Merge branch 'stable/for-jens-3.2' of git://oss.oracle.com/git/kwilk/xen into for-3.2/drivers

drivers/block/xen-blkback/blkback.c

@@ -39,6 +39,9 @@
 #include <linux/list.h>
 #include <linux/delay.h>
 #include <linux/freezer.h>
+#include <linux/loop.h>
+#include <linux/falloc.h>
+#include <linux/fs.h>
 
 #include <xen/events.h>
 #include <xen/page.h>
@@ -258,13 +261,16 @@ irqreturn_t xen_blkif_be_int(int irq, void *dev_id)
 
 static void print_stats(struct xen_blkif *blkif)
 {
-	pr_info("xen-blkback (%s): oo %3d  |  rd %4d  |  wr %4d  |  f %4d\n",
+	pr_info("xen-blkback (%s): oo %3d  |  rd %4d  |  wr %4d  |  f %4d"
+		 "  |  ds %4d\n",
 		 current->comm, blkif->st_oo_req,
-		 blkif->st_rd_req, blkif->st_wr_req, blkif->st_f_req);
+		 blkif->st_rd_req, blkif->st_wr_req,
+		 blkif->st_f_req, blkif->st_ds_req);
 	blkif->st_print = jiffies + msecs_to_jiffies(10 * 1000);
 	blkif->st_rd_req = 0;
 	blkif->st_wr_req = 0;
 	blkif->st_oo_req = 0;
+	blkif->st_ds_req = 0;
 }
 
 int xen_blkif_schedule(void *arg)
@@ -410,6 +416,59 @@ static int xen_blkbk_map(struct blkif_request *req,
 	return ret;
 }
 
+static void xen_blk_discard(struct xen_blkif *blkif, struct blkif_request *req)
+{
+	int err = 0;
+	int status = BLKIF_RSP_OKAY;
+	struct block_device *bdev = blkif->vbd.bdev;
+
+	if (blkif->blk_backend_type == BLKIF_BACKEND_PHY)
+		/* just forward the discard request */
+		err = blkdev_issue_discard(bdev,
+				req->u.discard.sector_number,
+				req->u.discard.nr_sectors,
+				GFP_KERNEL, 0);
+	else if (blkif->blk_backend_type == BLKIF_BACKEND_FILE) {
+		/* punch a hole in the backing file */
+		struct loop_device *lo = bdev->bd_disk->private_data;
+		struct file *file = lo->lo_backing_file;
+
+		if (file->f_op->fallocate)
+			err = file->f_op->fallocate(file,
+				FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE,
+				req->u.discard.sector_number << 9,
+				req->u.discard.nr_sectors << 9);
+		else
+			err = -EOPNOTSUPP;
+	} else
+		err = -EOPNOTSUPP;
+
+	if (err == -EOPNOTSUPP) {
+		pr_debug(DRV_PFX "discard op failed, not supported\n");
+		status = BLKIF_RSP_EOPNOTSUPP;
+	} else if (err)
+		status = BLKIF_RSP_ERROR;
+
+	make_response(blkif, req->id, req->operation, status);
+}
+
+static void xen_blk_drain_io(struct xen_blkif *blkif)
+{
+	atomic_set(&blkif->drain, 1);
+	do {
+		/* The initial value is one, and one refcnt taken at the
+		 * start of the xen_blkif_schedule thread. */
+		if (atomic_read(&blkif->refcnt) <= 2)
+			break;
+		wait_for_completion_interruptible_timeout(
+				&blkif->drain_complete, HZ);
+
+		if (!atomic_read(&blkif->drain))
+			break;
+	} while (!kthread_should_stop());
+	atomic_set(&blkif->drain, 0);
+}
+
 /*
  * Completion callback on the bio's. Called as bh->b_end_io()
  */
@@ -422,6 +481,11 @@ static void __end_block_io_op(struct pending_req *pending_req, int error)
 		pr_debug(DRV_PFX "flush diskcache op failed, not supported\n");
 		xen_blkbk_flush_diskcache(XBT_NIL, pending_req->blkif->be, 0);
 		pending_req->status = BLKIF_RSP_EOPNOTSUPP;
+	} else if ((pending_req->operation == BLKIF_OP_WRITE_BARRIER) &&
+		    (error == -EOPNOTSUPP)) {
+		pr_debug(DRV_PFX "write barrier op failed, not supported\n");
+		xen_blkbk_barrier(XBT_NIL, pending_req->blkif->be, 0);
+		pending_req->status = BLKIF_RSP_EOPNOTSUPP;
 	} else if (error) {
 		pr_debug(DRV_PFX "Buffer not up-to-date at end of operation,"
 			 " error=%d\n", error);
@@ -438,6 +502,10 @@ static void __end_block_io_op(struct pending_req *pending_req, int error)
 		make_response(pending_req->blkif, pending_req->id,
 			      pending_req->operation, pending_req->status);
 		xen_blkif_put(pending_req->blkif);
+		if (atomic_read(&pending_req->blkif->refcnt) <= 2) {
+			if (atomic_read(&pending_req->blkif->drain))
+				complete(&pending_req->blkif->drain_complete);
+		}
 		free_req(pending_req);
 	}
 }
@@ -532,7 +600,6 @@ do_block_io_op(struct xen_blkif *blkif)
 
