Re: IRQ and sleep_on

[Manfred Spraul <manfred@colorfullife.com>]

christophe barbe wrote:
> 
> I've missed the thread "avoiding bad sleeps" last week. I've had a similar problem
> and I would like to discuss the solution I've used to avoid it.
> 
> I want to wake up a sleeping process from an IRQ handler. In the process, if I use
> a interruptible_sleep_on(), I need first to restore flags (otherwise the process
> will sleep forever).
> 
> restore_flags(flags);
> // <<== here IRQ handler possibly call wake_up()
> interruptible_sleep_on(&my_queue);
>
> [...]
> I've written a modified version of  interruptible_sleep_on which takes an
> additionnal argument : flags to be restored.

That's possible, but it will crash on Sparc: you cannot restore the
interrupt flag saved in one function in another function.

The solution is very simple: do not call restore_flags() before
interruptible_sleep_on(), the schedule internally reenables interrupts.

>>>>>>>>>>
for(;;) {
	cli();
	if(condition) {
		sti();
		break;
	}
	interruptible_sleep_on();
	sti(); /* required! */
} 	
>>>>>>>>>

But if you are writing new code, then DO NOT USE sleep_on(), use
add_wait_queue(), and a spinlock instead of cli().
Look at wait_event_irq in <linux/raid/md_k.h> from the 2.4 kernel as an
example.

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