Skip to contentSkip to conversation

virtio-blk: support polling I/O and mq_ops->queue_rqs()

PATCH Superseded Suwan Kim 3 participants 11 messages Apr 5, 2022

Subsystems (3)

v4 of 6 Superseded Latest is v6 →

lore
Share & tools
lore

Suwan Kim

author
#0 lore raw

v4 cover virtio-blk: support polling I/O and mq_ops->queue_rqs()

Sorry. I forgot to fix module parameter name in patch1. So I resend it.

This patch serise adds support for polling I/O and mq_ops->queue_rqs() to virtio-blk driver.

Changes

v3 -> v4
- patch1 : virtblk_poll
- Add print the number of default/read/poll queues in init_vq()
- Add blk_mq_start_stopped_hw_queues() to virtblk_poll() virtblk_poll()
...
if (req_done)
blk_mq_start_stopped_hw_queues(vblk->disk->queue, true);
...

- patch2 : virtio_queue_rqs
- Modify virtio_queue_rqs() to hold lock only once when it adds requests to virtqueue just before virtqueue notify. It will guarantee that virtio_queue_rqs() will not use previous req again.

v2 -> v3
- Fix warning by kernel test robot

static int virtblk_poll()
...
if (!blk_mq_add_to_batch(req, iob, virtblk_result(vbr),
-> vbr->status,

v1 -> v2
- To receive the number of poll queues from user, use module parameter instead of QEMU uapi change.

- Add the comment about virtblk_map_queues().

- Add support for mq_ops->queue_rqs() to implement submit side batch.

Suwan Kim (2):
virtio-blk: support polling I/O
virtio-blk: support mq_ops->queue_rqs()

 drivers/block/virtio_blk.c | 222 ++++++++++++++++++++++++++++++++++---
 1 file changed, 207 insertions(+), 15 deletions(-)
Signature · 1 line
2.26.3
author
#1 lore raw

PATCH v4 1/2 virtio-blk: support polling I/O

This patch supports polling I/O via virtio-blk driver. Polling feature is enabled by module parameter "num_poll_queues" and it sets dedicated polling queues for virtio-blk. This patch improves the polling I/O throughput and latency.

The virtio-blk driver doesn't not have a poll function and a poll queue and it has been operating in interrupt driven method even if the polling function is called in the upper layer.

virtio-blk polling is implemented upon 'batched completion' of block layer. virtblk_poll() queues completed request to io_comp_batch->req_list and later, virtblk_complete_batch() calls unmap function and ends the requests in batch.

virtio-blk reads the number of poll queues from module parameter "num_poll_queues". If VM sets queue parameter as below, ("num-queues=N" [QEMU property], "num_poll_queues=M" [module parameter]) It allocates N virtqueues to virtio_blk->vqs[N] and it uses [0..(N-M-1)] as default queues and [(N-M)..(N-1)] as poll queues. Unlike the default queues, the poll queues have no callback function.

Regarding HW-SW queue mapping, the default queue mapping uses the existing method that condsiders MSI irq vector. But the poll queue doesn't have an irq, so it uses the regular blk-mq cpu mapping.

For verifying the improvement, I did Fio polling I/O performance test with io_uring engine with the options below.
(io_uring, hipri, randread, direct=1, bs=512, iodepth=64 numjobs=N) I set 4 vcpu and 4 virtio-blk queues - 2 default queues and 2 poll queues for VM.

As a result, IOPS and average latency improved about 10%.

Test result:

- Fio io_uring poll without virtio-blk poll support

	-- numjobs=1 : IOPS = 339K, avg latency = 188.33us
	-- numjobs=2 : IOPS = 367K, avg latency = 347.33us
	-- numjobs=4 : IOPS = 383K, avg latency = 682.06us

- Fio io_uring poll with virtio-blk poll support

	-- numjobs=1 : IOPS = 385K, avg latency = 165.94us
	-- numjobs=2 : IOPS = 408K, avg latency = 313.28us
	-- numjobs=4 : IOPS = 424K, avg latency = 613.05us
---
 drivers/block/virtio_blk.c | 112 +++++++++++++++++++++++++++++++++++--
 1 file changed, 108 insertions(+), 4 deletions(-)
drivers/block/virtio_blk.c +108 -4
index 8c415be86732..712579dcd3cc 100644
@@ -37,6 +37,10 @@ MODULE_PARM_DESC(num_request_queues,
 		 "0 for no limit. "
 		 "Values > nr_cpu_ids truncated to nr_cpu_ids.");
 
+static unsigned int poll_queues;
+module_param(poll_queues, uint, 0644);
+MODULE_PARM_DESC(poll_queues, "The number of dedicated virtqueues for polling I/O");
+
 static int major;
 static DEFINE_IDA(vd_index_ida);
 
@@ -81,6 +85,7 @@ struct virtio_blk {
 
 	/* num of vqs */
 	int num_vqs;
+	int io_queues[HCTX_MAX_TYPES];
 	struct virtio_blk_vq *vqs;
 };
 
@@ -548,6 +553,7 @@ static int init_vq(struct virtio_blk *vblk)
 	const char **names;
 	struct virtqueue **vqs;
 	unsigned short num_vqs;
+	unsigned int num_poll_vqs;
 	struct virtio_device *vdev = vblk->vdev;
 	struct irq_affinity desc = { 0, };
 
@@ -556,6 +562,7 @@ static int init_vq(struct virtio_blk *vblk)
 				   &num_vqs);
 	if (err)
 		num_vqs = 1;
+
 	if (!err && !num_vqs) {
 		dev_err(&vdev->dev, "MQ advertised but zero queues reported\n");
 		return -EINVAL;
@@ -565,6 +572,18 @@ static int init_vq(struct virtio_blk *vblk)
 			min_not_zero(num_request_queues, nr_cpu_ids),
 			num_vqs);
 
+	num_poll_vqs = min_t(unsigned int, poll_queues, num_vqs - 1);
+
+	memset(vblk->io_queues, 0, sizeof(int) * HCTX_MAX_TYPES);
+	vblk->io_queues[HCTX_TYPE_DEFAULT] = num_vqs - num_poll_vqs;
+	vblk->io_queues[HCTX_TYPE_READ] = 0;
+	vblk->io_queues[HCTX_TYPE_POLL] = num_poll_vqs;
+
+	dev_info(&vdev->dev, "%d/%d/%d default/read/poll queues\n",
+				vblk->io_queues[HCTX_TYPE_DEFAULT],
+				vblk->io_queues[HCTX_TYPE_READ],
+				vblk->io_queues[HCTX_TYPE_POLL]);
+
 	vblk->vqs = kmalloc_array(num_vqs, sizeof(*vblk->vqs), GFP_KERNEL);
 	if (!vblk->vqs)
 		return -ENOMEM;
@@ -578,8 +597,13 @@ static int init_vq(struct virtio_blk *vblk)
 	}
 
 	for (i = 0; i < num_vqs; i++) {
-		callbacks[i] = virtblk_done;
-		snprintf(vblk->vqs[i].name, VQ_NAME_LEN, "req.%d", i);
+		if (i < num_vqs - num_poll_vqs) {
+			callbacks[i] = virtblk_done;
+			snprintf(vblk->vqs[i].name, VQ_NAME_LEN, "req.%d", i);
+		} else {
+			callbacks[i] = NULL;
+			snprintf(vblk->vqs[i].name, VQ_NAME_LEN, "req_poll.%d", i);
+		}
 		names[i] = vblk->vqs[i].name;
 	}
 
