Given the prominence of the affected sites, I am upping the priority until we have a better handle on how widespread this is.

+ccameron since it's osx.

Since M57 roll out is in progress, let's RBS this to get TPM attention while we figure out how widespread it is. Can someone from the test team try and reproduce this?

+sandersd since this is likely H264 hardware decode.

Daniel, has this issue been seen in any versions other than M57?

Couple of questions

- Can you run this from the command line using the flag --show-mac-overlay-borders and take a video of the flickering?

- Can you run this from the command line using the flag --disable-avfoundation-overlays, and report whether or not the issue continues to reproduce?

To run from the command line, open the Terminal application, and type
  /Applications/Google\ Chrome.app/Contents/MacOS/Google\ Chrome
followed by the above flags.

Does Facebook have a reliable reproduction?

I own  Issue 701060  -- which is a regression drawing hardware-accelerated videos into the 2D canvas context, as well as uploading them to WebGL, on macOS. But that would be visibly and consistently broken.

I just took the M57 update on my MacBook Pro and don't see the issue.

Could you provide about:gpu from an affected system? Here's mine.

Graphics Feature Status
Canvas: Hardware accelerated
Flash: Hardware accelerated
Flash Stage3D: Hardware accelerated
Flash Stage3D Baseline profile: Hardware accelerated
Compositing: Hardware accelerated
Multiple Raster Threads: Enabled
Native GpuMemoryBuffers: Hardware accelerated
Rasterization: Hardware accelerated
Video Decode: Hardware accelerated
Video Encode: Hardware accelerated
VPx Video Decode: Hardware accelerated
WebGL: Hardware accelerated
WebGL2: Hardware accelerated
Driver Bug Workarounds
add_and_true_to_loop_condition
adjust_src_dst_region_for_blitframebuffer
decode_encode_srgb_for_generatemipmap
disable_framebuffer_cmaa
disable_multimonitor_multisampling
disable_webgl_rgb_multisampling_usage
emulate_abs_int_function
get_frag_data_info_bug
init_two_cube_map_levels_before_copyteximage
msaa_is_slow
pack_parameters_workaround_with_pack_buffer
rebind_transform_feedback_before_resume
regenerate_struct_names
remove_invariant_and_centroid_for_essl3
reset_base_mipmap_level_before_texstorage
rewrite_texelfetchoffset_to_texelfetch
scalarize_vec_and_mat_constructor_args
set_zero_level_before_generating_mipmap
unfold_short_circuit_as_ternary_operation
unpack_alignment_workaround_with_unpack_buffer
unpack_image_height_workaround_with_unpack_buffer
use_intermediary_for_copy_texture_image
use_unused_standard_shared_blocks
Problems Detected
Multisampling is buggy on OSX when multiple monitors are connected: 237931
Applied Workarounds: disable_multimonitor_multisampling
Unfold short circuit on Mac OS X: 307751
Applied Workarounds: unfold_short_circuit_as_ternary_operation
Always rewrite vec/mat constructors to be consistent: 398694
Applied Workarounds: scalarize_vec_and_mat_constructor_args
Mac drivers handle struct scopes incorrectly: 403957
Applied Workarounds: regenerate_struct_names
On Intel GPUs MSAA performance is not acceptable for GPU rasterization: 527565
Applied Workarounds: msaa_is_slow
glGenerateMipmap fails if the zero texture level is not set on some Mac drivers: 560499
Applied Workarounds: set_zero_level_before_generating_mipmap
Pack parameters work incorrectly with pack buffer bound: 563714
Applied Workarounds: pack_parameters_workaround_with_pack_buffer
Alignment works incorrectly with unpack buffer bound: 563714
Applied Workarounds: unpack_alignment_workaround_with_unpack_buffer
copyTexImage2D fails when reading from IOSurface on multiple GPU types.: 581777
Applied Workarounds: use_intermediary_for_copy_texture_image
Multisample renderbuffers with format GL_RGB8 have performance issues on Intel GPUs.: 607130
Applied Workarounds: disable_webgl_rgb_multisampling_usage
Limited enabling of Chromium GL_INTEL_framebuffer_CMAA: 535198
Applied Workarounds: disable_framebuffer_cmaa
glGetFragData{Location|Index} works incorrectly on Max: 638340
Applied Workarounds: get_frag_data_info_bug
glResumeTransformFeedback works incorrectly on Intel GPUs: 638514
Applied Workarounds: rebind_transform_feedback_before_resume
glTexStorage* are buggy when base mipmap level is not 0: 640506
Applied Workarounds: reset_base_mipmap_level_before_texstorage
Result of abs(i) where i is an integer in vertex shader is wrong: 642227
Applied Workarounds: emulate_abs_int_function
Rewrite texelFetchOffset to texelFetch for Intel Mac: 642605
Applied Workarounds: rewrite_texelfetchoffset_to_texelfetch
Rewrite condition in for and while loops for Intel Mac: 644669
Applied Workarounds: add_and_true_to_loop_condition
Decode and encode before generateMipmap for srgb format textures on macosx: 634519
Applied Workarounds: decode_encode_srgb_for_generatemipmap
Init first two levels before CopyTexImage2D for cube map texture on Intel Mac 10.12: 648197
Applied Workarounds: init_two_cube_map_levels_before_copyteximage
Insert statements to reference all members in unused std140/shared blocks on Mac: 618464
Applied Workarounds: use_unused_standard_shared_blocks
Tex(Sub)Image3D performs incorrectly when uploading from unpack buffer with GL_UNPACK_IMAGE_HEIGHT greater than zero on Intel Macs: 654258
Applied Workarounds: unpack_image_height_workaround_with_unpack_buffer
adjust src/dst region if blitting pixels outside read framebuffer on Mac: 644740
Applied Workarounds: adjust_src_dst_region_for_blitframebuffer
Mac driver GL 4.1 requires invariant and centroid to match between shaders: 639760, 641129
Applied Workarounds: remove_invariant_and_centroid_for_essl3
Disable KHR_blend_equation_advanced until cc shaders are updated: 661715
Version Information
Data exported	3/16/2017, 5:35:25 PM
Chrome version	Chrome/57.0.2987.110
Operating system	Mac OS X 10.12.3
Software rendering list version	12.13
Driver bug list version	9.29
ANGLE commit id	c1a5d16e964a
2D graphics backend	Skia/57 ae9cc5d3588d52f4b371b55845704b25d88cf06d
Command Line Args	Chrome.app/Contents/MacOS/Google Chrome --flag-switches-begin --flag-switches-end
Driver Information
Initialization time	28
In-process GPU	false
Passthrough Command Decoder	false
Sandboxed	true
GPU0	VENDOR = 0x10de, DEVICE= 0x0fe9
GPU1	VENDOR = 0x8086, DEVICE= 0x0d26 *ACTIVE*
Optimus	true
AMD switchable	false
Driver vendor	
Driver version	10.22.29
Driver date	
Pixel shader version	4.10
Vertex shader version	4.10
Max. MSAA samples	8
Machine model name	MacBookPro
Machine model version	11.3
GL_VENDOR	Intel Inc.
GL_RENDERER	Intel Iris Pro OpenGL Engine
GL_VERSION	4.1 INTEL-10.22.29
GL_EXTENSIONS	GL_ARB_blend_func_extended GL_ARB_draw_buffers_blend GL_ARB_draw_indirect GL_ARB_ES2_compatibility GL_ARB_explicit_attrib_location GL_ARB_gpu_shader_fp64 GL_ARB_gpu_shader5 GL_ARB_instanced_arrays GL_ARB_internalformat_query GL_ARB_occlusion_query2 GL_ARB_sample_shading GL_ARB_sampler_objects GL_ARB_separate_shader_objects GL_ARB_shader_bit_encoding GL_ARB_shader_subroutine GL_ARB_shading_language_include GL_ARB_tessellation_shader GL_ARB_texture_buffer_object_rgb32 GL_ARB_texture_cube_map_array GL_ARB_texture_gather GL_ARB_texture_query_lod GL_ARB_texture_rgb10_a2ui GL_ARB_texture_storage GL_ARB_texture_swizzle GL_ARB_timer_query GL_ARB_transform_feedback2 GL_ARB_transform_feedback3 GL_ARB_vertex_attrib_64bit GL_ARB_vertex_type_2_10_10_10_rev GL_ARB_viewport_array GL_EXT_debug_label GL_EXT_debug_marker GL_EXT_framebuffer_multisample_blit_scaled GL_EXT_texture_compression_s3tc GL_EXT_texture_filter_anisotropic GL_EXT_texture_sRGB_decode GL_APPLE_client_storage GL_APPLE_container_object_shareable GL_APPLE_flush_render GL_APPLE_object_purgeable GL_APPLE_rgb_422 GL_APPLE_row_bytes GL_APPLE_texture_range GL_ATI_texture_mirror_once GL_NV_texture_barrier
Disabled Extensions	GL_KHR_blend_equation_advanced GL_KHR_blend_equation_advanced_coherent
Window system binding vendor	
Window system binding version	
Window system binding extensions	
Direct rendering	Yes
Reset notification strategy	0x0000
GPU process crash count	0
Compositor Information
Tile Update Mode	Zero-copy
Partial Raster	Enabled
GpuMemoryBuffers Status
ATC	Software only
ATCIA	Software only
DXT1	Software only
DXT5	Software only
ETC1	Software only
R_8	GPU_READ_CPU_READ_WRITE, GPU_READ_CPU_READ_WRITE_PERSISTENT
RG_88	Software only
BGR_565	Software only
RGBA_4444	Software only
RGBX_8888	Software only
RGBA_8888	GPU_READ, SCANOUT
BGRX_8888	GPU_READ, SCANOUT
BGRA_8888	GPU_READ, SCANOUT, GPU_READ_CPU_READ_WRITE, GPU_READ_CPU_READ_WRITE_PERSISTENT
YVU_420	Software only
YUV_420_BIPLANAR	GPU_READ_CPU_READ_WRITE, GPU_READ_CPU_READ_WRITE_PERSISTENT
UYVY_422	GPU_READ_CPU_READ_WRITE, GPU_READ_CPU_READ_WRITE_PERSISTENT

