gpu: Allow waiting on sync tokens without sync token client.

gpu: Allow waiting on sync tokens without sync token client.

This CL adds SyncPointManager::Wait which takes a sync token and an
order number. This enables the upcoming gpu scheduler to wait on sync
tokens that a flush depends on without executing the flush. Such waits
are safe from deadlock if the order number of the message is used.

Misc cleanup in this CL:
1. Hide SyncPointClientState and instead use sync tokens for waiting.
2. The wait methods do not run the callback if the wait is invalid.
3. The cmd decoder wait sync token callback returns true on valid waits.
4. Added few more cmd buffer message types that don't need MakeCurrent.
5. Changed the sync point tests to use sync tokens everywhere.

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

Review-Url: https://codereview.chromium.org/2722883002
Cr-Commit-Position: refs/heads/master@{#454361}
Committed: https://chromium.googlesource.com/chromium/src/+/9147a531d69c35b928d4dc1a11684bb5baf1d6f9

[sunnyps, 2017-02-28 20:08:07]

Description was changed from

==========
gpu: Allow waiting on sync tokens without sync token client.

This CL adds SyncPointManager::Wait which takes a sync token and an
order number. This enables the upcoming gpu scheduler to wait on sync
tokens that a flush depends on without executing the flush. Such waits
are safe from deadlock if the order number of the message is used.

Misc cleanup in this CL:
1. Hide SyncPointClientState and instead sync tokens for waiting.
2. The wait methods do not run the callback if the wait is invalid.
3. The cmd decoder wait sync token callback returns true on valid waits.
4. Added few more cmd buffer message types that don't need MakeCurrent.
5. Changed the sync point tests to use sync tokens everywhere.

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

to

==========
gpu: Allow waiting on sync tokens without sync token client.

This CL adds SyncPointManager::Wait which takes a sync token and an
order number. This enables the upcoming gpu scheduler to wait on sync
tokens that a flush depends on without executing the flush. Such waits
are safe from deadlock if the order number of the message is used.

Misc cleanup in this CL:
1. Hide SyncPointClientState and instead sync tokens for waiting.
2. The wait methods do not run the callback if the wait is invalid.
3. The cmd decoder wait sync token callback returns true on valid waits.
4. Added few more cmd buffer message types that don't need MakeCurrent.
5. Changed the sync point tests to use sync tokens everywhere.

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, 2017-02-28 20:08:14]

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

[commit-bot: I haz the power, 2017-02-28 20:08:40]

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

[sunnyps, 2017-02-28 20:09:13]

PTAL

Let me know if you want me to split out any part of this CL.

[sunnyps, 2017-02-28 20:09:40]

Description was changed from

==========
gpu: Allow waiting on sync tokens without sync token client.

This CL adds SyncPointManager::Wait which takes a sync token and an
order number. This enables the upcoming gpu scheduler to wait on sync
tokens that a flush depends on without executing the flush. Such waits
are safe from deadlock if the order number of the message is used.

Misc cleanup in this CL:
1. Hide SyncPointClientState and instead sync tokens for waiting.
2. The wait methods do not run the callback if the wait is invalid.
3. The cmd decoder wait sync token callback returns true on valid waits.
4. Added few more cmd buffer message types that don't need MakeCurrent.
5. Changed the sync point tests to use sync tokens everywhere.

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: Allow waiting on sync tokens without sync token client.

This CL adds SyncPointManager::Wait which takes a sync token and an
order number. This enables the upcoming gpu scheduler to wait on sync
tokens that a flush depends on without executing the flush. Such waits
are safe from deadlock if the order number of the message is used.

Misc cleanup in this CL:
1. Hide SyncPointClientState and instead use sync tokens for waiting.
2. The wait methods do not run the callback if the wait is invalid.
3. The cmd decoder wait sync token callback returns true on valid waits.
4. Added few more cmd buffer message types that don't need MakeCurrent.
5. Changed the sync point tests to use sync tokens everywhere.

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
==========

[commit-bot: I haz the power, 2017-02-28 20:35:38]

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

[commit-bot: I haz the power, 2017-02-28 20:35:39]

Dry run: Try jobs failed on following builders:
  android_n5x_swarming_rel on master.tryserver.chromium.android (JOB_FAILED,
https://build.chromium.org/p/tryserver.chromium.android/builders/android_n5x_...)

[sunnyps, 2017-02-28 20:42:30]

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

[commit-bot: I haz the power, 2017-02-28 20:42:54]

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, 2017-02-28 21:28:50]

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

[commit-bot: I haz the power, 2017-02-28 21:28:51]

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, 2017-02-28 23:31:44]

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

[commit-bot: I haz the power, 2017-02-28 23:33:26]

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, 2017-03-01 00:12:35]

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

[commit-bot: I haz the power, 2017-03-01 00:12:36]

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, 2017-03-01 00:20:20]

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

[commit-bot: I haz the power, 2017-03-01 00:20:58]

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, 2017-03-01 01:16:48]

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

[commit-bot: I haz the power, 2017-03-01 01:16:49]

Dry run: Try jobs failed on following builders:
  android_n5x_swarming_rel on master.tryserver.chromium.android (JOB_FAILED,
https://build.chromium.org/p/tryserver.chromium.android/builders/android_n5x_...)

