Commit 20d0189b authored by Kent Overstreet's avatar Kent Overstreet
Browse files

block: Introduce new bio_split() 



The new bio_split() can split arbitrary bios - it's not restricted to
single page bios, like the old bio_split() (previously renamed to
bio_pair_split()). It also has different semantics - it doesn't allocate
a struct bio_pair, leaving it up to the caller to handle completions.

Then convert the existing bio_pair_split() users to the new bio_split()
- and also nvme, which was open coding bio splitting.

(We have to take that BUG_ON() out of bio_integrity_trim() because this
bio_split() needs to use it, and there's no reason it has to be used on
bios marked as cloned; BIO_CLONED doesn't seem to have clearly
documented semantics anyways.)

Signed-off-by: default avatarKent Overstreet <kmo@daterainc.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Martin K. Petersen <martin.petersen@oracle.com>
Cc: Matthew Wilcox <matthew.r.wilcox@intel.com>
Cc: Keith Busch <keith.busch@intel.com>
Cc: Vishal Verma <vishal.l.verma@intel.com>
Cc: Jiri Kosina <jkosina@suse.cz>
Cc: Neil Brown <neilb@suse.de>
parent ee67891b

drivers/block/nvme-core.c

+9 −97
Original line number Diff line number Diff line
@@ -441,104 +441,19 @@ int nvme_setup_prps(struct nvme_dev *dev, struct nvme_common_command *cmd, 
	return total_len;

}



struct nvme_bio_pair {

	struct bio b1, b2, *parent;

	struct bio_vec *bv1, *bv2;

	int err;

	atomic_t cnt;

};



static void nvme_bio_pair_endio(struct bio *bio, int err)

{

	struct nvme_bio_pair *bp = bio->bi_private;



	if (err)

		bp->err = err;



	if (atomic_dec_and_test(&bp->cnt)) {

		bio_endio(bp->parent, bp->err);

		kfree(bp->bv1);

		kfree(bp->bv2);

		kfree(bp);

	}

}



static struct nvme_bio_pair *nvme_bio_split(struct bio *bio, int idx,

							int len, int offset)

{

	struct nvme_bio_pair *bp;



	BUG_ON(len > bio->bi_iter.bi_size);

	BUG_ON(idx > bio->bi_vcnt);



	bp = kmalloc(sizeof(*bp), GFP_ATOMIC);

	if (!bp)

		return NULL;

	bp->err = 0;



	bp->b1 = *bio;

	bp->b2 = *bio;



	bp->b1.bi_iter.bi_size = len;

	bp->b2.bi_iter.bi_size -= len;

	bp->b1.bi_vcnt = idx;

	bp->b2.bi_iter.bi_idx = idx;

	bp->b2.bi_iter.bi_sector += len >> 9;



	if (offset) {

		bp->bv1 = kmalloc(bio->bi_max_vecs * sizeof(struct bio_vec),

								GFP_ATOMIC);

		if (!bp->bv1)

			goto split_fail_1;



		bp->bv2 = kmalloc(bio->bi_max_vecs * sizeof(struct bio_vec),

								GFP_ATOMIC);

		if (!bp->bv2)

			goto split_fail_2;



		memcpy(bp->bv1, bio->bi_io_vec,

			bio->bi_max_vecs * sizeof(struct bio_vec));

		memcpy(bp->bv2, bio->bi_io_vec,

			bio->bi_max_vecs * sizeof(struct bio_vec));



		bp->b1.bi_io_vec = bp->bv1;

		bp->b2.bi_io_vec = bp->bv2;

		bp->b2.bi_io_vec[idx].bv_offset += offset;

		bp->b2.bi_io_vec[idx].bv_len -= offset;

		bp->b1.bi_io_vec[idx].bv_len = offset;

		bp->b1.bi_vcnt++;

	} else

		bp->bv1 = bp->bv2 = NULL;



	bp->b1.bi_private = bp;

	bp->b2.bi_private = bp;



	bp->b1.bi_end_io = nvme_bio_pair_endio;

	bp->b2.bi_end_io = nvme_bio_pair_endio;



	bp->parent = bio;

	atomic_set(&bp->cnt, 2);



	return bp;



 split_fail_2:

	kfree(bp->bv1);

 split_fail_1:

	kfree(bp);

	return NULL;

