Unable to test this issue from TE end as we do not have minix U9-H device with us. Hence adding TE-Hardware-dependency label and routing it to dev team for further triaging of this.

Thanks!

Manage to get it.


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: Disabled
Native GpuMemoryBuffers: Software only. Hardware acceleration disabled
Rasterization: Hardware accelerated
Surface Synchronization: Disabled
Video Decode: Hardware accelerated
Viz Service Display Compositor: Disabled
WebGL: Hardware accelerated
WebGL2: Hardware accelerated
Driver Bug Workarounds
clear_uniforms_before_first_program_use
disable_framebuffer_cmaa
dont_disable_webgl_when_compositor_context_lost
max_msaa_sample_count_4
max_texture_size_limit_4096
scalarize_vec_and_mat_constructor_args
use_client_side_arrays_for_stream_buffers
use_virtualized_gl_contexts
Problems Detected
ARM driver doesn't like uploading lots of buffer data constantly: 178093
Applied Workarounds: use_client_side_arrays_for_stream_buffers
The Mali-Txxx driver does not guarantee flush ordering: 154715, 10068, 269829, 294779, 285292
Applied Workarounds: use_virtualized_gl_contexts
Clear uniforms before first program use on all platforms: 124764, 349137
Applied Workarounds: clear_uniforms_before_first_program_use
Always rewrite vec/mat constructors to be consistent: 398694
Applied Workarounds: scalarize_vec_and_mat_constructor_args
Limit max texure size to 4096 on all of Android
Applied Workarounds: max_texture_size_limit_4096
Use GL_INTEL_framebuffer_CMAA on ChromeOS: 535198
Applied Workarounds: disable_framebuffer_cmaa
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)
eglSwapBuffers intermittently fails on Android when app goes to background: 744678
Applied Workarounds: dont_disable_webgl_when_compositor_context_lost
Don't expose disjoint_timer_query extensions to WebGL: 808744
Limit MSAA to 4x on Android devices: 797243
Applied Workarounds: max_msaa_sample_count_4
Raster is using a single thread.
Disabled Features: multiple_raster_threads
Native GpuMemoryBuffers have been disabled, either via about:flags or command line.
Disabled Features: native_gpu_memory_buffers
Surface synchronization has been disabled by Finch trial or command line.
Disabled Features: surface_synchronization
Viz service display compositor is not enabled by default.
Disabled Features: viz_display_compositor
Checker-imaging has been disabled via finch trial or the command line.
Disabled Features: checker_imaging
Version Information
Data exported	2018-07-23T20:59:40.784Z
Chrome version	Chrome/67.0.3396.87
Operating system	Android 6.0.1
Software rendering list URL	https://chromium.googlesource.com/chromium/src/+/878cd31214ac27a3996927cd5c9c138b10c9fc8d/gpu/config/software_rendering_list.json
Driver bug list URL	https://chromium.googlesource.com/chromium/src/+/878cd31214ac27a3996927cd5c9c138b10c9fc8d/gpu/config/gpu_driver_bug_list.json
ANGLE commit id	702006f4a07e
2D graphics backend	Skia/67 78b60f4ff13b83da98ae2bca85aaef0a98b61098-
Command Line	--enable-pinch --enable-viewport --validate-input-event-stream --enable-longpress-drag-selection --touch-selection-strategy=direction --main-frame-resizes-are-orientation-changes --disable-composited-antialiasing --enable-dom-distiller --flag-switches-begin --flag-switches-end --top-controls-show-threshold=0.27 --top-controls-hide-threshold=0.17
Driver Information
Initialization time	249
In-process GPU	false
Passthrough Command Decoder	false
Direct Composition	false
Supports overlays	false
Sandboxed	false
GPU0	VENDOR = 0x0000 [ARM], DEVICE= 0x0000 [Mali-T820] *ACTIVE*
Optimus	false
AMD switchable	false
Driver vendor	
Driver version	1.
Driver date	
Pixel shader version	3.20
Vertex shader version	3.20
Max. MSAA samples	4
Machine model name	NEO-U9-H
Machine model version	
GL_VENDOR	ARM
GL_RENDERER	Mali-T820
GL_VERSION	OpenGL ES 3.2 v1.r11p0-00rel0.f83e33152085d4fe9267c1d35c6f9fca
