gpu: Add sync token dependencies to flush metadata.

gpu: Add sync token dependencies to flush metadata.

This CL adds a list of sync tokens, for which waits were inserted in the
command buffer, to the flush message. This information will be used by
the gpu scheduler to deschedule a command buffer until all sync tokens
of its next flush are released.

Also includes some minor cleanup.

R=jbauman@chromium.org
BUG=514813
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel;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

Review-Url: https://codereview.chromium.org/2727573003
Cr-Commit-Position: refs/heads/master@{#456961}
Committed: https://chromium.googlesource.com/chromium/src/+/7499629e5db5df76176f12b94dd4dffa69fbe668

[sunnyps, 2017-03-01 23:40:39]

Description was changed from

==========
gpu: Add sync token dependencies to flush metadata.

This CL adds a list of sync tokens, for which waits were inserted in the
command buffer, to the flush message. Also includes some minor cleanup.

R=jbauman@chromium.org
BUG=514813
==========

to

==========
gpu: Add sync token dependencies to flush metadata.

This CL adds a list of sync tokens, for which waits were inserted in the
command buffer, to the flush message. Also includes some minor cleanup.

R=jbauman@chromium.org
BUG=514813
CQ_INCLUDE_TRYBOTS=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
==========

[sunnyps, 2017-03-01 23:40:40]

Description was changed from

==========
gpu: Add sync token dependencies to flush metadata.

This CL adds a list of sync tokens, for which waits were inserted in the
command buffer, to the flush message. Also includes some minor cleanup.

R=jbauman@chromium.org
BUG=514813
CQ_INCLUDE_TRYBOTS=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
==========

to

==========
gpu: Add sync token dependencies to flush metadata.

This CL adds a list of sync tokens, for which waits were inserted in the
command buffer, to the flush message. Also includes some minor cleanup.

R=jbauman@chromium.org
BUG=514813
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_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
==========

[sunnyps, 2017-03-01 23:42:08]

Description was changed from

==========
gpu: Add sync token dependencies to flush metadata.

This CL adds a list of sync tokens, for which waits were inserted in the
command buffer, to the flush message. Also includes some minor cleanup.

R=jbauman@chromium.org
BUG=514813
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_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
==========

to

==========
gpu: Add sync token dependencies to flush metadata.

This CL adds a list of sync tokens, for which waits were inserted in the
command buffer, to the flush message. This information will be used by
the gpu scheduler to deschedule a command buffer until all sync tokens
of its next flush are released.

Also includes some minor cleanup.

R=jbauman@chromium.org
BUG=514813
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_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
==========

[sunnyps, 2017-03-01 23:42:12]

The CQ bit was checked by sunnyps@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2017-03-01 23:42:59]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[sunnyps, 2017-03-02 00:00:23]

The CQ bit was checked by sunnyps@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2017-03-02 00:00:31]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2017-03-02 00:35:15]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2017-03-02 00:35:16]

Dry run: Try jobs failed on following builders:
  cast_shell_linux on master.tryserver.chromium.linux (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/cast_shell_linu...)

[sunnyps, 2017-03-02 01:08:47]

The CQ bit was checked by sunnyps@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2017-03-02 01:09:28]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2017-03-02 02:46:33]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2017-03-02 02:46:33]

Dry run: Try jobs failed on following builders:
  linux_chromium_rel_ng on master.tryserver.chromium.linux (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/linux_chromium_...)

[sunnyps, 2017-03-02 03:09:05]

The CQ bit was checked by sunnyps@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2017-03-02 03:10:10]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2017-03-02 07:04:40]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2017-03-02 07:04:41]

Dry run: Try jobs failed on following builders:
  linux_android_rel_ng on master.tryserver.chromium.android (JOB_FAILED,
https://build.chromium.org/p/tryserver.chromium.android/builders/linux_androi...)

[sunnyps, 2017-03-02 19:27:05]

ptal

[jbauman, 2017-03-09 02:00:09]

https://codereview.chromium.org/2727573003/diff/60001/gpu/command_buffer/clie...
File gpu/command_buffer/client/gles2_implementation.cc (right):

https://codereview.chromium.org/2727573003/diff/60001/gpu/command_buffer/clie...
gpu/command_buffer/client/gles2_implementation.cc:6166:
gpu_control_->WaitSyncToken(sync_token);
I think this should actually go after the call to
helper_->WaitSyncTokenCHROMIUM. Otherwise if the helper needs to flush the
command buffer before putting in the command then the wait would go on the wrong
flush.

https://codereview.chromium.org/2727573003/diff/60001/gpu/command_buffer/clie...
File gpu/command_buffer/client/gpu_control.h (right):

https://codereview.chromium.org/2727573003/diff/60001/gpu/command_buffer/clie...
gpu/command_buffer/client/gpu_control.h:105: // Returns true if the wait is
valid.
Also comment that the wait command-buffer command should be used as well.

