Fix oinkmaster patch.

[people/pmueller/ipfire-2.x.git] / src / patches / suse-2.6.27.39 / patches.fixes / block-discard-requests

From b24db85c3e8a53ff0d3255b22e8e8b674572bdbc Mon Sep 17 00:00:00 2001
From: Hannes Reinecke <hare@suse.de>
Date: Thu, 23 Oct 2008 13:53:12 +0200
Subject: [PATCH] Implement block discard

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Signed-off-by: Hannes Reinecke <hare@suse.de>
---
 block/blk-barrier.c          |   69 +++++++++++++++++++++++++++++++++++++++
 block/blk-core.c             |   34 +++++++++++++------
 block/blk-merge.c            |   27 +++++++++------
 block/blk-settings.c         |   17 +++++++++
 block/blktrace.c             |   29 ++++------------
 block/compat_ioctl.c         |    1 
 block/elevator.c             |   12 +++++-
 block/ioctl.c                |   76 +++++++++++++++++++++++++++++++++++++++++++
 drivers/mtd/ftl.c            |   24 +++++++++++++
 drivers/mtd/mtd_blkdevs.c    |   16 +++++++++
 fs/fat/fatent.c              |   14 +++++++
 include/linux/bio.h          |   10 +++--
 include/linux/blkdev.h       |   22 +++++++++++-
 include/linux/blktrace_api.h |    6 ++-
 include/linux/fs.h           |    5 ++
 include/linux/mtd/blktrans.h |    2 +
 16 files changed, 314 insertions(+), 50 deletions(-)

--- a/block/blk-barrier.c
+++ b/block/blk-barrier.c
@@ -315,3 +315,72 @@ int blkdev_issue_flush(struct block_devi
 	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(DISCARD_BARRIER, 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);
--- a/block/blk-core.c
+++ b/block/blk-core.c
@@ -1077,7 +1077,12 @@ void init_request_from_bio(struct reques
 	/*
 	 * REQ_BARRIER implies no merging, but lets make it explicit
 	 */
-	if (unlikely(bio_barrier(bio)))
+	if (unlikely(bio_discard(bio))) {
+		req->cmd_flags |= REQ_DISCARD;
+		if (bio_barrier(bio))
+			req->cmd_flags |= REQ_SOFTBARRIER;
+		req->q->prepare_discard_fn(req->q, req);
+	} else if (unlikely(bio_barrier(bio)))
 		req->cmd_flags |= (REQ_HARDBARRIER | REQ_NOMERGE);
 
 	if (bio_sync(bio))
@@ -1095,7 +1100,7 @@ void init_request_from_bio(struct reques
 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;
@@ -1110,7 +1115,14 @@ static int __make_request(struct request
 	blk_queue_bounce(q, &bio);
 
 	barrier = bio_barrier(bio);
-	if (unlikely(barrier) && (q->next_ordered == QUEUE_ORDERED_NONE)) {
+	if (unlikely(barrier) && bio_has_data(bio) &&
+	    (q->next_ordered == QUEUE_ORDERED_NONE)) {
+		err = -EOPNOTSUPP;
+		goto end_io;
+	}
+
+	discard = bio_discard(bio);
+	if (unlikely(discard) && !q->prepare_discard_fn) {
 		err = -EOPNOTSUPP;
 		goto end_io;
 	}
@@ -1405,7 +1417,8 @@ end_io:
 
 		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 +1500,6 @@ 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 +1893,7 @@ static int blk_end_io(struct request *rq
 	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 +1951,7 @@ 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 +2024,14 @@ void blk_rq_bio_prep(struct request_queu
 	   we want BIO_RW_AHEAD (bit 1) to imply REQ_FAILFAST (bit 1). */
 	rq->cmd_flags |= (bio->bi_rw & 3);
 
-	rq->nr_phys_segments = bio_phys_segments(q, bio);
-	rq->nr_hw_segments = bio_hw_segments(q, bio);
+	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;
--- a/block/blk-merge.c
+++ b/block/blk-merge.c
@@ -11,7 +11,7 @@
 
 void blk_recalc_rq_sectors(struct request *rq, int nsect)
 {
-	if (blk_fs_request(rq)) {
+	if (blk_fs_request(rq) || blk_discard_rq(rq)) {
 		rq->hard_sector += nsect;
 		rq->hard_nr_sectors -= nsect;
 
