Skip to contentSkip to conversation

basic in order support for vhost_net

PATCH In review Jason Wang 2 participants 10 messages Jan 9, 2019

Subsystems (5)
lore
Share & tools
lore

Jason Wang

author
#0 lore raw

v2 cover basic in order support for vhost_net

Hi:

This series implement basic in order feature support for vhost_net. This feature requires both driver and device to use descriptors in order which can simplify the implementation and optimizaton for both side. The series also implement a simple optimization that avoid read available ring. Test shows 10% performance improvement at most.

More optimizations could be done on top.

Changes from V1:
- no code changes
- add result of SMAP off

Jason Wang (3):
virtio: introduce in order feature bit
vhost_net: support in order feature
vhost: don't touch avail ring if in_order is negotiated

 drivers/vhost/net.c                |  6 ++++--
 drivers/vhost/vhost.c              | 19 ++++++++++++-------
 include/uapi/linux/virtio_config.h |  6 ++++++
 3 files changed, 22 insertions(+), 9 deletions(-)
Signature · 1 line
2.17.1
author
#1 lore raw

PATCH v2 1/3 virtio: introduce in order feature bit

---
 include/uapi/linux/virtio_config.h | 6 ++++++
 1 file changed, 6 insertions(+)
include/uapi/linux/virtio_config.h +6 -0
index 1196e1c1d4f6..2698e069ed9e 100644
@@ -78,6 +78,12 @@
 /* This feature indicates support for the packed virtqueue layout. */
 #define VIRTIO_F_RING_PACKED		34
 
+/*
+ * Device uses buffers in the same order in which they have been
+ * available.
+ */
+#define VIRTIO_F_IN_ORDER		35
+
 /*
  * Does the device support Single Root I/O Virtualization?
  */
Signature · 1 line
2.17.1
#2 lore raw

Re: [PATCH net-next V2 1/3] virtio: introduce in order feature bit

Quoted 10 lines · Jason Wang
---
 include/uapi/linux/virtio_config.h | 6 ++++++
 1 file changed, 6 insertions(+)
include/uapi/linux/virtio_config.h +0 -0
index 1196e1c1d4f6..2698e069ed9e 100644

include/uapi/linux/virtio_config.h line 83

Quoted 6 lines
include/uapi/linux/virtio_config.h +3 -0
@@ -78,6 +78,12 @@
 /* This feature indicates support for the packed virtqueue layout. */
 #define VIRTIO_F_RING_PACKED		34
 
+/*
+ * Device uses buffers in the same order in which they have been
+ * available.

been *made* available

Quoted 8 lines
include/uapi/linux/virtio_config.h +3 -0
+ */
+#define VIRTIO_F_IN_ORDER		35
+
 /*
  * Does the device support Single Root I/O Virtualization?
  */
Signature · 1 line
2.17.1
#3 lore raw

Re: [PATCH net-next V2 1/3] virtio: introduce in order feature bit

In reply to Jason Wang
Quoted 10 lines · Jason Wang
---
 include/uapi/linux/virtio_config.h | 6 ++++++
 1 file changed, 6 insertions(+)
include/uapi/linux/virtio_config.h +0 -0
index 1196e1c1d4f6..2698e069ed9e 100644

include/uapi/linux/virtio_config.h line 83

Quoted 6 lines
include/uapi/linux/virtio_config.h +3 -0
@@ -78,6 +78,12 @@
 /* This feature indicates support for the packed virtqueue layout. */
 #define VIRTIO_F_RING_PACKED		34
 
+/*
+ * Device uses buffers in the same order in which they have been
+ * available.

been *made* available

Quoted 8 lines
include/uapi/linux/virtio_config.h +3 -0
+ */
+#define VIRTIO_F_IN_ORDER		35
+
 /*
  * Does the device support Single Root I/O Virtualization?
  */
Signature · 1 line
2.17.1
author
#4 lore raw

PATCH v2 1/3 virtio: introduce in order feature bit

In reply to Jason Wang
---
 include/uapi/linux/virtio_config.h | 6 ++++++
 1 file changed, 6 insertions(+)
include/uapi/linux/virtio_config.h +6 -0
index 1196e1c1d4f6..2698e069ed9e 100644
@@ -78,6 +78,12 @@
 /* This feature indicates support for the packed virtqueue layout. */
 #define VIRTIO_F_RING_PACKED		34
 
+/*
+ * Device uses buffers in the same order in which they have been
+ * available.
+ */
+#define VIRTIO_F_IN_ORDER		35
+
 /*
  * Does the device support Single Root I/O Virtualization?
  */
Signature · 1 line
2.17.1
author
#5 lore raw

PATCH v2 2/3 vhost_net: support in order feature

In reply to Jason Wang

This makes vhost_net to support in order feature. This is as simple as use datacopy path when it was negotiated. An alternative is not to advertise in order when zerocopy is enabled which tends to be suboptimal consider zerocopy may suffer from e.g HOL issues.

