Skip to contentSkip to conversation

2.6.25.10

INFO: inconsistent lock state Bernard Pidoux 2 participants 2 messages Sep 7, 2008 – Sep 8, 2008

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

Hi,

I observed this message recently on a system with acpi=off for the bios is quite old.

Apparently listen application and console are involved.

Is there anything I can try to resolve this situation ,

=================================
[ INFO: inconsistent lock state ]
2.6.25.10-patched-ax25 #1
---------------------------------
inconsistent {in-softirq-W} -> {softirq-on-W} usage. listen/7202 [HC0[0]:SC0[0]:HE1:SE1] takes:
(_xmit_ETHER){-+..}, at: [<c0262f08>] netpoll_send_skb+0x118/0x180
{in-softirq-W} state was registered at:

  [<c013af92>] __lock_acquire+0x3c2/0x1100
  [<c013bd50>] lock_acquire+0x80/0xa0
  [<c02beac3>] _spin_lock+0x33/0x60
  [<c0265c30>] __qdisc_run+0x130/0x1b0
  [<c025815c>] dev_queue_xmit+0x25c/0x330
  [<c8c4955f>] mld_sendpack+0x2cf/0x2f0 [ipv6]
  [<c8c4a6e5>] mld_ifc_timer_expire+0x175/0x260 [ipv6]
  [<c0122d19>] run_timer_softirq+0x149/0x1b0
  [<c011f054>] __do_softirq+0x54/0xc0
  [<c011f107>] do_softirq+0x47/0x50
  [<c011f557>] irq_exit+0x57/0x70
  [<c01062df>] do_IRQ+0x4f/0xa0
  [<c0104672>] common_interrupt+0x2e/0x34
  [<ffffffff>] 0xffffffff
irq event stamp: 8625
hardirqs last  enabled at (8625): [<c02bf0e7>] _spin_unlock_irqrestore+0x47/0x60
hardirqs last disabled at (8624): [<c02bef69>] _spin_lock_irqsave+0x19/0x70
softirqs last  enabled at (8620): [<c011f0a3>] __do_softirq+0xa3/0xc0
softirqs last disabled at (8579): [<c011f107>] do_softirq+0x47/0x50

other info that might help us debug this:
2 locks held by listen/7202:
#0: (&tty->atomic_write_lock){--..}, at: [<c020abac>]

tty_write_lock+0x1c/0x50
 #1:  (target_list_lock){--..}, at: [<c8abd10d>] write_msg+0x2d/0xe0 

[netconsole]

stack backtrace:
Pid: 7202, comm: listen Not tainted 2.6.25.10-patched-ax25 #1
 [<c0139226>] print_usage_bug+0x156/0x160
 [<c013a390>] mark_lock+0x400/0x5a0
 [<c0108327>] ? native_sched_clock+0x77/0xb0
 [<c013afd8>] __lock_acquire+0x408/0x1100
 [<c013a70d>] ? trace_hardirqs_on+0xbd/0x140
 [<c024f4e3>] ? skb_dequeue+0x43/0x60
 [<c013bd50>] lock_acquire+0x80/0xa0
 [<c0262f08>] ? netpoll_send_skb+0x118/0x180
 [<c02be81a>] __lock_text_start+0x3a/0x50
 [<c0262f08>] ? netpoll_send_skb+0x118/0x180
 [<c0262f08>] netpoll_send_skb+0x118/0x180
 [<c0263c97>] netpoll_send_udp+0x217/0x290
 [<c8abd17c>] write_msg+0x9c/0xe0 [netconsole]
 [<c8abd0e0>] ? write_msg+0x0/0xe0 [netconsole]
 [<c011a4a7>] __call_console_drivers+0x47/0x60
 [<c011a539>] _call_console_drivers+0x79/0x90
 [<c011a913>] release_console_sem+0xd3/0x1f0
 [<c0219181>] do_con_write+0x161/0x1a20
 [<c013a500>] ? mark_lock+0x570/0x5a0
 [<c02bf0e7>] ? _spin_unlock_irqrestore+0x47/0x60
 [<c011aa04>] ? release_console_sem+0x1c4/0x1f0
 [<c021aa84>] con_write+0x14/0x30
 [<c020d5d2>] write_chan+0x192/0x340
 [<c0115f40>] ? default_wake_function+0x0/0x10
 [<c020ad75>] tty_write+0x125/0x1b0
 [<c020d440>] ? write_chan+0x0/0x340
 [<c0175515>] vfs_write+0x95/0x120
 [<c020ac50>] ? tty_write+0x0/0x1b0
 [<c0175a6d>] sys_write+0x3d/0x70
 [<c0103bda>] sysenter_past_esp+0x5f/0xa5
 =======================
#1 lore raw
Quoted 10 lines · Bernard Pidoux

Hi,

I observed this message recently on a system with acpi=off for the bios is quite old.

Apparently listen application and console are involved.

Is there anything I can try to resolve this situation ,

Quoted 7 lines

=================================
[ INFO: inconsistent lock state ]
2.6.25.10-patched-ax25 #1
---------------------------------
inconsistent {in-softirq-W} -> {softirq-on-W} usage. listen/7202 [HC0[0]:SC0[0]:HE1:SE1] takes:
(_xmit_ETHER){-+..}, at: [<c0262f08>] netpoll_send_skb+0x118/0x180

Hi Bernard,

These warnings usually mean there is some local_irq_disable/enable() instead of local_irq_save/restore() etc. on the xmit path (maybe in a network driver used for netconsole). Anyway, as usual, it would be nice to test it with some more current kernel, and if it's still there, add some info about this driver and additional patches.

Thanks,
Jarek P.