Add 'discard' request handling

Some block devices benefit from a hint that they can forget the contents
of certain sectors. Add basic support for this to the block core, along
with a 'blkdev_issue_discard()' helper function which issues such
requests.

The caller doesn't get to provide an end_io functio, since
blkdev_issue_discard() will automatically split the request up into
multiple bios if appropriate. Neither does the function wait for
completion -- it's expected that callers won't care about when, or even
_if_, the request completes. It's only a hint to the device anyway. By
definition, the file system doesn't _care_ about these sectors any more.

[With feedback from OGAWA Hirofumi <hirofumi@mail.parknet.co.jp> and
Jens Axboe <jens.axboe@oracle.com]
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>

block/blk-barrier.c

@@ -315,3 +315,72 @@ int blkdev_issue_flush(struct block_device *bdev, sector_t *error_sector)
 	return ret;
 }
 EXPORT_SYMBOL(blkdev_issue_flush);
+
+static void blkdev_discard_end_io(struct bio *bio, int err)
+{
+	if (err) {
+		if (err == -EOPNOTSUPP)
+			set_bit(BIO_EOPNOTSUPP, &bio->bi_flags);
+		clear_bit(BIO_UPTODATE, &bio->bi_flags);
+	}
+
+	bio_put(bio);
+}
+
+/**
+ * blkdev_issue_discard - queue a discard
+ * @bdev:	blockdev to issue discard for
+ * @sector:	start sector
+ * @nr_sects:	number of sectors to discard
+ *
+ * Description:
+ *    Issue a discard request for the sectors in question. Does not wait.
+ */
+int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
+			 unsigned nr_sects)
+{
+	struct request_queue *q;
+	struct bio *bio;
+	int ret = 0;
+
+	if (bdev->bd_disk == NULL)
+		return -ENXIO;
+
+	q = bdev_get_queue(bdev);
+	if (!q)
+		return -ENXIO;
+
+	if (!q->prepare_discard_fn)
+		return -EOPNOTSUPP;
+
+	while (nr_sects && !ret) {
+		bio = bio_alloc(GFP_KERNEL, 0);
+		if (!bio)
+			return -ENOMEM;
+
+		bio->bi_end_io = blkdev_discard_end_io;
+		bio->bi_bdev = bdev;
+
+		bio->bi_sector = sector;
+
+		if (nr_sects > q->max_hw_sectors) {
+			bio->bi_size = q->max_hw_sectors << 9;
+			nr_sects -= q->max_hw_sectors;
+			sector += q->max_hw_sectors;
+		} else {
+			bio->bi_size = nr_sects << 9;
+			nr_sects = 0;
+		}
+		bio_get(bio);
+		submit_bio(WRITE_DISCARD, bio);
+
+		/* Check if it failed immediately */
+		if (bio_flagged(bio, BIO_EOPNOTSUPP))
+			ret = -EOPNOTSUPP;
+		else if (!bio_flagged(bio, BIO_UPTODATE))
+			ret = -EIO;
+		bio_put(bio);
+	}
+	return ret;
+}
+EXPORT_SYMBOL(blkdev_issue_discard);

block/blk-core.c

@@ -1079,6 +1079,10 @@ void init_request_from_bio(struct request *req, struct bio *bio)
 	 */
 	if (unlikely(bio_barrier(bio)))
 		req->cmd_flags |= (REQ_HARDBARRIER | REQ_NOMERGE);
+	if (unlikely(bio_discard(bio))) {
+		req->cmd_flags |= (REQ_SOFTBARRIER | REQ_DISCARD);
+		req->q->prepare_discard_fn(req->q, req);
+	}
 
 	if (bio_sync(bio))
 		req->cmd_flags |= REQ_RW_SYNC;
@@ -1095,7 +1099,7 @@ void init_request_from_bio(struct request *req, struct bio *bio)
 static int __make_request(struct request_queue *q, struct bio *bio)
 {
 	struct request *req;
-	int el_ret, nr_sectors, barrier, err;
+	int el_ret, nr_sectors, barrier, discard, err;
 	const unsigned short prio = bio_prio(bio);
 	const int sync = bio_sync(bio);
 	int rw_flags;
@@ -1115,6 +1119,12 @@ static int __make_request(struct request_queue *q, struct bio *bio)
 		goto end_io;
 	}
 
