[SCM] WebKit Debian packaging branch, debian/experimental, updated. upstream/1.3.3-9427-gc2be6fc
eric at webkit.org
eric at webkit.org
Wed Dec 22 12:57:07 UTC 2010
The following commit has been merged in the debian/experimental branch:
commit 131c80c903abcd425b80edd6aa787b8f8e2bb6a9
Author: eric at webkit.org <eric at webkit.org@268f45cc-cd09-0410-ab3c-d52691b4dbfc>
Date: Thu Sep 2 20:10:42 2010 +0000
2010-09-02 Eric Seidel <eric at webkit.org>
Reviewed by Alexey Proskuryakov.
fast/loader/recursive-before-unload-crash.html is flaky
https://bugs.webkit.org/show_bug.cgi?id=43840
Removed the frame loader callbacks as Brady suggested in
https://bugs.webkit.org/show_bug.cgi?id=38928#c27
This should stop the flake. As an added bonus this test no longer
needs platform specific results.
* fast/loader/recursive-before-unload-crash-expected.txt:
* fast/loader/recursive-before-unload-crash.html:
* platform/chromium-linux/fast/loader/recursive-before-unload-crash-expected.txt: Removed.
* platform/mac-tiger/fast/loader/recursive-before-unload-crash-expected.txt: Removed.
git-svn-id: http://svn.webkit.org/repository/webkit/trunk@66681 268f45cc-cd09-0410-ab3c-d52691b4dbfc
diff --git a/LayoutTests/ChangeLog b/LayoutTests/ChangeLog
index c62c187..b0f0a19 100644
--- a/LayoutTests/ChangeLog
+++ b/LayoutTests/ChangeLog
@@ -1,3 +1,20 @@
+2010-09-02 Eric Seidel <eric at webkit.org>
+
+ Reviewed by Alexey Proskuryakov.
+
+ fast/loader/recursive-before-unload-crash.html is flaky
+ https://bugs.webkit.org/show_bug.cgi?id=43840
+
+ Removed the frame loader callbacks as Brady suggested in
+ https://bugs.webkit.org/show_bug.cgi?id=38928#c27
+ This should stop the flake. As an added bonus this test no longer
+ needs platform specific results.
+
+ * fast/loader/recursive-before-unload-crash-expected.txt:
+ * fast/loader/recursive-before-unload-crash.html:
+ * platform/chromium-linux/fast/loader/recursive-before-unload-crash-expected.txt: Removed.
+ * platform/mac-tiger/fast/loader/recursive-before-unload-crash-expected.txt: Removed.
+
2010-09-02 Tony Chang <tony at chromium.org>
Unreviewed, update chromium test expectations.
diff --git a/LayoutTests/fast/loader/recursive-before-unload-crash-expected.txt b/LayoutTests/fast/loader/recursive-before-unload-crash-expected.txt
index 5dbf0a6..fa9cf2b 100644
--- a/LayoutTests/fast/loader/recursive-before-unload-crash-expected.txt
+++ b/LayoutTests/fast/loader/recursive-before-unload-crash-expected.txt
@@ -1,14 +1,5 @@
-main frame - didFinishDocumentLoadForFrame
-main frame - willPerformClientRedirectToURL: http://127.0.0.1:1234/
-main frame - didHandleOnloadEventsForFrame
-main frame - didFinishLoadForFrame
ALERT: onbeforeunload called, and iframe hasn't been added yet.
-main frame - didStartProvisionalLoadForFrame
ALERT: Adding iframe
-frame "<!--framePath //<!--frame0-->-->" - didStartProvisionalLoadForFrame
-main frame - didCancelClientRedirectForFrame
-frame "<!--framePath //<!--frame0-->-->" - didFailProvisionalLoadWithError
-main frame - didFailProvisionalLoadWithError
This test demonstrates a problem with our handling of the beforeunload event.
If a script manages to try and navigate the frame from beforeunload - when a navigation is already pending - we end up blowing out the stack by recursively consulting the policy delegate then running onbeforeunload repeatedly.