Also, please include any errors listed at the bottom of about:gpu that may appear.

Couple of things:

the oldest internal report of this that I can still find goes back to early November 2016. I have reports that I can certainly say are from Chrome 56, not sure if there are reports from older Chrome versions (some of the early reports have been by email / posts and do not have detailed browser version information).

We don't have a reliable way to reproduce it and I have never experienced this myself. We've been internally accumulating bug reports for this for several months now and only recently determined that this seems to be a problem that is not only affecting Facebook, but other sites, too.

This issue happens intermittently and there is no specific system that is affected. 

If I can get a about:gpu dump, I will forward it. It has been historically incredibly hard for us to get these from internal bug reports. The round-trip time for triaging internal bug reports is simply too high to reach out to a reporter at a later time to ask them for this dump.

Unable to reproduce this issue on Mac OS 10.12.3 using chrome #57.0.2987.98. Observed no flickering while watching videos in Youtube and Facebook sites. I have tested this issue by opening multiple webpages and multiple windows in the background. Tested the same in incognito mode as well still no luck to reproduce it.

Dropping RBS since it doesn't seem reproducible and M-57 has already rolled out. We'll continue to monitor and try to find something actionable.

I got some reports over the weekend with chrome://gpu dumps:

1st report
http://pastebin.com/FqwisbeE

2nd report
http://pastebin.com/r5N3165p

I'm not sure if the 2nd report snapshot was taken the same moment the issue occurred though.

Thanks Daniel for these about:gpu dumps.

The first one is from a pretty new machine with Intel Iris Graphics 500 GPU. Not sure whether we've tested on that hardware.

The second one, however, is an older MacBook Pro with dual Intel/NVIDIA GPUs, the NVIDIA one being the GeForce GT 750M OpenGL Engine. We test on these all the time, and one of those is my primary work machine. I've never seen this behavior so we really need to know any hints about how this might be reproduced.

Sadly we are not aware of any specific steps to reproduce it, we just know that we've been seeing frequent reports for this and it appears to be happening on other sites (YouTube) as well. 

The intermittently showing frames seem to be belonging to other UI elements on the same page, so maybe some kind of texture recycling is re-using a wrong texture or some other component is not correctly releasing a re-cycled texture.

