As a security precaution, we have turned on the setting "Require API key authentication for API requests" for everyone. If this has broken something, please contact
Last Comment Bug 415158 - Crash on opening a password-protected page
: Crash on opening a password-protected page
: crash
Product: Toolkit
Classification: Components
Component: Password Manager (show other bugs)
: Trunk
: x86 Linux
: -- critical with 1 vote (vote)
: ---
Assigned To: Nobody; OK to take it and work on it
: Matthew N. [:MattN] (PM me if requests are blocking you)
Depends on:
  Show dependency treegraph
Reported: 2008-01-31 13:58 PST by Horst H. von Brand
Modified: 2008-07-31 04:30 PDT (History)
2 users (show)
See Also:
Crash Signature:
QA Whiteboard:
Iteration: ---
Points: ---
Has Regression Range: ---
Has STR: ---

Resulting gecko-bugreport.txt file, zipped (23.69 KB, application/octet-stream)
2008-01-31 14:05 PST, Horst H. von Brand
no flags Details

Description User image Horst H. von Brand 2008-01-31 13:58:17 PST
User-Agent:       Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9b3pre) Gecko/2008012508 Fedora/3.0b3pre-0.beta2.12.nightly20080121.fc9 Minefield/3.0b3pre
Build Identifier: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9b3pre) Gecko/2008012508 Fedora/3.0b3pre-0.beta2.12.nightly20080121.fc9 Minefield/3.0b3pre

Accessing <>, Firefox crashes
and gives a backtrace. I have two saved usernames + passwords on this site, trying to enter the username (either from the pulldown menu or by hand) gives a crash.

Reproducible: Always

Steps to Reproduce:
Comment 1 User image Horst H. von Brand 2008-01-31 14:05:00 PST
Created attachment 300733 [details]
Resulting gecko-bugreport.txt file, zipped

The gecko-bugreport.txt file that results, zipped.
Comment 2 User image Matthias Versen [:Matti] 2008-01-31 17:17:45 PST
We accept only bug reports from official builds.
Please provide a Breakpad id of this crash.
Send the breakpad crash with the latest nightly and look in "about:crashes" for the id.
Comment 3 User image Christopher Aillon (sabbatical, not receiving bugmail) 2008-02-01 07:38:55 PST
(In reply to comment #2)
> We accept only bug reports from official builds.

Bull.  Tell that to anyone who cvs pulls and builds from source on their own then.  Like you know, a bunch of developers.  I'm only doing cvs pulls and builds.  I'm trying to _help_ by providing extra builds on linux of cvs snapshots to people.  Reporting bugs directly here is important.  Please treat this as normal.
Comment 4 User image Matthias Versen [:Matti] 2008-02-01 10:27:10 PST
There are reasons for this. Many vendors are adding additional patches.
How should we fix such bug, the bug reports will just stay unconfirmed in bugzilla for years. Just look at bug bug 415090 for an example with a bug in a SuSe build. Bug reports from the official source are of course valid, you can use only different build options and a different compiler in that case.

Comment 5 User image timeless 2008-02-03 22:53:02 PST
caillon: you're going to need to provide facilities for them to get useful stack traces. we do not have the ability to handle the provided stack trace.

when developers build from cvs (or hg or some other random system), we expect them to provide valid/useful stack traces. if they don't, we yell at them too.

does something prevent you from providing a working crash reporter server?

reporter: offhand, i'd have filed this in password manager. however your steps to reproduce are bad, please read about how to file a bug.

steps should be:
1. create a new profile
2. load
3. log in
4. when the log in succeeds, click the info banner to save the password
5. quit
6. run again (using the profile from 1)
7. load
8. log in using account 2
9. when the log in succeeds, click the info banner to save the password
10. quit
11. run again (using the profile from 1)
12. load
13. enter one character from one of the two usernames (from 3/8)
14. click on the username
15. click Ingresar
16. at some point here, firefox crashes

note: these steps aren't complete, your steps should be, you should indicate each time focus changes from field to field and how, if you click, press tab, press enter, or sneeze, that should be documented. if you press the down arrow, or the up arrow, or backspace, that should be documented. and you really should test with a new profile (see step 1).

we don't stick in a steps to reproduce field for nothing (nor a url field, which you left blank). and steps to reproduce shouldn't lie, you said that you crashed loading, but you then later go on to say that you selected a previously saved login.

we do welcome new contributors, but we ask that you follow the instructions.
Comment 6 User image Justin Dolske [:Dolske] 2008-02-04 01:21:14 PST
The attached bugreport isn't useful, it's a bunch of hex and generic process info, so I'm not sure what to do with it. A breakpad report or stack from a debug build would be far more useful.

Timeless: were you able to recreate the crash with those steps, or is that just an example?
Comment 7 User image timeless 2008-02-04 04:01:37 PST
sorry, that was an incomplete example of how to write /almost/ useful steps. if i had actually reproduced them, i'd have included an incident id or stack trace and confirmed this bug.
Comment 8 User image Andrew Schultz 2008-02-09 23:59:21 PST
Horst, to get a useful stacktrace from the Fedora build, you'll need to install the xulrunner-debuginfo (and probably firefox-debuginfo) packages and grab a stacktrace from gdb:

firefox -g -d gdb
(gdb) run
[make firefox crash]
(gdb) bt

please attach the output of bt to this bug.
Comment 9 User image Andrew Schultz 2008-02-13 23:51:56 PST
Comment #9 From Horst H. von Brand 	on 2008-01-31 21:09 EST 	[reply] 	 

Fixed in firefox-3.0-0.beta2.15.nightly20080130.fc9.i386

Note You need to log in before you can comment on or make changes to this bug.