FS#8454 - [slim] login manager hangs on X crash or Ctrl+Alt+Del

Attached to Project: Arch Linux
Opened by Anonymous Submitter - Monday, 29 October 2007, 18:27 GMT
Last edited by Gaetan Bisson (vesath) - Wednesday, 26 January 2011, 15:25 GMT
Task Type Bug Report
Category Packages: Extra
Status Closed
Assigned To Gaetan Bisson (vesath)
Architecture All
Severity Medium
Priority Normal
Reported Version 2007.08-2
Due in Version Undecided
Due Date Undecided
Percent Complete 100%
Votes 3
Private No

Details

Description:

For about a month now SLiM has been hanging on a Xorg crash or when Ctrl+Alt+Del is pressed, kicking back to VC/1 and staying there for up to 1 minute. Reverting packages has no effect which leads me to believe it may be due to one of its dependencies. This happens regardless of whether it was started via /etc/rc.conf or /etc/inittab.

Xorg is running the lastest stable fglrx package if that helps.

More info:

http://bbs.archlinux.org/viewtopic.php?id=37765
This task depends upon

Closed by  Gaetan Bisson (vesath)
Wednesday, 26 January 2011, 15:25 GMT
Reason for closing:  No response
Comment by Aaron Griffin (phrakture) - Tuesday, 30 October 2007, 22:58 GMT
Until I get a chance, could you possibly check the rc.d daemon as well? It uses the -d option to run as a daemon.

Comment by Alois Nespor (anespor) - Sunday, 27 January 2008, 13:21 GMT
confirm, on x86_64 - I use as a deamon, cca 10 min after crash slimd loadin CPU on 100% (both core)
Comment by Gavin Bisesi (Daenyth) - Friday, 25 April 2008, 03:33 GMT
Does this issue continue in the most recent version?
Comment by Alois Nespor (anespor) - Tuesday, 10 June 2008, 11:38 GMT
I tried and could not be repeated, probably solved
Comment by Piotr Krawiec (pkraw) - Tuesday, 24 June 2008, 06:45 GMT
  • Field changed: Percent Complete (100% → 0%)
Slim started in daemon mode, hangs when trying to stop it and taking 100% of CPU. Arch x86_64 here, so it seems I experience exactly the same problem as anespor's used to.
Comment by Piotr Krawiec (pkraw) - Tuesday, 24 June 2008, 07:19 GMT
BTW, my research about this bug is described here: http://bbs.archlinux.org/viewtopic.php?id=50360.
Comment by Greg (dolby) - Monday, 07 July 2008, 19:41 GMT
Did you contact the developers of the application?
Comment by MzE2OWM2 (warriant) - Friday, 31 October 2008, 13:43 GMT
Look at slim-pam and slim-cursor in AUR. I've included a patch from BerliOS, which fixes it probably.
Note: I don't use it as a daemon but I run it from inittab and it works perfectly, at least killall -KILL slim (with the patch).
Comment by Aaron Griffin (phrakture) - Thursday, 20 November 2008, 20:04 GMT
Ping? Alexander, are you interested in patching this, or should we wait for an upstream release?
Comment by name withheld (Gullible Jones) - Saturday, 22 November 2008, 20:55 GMT
The new upstream release (1.3.1) is out.
Comment by Alexander Fehr (pizzapunk) - Saturday, 22 November 2008, 22:28 GMT
I will update to the latest release. I think the patch is still needed.
Comment by Damjan Georgievski (damjan) - Thursday, 15 April 2010, 06:36 GMT
This bug still appears even with 1.3.1-7 (the latest package as of now).

I maybe have a working patch, if someone would want to test it (works for me).
Comment by Damjan Georgievski (damjan) - Thursday, 15 April 2010, 13:47 GMT
Hi all,
I've found that if I replace a single *vfork* call with a plain fork in app.cpp, slim starts to work ok in the case of killing X (ctrl-alt-backspace).

The vfork manpage mentions potential problems with signals so that might be the reason for the issue, and it also mentions that there's no reason whatsoever to use vfork on a modern Linux system.

I'm attaching the 1-char patch now, and if anyone can confirm it work, we could probably move it to the package?

   slim-vfork.patch (0.3 KiB)
Comment by Thomas Dziedzic (tomd123) - Saturday, 05 June 2010, 16:43 GMT
status of patch and bug?
Comment by Andrea Scarpino (BaSh) - Sunday, 04 July 2010, 23:37 GMT
orphan team
Comment by Gaetan Bisson (vesath) - Monday, 15 November 2010, 23:21 GMT
It works for me with slim-1.3.2-1 from [extra], so:
- If slim-1.3.2-1 still doesn't work for you, please confirm whether http://arch.vesath.org/all/slim-1.3.2-2-x86_64.pkg.tar.xz does.
- Otherwise (or if I don't hear back from you) I'll close this bug in a week or so.
Comment by Damjan Georgievski (damjan) - Tuesday, 16 November 2010, 12:17 GMT
I'm on Arch 32 bit

also the problem happens intermittently
Comment by Gaetan Bisson (vesath) - Tuesday, 16 November 2010, 12:24 GMT
Then could you test http://arch.vesath.org/all/slim-1.3.2-2-i686.pkg.tar.xz ?
Comment by Damjan Georgievski (damjan) - Tuesday, 16 November 2010, 13:05 GMT
We the above package installed still the same problem.

After I press Ctrl-Alt-Del, slim seems to get stuck.
This is the output of "ps axf"

21114 ? Ss 0:00 /usr/bin/slim -d
21117 ? Zs 0:00 \_ [X] <defunct>
Comment by Gaetan Bisson (vesath) - Tuesday, 16 November 2010, 13:36 GMT
Hum. That is too bad because this package featured the vfork==>fork patch...

Can you remind me of how are you running slim? Via inittab, right?
Comment by Gaetan Bisson (vesath) - Monday, 24 January 2011, 13:00 GMT
Since this problem only seems to affect certain people and we have no clear diagnosis or deterministic testcase, I will close this bug in a few days unless somebody who is affected provides more info on it.
Anyway, as this report was created three years ago, it is likely that the issues some people experience nowadays is not even the same as the one originally reported.

Loading...