Bug 126021 - Different and incorrect flags are passed to sendSync when accessibility is enabled, SpinRunLoopWhileWaitingForReply
Summary: Different and incorrect flags are passed to sendSync when accessibility is en...
Status: RESOLVED FIXED
Alias: None
Product: WebKit
Classification: Unclassified
Component: Accessibility (show other bugs)
Version: 528+ (Nightly build)
Hardware: Unspecified Unspecified
: P2 Normal
Assignee: mitz
URL:
Keywords: InRadar
Depends on: 127411 127429
Blocks:
  Show dependency tree / graph
 
Reported: 2013-12-19 15:06 PST by Alexey Proskuryakov
Modified: 2015-12-21 07:17 PST (History)
6 users (show)

See Also:

Attachments
Get rid of SpinRunLoopWhileWaitingForReply (14.46 KB, patch)
2015-12-20 08:23 PST, mitz 		darin: review+
Details | Formatted Diff | Diff
View All Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description Alexey Proskuryakov 2013-12-19 15:06:09 PST 
In WebKit2, we have multiple places where sendSync flags depend on whether accessibility is enabled.

    unsigned syncSendFlags = (WebCore::AXObjectCache::accessibilityEnabled()) ? CoreIPC::SpinRunLoopWhileWaitingForReply : 0;
    WebProcess::shared().parentProcessConnection()->sendSync(Messages::WebPageProxy::RunJavaScriptAlert(webFrame->frameID(), alertText), Messages::WebPageProxy::RunJavaScriptAlert::Reply(), m_page->pageID(), CoreIPC::Connection::NoTimeout, syncSendFlags);

It is unfortunate to have such differences, because we get to support two runtime behaviors instead of one. I think that we should either make SpinRunLoopWhileWaitingForReply unconditional, or stop passing SpinRunLoopWhileWaitingForReply, and find different fixes for bugs it was needed for.

Bug 126014 is one case where SpinRunLoopWhileWaitingForReply was found to cause significant trouble.
Comment 1 Alexey Proskuryakov 2013-12-19 15:32:09 PST 
See also: bug 123828.
Comment 2 Radar WebKit Bug Importer 2013-12-19 15:32:21 PST 
<rdar://problem/15703778>
Comment 3 Alexey Proskuryakov 2014-01-26 18:00:30 PST 
Removing "possibly" from the title, as it's certainly incorrect given the test crashes seen in blocking bugs.

Now that accessibility gets disabled after each test, this should affect testing, but it may cause instability in real life.
Comment 4 mitz 2015-12-20 08:23:54 PST 
Created attachment 267719 [details]
Get rid of SpinRunLoopWhileWaitingForReply
Comment 5 mitz 2015-12-21 07:17:37 PST 
Committed <http://trac.webkit.org/r194330>.