Re: [PATCH 3/4 V2] futex: free_pi_state outside of hb->lock sections

[Steven Rostedt <rostedt@goodmis.org>]

On Fri, 2010-07-09 at 15:55 -0700, Darren Hart wrote:
> Apologies, mangled a rebase and this patch had bits of 4/4 in it.
> Corrected below:
> 
> 
> >From f40b8e684df6ed2b1ba9167cad3922ab43f4a717 Mon Sep 17 00:00:00 2001
> From: Darren Hart <dvhltc@us.ibm.com>
> Date: Fri, 9 Jul 2010 17:07:10 -0400
> Subject: [PATCH 3/4 V2] futex: free_pi_state outside of hb->lock sections
> 
> free_pi_state() calls kfree() and might sleep. To prepare for raw hb->locks,
> get the calls to free_pi_state() out of the hb->lock() sections.
> 
> Signed-off-by: Darren Hart <dvhltc@us.ibm.com>
> Cc: Thomas Gleixner <tglx@linutronix.de>
> Cc: Peter Zijlstra <peterz@infradead.org>
> Cc: Ingo Molnar <mingo@elte.hu>
> Cc: Eric Dumazet <eric.dumazet@gmail.com>
> Cc: John Kacur <jkacur@redhat.com>
> Cc: Steven Rostedt <rostedt@goodmis.org>
> Cc: Mike Galbraith <efault@gmx.de>
> ---
>  kernel/futex.c |    8 +++++---
>  1 files changed, 5 insertions(+), 3 deletions(-)
> 
> diff --git a/kernel/futex.c b/kernel/futex.c
> index a6cec32..2cd58a2 100644
> --- a/kernel/futex.c
> +++ b/kernel/futex.c
> @@ -1339,7 +1339,6 @@ retry_private:

This is why I always add a space before labels. I have no idea what
function this was in.

>  			} else if (ret) {
>  				/* -EDEADLK */
>  				this->pi_state = NULL;
> -				free_pi_state(pi_state);

Where do we free the pi state here?

-- Steve

>  				goto out_unlock;
>  			}
>  		}
> @@ -1483,16 +1482,19 @@ retry:
>   */
>  static void unqueue_me_pi(struct futex_q *q)
>  {
> +	struct futex_pi_state *pi_state = NULL;
> +
>  	WARN_ON(plist_node_empty(&q->list));
>  	plist_del(&q->list, &q->list.plist);
>  
>  	BUG_ON(!q->pi_state);
> -	free_pi_state(q->pi_state);
> +	pi_state = q->pi_state;
>  	q->pi_state = NULL;
>  
>  	spin_unlock(q->lock_ptr);
> -
>  	drop_futex_key_refs(&q->key);
> +
> +	free_pi_state(pi_state);
>  }
>  
>  /*