 ps_3_0)
windowLayerManagerRemote: true
AzureCanvasBackend: skia
AzureContentBackend: cairo
AzureFallbackCanvasBackend: cairo
AzureSkiaAccelerated: 0

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

accessibility.typeaheadfind.flashBar: 0
browser.cache.disk.capacity: 358400
browser.cache.disk.smart_size.first_run: false
browser.cache.disk.smart_size.use_old_max: false
browser.cache.frecency_experiment: 4
browser.places.smartBookmarksVersion: 7
browser.sessionstore.upgradeBackup.latestBuildID: 20141108030205
browser.startup.homepage_override.buildID: 20141108030205
browser.startup.homepage_override.mstone: 36.0a1
dom.mozApps.used: true
extensions.lastAppVersion: 36.0a1
media.gmp-gmpopenh264.lastUpdate: 1415383716
media.gmp-gmpopenh264.version: 1.1
media.gmp-manager.lastCheck: 1415471800
network.cookie.prefsMigrated: true
places.database.lastMaintenance: 1415471388
places.history.expiration.transient_current_max_pages: 80279
plugin.disable_full_page_plugin_for_types: application/pdf
plugin.importedState: true
privacy.sanitize.migrateFx3Prefs: true
storage.vacuum.last.index: 1
storage.vacuum.last.places.sqlite: 1415384996

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

JavaScript
----------

Incremental GC: true

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

Activated: false
Prevent Accessibility: 0

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

NSPR
Expected minimum version: 4.10.7
Version in use: 4.10.7

NSS
Expected minimum version: 3.18 Basic ECC Beta
Version in use: 3.18 Basic ECC Beta

NSSSMIME
Expected minimum version: 3.18 Basic ECC Beta
Version in use: 3.18 Basic ECC Beta

NSSSSL
Expected minimum version: 3.18 Basic ECC Beta
Version in use: 3.18 Basic ECC Beta

NSSUTIL
Expected minimum version: 3.18 Beta
Version in use: 3.18 Beta

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

Comment 5

[Ronald Tilby]

The problem is gone if I uncheck Options > Advanced > Use hardware acceleration when available and restart the browser.

Comment 6

[TheRave]

Yes, this disabled E10s...

Comment 7

[Paul Silaghi, QA [:pauly]]

Couldn't repro on 36.0a1 (2014-11-17) Win 7 x64, AMD Radeon R7 200 Series.

Comment 8

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

I'm going to look at reproducing this but this makes me feel like something in the drivers aren't behaving if it's limited to this specific GPU/driver, and we may need to just blacklist accelerated layers on this configuration.

Comment 9

[TheRave]

Are there special debug builds or testcases to get more informations for the devs? Suspect it will be difficult to find the problem without a such graphic card on hand.

Comment 10

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

Is this limited only to e10s windows or do you see it when accelerated layers is enabled even on non-e10s?

Comment 11

[Ronald Tilby]

Firefox Nightly 36.0a1 (2014-11-20)
layers.acceleration.force-enabled: true
layers.acceleration.disabled: false
I get streaked text when scrolling


Firefox Nightly 36.0a1 (2014-11-20)
layers.acceleration.force-enabled: false
layers.acceleration.disabled:  false
I get streaked text when scrolling

Let me know of there are other config settings that you want me to test.

Comment 12

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

What if you disable e10s? https://wiki.mozilla.org/Electrolysis#Enabling_and_Disabling_Electrolysis

Comment 13

[TheRave]

layers.acceleration.force-enabled: true
layers.acceleration.disabled: false (default)

Result:
with E10S no change - again graphic glitches
without E10S: all ok

layers.acceleration.force-enabled: false (default)
layers.acceleration.disabled: true

Result:
with E10S: all ok
without E10S: all ok

Conclusion: problem in layer acceleration

Comment 14

[TheRave]

The recommendation from George for blacklisting layer accelerator in this hardware config is no solution. With disabled acceleration youtube HTML5 player stumbles through HD videos in Fullscreen. That is why i have updated my GPU driver some months ago to get finally the hardware acceleration. In fact the acceleration is working on non E10s.

What is the difference between layer acceleration in E10s and non E10s?

Comment 15

[Nicolas Silva [:nical]]

(In reply to TheRave from comment #14)
> What is the difference between layer acceleration in E10s and non E10s?

With e10s you share gpu surfaces across processes which could possibly reveal driver bugs that aren't present in the case where gpu surfaces stay within the same process (non-e10s).

Comment 16

[Ronald Tilby]

Firefox Nightly 36.0a1 (2014-11-24)

Enable E10S: On
layers.acceleration.force-enabled: true
layers.acceleration.disabled: false
I get streaked text when scrolling

Enable E10s: On
layers.acceleration.force-enabled: false
layers.acceleration.disabled: false
I get streaked text when scrolling

Enable E10s: On
layers.acceleration.force-enabled: false
layers.acceleration.disabled: true
Display is correct


Enable E10S: Off
layers.acceleration.force-enabled: true
layers.acceleration.disabled: false
Display is correct

Enable E10s: Off
layers.acceleration.force-enabled: false
layers.acceleration.disabled: false
Display is correct

Enable E10s: Off
layers.acceleration.force-enabled: false
layers.acceleration.disabled: true
Display is correct

Comment 17

[David Parks [:handyman]]

Ronald + TheRave,

You've mostly narrowed this down -- it seems pretty clear that you are running into issues with hardware acceleration + e10s... but there is one more data point I need.  Can you:

* Try the two failing e10s experiments (layers.acceleration.disabled=false, layers.acceleration.force-enabled=true or false) but post your data from about:support (just drop it in a comment in the bug) for both runs?  This will give us info on what the graphics compositor is doing.

* Try changing layers.offmainthreadcomposition.force-basic to true and setting layers.acceleration.disabled=false and layers.acceleration.force-enabled=false.  Let me know the results (about:support info is probably not necessary here).

As Nicolas mentioned, this may be a driver issue that is out of our hands but hopefully its something less serious.

Comment 18

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

I've been looking into how the driver blacklist works so that we can fallback to software if we think that's the correct solution here.