[sunnyps, 2017-03-10 03:19:38]

Description was changed from

==========
gpu: Add sync token dependencies to flush metadata.

This CL adds a list of sync tokens, for which waits were inserted in the
command buffer, to the flush message. This information will be used by
the gpu scheduler to deschedule a command buffer until all sync tokens
of its next flush are released.

Also includes some minor cleanup.

R=jbauman@chromium.org
BUG=514813
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_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
==========

to

==========
gpu: Add sync token dependencies to flush metadata.

This CL adds a list of sync tokens, for which waits were inserted in the
command buffer, to the flush message. This information will be used by
the gpu scheduler to deschedule a command buffer until all sync tokens
of its next flush are released.

Also includes some minor cleanup.

R=jbauman@chromium.org
BUG=514813
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel;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
==========

[sunnyps, 2017-03-10 03:25:17]

The CQ bit was checked by sunnyps@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2017-03-10 03:25:32]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[sunnyps, 2017-03-10 03:26:20]

Addressed comments. Renamed sync_token_dependencies to sync_token_fences.

PTAL. Thanks!

https://codereview.chromium.org/2727573003/diff/60001/gpu/command_buffer/clie...
File gpu/command_buffer/client/gles2_implementation.cc (right):

https://codereview.chromium.org/2727573003/diff/60001/gpu/command_buffer/clie...
gpu/command_buffer/client/gles2_implementation.cc:6166:
gpu_control_->WaitSyncToken(sync_token);
On 2017/03/09 02:00:09, jbauman wrote:
> I think this should actually go after the call to
> helper_->WaitSyncTokenCHROMIUM. Otherwise if the helper needs to flush the
> command buffer before putting in the command then the wait would go on the
wrong
> flush.

Nice catch!

https://codereview.chromium.org/2727573003/diff/60001/gpu/command_buffer/clie...
File gpu/command_buffer/client/gpu_control.h (right):

https://codereview.chromium.org/2727573003/diff/60001/gpu/command_buffer/clie...
gpu/command_buffer/client/gpu_control.h:105: // Returns true if the wait is
valid.
On 2017/03/09 02:00:09, jbauman wrote:
> Also comment that the wait command-buffer command should be used as well.

Done.

[commit-bot: I haz the power, 2017-03-10 04:49:17]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2017-03-10 04:49:18]

Dry run: This issue passed the CQ dry run.

[jbauman, 2017-03-10 22:08:12]

lgtm

[sunnyps, 2017-03-13 22:28:34]

sunnyps@chromium.org changed reviewers:
+ dcheng@chromium.org, piman@chromium.org

[sunnyps, 2017-03-13 22:28:35]

piman@ ptal
dcheng@ ptal at gpu_messages.h

[piman, 2017-03-13 23:17:39]

Overall LGTM, there's a few nits.

https://codereview.chromium.org/2727573003/diff/100001/gpu/command_buffer/cli...
File gpu/command_buffer/client/gpu_control.h (right):

https://codereview.chromium.org/2727573003/diff/100001/gpu/command_buffer/cli...
gpu/command_buffer/client/gpu_control.h:105: // Returns true if the wait is
valid. This is used in addition to
nit: it doesn't return anything.

https://codereview.chromium.org/2727573003/diff/100001/gpu/command_buffer/cli...
gpu/command_buffer/client/gpu_control.h:107: // texture updates from other
threads when used with MailboxManagerSync.
I don't think this one has anything to do with pulling texture updates - it's
about giving a hint that the next Flush will wait on this...

https://codereview.chromium.org/2727573003/diff/100001/gpu/ipc/client/command...
File gpu/ipc/client/command_buffer_proxy_impl.cc (right):

https://codereview.chromium.org/2727573003/diff/100001/gpu/ipc/client/command...
gpu/ipc/client/command_buffer_proxy_impl.cc:667:
verified_sync_token.SetVerifyFlush();
nit: I would suggest moving this to GLES2Implementation for consistency with
e.g. SignalSyncToken. It could even coalesce into a "bool
GetVerifiedSyncTokenForIPC(const SyncToken& sync_token, SyncToken* verified)"
helper that checks if it's verified, or can be waited without verification, and
sets the verified bit in the latter case.

