Re: [PATCH] hw breakpoint: Fix possible memory leak

[Namhyung Kim <namhyung@gmail.com>]

2012-02-27 (Mon), 12:04 +0100, Peter Zijlstra wrote:
> On Mon, 2012-02-27 at 11:44 +0100, Ingo Molnar wrote:
> > I edited that code earlier today - is the form below OK, or can 
> > you see a simpler method? It's not yet pushed out so can still 
> > edit it. 
> 
> I think something like the below should do, but then I didn't really
> think much about it, my thoughts went like:
> 
> ... *shees* that's ugly
>   ... that error path already does a loop
>     ... what the problem is!? -- reread changelog
>       ... err_cpu == cpu is placed wrong!
> 
> 
> So I replied and marked read.. waiting to either hear if there's a good
> reason to do ugly or find a new (tested) patch in my inbox.. :-)
> 
> ---
>  kernel/events/hw_breakpoint.c |    4 ++--
>  1 files changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/kernel/events/hw_breakpoint.c b/kernel/events/hw_breakpoint.c
> index b0309f7..3330022 100644
> --- a/kernel/events/hw_breakpoint.c
> +++ b/kernel/events/hw_breakpoint.c
> @@ -658,10 +658,10 @@ int __init init_hw_breakpoint(void)
>  
>   err_alloc:
>  	for_each_possible_cpu(err_cpu) {
> -		if (err_cpu == cpu)
> -			break;
>  		for (i = 0; i < TYPE_MAX; i++)
>  			kfree(per_cpu(nr_task_bp_pinned[i], cpu));
> +		if (err_cpu == cpu)
> +			break;
>  	}
>  
>  	return -ENOMEM;
> 

This would depend on the initial value of the percpu memory, and thus
have no problem in this case. It looks better to me, too :)


-- 
Regards,
Namhyung Kim