1150622 - (BLACK-YT) MSE for YouTube - Spike in "black video complaints"

Status: RESOLVED FIXED

(Core :: Audio/Video: Playback, defect, P1)

Description

[Sheila Mooney]

We received some feedback about a notable spike in "black video" complaints from YouTube. They indicated that could be caused by hardware acceleration, particularly on certain AMD drivers and suggested we look into comparing Chrome's list of blacklisted GPUs with ours.

Comment 1

[Jeff Muizelaar [:jrmuizel]]

Chrome's AMD DXVA blacklist is as follows:

driver version = 13.152.1.8000 
driver version = 8.810.4.5000..8.970.100.1100
all amd switchable

Comment 2

[Lawrence Mandel [:lmandel] (use needinfo)]

Tracking for 37+.

Comment 3

[Jeff Muizelaar [:jrmuizel]]

The list that the above entries are from is:
https://chromium.googlesource.com/chromium/deps/gpu/software_rendering_list/+/master/software_rendering_list.json

Comment 4

[Ralph Giles (:rillian)]

(In reply to Jeff Muizelaar [:jrmuizel] from comment #1)
> Chrome's AMD DXVA blacklist is as follows:
> 
> driver version = 13.152.1.8000 
> driver version = 8.810.4.5000..8.970.100.1100
> all amd switchable

The block list for windows video decoder acceleration also has:

- drivers older than January 2009 except device 8086:29a2, driver > 7.15.10.1624 and osmesa drivers
- S3 Trio Virtual PC device 5333:8811
- Parallels device 1ab8:* driver < 7
- ATI FireMV 2400 device 1002:3151
- NVidia device 10de:* driver 8.15.11.8593
- Anything on Windows version <= 6.0 (Vista and earlier)
- VMWare Fusion 4 device 15ad:* driver <= 7.14.1.1134 on Vista and later
- Intel device 8086:* driver <= 8.15.10.2702 (GPU sandbox issue)
- Intel device 8086:* driver 10.18.10.3308 (GPU sandbox issue)
- AMD device 1002:* driver 13.152.1.8000 (GPU sandbox issue)
- AMD device 1002:* drivers 8.810.4.5000 to 8.970.100.1100 (GPU sandbox issue)
- NVidia device 10de:* drivers 8.17.12.5729 to 8.17.12.8026 (GPU sandbox issue)
- NVidia device 10de:* drivers 9.18.13.783 to 9.18.13.1090 (GPU sandbox issue)
- All AMD switchable
- Intel device 1002:* driver 8.15.10.1749

From chromium src/gpu/config/software_rendering_list_json.cc

We don't need to worry about the vm issues for 37, and I don't know if the 'GPU sandbox issue' applies to us, but that's the full list.

Comment 5

[Jeff Muizelaar [:jrmuizel]]

Or rather:
https://chromium.googlesource.com/chromium/src/gpu/+/master/config/software_rendering_list_json.cc

I'm not sure what the difference between these two lists is.

Comment 6

[Jeff Muizelaar [:jrmuizel]]

All of the AMD entries (and a bunch of other ones) refer to this bug https://code.google.com/p/chromium/issues/detail?can=2&id=298968

Comment 7

[Jeff Muizelaar [:jrmuizel]]

Also, worth mentioning: Chrome detects AMD switchable configurations using a closed source piece of AMD software that they ship.

Comment 8

[Chris Peterson [:cpeterson]]

We don't know if 38+ are affected. This may be an IPv6 network problem like bug 1135078.

Comment 9

[Anthony Jones (:ajones, :kentuckyfriedtakahe, :k17e)]

There are some people reporting that disabling hardware acceleration fixes their issue. We do not have and graphics card models to reproduce the issue yet. We could try decoding a known chunk of data.

