Bug 14286

Summary: kernel BUG at fs/ext4/extents.c:2863
Product: File System Reporter: R.Nageswara Sastry (rnsastry)
Component: ext4Assignee: Theodore Tso (tytso)
Severity: normal CC: sandeen
Priority: P1    
Hardware: All   
OS: Linux   
Kernel Version: 2.6.32-rc1, 2.6.31 Tree: Mainline
Regression: No
Attachments: fsfuzzer image to recreate the kernel stack trace in tar gzipped format
fsfuzzer ext4 base image
Patch to fix this problem

Description R.Nageswara Sastry 2009-10-01 05:40:59 UTC
While working with fsfuzz encountered the following kernel stack traces.

Environment: 2.6.32-rc1 and 2.6.31 (kernel BUG at fs/ext4/extents.c:2833)
Architecture: s390

------------[ cut here ]------------
kernel BUG at fs/ext4/extents.c:2863!
illegal operation: 0001 [#1] SMP
Modules linked in: cbc md5 aes_s390 aes_generic ecb ecryptfs ext4 jbd2 crc16 autofs4 lockd sunrpc ipv6 loop qeth_l2 qeth qdio vmur ccwgroup dm_round_robin dm_multipath scsi_dh sd_mod scsi_mod multipath dm_snapshot dm_zero dm_mirror dm_region_hash dm_log dm_mod dasd_fba_mod dasd_eckd_mod dasd_mod ext3 jbd
CPU: 0 Not tainted 2.6.32-rc2 #1
Process fstest (pid: 5755, task: 0000000024a48038, ksp: 0000000011afbe20)
Krnl PSW : 0704000180000000 000003e00200f3f4 (ext4_ext_get_blocks+0x240/0x10cc [ext4])
           R:0 T:1 IO:1 EX:1 Key:0 M:1 W:0 P:0 AS:0 CC:0 PM:0 EA:3
Krnl GPRS: 0000000000000000 0000000000000000 0000000000002400 0000000000000000
           000000001416c000 000000000c8da000 000000002b539370 0000000000000000
           000000002b5390f0 000000002b5390f0 0000000011afba90 000000002b539040
           000003e001fe3000 000003e002023198 000003e00200f388 0000000011afb6c8
Krnl Code: 000003e00200f3e8: a71100ff           tmll    %r1,255
           000003e00200f3ec: a7740006           brc     7,3e00200f3f8
           000003e00200f3f0: a7f40001           brc     15,3e00200f3f2
          >000003e00200f3f4: a7f40000           brc     15,3e00200f3f4
           000003e00200f3f8: e3b040100004       lg      %r11,16(%r4)
           000003e00200f3fe: b90200bb           ltgr    %r11,%r11
           000003e00200f402: a78404ea           brc     8,3e00200fdd6
           000003e00200f406: 5810b000           l       %r1,0(%r11)
Call Trace:
([<000003e00200f388>] ext4_ext_get_blocks+0x1d4/0x10cc [ext4])
 [<000003e001fed31e>] ext4_get_blocks+0xba/0x3c0 [ext4]
 [<000003e001fee9d8>] ext4_get_block+0xcc/0x114 [ext4]
 [<00000000001449ba>] do_mpage_readpage+0x1fa/0x688
 [<0000000000144f7a>] mpage_readpages+0xae/0x100
 [<00000000000d49f8>] __do_page_cache_readahead+0x160/0x1f4
 [<00000000000d4acc>] ra_submit+0x40/0x54
 [<00000000000d4f94>] page_cache_sync_readahead+0x40/0x50
 [<00000000000cc334>] generic_file_aio_read+0x284/0x6a4
 [<000000000010ed34>] do_sync_read+0xd0/0x118
 [<000000000010fad0>] vfs_read+0xa8/0x174
 [<000000000010fc92>] SyS_read+0x56/0x84
 [<0000000000027f5a>] sysc_tracego+0xe/0x14
 [<0000004e53f12cc4>] 0x4e53f12cc4
Last Breaking-Event-Address:
 [<000003e00200f3f0>] ext4_ext_get_blocks+0x23c/0x10cc [ext4]

---[ end trace 7a3a53bbf5dda9be ]---

Related code:

   2858         /*
   2859          * consistent leaf must not be empty;
   2860          * this situation is possible, though, _during_ tree modification;
   2861          * this is why assert can't be put in ext4_ext_find_extent()
   2862          */
   2863         BUG_ON(path[depth].p_ext == NULL && depth != 0);
   2864         eh = path[depth].p_hdr;
Comment 1 R.Nageswara Sastry 2009-10-01 05:47:31 UTC
Created attachment 23213 [details]
fsfuzzer image to recreate the kernel stack trace in tar gzipped format
Comment 2 R.Nageswara Sastry 2009-10-01 05:50:50 UTC
Created attachment 23214 [details]
fsfuzzer ext4 base image
Comment 3 R.Nageswara Sastry 2009-10-01 05:58:01 UTC
unzip the above file(s) 

And download the fsfuzzer from URL 

Untar it
cd fsfuzzer-0.7.3

cp ext4.364.img to fsfuzzer-0.7.3/cfs
cp ext4.base to fsfuzzer-0.7.3/fs
and run the file named run_test in dir fsfuzzer-0.7.3
./run_test ext4 364

Check the dmesg.

Comment 4 R.Nageswara Sastry 2009-10-16 12:30:09 UTC
Any updates!!
Comment 5 Theodore Tso 2009-12-14 15:39:32 UTC
Created attachment 24181 [details]
Patch to fix this problem

Surbhi Palande has supplied the following patch to address this issue