@@ -138,14 +138,18 @@ static int blk_phys_contig_segment(struc
 	if (!test_bit(QUEUE_FLAG_CLUSTER, &q->queue_flags))
 		return 0;
 
-	if (!BIOVEC_PHYS_MERGEABLE(__BVEC_END(bio), __BVEC_START(nxt)))
-		return 0;
 	if (bio->bi_seg_back_size + nxt->bi_seg_front_size >
 	    q->max_segment_size)
 		return 0;
 
+	if (!bio_has_data(bio))
+		return 1;
+
+	if (!BIOVEC_PHYS_MERGEABLE(__BVEC_END(bio), __BVEC_START(nxt)))
+		return 0;
+
 	/*
-	 * bio and nxt are contigous in memory, check if the queue allows
+	 * bio and nxt are contiguous in memory, check if the queue allows
 	 * these two to be merged into one
 	 */
 	if (BIO_SEG_BOUNDARY(q, bio, nxt))
@@ -161,8 +165,9 @@ static int blk_hw_contig_segment(struct 
 		blk_recount_segments(q, bio);
 	if (!bio_flagged(nxt, BIO_SEG_VALID))
 		blk_recount_segments(q, nxt);
-	if (!BIOVEC_VIRT_MERGEABLE(__BVEC_END(bio), __BVEC_START(nxt)) ||
-	    BIOVEC_VIRT_OVERSIZE(bio->bi_hw_back_size + nxt->bi_hw_front_size))
+	if (bio_has_data(bio) &&
+	    (!BIOVEC_VIRT_MERGEABLE(__BVEC_END(bio), __BVEC_START(nxt)) ||
+	     BIOVEC_VIRT_OVERSIZE(bio->bi_hw_back_size + nxt->bi_hw_front_size)))
 		return 0;
 	if (bio->bi_hw_back_size + nxt->bi_hw_front_size > q->max_segment_size)
 		return 0;
@@ -325,8 +330,9 @@ int ll_back_merge_fn(struct request_queu
 	if (!bio_flagged(bio, BIO_SEG_VALID))
 		blk_recount_segments(q, bio);
 	len = req->biotail->bi_hw_back_size + bio->bi_hw_front_size;
-	if (BIOVEC_VIRT_MERGEABLE(__BVEC_END(req->biotail), __BVEC_START(bio))
-	    && !BIOVEC_VIRT_OVERSIZE(len)) {
+	if (!bio_has_data(bio) ||
+	    (BIOVEC_VIRT_MERGEABLE(__BVEC_END(req->biotail), __BVEC_START(bio))
+	     && !BIOVEC_VIRT_OVERSIZE(len))) {
 		int mergeable =  ll_new_mergeable(q, req, bio);
 
 		if (mergeable) {
@@ -364,8 +370,9 @@ int ll_front_merge_fn(struct request_que
 		blk_recount_segments(q, bio);
 	if (!bio_flagged(req->bio, BIO_SEG_VALID))
 		blk_recount_segments(q, req->bio);
-	if (BIOVEC_VIRT_MERGEABLE(__BVEC_END(bio), __BVEC_START(req->bio)) &&
-	    !BIOVEC_VIRT_OVERSIZE(len)) {
+	if (!bio_has_data(bio) ||
+	    (BIOVEC_VIRT_MERGEABLE(__BVEC_END(bio), __BVEC_START(req->bio)) &&
+	     !BIOVEC_VIRT_OVERSIZE(len))) {
 		int mergeable =  ll_new_mergeable(q, req, bio);
 
 		if (mergeable) {
--- a/block/blk-settings.c
+++ b/block/blk-settings.c
@@ -33,6 +33,23 @@ void blk_queue_prep_rq(struct request_qu
 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
  * @mbfn:	merge_bvec_fn
--- a/block/blktrace.c
+++ b/block/blktrace.c
@@ -111,23 +111,9 @@ static int act_log_check(struct blk_trac
  */
 static u32 ddir_act[2] __read_mostly = { BLK_TC_ACT(BLK_TC_READ), BLK_TC_ACT(BLK_TC_WRITE) };
 
