Re: XFS_WANT_CORRUPTED_GOTO at line 4528 of file fs/xfs/xfs_bmap.c

[Shailendra Tripathi <stripathi@agami.com>]

Eric Sandeen wrote:
> Anyway, the code in question that tripped here is:
>
>                 XFS_WANT_CORRUPTED_GOTO(
>                         XFS_BMAP_SANITY_CHECK(mp, block, level),
>                         error0);
>
> which is checking magic & other things.
>
OK. Thanks.
> The filesystem is corrupted; I'd guess this may be from 4kstacks + xfs 
> + lvm + raid.
>
Yes, the corruption is persisted. I'd like to think on another direction 
as well besides stack corruption isue and I have a reason for the same:

All the inodes which are corrupted are in the same region. The 
corruption starts from 1048704 which is a directory and other inodes 
which are corrupted are:

1048708
1048709
1048711
1048716
1048717
1048718
1048719
1048722
1048723

(gdb) p/x 1048704 % 64
$1 = 0x0
(gdb) p/x 1048708 % 64
$2 = 0x4
(gdb) p 1048723  % 64
$7 = 19
(gdb) p 1048723 - 1048704
$8 = 19

XFS flushes inodes in 8K buffer at a time. If I assume 256 byte inode(I 
couldn't get enough data to verify. I didn't hard to find if this is the 
case as well), so all these inodes are part of one flush buffer(32 
inodes will be flushed in one buffer). So, it could be possible that an 
I/O to the buffer is  lost somewhere. That's why, whole bunch of inodes 
in the same region are corrupted.