Saw the flicker once with 2012 Macbook Retina with OSX 10.12.2 which has dual GPU Intel/Nvidia GeForce GT 650 M , I saw the flicker once but again after testing for more then 40mins unable to see another flicker. 

Please find attached screenshots of GPU details before running any video and after.

Deleted: Before_playing_anyvideos.htm 50.0 KB

Before_playing_anyvideos.htm 50.0 KB View Download

Deleted: After_playing_videos.htm 48.7 KB

After_playing_videos.htm 48.7 KB View Download

Note: I saw this flicker this morning playing a YouTube video (on youtube.com, not embedded on another page). Unfortunately I left my personal laptop at home so can't gather about:gpu, but it's a 2012 era MacBook Pro with dual Intel and NVIDIA GeForce GT 650M GPUs. The browser session had been up a long time and survived multiple GPU process crashes (hangs and recovery via the watchdog -- possibly because of multiple suspend/resume cycles, though Chrome's supposed to deal more gracefully with that). I found a couple of GPU process crash IDs which I'll symbolize and see if there's anything suspicious. Overall, crashing the GPU process may be a way to provoke this.

I saw this over the weekend as well, also on YouTube.

I also grabbed about:gpu. It had lots of errors from several hours earlier, but no errors from the time at which I saw the flickering.

The flickering went away when I tab-switched to a different tab and returned.

I also attached an image of the "bad frame". It looks as though the resource for the play/pause button was somehow being used as a video frame incorrectly.

Deleted: about_gpu_after_flash.pdf 120 KB

about_gpu_after_flash.pdf 120 KB Download

Deleted: flicker.mov 2.7 MB

flicker.mov 2.7 MB Download

Deleted: Screen Shot 2017-03-27 at 11.12.45 AM.png 80.5 KB

	Screen Shot 2017-03-27 at 11.12.45 AM.png 80.5 KB View Download

 Issue 690598  has been merged into this issue.

Moving to Internals>Media>Video because it's not Blink>Media. Feel free to re-triage accordingly :)

Should we move this bug to available, or do we think it might not have found the right teams' backlog yet?

This is probably a GPU or compositing bug. At the very least the right people to contact are CC'd, but I don't think the component is correct yet.

There is a slim chance that if all of the contexts involved were virtual GL contexts, that the fix for  Issue 694359  may have fixed this.

Taking this for now.

Re #28, that sounds plausible. Youtube draws every a frame every few seconds where the video goes through a RenderPass (and hence hits the GL rendering path instead of CoreAnimation). So, I could believe that to be related to that issue.

ellyjones, are you still seeing this? Are other people still seeing this?

ccameron@ what are the next steps to take here?

I still see this.

That change hasn't made it to Chrome Stable yet. If you're testing with Chrome Dev Channel or Canary and still see this, please try to find reliable reproduction steps.

I see. I will test canary tomorrow and report back.

Re #30, I still don't have a repro of this (and I also haven't seen it in several months). If we can get an even-sometimes-reliable (like 1 out of 10 times) repro, then we can try debugging this. Without that, I'm not sure what can be done.

Agreed, reliable reproduction steps are still needed. I'm pretty sure I've seen this once recently on Chrome Canary, and reloading the page fixed it, so I doubt that the fix for  Issue 694359  fixed it.

At Facebook we have literally dozens of reports of this, but still no repro. 

The flickering is caused by the video briefly showing another element that is visible on the page instead of a frame of the video. This makes me speculate that this is likely a bug in some texture re-use code or something. It probably selects the wrong texture when painting the video frame and ends up painting the texture of another element on the page.

We are seeing this problem on our site soundslice.com, which plays video alongside a large <canvas> of music notation.

Example URL for your testing: https://www.soundslice.com/licks/328-rodrigo-gouveia-melodic-minor/

Here's a video demonstrating the flickering, made by one of our customers: https://www.youtube.com/watch?v=ZL2_nGVcjp4

In that video, you'll see that the "flicker" in the video displays Soundslice's music notation and player UI.

Unfortunately we don't have reliable reproduction steps yet. But it's definitely happening!

I think I've seen this once in the past month or so. I didn't get any useful information from about:gpu afterwards, IIRC.

This could be a bug in compositing or in CoreAnimation or in media. Still no way forward.

We are also having this issue and we can sometimes reproduce it on one of our dev's machine. But it doesn't always happen and we can't find a way to trigger it reliability.

We have screen recording when it happens but unfortunately we cannot release it because it contains sensitive information. 

I have attached some redacted screenshots.

Here are some of our observations we got:
It never happens on first clips.
The first flicker always happen at beginning of the clip, and it always repeat at constant rate (in once case, always 27 frames on a 24 fps video, but the rate can change, not sure if it is related the video been played)
When the flicker happens, the parts of the background colour of the page changes. See screenshot 2.
When the flicker repeats, the content it displays is always the snapshot of bottom part of the page when it happens for the first time with the changed background.

The first screenshot is the frame before the flicker happens. Everything is fine.
The second screenshot is the next frame of first screenshot. The bottom part the page is displayed as the video content.
Also the bottom part the background colour changed to black. Notice that the background changed part starts from the bottom of the video element and ends at bottom of the page. Also we had observation that sometimes it can end beyond the visible page area.
The last screenshot is 2 frames after second one (recording at 60FPS), the changed background is reverted back to normal. Still displaying the snapshot of the previous bottom part of the page.


We really want to have this bug fixed or need a workaround of it. Please let me know anything we can do to help resolve this issue.

