Fix workaround of https://github.com/otcshare/chromium-crosswalk/pull/49 upstream. #516

darktears · 2013-08-30T16:36:40Z

crosswalk-project/chromium-crosswalk#49 is way to intrusive, way too big to stay in chromium-crosswalk. The team aimed to reduce the patchset on top of Chromium while rebasing from 28 to 29 by upstreaming some of the patches, this big pull request goes in the opposite direction. It needs to be fixed ASAP by either solving the problem upstream (or by investigating a way to make the patch less intrusive but let's aim for upstream).

The team didn't have time to look in details, we landed this workaround but it needs to be investigated ASAP so we don't stay with it long.

tmpsantos · 2013-08-30T16:47:28Z

I really hope this get fixed ASAP. With the small set of patches we had before, rebasing into a new version was extremely difficult and time consuming. After this patch things should get way worse.

yongsheng · 2013-09-02T00:54:47Z

yes, we'll do that.

sqliu · 2013-09-02T01:18:08Z

I'll refine the patch.

jeez · 2013-10-02T15:56:01Z

Any news here? I'm rebasing to chromium v31 and this patch will probably be broken afterwards. It is so intrusive that it is hard to tell if the rebase was correct or not.

This issue was opened a month ago and nothing has been done about this yet?

sqliu · 2013-10-08T06:16:17Z

Hi @jeez, I'll re-submit another much simpler patch for fixing this issue.

jeez · 2013-10-08T12:02:33Z

Thanks for the news, but why don't you consider to submit such patch to upstream instead?

jeez · 2013-10-08T13:33:20Z

If I got it right from #824 we can now simply take this patch of of the tree... is that right?

jeez · 2013-10-08T13:52:12Z

And, by the way, if that is the case, for historical reason can someone please elaborate on why we needed that patch before and why we simply don't need it anymore? Thanks!

sqliu · 2013-10-09T04:57:03Z

@jeez thanks for reverting the patch.

And crosswalk-project/chromium-crosswalk#67 is the much simpler fix for the rendering issue.

For the upstream, this is not the case that upstream chromium/contentshell will consider (see commit log). So I don't think upstream will accept the fix.

ininallen · 2013-10-22T07:49:36Z

This bug has been migrated to JIRA, please track this issue in https://crosswalk-project.org/jira/browse/XWALK-43 . Thanks.

yongsheng · 2013-10-23T01:34:13Z

since it's not accepted by upstream, is this still p1 bug?
@darktears @baleboy

baleboy · 2013-10-23T02:11:31Z

@yongsheng could you point to the comments explaining why this wasn't accepted upstream? I'm not sure which commit logs I should look at. Thanks!

yongsheng · 2013-10-23T06:12:34Z

@baleboy It's for in process mode rendering. Currently we have a quick solution to fix this and merged in crosswalk-chromium. In upstream, we're tracking it and need a more complex solution. And we agree to track it but I don't think it's a P1(blocking) issue.
@darktears @sqliu

darktears · 2013-10-23T11:00:14Z

If we don't put it P1 it will root like last time. Can we modify the roadmap to add this work needed upstream. I would like to be solved before the next rebase that will happen in few weeks.

jeez · 2013-10-23T12:48:50Z

UPSTREAM WORK being tracked here: https://codereview.chromium.org/26293006/

(it was hard to find this link, had to look into 2 or 3 pull requests)

We should keep this as a P1 bug. There is already an upstream fix for this that we can cherry-pick as soon as it is landed -> https://codereview.chromium.org/35503004/

I'm glad we have move forward with the upstream work for this, but let's not forget we have only reached this point because we have kept this as a P1 bug and because we have been pushing it since the latest rebase.

baleboy · 2013-10-23T12:51:37Z

Putting back to P1 to ensure some closure on this issue. Hopefully we can cherry pick the upstream patch soon.

* In Chromium, single process mode is only for Android WebView, ContentShell and its derivatives only go to the multiple process mode. * The context design is only for multiple process mode(ContentShell/XWalk), if we force running it with single process mode, there are issues (black regions for video/canvas/webgl layers, and black screen for stop/restart actions). The reason is that Android port has browser-side compositor, and the shared context in browser-side compositor has conflicts with renderer's contexts if running in single process mode. * To run ContentShell/XWalk in single process mode, we have to resolve the shared context issue between renderer/browser. BUG=crosswalk-project/crosswalk#516 Conflicts: content/browser/renderer_host/compositor_impl_android.cc content/browser/renderer_host/image_transport_factory_android.cc content/common/gpu/client/webgraphicscontext3d_command_buffer_impl.cc content/common/gpu/client/webgraphicscontext3d_command_buffer_impl.h Rework with 32.0.1700.4 by Halton