@@ -728,16 +752,93 @@ static const struct attribute_group *virtblk_attr_groups[] = {
 static int virtblk_map_queues(struct blk_mq_tag_set *set)
 {
 	struct virtio_blk *vblk = set->driver_data;
+	int i, qoff;
+
+	for (i = 0, qoff = 0; i < set->nr_maps; i++) {
+		struct blk_mq_queue_map *map = &set->map[i];
+
+		map->nr_queues = vblk->io_queues[i];
+		map->queue_offset = qoff;
+		qoff += map->nr_queues;
+
+		if (map->nr_queues == 0)
+			continue;
+
+		/*
+		 * Regular queues have interrupts and hence CPU affinity is
+		 * defined by the core virtio code, but polling queues have
+		 * no interrupts so we let the block layer assign CPU affinity.
+		 */
+		if (i == HCTX_TYPE_DEFAULT)
+			blk_mq_virtio_map_queues(&set->map[i], vblk->vdev, 0);
+		else
+			blk_mq_map_queues(&set->map[i]);
+	}
+
+	return 0;
+}
+
+static void virtblk_complete_batch(struct io_comp_batch *iob)
+{
+	struct request *req;
+	struct virtblk_req *vbr;
+
+	rq_list_for_each(&iob->req_list, req) {
+		vbr = blk_mq_rq_to_pdu(req);
+		virtblk_unmap_data(req, vbr);
+		virtblk_cleanup_cmd(req);
+	}
+	blk_mq_end_request_batch(iob);
+}
+
+static int virtblk_poll(struct blk_mq_hw_ctx *hctx, struct io_comp_batch *iob)
+{
+	struct virtio_blk *vblk = hctx->queue->queuedata;
+	struct virtio_blk_vq *vq = hctx->driver_data;
+	struct virtblk_req *vbr;
+	bool req_done = false;
+	unsigned long flags;
+	unsigned int len;
+	int found = 0;
+
+	spin_lock_irqsave(&vq->lock, flags);
+
+	while ((vbr = virtqueue_get_buf(vq->vq, &len)) != NULL) {
+		struct request *req = blk_mq_rq_from_pdu(vbr);
 
-	return blk_mq_virtio_map_queues(&set->map[HCTX_TYPE_DEFAULT],
-					vblk->vdev, 0);
+		found++;
+		if (!blk_mq_add_to_batch(req, iob, vbr->status,
+						virtblk_complete_batch))
+			blk_mq_complete_request(req);
+		req_done = true;
+	}
+
+	if (req_done)
+		blk_mq_start_stopped_hw_queues(vblk->disk->queue, true);
+
+	spin_unlock_irqrestore(&vq->lock, flags);
+
+	return found;
+}
+
+static int virtblk_init_hctx(struct blk_mq_hw_ctx *hctx, void *data,
+			  unsigned int hctx_idx)
+{
+	struct virtio_blk *vblk = data;
+	struct virtio_blk_vq *vq = &vblk->vqs[hctx_idx];
+
+	WARN_ON(vblk->tag_set.tags[hctx_idx] != hctx->tags);
+	hctx->driver_data = vq;
+	return 0;
 }
 
 static const struct blk_mq_ops virtio_mq_ops = {
 	.queue_rq	= virtio_queue_rq,
 	.commit_rqs	= virtio_commit_rqs,
+	.init_hctx	= virtblk_init_hctx,
 	.complete	= virtblk_request_done,
 	.map_queues	= virtblk_map_queues,
+	.poll		= virtblk_poll,
 };
 
 static unsigned int virtblk_queue_depth;
@@ -816,6 +917,9 @@ static int virtblk_probe(struct virtio_device *vdev)
 		sizeof(struct scatterlist) * VIRTIO_BLK_INLINE_SG_CNT;
 	vblk->tag_set.driver_data = vblk;
 	vblk->tag_set.nr_hw_queues = vblk->num_vqs;
+	vblk->tag_set.nr_maps = 1;
+	if (vblk->io_queues[HCTX_TYPE_POLL])
+		vblk->tag_set.nr_maps = 3;
 
 	err = blk_mq_alloc_tag_set(&vblk->tag_set);
 	if (err)
Signature · 1 line
2.26.3
#2 lore raw

Re: [PATCH v4 1/2] virtio-blk: support polling I/O

Quoted 15 lines · Suwan Kim
drivers/block/virtio_blk.c +14 -0
+static int virtblk_poll(struct blk_mq_hw_ctx *hctx, struct io_comp_batch *iob)
+{
+	struct virtio_blk *vblk = hctx->queue->queuedata;
+	struct virtio_blk_vq *vq = hctx->driver_data;
+	struct virtblk_req *vbr;
+	bool req_done = false;
+	unsigned long flags;
+	unsigned int len;
+	int found = 0;
+
+	spin_lock_irqsave(&vq->lock, flags);
+
+	while ((vbr = virtqueue_get_buf(vq->vq, &len)) != NULL) {
+		struct request *req = blk_mq_rq_from_pdu(vbr);
 

drivers/block/virtio_blk.c static const struct attribute_group *virtblk_attr_groups[] = { · line 816

Quoted 10 lines
drivers/block/virtio_blk.c +8 -2
-	return blk_mq_virtio_map_queues(&set->map[HCTX_TYPE_DEFAULT],
-					vblk->vdev, 0);
+		found++;
+		if (!blk_mq_add_to_batch(req, iob, vbr->status,
+						virtblk_complete_batch))
+			blk_mq_complete_request(req);
+		req_done = true;
+	}
+
+	if (req_done)

Minor nit: req_done can be replaced with found > 0.

drivers/block/virtio_blk.c static const struct attribute_group *virtblk_attr_groups[] = { · line 822

Quoted 6 lines
drivers/block/virtio_blk.c +6 -0
+		blk_mq_start_stopped_hw_queues(vblk->disk->queue, true);
+
+	spin_unlock_irqrestore(&vq->lock, flags);
+
+	return found;
+}

_______________________________________________
Virtualization mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

author
#3 lore raw

Re: [PATCH v4 1/2] virtio-blk: support polling I/O

Quoted 16 lines · Stefan Hajnoczi
Quoted 15 lines · Suwan Kim
drivers/block/virtio_blk.c +14 -0
+static int virtblk_poll(struct blk_mq_hw_ctx *hctx, struct io_comp_batch *iob)
+{
+	struct virtio_blk *vblk = hctx->queue->queuedata;
+	struct virtio_blk_vq *vq = hctx->driver_data;
+	struct virtblk_req *vbr;
+	bool req_done = false;
+	unsigned long flags;
+	unsigned int len;
+	int found = 0;
+
+	spin_lock_irqsave(&vq->lock, flags);
+
+	while ((vbr = virtqueue_get_buf(vq->vq, &len)) != NULL) {
+		struct request *req = blk_mq_rq_from_pdu(vbr);
Quoted 12 lines

drivers/block/virtio_blk.c static const struct attribute_group *virtblk_attr_groups[] = { · line 816

Quoted 10 lines
drivers/block/virtio_blk.c +8 -2
-	return blk_mq_virtio_map_queues(&set->map[HCTX_TYPE_DEFAULT],
-					vblk->vdev, 0);
+		found++;
+		if (!blk_mq_add_to_batch(req, iob, vbr->status,
+						virtblk_complete_batch))
+			blk_mq_complete_request(req);
+		req_done = true;
+	}
+
+	if (req_done)

Minor nit: req_done can be replaced with found > 0.