 	return more_to_do;
 }
-
 /*
  * Transmutation of the 'struct blkif_request' to a proper 'struct bio'
  * and call the 'submit_bio' to pass it to the underlying storage.
@@ -549,6 +616,7 @@ static int dispatch_rw_block_io(struct xen_blkif *blkif,
 	int i, nbio = 0;
 	int operation;
 	struct blk_plug plug;
+	bool drain = false;
 
 	switch (req->operation) {
 	case BLKIF_OP_READ:
@@ -559,11 +627,16 @@ static int dispatch_rw_block_io(struct xen_blkif *blkif,
 		blkif->st_wr_req++;
 		operation = WRITE_ODIRECT;
 		break;
+	case BLKIF_OP_WRITE_BARRIER:
+		drain = true;
 	case BLKIF_OP_FLUSH_DISKCACHE:
 		blkif->st_f_req++;
 		operation = WRITE_FLUSH;
 		break;
-	case BLKIF_OP_WRITE_BARRIER:
+	case BLKIF_OP_DISCARD:
+		blkif->st_ds_req++;
+		operation = REQ_DISCARD;
+		break;
 	default:
 		operation = 0; /* make gcc happy */
 		goto fail_response;
@@ -572,7 +645,8 @@ static int dispatch_rw_block_io(struct xen_blkif *blkif,
 
 	/* Check that the number of segments is sane. */
 	nseg = req->nr_segments;
-	if (unlikely(nseg == 0 && operation != WRITE_FLUSH) ||
+	if (unlikely(nseg == 0 && operation != WRITE_FLUSH &&
+				operation != REQ_DISCARD) ||
 	    unlikely(nseg > BLKIF_MAX_SEGMENTS_PER_REQUEST)) {
 		pr_debug(DRV_PFX "Bad number of segments in request (%d)\n",
 			 nseg);
@@ -621,16 +695,25 @@ static int dispatch_rw_block_io(struct xen_blkif *blkif,
 		}
 	}
 
+	/* Wait on all outstanding I/O's and once that has been completed
+	 * issue the WRITE_FLUSH.
+	 */
+	if (drain)
+		xen_blk_drain_io(pending_req->blkif);
+
 	/*
 	 * If we have failed at this point, we need to undo the M2P override,
 	 * set gnttab_set_unmap_op on all of the grant references and perform
 	 * the hypercall to unmap the grants - that is all done in
 	 * xen_blkbk_unmap.
 	 */
-	if (xen_blkbk_map(req, pending_req, seg))
+	if (operation != REQ_DISCARD && xen_blkbk_map(req, pending_req, seg))
 		goto fail_flush;
 
-	/* This corresponding xen_blkif_put is done in __end_block_io_op */
+	/*
+	 * This corresponding xen_blkif_put is done in __end_block_io_op, or
+	 * below (in "!bio") if we are handling a BLKIF_OP_DISCARD.
+	 */
 	xen_blkif_get(blkif);
 
 	for (i = 0; i < nseg; i++) {
@@ -654,18 +737,25 @@ static int dispatch_rw_block_io(struct xen_blkif *blkif,
 		preq.sector_number += seg[i].nsec;
 	}
 
-	/* This will be hit if the operation was a flush. */
+	/* This will be hit if the operation was a flush or discard. */
 	if (!bio) {
-		BUG_ON(operation != WRITE_FLUSH);
+		BUG_ON(operation != WRITE_FLUSH && operation != REQ_DISCARD);
 
-		bio = bio_alloc(GFP_KERNEL, 0);
-		if (unlikely(bio == NULL))
-			goto fail_put_bio;
+		if (operation == WRITE_FLUSH) {
+			bio = bio_alloc(GFP_KERNEL, 0);
+			if (unlikely(bio == NULL))
+				goto fail_put_bio;
 
