[RFC PATCH V3 0/6] vhost: accelerate metadata access

[RFC PATCH V3 0/6] vhost: accelerate metadata access

[Jason Wang]

This series tries to access virtqueue metadata through kernel virtual
address instead of copy_user() friends since they had too much
overheads like checks, spec barriers or even hardware feature
toggling. This is done through setup kernel address through direct
mapping and co-opreate VM management with MMU notifiers.

Test shows about 23% improvement on TX PPS. TCP_STREAM doesn't see
obvious improvement.

Thanks

Changes from RFC V2:
- switch to use direct mapping instead of vmap()
- switch to use spinlock + RCU to synchronize MMU notifier and vhost
  data/control path
- set dirty pages in the invalidation callbacks
- always use copy_to/from_users() friends for the archs that may need
  flush_dcache_pages()
- various minor fixes
Changes from V4:
- use invalidate_range() instead of invalidate_range_start()
- track dirty pages
Changes from V3:
- don't try to use vmap for file backed pages
- rebase to master
Changes from V2:
- fix buggy range overlapping check
- tear down MMU notifier during vhost ioctl to make sure invalidation
  request can read metadata userspace address and vq size without
  holding vq mutex.
Changes from V1:
- instead of pinning pages, use MMU notifier to invalidate vmaps and
  remap duing metadata prefetch
- fix build warning on MIPS

Jason Wang (6):
  vhost: generalize adding used elem
  vhost: fine grain userspace memory accessors
  vhost: rename vq_iotlb_prefetch() to vq_meta_prefetch()
  vhost: introduce helpers to get the size of metadata area
  vhost: factor out setting vring addr and num
  vhost: access vq metadata through kernel virtual address

 drivers/vhost/net.c   |   4 +-
 drivers/vhost/vhost.c | 852 ++++++++++++++++++++++++++++++++++++------
 drivers/vhost/vhost.h |  34 +-
 3 files changed, 764 insertions(+), 126 deletions(-)

-- 
2.18.1

[RFC PATCH V3 1/6] vhost: generalize adding used elem

[Jason Wang]

Use one generic vhost_copy_to_user() instead of two dedicated
accessor. This will simplify the conversion to fine grain
accessors. About 2% improvement of PPS were seen during vitio-user
txonly test.

Signed-off-by: Jason Wang <jasowang@redhat.com>
---
 drivers/vhost/vhost.c | 11 +----------
 1 file changed, 1 insertion(+), 10 deletions(-)

diff --git a/drivers/vhost/vhost.c b/drivers/vhost/vhost.c
index 351af88231ad..6df76ac8200a 100644
--- a/drivers/vhost/vhost.c
+++ b/drivers/vhost/vhost.c
@@ -2255,16 +2255,7 @@ static int __vhost_add_used_n(struct vhost_virtqueue *vq,
 
 	start = vq->last_used_idx & (vq->num - 1);
 	used = vq->used->ring + start;
-	if (count == 1) {
-		if (vhost_put_user(vq, heads[0].id, &used->id)) {
-			vq_err(vq, "Failed to write used id");
-			return -EFAULT;
-		}
-		if (vhost_put_user(vq, heads[0].len, &used->len)) {
-			vq_err(vq, "Failed to write used len");
-			return -EFAULT;
-		}
-	} else if (vhost_copy_to_user(vq, used, heads, count * sizeof *used)) {
+	if (vhost_copy_to_user(vq, used, heads, count * sizeof *used)) {
 		vq_err(vq, "Failed to write used");
 		return -EFAULT;
 	}
-- 
2.18.1

[RFC PATCH V3 2/6] vhost: fine grain userspace memory accessors

[Jason Wang]

This is used to hide the metadata address from virtqueue helpers. This
will allow to implement a vmap based fast accessing to metadata.

Signed-off-by: Jason Wang <jasowang@redhat.com>
---
 drivers/vhost/vhost.c | 94 +++++++++++++++++++++++++++++++++++--------
 1 file changed, 77 insertions(+), 17 deletions(-)

diff --git a/drivers/vhost/vhost.c b/drivers/vhost/vhost.c
index 6df76ac8200a..7335f2275ed3 100644
--- a/drivers/vhost/vhost.c
+++ b/drivers/vhost/vhost.c
@@ -869,6 +869,34 @@ static inline void __user *__vhost_get_user(struct vhost_virtqueue *vq,
 	ret; \
 })
 