Comment 10

[Trevor Rowbotham [:rowbot]]

I have seen the black screen video issue intermittently using nightly the last few days, however, I am running an nvidia card so my issue may be different altogether.  What I've noticed in my situation is that:

-audio continues to play, but video remains black
-the |timeupdate| event is never dispatched
-the |currentTime| property is not updated
-the |buffered| property seems to be updated properly with new TimeRanges
-pausing the video and then playing it seems to force |currentTime| to update, but the |timeupdate| event is still not dispatched

Adapter Description	NVIDIA GeForce GTX 660
Adapter Drivers	nvd3dumx,nvwgf2umx,nvwgf2umx nvd3dum,nvwgf2um,nvwgf2um
Adapter RAM	2048
Asynchronous Pan/Zoom	none
Device ID	0x11c0
Direct2D Enabled	true
DirectWrite Enabled	true (6.2.9200.16492)
Driver Date	3-13-2015
Driver Version	9.18.13.4788
GPU #2 Active	false
GPU Accelerated Windows	1/1 Direct3D 11 (OMTC)
Subsys ID	30693842
Vendor ID	0x10de
WebGL Renderer	Google Inc. -- ANGLE (NVIDIA GeForce GTX 660 Direct3D11 vs_5_0 ps_5_0)
windowLayerManagerRemote	true
AzureCanvasBackend	direct2d 1.1
AzureContentBackend	direct2d 1.1
AzureFallbackCanvasBackend	cairo
AzureSkiaAccelerated	0

Comment 11

[Anthony Jones (:ajones, :kentuckyfriedtakahe, :k17e)]

This is a really common graphics adapter which is fixed by disabling hardware acceleration:

http://www.reddit.com/r/firefox/comments/318ivt/upgraded_to_ff_37_youtube_support_broke_again/cq047z8

Comment 12

[Jet Villegas (inactive)]

(In reply to Anthony Jones (:kentuckyfriedtakahe, :k17e) from comment #11)
> This is a really common graphics adapter which is fixed by disabling
> hardware acceleration:
> 
> http://www.reddit.com/r/firefox/comments/318ivt/
> upgraded_to_ff_37_youtube_support_broke_again/cq047z8

Since this driver does appear in the Chromium "accelerated_video_decode" blocklist, I'd like to go ahead and pull it (and the others on the Chromium list) from Firefox. Other ideas?

Comment 13

[Chris Peterson [:cpeterson]]

Do we still think the AMD drivers are an issue? I thought we had narrowed the problem to old versions of a specific Intel GPU driver?

Comment 14

[Anthony Jones (:ajones, :kentuckyfriedtakahe, :k17e)]

(In reply to Chris Peterson [:cpeterson] from comment #13)
> Do we still think the AMD drivers are an issue? I thought we had narrowed
> the problem to old versions of a specific Intel GPU driver?

Turns out to be Intel and nVidia. We'll keep this open as the meta-bug though.

Comment 15

[Anthony Jones (:ajones, :kentuckyfriedtakahe, :k17e)]

I've done a count of crashes of any kind on YouTube to give an indication of how common these GPU/driver versions are:

Count=968 with this driver version; I have a GPU at home I can test this driver version against:
- NVidia device 10de:* driver 8.15.11.8593

Count=2 with but both have more recent drivers
- Parallels device 1ab8:* driver < 7

Count=30 all with more recent drivers; I don't think this old driver version is significant because the driver version relates to logical hardware not physical hardware
- VMWare Fusion 4 device 15ad:* driver <= 7.14.1.1134 on Vista and later

Count=1 does this device relate to Windows XP mode in Win7 Professional?
- S3 Trio Virtual PC device 5333:8811

Count=0 must be fairly uncommon.
- ATI FireMV 2400 device 1002:3151

I don't know if we need to consider these:
- drivers older than January 2009 except device 8086:29a2, driver > 7.15.10.1624 and osmesa drivers

We don't support pre-vista:
- Anything on Windows version <= 6.0 (Vista and earlier)

These are described as sandbox issues and we have no reason to suspect that they affect us.
- Intel device 8086:* driver <= 8.15.10.2702 (GPU sandbox issue)
- Intel device 8086:* driver 10.18.10.3308 (GPU sandbox issue)
- AMD device 1002:* driver 13.152.1.8000 (GPU sandbox issue)
- AMD device 1002:* drivers 8.810.4.5000 to 8.970.100.1100 (GPU sandbox issue)
- NVidia device 10de:* drivers 8.17.12.5729 to 8.17.12.8026 (GPU sandbox issue)
- NVidia device 10de:* drivers 9.18.13.783 to 9.18.13.1090 (GPU sandbox issue)

I tested on an nVidia GTS 240 (10de:0607) and 
9.18.13.1090 works and so does 5.17.12.5869, even with clearkey EME which uses a sandbox.

