Chromium Code Reviews
chromiumcodereview-hr@appspot.gserviceaccount.com (chromiumcodereview-hr) | Please choose your nickname with Settings | Help | Chromium Project | Gerrit Changes | Sign out
(32)

Issue 2247173002: gpu: Use GpuMemoryBufferManager from GpuService when possible. (Closed)

Created:
4 years, 4 months ago by sadrul
Modified:
4 years, 2 months ago
Reviewers:
reveman, sky, piman
CC:
chromium-reviews, danakj+watch_chromium.org, darin-cc_chromium.org, jam, jbauman+watch_chromium.org, kalyank, mlamouri+watch-content_chromium.org, piman+watch_chromium.org, rjkroege, sievers+watch_chromium.org
Base URL:
https://chromium.googlesource.com/chromium/src.git@mus-gpu-memory-buffer-manager
Target Ref:
refs/pending/heads/master
Project:
chromium
Visibility:
Public.

Description

gpu: Use GpuMemoryBufferManager from GpuService when possible. ui::GpuService creates a GpuMemoryBufferManager instance that communicates with the gpu process over mojo (MojoGpuMemoryBufferManager). When running chrome as a muc client, use this version of the GpuMemoryBufferManager, instead of the one created in content. Add GpuChannelEstablishFactory::GetGpuMemoryBufferManager() for this. This makes it unnecessary to create the BrowserGpuChannelHostFactory or the BrowserGpuMemoryBufferManager singleton intances when running chrome as a mus client. BUG=637920 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 Committed: https://crrev.com/a78e33dec9fd79c95f94fa865cdd37518d31d245 Cr-Commit-Position: refs/heads/master@{#412316}

Patch Set 1 #

Unified diffs Side-by-side diffs Delta from patch set Stats (+25 lines, -9 lines) Patch
M content/browser/browser_main_loop.cc View 1 chunk +4 lines, -4 lines 0 comments Download
M content/browser/compositor/gpu_process_transport_factory.cc View 4 chunks +4 lines, -4 lines 0 comments Download
M content/browser/gpu/browser_gpu_channel_host_factory.h View 1 chunk +1 line, -0 lines 0 comments Download
M content/browser/gpu/browser_gpu_channel_host_factory.cc View 1 chunk +5 lines, -0 lines 0 comments Download
M content/renderer/render_thread_impl.cc View 1 chunk +5 lines, -1 line 0 comments Download
M gpu/ipc/client/gpu_channel_host.h View 1 chunk +1 line, -0 lines 0 comments Download
M services/ui/public/cpp/gpu_service.h View 1 chunk +1 line, -0 lines 0 comments Download
M services/ui/public/cpp/gpu_service.cc View 1 chunk +4 lines, -0 lines 0 comments Download

Depends on Patchset:

Messages

Total messages: 22 (10 generated)
sadrul
4 years, 4 months ago (2016-08-16 07:20:26 UTC) #7
piman
lgtm
4 years, 4 months ago (2016-08-16 17:09:46 UTC) #8
sadrul
+sky@ for services/ui changes.
4 years, 4 months ago (2016-08-16 17:51:21 UTC) #10
sky
LGTM
4 years, 4 months ago (2016-08-16 19:44:38 UTC) #11
commit-bot: I haz the power
CQ is trying da patch. Follow status at https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.org/2247173002/1
4 years, 4 months ago (2016-08-16 20:05:48 UTC) #13
commit-bot: I haz the power
Committed patchset #1 (id:1)
4 years, 4 months ago (2016-08-16 20:11:37 UTC) #14
commit-bot: I haz the power
Patchset 1 (id:??) landed as https://crrev.com/a78e33dec9fd79c95f94fa865cdd37518d31d245 Cr-Commit-Position: refs/heads/master@{#412316}
4 years, 4 months ago (2016-08-16 20:13:56 UTC) #16
reveman
MojoGpuMemoryBufferManager seem to be lacking native GpuMemoryBuffer support. How does this change not disable native ...
4 years, 2 months ago (2016-10-03 15:26:32 UTC) #18
sadrul
On 2016/10/03 15:26:32, reveman wrote: > MojoGpuMemoryBufferManager seem to be lacking native GpuMemoryBuffer support. > ...
4 years, 2 months ago (2016-10-03 17:28:54 UTC) #19
reveman
On 2016/10/03 at 17:28:54, sadrul wrote: > On 2016/10/03 15:26:32, reveman wrote: > > MojoGpuMemoryBufferManager ...
4 years, 2 months ago (2016-10-03 19:31:59 UTC) #20
sadrul
On 2016/10/03 19:31:59, reveman wrote: > On 2016/10/03 at 17:28:54, sadrul wrote: > > On ...
4 years, 2 months ago (2016-10-03 19:48:32 UTC) #21
reveman
4 years, 2 months ago (2016-10-03 19:52:46 UTC) #22
Message was sent while issue was closed.
On 2016/10/03 at 19:48:32, sadrul wrote:
> On 2016/10/03 19:31:59, reveman wrote:
> > On 2016/10/03 at 17:28:54, sadrul wrote:
> > > On 2016/10/03 15:26:32, reveman wrote:
> > > > MojoGpuMemoryBufferManager seem to be lacking native GpuMemoryBuffer
> > support.
> > > > How does this change not disable native gpu memory buffer support on
intel
> > based
> > > > ChromeOS devices where this is currently used?
> > > 
> > > It is missing. I am refactoring this part of the code so mus does what
chrome
> > is doing. (the hope is that we will use the same implementation on both
chrome
> > and mus)
> > 
> > Can we revert this for ChromeOS until fixed? This is currently causing a
> > performance/power regression in M54 for Broadwell, Haswell and Skylake based
> > Chromebooks. It's especially confusing as chrome://gpu is reporting that
native
> > GMBs are used but under the hood they are not.
> 
> Oh? MojoGpuMemoryBufferManager is not actually used on any releases yet. It
should not be affecting anything.

Ah I see now. It's protected by the IsRunningInMash condition:
https://cs.chromium.org/chromium/src/content/renderer/render_thread_impl.cc?r...

Sorry for the noise.

Powered by Google App Engine
This is Rietveld 408576698