On Wed, Feb 02 2005, Andrew Morton wrote:
> Dave Olien <[EMAIL PROTECTED]> wrote:
> >
> > +extern inline void zero_fill_bio(struct bio *bio)
> > +{
> > + unsigned long flags;
> > + struct bio_vec *bv;
> > + int i;
> > +
> > + bio_for_each_segment(bv, bio, i) {
> > + char *data = bvec_kmap_irq(bv, &flags);
> > + memset(data, 0, bv->bv_len);
> > + flush_dcache_page(bv->bv_page);
> > + bvec_kunmap_irq(data, &flags);
> > + }
> > +}
>
> heavens. Why was this made inline? And extern inline?
>
> It's too big for inlining (and is super-slow anyway) and will cause all
> sorts of unpleasant header file dependencies for all architectures. bio.h
> now needs to see the implementation of everyone's flush_dcache_page(), for
> example.
>
>
> Something like this?
>
> ---
> 25/include/linux/bio.h~add-local-bio-pool-support-and-modify-dm-uninline-zero_fill_bio
> 2005-02-02 18:17:18.225901376 -0800
> +++ 25-akpm/include/linux/bio.h 2005-02-02 18:17:18.230900616 -0800
> @@ -286,6 +286,7 @@ extern void bio_set_pages_dirty(struct b
> extern void bio_check_pages_dirty(struct bio *bio);
> extern struct bio *bio_copy_user(struct request_queue *, unsigned long,
> unsigned int, int);
> extern int bio_uncopy_user(struct bio *);
> +void zero_fill_bio(struct bio *bio);
>
> #ifdef CONFIG_HIGHMEM
> /*
> @@ -335,18 +336,4 @@ extern inline char *__bio_kmap_irq(struc
> __bio_kmap_irq((bio), (bio)->bi_idx, (flags))
> #define bio_kunmap_irq(buf,flags) __bio_kunmap_irq(buf, flags)
>
> -extern inline void zero_fill_bio(struct bio *bio)
> -{
> - unsigned long flags;
> - struct bio_vec *bv;
> - int i;
> -
> - bio_for_each_segment(bv, bio, i) {
> - char *data = bvec_kmap_irq(bv, &flags);
> - memset(data, 0, bv->bv_len);
> - flush_dcache_page(bv->bv_page);
> - bvec_kunmap_irq(data, &flags);
> - }
> -}
> -
> #endif /* __LINUX_BIO_H */
> diff -puN
> fs/bio.c~add-local-bio-pool-support-and-modify-dm-uninline-zero_fill_bio
> fs/bio.c
> ---
> 25/fs/bio.c~add-local-bio-pool-support-and-modify-dm-uninline-zero_fill_bio
> 2005-02-02 18:17:18.227901072 -0800
> +++ 25-akpm/fs/bio.c 2005-02-02 18:17:18.231900464 -0800
> @@ -182,6 +182,21 @@ struct bio *bio_alloc(int gfp_mask, int
> return bio_alloc_bioset(gfp_mask, nr_iovecs, fs_bio_set);
> }
>
> +void zero_fill_bio(struct bio *bio)
> +{
> + unsigned long flags;
> + struct bio_vec *bv;
> + int i;
> +
> + bio_for_each_segment(bv, bio, i) {
> + char *data = bvec_kmap_irq(bv, &flags);
> + memset(data, 0, bv->bv_len);
> + flush_dcache_page(bv->bv_page);
> + bvec_kunmap_irq(data, &flags);
> + }
> +}
> +EXPORT_SYMBOL(zero_fill_bio);
> +
> /**
> * bio_put - release a reference to a bio
> * @bio: bio to release reference to
> _
Yep looks good, thanks Andrew.
--
Jens Axboe
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