https://codereview.chromium.org/2727573003/diff/100001/gpu/ipc/client/gpu_cha...
File gpu/ipc/client/gpu_channel_host.cc (right):

https://codereview.chromium.org/2727573003/diff/100001/gpu/ipc/client/gpu_cha...
gpu/ipc/client/gpu_channel_host.cc:187: flush_info->latency_info,
flush_info->sync_token_fences));
nit: you can use std::move to avoid a copy (though you want to assign
flush_info->sync_token_fences to a new empty vector instead of .clear() below)

[dcheng, 2017-03-14 06:53:03]

ipc lgtm with some random comments

https://codereview.chromium.org/2727573003/diff/100001/gpu/gles2_conform_supp...
File gpu/gles2_conform_support/egl/context.cc (right):

https://codereview.chromium.org/2727573003/diff/100001/gpu/gles2_conform_supp...
gpu/gles2_conform_support/egl/context.cc:226: void Context::WaitSyncToken(const
gpu::SyncToken& sync_token) {}
Is this a NOTIMPLEMENTED kind of thing?

https://codereview.chromium.org/2727573003/diff/100001/gpu/ipc/client/gpu_cha...
File gpu/ipc/client/gpu_channel_host.cc (right):

https://codereview.chromium.org/2727573003/diff/100001/gpu/ipc/client/gpu_cha...
gpu/ipc/client/gpu_channel_host.cc:187: flush_info->latency_info,
flush_info->sync_token_fences));
On 2017/03/13 23:17:39, piman wrote:
> nit: you can use std::move to avoid a copy (though you want to assign
> flush_info->sync_token_fences to a new empty vector instead of .clear() below)

I don't think this copies, it serializes directly into the IPC pickle (though it
may not be sent immediately)

https://codereview.chromium.org/2727573003/diff/100001/ppapi/proxy/ppapi_comm...
File ppapi/proxy/ppapi_command_buffer_proxy.cc (right):

https://codereview.chromium.org/2727573003/diff/100001/ppapi/proxy/ppapi_comm...
ppapi/proxy/ppapi_command_buffer_proxy.cc:217: void
PpapiCommandBufferProxy::WaitSyncToken(const gpu::SyncToken& sync_token) {}
Is this a NOTIMPLEMENTED() kind of thing as well?

[sunnyps, 2017-03-15 00:03:08]

PTAL

https://codereview.chromium.org/2727573003/diff/100001/gpu/command_buffer/cli...
File gpu/command_buffer/client/gpu_control.h (right):

https://codereview.chromium.org/2727573003/diff/100001/gpu/command_buffer/cli...
gpu/command_buffer/client/gpu_control.h:105: // Returns true if the wait is
valid. This is used in addition to
On 2017/03/13 23:17:39, piman wrote:
> nit: it doesn't return anything.

Done.

https://codereview.chromium.org/2727573003/diff/100001/gpu/command_buffer/cli...
gpu/command_buffer/client/gpu_control.h:107: // texture updates from other
threads when used with MailboxManagerSync.
On 2017/03/13 23:17:39, piman wrote:
> I don't think this one has anything to do with pulling texture updates - it's
> about giving a hint that the next Flush will wait on this...

The comment was about why the WaitSyncToken command in the command buffer is
necessary. I've rewritten this and hopefully it's clear now.

https://codereview.chromium.org/2727573003/diff/100001/gpu/gles2_conform_supp...
File gpu/gles2_conform_support/egl/context.cc (right):

https://codereview.chromium.org/2727573003/diff/100001/gpu/gles2_conform_supp...
gpu/gles2_conform_support/egl/context.cc:226: void Context::WaitSyncToken(const
gpu::SyncToken& sync_token) {}
On 2017/03/14 06:53:03, dcheng wrote:
> Is this a NOTIMPLEMENTED kind of thing?

No, GLES2Implementation might call this. I don't know if WaitSyncToken is ever
called in this configuration, but it's OK to have a NOP implementation for this
method.