This is about:gpu

Graphics Feature Status
Canvas: Hardware accelerated
CheckerImaging: Disabled
Flash: Hardware accelerated
Flash Stage3D: Hardware accelerated
Flash Stage3D Baseline profile: Hardware accelerated
Compositing: Hardware accelerated
Multiple Raster Threads: Enabled
Native GpuMemoryBuffers: Hardware accelerated
Rasterization: Hardware accelerated
Video Decode: Hardware accelerated
Video Encode: Hardware accelerated
WebGL: Hardware accelerated
WebGL2: Hardware accelerated
Driver Bug Workarounds
add_and_true_to_loop_condition
adjust_src_dst_region_for_blitframebuffer
decode_encode_srgb_for_generatemipmap
disable_framebuffer_cmaa
disable_multimonitor_multisampling
disable_webgl_rgb_multisampling_usage
emulate_abs_int_function
get_frag_data_info_bug
init_two_cube_map_levels_before_copyteximage
msaa_is_slow
pack_parameters_workaround_with_pack_buffer
rebind_transform_feedback_before_resume
regenerate_struct_names
remove_invariant_and_centroid_for_essl3
rewrite_texelfetchoffset_to_texelfetch
scalarize_vec_and_mat_constructor_args
set_zero_level_before_generating_mipmap
unfold_short_circuit_as_ternary_operation
unpack_alignment_workaround_with_unpack_buffer
unpack_image_height_workaround_with_unpack_buffer
use_intermediary_for_copy_texture_image
use_unused_standard_shared_blocks
avoid_stencil_buffers
Problems Detected
Multisampling is buggy on OSX when multiple monitors are connected: 237931
Applied Workarounds: disable_multimonitor_multisampling
Unfold short circuit on Mac OS X: 307751
Applied Workarounds: unfold_short_circuit_as_ternary_operation
Always rewrite vec/mat constructors to be consistent: 398694
Applied Workarounds: scalarize_vec_and_mat_constructor_args
Mac drivers handle struct scopes incorrectly: 403957
Applied Workarounds: regenerate_struct_names
On Intel GPUs MSAA performance is not acceptable for GPU rasterization: 527565
Applied Workarounds: msaa_is_slow
glGenerateMipmap fails if the zero texture level is not set on some Mac drivers: 560499
Applied Workarounds: set_zero_level_before_generating_mipmap
Pack parameters work incorrectly with pack buffer bound: 563714
Applied Workarounds: pack_parameters_workaround_with_pack_buffer
Alignment works incorrectly with unpack buffer bound: 563714
Applied Workarounds: unpack_alignment_workaround_with_unpack_buffer
copyTexImage2D fails when reading from IOSurface on multiple GPU types.: 581777
Applied Workarounds: use_intermediary_for_copy_texture_image
Multisample renderbuffers with format GL_RGB8 have performance issues on Intel GPUs.: 607130
Applied Workarounds: disable_webgl_rgb_multisampling_usage
Limited enabling of Chromium GL_INTEL_framebuffer_CMAA: 535198
Applied Workarounds: disable_framebuffer_cmaa
glGetFragData{Location|Index} works incorrectly on Max: 638340
Applied Workarounds: get_frag_data_info_bug
glResumeTransformFeedback works incorrectly on Intel GPUs: 638514
Applied Workarounds: rebind_transform_feedback_before_resume
Result of abs(i) where i is an integer in vertex shader is wrong: 642227
Applied Workarounds: emulate_abs_int_function
Rewrite texelFetchOffset to texelFetch for Intel Mac: 642605
Applied Workarounds: rewrite_texelfetchoffset_to_texelfetch
Rewrite condition in for and while loops for Intel Mac: 644669
Applied Workarounds: add_and_true_to_loop_condition
Decode and encode before generateMipmap for srgb format textures on macosx: 634519
Applied Workarounds: decode_encode_srgb_for_generatemipmap
Init first two levels before CopyTexImage2D for cube map texture on Intel Mac 10.12: 648197
Applied Workarounds: init_two_cube_map_levels_before_copyteximage
Insert statements to reference all members in unused std140/shared blocks on Mac: 618464
Applied Workarounds: use_unused_standard_shared_blocks
Tex(Sub)Image3D performs incorrectly when uploading from unpack buffer with GL_UNPACK_IMAGE_HEIGHT greater than zero on Intel Macs: 654258
Applied Workarounds: unpack_image_height_workaround_with_unpack_buffer
adjust src/dst region if blitting pixels outside read framebuffer on Mac: 644740
Applied Workarounds: adjust_src_dst_region_for_blitframebuffer
Mac driver GL 4.1 requires invariant and centroid to match between shaders: 639760, 641129
Applied Workarounds: remove_invariant_and_centroid_for_essl3
Disable KHR_blend_equation_advanced until cc shaders are updated: 661715
Certain Apple devices leak stencil buffers: 713854
Applied Workarounds: avoid_stencil_buffers
Checker-imaging has been disabled via finch trial or the command line.
Disabled Features: checker_imaging
Version Information
Data exported   25/09/2017, 16:56:40
Chrome version  Chrome/60.0.3112.113
Operating system    Mac OS X 10.12.6
Software rendering list version 13.9
Driver bug list version 10.94
ANGLE commit id 3e6a61fecba9
2D graphics backend Skia/60 d1f2d15b36f6a6a9d199581b998a7ca924a1f1a8-
Command Line    /Applications/Google Chrome.app/Contents/MacOS/Google Chrome --flag-switches-begin --flag-switches-end
Driver Information
Initialization time 451
In-process GPU  false
Passthrough Command Decoder false
Supports overlays   false
Sandboxed   true
GPU0    VENDOR = 0x8086, DEVICE= 0x1626 *ACTIVE*
Optimus false
Optimus false
AMD switchable  false
Driver vendor   
Driver version  10.25.17
Driver date 
Pixel shader version    4.10
Vertex shader version   4.10
Max. MSAA samples   8
Machine model name  iMac
Machine model version   16.1
GL_VENDOR   Intel Inc.
GL_RENDERER Intel(R) HD Graphics 6000
GL_VERSION  4.1 INTEL-10.25.17
GL_EXTENSIONS   GL_ARB_blend_func_extended GL_ARB_draw_buffers_blend GL_ARB_draw_indirect GL_ARB_ES2_compatibility GL_ARB_explicit_attrib_location GL_ARB_gpu_shader_fp64 GL_ARB_gpu_shader5 GL_ARB_instanced_arrays GL_ARB_internalformat_query GL_ARB_occlusion_query2 GL_ARB_sample_shading GL_ARB_sampler_objects GL_ARB_separate_shader_objects GL_ARB_shader_bit_encoding GL_ARB_shader_subroutine GL_ARB_shading_language_include GL_ARB_tessellation_shader GL_ARB_texture_buffer_object_rgb32 GL_ARB_texture_cube_map_array GL_ARB_texture_gather GL_ARB_texture_query_lod GL_ARB_texture_rgb10_a2ui GL_ARB_texture_storage GL_ARB_texture_swizzle GL_ARB_timer_query GL_ARB_transform_feedback2 GL_ARB_transform_feedback3 GL_ARB_vertex_attrib_64bit GL_ARB_vertex_type_2_10_10_10_rev GL_ARB_viewport_array GL_EXT_debug_label GL_EXT_debug_marker GL_EXT_framebuffer_multisample_blit_scaled GL_EXT_texture_compression_s3tc GL_EXT_texture_filter_anisotropic GL_EXT_texture_sRGB_decode GL_APPLE_client_storage GL_APPLE_container_object_shareable GL_APPLE_flush_render GL_APPLE_object_purgeable GL_APPLE_rgb_422 GL_APPLE_row_bytes GL_APPLE_texture_range GL_ATI_texture_mirror_once GL_NV_texture_barrier
Disabled Extensions GL_KHR_blend_equation_advanced GL_KHR_blend_equation_advanced_coherent
Window system binding vendor    
Window system binding version   
Window system binding extensions    
Direct rendering    Yes
Reset notification strategy 0x0000
GPU process crash count 0
Compositor Information
Tile Update Mode    Zero-copy
Partial Raster  Enabled
GpuMemoryBuffers Status
ATC Software only
ATCIA   Software only
DXT1    Software only
DXT5    Software only
ETC1    Software only
R_8 GPU_READ_CPU_READ_WRITE, GPU_READ_CPU_READ_WRITE_PERSISTENT
RG_88   Software only
BGR_565 Software only
RGBA_4444   Software only
RGBX_8888   Software only
RGBA_8888   GPU_READ, SCANOUT, SCANOUT_CPU_READ_WRITE
BGRX_8888   GPU_READ, SCANOUT, SCANOUT_CPU_READ_WRITE
BGRA_8888   GPU_READ, SCANOUT, SCANOUT_CPU_READ_WRITE, GPU_READ_CPU_READ_WRITE, GPU_READ_CPU_READ_WRITE_PERSISTENT
RGBA_F16    GPU_READ_CPU_READ_WRITE, GPU_READ_CPU_READ_WRITE_PERSISTENT
YVU_420 Software only
YUV_420_BIPLANAR    GPU_READ_CPU_READ_WRITE, GPU_READ_CPU_READ_WRITE_PERSISTENT
UYVY_422    GPU_READ_CPU_READ_WRITE, GPU_READ_CPU_READ_WRITE_PERSISTENT

