297945 - Firefox keeps loading favicon.... forever

Status: RESOLVED FIXED

(Core :: DOM: Navigation, defect)

Description

[Peter van der Woude [:Peter6]]

Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.8b2) Gecko/20050616
Firefox/1.0+ ID:2005061612

repro:
1.remove/rename profile directory
2.start FF to create new profile
3.after this is done FF will start on http://www.mozilla.org/projects/deerpark/
4.close FF (don't change any prefs !important)
5.open Firefox 
6.http://www.mozilla.org/projects/deerpark/ will open and never finishes loading

Comment 1

[Peter van der Woude [:Peter6]]

*** This bug has been marked as a duplicate of 297941 ***

Comment 2

[Peter van der Woude [:Peter6]]

*** Bug 297941 has been marked as a duplicate of this bug. ***

Comment 3

[Peter van der Woude [:Peter6]]

regressionwindow:

works 20050613 2334pdt build
fails 20050614 0145pdt build

checkins
http://bonsai.mozilla.org/cvsquery.cgi?treeid=default&module=PhoenixTinderbox&branch=HEAD&branchtype=match&dir=&file=&filetype=match&who=&whotype=match&sortby=Date&hours=2&date=explicit&mindate=2005-06-13+22%3A50%3A00&maxdate=2005-06-14+01%3A25%3A00&cvsroot=%2Fcvsroot

possible regressor bug 109672 ??
CC: vlad

Comment 4

[Christian Schmidt]

This seems to be related to, whether the document is loaded from cache or not.

If you set the cache size to 0, or manually delete the contents of C:\Documents
and Settings\XXX\Local Settings\Application
Data\Mozilla\Firefox\Profiles\fooobar.default\Cache, the bug does not appear.

But if the document is loaded from cache, it looks as if onStateChange never
gets the STATE_STOP event, and that the load event never occurs.

Perhaps this could be related to bug 287669?

This bug may be exposed by the fix for bug 109672, but I think the real problem
is not in that bug but elsewhere.

Comment 5

[Christian Schmidt]

I have not been able to reproduce this on Linux.

The URL in question sends Expires-headers that are one hour into the future.
This allows Firefox to load the page without loading anything over the network.
This is unusual compared to other sites.

Comment 6

[Peter van der Woude [:Peter6]]

other example url
http://www.google.com/firefox?client=firefox-a&rls=org.mozilla:en-US:official

Comment 7

[Ryan Hill]

Confirmed.
Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8b2) Gecko/20050617 Firefox/1.0+

With BFCache on I can reproduce this easily.

1. have BFCache enabled.
2. navigate from any page to any other page using a hyperlink
3. press back.

result - throbber spins and spins, tab title says "Loading...", stop button is
clickable but has no effect.

notes - this does seem to be related to the cache.  while attempting to
reproduce i found that i could also trigger the bug by loading a page i had
already loaded this session.  eg. navigating to www.google.ca will now always
trigger it.

Comment 8

[Peter B]

It seems that ff doesn't send a GET request for the favicon. But waits forever
for one to come..

Temp fix: Use browser.jar and toolkit.jar from 050613 nighty...

Comment 9

[Peter van der Woude [:Peter6]]

Attachment: testcase

reproduce:

open testcase
close testcase
open testcase and watch FF never stop loading

Comment 10

[Peter van der Woude [:Peter6]]

Comment on attachment 186753 [details]
testcase

somehow it must have been that build i used.The testcase works again

Comment 11

[Christian Schmidt]

What I said about STATE_STOP not happening was wrong (at least I cannot 
reproduce it anymore).

If I remove the call to BookmarksUtils.loadFavIcon(), the bug does not appear. 
Appearently this method takes forever or kills the currently execution thread 
without triggering a warning.

If I change the loadFavIcon() method by adding an alert() just above the first 
line (the call to RDF.GetLiteral()), the alert box is displayed and the bug 
does not appear.

If I add an alert() just below the first line, the bug appears and the alert 
box is never opened.

If I add an alert() above and below the first line, both alerts are opened, and 
the bug does not appear.

Could this be some kind of concurrency issue in RDF.GetLiteral()?

Comment 12

[Brian Ryner (not reading)]

Ok, so since there were no bfcache-related checkins in the regression window, I
don't think this is a bfcache issue per se.  This bug should be considered
indpendent of bfcache.

Comment 13

[Christian Schmidt]

Attachment: Fix (made using Patch Maker)

The problem was that BookmarksUtils.loadFavIcon() was called before
initServices() and initBMService() in bookmarks.js were called from
delayedStartup() in browser.js.

loadFavIcon() now checks that the services have been initialised - otherwise it
just returns without updating the favicon. This (hopefully) isn't a big problem
- if the user has bookmarked his startup page, the favicon will be updated when
he loads the page using the bookmark.

Comment 14

[Vladimir Vukicevic [:vlad] [:vladv] (needinfo me, slow to respond)]

Comment on attachment 187040 [details] [diff] [review]
Fix (made using Patch Maker)

r=vladimir

Comment 15

[Asaf Romano (gone)]

Checking in bookmarks.js;
/cvsroot/mozilla/browser/components/bookmarks/content/bookmarks.js,v  <-- 
bookmarks.js
new revision: 1.101; previous revision: 1.100
done