Bug 52603 - Consider not allowing WebView::m_backingLayer to go into tiling mode
Summary: Consider not allowing WebView::m_backingLayer to go into tiling mode
Status: NEW
Alias: None
Product: WebKit
Classification: Unclassified
Component: Layout and Rendering (show other bugs)
Version: 528+ (Nightly build)
Hardware: PC Windows XP
: P2 Normal
Assignee: Nobody
URL:
Keywords: InRadar, PlatformOnly
Depends on:
Blocks:
 
Reported: 2011-01-17 17:31 PST by Adam Roben (:aroben)
Modified: 2011-01-18 09:38 PST (History)
3 users (show)

See Also:


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Adam Roben (:aroben) 2011-01-17 17:31:04 PST
The fix for bug 52582 made WebView draw into a GraphicsLayer when in accelerated compositing mode. This GraphicsLayer can enter tiling mode if the window is more than 2000px wide or tall. It's possible that this could adversely affect performance, and we know of at least one visual glitch that occurs in tiling mode (bug 52602). We should investigate not allowing this layer to go into tiling mode at all, or perhaps raise the size threshold at which it enters tiling mode.
Comment 1 Adam Roben (:aroben) 2011-01-17 17:40:57 PST
<rdar://problem/8876821>
Comment 2 Vangelis Kokkevis 2011-01-18 09:38:51 PST
(In reply to comment #0)
> The fix for bug 52582 made WebView draw into a GraphicsLayer when in accelerated compositing mode. This GraphicsLayer can enter tiling mode if the window is more than 2000px wide or tall. It's possible that this could adversely affect performance, and we know of at least one visual glitch that occurs in tiling mode (bug 52602). We should investigate not allowing this layer to go into tiling mode at all, or perhaps raise the size threshold at which it enters tiling mode.

Wouldn't tiling be beneficial for the root layer?  For the Chromium port, tiling the root layer allows for more efficient scrolling.  However, we have indeed noticed correctness side-effects with off-screen fixed position elements (http://code.google.com/p/chromium/issues/detail?id=69692) whose invalidations do seem to be getting lost.