Re: [Oops] 2.6.33-rc1: kernel NULL pointer dereference in check_preempt_wakeup

["Américo Wang" <xiyou.wangcong@gmail.com>]

On Wed, Dec 23, 2009 at 10:59 AM, Johannes Hirte
<johannes.hirte@fem.tu-ilmenau.de> wrote:
> Am Mittwoch 23 Dezember 2009 03:09:47 schrieb Américo Wang:
>> On Wed, Dec 23, 2009 at 8:20 AM, Johannes Hirte
>>
>> <johannes.hirte@fem.tu-ilmenau.de> wrote:
>> > With 2.6.33-rc1 I have random hangs. Even SysRq-Key works, only hard
> Aargh, should be: "Even SysRq-Key didn't work"
>
>> > reset. I've had it mostly within X, where I didn't get any messages but
>> > one time on console with the following trace:
>> >
>> > BUG: unable to handle kernel NULL pointer dereference at 0000000000000090
>> > IP [<ffffffff810263f1>1 check_preempt_wakeup+0xae/0x14e
>> > PGD 10efa1067 PUD 11767a067 PMD 0
>> > Oops: 0000 [#1] SMP
>> > last sysfs file: /sys/devices/pci0000:00/0000:00:10.3/tempt1_input
>> > CPU 0
>> > Pid: 20108, comm: ebuild.sh Not tainted 2.6.33-rc1 #1 TYAN Tiger K8W Dual
>> > AMD Opteron, S28755/To Be Filled By O.E.M.
>> > RIP: 0010:[<ffffffff810263f1>] [<ffffffff810263f1>]
>> > check_preempt_wakeup+0xae/0x14e RSP: 0018: ffff88006c0e3e50 EFALGS:
>> > 000010087
>> > RAX: ffff880028292420 RBX: 0000000000000000 RCX: 0000000000000000
>> > RDX: 0000000000000002 RSI: 0000000000000000 RDI: 0000000001200011
>> > RBP: ffff880105a3eae0 R08: 0000000000000000 R09: 0000000000000001
>> > R10: 0000000001200011 R11: ffff88011cdd0000 R12: ffff8800282123c0
>> > R13: 0000000000000001 R14: 0000000000000000 R15: 0000000000004ed0
>> > FS:  00007f767a6aa700(0000) GS:ffff880028200000(0000)
>> > knlGS:0000000000000000 CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
>> > CR2: 0000000000000090 CR§: 00000001061a3000 CR4: 00000000000006f0
>> > DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
>> > DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
>> > Process ebuild.sh (pid: 20108, threadinfo ffff88006c0e2000, task
>> > ffff880105a3eae0) Stack:
>> >  ffff8800202123c0 ffff88009734c740 0000000001200011 ffff88009734c740
>> > <0> 0000000000004ed0 ffffffff8102e329 0000000000000000 0000000000000286
>> > <0> 00007f767a6aa9d0 0000000000000000 0000000000000000 ffffffff8103042d
>> > Call Trace:
>> >  [<ffffffff8102e329>] ?  wake_up_new_task+0x66/0x8f
>> >  [<ffffffff8103042d>] ? do_fork+0x20f/0x273
>> >  [<ffffffff810a5a9s>] ? alloc_fd+0x69/10a
>> >  [<ffffffff8103a953>] ? sigprocmask+0x9f/0xc2
>> >  [<ffffffff810020d3>] ? stub_clone+0x13/0x20
>> >  [<ffffffff81001e6b>] ? system_call_fastpath+0x16/0x1b
>> > Code: 8b 89 88 00 00 00 ff c0 48 85 c9 75 f2 eb 09 40 0b 9b 88 00 00 00
>> > ff ca 39 c2 7f f3 eb 09 4d 8b b6 88 00 00 00 ff c8 39 d0 7f f3 <49> 8b 86
>> > 90 00 00 00 48 39 83 90 00 00 00 75 74 4d 85 f6 75 04
>> > RIP  [<ffffffff810263f1>] check_preempt_wakeup+0xae/0x14e
>> >  RSP <ffff88006c0e3e50>
>> > CR2: 0000000000000090
>> > ---[ end trace f6bafb05b72ae358 ]---
>>
>> Hmm, seems like the one Eric reported:
>> http://lkml.org/lkml/2009/12/21/272
>>
>> > Since the system didn't respond anymore, I had captured this with camera
>> > and copied by hand. I hope I haven't made any mistakes when copying. The
>> > picture wasn't very good.
>>
>> Wow, hard work. :)
>>
>> Can you try Peter's patch?
>>
>> http://lkml.org/lkml/2009/12/22/124
>>
>> Thanks!
>
> I'll try it. The problem is, I haven't a workload, I can reproduce the hang
> reliably. :(

Amazing...

If that patch doesn't work, please try run addr2line, something like:

 addr2line -e vmlinux ffffffff810263f1

Thanks.