On Fri, 2012-12-28 at 11:23 +0100, Mattias Eriksson wrote: > Hi! > > I'm using linux kernel 3.2.32 from backports, > linux-image-3.2.0-0.bpo.4-amd64, with squeeze. > The system is running xen (4.0.1 from squeeze) on a > VIA VB8001-16 motherboard and I get easily triggered kernel oops:es in > domU. Does any particular program or workload trigger this? > All PV guests are running Squeeze with linux 3.2.32 from bpo. > I also have a chroot:ed wheezy install, running under dom0, with > the latest openchrome xorg driver. > > [ 854.826235] alignment check: 0000 [#1] SMP > [ 854.826265] CPU 0 > [ 854.826273] Modules linked in: xenfs ext3 jbd evdev snd_pcm snd_timer > snd soundcore snd_page_alloc via_cputemp hwmon_vid pcspkr ext4 mbcache jbd2 crc16 dm_mod xen_netfront xen_blkfront > [ 854.826354] > [ 854.826365] Pid: 2348, comm: asterisk Not tainted 3.2.0-0.bpo.4-amd64 #1 Debian 3.2.32-1~bpo60+1 [...] > [ 854.826727] Call Trace: > [ 854.826747] [<ffffffff81007555>] ? get_phys_to_machine+0x16/0x58 > [ 854.826768] [<ffffffff8136d4b3>] ? stub_clone+0x13/0x20 > [ 854.826790] [<ffffffff8136d192>] ? system_call_fastpath+0x16/0x1b > [ 854.826808] [<ffffffff8100d722>] ? __switch_to+0x23b/0x2b1 > [ 854.826826] [<ffffffff8100e56d>] ? do_notify_resume+0x25/0x67 > [ 854.827041] [<ffffffff810470f2>] ? schedule_tail+0x24/0x56 > [ 854.827060] [<ffffffff8136d460>] ? int_signal+0x12/0x17 > [ 854.827075] Code: bd 38 01 00 00 e8 78 a7 04 00 85 c0 0f 85 70 02 00 > 00 65 48 8b 04 25 88 c6 00 00 8b 90 88 e0 ff ff c7 80 88 e0 ff ff 00 00 > 00 00 <48> 8b 05 fe e2 67 00 a9 00 00 00 04 74 0a 49 c7 45 08 01 00 00 > [ 854.827253] RIP [<ffffffff8100e19f>] do_signal+0x28c/0x635 > [ 854.827272] RSP <ffff880014d3be38> > [ 854.827299] ---[ end trace e5411f91d54565e6 ]--- Do you see a similar call trace each time? > The system does not freeze, but alignment errors doesn't feel all to > good. x86 processors should not generate the #AC (alignment check) interrupt unless it's specifically enabled (which it never is, because this feature is useless!). Even SSE instructions which require aligned operands generate #GP (general protection fault) and not #AC. > When using this setup with linux 2.6.32-46 > (linux-image-2.6.32-5-xen-amd64) all is fine except the slow > via-velocity driver. > Does anyone have any advice on how to debug this, or other hints? > > I'm fairly capable (I think) of debugging code using gdb in userspace, > but never tried kernel gdb though. I doubt it would help with something as weird as this. Ben. -- Ben Hutchings Never attribute to conspiracy what can adequately be explained by stupidity.
Attachment:
signature.asc
Description: This is a digitally signed message part