Thanks for the report. Could you please copy/paste the plaintext contents of about:gpu?

Filed as Radar 30727388 with Apple.

Marking ExternalDependency. We've raised this as a critical regression, and hopefully it will be fixed before 10.12.4 ships.

about:gpu:
--------------------
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	2017-2-27 08:55:57
Chrome version	Chrome/58.0.3024.0
Operating system	Mac OS X 10.12.4
Software rendering list version	12.16
Driver bug list version	9.31
ANGLE commit id	0a17c92c2927
2D graphics backend	Skia/58 7f15b682f4e3e0f61a56d3bbdb8163d3a54b997c
Command Line Args	Chrome Canary.app/Contents/MacOS/Google Chrome Canary --flag-switches-begin --enable-devtools-experiments --enable-features=AsmJsToWebAssembly,WebAssembly --flag-switches-end
Driver Information
Initialization time	30
In-process GPU	false
Passthrough Command Decoder	false
Sandboxed	true
GPU0	VENDOR = 0x8086, DEVICE= 0x0a2e *ACTIVE*
Optimus	false
AMD switchable	false
Driver vendor	
Driver version	10.24.38
Driver date	
Pixel shader version	4.10
Vertex shader version	4.10
Max. MSAA samples	8
Machine model name	MacBookPro
Machine model version	11.1
GL_VENDOR	Intel Inc.
GL_RENDERER	Intel Iris OpenGL Engine
GL_VERSION	4.1 INTEL-10.24.38
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
Log Messages
[84062:775:0226/170743.183610:ERROR:gles2_cmd_decoder.cc(17363)] : [.DisplayCompositor-0x7fb98863f600]GL ERROR :GL_INVALID_OPERATION : glCreateAndConsumeTextureCHROMIUM: invalid mailbox name
[84062:775:0226/170743.183739:ERROR:gles2_cmd_decoder.cc(11899)] : [.DisplayCompositor-0x7fb98863f600]GL ERROR :GL_INVALID_VALUE : glScheduleCALayerCHROMIUM: unsupported texture format
[84062:775:0226/170743.183761:ERROR:gles2_cmd_decoder.cc(17363)] : [.DisplayCompositor-0x7fb98863f600]GL ERROR :GL_INVALID_OPERATION : glCreateAndConsumeTextureCHROMIUM: invalid mailbox name
[84062:775:0226/170743.183774:ERROR:gles2_cmd_decoder.cc(11899)] : [.DisplayCompositor-0x7fb98863f600]GL ERROR :GL_INVALID_VALUE : glScheduleCALayerCHROMIUM: unsupported texture format

Link to related Firefox bug (same problem): https://bugzilla.mozilla.org/show_bug.cgi?id=1342666

Thanks. It looks like this is probably a bug isolated to Intel GPUs. Can try to confirm tomorrow on some of the machines in house.

It's reproducible on a MacBook Air with Intel HD 5000. Not the 6000, for some reason. Also reproducible on a dual-GPU MacBook Pro from last year with dual Iris Pro and AMD Radeon R9 M370X, but only if gfxCardStatus is used to force the machine onto the integrated GPU.

about:gpu follows. Attached are samples from the GPU process (I think -- haven't symbolized it yet) and the output of force quitting the browser process on the Intel HD 5000.

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
swizzle_rgba_for_async_readpixels
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
Async Readpixels with GL_BGRA format is broken on Haswell chipset on Macs: 265115
Applied Workarounds: swizzle_rgba_for_async_readpixels
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	2/27/2017, 12:07:19 PM
Chrome version	Chrome/57.0.2970.0
Operating system	Mac OS X 10.12.4
Software rendering list version	12.07
Driver bug list version	9.26
ANGLE commit id	38ef7c8c97d1
2D graphics backend	Skia/57 34194690328b612557f06ecec99b232d41888d4b
Command Line Args	Chrome Canary.app/Contents/MacOS/Google Chrome Canary --flag-switches-begin --flag-switches-end
Driver Information
Initialization time	37
In-process GPU	false
Passthrough Command Decoder	false
Sandboxed	true
GPU0	VENDOR = 0x8086, DEVICE= 0x0a26 *ACTIVE*
Optimus	false
AMD switchable	false
Driver vendor	
Driver version	10.24.38
Driver date	
Pixel shader version	4.10
Vertex shader version	4.10
Max. MSAA samples	8
Machine model name	MacBookAir
Machine model version	6.2
GL_VENDOR	Intel Inc.
GL_RENDERER	Intel HD Graphics 5000 OpenGL Engine
GL_VERSION	4.1 INTEL-10.24.38
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

