xen/block: add multi-page ring support

Extend xen/block to support multi-page ring, so that more requests can be
issued by using more than one pages as the request ring between blkfront
and backend.
As a result, the performance can get improved significantly.

We got some impressive improvements on our highend iscsi storage cluster
backend. If using 64 pages as the ring, the IOPS increased about 15 times
for the throughput testing and above doubled for the latency testing.

The reason was the limit on outstanding requests is 32 if use only one-page
ring, but in our case the iscsi lun was spread across about 100 physical
drives, 32 was really not enough to keep them busy.

Changes in v2:
 - Rebased to 4.0-rc6.
 - Document on how multi-page ring feature working to linux io/blkif.h.

Changes in v3:
 - Remove changes to linux io/blkif.h and follow the protocol defined
   in io/blkif.h of XEN tree.
 - Rebased to 4.1-rc3

Changes in v4:
 - Turn to use 'ring-page-order' and 'max-ring-page-order'.
 - A few comments from Roger.

Changes in v5:
 - Clarify with 4k granularity to comment
 - Address more comments from Roger

Signed-off-by: Bob Liu <bob.liu@oracle.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>

1 files changed, 96 insertions, 39 deletions

diff --git a/drivers/block/xen-blkfront.c b/drivers/block/xen-blkfront.c
index 88e23fd8c7f3..d3c1a9523d1f 100644
--- a/drivers/block/xen-blkfront.c
+++ b/drivers/block/xen-blkfront.c
@@ -98,7 +98,21 @@ static unsigned int xen_blkif_max_segments = 32;
 module_param_named(max, xen_blkif_max_segments, int, S_IRUGO);
 MODULE_PARM_DESC(max, "Maximum amount of segments in indirect requests (default is 32)");
 
-#define BLK_RING_SIZE __CONST_RING_SIZE(blkif, PAGE_SIZE)
+/*
+ * Maximum order of pages to be used for the shared ring between front and
+ * backend, 4KB page granularity is used.
+ */
+static unsigned int xen_blkif_max_ring_order;
+module_param_named(max_ring_page_order, xen_blkif_max_ring_order, int, S_IRUGO);
+MODULE_PARM_DESC(max_ring_page_order, "Maximum order of pages to be used for the shared ring");
+
+#define BLK_RING_SIZE(info) __CONST_RING_SIZE(blkif, PAGE_SIZE * (info)->nr_ring_pages)
+#define BLK_MAX_RING_SIZE __CONST_RING_SIZE(blkif, PAGE_SIZE * XENBUS_MAX_RING_PAGES)
+/*
+ * ring-ref%i i=(-1UL) would take 11 characters + 'ring-ref' is 8, so 19
+ * characters are enough. Define to 20 to keep consist with backend.
+ */
+#define RINGREF_NAME_LEN (20)
 
 /*
  * We have one of these per vbd, whether ide, scsi or 'other'.  They
@@ -114,13 +128,14 @@ struct blkfront_info
 	int vdevice;
 	blkif_vdev_t handle;
 	enum blkif_state connected;
-	int ring_ref;
+	int ring_ref[XENBUS_MAX_RING_PAGES];
+	unsigned int nr_ring_pages;
 	struct blkif_front_ring ring;
 	unsigned int evtchn, irq;
 	struct request_queue *rq;
 	struct work_struct work;
 	struct gnttab_free_callback callback;
-	struct blk_shadow shadow[BLK_RING_SIZE];
+	struct blk_shadow shadow[BLK_MAX_RING_SIZE];
 	struct list_head grants;
 	struct list_head indirect_pages;
 	unsigned int persistent_gnts_c;
@@ -139,8 +154,6 @@ static unsigned int nr_minors;
 static unsigned long *minors;
 static DEFINE_SPINLOCK(minor_lock);
 
-#define MAXIMUM_OUTSTANDING_BLOCK_REQS \
-	(BLKIF_MAX_SEGMENTS_PER_REQUEST * BLK_RING_SIZE)
 #define GRANT_INVALID_REF	0
 