Maybe this method should be renamed? piman@ WDYT?

https://codereview.chromium.org/2727573003/diff/100001/gpu/ipc/client/gpu_cha...
File gpu/ipc/client/gpu_channel_host.cc (right):

https://codereview.chromium.org/2727573003/diff/100001/gpu/ipc/client/gpu_cha...
gpu/ipc/client/gpu_channel_host.cc:187: flush_info->latency_info,
flush_info->sync_token_fences));
On 2017/03/14 06:53:03, dcheng wrote:
> On 2017/03/13 23:17:39, piman wrote:
> > nit: you can use std::move to avoid a copy (though you want to assign
> > flush_info->sync_token_fences to a new empty vector instead of .clear()
below)
> 
> I don't think this copies, it serializes directly into the IPC pickle (though
it
> may not be sent immediately)

I'll leave this as it is.

https://codereview.chromium.org/2727573003/diff/100001/ppapi/proxy/ppapi_comm...
File ppapi/proxy/ppapi_command_buffer_proxy.cc (right):

https://codereview.chromium.org/2727573003/diff/100001/ppapi/proxy/ppapi_comm...
ppapi/proxy/ppapi_command_buffer_proxy.cc:217: void
PpapiCommandBufferProxy::WaitSyncToken(const gpu::SyncToken& sync_token) {}
On 2017/03/14 06:53:03, dcheng wrote:
> Is this a NOTIMPLEMENTED() kind of thing as well?

No, pepper can wait on sync tokens and GLES2Implementation will call this method
in that case. We could annotate pepper's flushes with sync tokens like we do for
command buffer proxy but it's optional.

[sunnyps, 2017-03-15 00:15:54]

The CQ bit was checked by sunnyps@chromium.org to run a CQ dry run

[sunnyps, 2017-03-15 00:16:42]

I broke some tests because of last minute changes. Fixed now. PTAL

[commit-bot: I haz the power, 2017-03-15 00:16:53]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[piman, 2017-03-15 00:30:14]

lgtm

https://codereview.chromium.org/2727573003/diff/100001/gpu/gles2_conform_supp...
File gpu/gles2_conform_support/egl/context.cc (right):

https://codereview.chromium.org/2727573003/diff/100001/gpu/gles2_conform_supp...
gpu/gles2_conform_support/egl/context.cc:226: void Context::WaitSyncToken(const
gpu::SyncToken& sync_token) {}
On 2017/03/15 00:03:08, sunnyps wrote:
> On 2017/03/14 06:53:03, dcheng wrote:
> > Is this a NOTIMPLEMENTED kind of thing?
> 
> No, GLES2Implementation might call this. I don't know if WaitSyncToken is ever
> called in this configuration, but it's OK to have a NOP implementation for
this
> method.
> 
> Maybe this method should be renamed? piman@ WDYT?

It could be named something like WaitSyncTokenHint to indicate that it's a hint,
that can be used for scheduling, but doesn't have semantic meaning otherwise, so
it's ok to have it do nothing in some implementations?

https://codereview.chromium.org/2727573003/diff/100001/ppapi/proxy/ppapi_comm...
File ppapi/proxy/ppapi_command_buffer_proxy.cc (right):

https://codereview.chromium.org/2727573003/diff/100001/ppapi/proxy/ppapi_comm...
ppapi/proxy/ppapi_command_buffer_proxy.cc:217: void
PpapiCommandBufferProxy::WaitSyncToken(const gpu::SyncToken& sync_token) {}
On 2017/03/15 00:03:08, sunnyps wrote:
> On 2017/03/14 06:53:03, dcheng wrote:
> > Is this a NOTIMPLEMENTED() kind of thing as well?
> 
> No, pepper can wait on sync tokens and GLES2Implementation will call this
method
> in that case. We could annotate pepper's flushes with sync tokens like we do
for
> command buffer proxy but it's optional.

We could add a TODO to forward this to the renderer, but it's so seldom used
that I think it's ok to punt for now.

https://codereview.chromium.org/2727573003/diff/140001/gpu/command_buffer/cli...
File gpu/command_buffer/client/gpu_control.h (right):

https://codereview.chromium.org/2727573003/diff/140001/gpu/command_buffer/cli...
gpu/command_buffer/client/gpu_control.h:105: // This allows the command buffer
proxy to mark each flush with sync token
nit: maybe say "the next flush" instead of "each flush" to indicate that this
applies to the next flush?

