2007-12-13 16:19:54

by Jerome Marchand

Subject: [Patch 4/8] Enhanced partition statistics: cciss fix

Updates the enhanced partition statistics in cciss driver.

Signed-off-by: Jerome Marchand <[email protected]>
---
cciss.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff -urNp -X linux-2.6/Documentation/dontdiff linux-2.6.orig/drivers/block/cciss.c linux-2.6/drivers/block/cciss.c
--- linux-2.6.orig/drivers/block/cciss.c 2007-12-04 17:37:13.000000000 +0100
+++ linux-2.6/drivers/block/cciss.c 2007-12-05 13:46:29.000000000 +0100
@@ -1268,7 +1268,8 @@ static void cciss_softirq_done(struct re
if (blk_fs_request(rq)) {
const int rw = rq_data_dir(rq);

- disk_stat_add(rq->rq_disk, sectors[rw], rq->nr_sectors);
+ all_stat_add(rq->rq_disk, sectors[rw],
+ rq->nr_sectors, rq->sector);
}

#ifdef CCISS_DEBUG

2007-12-13 17:18:34

by Mike Miller

Subject: RE: [Patch 4/8] Enhanced partition statistics: cciss fix

>
> Updates the enhanced partition statistics in cciss driver.
>
> Signed-off-by: Jerome Marchand <[email protected]>
> ---
> cciss.c | 3 ++-
> 1 file changed, 2 insertions(+), 1 deletion(-) diff -urNp -X
> linux-2.6/Documentation/dontdiff
> linux-2.6.orig/drivers/block/cciss.c linux-2.6/drivers/block/cciss.c
> --- linux-2.6.orig/drivers/block/cciss.c 2007-12-04
> 17:37:13.000000000 +0100
> +++ linux-2.6/drivers/block/cciss.c 2007-12-05
> 13:46:29.000000000 +0100
> @@ -1268,7 +1268,8 @@ static void cciss_softirq_done(struct re
> if (blk_fs_request(rq)) {
> const int rw = rq_data_dir(rq);
>
> - disk_stat_add(rq->rq_disk, sectors[rw],
> rq->nr_sectors);
> + all_stat_add(rq->rq_disk, sectors[rw],
> + rq->nr_sectors, rq->sector);
> }

How does this mesh with the changes submitted by Kiyoshi Ueda? Any conflicts?

-- mikem

2007-12-13 19:31:44

by Kiyoshi Ueda

Subject: Re: [Patch 4/8] Enhanced partition statistics: cciss fix

Hi,

Thank you for noticing this, Mike.

On Thu, 13 Dec 2007 16:55:06 +0000, "Miller, Mike" wrote:
> > Updates the enhanced partition statistics in cciss driver.
> >
> > Signed-off-by: Jerome Marchand <[email protected]>
> > ---
> > cciss.c | 3 ++-
> > 1 file changed, 2 insertions(+), 1 deletion(-) diff -urNp -X
> > linux-2.6/Documentation/dontdiff
> > linux-2.6.orig/drivers/block/cciss.c linux-2.6/drivers/block/cciss.c
> > --- linux-2.6.orig/drivers/block/cciss.c 2007-12-04
> > 17:37:13.000000000 +0100
> > +++ linux-2.6/drivers/block/cciss.c 2007-12-05
> > 13:46:29.000000000 +0100
> > @@ -1268,7 +1268,8 @@ static void cciss_softirq_done(struct re
> > if (blk_fs_request(rq)) {
> > const int rw = rq_data_dir(rq);
> >
> > - disk_stat_add(rq->rq_disk, sectors[rw],
> > rq->nr_sectors);
> > + all_stat_add(rq->rq_disk, sectors[rw],
> > + rq->nr_sectors, rq->sector);
> > }
>
> How does this mesh with the changes submitted by Kiyoshi Ueda? Any conflicts?

I think no conflict with my patch technically.

Jerome's patch changes __end_that_request_first(), too, as same as
the change above.
And my patch changes cciss/cpqarray to use blk_end_request, which
uses __end_that_request_first() internally, instead of doing
disk_stat_add() by themselves.
(http://marc.info/?l=linux-kernel&m=119741542206258&w=2)

So Jerome can just drop the cciss/cpqarray changes from his patch-set,
if my patch-set is merged.

> -- mikem

Thanks,
Kiyoshi Ueda