2007-10-03 10:02:19

by Pekka Pietikäinen

[permalink] [raw]
Subject: 2.6.23-rc8-git2 possible recursive locking when running screen

Seen this a few times lately on a machine running rawhide when running
screen (and doing "something", it's not automatic. And box works just fine).
I think I saw this a few weeks back, so it's not a new regression.

=============================================
[ INFO: possible recursive locking detected ]
2.6.23-0.211.rc8.git2.fc8 #1
---------------------------------------------
screen/32227 is trying to acquire lock:
(&q->lock){++..}, at: [<c0426703>] __wake_up+0x15/0x42

but task is already holding lock:
(&q->lock){++..}, at: [<c0426703>] __wake_up+0x15/0x42

other info that might help us debug this:
2 locks held by screen/32227:
#0: (&tty->atomic_read_lock){--..}, at: [<c0554ca2>] read_chan+0x1a1/0x5af
#1: (&q->lock){++..}, at: [<c0426703>] __wake_up+0x15/0x42

stack backtrace:
[<c0406463>] show_trace_log_lvl+0x1a/0x2f
[<c0406e4d>] show_trace+0x12/0x14
[<c0406e65>] dump_stack+0x16/0x18
[<c0449c3a>] __lock_acquire+0x189/0xc67
[<c044ab92>] lock_acquire+0x7b/0x9e
[<c06334fa>] _spin_lock_irqsave+0x4a/0x77
[<c0426703>] __wake_up+0x15/0x42
[<c04aece2>] ep_poll_safewake+0x86/0xa8
[<c04af969>] ep_poll_callback+0x9f/0xaa
[<c042483e>] __wake_up_common+0x32/0x55
[<c042671f>] __wake_up+0x31/0x42
[<c054f864>] tty_wakeup+0x4f/0x54
[<c0555e9b>] pty_unthrottle+0x15/0x21
[<c055301b>] check_unthrottle+0x2e/0x30
[<c0554fad>] read_chan+0x4ac/0x5af
[<c0551c99>] tty_read+0x66/0xac
[<c0489fee>] vfs_read+0xad/0x167
[<c048a482>] sys_read+0x3d/0x61
[<c040522e>] syscall_call+0x7/0xb
=======================


2007-10-04 15:24:30

by Rafael J. Wysocki

[permalink] [raw]
Subject: Re: 2.6.23-rc8-git2 possible recursive locking when running screen

On Wednesday, 3 October 2007 11:24, Pekka Pietikainen wrote:
> Seen this a few times lately on a machine running rawhide when running
> screen (and doing "something", it's not automatic. And box works just fine).
> I think I saw this a few weeks back, so it's not a new regression.

Did it happen with 2.6.22?

Rafael