+	discard = bio_discard(bio);
+	if (unlikely(discard) && !q->prepare_discard_fn) {
+		err = -EOPNOTSUPP;
+		goto end_io;
+	}
+
 	spin_lock_irq(q->queue_lock);
 
 	if (unlikely(barrier) || elv_queue_empty(q))
@@ -1405,7 +1415,8 @@ static inline void __generic_make_request(struct bio *bio)
 
 		if (bio_check_eod(bio, nr_sectors))
 			goto end_io;
-		if (bio_empty_barrier(bio) && !q->prepare_flush_fn) {
+		if ((bio_empty_barrier(bio) && !q->prepare_flush_fn) ||
+		    (bio_discard(bio) && !q->prepare_discard_fn)) {
 			err = -EOPNOTSUPP;
 			goto end_io;
 		}
@@ -1487,7 +1498,6 @@ void submit_bio(int rw, struct bio *bio)
 	 * go through the normal accounting stuff before submission.
 	 */
 	if (bio_has_data(bio)) {
-
 		if (rw & WRITE) {
 			count_vm_events(PGPGOUT, count);
 		} else {
@@ -1881,7 +1891,7 @@ static int blk_end_io(struct request *rq, int error, unsigned int nr_bytes,
 	struct request_queue *q = rq->q;
 	unsigned long flags = 0UL;
 
-	if (bio_has_data(rq->bio)) {
+	if (bio_has_data(rq->bio) || blk_discard_rq(rq)) {
 		if (__end_that_request_first(rq, error, nr_bytes))
 			return 1;
 
@@ -1939,7 +1949,7 @@ EXPORT_SYMBOL_GPL(blk_end_request);
  **/
 int __blk_end_request(struct request *rq, int error, unsigned int nr_bytes)
 {
-	if (bio_has_data(rq->bio) &&
+	if ((bio_has_data(rq->bio) || blk_discard_rq(rq)) &&
 	    __end_that_request_first(rq, error, nr_bytes))
 		return 1;
 
@@ -2012,12 +2022,14 @@ void blk_rq_bio_prep(struct request_queue *q, struct request *rq,
 	   we want BIO_RW_AHEAD (bit 1) to imply REQ_FAILFAST (bit 1). */
 	rq->cmd_flags |= (bio->bi_rw & 3);
 
+	if (bio_has_data(bio)) {
 		rq->nr_phys_segments = bio_phys_segments(q, bio);
 		rq->nr_hw_segments = bio_hw_segments(q, bio);
+		rq->buffer = bio_data(bio);
+	}
 	rq->current_nr_sectors = bio_cur_sectors(bio);
 	rq->hard_cur_sectors = rq->current_nr_sectors;
 	rq->hard_nr_sectors = rq->nr_sectors = bio_sectors(bio);
-	rq->buffer = bio_data(bio);
 	rq->data_len = bio->bi_size;
 
 	rq->bio = rq->biotail = bio;

block/blk-settings.c

@@ -32,6 +32,23 @@ void blk_queue_prep_rq(struct request_queue *q, prep_rq_fn *pfn)
 }
 EXPORT_SYMBOL(blk_queue_prep_rq);
 
+/**
+ * blk_queue_set_discard - set a discard_sectors function for queue
+ * @q:		queue
+ * @dfn:	prepare_discard function
+ *
+ * It's possible for a queue to register a discard callback which is used
+ * to transform a discard request into the appropriate type for the
+ * hardware. If none is registered, then discard requests are failed
+ * with %EOPNOTSUPP.
+ *
+ */
+void blk_queue_set_discard(struct request_queue *q, prepare_discard_fn *dfn)
+{
+	q->prepare_discard_fn = dfn;
+}
+EXPORT_SYMBOL(blk_queue_set_discard);
+
 /**
  * blk_queue_merge_bvec - set a merge_bvec function for queue
  * @q:		queue

include/linux/bio.h

@@ -149,6 +149,8 @@ struct bio {
  * bit 2 -- barrier
  * bit 3 -- fail fast, don't want low level driver retries
  * bit 4 -- synchronous I/O hint: the block layer will unplug immediately
+ * bit 5 -- metadata request
+ * bit 6 -- discard sectors
  */
 #define BIO_RW		0	/* Must match RW in req flags (blkdev.h) */
 #define BIO_RW_AHEAD	1	/* Must match FAILFAST in req flags */
@@ -156,6 +158,7 @@ struct bio {
 #define BIO_RW_FAILFAST	3
 #define BIO_RW_SYNC	4
 #define BIO_RW_META	5
+#define BIO_RW_DISCARD	6
 
 /*
  * upper 16 bits of bi_rw define the io priority of this bio
@@ -186,13 +189,14 @@ struct bio {
 #define bio_rw_ahead(bio)	((bio)->bi_rw & (1 << BIO_RW_AHEAD))
 #define bio_rw_meta(bio)	((bio)->bi_rw & (1 << BIO_RW_META))
 #define bio_empty_barrier(bio)	(bio_barrier(bio) && !bio_has_data(bio))
+#define bio_discard(bio)	((bio)->bi_rw & (1 << BIO_RW_DISCARD))
 
 static inline unsigned int bio_cur_sectors(struct bio *bio)
 {
 	if (bio->bi_vcnt)
 		return bio_iovec(bio)->bv_len >> 9;
-
-	return 0;
+	else /* dataless requests such as discard */
+		return bio->bi_size >> 9;
 }
 
 static inline void *bio_data(struct bio *bio)

include/linux/blkdev.h

@@ -89,6 +89,7 @@ enum {
 enum rq_flag_bits {
 	__REQ_RW,		/* not set, read. set, write */
 	__REQ_FAILFAST,		/* no low level driver retries */
+	__REQ_DISCARD,		/* request to discard sectors */
 	__REQ_SORTED,		/* elevator knows about this request */
 	__REQ_SOFTBARRIER,	/* may not be passed by ioscheduler */
 	__REQ_HARDBARRIER,	/* may not be passed by drive either */
@@ -111,6 +112,7 @@ enum rq_flag_bits {
 };
 