 #define PARTS_PER_DISK		16
@@ -170,7 +183,7 @@ static int blkfront_setup_indirect(struct blkfront_info *info);
 static int get_id_from_freelist(struct blkfront_info *info)
 {
 	unsigned long free = info->shadow_free;
-	BUG_ON(free >= BLK_RING_SIZE);
+	BUG_ON(free >= BLK_RING_SIZE(info));
 	info->shadow_free = info->shadow[free].req.u.rw.id;
 	info->shadow[free].req.u.rw.id = 0x0fffffee; /* debug */
 	return free;
@@ -983,7 +996,7 @@ static void blkif_free(struct blkfront_info *info, int suspend)
 		}
 	}
 
-	for (i = 0; i < BLK_RING_SIZE; i++) {
+	for (i = 0; i < BLK_RING_SIZE(info); i++) {
 		/*
 		 * Clear persistent grants present in requests already
 		 * on the shared ring
@@ -1033,12 +1046,15 @@ free_shadow:
 	flush_work(&info->work);
 
 	/* Free resources associated with old device channel. */
-	if (info->ring_ref != GRANT_INVALID_REF) {
-		gnttab_end_foreign_access(info->ring_ref, 0,
-					  (unsigned long)info->ring.sring);
-		info->ring_ref = GRANT_INVALID_REF;
-		info->ring.sring = NULL;
+	for (i = 0; i < info->nr_ring_pages; i++) {
+		if (info->ring_ref[i] != GRANT_INVALID_REF) {
+			gnttab_end_foreign_access(info->ring_ref[i], 0, 0);
+			info->ring_ref[i] = GRANT_INVALID_REF;
+		}
 	}
+	free_pages((unsigned long)info->ring.sring, get_order(info->nr_ring_pages * PAGE_SIZE));
+	info->ring.sring = NULL;
+
 	if (info->irq)
 		unbind_from_irqhandler(info->irq, info);
 	info->evtchn = info->irq = 0;
@@ -1157,7 +1173,7 @@ static irqreturn_t blkif_interrupt(int irq, void *dev_id)
 		 * never have given to it (we stamp it up to BLK_RING_SIZE -
 		 * look in get_id_from_freelist.
 		 */
-		if (id >= BLK_RING_SIZE) {
+		if (id >= BLK_RING_SIZE(info)) {
 			WARN(1, "%s: response to %s has incorrect id (%ld)\n",
 			     info->gd->disk_name, op_name(bret->operation), id);
 			/* We can't safely get the 'struct request' as
@@ -1245,26 +1261,30 @@ static int setup_blkring(struct xenbus_device *dev,
 			 struct blkfront_info *info)
 {
 	struct blkif_sring *sring;
-	grant_ref_t gref;
-	int err;
+	int err, i;
+	unsigned long ring_size = info->nr_ring_pages * PAGE_SIZE;
+	grant_ref_t gref[XENBUS_MAX_RING_PAGES];
 
-	info->ring_ref = GRANT_INVALID_REF;
+	for (i = 0; i < info->nr_ring_pages; i++)
+		info->ring_ref[i] = GRANT_INVALID_REF;
 
-	sring = (struct blkif_sring *)__get_free_page(GFP_NOIO | __GFP_HIGH);
+	sring = (struct blkif_sring *)__get_free_pages(GFP_NOIO | __GFP_HIGH,
+						       get_order(ring_size));
 	if (!sring) {
 		xenbus_dev_fatal(dev, -ENOMEM, "allocating shared ring");
 		return -ENOMEM;
 	}
 	SHARED_RING_INIT(sring);
-	FRONT_RING_INIT(&info->ring, sring, PAGE_SIZE);
+	FRONT_RING_INIT(&info->ring, sring, ring_size);
 