}



static int nvme_split_and_submit(struct bio *bio, struct nvme_queue *nvmeq,

						int idx, int len, int offset)

				 int len)

{

	struct nvme_bio_pair *bp = nvme_bio_split(bio, idx, len, offset);

	if (!bp)

	struct bio *split = bio_split(bio, len >> 9, GFP_ATOMIC, NULL);

	if (!split)

		return -ENOMEM;



	bio_chain(split, bio);



	if (bio_list_empty(&nvmeq->sq_cong))

		add_wait_queue(&nvmeq->sq_full, &nvmeq->sq_cong_wait);

	bio_list_add(&nvmeq->sq_cong, &bp->b1);

	bio_list_add(&nvmeq->sq_cong, &bp->b2);

	bio_list_add(&nvmeq->sq_cong, split);

	bio_list_add(&nvmeq->sq_cong, bio);



	return 0;

}
@@ -568,8 +483,7 @@ static int nvme_map_bio(struct nvme_queue *nvmeq, struct nvme_iod *iod, 
		} else {

			if (!first && BIOVEC_NOT_VIRT_MERGEABLE(&bvprv, &bvec))

				return nvme_split_and_submit(bio, nvmeq,

							     iter.bi_idx,

							     length, 0);

							     length);



			sg = sg ? sg + 1 : iod->sg;

			sg_set_page(sg, bvec.bv_page,
@@ -578,9 +492,7 @@ static int nvme_map_bio(struct nvme_queue *nvmeq, struct nvme_iod *iod, 
		}



		if (split_len - length < bvec.bv_len)

			return nvme_split_and_submit(bio, nvmeq, iter.bi_idx,

						     split_len,

						     split_len - length);

			return nvme_split_and_submit(bio, nvmeq, split_len);

		length += bvec.bv_len;

		bvprv = bvec;

		first = 0;

drivers/block/pktcdvd.c

+74 −62
Original line number Diff line number Diff line
@@ -2338,26 +2338,8 @@ static void pkt_end_io_read_cloned(struct bio *bio, int err) 
	pkt_bio_finished(pd);

}



static void pkt_make_request(struct request_queue *q, struct bio *bio)

static void pkt_make_request_read(struct pktcdvd_device *pd, struct bio *bio)

{

	struct pktcdvd_device *pd;

	char b[BDEVNAME_SIZE];

	sector_t zone;

	struct packet_data *pkt;

	int was_empty, blocked_bio;

	struct pkt_rb_node *node;



	pd = q->queuedata;

	if (!pd) {

		pr_err("%s incorrect request queue\n",

		       bdevname(bio->bi_bdev, b));

		goto end_io;

	}



	/*

	 * Clone READ bios so we can have our own bi_end_io callback.

	 */

	if (bio_data_dir(bio) == READ) {

	struct bio *cloned_bio = bio_clone(bio, GFP_NOIO);

	struct packet_stacked_data *psd = mempool_alloc(psd_pool, GFP_NOIO);
@@ -2368,45 +2350,17 @@ static void pkt_make_request(struct request_queue *q, struct bio *bio) 
	cloned_bio->bi_end_io = pkt_end_io_read_cloned;

	pd->stats.secs_r += bio_sectors(bio);

	pkt_queue_bio(pd, cloned_bio);

		return;

	}



	if (!test_bit(PACKET_WRITABLE, &pd->flags)) {

		pkt_notice(pd, "WRITE for ro device (%llu)\n",

			   (unsigned long long)bio->bi_iter.bi_sector);

		goto end_io;

	}



	if (!bio->bi_iter.bi_size || (bio->bi_iter.bi_size % CD_FRAMESIZE)) {

		pkt_err(pd, "wrong bio size\n");

		goto end_io;

}



	blk_queue_bounce(q, &bio);



	zone = get_zone(bio->bi_iter.bi_sector, pd);

	pkt_dbg(2, pd, "start = %6llx stop = %6llx\n",

		(unsigned long long)bio->bi_iter.bi_sector,

		(unsigned long long)bio_end_sector(bio));



	/* Check if we have to split the bio */

static void pkt_make_request_write(struct request_queue *q, struct bio *bio)

