[PATCH] genirq: try_one_irq() should be called with irq disabled

[Yong Zhang]

Prarit report this:
Booting 2.6.32-rc5 on some IBM systems results in

Disabling IRQ #19

=================================
[ INFO: inconsistent lock state ]
2.6.32-rc5 #1
---------------------------------
inconsistent {IN-HARDIRQ-W} -> {HARDIRQ-ON-W} usage.
swapper/0 [HC0[0]:SC1[1]:HE1:SE0] takes:
 (&irq_desc_lock_class){?.-...}, at: [<ffffffff810c264e>] try_one_irq+0x32/0x138
{IN-HARDIRQ-W} state was registered at:
 [<ffffffff81095160>] __lock_acquire+0x2fc/0xd5d
 [<ffffffff81095cb4>] lock_acquire+0xf3/0x12d
 [<ffffffff814cdadd>] _spin_lock+0x40/0x89
 [<ffffffff810c3389>] handle_level_irq+0x30/0x105
 [<ffffffff81014e0e>] handle_irq+0x95/0xb7
 [<ffffffff810141bd>] do_IRQ+0x6a/0xe0
 [<ffffffff81012813>] ret_from_intr+0x0/0x16
irq event stamp: 195096
hardirqs last  enabled at (195096): [<ffffffff814cd7f7>]
_spin_unlock_irq+0x3a/0x5c
hardirqs last disabled at (195095): [<ffffffff814cdbdd>]
_spin_lock_irq+0x29/0x95
softirqs last  enabled at (195088): [<ffffffff81068c92>]
__do_softirq+0x1c1/0x1ef
softirqs last disabled at (195093): [<ffffffff8101304c>] call_softirq+0x1c/0x30

other info that might help us debug this:
1 lock held by swapper/0:
 #0:  (kernel/irq/spurious.c:21){+.-...}, at: [<ffffffff81070cf2>]
run_timer_softirq+0x1a9/0x315

stack backtrace:
Pid: 0, comm: swapper Not tainted 2.6.32-rc5 #1
Call Trace:
 <IRQ>  [<ffffffff81093e94>] valid_state+0x187/0x1ae
 [<ffffffff81096c7b>] ? check_usage_backwards+0x0/0xa3
 [<ffffffff81093fe4>] mark_lock+0x129/0x253
 [<ffffffff810951d4>] __lock_acquire+0x370/0xd5d
 [<ffffffff810c264e>] ? try_one_irq+0x32/0x138
 [<ffffffff8109329d>] ? save_trace+0x4e/0xcd
 [<ffffffff81095cb4>] lock_acquire+0xf3/0x12d
 [<ffffffff810c264e>] ? try_one_irq+0x32/0x138
 [<ffffffff81070cf2>] ? run_timer_softirq+0x1a9/0x315
 [<ffffffff810c264e>] ? try_one_irq+0x32/0x138
 [<ffffffff814cdadd>] _spin_lock+0x40/0x89
 [<ffffffff810c264e>] ? try_one_irq+0x32/0x138
 [<ffffffff810c264e>] try_one_irq+0x32/0x138
 [<ffffffff810c2795>] poll_all_shared_irqs+0x41/0x6d
 [<ffffffff810c27dd>] poll_spurious_irqs+0x1c/0x49
 [<ffffffff81070d82>] run_timer_softirq+0x239/0x315
 [<ffffffff81070cf2>] ? run_timer_softirq+0x1a9/0x315
 [<ffffffff810c27c1>] ? poll_spurious_irqs+0x0/0x49
 [<ffffffff81068bd3>] __do_softirq+0x102/0x1ef
 [<ffffffff8108eccf>] ? tick_dev_program_event+0x46/0xcc
 [<ffffffff8101304c>] call_softirq+0x1c/0x30
 [<ffffffff81014b65>] do_softirq+0x59/0xca
 [<ffffffff810686ad>] irq_exit+0x58/0xae
 [<ffffffff81029b84>] smp_apic_timer_interrupt+0x94/0xba
 [<ffffffff81012a33>] apic_timer_interrupt+0x13/0x20
 <EOI>  [<ffffffff8101a7b5>] ? mwait_idle+0x8c/0xb5
 [<ffffffff8101a7ac>] ? mwait_idle+0x83/0xb5
 [<ffffffff81010e55>] ? cpu_idle+0xbe/0x100
 [<ffffffff814c4270>] ? start_secondary+0x219/0x270

The reason is that try_one_irq() is called both from hardirq context
and softirq context. And by default the timer handler
poll_all_shared_irqs() is called with irq enabled.
Then the two usage will cause inconsistent.

Reported-by: Prarit Bhargava <prarit@xxxxxxxxxx>
Signed-off-by: Yong Zhang <yong.zhang0@xxxxxxxxx>
---
 kernel/irq/spurious.c |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)

diff --git a/kernel/irq/spurious.c b/kernel/irq/spurious.c
index 114e704..bd7273e 100644
--- a/kernel/irq/spurious.c
+++ b/kernel/irq/spurious.c
@@ -121,7 +121,9 @@ static void poll_all_shared_irqs(void)
 		if (!(status & IRQ_SPURIOUS_DISABLED))
 			continue;

+		local_irq_disable();
 		try_one_irq(i, desc);
+		local_irq_enable();
 	}
 }

-- 
1.6.3.3
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/