Possible XFS Corruption

Hi list,

I'm running a Gentoo-patched 2.6.7 kernel, and am experiencing possible XFS corruption on one of my partitions. I've included a sample of the logs I've managed to see here:

Filesystem "hda1": XFS internal error xfs_da_do_buf(2) at line 2273 of file fs/xfs/xfs_da_btree.c. Caller 0xc02420ec

[<c0241b2f>] xfs_da_do_buf+0x41f/0x960
[<c02420ec>] xfs_da_read_buf+0x3c/0x40
[<c02420ec>] xfs_da_read_buf+0x3c/0x40
[<c02420ec>] xfs_da_read_buf+0x3c/0x40
[<c0249611>] xfs_dir2_leaf_lookup_int+0x41/0x280
[<c0249611>] xfs_dir2_leaf_lookup_int+0x41/0x280
[<c0132a80>] do_generic_mapping_read+0x180/0x3a0
[<c02340e9>] xfs_bmap_last_offset+0xa9/0x120
[<c0249540>] xfs_dir2_leaf_lookup+0x20/0xb0
[<c02441f2>] xfs_dir2_lookup+0x112/0x130
[<c0132ca0>] file_read_actor+0x0/0xe0
[<c0283687>] xfs_read+0x197/0x250
[<c02736dc>] xfs_dir_lookup_int+0x2c/0xe0
[<c02781e7>] xfs_lookup+0x37/0x70
[<c0282aaa>] linvfs_lookup+0x4a/0x90
[<c015700f>] real_lookup+0xaf/0xd0
[<c0157218>] do_lookup+0x68/0x80
[<c0157628>] link_path_walk+0x3f8/0x7d0
[<c0157c06>] path_lookup+0x66/0x110
[<c015489f>] open_exec+0x1f/0xe0
[<c015499c>] kernel_read+0x3c/0x50
[<c016f8cf>] load_elf_binary+0xaef/0xb70
[<c011bd3d>] mm_init+0x8d/0xc0
[<c0135d52>] buffered_rmqueue+0xd2/0x180
[<c0135ead>] __alloc_pages+0xad/0x340
[<c013605b>] __alloc_pages+0x25b/0x340
[<c0291628>] __copy_from_user_ll+0x58/0x60
[<c0155411>] search_binary_handler+0x51/0x1a0
[<c015570b>] do_execve+0x1ab/0x230
[<c01049de>] sys_execve+0x2e/0x60
[<c0105d07>] syscall_call+0x7/0xb

This is not the first time I have seen this error, and am wondering if it is something anyone has experienced regularly. Any help in diagnosing this problem is greatly appreciated.

Please CC me any replies, as I'm not subscribed to the list at this address. Thanks in advance!

Callan

Is it reproducible with an unpatched kernel.org kernel?

thanks.

Nathan

Hi Nathan,

Unfortunately, I am unable to test this with a vanilla kernel. However, looking through the Gentoo patches, they did not touch any of the XFS code in a vanilla 2.6.7 kernel.

Is there any other way to diagnose this?

Thank you,

Callan

Oh?

I would be surprised if they had. A more likely source of problems would be changes in the VM subsystem (XFS metadata buffers are cached in the page cache).

The failure you see is XFS reporting corruption in a directory btree buffer which didn't have an appropriate magic number at its start when read in from disk. There's thousands of potential reasons why that may have happened; more often than not these days its an error thats occured outside of XFS though, and XFS is passing on the bad news.

If you can find a reproducible test case, you're half way there. If you can find a reproducible test case on a kernel.org kernel, you're 95% of the way there, cos then we can more easily help. ;)

cheers.

Nathan