It looks better. I will fix it.
Thanks for the feedback!

Regards,
Suwan Kim

#4 lore raw

Re: [PATCH v4 1/2] virtio-blk: support polling I/O

In reply to Suwan Kim
Quoted 110 lines · Suwan Kim

This patch supports polling I/O via virtio-blk driver. Polling feature is enabled by module parameter "num_poll_queues" and it sets dedicated polling queues for virtio-blk. This patch improves the polling I/O throughput and latency.

The virtio-blk driver doesn't not have a poll function and a poll queue and it has been operating in interrupt driven method even if the polling function is called in the upper layer.

virtio-blk polling is implemented upon 'batched completion' of block layer. virtblk_poll() queues completed request to io_comp_batch->req_list and later, virtblk_complete_batch() calls unmap function and ends the requests in batch.

virtio-blk reads the number of poll queues from module parameter "num_poll_queues". If VM sets queue parameter as below, ("num-queues=N" [QEMU property], "num_poll_queues=M" [module parameter]) It allocates N virtqueues to virtio_blk->vqs[N] and it uses [0..(N-M-1)] as default queues and [(N-M)..(N-1)] as poll queues. Unlike the default queues, the poll queues have no callback function.

Regarding HW-SW queue mapping, the default queue mapping uses the existing method that condsiders MSI irq vector. But the poll queue doesn't have an irq, so it uses the regular blk-mq cpu mapping.

For verifying the improvement, I did Fio polling I/O performance test with io_uring engine with the options below.
(io_uring, hipri, randread, direct=1, bs=512, iodepth=64 numjobs=N) I set 4 vcpu and 4 virtio-blk queues - 2 default queues and 2 poll queues for VM.

As a result, IOPS and average latency improved about 10%.

Test result:

- Fio io_uring poll without virtio-blk poll support

	-- numjobs=1 : IOPS = 339K, avg latency = 188.33us
	-- numjobs=2 : IOPS = 367K, avg latency = 347.33us
	-- numjobs=4 : IOPS = 383K, avg latency = 682.06us

- Fio io_uring poll with virtio-blk poll support

	-- numjobs=1 : IOPS = 385K, avg latency = 165.94us
	-- numjobs=2 : IOPS = 408K, avg latency = 313.28us
	-- numjobs=4 : IOPS = 424K, avg latency = 613.05us
---
 drivers/block/virtio_blk.c | 112 +++++++++++++++++++++++++++++++++++--
 1 file changed, 108 insertions(+), 4 deletions(-)
drivers/block/virtio_blk.c +19 -0
index 8c415be86732..712579dcd3cc 100644
@@ -37,6 +37,10 @@ MODULE_PARM_DESC(num_request_queues,
 		 "0 for no limit. "
 		 "Values > nr_cpu_ids truncated to nr_cpu_ids.");
 
+static unsigned int poll_queues;
+module_param(poll_queues, uint, 0644);
+MODULE_PARM_DESC(poll_queues, "The number of dedicated virtqueues for polling I/O");
+
 static int major;
 static DEFINE_IDA(vd_index_ida);
 
@@ -81,6 +85,7 @@ struct virtio_blk {
 
 	/* num of vqs */
 	int num_vqs;
+	int io_queues[HCTX_MAX_TYPES];
 	struct virtio_blk_vq *vqs;
 };
 
@@ -548,6 +553,7 @@ static int init_vq(struct virtio_blk *vblk)
 	const char **names;
 	struct virtqueue **vqs;
 	unsigned short num_vqs;
+	unsigned int num_poll_vqs;
 	struct virtio_device *vdev = vblk->vdev;
 	struct irq_affinity desc = { 0, };
 
@@ -556,6 +562,7 @@ static int init_vq(struct virtio_blk *vblk)
 				   &num_vqs);
 	if (err)
 		num_vqs = 1;
+
 	if (!err && !num_vqs) {
 		dev_err(&vdev->dev, "MQ advertised but zero queues reported\n");
 		return -EINVAL;
@@ -565,6 +572,18 @@ static int init_vq(struct virtio_blk *vblk)
 			min_not_zero(num_request_queues, nr_cpu_ids),
 			num_vqs);
 
+	num_poll_vqs = min_t(unsigned int, poll_queues, num_vqs - 1);
+
+	memset(vblk->io_queues, 0, sizeof(int) * HCTX_MAX_TYPES);
+	vblk->io_queues[HCTX_TYPE_DEFAULT] = num_vqs - num_poll_vqs;
+	vblk->io_queues[HCTX_TYPE_READ] = 0;
+	vblk->io_queues[HCTX_TYPE_POLL] = num_poll_vqs;
+
+	dev_info(&vdev->dev, "%d/%d/%d default/read/poll queues\n",
+				vblk->io_queues[HCTX_TYPE_DEFAULT],
+				vblk->io_queues[HCTX_TYPE_READ],
+				vblk->io_queues[HCTX_TYPE_POLL]);
+
 	vblk->vqs = kmalloc_array(num_vqs, sizeof(*vblk->vqs), GFP_KERNEL);
 	if (!vblk->vqs)
 		return -ENOMEM;
@@ -578,8 +597,13 @@ static int init_vq(struct virtio_blk *vblk)
 	}

drivers/block/virtio_blk.c static int init_vq(struct virtio_blk *vblk) · line 607

Quoted 10 lines
drivers/block/virtio_blk.c +7 -0
 
 	for (i = 0; i < num_vqs; i++) {
+		if (i < num_vqs - num_poll_vqs) {
+			callbacks[i] = virtblk_done;
+			snprintf(vblk->vqs[i].name, VQ_NAME_LEN, "req.%d", i);
+		} else {
+			callbacks[i] = NULL;
+			snprintf(vblk->vqs[i].name, VQ_NAME_LEN, "req_poll.%d", i);
+		}
 		names[i] = vblk->vqs[i].name;

This would look a little cleaner with two loops:

 	for (i = 0; i < num_vqs - num_poll_vqs; i++) {
		callbacks[i] = virtblk_done;
		snprintf(vblk->vqs[i].name, VQ_NAME_LEN, "req.%d", i);
 		names[i] = vblk->vqs[i].name;
	}
 	for (; i < num_vqs; i++) {
		callbacks[i] = NULL;
		snprintf(vblk->vqs[i].name, VQ_NAME_LEN, "req_poll.%d", i);
		names[i] = vblk->vqs[i].name;
	}

drivers/block/virtio_blk.c static const struct attribute_group *virtblk_attr_groups[] = { · line 772

Quoted 10 lines
drivers/block/virtio_blk.c +10 -0
+
+		if (map->nr_queues == 0)
+			continue;
+
+		/*
+		 * Regular queues have interrupts and hence CPU affinity is
+		 * defined by the core virtio code, but polling queues have
+		 * no interrupts so we let the block layer assign CPU affinity.
+		 */
+		if (i == HCTX_TYPE_DEFAULT)

I'd check for

		i != HCTX_TYPE_POLL

here instead to make the check a little more explicit and future proof for the potential addition of read queues (which would be a Linux only change without hypervisor or spec changes). In fact you might as well add that support now as doing it is completely trivial once a driver supports multiple map types.

drivers/block/virtio_blk.c static const struct attribute_group *virtblk_attr_groups[] = { · line 789

Quoted 9 lines
drivers/block/virtio_blk.c +9 -0
+static void virtblk_complete_batch(struct io_comp_batch *iob)
+{
+	struct request *req;
+	struct virtblk_req *vbr;
+
+	rq_list_for_each(&iob->req_list, req) {
+		vbr = blk_mq_rq_to_pdu(req);
+		virtblk_unmap_data(req, vbr);
+		virtblk_cleanup_cmd(req);

vbr is only used ones, so why not just:

		virtblk_unmap_data(req, blk_mq_rq_to_pdu);

?