---
 drivers/vhost/net.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
drivers/vhost/net.c +4 -2
index 36f3d0f49e60..0870f51a1c76 100644
@@ -74,7 +74,8 @@ enum {
 	VHOST_NET_FEATURES = VHOST_FEATURES |
 			 (1ULL << VHOST_NET_F_VIRTIO_NET_HDR) |
 			 (1ULL << VIRTIO_NET_F_MRG_RXBUF) |
-			 (1ULL << VIRTIO_F_IOMMU_PLATFORM)
+			 (1ULL << VIRTIO_F_IOMMU_PLATFORM) |
+	                 (1ULL << VIRTIO_F_IN_ORDER)
 };
 
 enum {
@@ -977,7 +978,8 @@ static void handle_tx(struct vhost_net *net)
 	vhost_disable_notify(&net->dev, vq);
 	vhost_net_disable_vq(net, vq);
 
-	if (vhost_sock_zcopy(sock))
+	if (vhost_sock_zcopy(sock) &&
+	    !vhost_has_feature(vq, VIRTIO_F_IN_ORDER))
 		handle_tx_zerocopy(net, sock);
 	else
 		handle_tx_copy(net, sock);
Signature · 1 line
2.17.1
author
#6 lore raw

PATCH v2 2/3 vhost_net: support in order feature

In reply to Jason Wang

This makes vhost_net to support in order feature. This is as simple as use datacopy path when it was negotiated. An alternative is not to advertise in order when zerocopy is enabled which tends to be suboptimal consider zerocopy may suffer from e.g HOL issues.

---
 drivers/vhost/net.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
drivers/vhost/net.c +4 -2
index 36f3d0f49e60..0870f51a1c76 100644
@@ -74,7 +74,8 @@ enum {
 	VHOST_NET_FEATURES = VHOST_FEATURES |
 			 (1ULL << VHOST_NET_F_VIRTIO_NET_HDR) |
 			 (1ULL << VIRTIO_NET_F_MRG_RXBUF) |
-			 (1ULL << VIRTIO_F_IOMMU_PLATFORM)
+			 (1ULL << VIRTIO_F_IOMMU_PLATFORM) |
+	                 (1ULL << VIRTIO_F_IN_ORDER)
 };
 
 enum {
@@ -977,7 +978,8 @@ static void handle_tx(struct vhost_net *net)
 	vhost_disable_notify(&net->dev, vq);
 	vhost_net_disable_vq(net, vq);
 
-	if (vhost_sock_zcopy(sock))
+	if (vhost_sock_zcopy(sock) &&
+	    !vhost_has_feature(vq, VIRTIO_F_IN_ORDER))
 		handle_tx_zerocopy(net, sock);
 	else
 		handle_tx_copy(net, sock);
Signature · 1 line
2.17.1
author
#7 lore raw

PATCH v2 3/3 vhost: don't touch avail ring if in_order is negotiated

In reply to Jason Wang

Device use descriptors table in order, so there's no need to read index from available ring. This eliminate the cache contention on available ring completely.

Virito-user + vhost_kernel + XDP_DROP on 2.60GHz Broadwell

Before /After

SMAP on:  4.8Mpps   5.3Mpps(+10%)
SMAP off: 6.6Mpps   7.0Mpps(+6%)
---
 drivers/vhost/vhost.c | 19 ++++++++++++-------
 1 file changed, 12 insertions(+), 7 deletions(-)
drivers/vhost/vhost.c +12 -7
index 55e5aa662ad5..ab0d05262235 100644
@@ -2012,6 +2012,7 @@ int vhost_get_vq_desc(struct vhost_virtqueue *vq,
 	__virtio16 avail_idx;
 	__virtio16 ring_head;
 	int ret, access;
+	bool in_order = vhost_has_feature(vq, VIRTIO_F_IN_ORDER);
 
 	/* Check it isn't doing very strange things with descriptor numbers. */
 	last_avail_idx = vq->last_avail_idx;
@@ -2044,15 +2045,19 @@ int vhost_get_vq_desc(struct vhost_virtqueue *vq,
 