-	err = xenbus_grant_ring(dev, info->ring.sring, 1, &gref);
+	err = xenbus_grant_ring(dev, info->ring.sring, info->nr_ring_pages, gref);
 	if (err < 0) {
-		free_page((unsigned long)sring);
+		free_pages((unsigned long)sring, get_order(ring_size));
 		info->ring.sring = NULL;
 		goto fail;
 	}
-	info->ring_ref = gref;
+	for (i = 0; i < info->nr_ring_pages; i++)
+		info->ring_ref[i] = gref[i];
 
 	err = xenbus_alloc_evtchn(dev, &info->evtchn);
 	if (err)
@@ -1292,7 +1312,18 @@ static int talk_to_blkback(struct xenbus_device *dev,
 {
 	const char *message = NULL;
 	struct xenbus_transaction xbt;
-	int err;
+	int err, i;
+	unsigned int max_page_order = 0;
+	unsigned int ring_page_order = 0;
+
+	err = xenbus_scanf(XBT_NIL, info->xbdev->otherend,
+			   "max-ring-page-order", "%u", &max_page_order);
+	if (err != 1)
+		info->nr_ring_pages = 1;
+	else {
+		ring_page_order = min(xen_blkif_max_ring_order, max_page_order);
+		info->nr_ring_pages = 1 << ring_page_order;
+	}
 
 	/* Create shared ring, alloc event channel. */
 	err = setup_blkring(dev, info);
@@ -1306,11 +1337,32 @@ again:
 		goto destroy_blkring;
 	}
 
-	err = xenbus_printf(xbt, dev->nodename,
-			    "ring-ref", "%u", info->ring_ref);
-	if (err) {
-		message = "writing ring-ref";
-		goto abort_transaction;
+	if (info->nr_ring_pages == 1) {
+		err = xenbus_printf(xbt, dev->nodename,
+				    "ring-ref", "%u", info->ring_ref[0]);
+		if (err) {
+			message = "writing ring-ref";
+			goto abort_transaction;
+		}
+	} else {
+		err = xenbus_printf(xbt, dev->nodename,
+				    "ring-page-order", "%u", ring_page_order);
+		if (err) {
+			message = "writing ring-page-order";
+			goto abort_transaction;
+		}
+
+		for (i = 0; i < info->nr_ring_pages; i++) {
+			char ring_ref_name[RINGREF_NAME_LEN];
+
+			snprintf(ring_ref_name, RINGREF_NAME_LEN, "ring-ref%u", i);
+			err = xenbus_printf(xbt, dev->nodename, ring_ref_name,
+					    "%u", info->ring_ref[i]);
+			if (err) {
+				message = "writing ring-ref";
+				goto abort_transaction;
+			}
+		}
 	}
 	err = xenbus_printf(xbt, dev->nodename,
 			    "event-channel", "%u", info->evtchn);
@@ -1338,6 +1390,9 @@ again:
 		goto destroy_blkring;
 	}
 
+	for (i = 0; i < BLK_RING_SIZE(info); i++)
+		info->shadow[i].req.u.rw.id = i+1;
+	info->shadow[BLK_RING_SIZE(info)-1].req.u.rw.id = 0x0fffffff;
 	xenbus_switch_state(dev, XenbusStateInitialised);
 
 	return 0;
@@ -1361,7 +1416,7 @@ again:
 static int blkfront_probe(struct xenbus_device *dev,
 			  const struct xenbus_device_id *id)
 {
-	int err, vdevice, i;
+	int err, vdevice;
 	struct blkfront_info *info;
 
 	/* FIXME: Use dynamic device id if this is not set. */
@@ -1422,10 +1477,6 @@ static int blkfront_probe(struct xenbus_device *dev,
 	info->connected = BLKIF_STATE_DISCONNECTED;
 	INIT_WORK(&info->work, blkif_restart_queue);
 