Or even better add a cleanup patch to just remove the vbr argument to virtblk_unmap_data as it is not needed at all.
_______________________________________________
Virtualization mailing list [email protected]
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

author
#5 lore raw

Re: [PATCH v4 1/2] virtio-blk: support polling I/O

Quoted 124 lines · Christoph Hellwig
Quoted 110 lines · Suwan Kim

This patch supports polling I/O via virtio-blk driver. Polling feature is enabled by module parameter "num_poll_queues" and it sets dedicated polling queues for virtio-blk. This patch improves the polling I/O throughput and latency.

The virtio-blk driver doesn't not have a poll function and a poll queue and it has been operating in interrupt driven method even if the polling function is called in the upper layer.

virtio-blk polling is implemented upon 'batched completion' of block layer. virtblk_poll() queues completed request to io_comp_batch->req_list and later, virtblk_complete_batch() calls unmap function and ends the requests in batch.

virtio-blk reads the number of poll queues from module parameter "num_poll_queues". If VM sets queue parameter as below, ("num-queues=N" [QEMU property], "num_poll_queues=M" [module parameter]) It allocates N virtqueues to virtio_blk->vqs[N] and it uses [0..(N-M-1)] as default queues and [(N-M)..(N-1)] as poll queues. Unlike the default queues, the poll queues have no callback function.

Regarding HW-SW queue mapping, the default queue mapping uses the existing method that condsiders MSI irq vector. But the poll queue doesn't have an irq, so it uses the regular blk-mq cpu mapping.

For verifying the improvement, I did Fio polling I/O performance test with io_uring engine with the options below.
(io_uring, hipri, randread, direct=1, bs=512, iodepth=64 numjobs=N) I set 4 vcpu and 4 virtio-blk queues - 2 default queues and 2 poll queues for VM.

As a result, IOPS and average latency improved about 10%.

Test result:

- Fio io_uring poll without virtio-blk poll support

	-- numjobs=1 : IOPS = 339K, avg latency = 188.33us
	-- numjobs=2 : IOPS = 367K, avg latency = 347.33us
	-- numjobs=4 : IOPS = 383K, avg latency = 682.06us

- Fio io_uring poll with virtio-blk poll support

	-- numjobs=1 : IOPS = 385K, avg latency = 165.94us
	-- numjobs=2 : IOPS = 408K, avg latency = 313.28us
	-- numjobs=4 : IOPS = 424K, avg latency = 613.05us
---
 drivers/block/virtio_blk.c | 112 +++++++++++++++++++++++++++++++++++--
 1 file changed, 108 insertions(+), 4 deletions(-)
drivers/block/virtio_blk.c +19 -0
index 8c415be86732..712579dcd3cc 100644
@@ -37,6 +37,10 @@ MODULE_PARM_DESC(num_request_queues,
 		 "0 for no limit. "
 		 "Values > nr_cpu_ids truncated to nr_cpu_ids.");
 
+static unsigned int poll_queues;
+module_param(poll_queues, uint, 0644);
+MODULE_PARM_DESC(poll_queues, "The number of dedicated virtqueues for polling I/O");
+
 static int major;
 static DEFINE_IDA(vd_index_ida);
 
@@ -81,6 +85,7 @@ struct virtio_blk {
 
 	/* num of vqs */
 	int num_vqs;
+	int io_queues[HCTX_MAX_TYPES];
 	struct virtio_blk_vq *vqs;
 };
 
@@ -548,6 +553,7 @@ static int init_vq(struct virtio_blk *vblk)
 	const char **names;
 	struct virtqueue **vqs;
 	unsigned short num_vqs;
+	unsigned int num_poll_vqs;
 	struct virtio_device *vdev = vblk->vdev;
 	struct irq_affinity desc = { 0, };
 
@@ -556,6 +562,7 @@ static int init_vq(struct virtio_blk *vblk)
 				   &num_vqs);
 	if (err)
 		num_vqs = 1;
+
 	if (!err && !num_vqs) {
 		dev_err(&vdev->dev, "MQ advertised but zero queues reported\n");
 		return -EINVAL;
@@ -565,6 +572,18 @@ static int init_vq(struct virtio_blk *vblk)
 			min_not_zero(num_request_queues, nr_cpu_ids),
 			num_vqs);
 
+	num_poll_vqs = min_t(unsigned int, poll_queues, num_vqs - 1);
+
+	memset(vblk->io_queues, 0, sizeof(int) * HCTX_MAX_TYPES);
+	vblk->io_queues[HCTX_TYPE_DEFAULT] = num_vqs - num_poll_vqs;
+	vblk->io_queues[HCTX_TYPE_READ] = 0;
+	vblk->io_queues[HCTX_TYPE_POLL] = num_poll_vqs;
+
+	dev_info(&vdev->dev, "%d/%d/%d default/read/poll queues\n",
+				vblk->io_queues[HCTX_TYPE_DEFAULT],
+				vblk->io_queues[HCTX_TYPE_READ],
+				vblk->io_queues[HCTX_TYPE_POLL]);
+
 	vblk->vqs = kmalloc_array(num_vqs, sizeof(*vblk->vqs), GFP_KERNEL);
 	if (!vblk->vqs)
 		return -ENOMEM;
@@ -578,8 +597,13 @@ static int init_vq(struct virtio_blk *vblk)
 	}

drivers/block/virtio_blk.c static int init_vq(struct virtio_blk *vblk) · line 607

Quoted 10 lines
drivers/block/virtio_blk.c +7 -0
 
 	for (i = 0; i < num_vqs; i++) {
+		if (i < num_vqs - num_poll_vqs) {
+			callbacks[i] = virtblk_done;
+			snprintf(vblk->vqs[i].name, VQ_NAME_LEN, "req.%d", i);
+		} else {
+			callbacks[i] = NULL;
+			snprintf(vblk->vqs[i].name, VQ_NAME_LEN, "req_poll.%d", i);
+		}
 		names[i] = vblk->vqs[i].name;

This would look a little cleaner with two loops:

Quoted 11 lines
 	for (i = 0; i < num_vqs - num_poll_vqs; i++) {
		callbacks[i] = virtblk_done;
		snprintf(vblk->vqs[i].name, VQ_NAME_LEN, "req.%d", i);
 		names[i] = vblk->vqs[i].name;
	}
 	for (; i < num_vqs; i++) {
		callbacks[i] = NULL;
		snprintf(vblk->vqs[i].name, VQ_NAME_LEN, "req_poll.%d", i);
		names[i] = vblk->vqs[i].name;
	}

It looks better. I will fix it.

