Re: killing a set of user processes causes system reboot

On Saturday 16 December 2006 18:36, John David Anglin wrote:
> > FYI I've been running gcc builds on my c3k to reproduce this, but I
> > haven't seen anything like this.
> 
> A little more info.
> 
> I'm seeing this almost every time with GCC 4.2.0 branch using
> vmlinux-2.6.18-rc7-pa1 or vmlinux-2.6.19-g211c7899-dirty.  The
> kernels are 32-bit with the default c3000 config.  It's happened
> in the last two builds and checks running the libjava testsuite.
> The tests timeout, so I would think the dejagnu/tcl framework
> would have tried to kill them when the timeout occured.
> 
> Within a group of processes that are stuck, I can kill all but one
> without causing a system reboot.  The processes appear stuck in the
> kernel.  I don't see any activity with strace.
> 
> I often see this with the Process_3.java test.  For example,
> http://gcc.gnu.org/ml/gcc-testresults/2006-12/msg00637.html

Hi Dave,

does it happen on 64bit kernel as well ?
If not, maybe you could try my (ugly & temporary) kernel patch from http://lists.parisc-linux.org/pipermail/parisc-linux/2006-December/030967.html ?

Helge