+static inline int vhost_put_avail_event(struct vhost_virtqueue *vq)
+{
+	return vhost_put_user(vq, cpu_to_vhost16(vq, vq->avail_idx),
+			      vhost_avail_event(vq));
+}
+
+static inline int vhost_put_used(struct vhost_virtqueue *vq,
+				 struct vring_used_elem *head, int idx,
+				 int count)
+{
+	return vhost_copy_to_user(vq, vq->used->ring + idx, head,
+				  count * sizeof(*head));
+}
+
+static inline int vhost_put_used_flags(struct vhost_virtqueue *vq)
+
+{
+	return vhost_put_user(vq, cpu_to_vhost16(vq, vq->used_flags),
+			      &vq->used->flags);
+}
+
+static inline int vhost_put_used_idx(struct vhost_virtqueue *vq)
+
+{
+	return vhost_put_user(vq, cpu_to_vhost16(vq, vq->last_used_idx),
+			      &vq->used->idx);
+}
+
 #define vhost_get_user(vq, x, ptr, type)		\
 ({ \
 	int ret; \
@@ -907,6 +935,43 @@ static void vhost_dev_unlock_vqs(struct vhost_dev *d)
 		mutex_unlock(&d->vqs[i]->mutex);
 }
 
+static inline int vhost_get_avail_idx(struct vhost_virtqueue *vq,
+				      __virtio16 *idx)
+{
+	return vhost_get_avail(vq, *idx, &vq->avail->idx);
+}
+
+static inline int vhost_get_avail_head(struct vhost_virtqueue *vq,
+				       __virtio16 *head, int idx)
+{
+	return vhost_get_avail(vq, *head,
+			       &vq->avail->ring[idx & (vq->num - 1)]);
+}
+
+static inline int vhost_get_avail_flags(struct vhost_virtqueue *vq,
+					__virtio16 *flags)
+{
+	return vhost_get_avail(vq, *flags, &vq->avail->flags);
+}
+
+static inline int vhost_get_used_event(struct vhost_virtqueue *vq,
+				       __virtio16 *event)
+{
+	return vhost_get_avail(vq, *event, vhost_used_event(vq));
+}
+
+static inline int vhost_get_used_idx(struct vhost_virtqueue *vq,
+				     __virtio16 *idx)
+{
+	return vhost_get_used(vq, *idx, &vq->used->idx);
+}
+
+static inline int vhost_get_desc(struct vhost_virtqueue *vq,
+				 struct vring_desc *desc, int idx)
+{
+	return vhost_copy_from_user(vq, desc, vq->desc + idx, sizeof(*desc));
+}
+
 static int vhost_new_umem_range(struct vhost_umem *umem,
 				u64 start, u64 size, u64 end,
 				u64 userspace_addr, int perm)
@@ -1844,8 +1909,7 @@ EXPORT_SYMBOL_GPL(vhost_log_write);
 static int vhost_update_used_flags(struct vhost_virtqueue *vq)
 {
 	void __user *used;
-	if (vhost_put_user(vq, cpu_to_vhost16(vq, vq->used_flags),
-			   &vq->used->flags) < 0)
+	if (vhost_put_used_flags(vq))
 		return -EFAULT;
 	if (unlikely(vq->log_used)) {
 		/* Make sure the flag is seen before log. */
@@ -1862,8 +1926,7 @@ static int vhost_update_used_flags(struct vhost_virtqueue *vq)
 
 static int vhost_update_avail_event(struct vhost_virtqueue *vq, u16 avail_event)
 {
-	if (vhost_put_user(vq, cpu_to_vhost16(vq, vq->avail_idx),
-			   vhost_avail_event(vq)))
+	if (vhost_put_avail_event(vq))
 		return -EFAULT;
 	if (unlikely(vq->log_used)) {
 		void __user *used;
@@ -1899,7 +1962,7 @@ int vhost_vq_init_access(struct vhost_virtqueue *vq)
 		r = -EFAULT;
 		goto err;
 	}
-	r = vhost_get_used(vq, last_used_idx, &vq->used->idx);
+	r = vhost_get_used_idx(vq, &last_used_idx);
 	if (r) {
 		vq_err(vq, "Can't access used idx at %p\n",
 		       &vq->used->idx);
@@ -2098,7 +2161,7 @@ int vhost_get_vq_desc(struct vhost_virtqueue *vq,
 	last_avail_idx = vq->last_avail_idx;
 