Comment 19

[TheRave]

At first the scenario with
layers.acceleration.force-enabled: & layers.acceleration.disabled: 


There are only two settings changing:
GPU Accelerated Windows: & gfx.direct2d.disabled:


with E10S
layers.acceleration.force-enabled: true
layers.acceleration.disabled: false (default)

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

Name: Firefox
Version: 36.0a1
User Agent: Mozilla/5.0 (Windows NT 6.3; rv:36.0) Gecko/20100101 Firefox/36.0
Multiprocess Windows: 1/1

Graphics
--------

Adapter Description: AMD Mobility Radeon X1600
Adapter Drivers: atiumdagB atidxx32 atiumdva
Adapter RAM: Unknown
Device ID: 0x71c5
DirectWrite Enabled: false (6.3.9600.17111)
Driver Date: 6-8-2014
Driver Version: 14.200.0.1
GPU #2 Active: false
GPU Accelerated Windows: 1/1 Direct3D 9 (OMTC)
Subsys ID: 10b21043
Vendor ID: 0x1002
WebGL Renderer: Google Inc. -- ANGLE (AMD Mobility Radeon X1600 Direct3D9Ex vs_3_0 ps_3_0)
windowLayerManagerRemote: true
AzureCanvasBackend: skia
AzureContentBackend: cairo
AzureFallbackCanvasBackend: cairo
AzureSkiaAccelerated: 0

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

browser.cache.disk.capacity: 358400
browser.cache.disk.smart_size.first_run: false
browser.cache.disk.smart_size.use_old_max: false
browser.cache.frecency_experiment: 2
browser.places.smartBookmarksVersion: 7
browser.sessionstore.upgradeBackup.latestBuildID: 20141126030207
browser.startup.homepage_override.buildID: 20141126030207
browser.startup.homepage_override.mstone: 36.0a1
browser.tabs.remote.autostart: true
browser.tabs.warnOnClose: false
dom.event.contextmenu.enabled: false
dom.mozApps.used: true
extensions.lastAppVersion: 36.0a1
layers.acceleration.force-enabled: true
media.gmp-gmpopenh264.lastUpdate: 1416580906
media.gmp-gmpopenh264.version: 1.1
media.gmp-manager.lastCheck: 1416953820
network.cookie.prefsMigrated: true
places.database.lastMaintenance: 1416953820
places.history.expiration.transient_current_max_pages: 77159
plugin.disable_full_page_plugin_for_types: application/pdf
plugin.importedState: true
privacy.sanitize.migrateFx3Prefs: true
storage.vacuum.last.index: 0
storage.vacuum.last.places.sqlite: 1416582293

+++++++++++
without E10S
layers.acceleration.force-enabled: true
layers.acceleration.disabled: false (default)
Application Basics
------------------

Name: Firefox
Version: 36.0a1
User Agent: Mozilla/5.0 (Windows NT 6.3; rv:36.0) Gecko/20100101 Firefox/36.0
Multiprocess Windows: 0/1

Graphics
--------

Adapter Description: AMD Mobility Radeon X1600
Adapter Drivers: atiumdagB atidxx32 atiumdva
Adapter RAM: Unknown
Device ID: 0x71c5
DirectWrite Enabled: false (6.3.9600.17111)
Driver Date: 6-8-2014
Driver Version: 14.200.0.1
GPU #2 Active: false
GPU Accelerated Windows: 1/1 Direct3D 9 (OMTC)
Subsys ID: 10b21043
Vendor ID: 0x1002
WebGL Renderer: Google Inc. -- ANGLE (AMD Mobility Radeon X1600 Direct3D9Ex vs_3_0 ps_3_0)
windowLayerManagerRemote: true
AzureCanvasBackend: skia
AzureContentBackend: cairo
AzureFallbackCanvasBackend: cairo
AzureSkiaAccelerated: 0

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

browser.cache.disk.capacity: 358400
browser.cache.disk.smart_size.first_run: false
browser.cache.disk.smart_size.use_old_max: false
browser.cache.frecency_experiment: 2
browser.places.smartBookmarksVersion: 7
browser.sessionstore.upgradeBackup.latestBuildID: 20141126030207
browser.startup.homepage_override.buildID: 20141126030207
browser.startup.homepage_override.mstone: 36.0a1
browser.tabs.warnOnClose: false
dom.event.contextmenu.enabled: false
dom.mozApps.used: true
extensions.lastAppVersion: 36.0a1
layers.acceleration.force-enabled: true
media.gmp-gmpopenh264.lastUpdate: 1416580906
media.gmp-gmpopenh264.version: 1.1
media.gmp-manager.lastCheck: 1416953820
network.cookie.prefsMigrated: true
places.database.lastMaintenance: 1416953820
places.history.expiration.transient_current_max_pages: 77159
plugin.disable_full_page_plugin_for_types: application/pdf
plugin.importedState: true
privacy.sanitize.migrateFx3Prefs: true
storage.vacuum.last.index: 0
storage.vacuum.last.places.sqlite: 1416582293

+++++++++++++++++
without E10S
layers.acceleration.force-enabled: false
layers.acceleration.disabled: true
Application Basics
------------------

Name: Firefox
Version: 36.0a1
User Agent: Mozilla/5.0 (Windows NT 6.3; rv:36.0) Gecko/20100101 Firefox/36.0
Multiprocess Windows: 1/1

Graphics
--------

Adapter Description: AMD Mobility Radeon X1600
Adapter Drivers: atiumdagB atidxx32 atiumdva
Adapter RAM: Unknown
Device ID: 0x71c5
DirectWrite Enabled: false (6.3.9600.17111)
Driver Date: 6-8-2014
Driver Version: 14.200.0.1
GPU #2 Active: false
GPU Accelerated Windows: 0/1 Basic (OMTC)
Subsys ID: 10b21043
Vendor ID: 0x1002
WebGL Renderer: Google Inc. -- ANGLE (AMD Mobility Radeon X1600 Direct3D9Ex vs_3_0 ps_3_0)
windowLayerManagerRemote: true
AzureCanvasBackend: skia
AzureContentBackend: cairo
AzureFallbackCanvasBackend: cairo
AzureSkiaAccelerated: 0

