diff mbox series

[v6,09/15] vhost: add shadow used ring support for packed rings

Message ID 20180702081629.29258-10-maxime.coquelin@redhat.com (mailing list archive)
State Superseded, archived
Delegated to: Maxime Coquelin
Headers show
Series Vhost: add support to packed ring layout | expand

Checks

Context Check Description
ci/Intel-compilation fail Compilation issues

Commit Message

Maxime Coquelin July 2, 2018, 8:16 a.m. UTC
Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>
---
 lib/librte_vhost/vhost.c      |  9 ++++--
 lib/librte_vhost/vhost.h      | 13 ++++++--
 lib/librte_vhost/vhost_user.c | 64 ++++++++++++++++++++++++++++----------
 lib/librte_vhost/virtio_net.c | 72 +++++++++++++++++++++++++++++++++++++++++--
 4 files changed, 134 insertions(+), 24 deletions(-)
diff mbox series

Patch

diff --git a/lib/librte_vhost/vhost.c b/lib/librte_vhost/vhost.c
index 05d2bb182..8538302c9 100644
--- a/lib/librte_vhost/vhost.c
+++ b/lib/librte_vhost/vhost.c
@@ -94,9 +94,12 @@  cleanup_device(struct virtio_net *dev, int destroy)
 }
 
 void
-free_vq(struct vhost_virtqueue *vq)
+free_vq(struct virtio_net *dev, struct vhost_virtqueue *vq)
 {
-	rte_free(vq->shadow_used_ring);
+	if (vq_is_packed(dev))
+		rte_free(vq->shadow_used_packed);
+	else
+		rte_free(vq->shadow_used_split);
 	rte_free(vq->batch_copy_elems);
 	rte_mempool_free(vq->iotlb_pool);
 	rte_free(vq);
@@ -111,7 +114,7 @@  free_device(struct virtio_net *dev)
 	uint32_t i;
 
 	for (i = 0; i < dev->nr_vring; i++)
-		free_vq(dev->virtqueue[i]);
+		free_vq(dev, dev->virtqueue[i]);
 
 	rte_free(dev);
 }
diff --git a/lib/librte_vhost/vhost.h b/lib/librte_vhost/vhost.h
index 35beff699..a7320469a 100644
--- a/lib/librte_vhost/vhost.h
+++ b/lib/librte_vhost/vhost.h
@@ -80,6 +80,12 @@  struct log_cache_entry {
 	unsigned long val;
 };
 
+struct vring_used_elem_packed {
+	uint16_t id;
+	uint32_t len;
+	uint32_t count;
+};
+
 /**
  * Structure contains variables relevant to RX/TX virtqueues.
  */
@@ -119,7 +125,10 @@  struct vhost_virtqueue {
 	struct zcopy_mbuf	*zmbufs;
 	struct zcopy_mbuf_list	zmbuf_list;
 
-	struct vring_used_elem  *shadow_used_ring;
+	union {
+		struct vring_used_elem  *shadow_used_split;
+		struct vring_used_elem_packed *shadow_used_packed;
+	};
 	uint16_t                shadow_used_idx;
 	struct vhost_vring_addr ring_addrs;
 
@@ -552,7 +561,7 @@  void reset_device(struct virtio_net *dev);
 void vhost_destroy_device(int);
 
 void cleanup_vq(struct vhost_virtqueue *vq, int destroy);
-void free_vq(struct vhost_virtqueue *vq);
+void free_vq(struct virtio_net *dev, struct vhost_virtqueue *vq);
 
 int alloc_vring_queue(struct virtio_net *dev, uint32_t vring_idx);
 
diff --git a/lib/librte_vhost/vhost_user.c b/lib/librte_vhost/vhost_user.c
index ae03ec29b..b2b57de57 100644
--- a/lib/librte_vhost/vhost_user.c
+++ b/lib/librte_vhost/vhost_user.c
@@ -244,7 +244,7 @@  vhost_user_set_features(struct virtio_net *dev, uint64_t features)
 
 			dev->virtqueue[dev->nr_vring] = NULL;
 			cleanup_vq(vq, 1);
-			free_vq(vq);
+			free_vq(dev, vq);
 		}
 	}
 