* In Chromium, single process mode is only for Android WebView, ContentShell and its derivatives only go to the multiple process mode. * The context design is only for multiple process mode(ContentShell/XWalk), if we force running it with single process mode, there are issues (black regions for video/canvas/webgl layers, and black screen for stop/restart actions). The reason is that Android port has browser-side compositor, and the shared context in browser-side compositor has conflicts with renderer's contexts if running in single process mode. * To run ContentShell/XWalk in single process mode, we have to resolve the shared context issue between renderer/browser. BUG=crosswalk-project/crosswalk#516

* In Chromium, single process mode is only for Android WebView, ContentShell and its derivatives only go to the multiple process mode. * The context design is only for multiple process mode(ContentShell/XWalk), if we force running it with single process mode, there are issues (black regions for video/canvas/webgl layers, and black screen for stop/restart actions). The reason is that Android port has browser-side compositor, and the shared context in browser-side compositor has conflicts with renderer's contexts if running in single process mode. * To run ContentShell/XWalk in single process mode, we have to resolve the shared context issue between renderer/browser. BUG=crosswalk-project/crosswalk#516 Conflicts: content/browser/renderer_host/compositor_impl_android.cc content/browser/renderer_host/image_transport_factory_android.cc content/common/gpu/client/webgraphicscontext3d_command_buffer_impl.cc content/common/gpu/client/webgraphicscontext3d_command_buffer_impl.h Rework with 32.0.1700.4 code base.

* In Chromium, single process mode is only for Android WebView, ContentShell and its derivatives only go to the multiple process mode. * The context design is only for multiple process mode(ContentShell/XWalk), if we force running it with single process mode, there are issues (black regions for video/canvas/webgl layers, and black screen for stop/restart actions). The reason is that Android port has browser-side compositor, and the shared context in browser-side compositor has conflicts with renderer's contexts if running in single process mode. * To run ContentShell/XWalk in single process mode, we have to resolve the shared context issue between renderer/browser. BUG=crosswalk-project/crosswalk#516

ghost assigned sqliu Aug 30, 2013

darktears mentioned this issue Aug 30, 2013

[Android] Fix a set of rendering issues introduced by single process mod... crosswalk-project/chromium-crosswalk#49

Merged

ghost assigned sqliu Sep 2, 2013

rakuco mentioned this issue Oct 4, 2013

Enable DIP for Aura DesktopScreenX11. crosswalk-project/chromium-crosswalk#66

Closed

sqliu mentioned this issue Oct 9, 2013

[Android] Fix the render issue introduced by single process mode. crosswalk-project/chromium-crosswalk#67

Merged

ibelem closed this as completed Nov 1, 2013

rakuco unassigned sqliu Mar 5, 2015

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Fix workaround of https://github.com/otcshare/chromium-crosswalk/pull/49 upstream. #516

Fix workaround of https://github.com/otcshare/chromium-crosswalk/pull/49 upstream. #516

darktears commented Aug 30, 2013

tmpsantos commented Aug 30, 2013

yongsheng commented Sep 2, 2013

sqliu commented Sep 2, 2013

jeez commented Oct 2, 2013

sqliu commented Oct 8, 2013

jeez commented Oct 8, 2013

jeez commented Oct 8, 2013

jeez commented Oct 8, 2013

sqliu commented Oct 9, 2013

ininallen commented Oct 22, 2013

yongsheng commented Oct 23, 2013

baleboy commented Oct 23, 2013

yongsheng commented Oct 23, 2013

darktears commented Oct 23, 2013

jeez commented Oct 23, 2013

baleboy commented Oct 23, 2013

Fix workaround of https://github.com/otcshare/chromium-crosswalk/pull/49 upstream. #516

Fix workaround of https://github.com/otcshare/chromium-crosswalk/pull/49 upstream. #516

Comments

darktears commented Aug 30, 2013

tmpsantos commented Aug 30, 2013

yongsheng commented Sep 2, 2013

sqliu commented Sep 2, 2013

jeez commented Oct 2, 2013

sqliu commented Oct 8, 2013

jeez commented Oct 8, 2013

jeez commented Oct 8, 2013

jeez commented Oct 8, 2013

sqliu commented Oct 9, 2013

ininallen commented Oct 22, 2013

yongsheng commented Oct 23, 2013

baleboy commented Oct 23, 2013

yongsheng commented Oct 23, 2013

darktears commented Oct 23, 2013

jeez commented Oct 23, 2013

baleboy commented Oct 23, 2013