Deleted: Screen Shot 2017-09-27 at 3.34.17 PM.png 889 KB

	Screen Shot 2017-09-27 at 3.34.17 PM.png 889 KB View Download

Deleted: Screen Shot 2017-09-27 at 3.34.31 PM.png 958 KB

	Screen Shot 2017-09-27 at 3.34.31 PM.png 958 KB View Download

Deleted: Screen Shot 2017-09-27 at 3.34.40 PM.png 965 KB

	Screen Shot 2017-09-27 at 3.34.40 PM.png 965 KB View Download

Re #40, if you can reproduce this frequently, can you take a screencapture of it when using the command line using the flag --show-mac-overlay-borders?

That will let us know what element is causing problems.

Once you get that, can you see if you can reproduce the bug using the flags
  --disable-avfoundation-overlays --show-mac-overlay-borders

And also, if you can get that, could you try with
  --disable-mac-overlays --show-mac-overlay-borders

Thank you! I know that this is very intermittent, but that would help us narrow down where this is occurring. My guess is that it is unrelated to Mac overlays, but that's an important first issue to be sure about.

Note: today's Canary on macOS:

Google Chrome	63.0.3227.0 (Official Build) canary (64-bit)
Revision	fae86d4eec4236e1a02db18228bd435402cc1020-refs/heads/master@{#505231}

contains the fix for  Issue 769488 , which may have implicitly solved this problem.

At the same time you're gathering data for ccameron@'s request in #42, above, it would be good to know if it looks like the problem is solved in Canary. I haven't been able to reproduce this often enough to conclusively tell.

Here is a recording with error reproduced with flag --show-mac-overlay-borders

We will try to reproduce it again with other flags and new Canary build.

Deleted: 1.mov 6.5 MB

1.mov 6.5 MB Download

It took us 40 mins to reproduce this bug for the first time and we tried 90 mins with flags --disable-avfoundation-overlays --show-mac-overlay-borders and another 90 mins with Canary 63.0.3229.0 build with flag --show-mac-overlay-borders and both of them doesn't show the issue.

Not sure it tells much given the natural of this bug...

This is recording with this bug happens on Canary 63.0.3229.0 build with flag --show-mac-overlay-borders

Deleted: 2.mov 5.2 MB

2.mov 5.2 MB Download

Thanks for the recording. Unfortunately it looks like the fix for  Issue 769488  didn't fix this.

Hopefully ccameron@ can narrow down the affected area now that the composited layer borders are known while the bug occurs.

We did one more test with Canary 63.0.3230.0 build with flags --disable-avfoundation-overlays --show-mac-overlay-borders and it is not happening any more.

We will do some more testing tomorrow to see if --disable-avfoundation-overlays does fixes the issue.

Is anything need to be worry about if we want to ship Electron app with with --disable-avfoundation-overlays flag?

--disable-avfoundation-overlays substantially increases h264 power usage, but should have no impact on stability.

This happens to me frequently as I browse imgur.com, as they encode/embed their gifs as mp4s. This may be redundant, but I figure more bug reports couldn't hurt! ¯\_(ツ)_/¯

Version 62.0.3202.45 (Official Build) beta (64-bit)

