Re: killing a set of user processes causes system reboot

On 12/13/06, John David Anglin <[email protected]> wrote:
> After a GCC build and check, I sometimes see some sleeping processes
> that didn't terminate:
>
> dave@hiauly6:~/gnu/gcc-4.2/objdir$ ps -ef|grep dave
> dave      2496     1  0 12:04 ?        00:00:00 /home/dave/gnu/gcc-4.2/objdir/hppa-linux/libjava/testsuite/Process_3.exe
> dave      2497     1  0 12:04 ?        00:00:00 /home/dave/gnu/gcc-4.2/objdir/hppa-linux/libjava/testsuite/Process_3.exe
> dave      2498     1 39 12:04 ?        01:59:20 /home/dave/gnu/gcc-4.2/objdir/hppa-linux/libjava/testsuite/Process_3.exe
> dave      2873     1  0 12:09 ?        00:00:00 /home/dave/gnu/gcc-4.2/objdir/hppa-linux/libjava/testsuite/Process_3.exe
> dave      2874     1  0 12:09 ?        00:00:00 /home/dave/gnu/gcc-4.2/objdir/hppa-linux/libjava/testsuite/Process_3.exe
> dave      2875     1 39 12:09 ?        01:56:30 /home/dave/gnu/gcc-4.2/objdir/hppa-linux/libjava/testsuite/Process_3.exe
>
> If I kill them, the system reboots with no messages as to why:
>
> root@hiauly6:/home/dave/gnu/gcc-4.2/objdir# kill -9 2496 2497 2498 2873 2874 2875
>
> I had strace attached to one of the processes but it gave no info:
> root@hiauly6:/home/dave# strace -p 2875
> Process 2875 attached - interrupt to quit
>
> I've seen this with 32-bit c3k kernel versions 2.6.18-rc7-pa1
> and 2.6.19-g211c7899-dirty.
>
> Thoughts?

Yes. We talked about this before. You have to set up a serial console
to your C3K, and issue a SYSRQ-T to show where the process is stuck in
the kernel.

FYI I've been running gcc builds on my c3k to reproduce this, but I
haven't seen anything like this.

Sorry to everyone who thinks I fell off the surface of the earth...
it's true. I'm back on my thesis writing schedule, which means hours
in the evenings are gone :(

I have a defense date in March! :)

c.