-		biolist[nbio++] = bio;
-		bio->bi_bdev    = preq.bdev;
-		bio->bi_private = pending_req;
-		bio->bi_end_io  = end_block_io_op;
+			biolist[nbio++] = bio;
+			bio->bi_bdev    = preq.bdev;
+			bio->bi_private = pending_req;
+			bio->bi_end_io  = end_block_io_op;
+		} else if (operation == REQ_DISCARD) {
+			xen_blk_discard(blkif, req);
+			xen_blkif_put(blkif);
+			free_req(pending_req);
+			return 0;
+		}
 	}
 
 	/*
@@ -685,7 +775,7 @@ static int dispatch_rw_block_io(struct xen_blkif *blkif,
 
 	if (operation == READ)
 		blkif->st_rd_sect += preq.nr_sects;
-	else if (operation == WRITE || operation == WRITE_FLUSH)
+	else if (operation & WRITE)
 		blkif->st_wr_sect += preq.nr_sects;
 
 	return 0;
@@ -765,9 +855,9 @@ static int __init xen_blkif_init(void)
 
 	mmap_pages = xen_blkif_reqs * BLKIF_MAX_SEGMENTS_PER_REQUEST;
 
-	blkbk->pending_reqs          = kmalloc(sizeof(blkbk->pending_reqs[0]) *
+	blkbk->pending_reqs          = kzalloc(sizeof(blkbk->pending_reqs[0]) *
 					xen_blkif_reqs, GFP_KERNEL);
-	blkbk->pending_grant_handles = kzalloc(sizeof(blkbk->pending_grant_handles[0]) *
+	blkbk->pending_grant_handles = kmalloc(sizeof(blkbk->pending_grant_handles[0]) *
 					mmap_pages, GFP_KERNEL);
 	blkbk->pending_pages         = kzalloc(sizeof(blkbk->pending_pages[0]) *
 					mmap_pages, GFP_KERNEL);
@@ -790,8 +880,6 @@ static int __init xen_blkif_init(void)
 	if (rc)
 		goto failed_init;
 
-	memset(blkbk->pending_reqs, 0, sizeof(blkbk->pending_reqs));
-
 	INIT_LIST_HEAD(&blkbk->pending_free);
 	spin_lock_init(&blkbk->pending_free_lock);
 	init_waitqueue_head(&blkbk->pending_free_wq);

drivers/block/xen-blkback/common.h

@@ -46,7 +46,7 @@
 
 #define DRV_PFX "xen-blkback:"
 #define DPRINTK(fmt, args...)				\
-	pr_debug(DRV_PFX "(%s:%d) " fmt ".\n",	\
+	pr_debug(DRV_PFX "(%s:%d) " fmt ".\n",		\
 		 __func__, __LINE__, ##args)
 
 
@@ -63,13 +63,26 @@ struct blkif_common_response {
 
 /* i386 protocol version */
 #pragma pack(push, 4)
+
+struct blkif_x86_32_request_rw {
+	blkif_sector_t sector_number;/* start sector idx on disk (r/w only)  */
+	struct blkif_request_segment seg[BLKIF_MAX_SEGMENTS_PER_REQUEST];
+};
+
+struct blkif_x86_32_request_discard {
+	blkif_sector_t sector_number;/* start sector idx on disk (r/w only)  */
+	uint64_t nr_sectors;
+};
+
 struct blkif_x86_32_request {
 	uint8_t        operation;    /* BLKIF_OP_???                         */
 	uint8_t        nr_segments;  /* number of segments                   */
 	blkif_vdev_t   handle;       /* only for read/write requests         */
 	uint64_t       id;           /* private guest value, echoed in resp  */
-	blkif_sector_t sector_number;/* start sector idx on disk (r/w only)  */
-	struct blkif_request_segment seg[BLKIF_MAX_SEGMENTS_PER_REQUEST];
+	union {
+		struct blkif_x86_32_request_rw rw;
+		struct blkif_x86_32_request_discard discard;
+	} u;
 };
 struct blkif_x86_32_response {
 	uint64_t        id;              /* copied from request */
@@ -79,13 +92,26 @@ struct blkif_x86_32_response {
 #pragma pack(pop)
 
