Re: [dm-devel] [PATCH v3 01/26] block: Fix a buffer overrun in bio_integrity_split()

[Vivek Goyal <vgoyal-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>]

On Mon, Sep 24, 2012 at 03:34:41PM -0700, Kent Overstreet wrote:
> bio_integrity_split() seemed to be confusing pointers and arrays -
> bip_vec in bio_integrity_payload is an array appended to the end of the
> payload, so the bio_vecs in struct bio_pair need to come immediately
> after the bio_integrity_payload they're for, and there was an assignment
> in bio_integrity_split() that didn't make any sense.
> 
> Signed-off-by: Kent Overstreet <koverstreet-hpIqsD4AKlfQT0dZR+AlfA@public.gmane.org>
> CC: Jens Axboe <axboe-tSWWG44O7X1aa/9Udqfwiw@public.gmane.org>
> CC: Martin K. Petersen <martin.petersen-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org>
> ---
>  fs/bio-integrity.c  | 3 ---
>  include/linux/bio.h | 6 ++++--
>  2 files changed, 4 insertions(+), 5 deletions(-)
> 
> diff --git a/fs/bio-integrity.c b/fs/bio-integrity.c
> index a3f28f3..c7b6b52 100644
> --- a/fs/bio-integrity.c
> +++ b/fs/bio-integrity.c
> @@ -697,9 +697,6 @@ void bio_integrity_split(struct bio *bio, struct bio_pair *bp, int sectors)
>  	bp->iv1 = bip->bip_vec[0];
>  	bp->iv2 = bip->bip_vec[0];
>  
> -	bp->bip1.bip_vec[0] = bp->iv1;
> -	bp->bip2.bip_vec[0] = bp->iv2;
> -
>  	bp->iv1.bv_len = sectors * bi->tuple_size;
>  	bp->iv2.bv_offset += sectors * bi->tuple_size;
>  	bp->iv2.bv_len -= sectors * bi->tuple_size;
> diff --git a/include/linux/bio.h b/include/linux/bio.h
> index b31036f..8e2d108 100644
> --- a/include/linux/bio.h
> +++ b/include/linux/bio.h
> @@ -200,8 +200,10 @@ struct bio_pair {
>  	struct bio			bio1, bio2;
>  	struct bio_vec			bv1, bv2;
>  #if defined(CONFIG_BLK_DEV_INTEGRITY)
> -	struct bio_integrity_payload	bip1, bip2;
> -	struct bio_vec			iv1, iv2;
> +	struct bio_integrity_payload	bip1;
> +	struct bio_vec			iv1;
> +	struct bio_integrity_payload	bip2;
> +	struct bio_vec			iv2;
>  #endif

I think it probably is a good idea to put a comment here so that we
know that certain elements of structure assume ordering.

Also I am wondering that what's the gurantee that there are no padding
bytes between bipi1 and iv1 (or bip2 or iv2). I think if there are padding
bytes then the assumption that bio_vec is always following bip will be
broken?

Also had a general question about split logic. We seem to have only one
global pool for bio pair (bio_split_pool). So in the IO stack if we split
a bio more than once, we have the deadlock possibility again?

Thanks
Vivek