2004-09-09 08:43:25

by Jens Axboe

Subject: [PATCH] fix leak with bounced bio's

Hi,

This might fix the last leak of memory reported with cd writing, the
current highmem bounce code will leak n-1 pages for any n page bio where
n > 1. CD writing typically uses 16 pages bios, so it is affected.

===== mm/highmem.c 1.51 vs edited =====
--- 1.51/mm/highmem.c 2004-07-29 06:58:32 +02:00
+++ edited/mm/highmem.c 2004-09-08 21:18:57 +02:00
@@ -284,7 +284,7 @@
struct bio_vec *tovec, *fromvec;
int i;

- bio_for_each_segment(tovec, to, i) {
+ __bio_for_each_segment(tovec, to, i, 0) {
fromvec = from->bi_io_vec + i;

/*
@@ -316,7 +316,7 @@
/*
* free up bounce indirect pages used
*/
- bio_for_each_segment(bvec, bio, i) {
+ __bio_for_each_segment(bvec, bio, i, 0) {
org_vec = bio_orig->bi_io_vec + i;
if (bvec->bv_page == org_vec->bv_page)
continue;

--
Jens Axboe

2004-09-09 14:34:29

by Frank Steiner

Subject: Re: [PATCH] fix leak with bounced bio's

Jens Axboe wrote:
> Hi,
>
> This might fix the last leak of memory reported with cd writing, the
> current highmem bounce code will leak n-1 pages for any n page bio where
> n > 1. CD writing typically uses 16 pages bios, so it is affected.


it fixes the leak for us that I described a while back
http://marc.theaimsgroup.com/?l=linux-kernel&m=109360958318479&w=2

Great :-))

cu,
Frank

--
Dipl.-Inform. Frank Steiner Web: http://www.bio.ifi.lmu.de/~steiner/
Lehrstuhl f. Bioinformatik Mail: http://www.bio.ifi.lmu.de/~steiner/m/
LMU, Amalienstr. 17 Phone: +49 89 2180-4049
80333 Muenchen, Germany Fax: +49 89 2180-99-4049

2004-09-09 14:41:23

by Jens Axboe

Subject: Re: [PATCH] fix leak with bounced bio's

On Thu, Sep 09 2004, Frank Steiner wrote:
> Jens Axboe wrote:
> >Hi,
> >
> >This might fix the last leak of memory reported with cd writing, the
> >current highmem bounce code will leak n-1 pages for any n page bio where
> >n > 1. CD writing typically uses 16 pages bios, so it is affected.
>
>
> it fixes the leak for us that I described a while back
> http://marc.theaimsgroup.com/?l=linux-kernel&m=109360958318479&w=2
>
> Great :-))

Great, thanks for confirming!

Maybe we should unconditionally bounce all bios for a while in a test
kernel, to make sure that all bouncing bugs are shaken out ;-)

--
Jens Axboe