Re: [RFC][PATCH -rt] irqd starvation on SMP by a single process?

[john stultz <johnstul@us.ibm.com>]

On Fri, 2006-05-12 at 07:50 +0200, Ingo Molnar wrote:
> +		if (!cpus_equal(current->cpus_allowed, irq_affinity[irq]));
> +			set_cpus_allowed(current, irq_affinity[irq]);

Gah! I introduced a terrible bug there. 

Note the semi-colon at the end of the if statement! Sorry about that!

The following patch (which I've tested as well) fixes that.

--- 2.6-rt/kernel/irq/manage.c	2006-05-11 18:37:36.000000000 -0500
+++ dev-rt/kernel/irq/manage.c	2006-05-12 12:55:56.000000000 -0500
@@ -724,6 +724,7 @@
 		set_current_state(TASK_INTERRUPTIBLE);
 		do_hardirq(desc);
 		cond_resched_all();
+		local_irq_disable();
 		__do_softirq();
 //		do_softirq_from_hardirq();
 		local_irq_enable();
@@ -731,10 +732,8 @@
 		/*
 		 * Did IRQ affinities change?
 		 */
-		if (!cpu_isset(smp_processor_id(), irq_affinity[irq])) {
-			mask = cpumask_of_cpu(any_online_cpu(irq_affinity[irq]));
-			set_cpus_allowed(current, mask);
-		}
+		if(!cpus_equal(current->cpus_allowed, irq_affinity[irq]))
+			set_cpus_allowed(current, irq_affinity[irq]);
 #endif
 		schedule();
 	}