Bug 190249 – [GTK][WPE] Incorrect rendering of layers whose backingStore hasn't changed

• Component: WebKitGTK
• Status: RESOLVED FIXED

[Miguel Gomez]

CoordinatedGraphicsLayer::flushCompositingStateForThisLayerOnly() sets m_nicosia.performLayerSync to true when the layer properties have changed and the layer needs to be synced. But after that, CoordinatedGraphicsLayer::updateContentBuffers() sets that value to false if the layer has backingStore that doesn't need to me modified. This causes that the layer properties are not synced if the backingStore isn't modified as well.

[Miguel Gomez]

Created attachment 351519 [details]
Patch

[Miguel Gomez]

Created attachment 351520 [details]
Patch

[WebKit Commit Bot]

Comment on attachment 351520 [details]
Patch

Clearing flags on attachment: 351520

Committed r236794: <https://trac.webkit.org/changeset/236794>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.

[Michael Catanzaro]

Is this the commit that might fix bug #189743?

Should this be added to https://trac.webkit.org/wiki/WebKitGTK/2.22.x for backport to 2.22? The Nicosia code has changed dramatically since branching, so I'm not sure without investigation.

[Adrian Perez]

(In reply to Michael Catanzaro from comment #5)
> Is this the commit that might fix bug #189743?
> 
> Should this be added to https://trac.webkit.org/wiki/WebKitGTK/2.22.x for
> backport to 2.22? The Nicosia code has changed dramatically since branching,
> so I'm not sure without investigation.

Yesterday I was chatting with Žan about this and found it that if we
also merge the patch for bug #188693 beforehand, then this will apply
cleanly on 2.22.x as well. And in that case we also will want the patch
for bug #189224 to avoid hitting a performance regression. Note that I
did *NOT* get to the point of trying to run a build with them applied,
and only checked what would be the minimum needed to get this merged,
in case we wanted.

[Michael Catanzaro]

Is the bug this fixes present on 2.22?

[Miguel Gomez]

In which version is the youtube (In reply to Michael Catanzaro from comment #7)
> Is the bug this fixes present on 2.22?

This fixes a regression that was introduced by the fix to bug #188693, which I think is not in 2.22, so this won't fix anything in that release.

[Michael Catanzaro]

Thanks Miguel. In that case, we surely don't need to backport this.