Re: Linux 2.6.31.10

[Dmitry Monakhov]

2010/1/8 Greg KH <gregkh@xxxxxxx>:
> On Thu, Jan 07, 2010 at 09:42:52PM +0100, Thomas Voegtle wrote:
>> On Thu, 7 Jan 2010, Thomas Voegtle wrote:
>>
>>> Done, between 2.6.31.9 and 2.6.31.11
>>>
>>>
>>> c169e13a6662cbf36fc0098508e7aa31b49cbf23 is first bad commit
>>> commit c169e13a6662cbf36fc0098508e7aa31b49cbf23
>>> Author: Dmitry Monakhov <dmonakhov@xxxxxxxxxx>
>>> Date: Â Mon Dec 14 15:21:13 2009 +0300
>>>
>>> Â Âquota: decouple fs reserved space from quota reservation
>>>
>>> Â Âcommit fd8fbfc1709822bd94247c5b2ab15a5f5041e103 upstream.
Ohhh No...
This is totaly my fault. after some cleanups i've forgot to test
ext3+quota scenario :( . The error is trivial, it is too sad what this
error
was found this late.'
calltrace:
->chown
    ->dquot_transfer
       ->inode_get_rsv_space
static qsize_t inode_get_rsv_space(struct inode *inode)
1378 {
1379         qsize_t ret;
<<< This is illigal to call this function from fs which has no quota
reservation support.
so we have to add simple if statment here like this  .
 if (!inode->i_sb->dq_op->get_reserved_space)
     /* quota reservation is no supported */
     return 0;
1380         spin_lock(&inode->i_lock);
1381         ret = *inode_reserved_space(inode);
1382         spin_unlock(&inode->i_lock);
1383         return ret;
1384 }

Will send patch in a minute.
>>>
>>>
>>>
>>> Reverting that patch on top of 2.6.31.11 fixes the problem for me.
>>
>>
>> Funny thing, I have the problem with 2.6.33-rc3, too:
>>
>>
>> ------------[ cut here ]------------
>> kernel BUG at fs/quota/dquot.c:1350!
>> invalid opcode: 0000 [#1]
>> last sysfs file: /sys/devices/pci0000:00/0000:00:1f.5/modalias
>>
>> Pid: 3061, comm: rpm Not tainted 2.6.33-rc3-BIG #73 P3TSSA/P3TSSA
>> EIP: 0060:[<c10c05d8>] EFLAGS: 00010246 CPU: 0
>> EIP is at inode_reserved_space+0x13/0x1b
>> EAX: cf25b5c4 EBX: cf25b5c4 ECX: 00000000 EDX: 00000000
>> ESI: cf25b5c4 EDI: 00000000 EBP: d3536ebc ESP: d3536ebc
>> ÂDS: 007b ES: 007b FS: 0000 GS: 00e0 SS: 0068
>> Process rpm (pid: 3061, ti=d3536000 task=d5c50000 task.ti=d3536000)
>> Stack:
>> Âd3536f1c c10c246a d79a4b00 d3536edc c10a4884 d3536ef4 c10c1fea 00000000
>> <0> 00000000 00000000 00000064 cf25b5c4 cf25b5c4 d701c800 d3536f1c 00000000
>> <0> d766b000 00000000 00000000 d766b0a0 0000b140 cf25b5c4 d3536f58 ffffff86
>> Call Trace:
>> Â[<c10c246a>] ? dquot_transfer+0xf2/0x48b
>> Â[<c10a4884>] ? mntput_no_expire+0x19/0x6c
>> Â[<c10c1fea>] ? dqget+0x210/0x260
>> Â[<c10c05aa>] ? vfs_dq_transfer+0x5f/0x7a
>> Â[<c10a2edd>] ? notify_change+0x177/0x25e
>> Â[<c10930c8>] ? chown_common+0x5e/0x70
>> Â[<c10931fe>] ? sys_lchown+0x3d/0x58
>> Â[<c10212a7>] ? sysenter_do_call+0x12/0x2c
>> Code: c0 0f b6 c0 eb 02 31 c0 5b 5e 5d c3 a8 08 75 b0 eb f4 a8 08 75 c9 eb
>> d5 55 8b 90 90 00 00 00 89 e5 8b 52 24 8b 52 44 85 d2 75 04 <0f> 0b eb fe
>> ff d2 5d c3 55 89 e5 57 89 c7 56 53 83 ec 1c 8b 35
>> EIP: [<c10c05d8>] inode_reserved_space+0x13/0x1b SS:ESP 0068:d3536ebc
>> ---[ end trace 0cbcf5a4693c0845 ]---
>
> Well, that's good, at least things are now consistant :)
>
> Jan, I think this is all yours...
>
> greg k-h
>
--
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/