Re: [PATCH 7/13] i386 sg: add support for chaining scatterlists

[Jens Axboe <jens.axboe@oracle.com>]

On Thu, May 10 2007, Benny Halevy wrote:
> Jens Axboe wrote:
> > +#define sg_is_chain(sg)		((unsigned long) (sg)->page & 0x01)
> > +#define sg_chain_ptr(sg)	\
> > +	((struct scatterlist *) ((unsigned long) (sg)->page & ~0x01))
> > +
> > +/*
> > + * We overload the meaning of ->page for sg chaining. If the LSB is
> > + * set, the page member contains a pointer to the next sgtable.
> > + */
> > +static inline struct scatterlist *sg_next(struct scatterlist *sg)
> > +{
> > +	if (sg_is_chain(sg))
> > +		return sg_chain_ptr(sg);
> > +
> > +	return sg + 1;
> > +}
> 
> Jens, should sg_next ever return the sg containing the chain link?
> If sg points at the entry right before the link entry, don't we
> want to skip it?
> 
> E.g.:
> 
> static inline struct scatterlist *sg_next(struct scatterlist *sg)
> {
> 	struct scatterlist *next;
> 
> 	/* just in case, shouldn't really ever be here */
> 	if (sg_is_chain(sg))
> 		return sg_chain_ptr(sg);
> 
> 	next = sg + 1;
> 
> 	if (sg_is_chain(next))
> 		return sg_chain_ptr(next);
> 
> 	return next;
> }

You are right, I got that mixed up, I changed the setup to overloading
->page right before posting it. So:

static inline struct scatterlist *sg_next(struct scatterlist *sg)
{
        sg++;

        if (unlikely(sg_is_chain(sg)))
                sg = sg_chain_ptr(sg);

        return sg;
}

-- 
Jens Axboe