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

browser.cache.disk.capacity: 358400
browser.cache.disk.smart_size.first_run: false
browser.cache.disk.smart_size.use_old_max: false
browser.cache.frecency_experiment: 2
browser.places.smartBookmarksVersion: 7
browser.sessionstore.upgradeBackup.latestBuildID: 20141126030207
browser.startup.homepage_override.buildID: 20141126030207
browser.startup.homepage_override.mstone: 36.0a1
browser.tabs.remote.autostart: true
browser.tabs.warnOnClose: false
dom.event.contextmenu.enabled: false
dom.mozApps.used: true
extensions.lastAppVersion: 36.0a1
gfx.direct2d.disabled: true
layers.acceleration.disabled: true
media.gmp-gmpopenh264.lastUpdate: 1416580906
media.gmp-gmpopenh264.version: 1.1
media.gmp-manager.lastCheck: 1416953820
network.cookie.prefsMigrated: true
places.database.lastMaintenance: 1416953820
places.history.expiration.transient_current_max_pages: 77159
plugin.disable_full_page_plugin_for_types: application/pdf
plugin.importedState: true
privacy.sanitize.migrateFx3Prefs: true
storage.vacuum.last.index: 0
storage.vacuum.last.places.sqlite: 1416582293

+++++++++++++++++

without E10S
layers.acceleration.force-enabled: false
layers.acceleration.disabled: true

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

Name: Firefox
Version: 36.0a1
User Agent: Mozilla/5.0 (Windows NT 6.3; rv:36.0) Gecko/20100101 Firefox/36.0
Multiprocess Windows: 0/1

Graphics
--------

Adapter Description: AMD Mobility Radeon X1600
Adapter Drivers: atiumdagB atidxx32 atiumdva
Adapter RAM: Unknown
Device ID: 0x71c5
DirectWrite Enabled: false (6.3.9600.17111)
Driver Date: 6-8-2014
Driver Version: 14.200.0.1
GPU #2 Active: false
GPU Accelerated Windows: 0/1 Basic (OMTC)
Subsys ID: 10b21043
Vendor ID: 0x1002
WebGL Renderer: Google Inc. -- ANGLE (AMD Mobility Radeon X1600 Direct3D9Ex vs_3_0 ps_3_0)
windowLayerManagerRemote: true
AzureCanvasBackend: skia
AzureContentBackend: cairo
AzureFallbackCanvasBackend: cairo
AzureSkiaAccelerated: 0

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

browser.cache.disk.capacity: 358400
browser.cache.disk.smart_size.first_run: false
browser.cache.disk.smart_size.use_old_max: false
browser.cache.frecency_experiment: 2
browser.places.smartBookmarksVersion: 7
browser.sessionstore.upgradeBackup.latestBuildID: 20141126030207
browser.startup.homepage_override.buildID: 20141126030207
browser.startup.homepage_override.mstone: 36.0a1
browser.tabs.warnOnClose: false
dom.event.contextmenu.enabled: false
dom.mozApps.used: true
extensions.lastAppVersion: 36.0a1
gfx.direct2d.disabled: true
layers.acceleration.disabled: true
media.gmp-gmpopenh264.lastUpdate: 1416580906
media.gmp-gmpopenh264.version: 1.1
media.gmp-manager.lastCheck: 1416953820
network.cookie.prefsMigrated: true
places.database.lastMaintenance: 1416953820
places.history.expiration.transient_current_max_pages: 77159
plugin.disable_full_page_plugin_for_types: application/pdf
plugin.importedState: true
privacy.sanitize.migrateFx3Prefs: true
storage.vacuum.last.index: 0
storage.vacuum.last.places.sqlite: 1416582293

Comment 20

[TheRave]

layers.offmainthreadcomposition.force-basic=true
layers.acceleration.disabled=false
layers.acceleration.force-enabled=false
+E10S

Same problems during scrolling.

Tried also:
layers.offmainthreadcomposition.enabled=false
layers.offmainthreadcomposition.force-basic=false

No changes.

Comment 21

[TheRave]

Played a little bit with some other settings.

layers.componentalpha.enabled;false seems to also to fix the graphic problem. Is this helpful?

Comment 22

[Ronald Tilby]

Enable E10S: True
layers.acceleration.disabled: False
layers.acceleration.force-enabled: False
Display result is BAD
Application Basics
------------------

Name: Firefox
Version: 36.0a1
User Agent: Mozilla/5.0 (Windows NT 6.0; rv:36.0) Gecko/20100101 Firefox/36.0
Multiprocess Windows: 1/1

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

All Crash Reports

Extensions
----------

Name: Microsoft .NET Framework Assistant
Version: 0.0.0
Enabled: false
ID: {20a82645-c095-46ed-80e3-08825760534b}

Name: Norton Toolbar
Version: 2014.7.9.8
Enabled: false
ID: {2D3F3651-74B9-4795-BDEC-6DA2F431CB62}

Graphics
--------

Adapter Description: Mobile Intel(R) 965 Express Chipset Family
Adapter Drivers: igdumdx32 igd10umd32
Adapter RAM: Unknown
Device ID: 0x2a02
DirectWrite Enabled: false (7.0.6002.23200)
Driver Date: 1-7-2010
Driver Version: 7.15.10.2039
GPU #2 Active: false
GPU Accelerated Windows: 1/1 Direct3D 9 (OMTC)
Subsys ID: 00000000
Vendor ID: 0x8086
WebGL Renderer: Google Inc. -- ANGLE (Mobile Intel(R) 965 Express Chipset Family Direct3D9Ex vs_3_0 ps_3_0)
windowLayerManagerRemote: true
AzureCanvasBackend: skia
AzureContentBackend: cairo
AzureFallbackCanvasBackend: cairo
AzureSkiaAccelerated: 0

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

browser.cache.disk.capacity: 358400
browser.cache.disk.smart_size.first_run: false
browser.cache.disk.smart_size.use_old_max: false
browser.cache.frecency_experiment: 4
browser.places.smartBookmarksVersion: 7
browser.sessionstore.upgradeBackup.latestBuildID: 20141126030207
browser.startup.homepage_override.buildID: 20141126030207
browser.startup.homepage_override.mstone: 36.0a1
browser.tabs.remote.autostart: true
dom.mozApps.used: true
extensions.lastAppVersion: 36.0a1
media.gmp-gmpopenh264.lastUpdate: 1417056409
media.gmp-gmpopenh264.version: 1.1
media.gmp-manager.lastCheck: 1417056409
network.cookie.prefsMigrated: true
places.database.lastMaintenance: 1417056378
places.history.expiration.transient_current_max_pages: 80279
plugin.disable_full_page_plugin_for_types: application/pdf
plugin.importedState: true
privacy.sanitize.migrateFx3Prefs: true
storage.vacuum.last.index: 0
storage.vacuum.last.places.sqlite: 1417056375

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

JavaScript
----------

Incremental GC: true

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

Activated: false
Prevent Accessibility: 0

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

NSPR
Expected minimum version: 4.10.7
Version in use: 4.10.7

NSS
Expected minimum version: 3.18 Basic ECC Beta
Version in use: 3.18 Basic ECC Beta