Graphics Feature Status
Canvas: Hardware accelerated
CheckerImaging: Disabled
Flash: Hardware accelerated
Flash Stage3D: Hardware accelerated
Flash Stage3D Baseline profile: Hardware accelerated
Compositing: Hardware accelerated
Multiple Raster Threads: Enabled
Native GpuMemoryBuffers: Hardware accelerated
Rasterization: Hardware accelerated
Video Decode: Hardware accelerated
Video Encode: Hardware accelerated
WebGL: Hardware accelerated
WebGL2: Hardware accelerated
Driver Bug Workarounds
add_and_true_to_loop_condition
adjust_src_dst_region_for_blitframebuffer
decode_encode_srgb_for_generatemipmap
disable_framebuffer_cmaa
disable_webgl_rgb_multisampling_usage
emulate_abs_int_function
get_frag_data_info_bug
init_two_cube_map_levels_before_copyteximage
msaa_is_slow
pack_parameters_workaround_with_pack_buffer
rebind_transform_feedback_before_resume
regenerate_struct_names
remove_invariant_and_centroid_for_essl3
reset_teximage2d_base_level
rewrite_texelfetchoffset_to_texelfetch
scalarize_vec_and_mat_constructor_args
set_zero_level_before_generating_mipmap
unfold_short_circuit_as_ternary_operation
unpack_alignment_workaround_with_unpack_buffer
unpack_image_height_workaround_with_unpack_buffer
use_intermediary_for_copy_texture_image
use_unused_standard_shared_blocks
Problems Detected
Unfold short circuit on Mac OS X: 307751
Applied Workarounds: unfold_short_circuit_as_ternary_operation
Always rewrite vec/mat constructors to be consistent: 398694
Applied Workarounds: scalarize_vec_and_mat_constructor_args
Mac drivers handle struct scopes incorrectly: 403957
Applied Workarounds: regenerate_struct_names
On Intel GPUs MSAA performance is not acceptable for GPU rasterization: 527565
Applied Workarounds: msaa_is_slow
glGenerateMipmap fails if the zero texture level is not set on some Mac drivers: 560499
Applied Workarounds: set_zero_level_before_generating_mipmap
Pack parameters work incorrectly with pack buffer bound: 563714
Applied Workarounds: pack_parameters_workaround_with_pack_buffer
Alignment works incorrectly with unpack buffer bound: 563714
Applied Workarounds: unpack_alignment_workaround_with_unpack_buffer
copyTexImage2D fails when reading from IOSurface on multiple GPU types.: 581777
Applied Workarounds: use_intermediary_for_copy_texture_image
Multisample renderbuffers with format GL_RGB8 have performance issues on Intel GPUs.: 607130
Applied Workarounds: disable_webgl_rgb_multisampling_usage
Limited enabling of Chromium GL_INTEL_framebuffer_CMAA: 535198
Applied Workarounds: disable_framebuffer_cmaa
glGetFragData{Location|Index} works incorrectly on Max: 638340
Applied Workarounds: get_frag_data_info_bug
glResumeTransformFeedback works incorrectly on Intel GPUs: 638514
Applied Workarounds: rebind_transform_feedback_before_resume
Result of abs(i) where i is an integer in vertex shader is wrong: 642227
Applied Workarounds: emulate_abs_int_function
Rewrite texelFetchOffset to texelFetch for Intel Mac: 642605
Applied Workarounds: rewrite_texelfetchoffset_to_texelfetch
Rewrite condition in for and while loops for Intel Mac: 644669
Applied Workarounds: add_and_true_to_loop_condition
Decode and encode before generateMipmap for srgb format textures on macosx: 634519
Applied Workarounds: decode_encode_srgb_for_generatemipmap
Init first two levels before CopyTexImage2D for cube map texture on Intel Mac 10.12: 648197
Applied Workarounds: init_two_cube_map_levels_before_copyteximage
Insert statements to reference all members in unused std140/shared blocks on Mac: 618464
Applied Workarounds: use_unused_standard_shared_blocks
Tex(Sub)Image3D performs incorrectly when uploading from unpack buffer with GL_UNPACK_IMAGE_HEIGHT greater than zero on Intel Macs: 654258
Applied Workarounds: unpack_image_height_workaround_with_unpack_buffer
adjust src/dst region if blitting pixels outside read framebuffer on Mac: 644740
Applied Workarounds: adjust_src_dst_region_for_blitframebuffer
Mac driver GL 4.1 requires invariant and centroid to match between shaders: 639760, 641129
Applied Workarounds: remove_invariant_and_centroid_for_essl3
Disable KHR_blend_equation_advanced until cc shaders are updated: 661715
Applied Workarounds: disable(GL_KHR_blend_equation_advanced), disable(GL_KHR_blend_equation_advanced_coherent)
On Intel GPUs MSAA performance is not acceptable for GPU rasterization: 527565, 751919
Applied Workarounds: msaa_is_slow
Reset TexImage2D base level to 0 on Intel Mac 10.12.4: 705865
Applied Workarounds: reset_teximage2d_base_level
Checker-imaging has been disabled via finch trial or the command line.
Disabled Features: checker_imaging
Version Information
Data exported	10/7/2017, 10:18:04 AM
Chrome version	Chrome/62.0.3202.45
Operating system	Mac OS X 10.13.0
Software rendering list version	13.12
Driver bug list version	10.30
ANGLE commit id	e8ef2bc4bd01
2D graphics backend	Skia/62 fc6dd6fa4c2105bfd9d91f0a1ab0a17e398a291a-
Command Line	/Applications/Google Chrome.app/Contents/MacOS/Google Chrome --show-mac-overlay-borders --flag-switches-begin --enable-devtools-experiments --enable-translate-new-ux --secondary-ui-md --disable-features=EnableManualFallbacksFilling --flag-switches-end
Driver Information
Initialization time	150
In-process GPU	false
Passthrough Command Decoder	false
Supports overlays	false
Sandboxed	true
GPU0	VENDOR = 0x8086, DEVICE= 0x5927 *ACTIVE*
Optimus	false
Optimus	false
AMD switchable	false
Driver vendor	
Driver version	10.28.26
Driver date	
Pixel shader version	4.10
Vertex shader version	4.10
Max. MSAA samples	16
Machine model name	MacBookPro
Machine model version	14.2
GL_VENDOR	Intel Inc.
GL_RENDERER	Intel(R) Iris(TM) Graphics 650
GL_VERSION	4.1 INTEL-10.28.26
GL_EXTENSIONS	GL_ARB_blend_func_extended GL_ARB_draw_buffers_blend GL_ARB_draw_indirect GL_ARB_ES2_compatibility GL_ARB_explicit_attrib_location GL_ARB_gpu_shader_fp64 GL_ARB_gpu_shader5 GL_ARB_instanced_arrays GL_ARB_internalformat_query GL_ARB_occlusion_query2 GL_ARB_sample_shading GL_ARB_sampler_objects GL_ARB_separate_shader_objects GL_ARB_shader_bit_encoding GL_ARB_shader_subroutine GL_ARB_shading_language_include GL_ARB_tessellation_shader GL_ARB_texture_buffer_object_rgb32 GL_ARB_texture_cube_map_array GL_ARB_texture_gather GL_ARB_texture_query_lod GL_ARB_texture_rgb10_a2ui GL_ARB_texture_storage GL_ARB_texture_swizzle GL_ARB_timer_query GL_ARB_transform_feedback2 GL_ARB_transform_feedback3 GL_ARB_vertex_attrib_64bit GL_ARB_vertex_type_2_10_10_10_rev GL_ARB_viewport_array GL_EXT_debug_label GL_EXT_debug_marker GL_EXT_framebuffer_multisample_blit_scaled GL_EXT_texture_compression_s3tc GL_EXT_texture_filter_anisotropic GL_EXT_texture_sRGB_decode GL_APPLE_client_storage GL_APPLE_container_object_shareable GL_APPLE_flush_render GL_APPLE_object_purgeable GL_APPLE_rgb_422 GL_APPLE_row_bytes GL_APPLE_texture_range GL_ATI_texture_mirror_once GL_NV_texture_barrier
Disabled Extensions	GL_KHR_blend_equation_advanced GL_KHR_blend_equation_advanced_coherent
Window system binding vendor	
Window system binding version	
Window system binding extensions	
Direct rendering	Yes
Reset notification strategy	0x0000
GPU process crash count	0
Compositor Information
Tile Update Mode	Zero-copy
Partial Raster	Enabled
GpuMemoryBuffers Status
ATC	Software only
ATCIA	Software only
DXT1	Software only
DXT5	Software only
ETC1	Software only
R_8	GPU_READ_CPU_READ_WRITE, GPU_READ_CPU_READ_WRITE_PERSISTENT
R_16	Software only
RG_88	Software only
BGR_565	Software only
RGBA_4444	Software only
RGBX_8888	Software only
RGBA_8888	GPU_READ, SCANOUT, SCANOUT_CPU_READ_WRITE
BGRX_8888	GPU_READ, SCANOUT, SCANOUT_CPU_READ_WRITE
BGRA_8888	GPU_READ, SCANOUT, SCANOUT_CPU_READ_WRITE, GPU_READ_CPU_READ_WRITE, GPU_READ_CPU_READ_WRITE_PERSISTENT
RGBA_F16	GPU_READ_CPU_READ_WRITE, GPU_READ_CPU_READ_WRITE_PERSISTENT
YVU_420	Software only
YUV_420_BIPLANAR	GPU_READ_CPU_READ_WRITE, GPU_READ_CPU_READ_WRITE_PERSISTENT
UYVY_422	GPU_READ_CPU_READ_WRITE, GPU_READ_CPU_READ_WRITE_PERSISTENT

