[2.6.32] task hung

Previous thread: [PATCH] Fix the x86_64 implementation of call_rwsem_wait() by David Howells on Tuesday, May 4, 2010 - 5:42 am. (2 messages)

Next thread: [RFC][PATCH 0/12] KVM, x86, ppc, asm-generic: moving dirty bitmaps to user space by Takuya Yoshikawa on Tuesday, May 4, 2010 - 5:56 am. (26 messages)
From: Folkert van Heusden
Date: Tuesday, May 4, 2010 - 5:49 am

Hi,

I'm copying 6GB of data onto an NBD device using a simple tool which
differs from dd/cat in that it measures how long it takes to write each
block and does a sync after each block.
Now during that run I get:

[45480.665074] INFO: task measurecopy:13780 blocked for more than 120 seconds.
[45480.665086] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
[45480.665090] measurecopy   D 0000000000000000     0 13780  30341 0x00000000
[45480.665102]  ffff8801bf06d4c0 0000000000000082 0000000000000000 ffff8801776d9d04
[45480.665116]  0000000000000001 ffff88000000f640 000000000000f8a0 ffff8801776d9fd8
[45480.665129]  00000000000155c0 00000000000155c0 ffff8801a867d4c0 ffff8801a867d7b8
[45480.665143] Call Trace:
[45480.665153]  [<ffffffff8106beb9>] ? ktime_get_ts+0x68/0xb2
[45480.665159]  [<ffffffff81098f06>] ? delayacct_end+0x74/0x7f
[45480.665168]  [<ffffffff810b3219>] ? sync_page+0x0/0x46
[45480.665175]  [<ffffffff812edf36>] ? io_schedule+0x73/0xb7
[45480.665181]  [<ffffffff810b325a>] ? sync_page+0x41/0x46
[45480.665186]  [<ffffffff812ee436>] ? __wait_on_bit+0x41/0x70
[45480.665192]  [<ffffffff810b33de>] ? wait_on_page_bit+0x6b/0x71
[45480.665199]  [<ffffffff81064a84>] ? wake_bit_function+0x0/0x23
[45480.665206]  [<ffffffff810baf36>] ? pagevec_lookup_tag+0x1a/0x21
[45480.665212]  [<ffffffff810b3bb4>] ? wait_on_page_writeback_range+0x69/0x11b
[45480.665218]  [<ffffffff810b3d26>] ? filemap_write_and_wait_range+0x43/0x52
[45480.665225]  [<ffffffff8110a176>] ? vfs_fsync_range+0x55/0x9e
[45480.665231]  [<ffffffff8110a23e>] ? do_fsync+0x28/0x39
[45480.665236]  [<ffffffff8110a25d>] ? sys_fdatasync+0xe/0x13
[45480.665243]  [<ffffffff81010b42>] ? system_call_fastpath+0x16/0x1b

command line:
./measurecopy -i /dev/VG01/LVHOME -o /dev/nbd0 -s -b 1048576

http://www.vanheusden.com/measurecopy/


Folkert van Heusden

-- 
Feeling generous? -> http://www.vanheusden.com/wishlist.php
----------------------------------------------------------------------
Phone: ...
From: Andrew Morton
Date: Thursday, May 6, 2010 - 3:18 pm

On Tue, 4 May 2010 14:49:50 +0200

end_page_writeback() never got run against a page.  My guess would be
that NBD got its knickers in a knot.
--

Previous thread: [PATCH] Fix the x86_64 implementation of call_rwsem_wait() by David Howells on Tuesday, May 4, 2010 - 5:42 am. (2 messages)

Next thread: [RFC][PATCH 0/12] KVM, x86, ppc, asm-generic: moving dirty bitmaps to user space by Takuya Yoshikawa on Tuesday, May 4, 2010 - 5:56 am. (26 messages)