@@ -293,13 +293,26 @@  vhost_user_set_vring_num(struct virtio_net *dev,
 		TAILQ_INIT(&vq->zmbuf_list);
 	}
 
-	vq->shadow_used_ring = rte_malloc(NULL,
+	if (vq_is_packed(dev)) {
+		vq->shadow_used_packed = rte_malloc(NULL,
+				vq->size *
+				sizeof(struct vring_used_elem_packed),
+				RTE_CACHE_LINE_SIZE);
+		if (!vq->shadow_used_packed) {
+			RTE_LOG(ERR, VHOST_CONFIG,
+					"failed to allocate memory for shadow used ring.\n");
+			return -1;
+		}
+
+	} else {
+		vq->shadow_used_split = rte_malloc(NULL,
 				vq->size * sizeof(struct vring_used_elem),
 				RTE_CACHE_LINE_SIZE);
-	if (!vq->shadow_used_ring) {
-		RTE_LOG(ERR, VHOST_CONFIG,
-			"failed to allocate memory for shadow used ring.\n");
-		return -1;
+		if (!vq->shadow_used_split) {
+			RTE_LOG(ERR, VHOST_CONFIG,
+					"failed to allocate memory for shadow used ring.\n");
+			return -1;
+		}
 	}
 
 	vq->batch_copy_elems = rte_malloc(NULL,
@@ -326,7 +339,8 @@  numa_realloc(struct virtio_net *dev, int index)
 	struct virtio_net *old_dev;
 	struct vhost_virtqueue *old_vq, *vq;
 	struct zcopy_mbuf *new_zmbuf;
-	struct vring_used_elem *new_shadow_used_ring;
+	struct vring_used_elem *new_shadow_used_split;
+	struct vring_used_elem_packed *new_shadow_used_packed;
 	struct batch_copy_elem *new_batch_copy_elems;
 	int ret;
 
@@ -361,13 +375,26 @@  numa_realloc(struct virtio_net *dev, int index)
 			vq->zmbufs = new_zmbuf;
 		}
 
-		new_shadow_used_ring = rte_malloc_socket(NULL,
-			vq->size * sizeof(struct vring_used_elem),
-			RTE_CACHE_LINE_SIZE,
-			newnode);
-		if (new_shadow_used_ring) {
-			rte_free(vq->shadow_used_ring);
-			vq->shadow_used_ring = new_shadow_used_ring;
+		if (vq_is_packed(dev)) {
+			new_shadow_used_packed = rte_malloc_socket(NULL,
+					vq->size *
+					sizeof(struct vring_used_elem_packed),
+					RTE_CACHE_LINE_SIZE,
+					newnode);
+			if (new_shadow_used_packed) {
+				rte_free(vq->shadow_used_packed);
+				vq->shadow_used_packed = new_shadow_used_packed;
+			}
+		} else {
+			new_shadow_used_split = rte_malloc_socket(NULL,
+					vq->size *
+					sizeof(struct vring_used_elem),
+					RTE_CACHE_LINE_SIZE,
+					newnode);
+			if (new_shadow_used_split) {
+				rte_free(vq->shadow_used_split);
+				vq->shadow_used_split = new_shadow_used_split;
+			}
 		}
 
 		new_batch_copy_elems = rte_malloc_socket(NULL,
@@ -1067,8 +1094,13 @@  vhost_user_get_vring_base(struct virtio_net *dev,
 
 	if (dev->dequeue_zero_copy)
 		free_zmbufs(vq);
-	rte_free(vq->shadow_used_ring);
-	vq->shadow_used_ring = NULL;
+	if (vq_is_packed(dev)) {
+		rte_free(vq->shadow_used_packed);
+		vq->shadow_used_packed = NULL;
+	} else {
+		rte_free(vq->shadow_used_split);
+		vq->shadow_used_split = NULL;
+	}
 
 	rte_free(vq->batch_copy_elems);
 	vq->batch_copy_elems = NULL;
diff --git a/lib/librte_vhost/virtio_net.c b/lib/librte_vhost/virtio_net.c
index 35f8cf90a..f0e2e6a1f 100644
--- a/lib/librte_vhost/virtio_net.c
+++ b/lib/librte_vhost/virtio_net.c
@@ -20,6 +20,7 @@ 
 
