Re: 2.6.15-rc5-rt2 won't boot on dual 933

["K.R. Foley" <kr@cybsft.com>]

Steven Rostedt wrote:
> On Sun, 2005-12-18 at 15:55 -0600, K.R. Foley wrote:
>> dump_stack+0x1e/0x20 (20)
>>  [<c011c9df>] preempt_enable_no_resched+0x5f/0x70 (20)
>>  [<c0100ff2>] cpu_idle+0xb2/0x100 (40)
>>  [<c0111446>] start_secondary+0x296/0x340<6> 
> 
> Ingo,
> 
> What's the reason for this printing out in cpu_idle?
> 
> #ifdef CONFIG_DEBUG_PREEMPT
> void notrace preempt_enable_no_resched(void)
> {
> 	static int once = 1;
> 
> 	barrier();
> 	dec_preempt_count();
> 
> 	if (once && !preempt_count()) {
> 		once = 0;
> 		printk(KERN_ERR "BUG: %s:%d task might have lost a preemption check!
> \n",
> 			current->comm, current->pid);
> 		dump_stack();
> 	}
> }
> 
> EXPORT_SYMBOL(preempt_enable_no_resched);
> #endif
> 
> I can understand the above when using preempt_enable_no_resched, when
> you know that you still have preemption on, but sometimes (as in
> cpu_idle) it is used just before calling schedule.  So this check is
> pretty much meaningless.
> 
> ---
> 
> KR, 
> 
> Does your kernel boot without CONFIG_DEBUG_PREEMPT?
> 
> -- Steve

No. Already tried and got a different dump. Not sure if it was just a
dump or a dump initiated by an oops. In an hour or so I will go down and
capture that.

-- 
   kr