 /* x86_64 protocol version */
+
+struct blkif_x86_64_request_rw {
+	blkif_sector_t sector_number;/* start sector idx on disk (r/w only)  */
+	struct blkif_request_segment seg[BLKIF_MAX_SEGMENTS_PER_REQUEST];
+};
+
+struct blkif_x86_64_request_discard {
+	blkif_sector_t sector_number;/* start sector idx on disk (r/w only)  */
+	uint64_t nr_sectors;
+};
+
 struct blkif_x86_64_request {
 	uint8_t        operation;    /* BLKIF_OP_???                         */
 	uint8_t        nr_segments;  /* number of segments                   */
 	blkif_vdev_t   handle;       /* only for read/write requests         */
 	uint64_t       __attribute__((__aligned__(8))) id;
-	blkif_sector_t sector_number;/* start sector idx on disk (r/w only)  */
-	struct blkif_request_segment seg[BLKIF_MAX_SEGMENTS_PER_REQUEST];
+	union {
+		struct blkif_x86_64_request_rw rw;
+		struct blkif_x86_64_request_discard discard;
+	} u;
 };
 struct blkif_x86_64_response {
 	uint64_t       __attribute__((__aligned__(8))) id;
@@ -113,6 +139,11 @@ enum blkif_protocol {
 	BLKIF_PROTOCOL_X86_64 = 3,
 };
 
+enum blkif_backend_type {
+	BLKIF_BACKEND_PHY  = 1,
+	BLKIF_BACKEND_FILE = 2,
+};
+
 struct xen_vbd {
 	/* What the domain refers to this vbd as. */
 	blkif_vdev_t		handle;
@@ -138,6 +169,7 @@ struct xen_blkif {
 	unsigned int		irq;
 	/* Comms information. */
 	enum blkif_protocol	blk_protocol;
+	enum blkif_backend_type blk_backend_type;
 	union blkif_back_rings	blk_rings;
 	struct vm_struct	*blk_ring_area;
 	/* The VBD attached to this interface. */
@@ -149,6 +181,9 @@ struct xen_blkif {
 	atomic_t		refcnt;
 
 	wait_queue_head_t	wq;
+	/* for barrier (drain) requests */
+	struct completion	drain_complete;
+	atomic_t		drain;
 	/* One thread per one blkif. */
 	struct task_struct	*xenblkd;
 	unsigned int		waiting_reqs;
@@ -159,6 +194,7 @@ struct xen_blkif {
 	int			st_wr_req;
 	int			st_oo_req;
 	int			st_f_req;
+	int			st_ds_req;
 	int			st_rd_sect;
 	int			st_wr_sect;
 
@@ -182,7 +218,7 @@ struct xen_blkif {
 
 struct phys_req {
 	unsigned short		dev;
-	unsigned short		nr_sects;
+	blkif_sector_t		nr_sects;
 	struct block_device	*bdev;
 	blkif_sector_t		sector_number;
 };
@@ -196,6 +232,8 @@ int xen_blkif_schedule(void *arg);
 int xen_blkbk_flush_diskcache(struct xenbus_transaction xbt,
 			      struct backend_info *be, int state);
 
+int xen_blkbk_barrier(struct xenbus_transaction xbt,
+		      struct backend_info *be, int state);
 struct xenbus_device *xen_blkbk_xenbus(struct backend_info *be);
 
 static inline void blkif_get_x86_32_req(struct blkif_request *dst,
@@ -206,12 +244,25 @@ static inline void blkif_get_x86_32_req(struct blkif_request *dst,
 	dst->nr_segments = src->nr_segments;
 	dst->handle = src->handle;
 	dst->id = src->id;
-	dst->u.rw.sector_number = src->sector_number;
-	barrier();
-	if (n > dst->nr_segments)
-		n = dst->nr_segments;
-	for (i = 0; i < n; i++)
-		dst->u.rw.seg[i] = src->seg[i];
+	switch (src->operation) {
+	case BLKIF_OP_READ:
+	case BLKIF_OP_WRITE:
+	case BLKIF_OP_WRITE_BARRIER:
+	case BLKIF_OP_FLUSH_DISKCACHE:
+		dst->u.rw.sector_number = src->u.rw.sector_number;
+		barrier();
+		if (n > dst->nr_segments)
+			n = dst->nr_segments;
+		for (i = 0; i < n; i++)
+			dst->u.rw.seg[i] = src->u.rw.seg[i];
+		break;
+	case BLKIF_OP_DISCARD:
+		dst->u.discard.sector_number = src->u.discard.sector_number;
+		dst->u.discard.nr_sectors = src->u.discard.nr_sectors;
+		break;
+	default:
+		break;
+	}
 }
 
