Bug#535571: kernel unaligned access copy_to_user_state with IPSec

2012-01-03 Thread ja nein
Hi Jonathan, ok, it took a while, but finally I've tested wheezy with latest kernel (linux-image-3.1.0-1-sparc64-smp) and the error's gone. It seems that somewhere to 3.x kernel the error was fixed. Regards, Sebastian On 27.09.2011 22:26, Jonathan Nieder wrote: ja nein wrote: I finally ma

Bug#535571: kernel unaligned access copy_to_user_state with IPSec

2011-09-27 Thread ja nein
Just to make sure, is the kernel you are reading symbols from the same build as the one that produced the message in dmesg? Yes, that's true. I've used the sources of debian via "apt-get source linux-image-2.6.32-5-sparc64-smp" and followed your instructions from there. I just had to activate th

Bug#535571: kernel unaligned access copy_to_user_state with IPSec

2011-09-27 Thread Jonathan Nieder
ja nein wrote: > I finally managed to compile some debug symbols into the kernel. Here we go: > > dmesg: > [3179998.695837] Kernel unaligned access at TPC [1066a958] > copy_to_user_state+0x54/0x9c [xfrm_user] > [3179998.811473] Kernel unaligned access at TPC[1066a958] > copy_to_user_state+0x54/0

Bug#535571: kernel unaligned access copy_to_user_state with IPSec

2011-09-27 Thread ja nein
The kernel modules gdb can use will be in .ko files scattered under the linux/ source dir which you can find with commands like $ find . -name .git -prune -o -name \*.ko -print Jonathan sorry, last gdb entry was incomplete: (gdb) list *(copy_to_user_state+0x9c0) 0x132c is in xfrm_add_sa_expire

Bug#535571: kernel unaligned access copy_to_user_state with IPSec

2011-09-27 Thread ja nein
The kernel modules gdb can use will be in .ko files scattered under the linux/ source dir which you can find with commands like $ find . -name .git -prune -o -name \*.ko -print Jonathan I finally managed to compile some debug symbols into the kernel. Here we go: dmesg: [3179998.695837] Kernel

Bug#535571: kernel unaligned access copy_to_user_state with IPSec

2011-09-27 Thread Jonathan Nieder
ja nein wrote: > Do i have to build them by my own or are they available as dpkg's > somewhere...? > > I've got a sparc machine as spare, so I could go and test, etc. Good catch. It seems that debug kernels are only automatically built for i386 and amd64 (presumably because those are the platfor

Bug#535571: kernel unaligned access copy_to_user_state with IPSec

2011-09-27 Thread Jonathan Nieder
sebastian wrote: > I tried this, but I don't get any info, maybe there are some symbols > missing. Do you have more details for me about addr2line? > > Are you hitting this bug, too? I suspect this happens on big endian systems. No, I haven't hit this, and alas I don't have a sparc to try it on.

Bug#535571: kernel unaligned access copy_to_user_state with IPSec

2011-09-03 Thread Jonathan Nieder
found 535571 linux-2.6/2.6.32-30 quit Hi, ja nein wrote: > [1046073.600266] Kernel unaligned access at TPC[104ea7fc] > copy_to_user_state+0x54/0x9c [xfrm_user] > [1046073.715755] Kernel unaligned access at TPC[104ea7fc] > copy_to_user_state+0x54/0x9c [xfrm_user] > > It happens as soon as there

Bug#535571: kernel unaligned access copy_to_user_state with IPSec

2010-02-20 Thread Moritz Muehlenhoff
On Fri, Jul 03, 2009 at 12:47:23PM +0200, Sebastian wrote: > Package: linux-image-2.6.26-2-sparc64-smp > Version: 2.6.26-17 > Severity: important > > I'm continiously getting these kernel messages while running several IPSec > tunnels: > > [3881706.284682] Kernel unaligned access at TPC[10276320

Bug#535571: kernel unaligned access copy_to_user_state with IPSec

2009-07-03 Thread Sebastian
Package: linux-image-2.6.26-2-sparc64-smp Version: 2.6.26-17 Severity: important I'm continiously getting these kernel messages while running several IPSec tunnels: [3881706.284682] Kernel unaligned access at TPC[10276320] copy_to_user_state+0x50/0x9c [xfrm_user] [3881706.403493] Kernel unalign