[sunnyps, 2017-03-02 00:11:07]

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

[commit-bot: I haz the power, 2017-03-02 00:12:00]

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

[jbauman, 2017-03-02 00:31:48]

https://codereview.chromium.org/2722883002/diff/60001/gpu/command_buffer/serv...
File gpu/command_buffer/service/sync_point_manager.h (right):

https://codereview.chromium.org/2722883002/diff/60001/gpu/command_buffer/serv...
gpu/command_buffer/service/sync_point_manager.h:144:
SyncPointClientState(scoped_refptr<SyncPointOrderData> order_data);
explicit

https://codereview.chromium.org/2722883002/diff/80001/gpu/ipc/in_process_comm...
File gpu/ipc/in_process_command_buffer.cc (right):

https://codereview.chromium.org/2722883002/diff/80001/gpu/ipc/in_process_comm...
gpu/ipc/in_process_command_buffer.cc:979: false,
Do you know why it's necessary to change this?

[sunnyps, 2017-03-02 00:34:29]

https://codereview.chromium.org/2722883002/diff/80001/gpu/ipc/in_process_comm...
File gpu/ipc/in_process_command_buffer.cc (right):

https://codereview.chromium.org/2722883002/diff/80001/gpu/ipc/in_process_comm...
gpu/ipc/in_process_command_buffer.cc:979: false,
On 2017/03/02 00:31:48, jbauman wrote:
> Do you know why it's necessary to change this?

I changed SignalSyncTokenOnGpuThread to use SyncPointClient::Wait instead of
WaitOutOfOrder. SyncPointClient::Wait has a DCHECK for IsProcessingOrderNumber.

[sunnyps, 2017-03-02 01:24:48]

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

[commit-bot: I haz the power, 2017-03-02 01:25:01]

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 01:26:45]

ptal

https://codereview.chromium.org/2722883002/diff/60001/gpu/command_buffer/serv...
File gpu/command_buffer/service/sync_point_manager.h (right):

https://codereview.chromium.org/2722883002/diff/60001/gpu/command_buffer/serv...
gpu/command_buffer/service/sync_point_manager.h:144:
SyncPointClientState(scoped_refptr<SyncPointOrderData> order_data);
On 2017/03/02 00:31:48, jbauman wrote:
> explicit

Done.

[jbauman, 2017-03-02 02:03:55]

lgtm

[sunnyps, 2017-03-02 02:41:09]

sunnyps@chromium.org changed reviewers:
+ boliu@chromium.org, fsamuel@chromium.org

[sunnyps, 2017-03-02 02:41:09]

boliu: please rs android_webview change

fsamuel: please rs services/ui/gpu change

allow_threaded_wait was unused in SyncPointManager

[Fady Samuel, 2017-03-02 02:43:10]

lgtm

[boliu, 2017-03-02 02:43:49]

rs lgtm

[sunnyps, 2017-03-02 02:44:10]

Thanks, everyone!

[sunnyps, 2017-03-02 02:44:20]

The CQ bit was unchecked by sunnyps@chromium.org

[sunnyps, 2017-03-02 02:44:21]

The CQ bit was checked by sunnyps@chromium.org

[commit-bot: I haz the power, 2017-03-02 02:45:28]

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 05:04:23]

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

[commit-bot: I haz the power, 2017-03-02 05:04:24]

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 19:26:31]

The CQ bit was checked by sunnyps@chromium.org

[commit-bot: I haz the power, 2017-03-02 19:27:11]

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 20:26:41]

CQ is committing da patch.

Bot data: {"patchset_id": 100001, "attempt_start_ts": 1488482791694030,
"parent_rev": "67e99111e67ac5aeae011a14154e054160bf98c4", "commit_rev":
"9147a531d69c35b928d4dc1a11684bb5baf1d6f9"}

[commit-bot: I haz the power, 2017-03-02 20:27:26]

Description was changed from

==========
gpu: Allow waiting on sync tokens without sync token client.

This CL adds SyncPointManager::Wait which takes a sync token and an
order number. This enables the upcoming gpu scheduler to wait on sync
tokens that a flush depends on without executing the flush. Such waits
are safe from deadlock if the order number of the message is used.

Misc cleanup in this CL:
1. Hide SyncPointClientState and instead use sync tokens for waiting.
2. The wait methods do not run the callback if the wait is invalid.
3. The cmd decoder wait sync token callback returns true on valid waits.
4. Added few more cmd buffer message types that don't need MakeCurrent.
5. Changed the sync point tests to use sync tokens everywhere.

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: Allow waiting on sync tokens without sync token client.

This CL adds SyncPointManager::Wait which takes a sync token and an
order number. This enables the upcoming gpu scheduler to wait on sync
tokens that a flush depends on without executing the flush. Such waits
are safe from deadlock if the order number of the message is used.

Misc cleanup in this CL:
1. Hide SyncPointClientState and instead use sync tokens for waiting.
2. The wait methods do not run the callback if the wait is invalid.
3. The cmd decoder wait sync token callback returns true on valid waits.
4. Added few more cmd buffer message types that don't need MakeCurrent.
5. Changed the sync point tests to use sync tokens everywhere.

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

Review-Url: https://codereview.chromium.org/2722883002
Cr-Commit-Position: refs/heads/master@{#454361}
Committed:
https://chromium.googlesource.com/chromium/src/+/9147a531d69c35b928d4dc1a1168...
==========

[commit-bot: I haz the power, 2017-03-02 20:27:31]

Committed patchset #6 (id:100001) as
https://chromium.googlesource.com/chromium/src/+/9147a531d69c35b928d4dc1a1168...