 static inline void blkif_get_x86_64_req(struct blkif_request *dst,
@@ -222,12 +273,25 @@ static inline void blkif_get_x86_64_req(struct blkif_request *dst,
 	dst->nr_segments = src->nr_segments;
 	dst->handle = src->handle;
 	dst->id = src->id;
-	dst->u.rw.sector_number = src->sector_number;
-	barrier();
-	if (n > dst->nr_segments)
-		n = dst->nr_segments;
-	for (i = 0; i < n; i++)
-		dst->u.rw.seg[i] = src->seg[i];
+	switch (src->operation) {
+	case BLKIF_OP_READ:
+	case BLKIF_OP_WRITE:
+	case BLKIF_OP_WRITE_BARRIER:
+	case BLKIF_OP_FLUSH_DISKCACHE:
+		dst->u.rw.sector_number = src->u.rw.sector_number;
+		barrier();
+		if (n > dst->nr_segments)
+			n = dst->nr_segments;
+		for (i = 0; i < n; i++)
+			dst->u.rw.seg[i] = src->u.rw.seg[i];
+		break;
+	case BLKIF_OP_DISCARD:
+		dst->u.discard.sector_number = src->u.discard.sector_number;
+		dst->u.discard.nr_sectors = src->u.discard.nr_sectors;
+		break;
+	default:
+		break;
+	}
 }
 
 #endif /* __XEN_BLKIF__BACKEND__COMMON_H__ */

drivers/block/xen-blkback/xenbus.c

@@ -114,6 +114,8 @@ static struct xen_blkif *xen_blkif_alloc(domid_t domid)
 	spin_lock_init(&blkif->blk_ring_lock);
 	atomic_set(&blkif->refcnt, 1);
 	init_waitqueue_head(&blkif->wq);
+	init_completion(&blkif->drain_complete);
+	atomic_set(&blkif->drain, 0);
 	blkif->st_print = jiffies;
 	init_waitqueue_head(&blkif->waiting_to_free);
 
@@ -272,6 +274,7 @@ VBD_SHOW(oo_req,  "%d\n", be->blkif->st_oo_req);
 VBD_SHOW(rd_req,  "%d\n", be->blkif->st_rd_req);
 VBD_SHOW(wr_req,  "%d\n", be->blkif->st_wr_req);
 VBD_SHOW(f_req,  "%d\n", be->blkif->st_f_req);
+VBD_SHOW(ds_req,  "%d\n", be->blkif->st_ds_req);
 VBD_SHOW(rd_sect, "%d\n", be->blkif->st_rd_sect);
 VBD_SHOW(wr_sect, "%d\n", be->blkif->st_wr_sect);
 
@@ -280,6 +283,7 @@ static struct attribute *xen_vbdstat_attrs[] = {
 	&dev_attr_rd_req.attr,
 	&dev_attr_wr_req.attr,
 	&dev_attr_f_req.attr,
+	&dev_attr_ds_req.attr,
 	&dev_attr_rd_sect.attr,
 	&dev_attr_wr_sect.attr,
 	NULL
@@ -419,6 +423,73 @@ int xen_blkbk_flush_diskcache(struct xenbus_transaction xbt,
 	return err;
 }
 
+int xen_blkbk_discard(struct xenbus_transaction xbt, struct backend_info *be)
+{
+	struct xenbus_device *dev = be->dev;
+	struct xen_blkif *blkif = be->blkif;
+	char *type;
+	int err;
+	int state = 0;
+
+	type = xenbus_read(XBT_NIL, dev->nodename, "type", NULL);
+	if (!IS_ERR(type)) {
+		if (strncmp(type, "file", 4) == 0) {
+			state = 1;
+			blkif->blk_backend_type = BLKIF_BACKEND_FILE;
+		}
+		if (strncmp(type, "phy", 3) == 0) {
+			struct block_device *bdev = be->blkif->vbd.bdev;
+			struct request_queue *q = bdev_get_queue(bdev);
+			if (blk_queue_discard(q)) {
+				err = xenbus_printf(xbt, dev->nodename,
+					"discard-granularity", "%u",
+					q->limits.discard_granularity);
+				if (err) {
+					xenbus_dev_fatal(dev, err,
+						"writing discard-granularity");
+					goto kfree;
+				}
+				err = xenbus_printf(xbt, dev->nodename,
+					"discard-alignment", "%u",
+					q->limits.discard_alignment);
+				if (err) {
+					xenbus_dev_fatal(dev, err,
+						"writing discard-alignment");
+					goto kfree;
+				}
+				state = 1;
+				blkif->blk_backend_type = BLKIF_BACKEND_PHY;
+			}
+		}
+	} else {
+		err = PTR_ERR(type);
+		xenbus_dev_fatal(dev, err, "reading type");
+		goto out;
+	}
+
+	err = xenbus_printf(xbt, dev->nodename, "feature-discard",
+			    "%d", state);
+	if (err)
+		xenbus_dev_fatal(dev, err, "writing feature-discard");
+kfree:
+	kfree(type);
+out:
+	return err;
+}
+int xen_blkbk_barrier(struct xenbus_transaction xbt,
+		      struct backend_info *be, int state)
+{
+	struct xenbus_device *dev = be->dev;
+	int err;
+
+	err = xenbus_printf(xbt, dev->nodename, "feature-barrier",
+			    "%d", state);
+	if (err)
+		xenbus_dev_fatal(dev, err, "writing feature-barrier");
+
+	return err;
+}
+
 /*
  * Entry point to this code when a new device is created.  Allocate the basic
  * structures, and watch the store waiting for the hotplug scripts to tell us
@@ -590,7 +661,7 @@ static void frontend_changed(struct xenbus_device *dev,
 