-/*
- * Bio action bits of interest
- */
-static u32 bio_act[9] __read_mostly = { 0, BLK_TC_ACT(BLK_TC_BARRIER), BLK_TC_ACT(BLK_TC_SYNC), 0, BLK_TC_ACT(BLK_TC_AHEAD), 0, 0, 0, BLK_TC_ACT(BLK_TC_META) };
-
-/*
- * More could be added as needed, taking care to increment the decrementer
- * to get correct indexing
- */
-#define trace_barrier_bit(rw)	\
-	(((rw) & (1 << BIO_RW_BARRIER)) >> (BIO_RW_BARRIER - 0))
-#define trace_sync_bit(rw)	\
-	(((rw) & (1 << BIO_RW_SYNC)) >> (BIO_RW_SYNC - 1))
-#define trace_ahead_bit(rw)	\
-	(((rw) & (1 << BIO_RW_AHEAD)) << (2 - BIO_RW_AHEAD))
-#define trace_meta_bit(rw)	\
-	(((rw) & (1 << BIO_RW_META)) >> (BIO_RW_META - 3))
+/* The ilog2() calls fall out because they're constant */
+#define MASK_TC_BIT(rw, __name) ( (rw & (1 << BIO_RW_ ## __name)) << \
+	  (ilog2(BLK_TC_ ## __name) + BLK_TC_SHIFT - BIO_RW_ ## __name) )
 
 /*
  * The worker for the various blk_add_trace*() types. Fills out a
@@ -147,10 +133,11 @@ void __blk_add_trace(struct blk_trace *b
 		return;
 
 	what |= ddir_act[rw & WRITE];
-	what |= bio_act[trace_barrier_bit(rw)];
-	what |= bio_act[trace_sync_bit(rw)];
-	what |= bio_act[trace_ahead_bit(rw)];
-	what |= bio_act[trace_meta_bit(rw)];
+	what |= MASK_TC_BIT(rw, BARRIER);
+	what |= MASK_TC_BIT(rw, SYNC);
+	what |= MASK_TC_BIT(rw, AHEAD);
+	what |= MASK_TC_BIT(rw, META);
+	what |= MASK_TC_BIT(rw, DISCARD);
 
 	pid = tsk->pid;
 	if (unlikely(act_log_check(bt, what, sector, pid)))
--- a/block/compat_ioctl.c
+++ b/block/compat_ioctl.c
@@ -788,6 +788,7 @@ long compat_blkdev_ioctl(struct file *fi
 		return compat_hdio_getgeo(disk, bdev, compat_ptr(arg));
 	case BLKFLSBUF:
 	case BLKROSET:
+	case BLKDISCARD:
 	/*
 	 * the ones below are implemented in blkdev_locked_ioctl,
 	 * but we call blkdev_ioctl, which gets the lock for us
--- a/block/elevator.c
+++ b/block/elevator.c
@@ -75,6 +75,12 @@ int elv_rq_merge_ok(struct request *rq, 
 		return 0;
 
 	/*
+	 * Don't merge file system requests and discard requests
+	 */
+	if (bio_discard(bio) != bio_discard(rq->bio))
+		return 0;
+
+	/*
 	 * different data direction or already started, don't merge
 	 */
 	if (bio_data_dir(bio) != rq_data_dir(rq))
@@ -438,6 +444,8 @@ void elv_dispatch_sort(struct request_qu
 	list_for_each_prev(entry, &q->queue_head) {
 		struct request *pos = list_entry_rq(entry);
 
+		if (blk_discard_rq(rq) != blk_discard_rq(pos))
+			break;
 		if (rq_data_dir(rq) != rq_data_dir(pos))
 			break;
 		if (pos->cmd_flags & stop_flags)
@@ -607,7 +615,7 @@ void elv_insert(struct request_queue *q,
 		break;
 
 	case ELEVATOR_INSERT_SORT:
-		BUG_ON(!blk_fs_request(rq));
+		BUG_ON(!blk_fs_request(rq) && !blk_discard_rq(rq));
 		rq->cmd_flags |= REQ_SORTED;
 		q->nr_sorted++;
 		if (rq_mergeable(rq)) {
@@ -692,7 +700,7 @@ void __elv_add_request(struct request_qu
 		 * this request is scheduling boundary, update
 		 * end_sector
 		 */
-		if (blk_fs_request(rq)) {
+		if (blk_fs_request(rq) || blk_discard_rq(rq)) {
 			q->end_sector = rq_end_sector(rq);
 			q->boundary_rq = rq;
 		}
--- a/block/ioctl.c
+++ b/block/ioctl.c
@@ -111,6 +111,69 @@ static int blkdev_reread_part(struct blo
 	return res;
 }
 