 	/* Grab the next descriptor number they're advertising, and increment
 	 * the index we've seen. */
-	if (unlikely(vhost_get_avail(vq, ring_head,
-		     &vq->avail->ring[last_avail_idx & (vq->num - 1)]))) {
-		vq_err(vq, "Failed to read head: idx %d address %p\n",
-		       last_avail_idx,
-		       &vq->avail->ring[last_avail_idx % vq->num]);
-		return -EFAULT;
+	if (!in_order) {
+		if (unlikely(vhost_get_avail(vq, ring_head,
+		    &vq->avail->ring[last_avail_idx & (vq->num - 1)]))) {
+			vq_err(vq, "Failed to read head: idx %d address %p\n",
+				last_avail_idx,
+				&vq->avail->ring[last_avail_idx % vq->num]);
+			return -EFAULT;
+		}
+		head = vhost16_to_cpu(vq, ring_head);
+	} else {
+		head = last_avail_idx & (vq->num - 1);
 	}
 
-	head = vhost16_to_cpu(vq, ring_head);
 
 	/* If their number is silly, that's an error. */
 	if (unlikely(head >= vq->num)) {
Signature · 1 line
2.17.1
author
#8 lore raw

PATCH v2 3/3 vhost: don't touch avail ring if in_order is negotiated

In reply to Jason Wang

Device use descriptors table in order, so there's no need to read index from available ring. This eliminate the cache contention on available ring completely.

Virito-user + vhost_kernel + XDP_DROP on 2.60GHz Broadwell

Before /After

SMAP on:  4.8Mpps   5.3Mpps(+10%)
SMAP off: 6.6Mpps   7.0Mpps(+6%)
---
 drivers/vhost/vhost.c | 19 ++++++++++++-------
 1 file changed, 12 insertions(+), 7 deletions(-)
drivers/vhost/vhost.c +12 -7
index 55e5aa662ad5..ab0d05262235 100644
@@ -2012,6 +2012,7 @@ int vhost_get_vq_desc(struct vhost_virtqueue *vq,
 	__virtio16 avail_idx;
 	__virtio16 ring_head;
 	int ret, access;
+	bool in_order = vhost_has_feature(vq, VIRTIO_F_IN_ORDER);
 
 	/* Check it isn't doing very strange things with descriptor numbers. */
 	last_avail_idx = vq->last_avail_idx;
@@ -2044,15 +2045,19 @@ int vhost_get_vq_desc(struct vhost_virtqueue *vq,
 
 	/* Grab the next descriptor number they're advertising, and increment
 	 * the index we've seen. */
-	if (unlikely(vhost_get_avail(vq, ring_head,
-		     &vq->avail->ring[last_avail_idx & (vq->num - 1)]))) {
-		vq_err(vq, "Failed to read head: idx %d address %p\n",
-		       last_avail_idx,
-		       &vq->avail->ring[last_avail_idx % vq->num]);
-		return -EFAULT;
+	if (!in_order) {
+		if (unlikely(vhost_get_avail(vq, ring_head,
+		    &vq->avail->ring[last_avail_idx & (vq->num - 1)]))) {
+			vq_err(vq, "Failed to read head: idx %d address %p\n",
+				last_avail_idx,
+				&vq->avail->ring[last_avail_idx % vq->num]);
+			return -EFAULT;
+		}
+		head = vhost16_to_cpu(vq, ring_head);
+	} else {
+		head = last_avail_idx & (vq->num - 1);
 	}
 
-	head = vhost16_to_cpu(vq, ring_head);
 
 	/* If their number is silly, that's an error. */
 	if (unlikely(head >= vq->num)) {
Signature · 1 line
2.17.1
author
#9 lore raw
In reply to Jason Wang
Quoted 9 lines · Jason Wang

Hi:

This series implement basic in order feature support for vhost_net. This feature requires both driver and device to use descriptors in order which can simplify the implementation and optimizaton for both side. The series also implement a simple optimization that avoid read available ring. Test shows 10% performance improvement at most.

Quoted 5 lines

More optimizations could be done on top.

Changes from V1:
- no code changes
- add result of SMAP off

Just notice the patch works only for some specific case e.g a buffer only contain one descriptor. Will respin.

Thanks

Quoted 11 lines

Jason Wang (3):
virtio: introduce in order feature bit
vhost_net: support in order feature
vhost: don't touch avail ring if in_order is negotiated

  drivers/vhost/net.c                |  6 ++++--
  drivers/vhost/vhost.c              | 19 ++++++++++++-------
  include/uapi/linux/virtio_config.h |  6 ++++++
  3 files changed, 22 insertions(+), 9 deletions(-)

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