Quoted 12 lines

drivers/block/virtio_blk.c static const struct attribute_group *virtblk_attr_groups[] = { · line 772

Quoted 10 lines
drivers/block/virtio_blk.c +10 -0
+
+		if (map->nr_queues == 0)
+			continue;
+
+		/*
+		 * Regular queues have interrupts and hence CPU affinity is
+		 * defined by the core virtio code, but polling queues have
+		 * no interrupts so we let the block layer assign CPU affinity.
+		 */
+		if (i == HCTX_TYPE_DEFAULT)

I'd check for

Quoted 7 lines
		i != HCTX_TYPE_POLL

here instead to make the check a little more explicit and future proof for the potential addition of read queues (which would be a Linux only change without hypervisor or spec changes). In fact you might as well add that support now as doing it is completely trivial once a driver supports multiple map types.

Your if condition is better choice when we consider future use of read queue. I will fix it too.

Quoted 12 lines

drivers/block/virtio_blk.c static const struct attribute_group *virtblk_attr_groups[] = { · line 789

Quoted 9 lines
drivers/block/virtio_blk.c +9 -0
+static void virtblk_complete_batch(struct io_comp_batch *iob)
+{
+	struct request *req;
+	struct virtblk_req *vbr;
+
+	rq_list_for_each(&iob->req_list, req) {
+		vbr = blk_mq_rq_to_pdu(req);
+		virtblk_unmap_data(req, vbr);
+		virtblk_cleanup_cmd(req);

vbr is only used ones, so why not just:

Quoted 4 lines
		virtblk_unmap_data(req, blk_mq_rq_to_pdu);

?
Or even better add a cleanup patch to just remove the vbr argument to virtblk_unmap_data as it is not needed at all.

Thanks for the feedback. I will add cleanup patch in this series.

Ragards,
Suwan Kim

author
#6 lore raw

Re: [PATCH v4 1/2] virtio-blk: support polling I/O

In reply to Christoph Hellwig
Quoted 167 lines · Christoph Hellwig
Quoted 110 lines · Suwan Kim

This patch supports polling I/O via virtio-blk driver. Polling feature is enabled by module parameter "num_poll_queues" and it sets dedicated polling queues for virtio-blk. This patch improves the polling I/O throughput and latency.

The virtio-blk driver doesn't not have a poll function and a poll queue and it has been operating in interrupt driven method even if the polling function is called in the upper layer.

virtio-blk polling is implemented upon 'batched completion' of block layer. virtblk_poll() queues completed request to io_comp_batch->req_list and later, virtblk_complete_batch() calls unmap function and ends the requests in batch.

virtio-blk reads the number of poll queues from module parameter "num_poll_queues". If VM sets queue parameter as below, ("num-queues=N" [QEMU property], "num_poll_queues=M" [module parameter]) It allocates N virtqueues to virtio_blk->vqs[N] and it uses [0..(N-M-1)] as default queues and [(N-M)..(N-1)] as poll queues. Unlike the default queues, the poll queues have no callback function.

Regarding HW-SW queue mapping, the default queue mapping uses the existing method that condsiders MSI irq vector. But the poll queue doesn't have an irq, so it uses the regular blk-mq cpu mapping.

For verifying the improvement, I did Fio polling I/O performance test with io_uring engine with the options below.
(io_uring, hipri, randread, direct=1, bs=512, iodepth=64 numjobs=N) I set 4 vcpu and 4 virtio-blk queues - 2 default queues and 2 poll queues for VM.

As a result, IOPS and average latency improved about 10%.

Test result:

- Fio io_uring poll without virtio-blk poll support

	-- numjobs=1 : IOPS = 339K, avg latency = 188.33us
	-- numjobs=2 : IOPS = 367K, avg latency = 347.33us
	-- numjobs=4 : IOPS = 383K, avg latency = 682.06us

- Fio io_uring poll with virtio-blk poll support

	-- numjobs=1 : IOPS = 385K, avg latency = 165.94us
	-- numjobs=2 : IOPS = 408K, avg latency = 313.28us
	-- numjobs=4 : IOPS = 424K, avg latency = 613.05us
---
 drivers/block/virtio_blk.c | 112 +++++++++++++++++++++++++++++++++++--
 1 file changed, 108 insertions(+), 4 deletions(-)
drivers/block/virtio_blk.c +19 -0
index 8c415be86732..712579dcd3cc 100644
@@ -37,6 +37,10 @@ MODULE_PARM_DESC(num_request_queues,
 		 "0 for no limit. "
 		 "Values > nr_cpu_ids truncated to nr_cpu_ids.");
 
+static unsigned int poll_queues;
+module_param(poll_queues, uint, 0644);
+MODULE_PARM_DESC(poll_queues, "The number of dedicated virtqueues for polling I/O");
+
 static int major;
 static DEFINE_IDA(vd_index_ida);
 
@@ -81,6 +85,7 @@ struct virtio_blk {
 
 	/* num of vqs */
 	int num_vqs;
+	int io_queues[HCTX_MAX_TYPES];
 	struct virtio_blk_vq *vqs;
 };
 
@@ -548,6 +553,7 @@ static int init_vq(struct virtio_blk *vblk)
 	const char **names;
 	struct virtqueue **vqs;
 	unsigned short num_vqs;
+	unsigned int num_poll_vqs;
 	struct virtio_device *vdev = vblk->vdev;
 	struct irq_affinity desc = { 0, };
 
@@ -556,6 +562,7 @@ static int init_vq(struct virtio_blk *vblk)
 				   &num_vqs);
 	if (err)
 		num_vqs = 1;
+
 	if (!err && !num_vqs) {
 		dev_err(&vdev->dev, "MQ advertised but zero queues reported\n");
 		return -EINVAL;
@@ -565,6 +572,18 @@ static int init_vq(struct virtio_blk *vblk)
 			min_not_zero(num_request_queues, nr_cpu_ids),
 			num_vqs);
 
+	num_poll_vqs = min_t(unsigned int, poll_queues, num_vqs - 1);
+
+	memset(vblk->io_queues, 0, sizeof(int) * HCTX_MAX_TYPES);
+	vblk->io_queues[HCTX_TYPE_DEFAULT] = num_vqs - num_poll_vqs;
+	vblk->io_queues[HCTX_TYPE_READ] = 0;
+	vblk->io_queues[HCTX_TYPE_POLL] = num_poll_vqs;
+
+	dev_info(&vdev->dev, "%d/%d/%d default/read/poll queues\n",
+				vblk->io_queues[HCTX_TYPE_DEFAULT],
+				vblk->io_queues[HCTX_TYPE_READ],
+				vblk->io_queues[HCTX_TYPE_POLL]);
+
 	vblk->vqs = kmalloc_array(num_vqs, sizeof(*vblk->vqs), GFP_KERNEL);
 	if (!vblk->vqs)
 		return -ENOMEM;
@@ -578,8 +597,13 @@ static int init_vq(struct virtio_blk *vblk)
 	}

drivers/block/virtio_blk.c static int init_vq(struct virtio_blk *vblk) · line 607

