IIRC TraceOutputter just creates a thread to generate a "fake" thread ID for tracing purposes. Since it's intended to represent the GPU timeline, a single one of these is really the best. But agreed, TraceOutputter doesn't account for multiple GL threads tracing at the same time, we need a bunch more logic. The simplest is absolutely one TraceOutputter per thread - it'll look like separate GPU timelines, but they should hopefully not overlap.

Then best option would be to make it similar to MailboxManager, have GpuChannelManager and InProcessCommandBuffer::Service own TraceOutputter and plumbing it through to where it's needed. Assuming TraceOutputter doesn't need to outlive these objects?

I don't think it needs to outlive those. I believe it's only called while decoding commands. So having one TraceOutputter for each of GpuChannelManager and InProcessCommandBuffer::Service seems like it would work.

piman@chromium.org: If nobody is working on this, can I take this one up?

Sure! I don't think anyone is working on this at the moment.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/07f05c09aec5a9a8299bfcb2dd0c97746e73e1f3

commit 07f05c09aec5a9a8299bfcb2dd0c97746e73e1f3
Author: Chandan Padhi <c.padhi@samsung.com>
Date: Tue Sep 19 02:52:53 2017

Create one gpu:gles2::TraceOutputter per thread

Currently, TraceOutputter is a global instance created in a
thread-unsafe way. This is a problem on android webview which
runs gpu service on two different threads in the same process.

With this CL, one TraceOutputter is created per thread for each
of GpuChannelManager and InProcessCommandBuffer::Service.
It also makes TraceOutputter non-ref-counted.

Bug:  752528 
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: I32758531c7ed7fbaf5dd24b3349c5e24c149b7cc
Reviewed-on: https://chromium-review.googlesource.com/667296
Reviewed-by: Chris Watkins <watk@chromium.org>
Reviewed-by: Antoine Labour <piman@chromium.org>
Commit-Queue: Chandan Padhi <c.padhi@samsung.com>
Cr-Commit-Position: refs/heads/master@{#502766}
[modify] https://crrev.com/07f05c09aec5a9a8299bfcb2dd0c97746e73e1f3/gpu/command_buffer/service/context_group_unittest.cc
[modify] https://crrev.com/07f05c09aec5a9a8299bfcb2dd0c97746e73e1f3/gpu/command_buffer/service/framebuffer_manager_unittest.cc
[modify] https://crrev.com/07f05c09aec5a9a8299bfcb2dd0c97746e73e1f3/gpu/command_buffer/service/gl_context_virtual_unittest.cc
[modify] https://crrev.com/07f05c09aec5a9a8299bfcb2dd0c97746e73e1f3/gpu/command_buffer/service/gles2_cmd_decoder.cc
[modify] https://crrev.com/07f05c09aec5a9a8299bfcb2dd0c97746e73e1f3/gpu/command_buffer/service/gles2_cmd_decoder.h
[modify] https://crrev.com/07f05c09aec5a9a8299bfcb2dd0c97746e73e1f3/gpu/command_buffer/service/gles2_cmd_decoder_mock.cc
[modify] https://crrev.com/07f05c09aec5a9a8299bfcb2dd0c97746e73e1f3/gpu/command_buffer/service/gles2_cmd_decoder_mock.h
[modify] https://crrev.com/07f05c09aec5a9a8299bfcb2dd0c97746e73e1f3/gpu/command_buffer/service/gles2_cmd_decoder_passthrough.cc
[modify] https://crrev.com/07f05c09aec5a9a8299bfcb2dd0c97746e73e1f3/gpu/command_buffer/service/gles2_cmd_decoder_passthrough.h
[modify] https://crrev.com/07f05c09aec5a9a8299bfcb2dd0c97746e73e1f3/gpu/command_buffer/service/gles2_cmd_decoder_unittest_base.cc
[modify] https://crrev.com/07f05c09aec5a9a8299bfcb2dd0c97746e73e1f3/gpu/command_buffer/service/gles2_cmd_decoder_unittest_base.h
[modify] https://crrev.com/07f05c09aec5a9a8299bfcb2dd0c97746e73e1f3/gpu/command_buffer/service/gpu_tracer.cc
[modify] https://crrev.com/07f05c09aec5a9a8299bfcb2dd0c97746e73e1f3/gpu/command_buffer/service/gpu_tracer.h
[modify] https://crrev.com/07f05c09aec5a9a8299bfcb2dd0c97746e73e1f3/gpu/command_buffer/service/gpu_tracer_unittest.cc
[modify] https://crrev.com/07f05c09aec5a9a8299bfcb2dd0c97746e73e1f3/gpu/command_buffer/service/query_manager_unittest.cc
[modify] https://crrev.com/07f05c09aec5a9a8299bfcb2dd0c97746e73e1f3/gpu/command_buffer/service/service_discardable_manager_unittest.cc
[modify] https://crrev.com/07f05c09aec5a9a8299bfcb2dd0c97746e73e1f3/gpu/command_buffer/service/texture_manager_unittest.cc
[modify] https://crrev.com/07f05c09aec5a9a8299bfcb2dd0c97746e73e1f3/gpu/command_buffer/tests/fuzzer_main.cc
[modify] https://crrev.com/07f05c09aec5a9a8299bfcb2dd0c97746e73e1f3/gpu/command_buffer/tests/gl_manager.cc
[modify] https://crrev.com/07f05c09aec5a9a8299bfcb2dd0c97746e73e1f3/gpu/command_buffer/tests/gl_manager.h
[modify] https://crrev.com/07f05c09aec5a9a8299bfcb2dd0c97746e73e1f3/gpu/gles2_conform_support/egl/context.cc
[modify] https://crrev.com/07f05c09aec5a9a8299bfcb2dd0c97746e73e1f3/gpu/gles2_conform_support/egl/context.h
[modify] https://crrev.com/07f05c09aec5a9a8299bfcb2dd0c97746e73e1f3/gpu/ipc/in_process_command_buffer.cc
[modify] https://crrev.com/07f05c09aec5a9a8299bfcb2dd0c97746e73e1f3/gpu/ipc/in_process_command_buffer.h
[modify] https://crrev.com/07f05c09aec5a9a8299bfcb2dd0c97746e73e1f3/gpu/ipc/service/gpu_channel_manager.cc
[modify] https://crrev.com/07f05c09aec5a9a8299bfcb2dd0c97746e73e1f3/gpu/ipc/service/gpu_channel_manager.h
[modify] https://crrev.com/07f05c09aec5a9a8299bfcb2dd0c97746e73e1f3/gpu/ipc/service/gpu_command_buffer_stub.cc
[modify] https://crrev.com/07f05c09aec5a9a8299bfcb2dd0c97746e73e1f3/media/gpu/android/android_video_decode_accelerator_unittest.cc

piman@chromium.org: Is there anything else to do w.r.t this bug? Shall we close this one?

Yes, I think we can close this. Thanks!