[sunnyps, 2017-03-15 00:56:12]

Thanks!

https://codereview.chromium.org/2727573003/diff/100001/gpu/gles2_conform_supp...
File gpu/gles2_conform_support/egl/context.cc (right):

https://codereview.chromium.org/2727573003/diff/100001/gpu/gles2_conform_supp...
gpu/gles2_conform_support/egl/context.cc:226: void Context::WaitSyncToken(const
gpu::SyncToken& sync_token) {}
On 2017/03/15 00:30:14, piman wrote:
> On 2017/03/15 00:03:08, sunnyps wrote:
> > On 2017/03/14 06:53:03, dcheng wrote:
> > > Is this a NOTIMPLEMENTED kind of thing?
> > 
> > No, GLES2Implementation might call this. I don't know if WaitSyncToken is
ever
> > called in this configuration, but it's OK to have a NOP implementation for
> this
> > method.
> > 
> > Maybe this method should be renamed? piman@ WDYT?
> 
> It could be named something like WaitSyncTokenHint to indicate that it's a
hint,
> that can be used for scheduling, but doesn't have semantic meaning otherwise,
so
> it's ok to have it do nothing in some implementations?

Done.

https://codereview.chromium.org/2727573003/diff/100001/ppapi/proxy/ppapi_comm...
File ppapi/proxy/ppapi_command_buffer_proxy.cc (right):

https://codereview.chromium.org/2727573003/diff/100001/ppapi/proxy/ppapi_comm...
ppapi/proxy/ppapi_command_buffer_proxy.cc:217: void
PpapiCommandBufferProxy::WaitSyncToken(const gpu::SyncToken& sync_token) {}
On 2017/03/15 00:30:14, piman wrote:
> On 2017/03/15 00:03:08, sunnyps wrote:
> > On 2017/03/14 06:53:03, dcheng wrote:
> > > Is this a NOTIMPLEMENTED() kind of thing as well?
> > 
> > No, pepper can wait on sync tokens and GLES2Implementation will call this
> method
> > in that case. We could annotate pepper's flushes with sync tokens like we do
> for
> > command buffer proxy but it's optional.
> 
> We could add a TODO to forward this to the renderer, but it's so seldom used
> that I think it's ok to punt for now.

Done.

https://codereview.chromium.org/2727573003/diff/140001/gpu/command_buffer/cli...
File gpu/command_buffer/client/gpu_control.h (right):

https://codereview.chromium.org/2727573003/diff/140001/gpu/command_buffer/cli...
gpu/command_buffer/client/gpu_control.h:105: // This allows the command buffer
proxy to mark each flush with sync token
On 2017/03/15 00:30:14, piman wrote:
> nit: maybe say "the next flush" instead of "each flush" to indicate that this
> applies to the next flush?

Done.

[sunnyps, 2017-03-15 00:57:41]

The CQ bit was checked by sunnyps@chromium.org

[sunnyps, 2017-03-15 00:57:42]

The patchset sent to the CQ was uploaded after l-g-t-m from
jbauman@chromium.org, dcheng@chromium.org, piman@chromium.org
Link to the patchset: https://codereview.chromium.org/2727573003/#ps160001
(title: "piman's review 2")

[commit-bot: I haz the power, 2017-03-15 00:58:19]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2017-03-15 02:35:52]

CQ is committing da patch.

Bot data: {"patchset_id": 160001, "attempt_start_ts": 1489539461511780,
"parent_rev": "a2ee9a3ffdc1e6ecb3d2563f668e0c25b57ef721", "commit_rev":
"7499629e5db5df76176f12b94dd4dffa69fbe668"}

[commit-bot: I haz the power, 2017-03-15 02:36:36]

Description was changed from

==========
gpu: Add sync token dependencies to flush metadata.

This CL adds a list of sync tokens, for which waits were inserted in the
command buffer, to the flush message. This information will be used by
the gpu scheduler to deschedule a command buffer until all sync tokens
of its next flush are released.

Also includes some minor cleanup.

R=jbauman@chromium.org
BUG=514813
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel;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
==========

to

==========
gpu: Add sync token dependencies to flush metadata.

This CL adds a list of sync tokens, for which waits were inserted in the
command buffer, to the flush message. This information will be used by
the gpu scheduler to deschedule a command buffer until all sync tokens
of its next flush are released.

Also includes some minor cleanup.

R=jbauman@chromium.org
BUG=514813
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel;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

Review-Url: https://codereview.chromium.org/2727573003
Cr-Commit-Position: refs/heads/master@{#456961}
Committed:
https://chromium.googlesource.com/chromium/src/+/7499629e5db5df76176f12b94dd4...
==========

[commit-bot: I haz the power, 2017-03-15 02:36:38]

Committed patchset #9 (id:160001) as
https://chromium.googlesource.com/chromium/src/+/7499629e5db5df76176f12b94dd4...