{

		struct bio_pair *bp;

		sector_t last_zone;

		int first_sectors;

	struct pktcdvd_device *pd = q->queuedata;

	sector_t zone;

	struct packet_data *pkt;

	int was_empty, blocked_bio;

	struct pkt_rb_node *node;



		last_zone = get_zone(bio_end_sector(bio) - 1, pd);

		if (last_zone != zone) {

			BUG_ON(last_zone != zone + pd->settings.size);

			first_sectors = last_zone - bio->bi_iter.bi_sector;

			bp = bio_pair_split(bio, first_sectors);

			BUG_ON(!bp);

			pkt_make_request(q, &bp->bio1);

			pkt_make_request(q, &bp->bio2);

			bio_pair_release(bp);

			return;

		}

	}

	zone = get_zone(bio->bi_iter.bi_sector, pd);



	/*

	 * If we find a matching packet in state WAITING or READ_WAIT, we can
@@ -2480,6 +2434,64 @@ static void pkt_make_request(struct request_queue *q, struct bio *bio) 
		 */

		wake_up(&pd->wqueue);

	}

}



static void pkt_make_request(struct request_queue *q, struct bio *bio)

{

	struct pktcdvd_device *pd;

	char b[BDEVNAME_SIZE];

	struct bio *split;



	pd = q->queuedata;

	if (!pd) {

		pr_err("%s incorrect request queue\n",

		       bdevname(bio->bi_bdev, b));

		goto end_io;

	}



	pkt_dbg(2, pd, "start = %6llx stop = %6llx\n",

		(unsigned long long)bio->bi_iter.bi_sector,

		(unsigned long long)bio_end_sector(bio));



	/*

	 * Clone READ bios so we can have our own bi_end_io callback.

	 */

	if (bio_data_dir(bio) == READ) {

		pkt_make_request_read(pd, bio);

		return;

	}



	if (!test_bit(PACKET_WRITABLE, &pd->flags)) {

		pkt_notice(pd, "WRITE for ro device (%llu)\n",

			   (unsigned long long)bio->bi_iter.bi_sector);

		goto end_io;

	}



	if (!bio->bi_iter.bi_size || (bio->bi_iter.bi_size % CD_FRAMESIZE)) {

		pkt_err(pd, "wrong bio size\n");

		goto end_io;

	}



	blk_queue_bounce(q, &bio);



	do {

		sector_t zone = get_zone(bio->bi_iter.bi_sector, pd);

		sector_t last_zone = get_zone(bio_end_sector(bio) - 1, pd);



		if (last_zone != zone) {

			BUG_ON(last_zone != zone + pd->settings.size);



			split = bio_split(bio, last_zone -

					  bio->bi_iter.bi_sector,

					  GFP_NOIO, fs_bio_set);

			bio_chain(split, bio);

		} else {

			split = bio;

		}



		pkt_make_request_write(q, split);

	} while (split != bio);



	return;

end_io:

	bio_io_error(bio);

drivers/md/bcache/bcache.h

+0 −1
Original line number Diff line number Diff line
@@ -901,7 +901,6 @@ void bch_bbio_endio(struct cache_set *, struct bio *, int, const char *); 
void bch_bbio_free(struct bio *, struct cache_set *);

struct bio *bch_bbio_alloc(struct cache_set *);



struct bio *bch_bio_split(struct bio *, int, gfp_t, struct bio_set *);

void bch_generic_make_request(struct bio *, struct bio_split_pool *);

void __bch_submit_bbio(struct bio *, struct cache_set *);

void bch_submit_bbio(struct bio *, struct cache_set *, struct bkey *, unsigned);

drivers/md/bcache/io.c

+2 −80
Original line number Diff line number Diff line
@@ -11,84 +11,6 @@ 


#include <linux/blkdev.h>



/**

 * bch_bio_split - split a bio

 * @bio:	bio to split

 * @sectors:	number of sectors to split from the front of @bio

 * @gfp:	gfp mask

 * @bs:		bio set to allocate from

 *

 * Allocates and returns a new bio which represents @sectors from the start of

 * @bio, and updates @bio to represent the remaining sectors.

 *

 * If bio_sectors(@bio) was less than or equal to @sectors, returns @bio

 * unchanged.

 *

 * The newly allocated bio will point to @bio's bi_io_vec, if the split was on a

 * bvec boundry; it is the caller's responsibility to ensure that @bio is not

 * freed before the split.

