Summary: | Loading of certain resources can hang when a website uses long running XMLHttpRequests and iframes | ||||||
---|---|---|---|---|---|---|---|
Product: | WebKit | Reporter: | Mario <mario.rasch> | ||||
Component: | Frames | Assignee: | Nobody <webkit-unassigned> | ||||
Status: | UNCONFIRMED --- | ||||||
Severity: | Critical | CC: | ap, beidson, medikoo+webkit.org | ||||
Priority: | P2 | ||||||
Version: | 528+ (Nightly build) | ||||||
Hardware: | Mac (Intel) | ||||||
OS: | OS X 10.5 | ||||||
Attachments: |
|
Description
Mario
2009-07-17 15:26:20 PDT
First off, why if the title says "crashes", please provide a crash report - We can't explore a crash without a crash report. Secondly, please try this again with the newest nightly. We've recently made fixes in loading code that seem relevant here, if not the exact same issue. Third, if the issue does still reproduce, it *will* be very difficult to explore it without a live example. Note that a custom crafted reduced test case qualifies... Created attachment 33021 [details]
crashlog from latest nightly build 46042
I just tried the latest bightly build again and it behaves a lot better than the current official Safari 4.0.2. Where Safari already hangs after the first or second click in the iframe (so that the page gets posted back), the nightly WebKit build behaved more stable. But still (it seems especially when initiating a new postback in the iframe before the last response in the iframe is completely loaded so that the progress bar in the address bar disappeared), the nightly WebKit hangs sometimes and finally crashed. The crash log is attached. I'll try to provide a reduced test case but this may take some days. A testcase is now available at http://xdf.c4b.de/devtest/serverpush.pages/lazytest.aspx. One can see the following: as soon as the XMLHTTPRequest is running and waits for a response, some images do not load completely when one clicks the "Refresh" button. Instead, loading hangs and does not continue until the XMLHTTPRequest gets a server response (e.g. a "KeepAlive" after 30 seconds approx. -> displayed in the debug console that appears when one clicks on the bug in the lower left corner). Alas, this testcase does not reproduce the crash in my case, so it seems there must apply some other conditions in my website to cause the crash. But at least, you should now be able to reproduce the hangs and possibly find a bugfix for this issue. Re-titling to reflect the only symptom that can be reproduced. |