Re: [PATCH 1/2]block cfq: don't use atomic_t for cfq_queue

[Vivek Goyal]

On Thu, Dec 23, 2010 at 10:45:33AM +0800, Shaohua Li wrote:
> cfq_queue->ref is used with queue_lock hold, so ref doesn't need to be an atomic
> and atomic operation is slower.
> 
> Signed-off-by: Shaohua Li <shaohua.li@xxxxxxxxx>

Looks good to me. 

Acked-by: Vivek Goyal <vgoyal@xxxxxxxxxx>

Vivek

> 
> ---
>  block/cfq-iosched.c |   27 ++++++++++++++++-----------
>  1 file changed, 16 insertions(+), 11 deletions(-)
> 
> Index: linux/block/cfq-iosched.c
> ===================================================================
> --- linux.orig/block/cfq-iosched.c	2010-12-22 21:14:15.000000000 +0800
> +++ linux/block/cfq-iosched.c	2010-12-23 10:44:23.000000000 +0800
> @@ -97,7 +97,7 @@ struct cfq_rb_root {
>   */
>  struct cfq_queue {
>  	/* reference count */
> -	atomic_t ref;
> +	int ref;
>  	/* various state flags, see below */
>  	unsigned int flags;
>  	/* parent cfq_data */
> @@ -2040,7 +2040,7 @@ static int cfqq_process_refs(struct cfq_
>  	int process_refs, io_refs;
>  
>  	io_refs = cfqq->allocated[READ] + cfqq->allocated[WRITE];
> -	process_refs = atomic_read(&cfqq->ref) - io_refs;
> +	process_refs = cfqq->ref - io_refs;
>  	BUG_ON(process_refs < 0);
>  	return process_refs;
>  }
> @@ -2080,10 +2080,10 @@ static void cfq_setup_merge(struct cfq_q
>  	 */
>  	if (new_process_refs >= process_refs) {
>  		cfqq->new_cfqq = new_cfqq;
> -		atomic_add(process_refs, &new_cfqq->ref);
> +		new_cfqq->ref += process_refs;
>  	} else {
>  		new_cfqq->new_cfqq = cfqq;
> -		atomic_add(new_process_refs, &cfqq->ref);
> +		cfqq->ref += new_process_refs;
>  	}
>  }
>  
> @@ -2538,9 +2538,10 @@ static void cfq_put_queue(struct cfq_que
>  	struct cfq_data *cfqd = cfqq->cfqd;
>  	struct cfq_group *cfqg, *orig_cfqg;
>  
> -	BUG_ON(atomic_read(&cfqq->ref) <= 0);
> +	BUG_ON(cfqq->ref <= 0);
>  
> -	if (!atomic_dec_and_test(&cfqq->ref))
> +	cfqq->ref--;
> +	if (cfqq->ref)
>  		return;
>  
>  	cfq_log_cfqq(cfqd, cfqq, "put_queue");
> @@ -2843,7 +2844,7 @@ static void cfq_init_cfqq(struct cfq_dat
>  	RB_CLEAR_NODE(&cfqq->p_node);
>  	INIT_LIST_HEAD(&cfqq->fifo);
>  
> -	atomic_set(&cfqq->ref, 0);
> +	cfqq->ref = 0;
>  	cfqq->cfqd = cfqd;
>  
>  	cfq_mark_cfqq_prio_changed(cfqq);
> @@ -2979,11 +2980,11 @@ cfq_get_queue(struct cfq_data *cfqd, boo
>  	 * pin the queue now that it's allocated, scheduler exit will prune it
>  	 */
>  	if (!is_sync && !(*async_cfqq)) {
> -		atomic_inc(&cfqq->ref);
> +		cfqq->ref++;
>  		*async_cfqq = cfqq;
>  	}
>  
> -	atomic_inc(&cfqq->ref);
> +	cfqq->ref++;
>  	return cfqq;
>  }
>  
> @@ -3681,7 +3682,7 @@ new_queue:
>  	}
>  
>  	cfqq->allocated[rw]++;
> -	atomic_inc(&cfqq->ref);
> +	cfqq->ref++;
>  
>  	spin_unlock_irqrestore(q->queue_lock, flags);
>  
> @@ -3862,6 +3863,10 @@ static void *cfq_init_queue(struct reque
>  	if (!cfqd)
>  		return NULL;
>  
> +	/*
> +	 * Don't need take queue_lock in the routine, since we are
> +	 * initializing the ioscheduler, and nobody is using cfqd
> +	 */
>  	cfqd->cic_index = i;
>  
>  	/* Init root service tree */
> @@ -3901,7 +3906,7 @@ static void *cfq_init_queue(struct reque
>  	 * will not attempt to free it.
>  	 */
>  	cfq_init_cfqq(cfqd, &cfqd->oom_cfqq, 1, 0);
> -	atomic_inc(&cfqd->oom_cfqq.ref);
> +	cfqd->oom_cfqq.ref++;
>  	cfq_link_cfqq_cfqg(&cfqd->oom_cfqq, &cfqd->root_group);
>  
>  	INIT_LIST_HEAD(&cfqd->cic_list);
> 
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/