Quoted 10 lines
drivers/block/virtio_blk.c +7 -0
 
 	for (i = 0; i < num_vqs; i++) {
+		if (i < num_vqs - num_poll_vqs) {
+			callbacks[i] = virtblk_done;
+			snprintf(vblk->vqs[i].name, VQ_NAME_LEN, "req.%d", i);
+		} else {
+			callbacks[i] = NULL;
+			snprintf(vblk->vqs[i].name, VQ_NAME_LEN, "req_poll.%d", i);
+		}
 		names[i] = vblk->vqs[i].name;

This would look a little cleaner with two loops:

 	for (i = 0; i < num_vqs - num_poll_vqs; i++) {
		callbacks[i] = virtblk_done;
		snprintf(vblk->vqs[i].name, VQ_NAME_LEN, "req.%d", i);
 		names[i] = vblk->vqs[i].name;
	}
 	for (; i < num_vqs; i++) {
		callbacks[i] = NULL;
		snprintf(vblk->vqs[i].name, VQ_NAME_LEN, "req_poll.%d", i);
		names[i] = vblk->vqs[i].name;
	}

drivers/block/virtio_blk.c static const struct attribute_group *virtblk_attr_groups[] = { · line 772

Quoted 10 lines
drivers/block/virtio_blk.c +10 -0
+
+		if (map->nr_queues == 0)
+			continue;
+
+		/*
+		 * Regular queues have interrupts and hence CPU affinity is
+		 * defined by the core virtio code, but polling queues have
+		 * no interrupts so we let the block layer assign CPU affinity.
+		 */
+		if (i == HCTX_TYPE_DEFAULT)

I'd check for

		i != HCTX_TYPE_POLL

here instead to make the check a little more explicit and future proof for the potential addition of read queues (which would be a Linux only change without hypervisor or spec changes). In fact you might as well add that support now as doing it is completely trivial once a driver supports multiple map types.

drivers/block/virtio_blk.c static const struct attribute_group *virtblk_attr_groups[] = { · line 789

Quoted 9 lines
drivers/block/virtio_blk.c +9 -0
+static void virtblk_complete_batch(struct io_comp_batch *iob)
+{
+	struct request *req;
+	struct virtblk_req *vbr;
+
+	rq_list_for_each(&iob->req_list, req) {
+		vbr = blk_mq_rq_to_pdu(req);
+		virtblk_unmap_data(req, vbr);
+		virtblk_cleanup_cmd(req);

vbr is only used ones, so why not just:

Quoted 5 lines
		virtblk_unmap_data(req, blk_mq_rq_to_pdu);

?

Or even better add a cleanup patch to just remove the vbr argument to virtblk_unmap_data as it is not needed at all.

I replied that I will add cleanup path but after thinking about it again, it seems better not to remove the argument because blk_mq_rq_to_pdu() needs type casting anyway.

So I will fix it to "virtblk_unmap_data(req, blk_mq_rq_to_pdu);" as you mentioned.

Regards,
Suwan Kim

author
#7 lore raw

PATCH v4 2/2 virtio-blk: support mq_ops->queue_rqs()

In reply to Suwan Kim

This patch supports mq_ops->queue_rqs() hook. It has an advantage of batch submission to virtio-blk driver. It also helps polling I/O because polling uses batched completion of block layer. Batch submission in queue_rqs() can boost polling performance.

In queue_rqs(), it iterates plug->mq_list, collects requests that belong to same HW queue until it encounters a request from other HW queue or sees the end of the list.
Then, virtio-blk adds requests into virtqueue and kicks virtqueue to submit requests.

If there is an error, it inserts error request to requeue_list and passes it to ordinary block layer path.

For verification, I did fio test.
(io_uring, randread, direct=1, bs=4K, iodepth=64 numjobs=N) I set 4 vcpu and 2 virtio-blk queues for VM and run fio test 5 times. It shows about 2% improvement.

| numjobs=2 | numjobs=4
-----------------------------------------------------------
fio without queue_rqs() | 291K IOPS | 238K IOPS
-----------------------------------------------------------
fio with queue_rqs() | 295K IOPS | 243K IOPS

For polling I/O performance, I also did fio test as below. (io_uring, hipri, randread, direct=1, bs=512, iodepth=64 numjobs=4) I set 4 vcpu and 2 poll queues for VM.
It shows about 2% improvement in polling I/O.

| IOPS | avg latency
-----------------------------------------------------------
fio poll without queue_rqs() | 424K | 613.05 usec
-----------------------------------------------------------
fio poll with queue_rqs() | 435K | 601.01 usec

---
 drivers/block/virtio_blk.c | 110 +++++++++++++++++++++++++++++++++----
 1 file changed, 99 insertions(+), 11 deletions(-)
drivers/block/virtio_blk.c +99 -11
index 712579dcd3cc..a091034bc551 100644
@@ -92,6 +92,7 @@ struct virtio_blk {
 struct virtblk_req {
 	struct virtio_blk_outhdr out_hdr;
 	u8 status;
+	int sg_num;
 	struct sg_table sg_table;
 	struct scatterlist sg[];
 };
@@ -311,18 +312,13 @@ static void virtio_commit_rqs(struct blk_mq_hw_ctx *hctx)
 		virtqueue_notify(vq->vq);
 }
 
-static blk_status_t virtio_queue_rq(struct blk_mq_hw_ctx *hctx,
-			   const struct blk_mq_queue_data *bd)
+static blk_status_t virtblk_prep_rq(struct blk_mq_hw_ctx *hctx,
+					struct virtio_blk *vblk,
+					struct request *req,
+					struct virtblk_req *vbr)
 {
-	struct virtio_blk *vblk = hctx->queue->queuedata;
-	struct request *req = bd->rq;
-	struct virtblk_req *vbr = blk_mq_rq_to_pdu(req);
-	unsigned long flags;
-	int num;
-	int qid = hctx->queue_num;
-	bool notify = false;
 	blk_status_t status;
-	int err;
+	int num;
 
 	status = virtblk_setup_cmd(vblk->vdev, req, vbr);
 	if (unlikely(status))
@@ -335,9 +331,30 @@ static blk_status_t virtio_queue_rq(struct blk_mq_hw_ctx *hctx,
 		virtblk_cleanup_cmd(req);
 		return BLK_STS_RESOURCE;
 	}
+	vbr->sg_num = num;
+
+	return BLK_STS_OK;
+}
+
+static blk_status_t virtio_queue_rq(struct blk_mq_hw_ctx *hctx,
+			   const struct blk_mq_queue_data *bd)
+{
+	struct virtio_blk *vblk = hctx->queue->queuedata;
+	struct request *req = bd->rq;
+	struct virtblk_req *vbr = blk_mq_rq_to_pdu(req);
+	unsigned long flags;
+	int qid = hctx->queue_num;
+	bool notify = false;
+	blk_status_t status;
+	int err;
+
+	status = virtblk_prep_rq(hctx, vblk, req, vbr);
+	if (unlikely(status))
+		return status;
 
 	spin_lock_irqsave(&vblk->vqs[qid].lock, flags);
-	err = virtblk_add_req(vblk->vqs[qid].vq, vbr, vbr->sg_table.sgl, num);
+	err = virtblk_add_req(vblk->vqs[qid].vq, vbr,
+				vbr->sg_table.sgl, vbr->sg_num);
 	if (err) {
 		virtqueue_kick(vblk->vqs[qid].vq);
 		/* Don't stop the queue if -ENOMEM: we may have failed to
@@ -367,6 +384,76 @@ static blk_status_t virtio_queue_rq(struct blk_mq_hw_ctx *hctx,
 	return BLK_STS_OK;
 }
 
