1082660 - Firefox UI freezes with OMTC

Status: RESOLVED DUPLICATE of bug 1064864

(Core :: Graphics, defect)

Description

[Janne H]

User Agent: Mozilla/5.0 (Windows NT 6.3; WOW64; rv:33.0) Gecko/20100101 Firefox/33.0
Build ID: 20141011015303

Steps to reproduce:

1. Install Firefox (I tested release version 33 and nightly 36)
1. Open https://www.mozilla.org/en-US/firefox/desktop/
2. Scroll down to "Faster at what you do most" section until gray bars are visible

Extra info: freeze doesn't happen on my slower machines or if I disable hardware accelerated rendering. Or if I turn off layers.offmainthreadcomposition.enabled.


Actual results:

Firefox UI freezes and stops updating.


Expected results:

Firefox should continue to work normally.

Comment 1

[Gian-Carlo Pascutto [:gcp]]

>freeze doesn't happen on my slower machines or if I disable hardware accelerated rendering.

What video card and driver version is this?

Comment 2

[Janne H]

Asus GTX Titan and Nvidia drivers 344.11.

Comment 3

[Loic]

Could you type about:support and paste the section "graphics", please/

Comment 4

[Gian-Carlo Pascutto [:gcp]]

Reproducible on my machine with Windows 7 too. Open that page in a tab, and scroll down quickly = total Firefox hang.

Adapter Description	NVIDIA GeForce GTX 670
Adapter Drivers	nvd3dumx,nvwgf2umx,nvwgf2umx nvd3dum,nvwgf2um,nvwgf2um
Adapter RAM	2048
Device ID	0x1189
Direct2D Enabled	true
DirectWrite Enabled	true (6.2.9200.16571)
Driver Date	9-13-2014
Driver Version	9.18.13.4411
GPU #2 Active	false
GPU Accelerated Windows	1/1 Direct3D 11 (OMTC)
Subsys ID	00000000
Vendor ID	0x10de
WebGL Renderer	Google Inc. -- ANGLE (NVIDIA GeForce GTX 670 Direct3D9Ex vs_3_0 ps_3_0)
windowLayerManagerRemote	true
AzureCanvasBackend	direct2d
AzureContentBackend	direct2d 1.1
AzureFallbackCanvasBackend	cairo
AzureSkiaAccelerated	0

Comment 5

[Janne H]

Adapter Description	NVIDIA GeForce GTX TITAN
Adapter Drivers	nvd3dumx,nvwgf2umx,nvwgf2umx nvd3dum,nvwgf2um,nvwgf2um
Adapter RAM	4095
Device ID	0x1005
Direct2D Enabled	true
DirectWrite Enabled	true (6.3.9600.17111)
Driver Date	9-13-2014
Driver Version	9.18.13.4411
GPU #2 Active	false
GPU Accelerated Windows	1/1 Direct3D 11 (OMTC)
Vendor ID	0x10de
WebGL Renderer	Google Inc. -- ANGLE (NVIDIA GeForce GTX TITAN Direct3D9Ex vs_3_0 ps_3_0)
windowLayerManagerRemote	true
AzureCanvasBackend	direct2d
AzureContentBackend	direct2d
AzureFallbackCanvasBackend	cairo
AzureSkiaAccelerated	0

Comment 6

[Gian-Carlo Pascutto [:gcp]]

Breaking into a hanging debug build didn't reveal much useful. We actually do still close if you click the Windows close button, so it's not really a total hang. Just the UI becoming totally unresponsive. The logs, however, are more revealing:

[4860] WARNING: Could not copy render target - source rect out of bounds: file c:/mozilla/mozilla-central/gfx/layers/d3d11/CompositorD3D11.cpp, line 478
[4860] WARNING: Could not copy render target - source rect out of bounds: file c:/mozilla/mozilla-central/gfx/layers/d3d11/CompositorD3D11.cpp, line 478
[4860] WARNING: Failed to open shared texture: file c:\mozilla\mozilla-central\gfx\layers\d3d11/TextureD3D11.cpp, line 435
[4860] WARNING: Failed to open shared texture: file c:\mozilla\mozilla-central\gfx\layers\d3d11/TextureD3D11.cpp, line 435
[4860] WARNING: Failed to open shared texture: file c:\mozilla\mozilla-central\gfx\layers\d3d11/TextureD3D11.cpp, line 435
[4860] WARNING: Failed to open shared texture: file c:\mozilla\mozilla-central\gfx\layers\d3d11/TextureD3D11.cpp, line 435
[4860] WARNING: Failed to open shared texture: file c:\mozilla\mozilla-central\gfx\layers\d3d11/TextureD3D11.cpp, line 435
[4860] WARNING: Failed to open shared texture: file c:\mozilla\mozilla-central\gfx\layers\d3d11/TextureD3D11.cpp, line 435
[4860] WARNING: Failed to open shared texture: file c:\mozilla\mozilla-central\gfx\layers\d3d11/TextureD3D11.cpp, line 435
[4860] WARNING: Failed to open shared texture: file c:\mozilla\mozilla-central\gfx\layers\d3d11/TextureD3D11.cpp, line 435
[4860] WARNING: Failed to open shared texture: file c:\mozilla\mozilla-central\gfx\layers\d3d11/TextureD3D11.cpp, line 435
[4860] WARNING: Failed to open shared texture: file c:\mozilla\mozilla-central\gfx\layers\d3d11/TextureD3D11.cpp, line 435
[4860] ###!!! ASSERTION: Could not create texture: 'texture', file c:/mozilla/mozilla-central/gfx/layers/d3d11/CompositorD3D11.cpp, line 451

This goes on for thousands of lines.

Comment 7

[Lars Viklund]

A small nitpick on srcBox.back at CompositorD3D11.cpp:34, pixel ranges in D3D_BOX are half-open, so if front is 0, back shall be 1 unless you want a 0-pixel thick slice.

http://msdn.microsoft.com/en-us/library/windows/desktop/ff476089%28v=vs.85%29.aspx

Comment 8

[Gian-Carlo Pascutto [:gcp]]

(In reply to Lars Viklund from comment #7)
> A small nitpick on srcBox.back at CompositorD3D11.cpp:34, pixel ranges in
> D3D_BOX are half-open, so if front is 0, back shall be 1 unless you want a
> 0-pixel thick slice.

Fixed in bug 1057716. You're looking at an outdated source tree.

Comment 9

[Gian-Carlo Pascutto [:gcp]]

It looks like the "Could not copy render target - source rect out of bounds" warning is unrelated, I don't always get that if I hang. Here's the backtrace for the failing CreateTexture:

D3D11_TEXTURE2D_DESC = {Width=260 Height=52 MipLevels=1 ...}

>	xul.dll!mozilla::layers::CompositorD3D11::CreateRenderTargetFromSource(const mozilla::gfx::IntRectTyped<mozilla::gfx::UnknownUnits> & aRect, const mozilla::layers::CompositingRenderTarget * aSource, const mozilla::gfx::IntPointTyped<mozilla::gfx::UnknownUnits> & aSourcePoint)  Line 453	C++
 	xul.dll!mozilla::layers::CreateTemporaryTargetAndCopyFromBackground<mozilla::layers::ContainerLayerComposite>(mozilla::layers::ContainerLayerComposite * aContainer, mozilla::layers::LayerManagerComposite * aManager)  Line 397 + 0x16 bytes	C++
 	xul.dll!mozilla::layers::ContainerRender<mozilla::layers::ContainerLayerComposite>(mozilla::layers::ContainerLayerComposite * aContainer, mozilla::layers::LayerManagerComposite * aManager, const nsIntRect & aClipRect)  Line 430 + 0xd bytes	C++
 	xul.dll!mozilla::layers::ContainerLayerComposite::RenderLayer(const nsIntRect & aClipRect)  Line 545 + 0x15 bytes	C++
 	xul.dll!mozilla::layers::RenderLayers<mozilla::layers::ContainerLayerComposite>(mozilla::layers::ContainerLayerComposite * aContainer, mozilla::layers::LayerManagerComposite * aManager, const mozilla::gfx::IntRectTyped<mozilla::RenderTargetPixel> & aClipRect)  Line 325 + 0x32 bytes	C++
 	xul.dll!mozilla::layers::ContainerRender<mozilla::layers::ContainerLayerComposite>(mozilla::layers::ContainerLayerComposite * aContainer, mozilla::layers::LayerManagerComposite * aManager, const nsIntRect & aClipRect)  Line 471 + 0x22 bytes	C++
 	xul.dll!mozilla::layers::ContainerLayerComposite::RenderLayer(const nsIntRect & aClipRect)  Line 545 + 0x15 bytes	C++
 	xul.dll!mozilla::layers::RenderLayers<mozilla::layers::ContainerLayerComposite>(mozilla::layers::ContainerLayerComposite * aContainer, mozilla::layers::LayerManagerComposite * aManager, const mozilla::gfx::IntRectTyped<mozilla::RenderTargetPixel> & aClipRect)  Line 325 + 0x32 bytes	C++
 	xul.dll!mozilla::layers::ContainerRender<mozilla::layers::ContainerLayerComposite>(mozilla::layers::ContainerLayerComposite * aContainer, mozilla::layers::LayerManagerComposite * aManager, const nsIntRect & aClipRect)  Line 471 + 0x22 bytes	C++
 	xul.dll!mozilla::layers::ContainerLayerComposite::RenderLayer(const nsIntRect & aClipRect)  Line 545 + 0x15 bytes	C++
 	xul.dll!mozilla::layers::RenderLayers<mozilla::layers::ContainerLayerComposite>(mozilla::layers::ContainerLayerComposite * aContainer, mozilla::layers::LayerManagerComposite * aManager, const mozilla::gfx::IntRectTyped<mozilla::RenderTargetPixel> & aClipRect)  Line 325 + 0x32 bytes	C++
 	xul.dll!mozilla::layers::ContainerRender<mozilla::layers::ContainerLayerComposite>(mozilla::layers::ContainerLayerComposite * aContainer, mozilla::layers::LayerManagerComposite * aManager, const nsIntRect & aClipRect)  Line 471 + 0x22 bytes	C++
 	xul.dll!mozilla::layers::ContainerLayerComposite::RenderLayer(const nsIntRect & aClipRect)  Line 545 + 0x15 bytes	C++
 	xul.dll!mozilla::layers::RenderLayers<mozilla::layers::ContainerLayerComposite>(mozilla::layers::ContainerLayerComposite * aContainer, mozilla::layers::LayerManagerComposite * aManager, const mozilla::gfx::IntRectTyped<mozilla::RenderTargetPixel> & aClipRect)  Line 325 + 0x32 bytes	C++
 	xul.dll!mozilla::layers::ContainerRender<mozilla::layers::ContainerLayerComposite>(mozilla::layers::ContainerLayerComposite * aContainer, mozilla::layers::LayerManagerComposite * aManager, const nsIntRect & aClipRect)  Line 471 + 0x22 bytes	C++
 	xul.dll!mozilla::layers::ContainerLayerComposite::RenderLayer(const nsIntRect & aClipRect)  Line 545 + 0x15 bytes	C++
 	xul.dll!mozilla::layers::LayerManagerComposite::Render()  Line 676	C++
 	xul.dll!mozilla::layers::LayerManagerComposite::EndTransaction(void (mozilla::layers::PaintedLayer *, gfxContext *, const nsIntRegion &, mozilla::layers::DrawRegionClip, const nsIntRegion &, void *)* aCallback, void * aCallbackData, mozilla::layers::LayerManager::EndTransactionFlags aFlags)  Line 261	C++
 	xul.dll!mozilla::layers::LayerManagerComposite::EndEmptyTransaction(mozilla::layers::LayerManager::EndTransactionFlags aFlags)  Line 211	C++
 	xul.dll!mozilla::layers::CompositorParent::CompositeToTarget(mozilla::gfx::DrawTarget * aTarget, const nsIntRect * aRect)  Line 711	C++
 	xul.dll!mozilla::layers::CompositorParent::CompositeCallback()  Line 646	C++
 	xul.dll!MessageLoop::RunTask(Task * task)  Line 359	C++
 	xul.dll!MessageLoop::DeferOrRunPendingTask(const MessageLoop::PendingTask & pending_task)  Line 369	C++
 	xul.dll!MessageLoop::DoWork()  Line 444 + 0x5 bytes	C++
 	xul.dll!base::MessagePumpForUI::DoRunLoop()  Line 216 + 0xa bytes	C++
 	xul.dll!base::MessagePumpWin::RunWithDispatcher(base::MessagePump::Delegate * delegate, base::MessagePumpWin::Dispatcher * dispatcher)  Line 56	C++
 	xul.dll!base::MessagePumpWin::Run(base::MessagePump::Delegate * delegate)  Line 78 + 0xd bytes	C++
 	xul.dll!MessageLoop::RunInternal()  Line 230 + 0xf bytes	C++
 	xul.dll!MessageLoop::RunHandler()  Line 224	C++
 	xul.dll!MessageLoop::Run()  Line 198	C++
 	xul.dll!base::Thread::ThreadMain()  Line 173	C++
 	xul.dll!`anonymous namespace'::ThreadFunc(void * closure)  Line 27	C++
 	kernel32.dll!@BaseThreadInitThunk@12()  + 0x12 bytes	
 	ntdll.dll!___RtlUserThreadStart@8()  + 0x27 bytes	
 	ntdll.dll!__RtlUserThreadStart@8()  + 0x1b bytes