NSSSMIME
Expected minimum version: 3.18 Basic ECC Beta
Version in use: 3.18 Basic ECC Beta

NSSSSL
Expected minimum version: 3.18 Basic ECC Beta
Version in use: 3.18 Basic ECC Beta

NSSUTIL
Expected minimum version: 3.18 Beta
Version in use: 3.18 Beta

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

Comment 23

[Ronald Tilby]

Enable E10S: True
layers.acceleration.disabled: False
layers.acceleration.force-enabled: True
Display is BAD

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

Name: Firefox
Version: 36.0a1
User Agent: Mozilla/5.0 (Windows NT 6.0; rv:36.0) Gecko/20100101 Firefox/36.0
Multiprocess Windows: 1/1

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

All Crash Reports

Extensions
----------

Name: Microsoft .NET Framework Assistant
Version: 0.0.0
Enabled: false
ID: {20a82645-c095-46ed-80e3-08825760534b}

Name: Norton Toolbar
Version: 2014.7.9.8
Enabled: false
ID: {2D3F3651-74B9-4795-BDEC-6DA2F431CB62}

Graphics
--------

Adapter Description: Mobile Intel(R) 965 Express Chipset Family
Adapter Drivers: igdumdx32 igd10umd32
Adapter RAM: Unknown
Device ID: 0x2a02
DirectWrite Enabled: false (7.0.6002.23200)
Driver Date: 1-7-2010
Driver Version: 7.15.10.2039
GPU #2 Active: false
GPU Accelerated Windows: 1/1 Direct3D 9 (OMTC)
Subsys ID: 00000000
Vendor ID: 0x8086
WebGL Renderer: Google Inc. -- ANGLE (Mobile Intel(R) 965 Express Chipset Family Direct3D9Ex vs_3_0 ps_3_0)
windowLayerManagerRemote: true
AzureCanvasBackend: skia
AzureContentBackend: cairo
AzureFallbackCanvasBackend: cairo
AzureSkiaAccelerated: 0

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

browser.cache.disk.capacity: 358400
browser.cache.disk.smart_size.first_run: false
browser.cache.disk.smart_size.use_old_max: false
browser.cache.frecency_experiment: 4
browser.places.smartBookmarksVersion: 7
browser.sessionstore.upgradeBackup.latestBuildID: 20141126030207
browser.startup.homepage_override.buildID: 20141126030207
browser.startup.homepage_override.mstone: 36.0a1
browser.tabs.remote.autostart: true
dom.mozApps.used: true
extensions.lastAppVersion: 36.0a1
layers.acceleration.force-enabled: true
media.gmp-gmpopenh264.lastUpdate: 1417056409
media.gmp-gmpopenh264.version: 1.1
media.gmp-manager.lastCheck: 1417056409
network.cookie.prefsMigrated: true
places.database.lastMaintenance: 1417056378
places.history.expiration.transient_current_max_pages: 80279
plugin.disable_full_page_plugin_for_types: application/pdf
plugin.importedState: true
privacy.sanitize.migrateFx3Prefs: true
storage.vacuum.last.index: 0
storage.vacuum.last.places.sqlite: 1417056375

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

JavaScript
----------

Incremental GC: true

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

Activated: false
Prevent Accessibility: 0

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

NSPR
Expected minimum version: 4.10.7
Version in use: 4.10.7

NSS
Expected minimum version: 3.18 Basic ECC Beta
Version in use: 3.18 Basic ECC Beta

NSSSMIME
Expected minimum version: 3.18 Basic ECC Beta
Version in use: 3.18 Basic ECC Beta

NSSSSL
Expected minimum version: 3.18 Basic ECC Beta
Version in use: 3.18 Basic ECC Beta

NSSUTIL
Expected minimum version: 3.18 Beta
Version in use: 3.18 Beta

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

Comment 24

[Ronald Tilby]

Enable E10S: True
layers.acceleration.disabled: false
layers.acceleration.force-enabled: false
layers.offmainthreadcomposition.force-basic: true
Display is BAD

Comment 25

[TheRave]

Something is changed. Graphic glitches are gone. But youtube in full screen stucks now. Missing Hardware acceleration.

Before:
GPU Accelerated Windows	1/1 Direct3D 9 (OMTC)

Now:
GPU Accelerated Windows	0/2 Basic (OMTC)

Comment 26

[TheRave]

This was a christmas present as expected.
 
http://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=0532f2509f3f&tochange=b17e7747d3fb

Don't enable the D3D9 compositing backend except on Windows XP
https://bugzilla.mozilla.org/show_bug.cgi?id=1105088

Set layers.prefer-d3d9 to true now and hwa works. 

And related to this bug here, i accept a not useable(disabled) E10s but not a disabled hardware acceleration.

Comment 27

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

:nical, is our logic for bug 1105088 still valid in this scenario - Windows 8.1, D3D9 card, up to date drivers?

Comment 28

[Nicolas Silva [:nical]]

(In reply to Milan Sreckovic [:milan] from comment #27)
> :nical, is our logic for bug 1105088 still valid in this scenario - Windows
> 8.1, D3D9 card, up to date drivers?

Yes, the intent of bug 1105088 is to reduce the number of default configurations we ship. Non-xp users who want to use D3D9 can enable it through prefs, but since we don't have the resource to properly support D3D9 on these configurations it has to be enabled manually. We can easily come back on this decision if need be.

Comment 29

[TheRave]

HWA is also only a problem with HTML5 players, with "Flash" a HWA driver was not necessary. So the E10s graphic trouble for me was only due the HTML5 player and the necessary driver update. 

But maybe somebody can still reply to my comment 21. Maybe we find here a solution for this bug.

Comment 30

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

(In reply to TheRave from comment #29)
> ... 
> But maybe somebody can still reply to my comment 21. Maybe we find here a
> solution for this bug.

(In reply to TheRave from comment #21)
> Played a little bit with some other settings.
> 
> layers.componentalpha.enabled;false seems to also to fix the graphic
> problem. Is this helpful?

Since this is a bit old and things changed in the meantime, I just want to catch up where we are.  Today, with nightly, by default, we don't get the original correctness problem, but YouTube video is clunky as we're running software.  Correct?

Setting layers.prefer-d3d9 to true helps with the YouTube video performance, but re-introduces the original correctness problem.