GL_EXTENSIONS	GL_EXT_debug_marker GL_ARM_rgba8 GL_ARM_mali_shader_binary GL_OES_depth24 GL_OES_depth_texture GL_OES_depth_texture_cube_map GL_OES_packed_depth_stencil GL_OES_rgb8_rgba8 GL_EXT_read_format_bgra GL_OES_compressed_paletted_texture GL_OES_compressed_ETC1_RGB8_texture GL_OES_standard_derivatives GL_OES_EGL_image GL_OES_EGL_image_external GL_OES_EGL_image_external_essl3 GL_OES_EGL_sync GL_OES_texture_npot GL_OES_vertex_half_float GL_OES_required_internalformat GL_OES_vertex_array_object GL_OES_mapbuffer GL_EXT_texture_format_BGRA8888 GL_EXT_texture_rg GL_EXT_texture_type_2_10_10_10_REV GL_OES_fbo_render_mipmap GL_OES_element_index_uint GL_EXT_shadow_samplers GL_OES_texture_compression_astc GL_KHR_texture_compression_astc_ldr GL_KHR_texture_compression_astc_hdr GL_KHR_texture_compression_astc_sliced_3d GL_KHR_debug GL_EXT_occlusion_query_boolean GL_EXT_disjoint_timer_query GL_EXT_blend_minmax GL_EXT_discard_framebuffer GL_OES_get_program_binary GL_OES_texture_3D GL_EXT_texture_storage GL_EXT_multisampled_render_to_texture GL_OES_surfaceless_context GL_OES_texture_stencil8 GL_EXT_shader_pixel_local_storage GL_ARM_shader_framebuffer_fetch GL_ARM_shader_framebuffer_fetch_depth_stencil GL_ARM_mali_program_binary GL_EXT_sRGB GL_EXT_sRGB_write_control GL_EXT_texture_sRGB_decode GL_KHR_blend_equation_advanced GL_KHR_blend_equation_advanced_coherent GL_OES_texture_storage_multisample_2d_array GL_OES_shader_image_atomic GL_EXT_robustness GL_EXT_draw_buffers_indexed GL_OES_draw_buffers_indexed GL_EXT_texture_border_clamp GL_OES_texture_border_clamp GL_EXT_texture_cube_map_array GL_OES_texture_cube_map_array GL_OES_sample_variables GL_OES_sample_shading GL_OES_shader_multisample_interpolation GL_EXT_shader_io_blocks GL_OES_shader_io_blocks GL_EXT_tessellation_shader GL_OES_tessellation_shader GL_EXT_primitive_bounding_box GL_OES_primitive_bounding_box GL_EXT_geometry_shader GL_OES_geometry_shader GL_ANDROID_extension_pack_es31a GL_EXT_gpu_shader5 GL_OES_gpu_shader5 GL_EXT_texture_buffer GL_OES_texture_buffer GL_EXT_copy_image GL_OES_copy_image GL_EXT_color_buffer_half_float GL_EXT_color_buffer_float GL_OVR_multiview GL_OVR_multiview2 GL_OVR_multiview_multisampled_render_to_texture GL_ARM_packed_arithmetic GL_KHR_robustness GL_KHR_robust_buffer_access_behavior GL_EXT_draw_elements_base_vertex GL_OES_draw_elements_base_vertex
Disabled Extensions	GL_KHR_blend_equation_advanced GL_KHR_blend_equation_advanced_coherent
Disabled WebGL Extensions	EXT_disjoint_timer_query EXT_disjoint_timer_query_webgl2
Window system binding vendor	
Window system binding version	
Window system binding extensions	
Direct rendering	Yes
Reset notification strategy	0x8252
GPU process crash count	0
Compositor Information
Tile Update Mode	One-copy
Partial Raster	Enabled
GpuMemoryBuffers Status
ATC	Software only
ATCIA	Software only
DXT1	Software only
DXT5	Software only
ETC1	Software only
R_8	Software only
R_16	Software only
RG_88	Software only
BGR_565	Software only
RGBA_4444	Software only
RGBX_8888	Software only
RGBA_8888	Software only
BGRX_8888	Software only
BGRX_1010102	Software only
RGBX_1010102	Software only
BGRA_8888	Software only
RGBA_F16	Software only
YVU_420	Software only
YUV_420_BIPLANAR	Software only
UYVY_422	Software only
Display(s) Information
Info	Display[0] bounds=[0,0 1280x720], workarea=[0,0 1280x720], scale=1.5, external.
Color space information	{primaries:BT709, transfer:IEC61966_2_1, matrix:RGB, range:FULL}
Bits per color component	8
Bits per pixel	24
Video Acceleration Information
Decode vp8	up to 3840x2160 pixels (encrypted)
Decode vp8	up to 3840x2160 pixels (encrypted)
Decode vp9 profile0	up to 3840x2160 pixels
Decode vp9 profile0	up to 3840x2160 pixels (encrypted)
Decode vp9 profile1	up to 3840x2160 pixels
Decode vp9 profile1	up to 3840x2160 pixels (encrypted)
Decode vp9 profile2	up to 3840x2160 pixels
Decode vp9 profile2	up to 3840x2160 pixels (encrypted)
Decode vp9 profile3	up to 3840x2160 pixels
Decode vp9 profile3	up to 3840x2160 pixels (encrypted)
Decode h264 baseline	up to 3840x2160 pixels
Decode h264 main	up to 3840x2160 pixels
Decode h264 extended	up to 3840x2160 pixels
Decode h264 high	up to 3840x2160 pixels
Decode h264 high 10	up to 3840x2160 pixels
Decode h264 high 4:2:2	up to 3840x2160 pixels
Decode h264 high 4:4:4 predictive	up to 3840x2160 pixels
Decode h264 scalable baseline	up to 3840x2160 pixels
Decode h264 scalable high	up to 3840x2160 pixels
Decode h264 stereo high	up to 3840x2160 pixels
Decode h264 multiview high	up to 3840x2160 pixels