 */

struct bio *bch_bio_split(struct bio *bio, int sectors,

			  gfp_t gfp, struct bio_set *bs)

{

	unsigned vcnt = 0, nbytes = sectors << 9;

	struct bio_vec bv;

	struct bvec_iter iter;

	struct bio *ret = NULL;



	BUG_ON(sectors <= 0);



	if (sectors >= bio_sectors(bio))

		return bio;



	if (bio->bi_rw & REQ_DISCARD) {

		ret = bio_alloc_bioset(gfp, 1, bs);

		if (!ret)

			return NULL;

		goto out;

	}



	bio_for_each_segment(bv, bio, iter) {

		vcnt++;



		if (nbytes <= bv.bv_len)

			break;



		nbytes -= bv.bv_len;

	}



	ret = bio_alloc_bioset(gfp, vcnt, bs);

	if (!ret)

		return NULL;



	bio_for_each_segment(bv, bio, iter) {

		ret->bi_io_vec[ret->bi_vcnt++] = bv;



		if (ret->bi_vcnt == vcnt)

			break;

	}



	ret->bi_io_vec[ret->bi_vcnt - 1].bv_len = nbytes;

out:

	ret->bi_bdev	= bio->bi_bdev;

	ret->bi_iter.bi_sector	= bio->bi_iter.bi_sector;

	ret->bi_iter.bi_size	= sectors << 9;

	ret->bi_rw	= bio->bi_rw;



	if (bio_integrity(bio)) {

		if (bio_integrity_clone(ret, bio, gfp)) {

			bio_put(ret);

			return NULL;

		}



		bio_integrity_trim(ret, 0, bio_sectors(ret));

	}



	bio_advance(bio, ret->bi_iter.bi_size);



	return ret;

}



static unsigned bch_bio_max_sectors(struct bio *bio)

{

	struct request_queue *q = bdev_get_queue(bio->bi_bdev);
@@ -172,7 +94,7 @@ void bch_generic_make_request(struct bio *bio, struct bio_split_pool *p) 
	bio_get(bio);



	do {

		n = bch_bio_split(bio, bch_bio_max_sectors(bio),

		n = bio_next_split(bio, bch_bio_max_sectors(bio),

				   GFP_NOIO, s->p->bio_split);



		n->bi_end_io	= bch_bio_submit_split_endio;

drivers/md/bcache/request.c

+6 −6
Original line number Diff line number Diff line
@@ -371,7 +371,7 @@ static void bch_data_insert_start(struct closure *cl) 
				       op->writeback))

			goto err;



		n = bch_bio_split(bio, KEY_SIZE(k), GFP_NOIO, split);

		n = bio_next_split(bio, KEY_SIZE(k), GFP_NOIO, split);



		n->bi_end_io	= bch_data_insert_endio;

		n->bi_private	= cl;
@@ -679,7 +679,7 @@ static int cache_lookup_fn(struct btree_op *op, struct btree *b, struct bkey *k) 
	if (KEY_DIRTY(k))

		s->read_dirty_data = true;



	n = bch_bio_split(bio, min_t(uint64_t, INT_MAX,

	n = bio_next_split(bio, min_t(uint64_t, INT_MAX,

				      KEY_OFFSET(k) - bio->bi_iter.bi_sector),

			   GFP_NOIO, s->d->bio_split);
@@ -920,7 +920,7 @@ static int cached_dev_cache_miss(struct btree *b, struct search *s, 
	struct bio *miss, *cache_bio;



	if (s->cache_miss || s->iop.bypass) {

		miss = bch_bio_split(bio, sectors, GFP_NOIO, s->d->bio_split);

		miss = bio_next_split(bio, sectors, GFP_NOIO, s->d->bio_split);

		ret = miss == bio ? MAP_DONE : MAP_CONTINUE;

		goto out_submit;

	}
@@ -943,7 +943,7 @@ static int cached_dev_cache_miss(struct btree *b, struct search *s, 


	s->iop.replace = true;



	miss = bch_bio_split(bio, sectors, GFP_NOIO, s->d->bio_split);

	miss = bio_next_split(bio, sectors, GFP_NOIO, s->d->bio_split);



	/* btree_search_recurse()'s btree iterator is no good anymore */

	ret = miss == bio ? MAP_DONE : -EINTR;

CRA Git | Maintained and supported by SUSTech CRA and CCSE