[Top] [All Lists]

kernel panic with XFS ACL inheritance on a FULL file system

To: acl-devel-owner@xxxxxxxxxxx, "'linux-xfs@xxxxxxxxxxx'" <linux-xfs@xxxxxxxxxxx>
Subject: kernel panic with XFS ACL inheritance on a FULL file system
From: Srikumar Subramanian <SrikumarS@xxxxxxx>
Date: Tue, 18 Oct 2005 03:38:54 -0400
Cc: ag@xxxxxxxxxxx, Srikumar Subramanian <SrikumarS@xxxxxxx>, Prabhakar Krishnan <kpkar@xxxxxxx>
Sender: linux-xfs-bounce@xxxxxxxxxxx
Hi All,

We are testing XFS with ACL and got a issue when file system is full.

I create a base directory (base/) with 10 access ACL and 10 default ACL. The
file system size is 2 GB. Under base/ directory, i have create lot of
subdir, sub-subdirs, and files (each 1KB size) to fill the filesystem.

FC3; XFS 1.3; kernel 2.6.9-1.667smp

During the test, there is no problem in inheriting acl from parent

But when the XFS file system goes out of space, it stops with a panic. 
It is reproducible consistently. It happens both in UP and SMP kernel.

kernel stack trace:
Unable to handle kernel NULL pointer dereference at virtual address 00000004
printing eip:
*pde = 00004001
Oops: 0002 [#1]
Modules linked in: tbiscsi(U) tbblock(U) xfs i2c_i801 cache(U) dvm(U) sg st
osst sr_mod md5 ipv6 parport_pc lp parport autofs4 i2c_dev i2c_core sunrpc
iptable_filter ip_tables dm_mod button battery ac uhci_hcd ehci_hcd
hw_random e1000 floppy ext3 jbd bcraid aic79xx sd_mod scsi_mod
CPU:    0
EIP:    0060:[<8560dbbc>]    Tainted: PF  VLI
EFLAGS: 00010286   (2.6.9-1.667smp) 
EIP is at xfs_ail_insert+0x83/0x8e [xfs]
eax: 00000000   ebx: 76281b40   ecx: 70e244b0   edx: 00000000
esi: fffffc19   edi: ffffffff   ebp: 7ebd201c   esp: 79f46e58
ds: 007b   es: 007b   ss: 0068
Process xfslogd/0 (pid: 3682, threadinfo=79f46000 task=7995ce90)
Stack: 00008250 00000002 00000000 00000000 7ebd2014 7ebd201c 7ebd2000
       8560d9dd 00000000 00008250 00000002 00000002 70e244b0 00008250
       8560d675 00008250 00000002 00000000 00000001 7ebd2000 73b48ad0
Call Trace:
 [<8560d9dd>] xfs_trans_update_ail+0x63/0x9e [xfs]
 [<8560d675>] xfs_trans_chunk_committed+0x141/0x188 [xfs]
 [<8560d4b3>] xfs_trans_committed+0x38/0xb9 [xfs]
 [<85602a68>] xlog_state_do_callback+0x16f/0x24e [xfs]
 [<8560188f>] xlog_iodone+0x7e/0x9b [xfs]
 [<856193fd>] pagebuf_iodone_work+0xf/0x2f [xfs]
 [<0212cd9f>] worker_thread+0x168/0x1d5
 [<856193ee>] pagebuf_iodone_work+0x0/0x2f [xfs]
 [<0211cbc5>] default_wake_function+0x0/0xc
 [<0211cbc5>] default_wake_function+0x0/0xc
 [<0212cc37>] worker_thread+0x0/0x1d5
 [<021300dd>] kthread+0x73/0x9b
 [<0213006a>] kthread+0x0/0x9b
 [<021041f1>] kernel_thread_helper+0x5/0xb
Code: 74 1e 83 cf ff 39 c2 be 19 fc ff ff 72 07 be e7 03 00 00 31 ff 85 ff
7c 07 7f ae 83 fe 00 77 a9 89 59 04 8b 03 89 01 89 0b 8b 01 <89> 48 04 83 c4
10 5b 5e 5f 5d c3 89 d0 8b 0a 8b 52 04 89 51 04

The same test went good without any panic in XFS 1.2 (Kernel 2.4.20)

Any help or kernel patch is much appreciated.

Please CC me in your reply.

Thanks & Regards,
Srikumar Subramanian

PS: Hi Andreas, we didnt get any response from the mailing list, I hope you
will respond. Thanks.

<Prev in Thread] Current Thread [Next in Thread>