 		/*
 		 * Enforce precondition before potential leak point.
-		 * blkif_disconnect() is idempotent.
+		 * xen_blkif_disconnect() is idempotent.
 		 */
 		xen_blkif_disconnect(be->blkif);
 
@@ -611,7 +682,7 @@ static void frontend_changed(struct xenbus_device *dev,
 			break;
 		/* fall through if not online */
 	case XenbusStateUnknown:
-		/* implies blkif_disconnect() via blkback_remove() */
+		/* implies xen_blkif_disconnect() via xen_blkbk_remove() */
 		device_unregister(&dev->dev);
 		break;
 
@@ -650,6 +721,11 @@ static void connect(struct backend_info *be)
 	if (err)
 		goto abort;
 
+	err = xen_blkbk_discard(xbt, be);
+
+	/* If we can't advertise it is OK. */
+	err = xen_blkbk_barrier(xbt, be, be->blkif->vbd.flush_support);
+
 	err = xenbus_printf(xbt, dev->nodename, "sectors", "%llu",
 			    (unsigned long long)vbd_sz(&be->blkif->vbd));
 	if (err) {

drivers/block/xen-blkfront.c

@@ -98,6 +98,9 @@ struct blkfront_info
 	unsigned long shadow_free;
 	unsigned int feature_flush;
 	unsigned int flush_op;
+	unsigned int feature_discard;
+	unsigned int discard_granularity;
+	unsigned int discard_alignment;
 	int is_ready;
 };
 
@@ -302,29 +305,36 @@ static int blkif_queue_request(struct request *req)
 		ring_req->operation = info->flush_op;
 	}
 
-	ring_req->nr_segments = blk_rq_map_sg(req->q, req, info->sg);
-	BUG_ON(ring_req->nr_segments > BLKIF_MAX_SEGMENTS_PER_REQUEST);
+	if (unlikely(req->cmd_flags & REQ_DISCARD)) {
+		/* id, sector_number and handle are set above. */
+		ring_req->operation = BLKIF_OP_DISCARD;
+		ring_req->nr_segments = 0;
+		ring_req->u.discard.nr_sectors = blk_rq_sectors(req);
+	} else {
+		ring_req->nr_segments = blk_rq_map_sg(req->q, req, info->sg);
+		BUG_ON(ring_req->nr_segments > BLKIF_MAX_SEGMENTS_PER_REQUEST);
 
-	for_each_sg(info->sg, sg, ring_req->nr_segments, i) {
-		buffer_mfn = pfn_to_mfn(page_to_pfn(sg_page(sg)));
-		fsect = sg->offset >> 9;
-		lsect = fsect + (sg->length >> 9) - 1;
-		/* install a grant reference. */
-		ref = gnttab_claim_grant_reference(&gref_head);
-		BUG_ON(ref == -ENOSPC);
+		for_each_sg(info->sg, sg, ring_req->nr_segments, i) {
+			buffer_mfn = pfn_to_mfn(page_to_pfn(sg_page(sg)));
+			fsect = sg->offset >> 9;
+			lsect = fsect + (sg->length >> 9) - 1;
+			/* install a grant reference. */
+			ref = gnttab_claim_grant_reference(&gref_head);
+			BUG_ON(ref == -ENOSPC);
 
