[PATCH] md_make_request: don't touch the bio after calling make_request

[PATCH] md_make_request: don't touch the bio after calling make_request

[Chris Mason]

md_make_request was calling bio_sectors() for part_stat_add
after it was calling the make_request function.  This is
bad because the make_request function can free the bio and
because the bi_size field can change around.

The fix here was suggested by Jens Axboe.  It saves the
sector count before the make_request call.  I hit this
with CONFIG_DEBUG_PAGEALLOC turned on while trying to break
his pretty fusionio card.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
---
 drivers/md/md.c |    9 +++++++--
 1 files changed, 7 insertions(+), 2 deletions(-)

diff --git a/drivers/md/md.c b/drivers/md/md.c
index b76cfc8..a7d3c3c 100644
--- a/drivers/md/md.c
+++ b/drivers/md/md.c
@@ -287,6 +287,7 @@ static int md_make_request(struct request_queue *q, struct bio *bio)
 	mddev_t *mddev = q->queuedata;
 	int rv;
 	int cpu;
+	unsigned int sectors;
 
 	if (mddev == NULL || mddev->pers == NULL
 	    || !mddev->ready) {
@@ -311,12 +312,16 @@ static int md_make_request(struct request_queue *q, struct bio *bio)
 	atomic_inc(&mddev->active_io);
 	rcu_read_unlock();
 
+	/*
+	 * save the sectors now since our bio can
+	 * go away inside make_request
+	 */
+	sectors = bio_sectors(bio);
 	rv = mddev->pers->make_request(mddev, bio);
 
 	cpu = part_stat_lock();
 	part_stat_inc(cpu, &mddev->gendisk->part0, ios[rw]);
-	part_stat_add(cpu, &mddev->gendisk->part0, sectors[rw],
-		      bio_sectors(bio));
+	part_stat_add(cpu, &mddev->gendisk->part0, sectors[rw], sectors);
 	part_stat_unlock();
 
 	if (atomic_dec_and_test(&mddev->active_io) && mddev->suspended)
-- 
1.7.1.1

Re: [PATCH] md_make_request: don't touch the bio after calling make_request

[NeilBrown]

On Mon, 07 Feb 2011 14:21:48 -0500 Chris Mason <chris.mason@oracle.com> wrote:

> md_make_request was calling bio_sectors() for part_stat_add
> after it was calling the make_request function.  This is
> bad because the make_request function can free the bio and
> because the bi_size field can change around.
> 
> The fix here was suggested by Jens Axboe.  It saves the
> sector count before the make_request call.  I hit this
> with CONFIG_DEBUG_PAGEALLOC turned on while trying to break
> his pretty fusionio card.

Thanks Chris!!

I've added it to my queue of md fixes (which I really must send of soon) and
tagged it for -stable.

NeilBrown


> 
> Signed-off-by: Chris Mason <chris.mason@oracle.com>
> ---
>  drivers/md/md.c |    9 +++++++--
>  1 files changed, 7 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/md/md.c b/drivers/md/md.c
> index b76cfc8..a7d3c3c 100644
> --- a/drivers/md/md.c
> +++ b/drivers/md/md.c
> @@ -287,6 +287,7 @@ static int md_make_request(struct request_queue *q, struct bio *bio)
>  	mddev_t *mddev = q->queuedata;
>  	int rv;
>  	int cpu;
> +	unsigned int sectors;
>  
>  	if (mddev == NULL || mddev->pers == NULL
>  	    || !mddev->ready) {
> @@ -311,12 +312,16 @@ static int md_make_request(struct request_queue *q, struct bio *bio)
>  	atomic_inc(&mddev->active_io);
>  	rcu_read_unlock();
>  
> +	/*
> +	 * save the sectors now since our bio can
> +	 * go away inside make_request
> +	 */
> +	sectors = bio_sectors(bio);
>  	rv = mddev->pers->make_request(mddev, bio);
>  
>  	cpu = part_stat_lock();
>  	part_stat_inc(cpu, &mddev->gendisk->part0, ios[rw]);
> -	part_stat_add(cpu, &mddev->gendisk->part0, sectors[rw],
> -		      bio_sectors(bio));
> +	part_stat_add(cpu, &mddev->gendisk->part0, sectors[rw], sectors);
>  	part_stat_unlock();
>  
>  	if (atomic_dec_and_test(&mddev->active_io) && mddev->suspended)