+static bool virtblk_prep_rq_batch(struct request *req)
+{
+	struct virtio_blk *vblk = req->mq_hctx->queue->queuedata;
+	struct virtblk_req *vbr = blk_mq_rq_to_pdu(req);
+
+	req->mq_hctx->tags->rqs[req->tag] = req;
+
+	return virtblk_prep_rq(req->mq_hctx, vblk, req, vbr) == BLK_STS_OK;
+}
+
+static bool virtblk_add_req_batch(struct virtio_blk_vq *vq,
+					struct request **rqlist,
+					struct request **requeue_list)
+{
+	unsigned long flags;
+	int err;
+	bool kick;
+
+	spin_lock_irqsave(&vq->lock, flags);
+
+	while (!rq_list_empty(*rqlist)) {
+		struct request *req = rq_list_pop(rqlist);
+		struct virtblk_req *vbr = blk_mq_rq_to_pdu(req);
+
+		err = virtblk_add_req(vq->vq, vbr,
+					vbr->sg_table.sgl, vbr->sg_num);
+		if (err) {
+			virtblk_unmap_data(req, vbr);
+			virtblk_cleanup_cmd(req);
+			rq_list_add(requeue_list, req);
+		}
+	}
+
+	kick = virtqueue_kick_prepare(vq->vq);
+	spin_unlock_irqrestore(&vq->lock, flags);
+
+	return kick;
+}
+
+static void virtio_queue_rqs(struct request **rqlist)
+{
+	struct request *req, *next, *prev = NULL;
+	struct request *requeue_list = NULL;
+
+	rq_list_for_each_safe(rqlist, req, next) {
+		struct virtio_blk_vq *vq = req->mq_hctx->driver_data;
+		bool kick;
+
+		if (!virtblk_prep_rq_batch(req)) {
+			rq_list_move(rqlist, &requeue_list, req, prev);
+			req = prev;
+			if (!req)
+				continue;
+		}
+
+		if (!next || req->mq_hctx != next->mq_hctx) {
+			req->rq_next = NULL;
+			kick = virtblk_add_req_batch(vq, rqlist, &requeue_list);
+			if (kick)
+				virtqueue_notify(vq->vq);
+
+			*rqlist = next;
+			prev = NULL;
+		} else
+			prev = req;
+	}
+
+	*rqlist = requeue_list;
+}
+
 /* return id (s/n) string for *disk to *id_str
  */
 static int virtblk_get_id(struct gendisk *disk, char *id_str)
@@ -834,6 +921,7 @@ static int virtblk_init_hctx(struct blk_mq_hw_ctx *hctx, void *data,
 
 static const struct blk_mq_ops virtio_mq_ops = {
 	.queue_rq	= virtio_queue_rq,
+	.queue_rqs	= virtio_queue_rqs,
 	.commit_rqs	= virtio_commit_rqs,
 	.init_hctx	= virtblk_init_hctx,
 	.complete	= virtblk_request_done,
Signature · 1 line
2.26.3
#8 lore raw

Re: [PATCH v4 2/2] virtio-blk: support mq_ops->queue_rqs()

Quoted 78 lines · Suwan Kim

This patch supports mq_ops->queue_rqs() hook. It has an advantage of batch submission to virtio-blk driver. It also helps polling I/O because polling uses batched completion of block layer. Batch submission in queue_rqs() can boost polling performance.

In queue_rqs(), it iterates plug->mq_list, collects requests that belong to same HW queue until it encounters a request from other HW queue or sees the end of the list.
Then, virtio-blk adds requests into virtqueue and kicks virtqueue to submit requests.

If there is an error, it inserts error request to requeue_list and passes it to ordinary block layer path.

For verification, I did fio test.
(io_uring, randread, direct=1, bs=4K, iodepth=64 numjobs=N) I set 4 vcpu and 2 virtio-blk queues for VM and run fio test 5 times. It shows about 2% improvement.

| numjobs=2 | numjobs=4
-----------------------------------------------------------
fio without queue_rqs() | 291K IOPS | 238K IOPS
-----------------------------------------------------------
fio with queue_rqs() | 295K IOPS | 243K IOPS

For polling I/O performance, I also did fio test as below. (io_uring, hipri, randread, direct=1, bs=512, iodepth=64 numjobs=4) I set 4 vcpu and 2 poll queues for VM.
It shows about 2% improvement in polling I/O.

| IOPS | avg latency
-----------------------------------------------------------
fio poll without queue_rqs() | 424K | 613.05 usec
-----------------------------------------------------------
fio poll with queue_rqs() | 435K | 601.01 usec

---
 drivers/block/virtio_blk.c | 110 +++++++++++++++++++++++++++++++++----
 1 file changed, 99 insertions(+), 11 deletions(-)
drivers/block/virtio_blk.c +6 -10
index 712579dcd3cc..a091034bc551 100644
@@ -92,6 +92,7 @@ struct virtio_blk {
 struct virtblk_req {
 	struct virtio_blk_outhdr out_hdr;
 	u8 status;
+	int sg_num;
 	struct sg_table sg_table;
 	struct scatterlist sg[];
 };
@@ -311,18 +312,13 @@ static void virtio_commit_rqs(struct blk_mq_hw_ctx *hctx)
 		virtqueue_notify(vq->vq);
 }
 
-static blk_status_t virtio_queue_rq(struct blk_mq_hw_ctx *hctx,
-			   const struct blk_mq_queue_data *bd)
+static blk_status_t virtblk_prep_rq(struct blk_mq_hw_ctx *hctx,
+					struct virtio_blk *vblk,
+					struct request *req,
+					struct virtblk_req *vbr)
 {
-	struct virtio_blk *vblk = hctx->queue->queuedata;
-	struct request *req = bd->rq;
-	struct virtblk_req *vbr = blk_mq_rq_to_pdu(req);
-	unsigned long flags;
-	int num;
-	int qid = hctx->queue_num;
-	bool notify = false;
 	blk_status_t status;
-	int err;
+	int num;
 
 	status = virtblk_setup_cmd(vblk->vdev, req, vbr);
 	if (unlikely(status))

drivers/block/virtio_blk.c static blk_status_t virtio_queue_rq(struct blk_mq_hw_ctx *hctx, · line 334

Quoted 4 lines
drivers/block/virtio_blk.c +1 -0
@@ -335,9 +331,30 @@ static blk_status_t virtio_queue_rq(struct blk_mq_hw_ctx *hctx,
 		virtblk_cleanup_cmd(req);
 		return BLK_STS_RESOURCE;
 	}
+	vbr->sg_num = num;

This can go into the nents field of vbr->sg_table.

drivers/block/virtio_blk.c static blk_status_t virtio_queue_rq(struct blk_mq_hw_ctx *hctx, · line 357

Quoted 10 lines
drivers/block/virtio_blk.c +7 -1
+	int err;
+
+	status = virtblk_prep_rq(hctx, vblk, req, vbr);
+	if (unlikely(status))
+		return status;
 
 	spin_lock_irqsave(&vblk->vqs[qid].lock, flags);
-	err = virtblk_add_req(vblk->vqs[qid].vq, vbr, vbr->sg_table.sgl, num);
+	err = virtblk_add_req(vblk->vqs[qid].vq, vbr,
+				vbr->sg_table.sgl, vbr->sg_num);

And while we're at it - virtblk_add_req can lose the data_sg and have_data arguments as they can be derived from vbr.
_______________________________________________
Virtualization mailing list [email protected]
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

author
#9 lore raw

Re: [PATCH v4 2/2] virtio-blk: support mq_ops->queue_rqs()

Quoted 92 lines · Christoph Hellwig
Quoted 78 lines · Suwan Kim

This patch supports mq_ops->queue_rqs() hook. It has an advantage of batch submission to virtio-blk driver. It also helps polling I/O because polling uses batched completion of block layer. Batch submission in queue_rqs() can boost polling performance.