-		gnttab_grant_foreign_access_ref(
-				ref,
-				info->xbdev->otherend_id,
-				buffer_mfn,
-				rq_data_dir(req) );
-
-		info->shadow[id].frame[i] = mfn_to_pfn(buffer_mfn);
-		ring_req->u.rw.seg[i] =
-				(struct blkif_request_segment) {
-					.gref       = ref,
-					.first_sect = fsect,
-					.last_sect  = lsect };
+			gnttab_grant_foreign_access_ref(
+					ref,
+					info->xbdev->otherend_id,
+					buffer_mfn,
+					rq_data_dir(req));
+
+			info->shadow[id].frame[i] = mfn_to_pfn(buffer_mfn);
+			ring_req->u.rw.seg[i] =
+					(struct blkif_request_segment) {
+						.gref       = ref,
+						.first_sect = fsect,
+						.last_sect  = lsect };
+		}
 	}
 
 	info->ring.req_prod_pvt++;
@@ -370,7 +380,9 @@ static void do_blkif_request(struct request_queue *rq)
 
 		blk_start_request(req);
 
-		if (req->cmd_type != REQ_TYPE_FS) {
+		if ((req->cmd_type != REQ_TYPE_FS) ||
+		    ((req->cmd_flags & (REQ_FLUSH | REQ_FUA)) &&
+		    !info->flush_op)) {
 			__blk_end_request_all(req, -EIO);
 			continue;
 		}
@@ -399,6 +411,7 @@ static void do_blkif_request(struct request_queue *rq)
 static int xlvbd_init_blk_queue(struct gendisk *gd, u16 sector_size)
 {
 	struct request_queue *rq;
+	struct blkfront_info *info = gd->private_data;
 
 	rq = blk_init_queue(do_blkif_request, &blkif_io_lock);
 	if (rq == NULL)
@@ -406,6 +419,13 @@ static int xlvbd_init_blk_queue(struct gendisk *gd, u16 sector_size)
 
 	queue_flag_set_unlocked(QUEUE_FLAG_VIRT, rq);
 
+	if (info->feature_discard) {
+		queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, rq);
+		blk_queue_max_discard_sectors(rq, get_capacity(gd));
+		rq->limits.discard_granularity = info->discard_granularity;
+		rq->limits.discard_alignment = info->discard_alignment;
+	}
+
 	/* Hard sector size and max sectors impersonate the equiv. hardware. */
 	blk_queue_logical_block_size(rq, sector_size);
 	blk_queue_max_hw_sectors(rq, 512);
@@ -722,6 +742,17 @@ static irqreturn_t blkif_interrupt(int irq, void *dev_id)
 
 		error = (bret->status == BLKIF_RSP_OKAY) ? 0 : -EIO;
 		switch (bret->operation) {
+		case BLKIF_OP_DISCARD:
+			if (unlikely(bret->status == BLKIF_RSP_EOPNOTSUPP)) {
+				struct request_queue *rq = info->rq;
+				printk(KERN_WARNING "blkfront: %s: discard op failed\n",
+					   info->gd->disk_name);
+				error = -EOPNOTSUPP;
+				info->feature_discard = 0;
+				queue_flag_clear(QUEUE_FLAG_DISCARD, rq);
+			}
+			__blk_end_request_all(req, error);
+			break;
 		case BLKIF_OP_FLUSH_DISKCACHE:
 		case BLKIF_OP_WRITE_BARRIER:
 			if (unlikely(bret->status == BLKIF_RSP_EOPNOTSUPP)) {
@@ -1098,6 +1129,33 @@ blkfront_closing(struct blkfront_info *info)
 	bdput(bdev);
 }
 
+static void blkfront_setup_discard(struct blkfront_info *info)
+{
+	int err;
+	char *type;
+	unsigned int discard_granularity;
+	unsigned int discard_alignment;
+
+	type = xenbus_read(XBT_NIL, info->xbdev->otherend, "type", NULL);
+	if (IS_ERR(type))
+		return;
+
+	if (strncmp(type, "phy", 3) == 0) {
+		err = xenbus_gather(XBT_NIL, info->xbdev->otherend,
+			"discard-granularity", "%u", &discard_granularity,
+			"discard-alignment", "%u", &discard_alignment,
+			NULL);
+		if (!err) {
+			info->feature_discard = 1;
+			info->discard_granularity = discard_granularity;
+			info->discard_alignment = discard_alignment;
+		}
+	} else if (strncmp(type, "file", 4) == 0)
+		info->feature_discard = 1;
+
+	kfree(type);
+}
+
 /*
  * Invoked when the backend is finally 'ready' (and has told produced
  * the details about the physical device - #sectors, size, etc).
@@ -1108,7 +1166,7 @@ static void blkfront_connect(struct blkfront_info *info)
 	unsigned long sector_size;
 	unsigned int binfo;
 	int err;
-	int barrier, flush;
+	int barrier, flush, discard;
 
