Re: unable to handle paging request in worker_thread on apm resume

[Andrew Morton]

"J. Bruce Fields" <bfields@xxxxxxxxxxxx> wrote:
>
> I got the following after an apm resume on a thinkpad X31, with
>  2.6.12-rc1 plus some (hopefully unrelated) NFS patches.  Any ideas?
> 
>  --Bruce Fields
> 
> 
>  Mar 21 18:22:44 puzzle apmd[1815]: Suspending now
>  Mar 21 22:37:36 puzzle kernel: PCI: Setting latency timer of device 0000:00:1d.0 to 64
>  Mar 21 22:37:36 puzzle kernel: PCI: Setting latency timer of device 0000:00:1d.1 to 64
>  Mar 21 22:37:36 puzzle kernel: PCI: Setting latency timer of device 0000:00:1d.2 to 64
>  Mar 21 22:37:39 puzzle kernel: Unable to handle kernel paging request at virtual address d28c9dbc
>  Mar 21 22:37:39 puzzle kernel:  printing eip:
>  Mar 21 22:37:39 puzzle kernel: c01331a6
>  Mar 21 22:37:39 puzzle kernel: *pde = 00048067
>  Mar 21 22:37:39 puzzle kernel: *pte = 128c9000
>  Mar 21 22:37:39 puzzle kernel: Oops: 0002 [#1]
>  Mar 21 22:37:39 puzzle kernel: PREEMPT DEBUG_PAGEALLOC
>  Mar 21 22:37:39 puzzle kernel: Modules linked in:
>  Mar 21 22:37:39 puzzle kernel: CPU:    0
>  Mar 21 22:37:39 puzzle kernel: EIP:    0060:[<c01331a6>]    Not tainted VLI
>  Mar 21 22:37:39 puzzle kernel: EFLAGS: 00010093   (2.6.12-rc1-CITI_NFS4_ALL-1) 
>  Mar 21 22:37:39 puzzle kernel: EIP is at worker_thread+0x1a6/0x400
>  Mar 21 22:37:39 puzzle kernel: eax: d28c9db8   ebx: 00000246   ecx: c0680ea4   edx: dff19f58
>  Mar 21 22:37:39 puzzle kernel: esi: dff19f38   edi: c0680ea0   ebp: dff0ffc4   esp: dff0ff5c
>  Mar 21 22:37:39 puzzle kernel: ds: 007b   es: 007b   ss: 0068
>  Mar 21 22:37:39 puzzle kernel: Process events/0 (pid: 3, threadinfo=dff0f000 task=dff1aa90)
>  Mar 21 22:37:39 puzzle kernel: Stack: dff19f80 00000000 c05643e0 dff0f000 dff19f58 ffffffff ffffffff 00000001 
>  Mar 21 22:37:39 puzzle kernel:        00000000 c0115ec0 00010000 00000000 dff07e44 c0571eff dff0ffc8 00000000 
>  Mar 21 22:37:39 puzzle kernel:        dff1aa90 c0115ec0 00100100 00200200 e1b0835a 00000000 dff1aa90 dff0f000 
>  Mar 21 22:37:39 puzzle kernel: Call Trace:
>  Mar 21 22:37:39 puzzle kernel:  [<c0103a75>] show_stack+0x75/0x90
>  Mar 21 22:37:39 puzzle kernel:  [<c0103bcb>] show_registers+0x11b/0x180
>  Mar 21 22:37:39 puzzle kernel:  [<c0103dfd>] die+0x13d/0x290
>  Mar 21 22:37:39 puzzle kernel:  [<c011375f>] do_page_fault+0x30f/0x68e
>  Mar 21 22:37:39 puzzle kernel:  [<c01035cb>] error_code+0x2b/0x30
>  Mar 21 22:37:39 puzzle kernel:  [<c01399e8>] kthread+0x98/0xa0
>  Mar 21 22:37:39 puzzle kernel:  [<c0100c49>] kernel_thread_helper+0x5/0xc

Seems that there's a `struct work_struct' which is still registered but its
memory has been freed.  It's likely that CONFIG_DEBUG_PAGEALLOC caught this.

Either that, or some module got unloaded without flushing its workqueue.

Are you using any modules which do schedule_work()?

Have you added any code which does schedule_work()?

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/