Re: 2.6.1-mm1 - OOPs and hangs during modprobe

[Andrew Morton <akpm _at_ xxxxxxxx>]

Valdis.Kletnieks _at_ vt.edu wrote:
>
> Summary: 2.6.1-mm1 gives an OOPs while doing a modprobe.  Subsequent
> references to /proc/modules hang (causing hangs while doing a 'shutdown'
> because of scripts trying to rmmod modules.  lsmod and 'cat /proc/modules'
> hang as well after the oops.
> 
> This one's from trying to load ip_conntrack_ftp:
> 
> Module len 6897 truncated
> Unable to handle kernel NULL pointer dereference at virtual address 00000004
>  printing eip:
> c013040b
> *pde = 00000000
> Oops: 0002 [#1]
> PREEMPT
> CPU:    0
> EIP:    0060:[<c013040b>]    Not tainted VLI
> EFLAGS: 00010002
> EIP is at sys_init_module+0x90/0x225
> eax: 00000004   ebx: 0807a3e8   ecx: c03d5c30   edx: d187b104
> esi: 00000000   edi: cf220000   ebp: cf221fbc   esp: cf221fb0
> ds: 007b   es: 007b   ss: 0068
> Process modprobe (pid: 172, threadinfo=cf220000 task=cf4606c0)
> Stack: 0807a3e8 00000002 080573a0 cf220000 c03548de 0807a3e8 00001af1 0807a088
>        00000002 080573a0 bfffe8c0 00000080 0000007b 0000007b 00000080 ffffd41a
>        00000073 00000287 bfffe8c0 0000007b
> Call Trace:
>  [<c03548de>] sysenter_past_esp+0x43/0x65
> 
> Code: d8 57 3d c0 ff 05 d8 57 3d c0 0f 8e ac 06 00 00 89 c2 e9 9f 01 00 00 fa bf 00 e0 ff ff 21 e7 ff 47 14 8b 15 e8 57 3d c0 8d 40 04 <89> 56 04 89 42 04 a3 e8 57 3d c0 c7 40 04 e8 57 3d c0 fb 8b 47
>  <6>note: modprobe[172] exited with preempt_count 1
> Debug: sleeping function called from invalid context at include/linux/rwsem.h:43
> in_atomic():1, irqs_disabled():0
> Call Trace:
>  [<c011b50f>] __might_sleep+0xa4/0xac
>  [<c011f16f>] do_exit+0xd9/0x389
>  [<c010c5d2>] do_divide_error+0x0/0xad
>  [<c011893d>] do_page_fault+0x35f/0x4b2
>  [<c014576f>] unmap_vm_area+0x2c/0x73
>  [<c0145a81>] vfree+0x25/0x27
>  [<c0130354>] load_module+0x790/0x7b7
>  [<c01185de>] do_page_fault+0x0/0x4b2
>  [<c035535f>] error_code+0x2f/0x38
>  [<c013040b>] sys_init_module+0x90/0x225
>  [<c03548de>] sysenter_past_esp+0x43/0x65
> 
> My original thought was that the ip_conntrack_ftp.ko got corrupted,

My original thought is about Rusty. 

The `check for truncated module' patch is clearly triggering when it should
not be.  It then incorrectly returns "success" from load_module() even
though load_module() failed.

A `patch -R' of

ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.1/2.6.1-mm1/broken-out/check-for-truncated-modules.patch

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