Deleted: chrome_x265.mp4 2.3 MB

	chrome_x265.mp4 2.3 MB View Download

 Issue 776523  has been merged into this issue.

 Issue 781598  has been merged into this issue.

Hi there,
I'm experiencing the same issue for a long time. Still present on 63.0.3239.84

Not sure what was change on youtube, but recently it's happening very rarely on youtube, on the other hand it happens a lot on facebook and almost every other websites.

My laptop is Macbook Pro 13" Late 2013

I continue to experience this very intermittently. I never get any errors from about:gpu. I'm not 100% certain if the affected videos were software decode or hardware decode.

For me, the problem completely disappears when I turn hardware acceleration off. (though I would like to be able to turn in on, of course)

 Issue 800231  has been merged into this issue.

Has there been any progress on this? I can have it happen very consistently (within a few mins of each other) so if any testing is needed just let me know.

I've gotten this with hardware decode and noticed that when the AVSampleBufferDisplayLayer is recreated, the issue seems to go away.

I will add some LOG spew from -[AVSampleBufferDisplayLayer error], in case that is the source of problems.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/dd0002c774dca814c36ed931d3cf8d696c085e4a

commit dd0002c774dca814c36ed931d3cf8d696c085e4a
Author: Christopher Cameron <ccameron@chromium.org>
Date: Fri Jan 12 04:46:49 2018

Log AVSampleBufferDisplayLayer status

Dump this status so that it will appear in about:gpu. Hopefully this
will give a hint as to the cause of intermittent video flickering.

Bug:  702369 
Change-Id: I4d31adf1ba6086f6de43bc3ed27cda8afe068cc9
Reviewed-on: https://chromium-review.googlesource.com/863222
Reviewed-by: Trent Apted <tapted@chromium.org>
Commit-Queue: ccameron <ccameron@chromium.org>
Cr-Commit-Position: refs/heads/master@{#528880}
[modify] https://crrev.com/dd0002c774dca814c36ed931d3cf8d696c085e4a/ui/accelerated_widget_mac/ca_renderer_layer_tree.mm

Reproduced the issue with 65.0.3321.0 (r529180), and none of the spew from the above r528880 showed up. This was with h264 accelerated decode (AFAICT).

The machine was briefly in a state where it would reproduce this bug pretty often (about 1 in 5 video plays) -- I even got it to reproduce in a Chromium release build, but the machine got out of that state and the last ~100 runs haven't managed to reproduce.

The video in question was: https://www.youtube.com/watch?v=4cs3C6Z_3gU

So I'm still pretty stumped here. This seems to go away when we don't use an AVSampleBufferDisplayLayer -- it may be that we can just destroy-then-re-create the AVSampleBufferDisplayLayer very ~1 second so that, if we ever end up in this state, we at least exit it "soonish". We'll need to verify the power implications for this ...

I can reproduce the issue almost every time. I just open multiple videos and live streams in few dozen tabs and switch tabs while all videos are playing. The issue appears in multiple videos while it was NOT affected when I started playing the videos. 

I played with this for more hours than I want to think of, and now...
- I could now reproduce this almost-every time in Chrome Canary (with a particular video and opening sequence)
- I could rarely reproduce this with Chrome (same account and everything)
- I could rarely reproduce this with Chromium
- I tried using Chromium with my  Chrome Canary user data dir and it happens almost-every-time!!

I can now actually debug this issue!!

I've made this not happen(*) for this ONE particular repro.

This is a video that is decided in hardware, and gives us a CVPixelBuffer directly from the VTDecompressionSession.

Our AVSampleBufferDisplayLayer consumes CVPixelBuffers directly, so we just send the CVPixelBuffer directly in.