We have a detect and fallback for AMD that covers this:
- All AMD switchable

This is covered by our recent re-blacklisting:
- Intel device 8086:* driver 8.15.10.1749 (bug says Intel device 1002:* driver 8.15.10.1749)

Comment 16

[DaBiXILI]

I have a similar issue: https://bugzilla.mozilla.org/show_bug.cgi?id=1155608

Comment 17

[Axel Nennker]

Black youtube video using Nightly on:
Name	Intel(R) HD Graphics Family
PNP Device ID	PCI\VEN_8086&DEV_0126&SUBSYS_15E910CF&REV_09\3&11583659&1&10
Adapter Type	Intel(R) HD Graphics Family, Intel Corporation compatible
Adapter Description	Intel(R) HD Graphics Family
Adapter RAM	1,69 GB (1.810.044.928 bytes)
Installed Drivers	igdumd64.dll,igd10umd64.dll,igd10umd64.dll,igdumdx32,igd10umd32,igd10umd32
Driver Version	8.15.10.2372
INF File	oem13.inf (iSNBM0 section)
Color Planes	Not Available
Color Table Entries	4294967296
Resolution	1366 x 768 x 60 hertz
Bits/Pixel	32
Memory Address	0xE0000000-0xE03FFFFF
Memory Address	0xD0000000-0xDFFFFFFF
I/O Port	0x00003000-0x0000303F
IRQ Channel	IRQ 4294967284
I/O Port	0x000003B0-0x000003BB
I/O Port	0x000003C0-0x000003DF
Memory Address	0xA0000-0xBFFFF
Driver	c:\windows\system32\drivers\igdkmd64.sys (8.15.10.2372, 11,66 MB (12.228.128 bytes), 30.08.2011 20:40)

Comment 18

[Sylvestre Ledru [:Sylvestre]]

Anthony, this bug is so general that I don't think there is anything we can do here. I would like to untrack it, ok with you? Thanks

Comment 19

[Anthony Jones (:ajones, :kentuckyfriedtakahe, :k17e)]

Yes. Untrack it.

Comment 20

[Matt Woodrow (:mattwoodrow)]

(In reply to Axel Nennker from comment #17)
> Black youtube video using Nightly on:
> Name	Intel(R) HD Graphics Family
> PNP Device ID	PCI\VEN_8086&DEV_0126&SUBSYS_15E910CF&REV_09\3&11583659&1&10
> Adapter Type	Intel(R) HD Graphics Family, Intel Corporation compatible
> Adapter Description	Intel(R) HD Graphics Family
> Adapter RAM	1,69 GB (1.810.044.928 bytes)
> Installed Drivers
> igdumd64.dll,igd10umd64.dll,igd10umd64.dll,igdumdx32,igd10umd32,igd10umd32
> Driver Version	8.15.10.2372
> INF File	oem13.inf (iSNBM0 section)
> Color Planes	Not Available
> Color Table Entries	4294967296
> Resolution	1366 x 768 x 60 hertz
> Bits/Pixel	32
> Memory Address	0xE0000000-0xE03FFFFF
> Memory Address	0xD0000000-0xDFFFFFFF
> I/O Port	0x00003000-0x0000303F
> IRQ Channel	IRQ 4294967284
> I/O Port	0x000003B0-0x000003BB
> I/O Port	0x000003C0-0x000003DF
> Memory Address	0xA0000-0xBFFFF
> Driver	c:\windows\system32\drivers\igdkmd64.sys (8.15.10.2372, 11,66 MB
> (12.228.128 bytes), 30.08.2011 20:40)

Can you please copy the graphics section of about:support from  the latest nightly?

Comment 21

[Axel Nennker]