 	switch (info->connected) {
 	case BLKIF_STATE_CONNECTED:
@@ -1178,7 +1236,14 @@ static void blkfront_connect(struct blkfront_info *info)
 		info->feature_flush = REQ_FLUSH;
 		info->flush_op = BLKIF_OP_FLUSH_DISKCACHE;
 	}
-		
+
+	err = xenbus_gather(XBT_NIL, info->xbdev->otherend,
+			    "feature-discard", "%d", &discard,
+			    NULL);
+
+	if (!err && discard)
+		blkfront_setup_discard(info);
+
 	err = xlvbd_alloc_gendisk(sectors, info, binfo, sector_size);
 	if (err) {
 		xenbus_dev_fatal(info->xbdev, err, "xlvbd_add at %s",
@@ -1385,6 +1450,8 @@ static struct xenbus_driver blkfront = {
 
 static int __init xlblk_init(void)
 {
+	int ret;
+
 	if (!xen_domain())
 		return -ENODEV;
 
@@ -1394,7 +1461,13 @@ static int __init xlblk_init(void)
 		return -ENODEV;
 	}
 
-	return xenbus_register_frontend(&blkfront);
+	ret = xenbus_register_frontend(&blkfront);
+	if (ret) {
+		unregister_blkdev(XENVBD_MAJOR, DEV_NAME);
+		return ret;
+	}
+
+	return 0;
 }
 module_init(xlblk_init);
 

include/xen/interface/io/blkif.h

@@ -57,6 +57,36 @@ typedef uint64_t blkif_sector_t;
  * "feature-flush-cache" node!
  */
 #define BLKIF_OP_FLUSH_DISKCACHE   3
+
+/*
+ * Recognised only if "feature-discard" is present in backend xenbus info.
+ * The "feature-discard" node contains a boolean indicating whether trim
+ * (ATA) or unmap (SCSI) - conviently called discard requests are likely
+ * to succeed or fail. Either way, a discard request
+ * may fail at any time with BLKIF_RSP_EOPNOTSUPP if it is unsupported by
+ * the underlying block-device hardware. The boolean simply indicates whether
+ * or not it is worthwhile for the frontend to attempt discard requests.
+ * If a backend does not recognise BLKIF_OP_DISCARD, it should *not*
+ * create the "feature-discard" node!
+ *
+ * Discard operation is a request for the underlying block device to mark
+ * extents to be erased. However, discard does not guarantee that the blocks
+ * will be erased from the device - it is just a hint to the device
+ * controller that these blocks are no longer in use. What the device
+ * controller does with that information is left to the controller.
+ * Discard operations are passed with sector_number as the
+ * sector index to begin discard operations at and nr_sectors as the number of
+ * sectors to be discarded. The specified sectors should be discarded if the
+ * underlying block device supports trim (ATA) or unmap (SCSI) operations,
+ * or a BLKIF_RSP_EOPNOTSUPP  should be returned.
+ * More information about trim/unmap operations at:
+ * http://t13.org/Documents/UploadedDocuments/docs2008/
+ *     e07154r6-Data_Set_Management_Proposal_for_ATA-ACS2.doc
+ * http://www.seagate.com/staticfiles/support/disc/manuals/
+ *     Interface%20manuals/100293068c.pdf
+ */
+#define BLKIF_OP_DISCARD           5
+
 /*
  * Maximum scatter/gather segments per request.
  * This is carefully chosen so that sizeof(struct blkif_ring) <= PAGE_SIZE.
@@ -74,6 +104,11 @@ struct blkif_request_rw {
 	} seg[BLKIF_MAX_SEGMENTS_PER_REQUEST];
 };
 
+struct blkif_request_discard {
+	blkif_sector_t sector_number;
+	uint64_t nr_sectors;
+};
+
 struct blkif_request {
 	uint8_t        operation;    /* BLKIF_OP_???                         */
 	uint8_t        nr_segments;  /* number of segments                   */
@@ -81,6 +116,7 @@ struct blkif_request {
 	uint64_t       id;           /* private guest value, echoed in resp  */
 	union {
 		struct blkif_request_rw rw;
+		struct blkif_request_discard discard;
 	} u;
 };