This is in contrast with software decoded videos, we get an IOSurface, and so we wrap the IOSurface in a CVPixelBuffer to feed it to the AVSampleBufferDisplayLayer.

The "fix" is taking the CVPixelBuffer we get from the VTDecompressionSession, extracting its IOSurface, then wrapping that in a new CVPixelBuffer, and feeding the new CVPixelBuffer to the AVSampleBufferDisplayLayer.

This does not strike me as the "right" thing to do. In fact, I was expecting that doing the opposite (restricting AVSampleBufferDisplayLayer use to directly-handed-to-us CVPixelBuffers was the fix)

(*) I have seen single-frame flashes of wrong content, but it's just one frame and then never appears again. In all previous instances of this, the flash would happen every ~10 frames forever.

I went an ran this 50 times in both configurations (which was indeed as tedious as it sounds), and the results are as follows:

                                    | IOSurface | CVPixelBuffer
 -----------------------------------+-----------+---------------
  Plays w/no error:                 |        46 |            39
  Plays w/exactly 1 flashed bug:    |         4 |             0
  Plays w/constantly flashing bug:  |         0 |            21

When using the decoded CVPixelBuffers, we re-use the same pool of CVPixelBuffers over and over again. When re-wrapping the CVPixelBuffers, we mostly re-use the same underlying IOSurfaces over and over again, but the CVPixelBuffers are new every frame.

I can believe the narrative that some of the CVPixelBuffers are somehow "broken" and that re-wrapping them ensures a "broken" CVPixelBuffers only affect one frame, rather than re-appearing over and over.

(again it's odd that the thing I did to try to make this happen more often ended up causing it to stop happening)

(that should read 29, not 39 above)

Tested this issue on Mac 10.12.6 using chrome latest-65.0.3322.0 as per C#0.Observed no flickering while playing youtube.com and facebook.com videos.

Please find the attached screencast & GPU details for reference.

daniel.baulig@,Could you please check this issue and confirm the fix.

Thanks..1

Deleted: 702369-GPU.rtf 74.5 KB

702369-GPU.rtf 74.5 KB Download

Deleted: 702369-Mac.mp4 11.5 MB

	702369-Mac.mp4 11.5 MB View Download

Two small observations (I can sometimes reproduce it with the video you linked, ccameron@):

1. Scrolling down a bit makes the issue disappear.
2. I can pause the video on the bad frame and the bad image stays, but scrolling down while it's paused makes it switch back to the right image (and the issue goes away).

Re #68: We destroy and re-create the AVSampleBufferDisplayLayer when it goes off-screen. So, that's why I did the patch in #60 -- I assumed something was broken with the AVSampleBufferDisplayLayer.

Now it seems like it's something busted with the CVPixelBuffer :/

I'll try to land the "fix" ASAP

Re #69: Just to clarify, it happens when even a small amount of it is offscreen. Here's a video.

Deleted: fix_on_scroll.mp4 1.8 MB

	fix_on_scroll.mp4 1.8 MB View Download

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/1b8d9dce9e641cd9aa646aa293b1a0441fe8d071

commit 1b8d9dce9e641cd9aa646aa293b1a0441fe8d071
Author: Christopher Cameron <ccameron@chromium.org>
Date: Thu Jan 18 23:59:48 2018

Never directly use CVPixelBufferRefs from VTDecompressionSession

Re-wrapping the CVPixelBufferRefs's IOSurface in pristine
CVPixelBufferRefs makes a bug where frames repeatly flash incorrect
content go away.

There is no clear or documented reason why this should be required.
Rather, it seems more intuitive to use the VTDecompressionSession
CVPixelbufferRefs directly.

This was observed only in a single fairly consistent repro case, so it
is not clear that this indeed fixes the issue -- there was still
occasionally a single flashed incorrect frame even with this fix in
place.

It may be that this fix makes the underlying issue less frequent, or it
may be that this fix makes it so that a single "bad" CVPixelBuffer that
we inevitably encounter (say, due to some peculiarity of the decoder)
does not get repeatedly reused.

This fix to increases fullscreen hardware decode video power consumption
from 1.57W to 1.65W, entirely due to an increase in CPU power
consumption. If this does indeed fix the bug, I consider this to be an
acceptable price. This fix has no appreciable effect on non-fullscreen
video power consumption.

Bug:  702369 
Cq-Include-Trybots: master.tryserver.chromium.android:android_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel
Change-Id: Ibd5f937695bbd7bb9c16084c3b23dc1968ee83c4
Reviewed-on: https://chromium-review.googlesource.com/867413
Commit-Queue: ccameron <ccameron@chromium.org>
Reviewed-by: Sidney San Martín <sdy@chromium.org>
Cr-Commit-Position: refs/heads/master@{#530351}
[modify] https://crrev.com/1b8d9dce9e641cd9aa646aa293b1a0441fe8d071/ui/accelerated_widget_mac/ca_renderer_layer_tree.mm

Can folks on 65.0.3325.0+ chime in here if the issue has been resolved for them?

I'm on 66.0.3332.0 canary. Seems to be fixed for me, at least I failed to reproduce it for 10 minutes. Will write here if it happens again.
So what was the cause of the issue?

Thanks! See comment #71 for Chris' explanation.

So far I only got occasional flickers which go away after less than a second as mentioned by Chris. It's bearable compared to frequent and persistent mess ups.

Thanks for verifying -- I've seen a few single-frame-mess-ups, but nothing persistent.

Ultimately I don't have an explanation for why this should have fixed the issue. I would imagine it to be a bug in AVFoundation.

Thanks for fixing this, ccameron :)

Thank you for fixing this indeed. I have noticed similar flickering on latest Firefox too, so I guess this is related to AVFoundation or gpu drivers.

Do we have a radar open for this if we think it's an AVFoundation bug?

Unfortunately just observed persistent flickering again while watching video on facebook.

Chrome Version 64.0.3282.140

It happened to me too yesterday on youtube, on Chrome 64.0.3282.140

Am I correct in assuming this should be fixed in Chrome 65?

It seems to be first available in 65.0.3325.0:
https://storage.googleapis.com/chromium-find-releases-static/1b8.html#1b8d9dce9e641cd9aa646aa293b1a0441fe8d071

 Issue 808794  has been merged into this issue.