Setting layers.prefer-d3d9 to true and also setting layers.componentalpha.enabled to false fixes the original correctness problem, and YouTube video performance is "normal".

Did I get it wrong?

Comment 31

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

I think another vital piece of information here (correct me if I'm wrong) is that without e10s enabled, there is no correctness problem with d3d9. Also, since nical's patch to remove d3d9 as a supported configuration on !winxp, nightly users don't have hardware acceleration at all, whether on e10s or otherwise.

One possible thing that might need to be considered here is whether the blocklist should be expanded to cover "these drivers work for non-e10s but break when doing any cross process stuff", related to nical's comment 15.

Comment 32

[TheRave]

It took some time to get the testcase working....

All right so. -> with build 26.11. it works with exactly this constellation.
HWA active, E10s working, Youtube smoothly in fullscreen

But there is a new dependency. layers.prefer-d3d9 must be set to falsem also in build 26.11. In other case it works not on 26.11 with componentalpha set to false. This was still true from my yesterday tests and it takes some time to find it. 

Of course i cant get the complete constellation working in todays nightly because d3d9d must be false. And if i set it to false HWA is off and E10s works in any case.

Comment 33

[TheRave]

(In reply to George Wright (:gw280) from comment #31)
> I think another vital piece of information here (correct me if I'm wrong) is
> that without e10s enabled, there is no correctness problem with d3d9. Also,
> since nical's patch to remove d3d9 as a supported configuration on !winxp,
> nightly users don't have hardware acceleration at all, whether on e10s or
> otherwise.

^^Confirmed.
 
> One possible thing that might need to be considered here is whether the
> blocklist should be expanded to cover "these drivers work for non-e10s but
> break when doing any cross process stuff", related to nical's comment 15.

Comment 34

[TheRave]

As conclusion for this bug:

layers.componentalpha.enabled;false fixed the issue until this came:

Don't enable the D3D9 compositing backend except on Windows XP
https://bugzilla.mozilla.org/show_bug.cgi?id=1105088

Its a pitty that in the meantime no further test was possible here, if newer changes have additional influences...

Comment 35

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

NI for investigation

Comment 36

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

Is your about:support still showing that you're using Basic (OMTC)? I think now you should be on D3D11 WARP instead. If you're not on WARP, do you get WARP if you load up with a clean profile? If so, does that work as expected?

In any case, I think this can be kicked out of e10s triage because the default shipping configuration appears to be functioning correctly, just it lacks hardware acceleration (see comment 25).

Comment 37

[TheRave]

No warp is here, i have 0/2 Basic (OMTC) active and with clean profile its the same.
Of course i have no graphic glitches anymore. But also no HWA.

----------
If i recall correct here since bug 1105088 i need layers.prefer-d3d9 set to true manually to get D3D9. This is ok, HWA is working with d3d9 true. 

But before 110588 it was ok to have it prefer d3d9 to false and set layers.componentalpha.enabled;false to get rid off Graphic problems. 

Since Bug 110588 componentalpha to false have no influence more to get rid off the graphic glitches. 

In fact d3d9 manually set to true and componentalpha to false have not the same result as automatically d3d9 active before Bug 110588 and componentalpha to false.

Comment 38

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

(In reply to TheRave from comment #37)
> No warp is here, i have 0/2 Basic (OMTC) active and with clean profile its
> the same.
> Of course i have no graphic glitches anymore. But also no HWA.

No WARP even with a clean profile? As I understand it, you shouldn't be getting Basic (OMTC) *at all* on Windows anymore, and it's considered "unsupported".

Comment 39

[TheRave]

No Warp on clean profile.

Comment 40

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

Going to morph this bug then, as it seems the old issue with corruption is no longer present on supported configurations with e10s. Reassigning to Bas so he can look into why we're not getting WARP here.

Comment 41

[Bas Schouten (:bas.schouten)]

(In reply to TheRave from comment #39)
> No Warp on clean profile.

Is this on nightly?

Comment 42

[TheRave]

Yes this is nightly. As long i can avoid use of E10s i will use nightlies.

Comment 43

[TheRave]

I have compiled a own build for testing behavior after revert changes of bug 1105088 with current nightly in relation of E10s and HWA on my hardware. 

With my custom build it is possible to get HWA back on E10s without graphic glitches. For this it is mandatory to set d3d9 to false and layers.componentalpha.enabled to false. As soon i set the d3d9 pref to true i get the graphic glitches back also if layers.componentalpha.enabled is false.

As i understand Bug 1105088 this was done to limit the default configurations which mozilla want to ship. Is it possible to get a separate pref instead to use the d3d9 pref here for this limitation?

Comment 44

[Nicolas Silva [:nical]]

(In reply to TheRave from comment #43)
> As i understand Bug 1105088 this was done to limit the default
> configurations which mozilla want to ship.

Yes

> Is it possible to get a separate
> pref instead to use the d3d9 pref here for this limitation?

I am probably lacking a bit of context, but why do you need a separate pref, since this is already a pref about whether or not to enable d3d9? We currently don't want non-xp users to get d3d9 by default because we don't currently have the resources to support this properly, but we do have the option to opt into d3d9 manually (for users who know what they are doing and are willing to take some risks) through this pref. What would the separate pref add?

Comment 45

[TheRave]

Basically you force me to use the d3d9 switch which is due graphic glitches not correct working with E10s. This force me to disable d3d9 pref on E10s and the GPU Mode will change to 0/3 Basic.

On my custom build with d3d9=false and componentalpha=false the graphic problems are gone.

This are the options:
a) disable E10s will only work as long there is a non E10s mode
b) find the reason why without Bug 1105088 => d3d9=false (this results in d3d11 acc troubleshoot informations)+componentalpha=false is working in E10s and d3d9=true+componentalpha=false is not working in E10s
c) a separate pref to restrict the default shipment of configs so that we can let the d3d9 pref untouched,
d) downgrade to XP
e) new Hardware
f) a custom build is only for me an option and an annoying task then i would tend to option e)

Comment 46

[Nicolas Silva [:nical]]

(In reply to TheRave from comment #45)
> b) find the reason why without Bug 1105088 => d3d9=false (this results in
> d3d11 acc troubleshoot informations)+componentalpha=false is working in E10s
> and d3d9=true+componentalpha=false is not working in E10s

Okay, that's intriguing. Bug 1105088 should have just been equivalent to d3d9 off by default for non-xp users. The interaction with component alpha is unexpected and we need to figure this out.

