Skip to contentSkip to conversation

Warning in record_one_backref()

DISCUSSION Pedro Fonseca 3 participants 3 messages Nov 19, 2013 – Nov 28, 2013 Btrfs file system

lore
Share & tools
lore
#0 lore raw

Hi,

In another test, I've encountered a few situations that triggered a warning message in "record_one_backref()". I'm not sure if it's serious but it is probably related to the concurrent defragment operations executed during the test.

Warning dump:

Quoted 21 lines
[  147.558178] ------------[ cut here ]------------
[  147.579146] WARNING: CPU: 1 PID: 2795 at [...]/linux-3.11.1-fs-static/fs/btrfs/inode.c:2206 record_one_backref+0x3cb/0x3e4()
[  147.582925] Modules linked in: loop rtc_cmos pcspkr tpm_tis freq_table mperf i2c_piix4
[  147.589360] CPU: 1 PID: 2795 Comm: btrfs-endio-wri Not tainted 3.11.1 #2
[  147.589731] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2007
[  147.590523]  0000089e c4059d1c c176917a c1a07b61 c4059d34 c102d1c0 c12b8c4b 00000007
[  147.591240]  df3c8f18 c5e14800 c4059d44 c102d1e6 00000009 00000000 c4059db4 c12b8c4b
[  147.595457]  df45f580 00000286 c5e14800 c10a70fe de4c6c00 c4059d97 00000000 00000106
[  147.596156] Call Trace:
[  147.597365]  [<c176917a>] dump_stack+0x41/0x57
[  147.597684]  [<c102d1c0>] warn_slowpath_common+0x5e/0x75
[  147.597984]  [<c12b8c4b>] ? record_one_backref+0x3cb/0x3e4
[  147.598275]  [<c102d1e6>] warn_slowpath_null+0xf/0x13
[  147.602112]  [<c12b8c4b>] record_one_backref+0x3cb/0x3e4
[  147.602392]  [<c10a70fe>] ? __cache_free+0xa2/0xaa
[  147.602667]  [<c12fe908>] iterate_extent_inodes+0x132/0x1b2
[  147.602964]  [<c12feba9>] iterate_inodes_from_logical+0x88/0x97
[  147.603256]  [<c12b8880>] ? record_extent_backrefs+0x8b/0x8b
[  147.603547]  [<c12b8845>] record_extent_backrefs+0x50/0x8b
[  147.603822]  [<c12b8880>] ? record_extent_backrefs+0x8b/0x8b
[  147.604107]  [<c12c1f8e>] btrfs_finish_ordered_io+0x77e/0x876
Quoted 8 lines
[  147.604409]  [<c12c2091>] finish_ordered_fn+0xb/0xd
[  147.604665]  [<c12db161>] worker_loop+0x111/0x3ec
[  147.604929]  [<c104a2c4>] ? default_wake_function+0xb/0xd
[  147.608784]  [<c1040c43>] kthread+0x90/0x95
[  147.609021]  [<c12db050>] ? btrfs_queue_worker+0x1c8/0x1c8
[  147.609311]  [<c176c277>] ret_from_kernel_thread+0x1b/0x28
[  147.609589]  [<c1040bb3>] ? kthread_freezable_should_stop+0x4f/0x4f
[  147.610452] ---[ end trace 626899e11111abbe ]---

The warning occurred at the end of the execution of the following file system operations:

Quoted 12 lines
   CPU: 1 Op: mknod
   CPU: 1 Op: readlink
CPU: 0 Op: mknod
CPU: 0 Op: readlink
CPU: 0 Op: creat
CPU: 0 Op: symlink
CPU: 0 Op: dwrite
CPU: 0 Op: creat
CPU: 0 Op: write
CPU: 0 Op: dwrite
   CPU: 1 Op: creat
   CPU: 1 Op: symlink
Quoted 8 lines
   CPU: 1 Op: dwrite
   CPU: 1 Op: creat
   CPU: 1 Op: write
   CPU: 1 Op: dwrite
CPU: 0 Op: write
CPU: 0 Op: btrfs_ioctl_filesystem_defragment
   CPU: 1 Op: write
   CPU: 1 Op: btrfs_ioctl_filesystem_defragment

Pedro

#2 lore raw
Quoted 11 lines · Emil Karlson
Quoted 7 lines · Pedro Fonseca

Hi,

In another test, I've encountered a few situations that triggered a warning message in "record_one_backref()". I'm not sure if it's serious but it is probably related to the concurrent defragment operations executed during the test.

I think You'll need this patch or later linux-3.11 or linux-3.12

https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=50f1319cb5f7690e4d9de18d1a75ea89296d0e53

The patch has been added to stable kernel 3.11.6.