Summary: | REGRESSION: Crash closing live web cam viewer page | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
Product: | WebKit | Reporter: | Gibbons Burke <gibbonsb> | ||||||||
Component: | Page Loading | Assignee: | Nobody <webkit-unassigned> | ||||||||
Status: | RESOLVED FIXED | ||||||||||
Severity: | Critical | CC: | beidson, ddkilzer, justin.garcia, mitz, p.sipkes | ||||||||
Priority: | P1 | Keywords: | InRadar, Regression | ||||||||
Version: | 523.x (Safari 3) | ||||||||||
Hardware: | Mac | ||||||||||
OS: | OS X 10.4 | ||||||||||
URL: | http://webcam.sewanee.edu/view/view.shtml | ||||||||||
Attachments: |
|
Description
Gibbons Burke
2007-04-16 04:39:45 PDT
ASSERTION FAILED: m_loadingResources.contains(identifier) (/WebKit/WebKit/WebView/WebDocumentLoaderMac.mm:93 decreaseLoadCount) This seems almost certainly related to http://bugs.webkit.org/show_bug.cgi?id=13365 *** Bug 13540 has been marked as a duplicate of this bug. *** *** Bug 13573 has been marked as a duplicate of this bug. *** *** Bug 13580 has been marked as a duplicate of this bug. *** Created attachment 14333 [details]
Crash report
Created attachment 14367 [details]
Avoid calling decreaseLoadCount() twice for multipart subresources
This patch fixes the crash, but does not address other problems with multipart subresources in TOT (at least with Safari 2), namely that the progress bar and status bar remain in "loading" state after the first part is received, while the Activity panel does not report the additional data received. The behavior of shipping WebKit is the opposite: the progress bar and status bar indicate that the page has finished loading, but the size of the resource in the Activity panel keeps increasing.
The mjpg source still does not render smoothly. It's like watching a television with bad reception. Latest nightly does not crash however. Mitz, I think your patch is dead on, but I didn't know it'd been sitting here cause you didn't put it up for review! :) Now comes the layout test... If we don't already have a "multipart resource" in the http tests we might need one. Peter, The webcam view not being smooth is a seperate bug - if you still see it, please file a new bug to track that issue. That said, I do *not* see any problems with the smoothness/rendering. Is it possible its related to your network connection and the server is configured to send a lower-quality image down slow/error-hindered network connections? I tried the webcam with Safari's latest version from Apple (419.3) and it works fine on my network. Same with Firefox and even IE. Only Webkit doesn't render smoothly and still crashes. Should I file a new bug, or do we consider this one to be still open? Thanks. (In reply to comment #11) > I tried the webcam with Safari's latest version from Apple (419.3) and it works > fine on my network. Same with Firefox and even IE. Only Webkit doesn't render > smoothly and still crashes. Should I file a new bug, or do we consider this one > to be still open? Thanks. Peter, please file a new bug for the "rendering smoothly" issue. Thanks! (In reply to comment #12) > Peter, please file a new bug for the "rendering smoothly" issue. Thanks! Bug 13642. Created attachment 14574 [details]
Avoid calling decreaseLoadCount() twice for multipart subresources
Now with change log and an http test.
Comment on attachment 14574 [details]
Avoid calling decreaseLoadCount() twice for multipart subresources
r=me. I know there are still issues with multipart, including some in WebCore that we can actually address, but I think this is still an important and correct fix for this crasher
Landed in r21497 |