In queue_rqs(), it iterates plug->mq_list, collects requests that belong to same HW queue until it encounters a request from other HW queue or sees the end of the list.
Then, virtio-blk adds requests into virtqueue and kicks virtqueue to submit requests.

If there is an error, it inserts error request to requeue_list and passes it to ordinary block layer path.

For verification, I did fio test.
(io_uring, randread, direct=1, bs=4K, iodepth=64 numjobs=N) I set 4 vcpu and 2 virtio-blk queues for VM and run fio test 5 times. It shows about 2% improvement.

| numjobs=2 | numjobs=4
-----------------------------------------------------------
fio without queue_rqs() | 291K IOPS | 238K IOPS
-----------------------------------------------------------
fio with queue_rqs() | 295K IOPS | 243K IOPS

For polling I/O performance, I also did fio test as below. (io_uring, hipri, randread, direct=1, bs=512, iodepth=64 numjobs=4) I set 4 vcpu and 2 poll queues for VM.
It shows about 2% improvement in polling I/O.

| IOPS | avg latency
-----------------------------------------------------------
fio poll without queue_rqs() | 424K | 613.05 usec
-----------------------------------------------------------
fio poll with queue_rqs() | 435K | 601.01 usec

---
 drivers/block/virtio_blk.c | 110 +++++++++++++++++++++++++++++++++----
 1 file changed, 99 insertions(+), 11 deletions(-)
drivers/block/virtio_blk.c +6 -10
index 712579dcd3cc..a091034bc551 100644
@@ -92,6 +92,7 @@ struct virtio_blk {
 struct virtblk_req {
 	struct virtio_blk_outhdr out_hdr;
 	u8 status;
+	int sg_num;
 	struct sg_table sg_table;
 	struct scatterlist sg[];
 };
@@ -311,18 +312,13 @@ static void virtio_commit_rqs(struct blk_mq_hw_ctx *hctx)
 		virtqueue_notify(vq->vq);
 }
 
-static blk_status_t virtio_queue_rq(struct blk_mq_hw_ctx *hctx,
-			   const struct blk_mq_queue_data *bd)
+static blk_status_t virtblk_prep_rq(struct blk_mq_hw_ctx *hctx,
+					struct virtio_blk *vblk,
+					struct request *req,
+					struct virtblk_req *vbr)
 {
-	struct virtio_blk *vblk = hctx->queue->queuedata;
-	struct request *req = bd->rq;
-	struct virtblk_req *vbr = blk_mq_rq_to_pdu(req);
-	unsigned long flags;
-	int num;
-	int qid = hctx->queue_num;
-	bool notify = false;
 	blk_status_t status;
-	int err;
+	int num;
 
 	status = virtblk_setup_cmd(vblk->vdev, req, vbr);
 	if (unlikely(status))

drivers/block/virtio_blk.c static blk_status_t virtio_queue_rq(struct blk_mq_hw_ctx *hctx, · line 334

Quoted 4 lines
drivers/block/virtio_blk.c +1 -0
@@ -335,9 +331,30 @@ static blk_status_t virtio_queue_rq(struct blk_mq_hw_ctx *hctx,
 		virtblk_cleanup_cmd(req);
 		return BLK_STS_RESOURCE;
 	}
+	vbr->sg_num = num;

This can go into the nents field of vbr->sg_table.

Quoted 6 lines
drivers/block/virtio_blk.c +5 -0
+	int err;
+
+	status = virtblk_prep_rq(hctx, vblk, req, vbr);
+	if (unlikely(status))
+		return status;
 
Quoted 7 lines

drivers/block/virtio_blk.c static blk_status_t virtio_queue_rq(struct blk_mq_hw_ctx *hctx, · line 357

Quoted 4 lines
drivers/block/virtio_blk.c +2 -1
 	spin_lock_irqsave(&vblk->vqs[qid].lock, flags);
-	err = virtblk_add_req(vblk->vqs[qid].vq, vbr, vbr->sg_table.sgl, num);
+	err = virtblk_add_req(vblk->vqs[qid].vq, vbr,
+				vbr->sg_table.sgl, vbr->sg_num);

And while we're at it - virtblk_add_req can lose the data_sg and have_data arguments as they can be derived from vbr.

Ok. I will remove vbr->sg_num and save it to vbr->sg_table.nents.

Regards,
Suwan Kim

reviewer Reviewed-by on v4 2/2
#10 lore raw

Re: [PATCH v4 2/2] virtio-blk: support mq_ops->queue_rqs()

In reply to Suwan Kim
Quoted 36 lines · Suwan Kim

This patch supports mq_ops->queue_rqs() hook. It has an advantage of batch submission to virtio-blk driver. It also helps polling I/O because polling uses batched completion of block layer. Batch submission in queue_rqs() can boost polling performance.

In queue_rqs(), it iterates plug->mq_list, collects requests that belong to same HW queue until it encounters a request from other HW queue or sees the end of the list.
Then, virtio-blk adds requests into virtqueue and kicks virtqueue to submit requests.

If there is an error, it inserts error request to requeue_list and passes it to ordinary block layer path.

For verification, I did fio test.
(io_uring, randread, direct=1, bs=4K, iodepth=64 numjobs=N) I set 4 vcpu and 2 virtio-blk queues for VM and run fio test 5 times. It shows about 2% improvement.

| numjobs=2 | numjobs=4
-----------------------------------------------------------
fio without queue_rqs() | 291K IOPS | 238K IOPS
-----------------------------------------------------------
fio with queue_rqs() | 295K IOPS | 243K IOPS

For polling I/O performance, I also did fio test as below. (io_uring, hipri, randread, direct=1, bs=512, iodepth=64 numjobs=4) I set 4 vcpu and 2 poll queues for VM.
It shows about 2% improvement in polling I/O.

| IOPS | avg latency
-----------------------------------------------------------
fio poll without queue_rqs() | 424K | 613.05 usec
-----------------------------------------------------------
fio poll with queue_rqs() | 435K | 601.01 usec

Quoted 4 lines
---
 drivers/block/virtio_blk.c | 110 +++++++++++++++++++++++++++++++++----
 1 file changed, 99 insertions(+), 11 deletions(-)

_______________________________________________
Virtualization mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/virtualization