The same bug happen with my android mobile phone.

Thanks. Everything looks okay in there. If you can grab chrome://media-internals when the playback goes bad that'd be helpful. Can you also check if there are any entries in chrome://crashes ?

Youtube tv
1. Play a video of youtube tv in landscape mode
2. Click "full screen" button
3. Then you can find screen frozen

More Info:
url: tv.youtube.com
device: any android mobile phone
chrmoium version: M68

ffpax91: what android phone do you have?

I try it on some different android phones and it always happen.  Besides for the same android phone, it works well with firefox.

Le Max2, Le x526, n6p, xiaomi 8

chelamcherla: do you have any of these devices? We don't over here.

Maybe you can try it on any android phone. It always happen.

We did, but are unable to reproduce on common phones. E.g., it does not happen on any Pixel device I've tried.

As per comment#12 tested this issue on Nexus 6p and xiaomi redmi note 3 with steps mentioned in comment#1 and unable to reproduce this issue.

Unable to test with tv.youtube.com as in India we do not have access to youtube tv. Hence requesting someone from MTV team to please look into this issue.

Thanks!

In landscape mode and full screen,it works well with video of www.youtube.com,but bug happens when play video of tv.youtube.com.

"We did, but are unable to reproduce on common phones. E.g., it does not happen on any Pixel device I've tried." I want to know how you do it?

+xhwang. I only tested normal YT not YTTV. Sorry didn't notice your link was different then the reporters. Sounds like it might be EME related, do you have a TV test account still xhwang?

+strobe for routing before we go expense an account for YTTV.

- MTV label since the account expense is already taken care.

Is there any progress,please?

Sorry haven't had a chance to dig further yet. Will try to get to it this week.

+dougman, does the test team have a YT TV account and Android device they can test?

 By FireFox， video of youtube tv plays well with full screen in landscape mode.

Okay, I can repro the issue with YouTube TV. You need to request desktop site and then hit fullscreen button (upper right corner). If you keep portrait orientation everything works, but as soon as you attempt to rotate to landscape the video scrolls off the screen and doesn't come back.

You can hear audio for a little while after this, but eventually it freezes and then you're unable to exit fullscreen or navigate to any other domain once in this state. Inspect lets you connect, but then you can't see any elements. This seems like a pretty hard crash on fullscreen EME content.

=>liberato@ to take a look. 

Continually tapping the back button eventually gets you out of this state. If nothing else we shouldn't lock up the browser like this.

Oh nothing interesting in the adb log either.

re c#25, which device are you using?  did you end up being able to repro it on a pixel?

if it's switching to overlay, and it's offscreen, i have no idea what will happen.  if i had to guess, gl_renderer would skip the overlay promotion, just like it skips the draw for non-overlay offscreen videos.  we should still return the video frames and release the back buffers without rendering.

kinda sounds like that's not happening, though.

i wonder if one scrolls a non-fullscreen secure video off the screen if it still breaks.

i'll try to repro.

Pixel2 XL. It only happens on YouTube TV. Not the main site.

Is there any way to fix this bug, please?

i'm trying to repro locally.

i can finally get this to show up locally on my S8+.

one thing i notice about youtube tv is that one must tap the video to get the controls overlay.  tapping the background around the video seems to do nothing. that seems like a site issue, and would explain why the site seems to be locked up when the video is off-screen.

now, i need to figure out why playback stops.

separately, i tried scrolling L1 content off-screen with the shaka demo player.  seemed fine there.

also, i tried a different video (Saturday Night Live), and the video did not scroll entirely off the screen when i switched to fullscreen landscape.  it was definitely clipped, but playing normally otherwise.

it did, eventually, freeze.

it looks like this is an app issue.  at some point, chrome tries to read from the video stream, but the demuxer never returns any.

it does continue to get audio buffers for quite some time, though.

the decoder seems to be working fine, video frames are being returned normally, and everything else looks correct.  we just stop getting video data from the demuxer at some point.

