panic: fix a possible deadlock in panic()
authorVikram Mulukutla <markivx@codeaurora.org>
Mon, 30 Jul 2012 21:39:58 +0000 (14:39 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Tue, 31 Jul 2012 00:25:13 +0000 (17:25 -0700)
commit190320c3b6640d4104650f55ff69611e050ea06b
tree94d43c11e025e4b4efc577a2cf88a29250d4f510
parent6c55845e71ba7a862395f504a240a5b62632e0d6
panic: fix a possible deadlock in panic()

panic_lock is meant to ensure that panic processing takes place only on
one cpu; if any of the other cpus encounter a panic, they will spin
waiting to be shut down.

However, this causes a regression in this scenario:

1. Cpu 0 encounters a panic and acquires the panic_lock
   and proceeds with the panic processing.
2. There is an interrupt on cpu 0 that also encounters
   an error condition and invokes panic.
3. This second invocation fails to acquire the panic_lock
   and enters the infinite while loop in panic_smp_self_stop.

Thus all panic processing is stopped, and the cpu is stuck for eternity
in the while(1) inside panic_smp_self_stop.

To address this, disable local interrupts with local_irq_disable before
acquiring the panic_lock.  This will prevent interrupt handlers from
executing during the panic processing, thus avoiding this particular
problem.

Signed-off-by: Vikram Mulukutla <markivx@codeaurora.org>
Reviewed-by: Stephen Boyd <sboyd@codeaurora.org>
Cc: Michael Holzheu <holzheu@linux.vnet.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
kernel/panic.c