+static void blk_ioc_discard_endio(struct bio *bio, int err)
+{
+	if (err) {
+		if (err == -EOPNOTSUPP)
+			set_bit(BIO_EOPNOTSUPP, &bio->bi_flags);
+		clear_bit(BIO_UPTODATE, &bio->bi_flags);
+	}
+	complete(bio->bi_private);
+}
+
+static int blk_ioctl_discard(struct block_device *bdev, uint64_t start,
+			     uint64_t len)
+{
+	struct request_queue *q = bdev_get_queue(bdev);
+	int ret = 0;
+
+	if (start & 511)
+		return -EINVAL;
+	if (len & 511)
+		return -EINVAL;
+	start >>= 9;
+	len >>= 9;
+
+	if (start + len > (bdev->bd_inode->i_size >> 9))
+		return -EINVAL;
+
+	if (!q->prepare_discard_fn)
+		return -EOPNOTSUPP;
+
+	while (len && !ret) {
+		DECLARE_COMPLETION_ONSTACK(wait);
+		struct bio *bio;
+
+		bio = bio_alloc(GFP_KERNEL, 0);
+		if (!bio)
+			return -ENOMEM;
+
+		bio->bi_end_io = blk_ioc_discard_endio;
+		bio->bi_bdev = bdev;
+		bio->bi_private = &wait;
+		bio->bi_sector = start;
+
+		if (len > q->max_hw_sectors) {
+			bio->bi_size = q->max_hw_sectors << 9;
+			len -= q->max_hw_sectors;
+			start += q->max_hw_sectors;
+		} else {
+			bio->bi_size = len << 9;
+			len = 0;
+		}
+		submit_bio(DISCARD_NOBARRIER, bio);
+
+		wait_for_completion(&wait);
+
+		if (bio_flagged(bio, BIO_EOPNOTSUPP))
+			ret = -EOPNOTSUPP;
+		else if (!bio_flagged(bio, BIO_UPTODATE))
+			ret = -EIO;
+		bio_put(bio);
+	}
+	return ret;
+}
+
 static int put_ushort(unsigned long arg, unsigned short val)
 {
 	return put_user(val, (unsigned short __user *)arg);
@@ -258,6 +321,19 @@ int blkdev_ioctl(struct inode *inode, st
 		set_device_ro(bdev, n);
 		unlock_kernel();
 		return 0;
+
+	case BLKDISCARD: {
+		uint64_t range[2];
+
+		if (!(file->f_mode & FMODE_WRITE))
+			return -EBADF;
+
+		if (copy_from_user(range, (void __user *)arg, sizeof(range)))
+			return -EFAULT;
+
+		return blk_ioctl_discard(bdev, range[0], range[1]);
+	}
+
 	case HDIO_GETGEO: {
 		struct hd_geometry geo;
 
--- a/drivers/mtd/ftl.c
+++ b/drivers/mtd/ftl.c
@@ -1005,6 +1005,29 @@ static int ftl_writesect(struct mtd_blkt
 	return ftl_write((void *)dev, buf, block, 1);
 }
 
+static int ftl_discardsect(struct mtd_blktrans_dev *dev,
+			   unsigned long sector, unsigned nr_sects)
+{
+	partition_t *part = (void *)dev;
+	uint32_t bsize = 1 << part->header.EraseUnitSize;
+
+	DEBUG(1, "FTL erase sector %ld for %d sectors\n",
+	      sector, nr_sects);
+
+	while (nr_sects) {
+		uint32_t old_addr = part->VirtualBlockMap[sector];
+		if (old_addr != 0xffffffff) {
+			part->VirtualBlockMap[sector] = 0xffffffff;
+			part->EUNInfo[old_addr/bsize].Deleted++;
+			if (set_bam_entry(part, old_addr, 0))
+				return -EIO;
+		}
+		nr_sects--;
+		sector++;
+	}
+
+	return 0;
+}
 /*====================================================================*/
 
 static void ftl_freepart(partition_t *part)
@@ -1069,6 +1092,7 @@ static struct mtd_blktrans_ops ftl_tr = 
 	.blksize 	= SECTOR_SIZE,
 	.readsect	= ftl_readsect,
 	.writesect	= ftl_writesect,
+	.discard	= ftl_discardsect,
 	.getgeo		= ftl_getgeo,
 	.add_mtd	= ftl_add_mtd,
 	.remove_dev	= ftl_remove_dev,
--- a/drivers/mtd/mtd_blkdevs.c
+++ b/drivers/mtd/mtd_blkdevs.c
@@ -32,6 +32,14 @@ struct mtd_blkcore_priv {
 	spinlock_t queue_lock;
 };
 