 #include "iotlb.h"
 #include "vhost.h"
+#include "virtio-packed.h"
 
 #define MAX_PKT_BURST 32
 
@@ -82,7 +83,7 @@  do_flush_shadow_used_ring_split(struct virtio_net *dev,
 			uint16_t to, uint16_t from, uint16_t size)
 {
 	rte_memcpy(&vq->used->ring[to],
-			&vq->shadow_used_ring[from],
+			&vq->shadow_used_split[from],
 			size * sizeof(struct vring_used_elem));
 	vhost_log_cache_used_vring(dev, vq,
 			offsetof(struct vring_used, ring[to]),
@@ -126,8 +127,73 @@  update_shadow_used_ring_split(struct vhost_virtqueue *vq,
 {
 	uint16_t i = vq->shadow_used_idx++;
 
-	vq->shadow_used_ring[i].id  = desc_idx;
-	vq->shadow_used_ring[i].len = len;
+	vq->shadow_used_split[i].id  = desc_idx;
+	vq->shadow_used_split[i].len = len;
+}
+
+static __rte_always_inline void
+flush_shadow_used_ring_packed(struct virtio_net *dev,
+			struct vhost_virtqueue *vq)
+{
+	int i;
+	uint16_t used_idx = vq->last_used_idx;
+
+	/* Split loop in two to save memory barriers */
+	for (i = 0; i < vq->shadow_used_idx; i++) {
+		vq->desc_packed[used_idx].index = vq->shadow_used_packed[i].id;
+		vq->desc_packed[used_idx].len = vq->shadow_used_packed[i].len;
+
+		used_idx += vq->shadow_used_packed[i].count;
+		if (used_idx >= vq->size)
+			used_idx -= vq->size;
+	}
+
+	rte_smp_wmb();
+
+	for (i = 0; i < vq->shadow_used_idx; i++) {
+		uint16_t flags;
+
+		if (vq->shadow_used_packed[i].len)
+			flags = VRING_DESC_F_WRITE;
+		else
+			flags = 0;
+
+		if (vq->used_wrap_counter) {
+			flags |= VRING_DESC_F_USED;
+			flags |= VRING_DESC_F_AVAIL;
+		} else {
+			flags &= ~VRING_DESC_F_USED;
+			flags &= ~VRING_DESC_F_AVAIL;
+		}
+
+		vq->desc_packed[vq->last_used_idx].flags = flags;
+
+		vhost_log_cache_used_vring(dev, vq,
+					vq->last_used_idx *
+					sizeof(struct vring_desc_packed),
+					sizeof(struct vring_desc_packed));
+
+		vq->last_used_idx += vq->shadow_used_packed[i].count;
+		if (vq->last_used_idx >= vq->size) {
+			vq->used_wrap_counter ^= 1;
+			vq->last_used_idx -= vq->size;
+		}
+	}
+
+	rte_smp_wmb();
+	vq->shadow_used_idx = 0;
+	vhost_log_cache_sync(dev, vq);
+}
+
+static __rte_always_inline void
+update_shadow_used_ring_packed(struct vhost_virtqueue *vq,
+			 uint16_t desc_idx, uint16_t len, uint16_t count)
+{
+	uint16_t i = vq->shadow_used_idx++;
+
+	vq->shadow_used_packed[i].id  = desc_idx;
+	vq->shadow_used_packed[i].len = len;
+	vq->shadow_used_packed[i].count = count;
 }
 
 static inline void