I saw a crash on my build in DocumentLoader::stopLoading() because m_frame was set to NULL. stopLoading() assumes that if m_committed is true, m_frame is not NULL, but this does not seem to be enforced elsewhere in the code. Therefore, I've put together the attached patch which adds a NULL check for m_frame.
Created attachment 13339 [details] Check for m_frame == NULL patch
Comment on attachment 13339 [details] Check for m_frame == NULL patch Looks like a good fix, but this patch needs a test case. r- for lack of test case.
I'm not sure how useful this bug is. Someone with CrashReporter access (apple employee) could check to see if this is a common crash point. If it isn't, then we should probably just assume this has already been fixed by some other fix and close the bug. Thanks for the report!
(In reply to comment #3) > I'm not sure how useful this bug is. Someone with CrashReporter access (apple > employee) could check to see if this is a common crash point. If it isn't, > then we should probably just assume this has already been fixed by some other > fix and close the bug. Thanks for the report! > Any progress on this bug? What further action should be performed to proceed with this report?
Frameless documents don't do loading, so such a crash would indicate a problem somewhere else. But it's likely that is has been fixed since the bug was filed. Marking INVALID, since this could never be confirmed.