+static int blktrans_discard_request(struct request_queue *q,
+				    struct request *req)
+{
+	req->cmd_type = REQ_TYPE_LINUX_BLOCK;
+	req->cmd[0] = REQ_LB_OP_DISCARD;
+	return 0;
+}
+
 static int do_blktrans_request(struct mtd_blktrans_ops *tr,
 			       struct mtd_blktrans_dev *dev,
 			       struct request *req)
@@ -44,6 +52,10 @@ static int do_blktrans_request(struct mt
 
 	buf = req->buffer;
 
+	if (req->cmd_type == REQ_TYPE_LINUX_BLOCK &&
+	    req->cmd[0] == REQ_LB_OP_DISCARD)
+		return !tr->discard(dev, block, nsect);
+
 	if (!blk_fs_request(req))
 		return 0;
 
@@ -367,6 +379,10 @@ int register_mtd_blktrans(struct mtd_blk
 
 	tr->blkcore_priv->rq->queuedata = tr;
 	blk_queue_hardsect_size(tr->blkcore_priv->rq, tr->blksize);
+	if (tr->discard)
+		blk_queue_set_discard(tr->blkcore_priv->rq,
+				      blktrans_discard_request);
+
 	tr->blkshift = ffs(tr->blksize) - 1;
 
 	tr->blkcore_priv->thread = kthread_run(mtd_blktrans_thread, tr,
--- a/fs/fat/fatent.c
+++ b/fs/fat/fatent.c
@@ -6,6 +6,7 @@
 #include <linux/module.h>
 #include <linux/fs.h>
 #include <linux/msdos_fs.h>
+#include <linux/blkdev.h>
 
 struct fatent_operations {
 	void (*ent_blocknr)(struct super_block *, int, int *, sector_t *);
@@ -535,6 +536,7 @@ int fat_free_clusters(struct inode *inod
 	struct fat_entry fatent;
 	struct buffer_head *bhs[MAX_BUF_PER_PAGE];
 	int i, err, nr_bhs;
+	int first_cl = cluster;
 
 	nr_bhs = 0;
 	fatent_init(&fatent);
@@ -551,6 +553,18 @@ int fat_free_clusters(struct inode *inod
 			goto error;
 		}
 
+		/*
+		 * Issue discard for the sectors we no longer care about,
+		 * batching contiguous clusters into one request
+		 */
+		if (cluster != fatent.entry + 1) {
+			int nr_clus = fatent.entry - first_cl + 1;
+
+			sb_issue_discard(sb, fat_clus_to_blknr(sbi, first_cl),
+					 nr_clus * sbi->sec_per_clus);
+			first_cl = cluster;
+		}
+
 		ops->ent_put(&fatent, FAT_ENT_FREE);
 		if (sbi->free_clusters != -1) {
 			sbi->free_clusters++;
--- a/include/linux/bio.h
+++ b/include/linux/bio.h
@@ -156,6 +156,8 @@ 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 */
@@ -163,6 +165,7 @@ 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
@@ -192,14 +195,15 @@ struct bio {
 #define bio_failfast(bio)	((bio)->bi_rw & (1 << BIO_RW_FAILFAST))
 #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))
+#define bio_empty_barrier(bio)	(bio_barrier(bio) && !bio_has_data(bio) && !bio_discard(bio))
 
 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)
--- a/include/linux/blkdev.h
+++ b/include/linux/blkdev.h
@@ -81,6 +81,7 @@ enum {
 	 */
 	REQ_LB_OP_EJECT	= 0x40,		/* eject request */
 	REQ_LB_OP_FLUSH = 0x41,		/* flush device */
+	REQ_LB_OP_DISCARD = 0x42,	/* discard sectors */
 };
 