 #define REQ_RW		(1 << __REQ_RW)
+#define REQ_DISCARD	(1 << __REQ_DISCARD)
 #define REQ_FAILFAST	(1 << __REQ_FAILFAST)
 #define REQ_SORTED	(1 << __REQ_SORTED)
 #define REQ_SOFTBARRIER	(1 << __REQ_SOFTBARRIER)
@@ -252,6 +254,7 @@ typedef void (request_fn_proc) (struct request_queue *q);
 typedef int (make_request_fn) (struct request_queue *q, struct bio *bio);
 typedef int (prep_rq_fn) (struct request_queue *, struct request *);
 typedef void (unplug_fn) (struct request_queue *);
+typedef int (prepare_discard_fn) (struct request_queue *, struct request *);
 
 struct bio_vec;
 struct bvec_merge_data {
@@ -307,6 +310,7 @@ struct request_queue
 	make_request_fn		*make_request_fn;
 	prep_rq_fn		*prep_rq_fn;
 	unplug_fn		*unplug_fn;
+	prepare_discard_fn	*prepare_discard_fn;
 	merge_bvec_fn		*merge_bvec_fn;
 	prepare_flush_fn	*prepare_flush_fn;
 	softirq_done_fn		*softirq_done_fn;
@@ -546,6 +550,7 @@ enum {
 #define blk_sorted_rq(rq)	((rq)->cmd_flags & REQ_SORTED)
 #define blk_barrier_rq(rq)	((rq)->cmd_flags & REQ_HARDBARRIER)
 #define blk_fua_rq(rq)		((rq)->cmd_flags & REQ_FUA)
+#define blk_discard_rq(rq)	((rq)->cmd_flags & REQ_DISCARD)
 #define blk_bidi_rq(rq)		((rq)->next_rq != NULL)
 #define blk_empty_barrier(rq)	(blk_barrier_rq(rq) && blk_fs_request(rq) && !(rq)->hard_nr_sectors)
 /* rq->queuelist of dequeued request must be list_empty() */
@@ -796,6 +801,7 @@ extern void blk_queue_merge_bvec(struct request_queue *, merge_bvec_fn *);
 extern void blk_queue_dma_alignment(struct request_queue *, int);
 extern void blk_queue_update_dma_alignment(struct request_queue *, int);
 extern void blk_queue_softirq_done(struct request_queue *, softirq_done_fn *);
+extern void blk_queue_set_discard(struct request_queue *, prepare_discard_fn *);
 extern struct backing_dev_info *blk_get_backing_dev_info(struct block_device *bdev);
 extern int blk_queue_ordered(struct request_queue *, unsigned, prepare_flush_fn *);
 extern int blk_do_ordered(struct request_queue *, struct request **);
@@ -837,6 +843,16 @@ static inline struct request *blk_map_queue_find_tag(struct blk_queue_tag *bqt,
 }
 
 extern int blkdev_issue_flush(struct block_device *, sector_t *);
+extern int blkdev_issue_discard(struct block_device *, sector_t sector,
+				unsigned nr_sects);
+
+static inline int sb_issue_discard(struct super_block *sb,
+				   sector_t block, unsigned nr_blocks)
+{
+	block <<= (sb->s_blocksize_bits - 9);
+	nr_blocks <<= (sb->s_blocksize_bits - 9);
+	return blkdev_issue_discard(sb->s_bdev, block, nr_blocks);
+}
 
 /*
 * command filter functions

include/linux/fs.h

@@ -86,7 +86,8 @@ extern int dir_notify_enable;
 #define READ_META	(READ | (1 << BIO_RW_META))
 #define WRITE_SYNC	(WRITE | (1 << BIO_RW_SYNC))
 #define SWRITE_SYNC	(SWRITE | (1 << BIO_RW_SYNC))
-#define WRITE_BARRIER	((1 << BIO_RW) | (1 << BIO_RW_BARRIER))
+#define WRITE_BARRIER	(WRITE | (1 << BIO_RW_BARRIER))
+#define WRITE_DISCARD	(WRITE | (1 << BIO_RW_DISCARD))
 
 #define SEL_IN		1
 #define SEL_OUT		2