Title: [201263] trunk/Tools
- Revision
- 201263
- Author
- [email protected]
- Date
- 2016-05-22 15:55:23 -0700 (Sun, 22 May 2016)
Log Message
RELEASE_ASSERT(TestController::singleton().isCurrentInvocation(testInvocation)) is sometimes hit when running tests
https://bugs.webkit.org/show_bug.cgi?id=157977
rdar://problem/20874156
Reviewed by David Kilzer.
* WebKitTestRunner/TestInvocation.cpp:
(WTR::TestInvocation::invoke): Be more certain to launch a new process when there was
a timeout detected. If we don't, there is a chance that the process is not quite frozen,
and will still send a response that we no longer expect.
(WTR::TestInvocation::dumpWebProcessUnresponsiveness): Print timeout details to stderr
too, as stdout is not available, at least not in the case that I was observing.
Modified Paths
Diff
Modified: trunk/Tools/ChangeLog (201262 => 201263)
--- trunk/Tools/ChangeLog 2016-05-22 22:36:16 UTC (rev 201262)
+++ trunk/Tools/ChangeLog 2016-05-22 22:55:23 UTC (rev 201263)
@@ -1,3 +1,18 @@
+2016-05-22 Alexey Proskuryakov <[email protected]>
+
+ RELEASE_ASSERT(TestController::singleton().isCurrentInvocation(testInvocation)) is sometimes hit when running tests
+ https://bugs.webkit.org/show_bug.cgi?id=157977
+ rdar://problem/20874156
+
+ Reviewed by David Kilzer.
+
+ * WebKitTestRunner/TestInvocation.cpp:
+ (WTR::TestInvocation::invoke): Be more certain to launch a new process when there was
+ a timeout detected. If we don't, there is a chance that the process is not quite frozen,
+ and will still send a response that we no longer expect.
+ (WTR::TestInvocation::dumpWebProcessUnresponsiveness): Print timeout details to stderr
+ too, as stdout is not available, at least not in the case that I was observing.
+
2016-05-22 Brady Eidson <[email protected]>
Move to C++14.
Modified: trunk/Tools/WebKitTestRunner/TestInvocation.cpp (201262 => 201263)
--- trunk/Tools/WebKitTestRunner/TestInvocation.cpp 2016-05-22 22:36:16 UTC (rev 201262)
+++ trunk/Tools/WebKitTestRunner/TestInvocation.cpp 2016-05-22 22:55:23 UTC (rev 201263)
@@ -172,13 +172,15 @@
if (m_webProcessIsUnresponsive)
dumpWebProcessUnresponsiveness();
- else if (!TestController::singleton().resetStateToConsistentValues(m_options)) {
- // The process froze while loading about:blank, let's start a fresh one.
- // It would be nice to report that the previous test froze after dumping results, but we have no way to do that.
- TestController::singleton().terminateWebContentProcess();
- // Make sure that we have a process, as invoke() will need one to send bundle messages for the next test.
- TestController::singleton().reattachPageToWebProcess();
- }
+ else if (TestController::singleton().resetStateToConsistentValues(m_options))
+ return;
+ else
+ dumpWebProcessUnresponsiveness("TestController::resetStateToConsistentValues failed\n");
+
+ // The process is unresponsive, so let's start a new one.
+ TestController::singleton().terminateWebContentProcess();
+ // Make sure that we have a process, as invoke() will need one to send bundle messages for the next test.
+ TestController::singleton().reattachPageToWebProcess();
}
void TestInvocation::dumpWebProcessUnresponsiveness()
@@ -188,6 +190,7 @@
void TestInvocation::dumpWebProcessUnresponsiveness(const char* errorMessage)
{
+ fprintf(stderr, "%s", errorMessage);
char errorMessageToStderr[1024];
#if PLATFORM(COCOA)
pid_t pid = WKPageGetProcessIdentifier(TestController::singleton().mainWebView()->page());
_______________________________________________
webkit-changes mailing list
[email protected]
https://lists.webkit.org/mailman/listinfo/webkit-changes