 /*
@@ -89,6 +90,7 @@ 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 +113,7 @@ 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 +255,7 @@ typedef void (request_fn_proc) (struct r
 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 +311,7 @@ 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;
@@ -536,7 +541,7 @@ enum {
 #define blk_noretry_request(rq)	((rq)->cmd_flags & REQ_FAILFAST)
 #define blk_rq_started(rq)	((rq)->cmd_flags & REQ_STARTED)
 
-#define blk_account_rq(rq)	(blk_rq_started(rq) && blk_fs_request(rq))
+#define blk_account_rq(rq)	(blk_rq_started(rq) && (blk_fs_request(rq) || blk_discard_rq(rq)))
 
 #define blk_pm_suspend_request(rq)	((rq)->cmd_type == REQ_TYPE_PM_SUSPEND)
 #define blk_pm_resume_request(rq)	((rq)->cmd_type == REQ_TYPE_PM_RESUME)
@@ -546,6 +551,7 @@ 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() */
@@ -592,7 +598,8 @@ static inline void blk_clear_queue_full(
 #define RQ_NOMERGE_FLAGS	\
 	(REQ_NOMERGE | REQ_STARTED | REQ_HARDBARRIER | REQ_SOFTBARRIER)
 #define rq_mergeable(rq)	\
-	(!((rq)->cmd_flags & RQ_NOMERGE_FLAGS) && blk_fs_request((rq)))
+	(!((rq)->cmd_flags & RQ_NOMERGE_FLAGS) && \
+	 (blk_discard_rq(rq) || blk_fs_request((rq))))
 
 /*
  * q->prep_rq_fn return values
@@ -797,6 +804,7 @@ extern void blk_queue_merge_bvec(struct 
 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 **);
@@ -838,6 +846,16 @@ static inline struct request *blk_map_qu
 }
 
 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
--- a/include/linux/blktrace_api.h
+++ b/include/linux/blktrace_api.h
@@ -23,7 +23,8 @@ enum blktrace_cat {
 	BLK_TC_NOTIFY	= 1 << 10,	/* special message */
 	BLK_TC_AHEAD	= 1 << 11,	/* readahead */
 	BLK_TC_META	= 1 << 12,	/* metadata */
-	BLK_TC_DRV_DATA	= 1 << 13,	/* binary per-driver data */
+	BLK_TC_DISCARD	= 1 << 13,	/* discard requests */
+	BLK_TC_DRV_DATA	= 1 << 14,	/* binary per-drivers data */
 
 	BLK_TC_END	= 1 << 15,	/* only 16-bits, reminder */
 };
@@ -204,6 +205,9 @@ static inline void blk_add_trace_rq(stru
 	if (likely(!bt))
 		return;
 
+	if (blk_discard_rq(rq))
+		rw |= (1 << BIO_RW_DISCARD);
+
 	if (blk_pc_request(rq)) {
 		what |= BLK_TC_ACT(BLK_TC_PC);
 		__blk_add_trace(bt, 0, rq->data_len, rw, what, rq->errors, sizeof(rq->cmd), rq->cmd);
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -91,7 +91,9 @@ 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 DISCARD_NOBARRIER (1 << BIO_RW_DISCARD)
+#define DISCARD_BARRIER ((1 << BIO_RW_DISCARD) | (1 << BIO_RW_BARRIER))
 
 #define SEL_IN		1
 #define SEL_OUT		2
@@ -229,6 +231,7 @@ extern int dir_notify_enable;
 #define BLKTRACESTART _IO(0x12,116)
 #define BLKTRACESTOP _IO(0x12,117)
 #define BLKTRACETEARDOWN _IO(0x12,118)
+#define BLKDISCARD _IO(0x12,119)
 
 #define BMAP_IOCTL 1		/* obsolete - kept for compatibility */
 #define FIBMAP	   _IO(0x00,1)	/* bmap access */
--- a/include/linux/mtd/blktrans.h
+++ b/include/linux/mtd/blktrans.h
@@ -41,6 +41,8 @@ struct mtd_blktrans_ops {
 		    unsigned long block, char *buffer);
 	int (*writesect)(struct mtd_blktrans_dev *dev,
 		     unsigned long block, char *buffer);
+	int (*discard)(struct mtd_blktrans_dev *dev,
+		       unsigned long block, unsigned nr_blocks);
 
 	/* Block layer ioctls */
 	int (*getgeo)(struct mtd_blktrans_dev *dev, struct hd_geometry *geo);