Comment 10

[Gian-Carlo Pascutto [:gcp]]

Well I'll be damned. The fix pointed out in comment 7 is actually the *cause* of this bug!

$ hg bisect -b
The first bad revision is:
changeset:   201728:b5ccef6b01ec
user:        Jeff Muizelaar <[e-mail]>
date:        Tue Aug 26 14:33:03 2014 -0400
summary:     Bug 1057716. d3d11: Properly copy the background. r=bas

Comment 11

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

Jeff, Bas, let's sort this out quickly and uplift if necessary?

Comment 12

[Lars Viklund]

I would not say cause, but a trigger revealing underlying bogus code.
An empty box is documented to be a no-op for functions like CopySubresourceRegion, so before that commit, the operations did nothing.

MSDN on ID3D11DeviceContext::CopySubresourceRegion:
http://msdn.microsoft.com/en-us/library/windows/desktop/ff476394%28v=vs.85%29.aspx

> An empty box results in a no-op.  A box is empty if the top value is
> greater than or equal to the bottom value, or the left value is
> greater than or equal to the right value, or the front value is
> greater than or equal to the back value. When the box is empty,
> CopySubresourceRegion doesn't perform a copy operation.

> Notice, that for a 2D texture, front and back are set to 0 and 1
> respectively.

Comment 13

[Elbart]

bug 1064864 and bug 1067317 have similar STRs.

Dupes?

Comment 14

[Gian-Carlo Pascutto [:gcp]]

(In reply to Elbart from comment #13)
> bug 1064864 and bug 1067317 have similar STRs.
> 
> Dupes?

Possible, though they seem to end up with a white window instead of frozen content. Checking whether a Nightly immediately before and after the commit identified in comment 10 triggers them would be useful.

Comment 15

[Virtual_ManPL [:Virtual] 🇵🇱 - (please needinfo? me - so I will see your comment/reply/question/etc.)]

(In reply to Gian-Carlo Pascutto [:gcp] from comment #14)
> (In reply to Elbart from comment #13)
> > bug 1064864 and bug 1067317 have similar STRs.
> > 
> > Dupes?
> 
> Possible, though they seem to end up with a white window instead of frozen content.

Looks like a dupe of bug #1064864 - Window with few tabs is LOCKED, FROZEN and ALL WHITE in some odd cases.
If you think otherwise please reopen.

Comment 16

[Jeff Muizelaar [:jrmuizel]]

Is it possible for you to make a reduced test case of the page that causes the hang?

Comment 17

[Gian-Carlo Pascutto [:gcp]]

That page is far beyond my HTML/CSS skills. Maybe you can find whoever made it at Moz and see if they have a graphics card that can reproduce it, though.