> the image he provided has a clean log...
> XFS mounting filesystem loop0
> Ending clean XFS mount for filesystem: loop0
>
> this file is one of the ones in bad shape as he mentioned:
>
> [root@link-07 tmp]# ls -i mnt/lib/rpm/Packages
> 1048708 mnt/lib/rpm/Packages
>
> in my case if I try to read it I get:
>
> attempt to access beyond end of device
> loop0: rw=0, want=1546188226568, limit=4194304
> I/O error in filesystem ("loop0") meta-data dev loop0 block 
> 0x16800000000 ("xfs_trans_read_buf") error 5 buf count 4096
> attempt to access beyond end of device
>
> the inode looks like :
>
> xfs_db> p
> core.magic = 0x494e
> core.mode = 0100644
> core.version = 1
> core.format = 3 (btree)
> core.nlinkv1 = 1
> core.uid = 37
> core.gid = 37
> core.flushiter = 121
> core.atime.sec = Sun Oct 15 17:17:38 2006
> core.atime.nsec = 403299952
> core.mtime.sec = Sun Oct 15 17:17:38 2006
> core.mtime.nsec = 403299952
> core.ctime.sec = Sun Oct 15 17:17:45 2006
> core.ctime.nsec = 983773702
> core.size = 21340160
> core.nblocks = 5212
> core.extsize = 0
> core.nextents = 154
> core.naextents = 1
> core.forkoff = 15
> core.aformat = 2 (extents)
> core.dmevmask = 0
> core.dmstate = 0
> core.newrtbm = 0
> core.prealloc = 0
> core.realtime = 0
> core.immutable = 0
> core.append = 0
> core.sync = 0
> core.noatime = 0
> core.nodump = 0
> core.rtinherit = 0
> core.projinherit = 0
> core.nosymlinks = 0
> core.extsz = 0
> core.extszinherit = 0
> core.nodefrag = 0
> core.gen = 0
> next_unlinked = null
> u.bmbt.level = 1
> u.bmbt.numrecs = 1
> u.bmbt.keys[1] = [startoff] 1:[0]
> u.bmbt.ptrs[1] = 1:1233986491173044224
> a.bmx[0] = [startoff,startblock,blockcount,extentflag] 0:[0,72071,1,0]
>
> repair output:
>
> repair output follows.
>
> [root@link-07 tmp]# xfs_repair -n image
> Phase 1 - find and verify superblock...
> Phase 2 - using internal log
>         - scan filesystem freespace and inode maps...
> would zero unused portion of primary superblock (AG #0)
> would zero unused portion of secondary superblock (AG #1)
> would zero unused portion of secondary superblock (AG #2)
> would zero unused portion of secondary superblock (AG #3)
> would zero unused portion of secondary superblock (AG #4)
> would zero unused portion of secondary superblock (AG #5)
> would zero unused portion of secondary superblock (AG #6)
> would zero unused portion of secondary superblock (AG #7)
>         - found root inode chunk
> Phase 3 - for each AG...
>         - scan (but don't clear) agi unlinked lists...
>         - process known inodes and perform inode discovery...
>         - agno = 0
>         - agno = 1
> bad bmap btree ptr 0x1120002d00000000 in ino 1048708
> bad data fork in inode 1048708
> would have cleared inode 1048708
> bad bmap btree ptr 0x8000000300000000 in ino 1048709
> bad data fork in inode 1048709
> would have cleared inode 1048709
> bad bmap btree ptr 0x57e0001700000000 in ino 1048711
> bad data fork in inode 1048711
> would have cleared inode 1048711
> bad bmap btree ptr 0x9b00000200000000 in ino 1048716
> bad data fork in inode 1048716
> would have cleared inode 1048716
> bad bmap btree ptr 0x9140000100000000 in ino 1048717
> bad data fork in inode 1048717
> would have cleared inode 1048717
> bad bmap btree ptr 0xbf00000100000000 in ino 1048718
> bad data fork in inode 1048718
> would have cleared inode 1048718
> bad bmap btree ptr 0xa840000100000000 in ino 1048719
> bad data fork in inode 1048719
> would have cleared inode 1048719
> bad bmap btree ptr 0x6b00000100000000 in ino 1048722
> bad data fork in inode 1048722
> would have cleared inode 1048722
> bad bmap btree ptr 0x98c0000100000000 in ino 1048723
> bad data fork in inode 1048723
> would have cleared inode 1048723
>         - agno = 2
>         - agno = 3
>         - agno = 4
>         - agno = 5
>         - agno = 6
>         - agno = 7
>         - process newly discovered inodes...
> Phase 4 - check for duplicate blocks...
>         - setting up duplicate extent list...
>         - check for inodes claiming duplicate blocks...
>         - agno = 0
>         - agno = 1
> entry "Packages" at block 0 offset 144 in directory inode 1048704 
> references free inode 1048708
>         would clear inode number in entry at offset 144...
> entry "Providename" at block 0 offset 200 in directory inode 1048704 
> references free inode 1048709
>         would clear inode number in entry at offset 200...
> entry "Basenames" at block 0 offset 312 in directory inode 1048704 
> references free inode 1048711
>         would clear inode number in entry at offset 312...
> entry "Requirename" at block 0 offset 520 in directory inode 1048704 
> references free inode 1048716
>         would clear inode number in entry at offset 520...
> entry "Dirnames" at block 0 offset 568 in directory inode 1048704 
> references free inode 1048717
>         would clear inode number in entry at offset 568...
> entry "Requireversion" at block 0 offset 624 in directory inode 
> 1048704 references free inode 1048718
>         would clear inode number in entry at offset 624...
> entry "Provideversion" at block 0 offset 688 in directory inode 
> 1048704 references free inode 1048719
>         would clear inode number in entry at offset 688...
> entry "Sha1header" at block 0 offset 856 in directory inode 1048704 
> references free inode 1048722
>         would clear inode number in entry at offset 856...
> entry "Filemd5s" at block 0 offset 904 in directory inode 1048704 
> references free inode 1048723
>         would clear inode number in entry at offset 904...
> bad bmap btree ptr 0x1120002d00000000 in ino 1048708
> bad data fork in inode 1048708
> would have cleared inode 1048708
> bad bmap btree ptr 0x8000000300000000 in ino 1048709
> bad data fork in inode 1048709
> would have cleared inode 1048709
> bad bmap btree ptr 0x57e0001700000000 in ino 1048711
> bad data fork in inode 1048711
> would have cleared inode 1048711
> bad bmap btree ptr 0x9b00000200000000 in ino 1048716
> bad data fork in inode 1048716
> would have cleared inode 1048716
> bad bmap btree ptr 0x9140000100000000 in ino 1048717
> bad data fork in inode 1048717
> would have cleared inode 1048717
> bad bmap btree ptr 0xbf00000100000000 in ino 1048718
> bad data fork in inode 1048718
> would have cleared inode 1048718
> bad bmap btree ptr 0xa840000100000000 in ino 1048719
> bad data fork in inode 1048719
> would have cleared inode 1048719
> bad bmap btree ptr 0x6b00000100000000 in ino 1048722
> bad data fork in inode 1048722
> would have cleared inode 1048722
> bad bmap btree ptr 0x98c0000100000000 in ino 1048723
> bad data fork in inode 1048723
> would have cleared inode 1048723
>         - agno = 2
>         - agno = 3
>         - agno = 4
>         - agno = 5
>         - agno = 6
>         - agno = 7
> No modify flag set, skipping phase 5
> Phase 6 - check inode connectivity...
>         - traversing filesystem starting at / ...
> entry "Packages" in directory inode 1048704 points to free inode 
> 1048708, would junk entry
> entry "Providename" in directory inode 1048704 points to free inode 
> 1048709, would junk entry
> entry "Basenames" in directory inode 1048704 points to free inode 
> 1048711, would junk entry
> entry "Requirename" in directory inode 1048704 points to free inode 
> 1048716, would junk entry
> entry "Dirnames" in directory inode 1048704 points to free inode 
> 1048717, would junk entry
> entry "Requireversion" in directory inode 1048704 points to free inode 
> 1048718, would junk entry
> entry "Provideversion" in directory inode 1048704 points to free inode 
> 1048719, would junk entry
> entry "Sha1header" in directory inode 1048704 points to free inode 
> 1048722, would junk entry
> entry "Filemd5s" in directory inode 1048704 points to free inode 
> 1048723, would junk entry
>         - traversal finished ...
>         - traversing all unattached subtrees ...
>         - traversals finished ...
>         - moving disconnected inodes to lost+found ...
> Phase 7 - verify link counts...
> No modify flag set, skipping filesystem flush and exiting.