RESOLVED FIXED 151987
[CoordinatedGraphics] layerTreeHost always exist in CoordinatedDrawingArea
https://bugs.webkit.org/show_bug.cgi?id=151987
Summary [CoordinatedGraphics] layerTreeHost always exist in CoordinatedDrawingArea
Ryuan Choi
Reported 2015-12-08 02:48:25 PST
Clean up the unnecessary check routine and dead code in CoordinatedDrawingArea. CoordinatedGraphics using UI side compositor always forces accelerated compositing.
Attachments
Patch (25.62 KB, patch)
2015-12-08 02:51 PST, Ryuan Choi
no flags
Patch (35.04 KB, patch)
2015-12-08 08:43 PST, Ryuan Choi
no flags
Patch (36.22 KB, patch)
2015-12-08 09:14 PST, Ryuan Choi
no flags
Ryuan Choi
Comment 1 2015-12-08 02:51:47 PST
Gwang Yoon Hwang
Comment 2 2015-12-08 05:41:23 PST
Comment on attachment 266874 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=266874&action=review > Source/WebKit2/UIProcess/CoordinatedGraphics/CoordinatedDrawingAreaProxy.cpp:58 > , m_discardBackingStoreTimer(RunLoop::current(), this, &CoordinatedDrawingAreaProxy::discardBackingStore) Because we are not going into the non-accelerated code path anymore, you can remove all codes related with backing store states and updates.
Ryuan Choi
Comment 3 2015-12-08 06:50:06 PST
Comment on attachment 266874 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=266874&action=review >> Source/WebKit2/UIProcess/CoordinatedGraphics/CoordinatedDrawingAreaProxy.cpp:58 >> , m_discardBackingStoreTimer(RunLoop::current(), this, &CoordinatedDrawingAreaProxy::discardBackingStore) > > Because we are not going into the non-accelerated code path anymore, you can remove all codes related with backing store states and updates. I agree. but I am little bit worried that patch become difficult to be reviewed. How about separating it from this patch? If it doesn't matter, I will update the patch with more cleanup.
Gwang Yoon Hwang
Comment 4 2015-12-08 07:05:55 PST
I prefer to remove all codes related with backing store updates. And I think it would be easier to review :) (In reply to comment #3) > Comment on attachment 266874 [details] > Patch > > View in context: > https://bugs.webkit.org/attachment.cgi?id=266874&action=review > > >> Source/WebKit2/UIProcess/CoordinatedGraphics/CoordinatedDrawingAreaProxy.cpp:58 > >> , m_discardBackingStoreTimer(RunLoop::current(), this, &CoordinatedDrawingAreaProxy::discardBackingStore) > > > > Because we are not going into the non-accelerated code path anymore, you can remove all codes related with backing store states and updates. > > I agree. but I am little bit worried that patch become difficult to be > reviewed. > How about separating it from this patch? > > If it doesn't matter, I will update the patch with more cleanup.
Ryuan Choi
Comment 5 2015-12-08 08:43:28 PST
Ryuan Choi
Comment 6 2015-12-08 09:14:25 PST
Gwang Yoon Hwang
Comment 7 2015-12-09 01:33:51 PST
Comment on attachment 266895 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=266895&action=review It looks good to me except some nitpics. nice work. > Source/WebKit2/UIProcess/CoordinatedGraphics/CoordinatedDrawingAreaProxy.h:69 > + virtual void setBackingStoreIsDiscardable(bool) { } You can delete this line because it is already defined at DrawingAreaProxy > Source/WebKit2/WebProcess/WebPage/CoordinatedGraphics/CoordinatedDrawingArea.cpp:85 > + ASSERT(m_dirtyRegion.isEmpty()); I think you can remove m_dirtyRegion as well because we are not going to update this anymore after this patch. > Source/WebKit2/WebProcess/WebPage/CoordinatedGraphics/CoordinatedDrawingArea.cpp:94 > + ASSERT(m_dirtyRegion.isEmpty()); ditto. > Source/WebKit2/WebProcess/WebPage/CoordinatedGraphics/CoordinatedDrawingArea.cpp:104 > + ASSERT(m_scrollOffset.isEmpty()); We can remove m_scrollRect and m_scrollOffset as well.
Ryuan Choi
Comment 8 2015-12-09 01:43:04 PST
Ryuan Choi
Comment 9 2015-12-09 01:43:49 PST
Comment on attachment 266895 [details] Patch Clearing flags after landed manually with following yoon's comments.
Note You need to log in before you can comment on or make changes to this bug.