After this happens, the FrameLoader is in a bogus state where it thinks it is in the middle of a provisional load, but it doesn't have a provisional document loader.
diff --git a/LayoutTests/fast/loader/recursive-before-unload-crash.html b/LayoutTests/fast/loader/recursive-before-unload-crash.html
index 41da730..bb4f0d8 100644
--- a/LayoutTests/fast/loader/recursive-before-unload-crash.html
+++ b/LayoutTests/fast/loader/recursive-before-unload-crash.html
@@ -2,7 +2,6 @@
if (window.layoutTestController) {
layoutTestController.waitUntilDone();
layoutTestController.dumpAsText();
- layoutTestController.dumpFrameLoadCallbacks();
}
var haveAddedIFrame = false;
diff --git a/LayoutTests/platform/chromium-linux/fast/loader/recursive-before-unload-crash-expected.txt b/LayoutTests/platform/chromium-linux/fast/loader/recursive-before-unload-crash-expected.txt
deleted file mode 100644
index a76016d..0000000
--- a/LayoutTests/platform/chromium-linux/fast/loader/recursive-before-unload-crash-expected.txt
+++ /dev/null
@@ -1,18 +0,0 @@
-main frame - didFinishDocumentLoadForFrame
-main frame - willPerformClientRedirectToURL: http://127.0.0.1:1234/
-main frame - didHandleOnloadEventsForFrame
-main frame - didFinishLoadForFrame
-ALERT: onbeforeunload called, and iframe hasn't been added yet.
-main frame - didStartProvisionalLoadForFrame
-main frame - didCancelClientRedirectForFrame
-main frame - didFailProvisionalLoadWithError
-ALERT: Adding iframe
-frame "<!--framePath //<!--frame0-->-->" - didStartProvisionalLoadForFrame
-frame "<!--framePath //<!--frame0-->-->" - didFailProvisionalLoadWithError
-This test demonstrates a problem with our handling of the beforeunload event.
-If a script manages to try and navigate the frame from beforeunload - when a navigation is already pending - we end up blowing out the stack by recursively consulting the policy delegate then running onbeforeunload repeatedly.
-After this happens, the FrameLoader is in a bogus state where it thinks it is in the middle of a provisional load, but it doesn't have a provisional document loader.
-In this state, the frame is very difficult to navigate anywhere else, and attempts to load new things within the frame can result in a crash.
-This was reproducibly identified on sears.com following a bizarre Safari specific code path.
-Click here to run the beforeunload test and blow out the stack
-Click here to append an iframe and crash
diff --git a/LayoutTests/platform/mac-tiger/fast/loader/recursive-before-unload-crash-expected.txt b/LayoutTests/platform/mac-tiger/fast/loader/recursive-before-unload-crash-expected.txt
deleted file mode 100644
index 408fe6e..0000000
--- a/LayoutTests/platform/mac-tiger/fast/loader/recursive-before-unload-crash-expected.txt
+++ /dev/null
@@ -1,19 +0,0 @@
-main frame - didFinishDocumentLoadForFrame
-main frame - willPerformClientRedirectToURL: http://127.0.0.1:1234/
-main frame - didHandleOnloadEventsForFrame
-main frame - didFinishLoadForFrame
-ALERT: onbeforeunload called, and iframe hasn't been added yet.
-main frame - didStartProvisionalLoadForFrame
-ALERT: Adding iframe
-frame "<!--framePath //<!--frame0-->-->" - didStartProvisionalLoadForFrame
-frame "<!--framePath //<!--frame0-->-->" - didFailProvisionalLoadWithError
-main frame - didCancelClientRedirectForFrame
-main frame - didFailProvisionalLoadWithError
-This test demonstrates a problem with our handling of the beforeunload event.
-If a script manages to try and navigate the frame from beforeunload - when a navigation is already pending - we end up blowing out the stack by recursively consulting the policy delegate then running onbeforeunload repeatedly.
-After this happens, the FrameLoader is in a bogus state where it thinks it is in the middle of a provisional load, but it doesn't have a provisional document loader.
-In this state, the frame is very difficult to navigate anywhere else, and attempts to load new things within the frame can result in a crash.
-This was reproducibly identified on sears.com following a bizarre Safari specific code path.
-Click here to run the beforeunload test and blow out the stack
-Click here to append an iframe and crash
-
--
WebKit Debian packaging
More information about the Pkg-webkit-commits
mailing list