[webkit-changes] [WebKit/WebKit] 3f7af1: [UI-side compositing] Guard against UI process han...

Simon Fraser noreply at github.com
Mon May 8 11:18:48 PDT 2023


  Branch: refs/heads/main
  Home:   https://github.com/WebKit/WebKit
  Commit: 3f7af1859b92db6b242e50fa76450385e91628e0
      https://github.com/WebKit/WebKit/commit/3f7af1859b92db6b242e50fa76450385e91628e0
  Author: Simon Fraser <simon.fraser at apple.com>
  Date:   2023-05-08 (Mon, 08 May 2023)

  Changed paths:
    M Source/WebKit/UIProcess/RemoteLayerTree/mac/RemoteLayerTreeEventDispatcher.cpp

  Log Message:
  -----------
  [UI-side compositing] Guard against UI process hangs caused by incorrect frame rate data
https://bugs.webkit.org/show_bug.cgi?id=256399
rdar://108954321

Reviewed by Chris Dumez.

The spin report in the radar suggests that the Safari main thread is blocked on
RemoteLayerTreeEventDispatcher::mainThreadDisplayDidRefresh() because the scrolling thread
is stuck in RemoteLayerTreeEventDispatcher::waitForRenderingUpdateCompletionOrTimeout().
The only way this can happen is that we send a long timeout to m_stateCondition.waitUntil(),
but this timeout should be at most half of the frame duration.

So protect against bad m_scrollingTree->frameDuration() data by imposing a cap
at at most 32ms on this timeout, and release log if it seems wrong.

* Source/WebKit/UIProcess/RemoteLayerTree/mac/RemoteLayerTreeEventDispatcher.cpp:
(WebKit::RemoteLayerTreeEventDispatcher::waitForRenderingUpdateCompletionOrTimeout):

Canonical link: https://commits.webkit.org/263806@main




More information about the webkit-changes mailing list