Deleted: browser-process-force-quit.txt 857 KB

browser-process-force-quit.txt 857 KB View Download

Deleted: gpu-process-sample.txt 79.2 KB

gpu-process-sample.txt 79.2 KB View Download

rsesek@: having difficulty symbolizing the stack traces produced after force quitting the browser process. See attached, which comes from the current Canary. Do you have any advice? Can I massage the output somehow to get crsym to handle it?

Deleted: browser-force-quit.txt 856 KB

browser-force-quit.txt 856 KB View Download

Argh, yes this format is tricky because it contains samples from all processes. Copy from lines 1 to 672 (end of the first process report) and it should work.

Thanks rsesek@, that's awesome. I tried that with an earlier dump but must have made a mistake. Here's the symbolized stack trace. The browser process is waiting on the GPU process, probably because the latter is wedged in the graphics driver.

Deleted: browser-force-quit-symbolized.txt 68.4 KB

browser-force-quit-symbolized.txt 68.4 KB View Download

For completeness, here's the symbolized "sample" output from the GPU process from above, which was from an older Chromium version. The GPU process doesn't appear wedged at this point. I'm not sure whether our watchdog might have killed and restarted it, and that the sample is from the restarted process.

Deleted: gpu-process-sample-symbolized.txt 82.8 KB

gpu-process-sample-symbolized.txt 82.8 KB View Download

Some more digging from my side (copy-paste from an email to kbr):

attached is a annotated GL call capture of the initialisation and
first 2 frames I could extract from the iOS version (the code runs
fine on native iOS btw). This is the minimal code which causes these
problems on OSX, basically just clearing, no shaders or draw calls).

The problem revolves around MSAA MRT offscreen render targets which
are resolved into color textures, and after that a clear operation on
the default frame buffer must happen. I was first suspecting the MSAA
resolve via glBlitFramebuffer calls, but the actual culprit seems to
be:

::glClearBufferfv(GL_COLOR, i, clearValues);

...which is called on the offscreen MSAA render buffers, if those
buffers are not MSAA the problem seems to disappear. Removing those
three calls to glClearBufferv() fixes the problem.

Deleted: multiplerendertarget_log.txt 21.8 KB

multiplerendertarget_log.txt 21.8 KB View Download

Thanks floooh@ for the call trace. Working on translating it to JavaScript and trying to reproduce the problem on the target hardware with it. I've gotten to the clearBuffer[fi]v calls already, but they themselves don't seem to provoke the problem. Maybe it's caused by rendering multiple frames. Will post another update tomorrow.

I think the problem was only triggered when the default framebuffer is also cleared, which happens after after clearing and resolving the MSAA offscreen render targets. The actual hangs are then happening at the end when the frame is presented (in the OSX native code version this calls glfwSwapBuffers)

I will also continue investigating this in parallel in the evening.

Good luck!

I was able to reproduce the problem in JS, see attached file (there's an index.html and a commented JS in it, just open the index.html in Chrome on the right hw/sw config and the whole window system should freeze for 10..20 seconds (after a while it is possible to kill Chrome/FF).

Deleted: issue696187_repro.zip 2.1 KB

issue696187_repro.zip 2.1 KB Download

floooh@: thank you for the reduced test case! I didn't have time yesterday to continue work on it. This is perfect. I confirm that it causes the hang you mention on 10.12.4 beta 3, and works OK on 10.12.4 beta 4. Will work on getting it in a form of a pass/fail test for the WebGL conformance suite and be back in touch.

The conformance test for this bug was integrated in https://github.com/KhronosGroup/WebGL/pull/2362 and rolled in to Chrome's automated test suite in https://codereview.chromium.org/2806313004/ .

Apple indicated in Radar 30727388 that the bug on their side was already fixed before 10.12.4 shipped.

I hope that they'll run the WebGL 2.0 conformance tests in-house before shipping driver updates so that regressions like this one will be caught.

Since there were no associated code changes in Chromium for this fix, closing as WontFix. Thank you for reporting this bug and for the great test case.