(In reply to Matt Woodrow (:mattwoodrow) from comment #20)
> (In reply to Axel Nennker from comment #17)
> > Black youtube video using Nightly on:
> > Name	Intel(R) HD Graphics Family
> > PNP Device ID	PCI\VEN_8086&DEV_0126&SUBSYS_15E910CF&REV_09\3&11583659&1&10
> > Adapter Type	Intel(R) HD Graphics Family, Intel Corporation compatible
> > Adapter Description	Intel(R) HD Graphics Family
> > Adapter RAM	1,69 GB (1.810.044.928 bytes)
> > Installed Drivers
> > igdumd64.dll,igd10umd64.dll,igd10umd64.dll,igdumdx32,igd10umd32,igd10umd32
> > Driver Version	8.15.10.2372
> > INF File	oem13.inf (iSNBM0 section)
> > Color Planes	Not Available
> > Color Table Entries	4294967296
> > Resolution	1366 x 768 x 60 hertz
> > Bits/Pixel	32
> > Memory Address	0xE0000000-0xE03FFFFF
> > Memory Address	0xD0000000-0xDFFFFFFF
> > I/O Port	0x00003000-0x0000303F
> > IRQ Channel	IRQ 4294967284
> > I/O Port	0x000003B0-0x000003BB
> > I/O Port	0x000003C0-0x000003DF
> > Memory Address	0xA0000-0xBFFFF
> > Driver	c:\windows\system32\drivers\igdkmd64.sys (8.15.10.2372, 11,66 MB
> > (12.228.128 bytes), 30.08.2011 20:40)
> 
> Can you please copy the graphics section of about:support from  the latest
> nightly?

Graphics
Adapter Description	Intel(R) HD Graphics Family
Adapter Drivers	igdumd64 igd10umd64 igd10umd64 igdumdx32 igd10umd32 igd10umd32
Adapter RAM	Unknown
Asynchronous Pan/Zoom	none
Device ID	0x0126
Direct2D Enabled	true
DirectWrite Enabled	true (6.2.9200.16571)
Driver Date	4-15-2011
Driver Version	8.15.10.2372
GPU #2 Active	false
GPU Accelerated Windows	1/1 Direct3D 11 (OMTC)
Subsys ID	15e910cf
Vendor ID	0x8086
WebGL Renderer	Google Inc. -- ANGLE (Intel(R) HD Graphics Family Direct3D9Ex vs_3_0 ps_3_0)
windowLayerManagerRemote	true
AzureCanvasBackend	direct2d 1.1
AzureContentBackend	direct2d 1.1
AzureFallbackCanvasBackend	cairo
AzureSkiaAccelerated	0

Comment 22

[Barry Marshall]

My Thinkpad T420 with Intel / NVIDIA on Wn7 64 bit on Lenovo's latest driver package also has problems:

Graphics
Adapter Description	Intel(R) HD Graphics Family
Adapter Description (GPU #2)	NVIDIA NVS 4200M
Adapter Drivers	igdumd64 igd10umd64 igd10umd64 igdumdx32 igd10umd32 igd10umd32
Adapter Drivers (GPU #2)	nvd3dumx,nvwgf2umx,nvwgf2umx nvd3dum,nvwgf2um,nvwgf2um
Adapter RAM	Unknown
Adapter RAM (GPU #2)	1024
Asynchronous Pan/Zoom	none
Device ID	0x0126
Device ID (GPU #2)	0x1057
Direct2D Enabled	true
DirectWrite Enabled	true (6.2.9200.16571)
Driver Date	9-26-2011
Driver Date (GPU #2)	2-4-2015
Driver Version	8.15.10.2538
Driver Version (GPU #2)	9.18.13.4520
GPU #2 Active	false
GPU Accelerated Windows	3/3 Direct3D 11 (OMTC)
Subsys ID	21d017aa
Subsys ID (GPU #2)	21d017aa
Vendor ID	0x8086
Vendor ID (GPU #2)	0x10de
WebGL Renderer	Google Inc. -- ANGLE (Intel(R) HD Graphics Family Direct3D9Ex vs_3_0 ps_3_0)
windowLayerManagerRemote	true
AzureCanvasBackend	direct2d 1.1
AzureContentBackend	direct2d 1.1
AzureFallbackCanvasBackend	cairo
AzureSkiaAccelerated	0

Comment 23

[Liz Henry (:lizzard) (relman/hg->git project)]

Note from discussion with Anthony: This is a tracking bug which we'll close after bug 1156135 lands.

Comment 24

[Liz Henry (:lizzard) (relman/hg->git project)]

Note: I went through the bugs dependent on this, tracked most of them, and asked how far they are intended to be uplifted. Fixes are just now rolling in.

Comment 25

[Liz Henry (:lizzard) (relman/hg->git project)]

Untracking this - while I do track some meta bugs, at this point, we should track the individual bugs as they come up.

Comment 26

[Sylvestre Ledru [:Sylvestre]]

Ditto for 40.

Comment 27

[Anthony Jones (:ajones, :kentuckyfriedtakahe, :k17e)]

I'm going to mark this as fixed because the dependent bugs are fixed.