linux-kernel - Re: [RFC PATCH 1/2] sg_ring instead of scatterlist chaining

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-Id: <20071105084042.83149a5c.randy.dunlap@oracle.com>
Date:	Mon, 5 Nov 2007 08:40:42 -0800
From:	Randy Dunlap <randy.dunlap@...cle.com>
To:	Rusty Russell <rusty@...tcorp.com.au>
Cc:	Linus Torvalds <torvalds@...ux-foundation.org>,
	Paul Mackerras <paulus@...ba.org>,
	Boaz Harrosh <bharrosh@...asas.com>,
	Jens Axboe <jens.axboe@...cle.com>,
	Alan Cox <alan@...rguk.ukuu.org.uk>,
	Geert Uytterhoeven <geert@...ux-m68k.org>,
	Linux Kernel Development <linux-kernel@...r.kernel.org>,
	mingo@...e.hu
Subject: Re: [RFC PATCH 1/2] sg_ring instead of scatterlist chaining

On Mon, 5 Nov 2007 17:11:55 +1100 Rusty Russell wrote:

> Hi all,
> 
> 	This patch implements a header for a linked list of scatterlist
>  arrays, rather than using an extra entry and low pointer bits to chain them
> together.  I've tested that it's sane for virtio (which uses struct
> scatterlist).
> 
> Features:
> 1) Neatens code by including length in structure.
> 2) Avoids end ambiguity by including maximum length too.
> 3) Works fine with old "sg is an array" interfaces.
> 4) Kinda icky for stack declaration, so hence a helper is created.
> 5) Lacks magic.
> 
> I reverted (most of?) the scatterlist chaining changes to create these
> patches, so it won't apply to your kernels.  The reversion patch isn't
> interesting, so I haven't posted it.
> 
> Thanks,
> Rusty.
> 
> diff --git a/include/linux/scatterlist.h b/include/linux/scatterlist.h
> index 4efbd9c..ce7e581 100644
> --- a/include/linux/scatterlist.h
> +++ b/include/linux/scatterlist.h
> @@ -5,6 +5,51 @@
>  #include <linux/mm.h>
>  #include <linux/string.h>
>  
> +/**
> + * struct sg_ring - a ring of scatterlists
> + * @list: the list_head chaining them together
> + * @num: the number of valid sg entries
> + * @max: the maximum number of sg entries (size of the sg array).
> + * @sg: the array of scatterlist entries.
> + *
> + * This provides a convenient encapsulation of one or more scatter gather
> + * arrays. */

Hi Rusty,

I don't know where these patches are going, but please put the
trailing */ in all of these kernel-doc descriptions (nice ones :)
on a separate line.  Thanks.

> +struct sg_ring
> +{
> +	struct list_head list;
> +	unsigned int num, max;
> +	struct scatterlist sg[0];
> +};
> +
> +/* This helper declares an sg ring on the stack or in a struct. */
> +#define DECLARE_SG(name, max)			\
> +	struct {				\
> +		struct sg_ring ring;		\
> +		struct scatterlist sg[max];	\
> +	} name
> +
> +/**
> + * sg_ring_init - initialize a scatterlist ring.
> + * @sg: the sg_ring.
> + * @max: the size of the trailing sg array.
> + *
> + * After initialization sg is alone in the ring. */
> +static inline void sg_ring_init(struct sg_ring *sg, unsigned int max)
> +{
> +	INIT_LIST_HEAD(&sg->list);
> +	sg->max = max;
> +}
> +
> +/**
> + * sg_ring_next - next array in a scatterlist ring.
> + * @sg: the sg_ring.
> + *
> + * After initialization sg is alone in the ring. */
> +static inline struct sg_ring *sg_ring_next(struct sg_ring *sg)
> +{
> +	return list_first_entry(&sg->list, struct sg_ring, list);
> +}
> +
>  static inline void sg_set_buf(struct scatterlist *sg, const void *buf,
>  			      unsigned int buflen)
>  {
> @@ -20,4 +65,20 @@ static inline void sg_init_one(struct scatterlist *sg, const void *buf,
>  	sg_set_buf(sg, buf, buflen);
>  }
>  
> +/**
> + * sg_init_single - initialize a one-element scatterlist ring.
> + * @sg: the sg_ring.
> + * @buf: the pointer to the buffer.
> + * @buflen: the length of the buffer.
> + *
> + * Does sg_ring_init and also sets up first (and only) sg element. */
> +static inline void sg_init_single(struct sg_ring *sg,
> +				  const void *buf,
> +				  unsigned int buflen)
> +{
> +	sg_ring_init(sg, 1);
> +	sg->num = 1;
> +	sg_init_one(&sg->sg[0], buf, buflen);
> +}
> +
>  #endif /* _LINUX_SCATTERLIST_H */

---
~Randy
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/