By default, Bugzilla does not search the list of RESOLVED bugs.
You can force it to do so by putting the upper-case word ALL in front of your search query, e.g.: ALL tdelibs
We recommend searching for bugs this way, as you may discover that your bug has already been resolved and fixed in a later release.
Bug 1275 - SAK in KDM breaks reboot/poweroff: init: no more processes left in this runlevel
Summary: SAK in KDM breaks reboot/poweroff: init: no more processes left in this runlevel
Status: RESOLVED FIXED
Alias: None
Product: TDE
Classification: Unclassified
Component: tdebase (show other bugs)
Version: 3.5.13 [Trinity]
Hardware: Other Linux
: P5 blocker
Assignee: Timothy Pearson
URL:
Depends on:
Blocks:
 
Reported: 2012-10-17 07:00 CDT by linux
Modified: 2013-04-06 21:02 CDT (History)
3 users (show)

See Also:
Compiler Version:
TDE Version String:
Application Version:
Application Name:

Attachments
strace (93.54 KB, text/plain)
2012-10-18 02:21 CDT, linux 		Details
lspci (1.76 KB, text/plain)
2012-10-18 02:50 CDT, linux 		Details
lsusb (407 bytes, text/plain)
2012-10-18 02:51 CDT, linux 		Details
dmesg (37.02 KB, text/plain)
2012-10-18 02:51 CDT, linux 		Details
View All Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description linux 2012-10-17 07:00:24 CDT 
Upgraded from 3.5.13 to 3.5.13.1 (in Debian Squeeze, 2.6.32-5-686 kernel) and now the system fails to reboot or poweroff. Trinity ends and then the system hangs with "init: no more processes left in this runlevel" message.

Noticed that when the kdm-trinity is stopped manually, two tsak processes remain running. If I kill them, it kills all other processes too! That's why init complains.

Disabling SAK in KDM configuration worked-around the problem. Now SAK can't even be enabled - it says "Secure Attention Key support is not available on your system. Please check for the presence of evdev and uinput". I see tsak segfault in dmesg.

The weird thing is that only one system out of 4 (all with Debian and 3.5.13.1) is affected.
Comment 1 Slávek Banko 2012-10-17 10:55:26 CDT 
Interesting finding. I was with kernel 2.6.32 always without SAK and I assumed that simply does this kernel not have support uinput. But now I find, that uinput is builded as module and is not loaded at boot. When I add uinpout into /etc/modules, SAK is working.

However, problem with "init: no more processes left in this runlevel" I still cannot reproduce.
Comment 2 linux 2012-10-18 02:07:53 CDT 
Seems that tsak does not like something on this system (Asus Eee PC 701):

# /opt/trinity/bin/tsak
Found 119 keyboard(s)
Reading from keyboard: (AT Translated Set 2 keyboard)
Reading from keyboard: (ACPI Virtual Keyboard Device)
Reading from keyboard: (ACPI Virtual Keyboard Device)
Reading from keyboard: (ACPI Virtual Keyboard Device)
...
Reading from keyboard: (ACPI Virtual Keyboard Device)
AT Translated Set 2 keyboard+tsak
Device created.
ACPI Virtual Keyboard Device+tsak
Device created.
Failed to grab exclusive input device lockSegmentation fault
Comment 3 linux 2012-10-18 02:21:04 CDT 
Created attachment 906 [details]
strace
Comment 4 Timothy Pearson 2012-10-18 02:27:17 CDT 
(In reply to comment #2)
> Seems that tsak does not like something on this system (Asus Eee PC 701):
> 
> # /opt/trinity/bin/tsak
> Found 119 keyboard(s)
> Reading from keyboard: (AT Translated Set 2 keyboard)
> Reading from keyboard: (ACPI Virtual Keyboard Device)
<snip>

That looks very, very wrong.  119 keyboards?

Can you please post the output of lspci and lsusb?

Thanks!

Tim
Comment 5 Slávek Banko 2012-10-18 02:34:42 CDT 
Please, you have /dev/uinput? If not (which corresponds to the 2.6.32 kernel), first try: modprobe uinput, or add uinput into /etc/modules.
Comment 6 linux 2012-10-18 02:50:18 CDT 
Created attachment 907 [details]
lspci
Comment 7 linux 2012-10-18 02:51:30 CDT 
Created attachment 908 [details]
lsusb
Comment 8 linux 2012-10-18 02:51:57 CDT 
Created attachment 909 [details]
dmesg
Comment 9 linux 2012-10-18 03:01:18 CDT 
I have /dev/uinput, the uinput module is loaded automatically.

Seems like a bug in tsak - when I delete /dev/input/event10 (which is the ACPI Virtual Keyboard Device), it shows only 1 keyboard and does not crash.
Comment 10 linux 2012-10-18 03:03:50 CDT 
Oh and that device is created by acpi-fakekey daemon. And it' also loading the uinput module.
Comment 11 Timothy Pearson 2012-10-18 12:12:06 CDT 
(In reply to comment #9)
> I have /dev/uinput, the uinput module is loaded automatically.
> 
> Seems like a bug in tsak - when I delete /dev/input/event10 (which is the ACPI
> Virtual Keyboard Device), it shows only 1 keyboard and does not crash.

Yes, that is what I thought.  tsak needs to ignore all fake keyboard devices.

Raising priority to BLOCKER.

Tim
Comment 12 Timothy Pearson 2013-04-06 21:02:13 CDT 
(In reply to comment #11)
> (In reply to comment #9)
> > I have /dev/uinput, the uinput module is loaded automatically.
> > 
> > Seems like a bug in tsak - when I delete /dev/input/event10 (which is the ACPI
> > Virtual Keyboard Device), it shows only 1 keyboard and does not crash.
> 
> Yes, that is what I thought.  tsak needs to ignore all fake keyboard devices.
> 
> Raising priority to BLOCKER.
> 
> Tim

Fixed in GIT hash 8d2f9ae.

Thanks for reporting!