-	for (i = 0; i < BLK_RING_SIZE; i++)
-		info->shadow[i].req.u.rw.id = i+1;
-	info->shadow[BLK_RING_SIZE-1].req.u.rw.id = 0x0fffffff;
-
 	/* Front end dir is a number, which is used as the id. */
 	info->handle = simple_strtoul(strrchr(dev->nodename, '/')+1, NULL, 0);
 	dev_set_drvdata(&dev->dev, info);
@@ -1469,10 +1520,10 @@ static int blkif_recover(struct blkfront_info *info)
 
 	/* Stage 2: Set up free list. */
 	memset(&info->shadow, 0, sizeof(info->shadow));
-	for (i = 0; i < BLK_RING_SIZE; i++)
+	for (i = 0; i < BLK_RING_SIZE(info); i++)
 		info->shadow[i].req.u.rw.id = i+1;
 	info->shadow_free = info->ring.req_prod_pvt;
-	info->shadow[BLK_RING_SIZE-1].req.u.rw.id = 0x0fffffff;
+	info->shadow[BLK_RING_SIZE(info)-1].req.u.rw.id = 0x0fffffff;
 
 	rc = blkfront_setup_indirect(info);
 	if (rc) {
@@ -1484,7 +1535,7 @@ static int blkif_recover(struct blkfront_info *info)
 	blk_queue_max_segments(info->rq, segs);
 	bio_list_init(&bio_list);
 	INIT_LIST_HEAD(&requests);
-	for (i = 0; i < BLK_RING_SIZE; i++) {
+	for (i = 0; i < BLK_RING_SIZE(info); i++) {
 		/* Not in use? */
 		if (!copy[i].request)
 			continue;
@@ -1690,7 +1741,7 @@ static int blkfront_setup_indirect(struct blkfront_info *info)
 		segs = info->max_indirect_segments;
 	}
 
-	err = fill_grant_buffer(info, (segs + INDIRECT_GREFS(segs)) * BLK_RING_SIZE);
+	err = fill_grant_buffer(info, (segs + INDIRECT_GREFS(segs)) * BLK_RING_SIZE(info));
 	if (err)
 		goto out_of_memory;
 
@@ -1700,7 +1751,7 @@ static int blkfront_setup_indirect(struct blkfront_info *info)
 		 * grants, we need to allocate a set of pages that can be
 		 * used for mapping indirect grefs
 		 */
-		int num = INDIRECT_GREFS(segs) * BLK_RING_SIZE;
+		int num = INDIRECT_GREFS(segs) * BLK_RING_SIZE(info);
 
 		BUG_ON(!list_empty(&info->indirect_pages));
 		for (i = 0; i < num; i++) {
@@ -1711,7 +1762,7 @@ static int blkfront_setup_indirect(struct blkfront_info *info)
 		}
 	}
 
-	for (i = 0; i < BLK_RING_SIZE; i++) {
+	for (i = 0; i < BLK_RING_SIZE(info); i++) {
 		info->shadow[i].grants_used = kzalloc(
 			sizeof(info->shadow[i].grants_used[0]) * segs,
 			GFP_NOIO);
@@ -1733,7 +1784,7 @@ static int blkfront_setup_indirect(struct blkfront_info *info)
 	return 0;
 
 out_of_memory:
-	for (i = 0; i < BLK_RING_SIZE; i++) {
+	for (i = 0; i < BLK_RING_SIZE(info); i++) {
 		kfree(info->shadow[i].grants_used);
 		info->shadow[i].grants_used = NULL;
 		kfree(info->shadow[i].sg);
@@ -2089,6 +2140,12 @@ static int __init xlblk_init(void)
 	if (!xen_domain())
 		return -ENODEV;
 
+	if (xen_blkif_max_ring_order > XENBUS_MAX_RING_PAGE_ORDER) {
+		pr_info("Invalid max_ring_order (%d), will use default max: %d.\n",
+			xen_blkif_max_ring_order, XENBUS_MAX_RING_PAGE_ORDER);
+		xen_blkif_max_ring_order = 0;
+	}
+
 	if (!xen_has_pv_disk_devices())
 		return -ENODEV;