> c) a separate pref to restrict the default shipment of configs so that we
> can let the d3d9 pref untouched,

I don't see how adding a pref will affect the issue you are experiencing.

(In reply to TheRave from comment #43)
> With my custom build it is possible to get HWA back on E10s without graphic
> glitches. For this it is mandatory to set d3d9 to false and
> layers.componentalpha.enabled to false.

So your custom build is running with D3D11 compositing (what is the accelerated window backend in about:support)?

Comment 47

[TheRave]

>I don't see how adding a pref will affect the issue you are experiencing.

d3d9 pref = true -> side effects under E10s.


>So your custom build is running with D3D11 compositing (what is the accelerated window backend in >about:support)?

Adapter Description	AMD Mobility Radeon X1600
Adapter Drivers	atiumdagB atidxx32 atiumdva
Adapter RAM	Unknown
Asynchronous Pan/Zoom	wheel input enabled
Device ID	0x71c5
DirectWrite Enabled	false (10.0.10586.0)
Driver Date	6-8-2014
Driver Version	14.200.0.1
GPU #2 Active	false
GPU Accelerated Windows	2/2 Direct3D 11 (OMTC)
Subsys ID	10b21043
Supports Hardware H264 Decoding	No; Hardware video decoding disabled or blacklisted
Vendor ID	0x1002
WebGL Renderer	Google Inc. -- ANGLE (AMD Mobility Radeon X1600 Direct3D9Ex vs_3_0 ps_3_0)
windowLayerManagerRemote	true
AzureCanvasBackend	skia
AzureContentBackend	cairo
AzureFallbackCanvasBackend	cairo
AzureSkiaAccelerated	0

Comment 48

[TheRave]

Since todays nightly the graphic glitches are back in Win 10 with E10s. 
layers.prefer-d3d9 is false of course. Is there a new switch introduced?

Build 9.5.16

Features
Compositing	Direct3D 9
Asynchronous Pan/Zoom	wheel input enabled; touch input enabled
WebGL Renderer	Google Inc. -- ANGLE (AMD Mobility Radeon X1600 Direct3D9Ex vs_3_0 ps_3_0)
Hardware H264 Decoding	No; Hardware video decoding disabled or blacklisted
DirectWrite	false (10.0.10586.0)
GPU #1
Active	Yes
Description	AMD Mobility Radeon X1600
Vendor ID	0x1002
Device ID	0x71c5
Driver Version	14.200.0.1
Driver Date	6-8-2014
Drivers	atiumdagB atidxx32 atiumdva
Subsys ID	10b21043
RAM	Unknown
Diagnostics
AzureCanvasAccelerated	0
AzureCanvasBackend	skia
AzureContentBackend	cairo
AzureFallbackCanvasBackend	cairo
Decision Log
HW_COMPOSITING	
force_enabled by user: Force-enabled by pref
D3D9_COMPOSITING	
force_enabled by user: Hardware compositing is force-enabled
DIRECT2D	
failed by runtime: Direct3D11 device does not support texture sharing
Failure Log
(#0) Error	[D3D11] create component blender failed: 0x80070057, 0
(#1) Error	[D3D11] create component blender failed: 0x80070057, 0
(#2) Error	[D3D11] create component blender failed: 0x80070057, 0
(#3) Error	[D3D11] create component blender failed: 0x80070057, 0
(#4) Error	[D3D11] create component blender failed: 0x80070057, 0
(#5) Error	[D3D11] create component blender failed: 0x80070057, 0
(#6) Error	[D3D11] create component blender failed: 0x80070057, 0
(#7) Error	[D3D11] create component blender failed: 0x80070057, 0

Build 8.5.16

Features
Compositing	Basic
Asynchronous Pan/Zoom	wheel input enabled; touch input enabled
WebGL Renderer	Google Inc. -- ANGLE (AMD Mobility Radeon X1600 Direct3D9Ex vs_3_0 ps_3_0)
Hardware H264 Decoding	No; Hardware video decoding disabled or blacklisted
DirectWrite	false (10.0.10586.0)
GPU #1
Active	Yes
Description	AMD Mobility Radeon X1600
Vendor ID	0x1002
Device ID	0x71c5
Driver Version	14.200.0.1
Driver Date	6-8-2014
Drivers	atiumdagB atidxx32 atiumdva
Subsys ID	10b21043
RAM	Unknown
Diagnostics
AzureCanvasAccelerated	0
AzureCanvasBackend	skia
AzureContentBackend	cairo
AzureFallbackCanvasBackend	cairo
Decision Log
HW_COMPOSITING	
force_enabled by user: Force-enabled by pref
DIRECT2D	
failed by runtime: Direct3D11 device does not support texture sharing
Failure Log
(#0) Error	[D3D11] create component blender failed: 0x80070057, 0
(#1) Error	[D3D11] create component blender failed: 0x80070057, 0
(#2) Error	[D3D11] create component blender failed: 0x80070057, 0

Comment 49

[Jeff Muizelaar [:jrmuizel]]

I bet this was caused by dvander's d3d9 change. It looks like you have layers.acceleration.force-enabled set to true. Do the problems go away if you unset that pref?

Comment 50

[TheRave]

What i can say is that layers.componentalpha.enabled=false still removes the graphic glitches. So there are no changes since reporting this bug.

Yes, layers.acceleration.force-enabled is set to true (as since years ago). But changing this value makes no difference. D3D9 is still active here. But i have no objections against this new "bug" as expected componentalpha=false will fix this for me and i have HWA back in e10s and no problems.

If i see by Anthony's comment 4 in bug 1237769 then it looks like you have same problems in XP with Nvidia cards. Are you sure that do we still talking about a driver problem?

Maybe somebody should try to find the real reason for this, instead the hacks for disabling d3d9 and E10s vice versa.

Comment 51

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

Looks like Jeff already answered your questions

Comment 52

[Jeff Muizelaar [:jrmuizel]]

(In reply to TheRave from comment #50)
> What i can say is that layers.componentalpha.enabled=false still removes the
> graphic glitches. So there are no changes since reporting this bug.
> 
> Yes, layers.acceleration.force-enabled is set to true (as since years ago).
> But changing this value makes no difference. D3D9 is still active here. But
> i have no objections against this new "bug" as expected componentalpha=false
> will fix this for me and i have HWA back in e10s and no problems.
> 
> If i see by Anthony's comment 4 in bug 1237769 then it looks like you have
> same problems in XP with Nvidia cards. Are you sure that do we still talking
> about a driver problem?
> 
> Maybe somebody should try to find the real reason for this, instead the
> hacks for disabling d3d9 and E10s vice versa.

