XFS_WANT_CORRUPTED_GOTO at line 1545 of file fs/xfs/xfs_alloc.c

[Mark Tinguely]

On 04/20/15 23:10, beebol wrote:
> SOS！！！
> This is what causes the problem, whether it can be reproduced?
> How to fix this problem?
>
> Looking forward to your reply.
> information:
> #cat /etc/redhat-release
> Red Hat Enterprise Linux Server release 5.3 (Tikanga)
>
>   #uname -a
> Linux 1046_qd_119_cnc 2.6.18-194.el5 #1 SMP Tue Mar 16 21:52:39 EDT 2010
> x86_64 x86_64 x86_64 GNU/Linux
>
> # df -h
> Filesystem            Size  Used Avail Use% Mounted on
> /dev/sda7             5.9G  712M  4.9G  13% /
> /dev/sda3             3.9G  825M  2.9G  22% /var
> /dev/sda2             5.9G  2.2G  3.4G  39% /usr
> /dev/sda1             122M   18M   99M  15% /boot
> tmpfs                  16G     0   16G   0% /dev/shm
> /dev/sda8             1.7T  512G  1.2T  32% /home
>
> #fstab
> LABEL=/home             /home                   xfs
> defaults,noatime,nodiratime        1 2
>
> install packages:
> xfsprogs-2.9.4-4.el5.x86_64.rpm
> xfsprogs-devel-2.9.4-4.el5.x86_64.rpm
> kernel-module-xfs-2.6.18-128.el5-0.4-4.slc5.x86_64.rpm
>
> /var/log/message:
>
> Apr 20 12:07:17 1046_qd_119 kernel: XFS internal error
> XFS_WANT_CORRUPTED_GOTO at line 1545 of file fs/xfs/xfs_alloc.c.  Caller
> 0xffffffff882bc961

Somehow blocks are in the free list and are allocated at the same time. 
The corruption can happen long before it is tripped over in the free. A 
form of this problem has been around for a long time and has even popped 
in more recent community XFS in the past year.

If you want to make a metadata dump, I will look at it. I bet there are 
other similarly free/allocated or duplicately allocated blocks, but 
won't show how they got into that condition.

You will have to do an "xfs_repair -L". Use a more recent xfs_repair to 
get all of the problems resolved (like zeroed startblocks).

--Mark Tinguely.