Re: [GIT pull] genirq fixes for 2.6.31

[Kevin Winchester <kjwinchester@gmail.com>]

Kevin Winchester wrote:
> Thomas Gleixner wrote:
>> +/*
>>   * Interrupt handler thread
>>   */
>>  static int irq_thread(void *data)
>> @@ -458,6 +494,8 @@ static int irq_thread(void *data)
>>  
>>  	while (!irq_wait_for_interrupt(action)) {
>>  
>> +		irq_thread_check_affinity(desc, action);
>> +
>>  		atomic_inc(&desc->threads_active);
>>  
>>  		spin_lock_irq(&desc->lock);
> 
> Any chance we could do this in a way that doesn't break the build for CONFIG_SMP=n? :)
> 
> Should this call simply be wrapped in an #ifdef, or should the function be defined for !SMP?
> 

Actually, I guess the function is defined always, but it references desc->affinity unconditionally.

Perhaps something like the following (likely whitespace damaged) patch:

diff --git a/kernel/irq/manage.c b/kernel/irq/manage.c
index f0de36f..353e335 100644
--- a/kernel/irq/manage.c
+++ b/kernel/irq/manage.c
@@ -451,6 +451,8 @@ static int irq_wait_for_interrupt(struct irqaction *action)
        return -1;
 }
 
+#ifdef CONFIG_SMP
+
 /*
  * Check whether we need to change the affinity of the interrupt thread.
  */
@@ -479,6 +481,15 @@ irq_thread_check_affinity(struct irq_desc *desc, struct irqaction *action)
        free_cpumask_var(mask);
 }
 
+#else
+
+static void
+irq_thread_check_affinity(struct irq_desc *desc, struct irqaction *action)
+{
+}
+
+#endif
+
 /*
  * Interrupt handler thread
  */