Sure. Do you mind filing a fresh bug and cc'ing me there?

Comment 53

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

(In reply to Jeff Muizelaar [:jrmuizel] from comment #49)
> I bet this was caused by dvander's d3d9 change. It looks like you have
> layers.acceleration.force-enabled set to true. Do the problems go away if
> you unset that pref?

Yes, I did changed this semi-intentionally, since otherwise "force enabled" doesn't really do anything. (For example it does not bypass the D3D11 blocklist, but it does bypass texture sharing checks). Do you think that change should be reverted?

Comment 54

[TheRave]

Only for protocol, graphic glitches are back with today's nightly and E10s. Looks like this one https://bugzilla.mozilla.org/show_bug.cgi?id=1292498

Compositing	Direct3D 9
Asynchronous Pan/Zoom	none
WebGL Renderer	Google Inc. -- ANGLE (AMD Mobility Radeon X1600 Direct3D9Ex vs_3_0 ps_3_0)
WebGL2 Renderer	WebGL creation failed: * WebGL 2 requires support for the following features: clear_buffers copy_buffer draw_instanced frag_color_float get_integer_indexed get_integer64_indexed gpu_shader4 instanced_arrays instanced_non_arrays occlusion_query2 sRGB_framebuffer sampler_objects texture_storage transform_feedback2 uniform_buffer_object prim_restart_fixed * Exhausted GL driver options.
Hardware H264 Decoding	No; Hardware video decoding disabled or blacklisted
Audio Backend	wasapi
DirectWrite	false (10.0.10586.494)
GPU #1
Active	Yes
Description	AMD Mobility Radeon X1600
Vendor ID	0x1002
Device ID	0x71c5
Driver Version	14.200.0.1
Driver Date	6-8-2014
Drivers	atiumdagB atidxx32 atiumdva
Subsys ID	10b21043
RAM	Unknown
Diagnostics
AzureCanvasAccelerated	0
AzureCanvasBackend	skia
AzureContentBackend	cairo
AzureFallbackCanvasBackend	cairo
failures	[GFX1]: D3D11 device creation failed: 0x887a0004
Decision Log
HW_COMPOSITING	
force_enabled by user: Force-enabled by pref
D3D11_COMPOSITING	
failed by runtime: Failed to acquire a D3D11 device
D3D9_COMPOSITING	
force_enabled by user: Hardware compositing is force-enabled
DIRECT2D	
failed by runtime: Failed to acquire a Direct3D 11 content device
D3D11_HW_ANGLE	
disabled by env: D3D11 compositing is disabled
Failure Log
(#0) Assert	D3D11 device creation failed: 0x887a0004

Comment 55

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

(In reply to TheRave from comment #54)
> Only for protocol, graphic glitches are back with today's nightly and E10s.
> Looks like this one https://bugzilla.mozilla.org/show_bug.cgi?id=1292498

What does your about:support look like from before that change? (Either with an Aurora build or an earlier Nightly build?)

Could you also paste some Telemetry data? Go to about:telemetry, click "Environment", then "System", and copy everything starting with "gfx".

Comment 56

[TheRave]

09.08.16

Compositing	Direct3D 11
Asynchronous Pan/Zoom	none
WebGL Renderer	Google Inc. -- ANGLE (AMD Mobility Radeon X1600 Direct3D9Ex vs_3_0 ps_3_0)
WebGL2 Renderer	WebGL creation failed: * WebGL 2 requires support for the following features: clear_buffers copy_buffer draw_instanced frag_color_float get_integer_indexed get_integer64_indexed gpu_shader4 instanced_arrays instanced_non_arrays occlusion_query2 sRGB_framebuffer sampler_objects texture_storage transform_feedback2 uniform_buffer_object prim_restart_fixed * Exhausted GL driver options.
Hardware H264 Decoding	No; Hardware video decoding disabled or blacklisted
Audio Backend	wasapi
DirectWrite	false (10.0.10586.494)
GPU #1
Active	Yes
Description	AMD Mobility Radeon X1600
Vendor ID	0x1002
Device ID	0x71c5
Driver Version	14.200.0.1
Driver Date	6-8-2014
Drivers	atiumdagB atidxx32 atiumdva
Subsys ID	10b21043
RAM	Unknown
Diagnostics
AzureCanvasAccelerated	0
AzureCanvasBackend	skia
AzureContentBackend	cairo
AzureFallbackCanvasBackend	cairo
failures	[GFX1-]: VendorIDMismatch V 0x1002 0
Decision Log
HW_COMPOSITING	
force_enabled by user: Force-enabled by pref
D3D9_COMPOSITING	
force_enabled by user: Hardware compositing is force-enabled
DIRECT2D	
failed by runtime: Direct3D11 device does not support texture sharing
D3D11_HW_ANGLE	
broken by runtime: RenderTargetViews need recreating
Failure Log
(#0) Error	VendorIDMismatch V 0x1002 0

gfx.D2DEnabled	false
gfx.DWriteEnabled	false
gfx.adapters.[0].description	AMD Mobility Radeon X1600
gfx.adapters.[0].vendorID	0x1002
gfx.adapters.[0].deviceID	0x71c5
gfx.adapters.[0].subsysID	10b21043
gfx.adapters.[0].RAM	null
gfx.adapters.[0].driver	atiumdagB atidxx32 atiumdva
gfx.adapters.[0].driverVersion	14.200.0.1
gfx.adapters.[0].driverDate	6-8-2014
gfx.adapters.[0].GPUActive	true
gfx.monitors.[0].screenWidth	1440
gfx.monitors.[0].screenHeight	900
gfx.monitors.[0].refreshRate	60
gfx.monitors.[0].pseudoDisplay	false
gfx.features.compositor	d3d11
gfx.features.d3d11.status	available
gfx.features.d3d11.version	37632
gfx.features.d3d11.warp	false
gfx.features.d3d11.textureSharing	false
gfx.features.d3d11.blacklisted	false
gfx.features.d2d.status	failed
gfx.features.d2d.version	1.1

