Re: Remaining arch problems in cpu_idle

[Nick Piggin <nickpiggin@yahoo.com.au>]

Luck, Tony wrote:
>>I don't suppose safe_halt can be called with interrupts off
>>safely, like the i386 function of the same name?
> 
> 
> Yes.  Looking at the code for safe_halt() [chasing through
> ia64_do_halt_light() to PAL_CALL() to ia64_pal_call_static] I see
> that we disable interrupts before calling into the PAL, and restore
> the state after we return.  So somewhere in the depths of the PAL
> code the cpu is checking to see whether an external interrupt is
> pending, even though we have interrupts blocked.
> 

OK, in which case we should be able to disable interrupts before
checking need_resched() ?

Something like the following:

  void
  default_idle (void)
  {
-       while (!need_resched())
-               if (can_do_pal_halt)
-                       safe_halt();
-               else
+       if (can_do_pal_halt) {
+               while (!need_resched()) {
+                       local_irq_disable();
+                       if (!need_resched())
+                               safe_halt();
+                       local_irq_enable();
+       } else {
+               while (!need_resched())
                         cpu_relax();
+       }
  }



-- 
SUSE Labs, Novell Inc.

Send instant messages to your online friends http://au.messenger.yahoo.com