Re: [PATCH] klist: del waiter from klist_remove_waiters before wakeup waitting process

[Peter Zijlstra <peterz@infradead.org>]

On Fri, May 03, 2013 at 03:06:36PM +0800, wangbiao wrote:
> From: "wang, biao" <biao.wang@intel.com>
> Date: Fri, 3 May 2013 14:18:34 +0800
> Subject: [PATCH] klist: del waiter from klist_remove_waiters before wakeup
>  waitting process
> 
> There is a race between klist_remove and klist_release. klist_remove
> uses a local var waiter saved on stack. When klist_release calls
> wake_up_process(waiter->process) to wake up the waiter, waiter might run
> immediately and reuse the stack. Then, klist_release calls
> list_del(&waiter->list) to change previous
> wait data and cause prior waiter thread corrupt.
> 
> The patch fixes it against kernel 3.9.


I've never seen that code before in my life; but after a quick look you appear
to be completely right.

> Signed-off-by: wang, biao <biao.wang@intel.com>

Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>

> ---
>  lib/klist.c |    2 +-
>  1 files changed, 1 insertions(+), 1 deletions(-)
> 
> diff --git a/lib/klist.c b/lib/klist.c
> index 0874e41..358a368 100644
> --- a/lib/klist.c
> +++ b/lib/klist.c
> @@ -193,10 +193,10 @@ static void klist_release(struct kref *kref)
>  		if (waiter->node != n)
>  			continue;
>  
> +		list_del(&waiter->list);
>  		waiter->woken = 1;
>  		mb();
>  		wake_up_process(waiter->process);
> -		list_del(&waiter->list);
>  	}
>  	spin_unlock(&klist_remove_lock);
>  	knode_set_klist(n, NULL);
> -- 
> 1.7.6
> 
> 
>