Closed Bug 547472 Opened 15 years ago Closed 12 years ago

Firefox 3.6 hangs on Windows 7 when used with DownThemAll!

Categories

(Firefox :: General, defect)

3.6 Branch
All
Windows 7
defect
Not set
major

Tracking

()

RESOLVED INCOMPLETE

People

(Reporter: nmaier, Unassigned)

References

()

Details

(Keywords: hang)

Some of our users reported that Firefox becomes unresponsive after some time when our DownThemAll! extension is in use. It seems the UI thread is locked (deadlocked or whatever) in some way. The usual suspect, Antivirus Software, is not to blame this time. Since dTa is javascript only, it cannot be some random accidental endless-loop as this would produce a slow script warning. Furthermore this is a regression from Firefox 3.5, as the latest 3.5 builds do not produce any hangs (users reported a downgrade works). I cannot decently debug or even reproduce this, so I'm hoping that somebody here can. See http://bugs.code.downthemall.net/trac/ticket/1428 and dupe'd tickets for more gory details. I also attached a .dmp file as produced by MS debugging tools (MiniDump Tool, http://bill.atwill.com/) a user kindly provided of a running but unresponsive Fx instance. For whatever reason my MSVC2005/2008 refuses to load most of the symbols, so I don't get any information about the call stacks. (As the process does not actually crash, there won't be any crash reports unfortunately) Reproduce (according to users): * Install DownThemAll! (release from AMO and latest nightly from http://bugs.code.downthemall.net/trac/wiki/NightlyBuilds both produce such soft-crashes) * Use DownThemAll! to download some file(s). As users reported the "crash" will only occur after dTa has the download running for some time be sure to download a multi-MB file. * If you're affected the Firefox process will become unresponsive.
Flags: blocking1.9.0.19?
Flags: blocking1.8.1.next?
Firefox becoming unresponsive is not the same thing as a crash. As the developer of downthemall been contacted? I do believe (may not be correct) that pure JS extensions can cause a hang. If it was a crash, then yes, it's our fault, not so sure about a hang.
Severity: critical → major
Keywords: hang
Summary: Firefox 3.6 becomes unresponsive (soft-crash) on Windows 7 when used with DownThemAll! → Firefox 3.6 hangs on Windows 7 when used with DownThemAll!
(In reply to comment #1) > Firefox becoming unresponsive is not the same thing as a crash. Yep, but I call it a "soft-crash" ;) > As the developer of downthemall been contacted? I am the developer of DownThemAll!. > I do believe (may not be > correct) that pure JS extensions can cause a hang. If it was a crash, then yes, > it's our fault, not so sure about a hang. Yep, there are certain ways for an extension to trigger a hang or even a crash. I don't think, however, that this hang is because of DownThemAll! code. Many times what actually happens then is the extension triggers a bug in mozilla code, some loaded third party code (Antivirus ;) or even the OS itself. As to here, I'm not certain what exactly triggers this class of hangs. Most likely it is either some IO operation (disk or net) or windows painting (unlikely, but you never know with Win7 having a new window manager and all). As DownThemAll! works with moz-1.9.1 builds but not moz-1.9.2 for some users I consider this a regression in Firefox until I get proof that the extension code is to blame. I'd be more than happy to fix any dTa! bugs or at least implement a work-around. Unfortunately I cannot reproduce these hangs at all myself, or else I could have fired up the debugger or at least tracked down the regression range.
Based on the information to date, this isn't a blocker, especially not on these older branches.
Flags: blocking1.9.0.19?
Flags: blocking1.9.0.19-
Flags: blocking1.8.1.next?
Flags: blocking1.8.1.next-
Is this still an issue ?
Flags: needinfo?(MaierMan)
I cannot tell if "this" is still an issue because I'm not exactly sure what "this" is. But since the reports where about rather old code (in terms of Firefox and DownThemAll!) much of which has changed in the meantime and neither I nor anybody else was able to track this down, let alone reproduce it, lets close it as incomplete. That's at least what I did with the bug in the DownThemAll! trac. I'll reopen or refile in the unlikely event I get recent and more detailed information.
Status: NEW → RESOLVED
Closed: 12 years ago
Flags: needinfo?(MaierMan)
Resolution: --- → INCOMPLETE
You need to log in before you can comment on or make changes to this bug.