2007-08-11 02:40:29

by Russ Dill

Subject: Process stuck in md_wakeup_thread

On 2.6.22 from debian (stock), I have a process (dpkg) stuck with the following
calltrace:

SysRq : Show Blocked State

free sibling
task PC stack pid father child younger older
dpkg D 00000003 0 26040 20765 (NOTLB)
e57d5e30 00200082 00000000 00000003 dfc48ba8 00000000 dfc48ba8 00000000
00000007 e0af45c0 e8ce17aa 0002827f 00051ec2 e0af46cc c1809980 00000000
e8ce1324 0002827f 00200082 f881cd4c 00200286 f8ba2c85 c1809980 e57d5e60
Call Trace:
[<f881cd4c>] md_wakeup_thread+0x26/0x28 [md_mod]
[<f8ba2c85>] raid5_unplug_device+0x4e/0x5a [raid456]
[<c02a786e>] io_schedule+0x1d/0x27
[<c014f3c5>] sync_page+0x0/0x3b
[<c014f3fd>] sync_page+0x38/0x3b
[<c02a7a6f>] __wait_on_bit_lock+0x2a/0x52
[<c014f3b7>] __lock_page+0x58/0x5e
[<c01333aa>] wake_bit_function+0x0/0x3c
[<c0155daa>] truncate_inode_pages_range+0x201/0x256
[<c0155e16>] truncate_inode_pages+0x17/0x1a
[<f8c867eb>] reiserfs_delete_inode+0x36/0xdd [reiserfs]
[<f8c867b5>] reiserfs_delete_inode+0x0/0xdd [reiserfs]
[<c017acc5>] generic_delete_inode+0xa0/0x105
[<c017a4ad>] iput+0x60/0x62
[<c0172f41>] do_unlinkat+0xb6/0x126
[<c0103d7e>] syscall_call+0x7/0xb
=======================

My system is still up and running.

2007-08-12 08:23:36

by Andrew Morton

Subject: Re: Process stuck in md_wakeup_thread

On Sat, 11 Aug 2007 02:34:34 +0000 (UTC) Russ Dill <[email protected]> wrote:

> On 2.6.22 from debian (stock), I have a process (dpkg) stuck with the following
> calltrace:
>
> SysRq : Show Blocked State
>
> free sibling
> task PC stack pid father child younger older
> dpkg D 00000003 0 26040 20765 (NOTLB)
> e57d5e30 00200082 00000000 00000003 dfc48ba8 00000000 dfc48ba8 00000000
> 00000007 e0af45c0 e8ce17aa 0002827f 00051ec2 e0af46cc c1809980 00000000
> e8ce1324 0002827f 00200082 f881cd4c 00200286 f8ba2c85 c1809980 e57d5e60
> Call Trace:
> [<f881cd4c>] md_wakeup_thread+0x26/0x28 [md_mod]
> [<f8ba2c85>] raid5_unplug_device+0x4e/0x5a [raid456]

The above is stack gunk.

> [<c02a786e>] io_schedule+0x1d/0x27
> [<c014f3c5>] sync_page+0x0/0x3b
> [<c014f3fd>] sync_page+0x38/0x3b
> [<c02a7a6f>] __wait_on_bit_lock+0x2a/0x52
> [<c014f3b7>] __lock_page+0x58/0x5e
> [<c01333aa>] wake_bit_function+0x0/0x3c
> [<c0155daa>] truncate_inode_pages_range+0x201/0x256
> [<c0155e16>] truncate_inode_pages+0x17/0x1a
> [<f8c867eb>] reiserfs_delete_inode+0x36/0xdd [reiserfs]
> [<f8c867b5>] reiserfs_delete_inode+0x0/0xdd [reiserfs]
> [<c017acc5>] generic_delete_inode+0xa0/0x105
> [<c017a4ad>] iput+0x60/0x62
> [<c0172f41>] do_unlinkat+0xb6/0x126
> [<c0103d7e>] syscall_call+0x7/0xb
> =======================
>
> My system is still up and running.
>

It's stuck waiting for a page to come unlocked. The most likely
explanation is that an IO got lost.

2007-08-13 22:28:05

by Russ Dill

Subject: Re: Process stuck in md_wakeup_thread

On 8/12/07, Andrew Morton <[email protected]> wrote:
> On Sat, 11 Aug 2007 02:34:34 +0000 (UTC) Russ Dill <[email protected]> wrote:
>
> > On 2.6.22 from debian (stock), I have a process (dpkg) stuck with the following
> > calltrace:
> >
> > SysRq : Show Blocked State
> >
> > free sibling
> > task PC stack pid father child younger older
> > dpkg D 00000003 0 26040 20765 (NOTLB)
> > e57d5e30 00200082 00000000 00000003 dfc48ba8 00000000 dfc48ba8 00000000
> > 00000007 e0af45c0 e8ce17aa 0002827f 00051ec2 e0af46cc c1809980 00000000
> > e8ce1324 0002827f 00200082 f881cd4c 00200286 f8ba2c85 c1809980 e57d5e60
> > Call Trace:
> > [<f881cd4c>] md_wakeup_thread+0x26/0x28 [md_mod]
> > [<f8ba2c85>] raid5_unplug_device+0x4e/0x5a [raid456]
>
> The above is stack gunk.
>
> > [<c02a786e>] io_schedule+0x1d/0x27
> > [<c014f3c5>] sync_page+0x0/0x3b
> > [<c014f3fd>] sync_page+0x38/0x3b
> > [<c02a7a6f>] __wait_on_bit_lock+0x2a/0x52
> > [<c014f3b7>] __lock_page+0x58/0x5e
> > [<c01333aa>] wake_bit_function+0x0/0x3c
> > [<c0155daa>] truncate_inode_pages_range+0x201/0x256
> > [<c0155e16>] truncate_inode_pages+0x17/0x1a
> > [<f8c867eb>] reiserfs_delete_inode+0x36/0xdd [reiserfs]
> > [<f8c867b5>] reiserfs_delete_inode+0x0/0xdd [reiserfs]
> > [<c017acc5>] generic_delete_inode+0xa0/0x105
> > [<c017a4ad>] iput+0x60/0x62
> > [<c0172f41>] do_unlinkat+0xb6/0x126
> > [<c0103d7e>] syscall_call+0x7/0xb
> > =======================
> >
> > My system is still up and running.
> >
>
> It's stuck waiting for a page to come unlocked. The most likely
> explanation is that an IO got lost.
>

I'm not sure how an IO becomes lost. I'm not running on exotic
hardware or anything. What can I do to provide more useful debugging
output in case this occurs again?