That shouldn't cause things to hang that bad though. Like I can't even connect dev tools when this occurs, which seems like we dead lock the main thread maybe.

(Which would also explain why demuxer runs out of data :)

i didn't notice that it was that broken.  i don't remember if the back button exited full screen or not.

i should point out that the demuxer didn't run out of audio data for a while -- it got back a dozen or so audio buffers before it quit.

i'll poke at it more.

Probably just can't append new data, rest of calls happen on media thread.

By FireFox，follow the same steps and video of youtube tv plays well with full screen in landscape mode.

re c#34: good point.  the renderer main thread is spinning at 99% when this happens.

i'll see what's doing that.

the main thread seems to be calling into some v8 function that never returns on a dom timer.  i'm trying to figure out how to convert a v8::Function into some sort of name or line number or something.

It looks like it should be really easy, what with methods like GetDebugName(), but it isn't.  :)

+V8 folks for help.

the call is something called Debouncer.boundComplete @ line 137, column 146 .

i have no idea what this function is doing.  it's something inside polymer:

this.boundComplete=function(){self.complete()}};

Hmm, I guess it's a site issue then. I didn't think a rogue JS script could bring down the renderer that hard though.

Seems like WontFix from our side then. Thanks for looking into it more. ping strobe@ in case it's interesting that YouTube TV on Android (via request desktop site) causes hangs.

It is interesting that it works in Firefox mweb though.

Debouncer just aggregates multiple calls in one microtask into a single call to some user-supplied function.

so, pretty much i've discovered that it's "doing something".

Why？ Why wontfix？This problem is real！
By FireFox，follow the same steps and video of youtube tv plays well with full screen in landscape mode.
Can you give me a reasonable explanation?

@liberato: I'm still surprised that this brings down the browser so hard. I would expect a "page not responding" dialog or something - but as is you can't even change tabs or do anything when this happens. +boliu who might know the right people to route this issue to if it's a page hang. (repro steps are in c#25)

do you have a trace of when this happens? is the main thread actually not yeilding the message loop, or it's just doing something super expensive constantly?

I think task scheduling puts frame production tasks above js tasks, so as long as js is yielding the loop, it shouldn't hang that badly. And if it is not yielding the loop, then yeah, the "page not responding" thing should come up.

c#50: sorry, missed the last comment.  i don't have a chrome://tracing trace, but based on logging i think it was not yielding the message loop.  the message loop would run a posted task => blink => v8 => polymer, but the posted task never returned.  so the main thread is stuck.

re c#48: i marked it WontFix only because it looks like chromium is doing what the site is telling it to do.  to get the issue resolved, my best guess is that the next place to look is in the youtube tv site rather than chromium.

i've notified the youtube tv folks, who might have some insight about why this is happening.

as for why it's different between browsers, it's possible that the site (or, possibly, a library that the site uses) has code that behaves differently on different browsers.  it's common for sites to tweak their behavior to work around the quirks of whatever browser is being used.

sorry about the confusion.

 Issue 875701  has been merged into this issue.

More info about the "page not responding" thing in case you want to investigate. It monitors timeout on input ack, so requires user trying to interact with the page to show up. Code is here in case you want to check what exactly happens:

https://cs.chromium.org/chromium/src/content/browser/renderer_host/render_widget_host_impl.cc?rcl=17a40397711ee604334d86f9533f1ea1cc327c3f&l=408

Maybe in fullscreen that dialog is not hooked up? Or maybe input is not going to blink main thread, and is ack-ed by compositor thread? I dunno these things in detail

Hmm, can't seem to get a trace out of this, records and progress increases in percentage and then drops to 0% and stop does nothing :/ No page not responding is coming up, and I can trigger this issue without entering fullscreen.

Attempting a second trace causes trace to show "no categories found; recording will use default categories" so some sort of trace state gets wiped too.

Oh hey, this happens on desktop linux too if you use dev tools and force "pixel 2 xl" device and click rotate a bunch of times. The browser process is still fairly hosed, it can't start tracing or anything, YT tab is showed at 130% cpu. Forced some crashes that I'm looking through now. I'm still not sure why not-responding doesn't come up, but it's not just a mobile issue.

Curiously smaller resolution devices don't cause any problems. It's just the p2xl size that breaks.

This is just a YouTube TV bug as Frank found. I'll try on my Pixelbook later today and see if it repros there, if so we can file a bug against YTTV team to take a look, but otherwise YT TV isn't launched on mweb, so this will likely be WontFix.

YouTube folk are cc'd in case they launch on mweb later, but for now you should just use the app.