Summary: | REGRESSION (r143619): Crashes in three layout tests | ||||||
---|---|---|---|---|---|---|---|
Product: | WebKit | Reporter: | Zan Dobersek <zan> | ||||
Component: | WebKitGTK | Assignee: | Martin Robinson <mrobinson> | ||||
Status: | RESOLVED FIXED | ||||||
Severity: | Normal | CC: | cgarcia, george.mccollister, mrobinson, plaes, webkit.review.bot | ||||
Priority: | P2 | Keywords: | Gtk, LayoutTestFailure, Regression | ||||
Version: | 528+ (Nightly build) | ||||||
Hardware: | Unspecified | ||||||
OS: | Unspecified | ||||||
Attachments: |
|
Description
Zan Dobersek
2013-02-22 04:42:10 PST
Created attachment 190109 [details]
Patch
Comment on attachment 190109 [details]
Patch
Makes sense, took me a while to understand which g_str_equal it was crashing on.
Comment on attachment 190109 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=190109&action=review > Source/WebKit/gtk/WebCoreSupport/FrameLoaderClientGtk.cpp:1024 > - if (!g_str_equal(identifierString.get(), webView->priv->mainResourceIdentifier.data())) > - webkit_web_view_remove_resource(webView, identifierString.get()); > + webkitWebViewRemoveSubresource(webView, identifierString.get()); I think that at this point identifierString should never be null, this might be hiding a bug in the WebCore loader. > Source/WebKit/gtk/WebCoreSupport/FrameLoaderClientGtk.cpp:1048 > - if (!g_str_equal(identifierString.get(), webView->priv->mainResourceIdentifier.data())) > - webkit_web_view_remove_resource(webView, identifierString.get()); > + webkitWebViewRemoveSubresource(webView, identifierString.get()); Ditto. > Source/WebKit/gtk/webkit/webkitwebview.cpp:5075 > + if (!mainResource.isNull() && g_str_equal(identifier, mainResource.data())) Instead of checking null and then compare we might use g_strcmp0 that already handles null pointers gracefully. Comment on attachment 190109 [details] Patch Clearing flags on attachment: 190109 Committed r144055: <http://trac.webkit.org/changeset/144055> All reviewed patches have been landed. Closing bug. Comment on attachment 190109 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=190109&action=review >> Source/WebKit/gtk/WebCoreSupport/FrameLoaderClientGtk.cpp:1024 >> + webkitWebViewRemoveSubresource(webView, identifierString.get()); > > I think that at this point identifierString should never be null, this might be hiding a bug in the WebCore loader. Possibly. >> Source/WebKit/gtk/webkit/webkitwebview.cpp:5075 >> + if (!mainResource.isNull() && g_str_equal(identifier, mainResource.data())) > > Instead of checking null and then compare we might use g_strcmp0 that already handles null pointers gracefully. I did initially use g_strcmp0, but decided on this approach, because it seemed clearer. If mainResource wasn't a CString I would have used g_strcmp0. *** Bug 111870 has been marked as a duplicate of this bug. *** |