Re: Oops with 2.6.15-mm1

[JANAK DESAI <janak@us.ibm.com>]

Andrew Morton wrote:

>simon schuler <simon.schuler@gmx.net> wrote:
>  
>
>>I'm getting an oops sometimes (about 70% probability) when I try to 
>>start a wine application.
>>last working kernel version: 2.6.15-rc5-mm1
>>not working: 2.6.15-mm1, 2.6.15-rc5-mm3
>>not tested: 2.6.15-rc5-mm2
>>
>>.config is attached.
>>I don't know if this is a known issue...
>>If you need more information, let me know.
>>
>>Unable to handle kernel NULL pointer dereference at virtual address 00000004
>> printing eip:
>>c0118755
>>*pde = 00000000
>>Oops: 0000 [#1]
>>last sysfs file: /class/vc/vcsa1/dev
>>Modules linked in:
>>CPU:    0
>>EIP:    0060:[<c0118755>]    Not tainted VLI
>>EFLAGS: 00210282   (2.6.15-mm1)
>>EIP is at dup_fd+0x25/0x290
>>eax: eff40c20   ebx: eff40c28   ecx: eff40c20   edx: eff40c58
>>esi: ead0e000   edi: 00000000   ebp: eff7aaa0   esp: ead0fe48
>>ds: 007b   es: 007b   ss: 0068
>>Process wine (pid: 769, threadinfo=ead0e000 task=eb468ab0)
>>Stack: <0>00000000 ebbe8854 00000001 ec1abd0c eff40c20 c015b937 ec1abd0c 
>>00000001
>>       <0>00000000 eb468ab0 ead0e000 c0423ebe eff7aaa0 c0118a1c eb468ab0 
>>ead0fe88
>>       <0>fffffff4 eff40ce0 c0118a73 00000000 eb468ab0 00000060 ead0e000 
>>c01867e4
>>Call Trace:
>> [<c015b937>] vfs_read+0x127/0x190
>> [<c0118a1c>] copy_files+0x5c/0x80
>> [<c0118a73>] unshare_files+0x33/0x70
>> [<c01867e4>] load_elf_binary+0x174/0xe10
>> [<c014102f>] get_page_from_freelist+0x7f/0xd0
>> [<c0259d0f>] copy_from_user+0x3f/0xa0
>> [<c0259d0f>] copy_from_user+0x3f/0xa0
>> [<c0166063>] copy_strings+0x163/0x230
>> [<c0167160>] search_binary_handler+0x50/0x180
>> [<c0167414>] do_execve+0x184/0x220
>> [<c0101c4c>] sys_execve+0x3c/0x80
>> [<c010300b>] sysenter_past_esp+0x54/0x75
>>Code: bc 27 00 00 00 00 55 57 56 53 83 ec 24 8b 44 24 38 8b b8 64 04 00 
>>00 e8 6a ff ff ff 85 c0 89 44 24 10 0f 84 b6 01 00 00 8b 58 04 <8b> 77 
>>04 89 34 24 e8 d0 fe ff ff 89 44 24 18 31 c0 8b 54 24 18
>>
>>    
>>
>
>OK, thanks.  It looks like a problem in the new unsharing code...
>
>
>  
>
Thanks. I am investigating the problem now and will send updated patches
as soon as I fix the problem. I am also writing up detailed 
justification, design
considerations and test plan. I will include it in the patch set as well.

-Janak