-----------------------------------------

15.8.16
Compositing	Direct3D 9
Asynchronous Pan/Zoom	none
WebGL Renderer	Google Inc. -- ANGLE (AMD Mobility Radeon X1600 Direct3D9Ex vs_3_0 ps_3_0)
WebGL2 Renderer	WebGL creation failed: * WebGL 2 requires support for the following features: clear_buffers copy_buffer draw_instanced frag_color_float get_integer_indexed get_integer64_indexed gpu_shader4 instanced_arrays instanced_non_arrays occlusion_query2 sRGB_framebuffer sampler_objects texture_storage transform_feedback2 uniform_buffer_object prim_restart_fixed * Exhausted GL driver options.
Hardware H264 Decoding	No; Hardware video decoding disabled or blacklisted
Audio Backend	wasapi
DirectWrite	false (10.0.10586.494)
GPU #1
Active	Yes
Description	AMD Mobility Radeon X1600
Vendor ID	0x1002
Device ID	0x71c5
Driver Version	14.200.0.1
Driver Date	6-8-2014
Drivers	atiumdagB atidxx32 atiumdva
Subsys ID	10b21043
RAM	Unknown
Diagnostics
AzureCanvasAccelerated	0
AzureCanvasBackend	skia
AzureContentBackend	cairo
AzureFallbackCanvasBackend	cairo
failures	[GFX1]: D3D11 device creation failed: 0x887a0004
Decision Log
HW_COMPOSITING	
force_enabled by user: Force-enabled by pref
D3D11_COMPOSITING	
failed by runtime: Failed to acquire a D3D11 device
D3D9_COMPOSITING	
force_enabled by user: Hardware compositing is force-enabled
DIRECT2D	
failed by runtime: Failed to acquire a Direct3D 11 content device
D3D11_HW_ANGLE	
disabled by env: D3D11 compositing is disabled
Failure Log
(#0) Assert	D3D11 device creation failed: 0x887a0004

gfx.D2DEnabled	false
gfx.DWriteEnabled	false
gfx.adapters.[0].description	AMD Mobility Radeon X1600
gfx.adapters.[0].vendorID	0x1002
gfx.adapters.[0].deviceID	0x71c5
gfx.adapters.[0].subsysID	10b21043
gfx.adapters.[0].RAM	null
gfx.adapters.[0].driver	atiumdagB atidxx32 atiumdva
gfx.adapters.[0].driverVersion	14.200.0.1
gfx.adapters.[0].driverDate	6-8-2014
gfx.adapters.[0].GPUActive	true
gfx.monitors.[0].screenWidth	1440
gfx.monitors.[0].screenHeight	900
gfx.monitors.[0].refreshRate	60
gfx.monitors.[0].pseudoDisplay	false
gfx.features.compositor	d3d9
gfx.features.d3d11.status	failed
gfx.features.d2d.status	failed
gfx.features.d2d.version	1.1

Comment 57

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

Do the glitches go away if you turn off acceleration.force-enabled?

Comment 58

[TheRave]

No, acceleration.force-enabled have no influence.

Comment 59

[Sotaro Ikeda [:sotaro]]

Basic (OMTC) tend to weak to glitch. It is going to be mitigated by Bug 1259571.

Comment 60

[TheRave]

Since today's nightly also non e10s results in graphic glitches after scrolling.

2016-09-03T23:51:37: INFO : Narrowed inbound regression window from [0cfb8480, b7e5426f] (3 revisions) to [0cfb8480, 657ba2f2] (2 revisions) (~1 steps left)
2016-09-03T23:51:37: DEBUG : Starting merge handling...
2016-09-03T23:51:37: DEBUG : Using url: https://hg.mozilla.org/integration/mozilla-inbound/json-pushes?changeset=657ba2f2611c714d718fb0268acfd5e60ff44310&full=1
2016-09-03T23:51:38: DEBUG : Found commit message:
Bug 1299738 - warn instead of crash if DrawTargetSkia's GetBitmapForSurface fails. r=mchang

2016-09-03T23:51:38: INFO : The bisection is done.
2016-09-03T23:51:38: INFO : Stopped

Comment 61

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

Interesting ^^

Comment 62

[Lee Salzman [:lsalzman]]

(In reply to TheRave from comment #60)
> Since today's nightly also non e10s results in graphic glitches after
> scrolling.
> 
> 2016-09-03T23:51:37: INFO : Narrowed inbound regression window from
> [0cfb8480, b7e5426f] (3 revisions) to [0cfb8480, 657ba2f2] (2 revisions) (~1
> steps left)
> 2016-09-03T23:51:37: DEBUG : Starting merge handling...
> 2016-09-03T23:51:37: DEBUG : Using url:
> https://hg.mozilla.org/integration/mozilla-inbound/json-
> pushes?changeset=657ba2f2611c714d718fb0268acfd5e60ff44310&full=1
> 2016-09-03T23:51:38: DEBUG : Found commit message:
> Bug 1299738 - warn instead of crash if DrawTargetSkia's GetBitmapForSurface
> fails. r=mchang
> 
> 2016-09-03T23:51:38: INFO : The bisection is done.
> 2016-09-03T23:51:38: INFO : Stopped

When the error occurs, are there is a warning in the debug log printed like "Failed getting DataSourceSurface for Skia bitmap"?

That's the only reason I could maybe see this occurring. In that while previously a debug build would crash in that situation with Skia content, now it is allowed to continue like the other backends. So at least, we should hypothetically see evidence of that in the logs. That would be fallout from something like OOM, maybe a device reset, etc.

That's my only guess as to what could be going on here if that patch somehow caused something like that.

Comment 63

[TheRave]

Do i need a debug build here? Debug builds are crashing immediately at startup here.

Comment 64

[TheRave]

^^

Comment 65

[TheRave]

I have tested today randomly a nightly and non E10s was now working properly.
Mozregression leads to:

2016-09-15T21:39:06: DEBUG : Found commit message:
Bug 1302225 - Backout skia windows. r=me

Comment 66

[Lee Salzman [:lsalzman]]

Related to "Bug 1299738 - warn instead of crash if DrawTargetSkia's GetBitmapForSurface fails. r=mchang", Mason did find a sneaky little bug in my patch which he corrected in bug 1301920 (https://hg.mozilla.org/mozilla-central/rev/fece9aa991e6). So at least Mason managed to absolve me of my sins in this case. :)

Comment 67

[TheRave]

I have set gfx.content.azure.backends to direct2d1.1,skia,cairo and it seems to work.

Comment 68

[Mason Chang [Inactive] [:mchang]]

From comment 67, resolving as WFM. There seems to be a bunch of different bugs in this one bug, so please file a new one if something else comes up. This seems to be the approach in comment 52 as well. Thanks!