Re: Stack traces in 2.6.9-rc2-mm4

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

Ingo Molnar wrote:
> * J.A. Magallon <jamagallon@able.es> wrote:
> 
> 
>>I got the same on another place...
>>
>>Sep 28 00:48:51 werewolf kernel: using smp_processor_id() in preemptible 
>>code: X/4012
>>Sep 28 00:48:51 werewolf kernel:  [smp_processor_id+135/141] 
>>smp_processor_id+0x87/0x8d
>>Sep 28 00:48:51 werewolf kernel:  [add_timer_randomness+314/365] 
>>add_timer_randomness+0x13a/0x16d
>>Sep 28 00:48:51 werewolf kernel:  [<b0206733>] 
>>add_timer_randomness+0x13a/0x16d
>>Sep 28 00:48:51 werewolf kernel:  [input_event+85/999] 
>>input_event+0x55/0x3e7
> 
> 
> the one below should fix this: a certain codepath in the random driver
> relied on vt_ioctl() being under the BKL and implicitly disabling
> preemption. The code wasnt buggy upstream but it's slighly unrobust so i
> think we want the fix upstream too, independently of the remove-bkl
> patch. This change has been in the -VP patchset for some time so it
> should work fine.
> 
> 	Ingo
> 
> --- linux/drivers/char/random.c.orig	
> +++ linux/drivers/char/random.c	
> @@ -807,10 +807,11 @@ static void add_timer_randomness(struct 
>  	long		delta, delta2, delta3;
>  	int		entropy = 0;
>  
> +	preempt_disable();
>  	/* if over the trickle threshold, use only 1 in 4096 samples */
>  	if ( random_state->entropy_count > trickle_thresh &&
>  	     (__get_cpu_var(trickle_count)++ & 0xfff))
> -		return;
> +		goto out;
>  

It looks like upstream code *is* buggy because that is a non-atomic
RMW operation on the per-cpu var, no? Hence you must disable preempt.