 	if (vq->avail_idx == vq->last_avail_idx) {
-		if (unlikely(vhost_get_avail(vq, avail_idx, &vq->avail->idx))) {
+		if (unlikely(vhost_get_avail_idx(vq, &avail_idx))) {
 			vq_err(vq, "Failed to access avail idx at %p\n",
 				&vq->avail->idx);
 			return -EFAULT;
@@ -2125,8 +2188,7 @@ 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)]))) {
+	if (unlikely(vhost_get_avail_head(vq, &ring_head, last_avail_idx))) {
 		vq_err(vq, "Failed to read head: idx %d address %p\n",
 		       last_avail_idx,
 		       &vq->avail->ring[last_avail_idx % vq->num]);
@@ -2161,8 +2223,7 @@ int vhost_get_vq_desc(struct vhost_virtqueue *vq,
 			       i, vq->num, head);
 			return -EINVAL;
 		}
-		ret = vhost_copy_from_user(vq, &desc, vq->desc + i,
-					   sizeof desc);
+		ret = vhost_get_desc(vq, &desc, i);
 		if (unlikely(ret)) {
 			vq_err(vq, "Failed to get descriptor: idx %d addr %p\n",
 			       i, vq->desc + i);
@@ -2255,7 +2316,7 @@ static int __vhost_add_used_n(struct vhost_virtqueue *vq,
 
 	start = vq->last_used_idx & (vq->num - 1);
 	used = vq->used->ring + start;
-	if (vhost_copy_to_user(vq, used, heads, count * sizeof *used)) {
+	if (vhost_put_used(vq, heads, start, count)) {
 		vq_err(vq, "Failed to write used");
 		return -EFAULT;
 	}
@@ -2297,8 +2358,7 @@ int vhost_add_used_n(struct vhost_virtqueue *vq, struct vring_used_elem *heads,
 
 	/* Make sure buffer is written before we update index. */
 	smp_wmb();
-	if (vhost_put_user(vq, cpu_to_vhost16(vq, vq->last_used_idx),
-			   &vq->used->idx)) {
+	if (vhost_put_used_idx(vq)) {
 		vq_err(vq, "Failed to increment used idx");
 		return -EFAULT;
 	}
@@ -2331,7 +2391,7 @@ static bool vhost_notify(struct vhost_dev *dev, struct vhost_virtqueue *vq)
 
 	if (!vhost_has_feature(vq, VIRTIO_RING_F_EVENT_IDX)) {
 		__virtio16 flags;
-		if (vhost_get_avail(vq, flags, &vq->avail->flags)) {
+		if (vhost_get_avail_flags(vq, &flags)) {
 			vq_err(vq, "Failed to get flags");
 			return true;
 		}
@@ -2345,7 +2405,7 @@ static bool vhost_notify(struct vhost_dev *dev, struct vhost_virtqueue *vq)
 	if (unlikely(!v))
 		return true;
 
-	if (vhost_get_avail(vq, event, vhost_used_event(vq))) {
+	if (vhost_get_used_event(vq, &event)) {
 		vq_err(vq, "Failed to get used event idx");
 		return true;
 	}
@@ -2390,7 +2450,7 @@ bool vhost_vq_avail_empty(struct vhost_dev *dev, struct vhost_virtqueue *vq)
 	if (vq->avail_idx != vq->last_avail_idx)
 		return false;
 
-	r = vhost_get_avail(vq, avail_idx, &vq->avail->idx);
+	r = vhost_get_avail_idx(vq, &avail_idx);
 	if (unlikely(r))
 		return false;
 	vq->avail_idx = vhost16_to_cpu(vq, avail_idx);
@@ -2426,7 +2486,7 @@ bool vhost_enable_notify(struct vhost_dev *dev, struct vhost_virtqueue *vq)
 	/* They could have slipped one in as we were doing that: make
 	 * sure it's written, then check again. */
 	smp_mb();
-	r = vhost_get_avail(vq, avail_idx, &vq->avail->idx);
+	r = vhost_get_avail_idx(vq, &avail_idx);
 	if (r) {
 		vq_err(vq, "Failed to check avail idx at %p: %d\n",
 		       &vq->avail->idx, r);
-- 
2.18.1

[RFC PATCH V3 3/6] vhost: rename vq_iotlb_prefetch() to vq_meta_prefetch()

[Jason Wang]

Rename the function to be more accurate since it actually tries to
prefetch vq metadata address in IOTLB. And this will be used by
following patch to prefetch metadata virtual addresses.

Signed-off-by: Jason Wang <jasowang@redhat.com>
---
 drivers/vhost/net.c   | 4 ++--
 drivers/vhost/vhost.c | 4 ++--
 drivers/vhost/vhost.h | 2 +-
 3 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/drivers/vhost/net.c b/drivers/vhost/net.c
index df51a35cf537..bf55f995ebae 100644
--- a/drivers/vhost/net.c
+++ b/drivers/vhost/net.c
@@ -971,7 +971,7 @@ static void handle_tx(struct vhost_net *net)
 	if (!sock)
 		goto out;
 
-	if (!vq_iotlb_prefetch(vq))
+	if (!vq_meta_prefetch(vq))
 		goto out;
 
 	vhost_disable_notify(&net->dev, vq);
@@ -1140,7 +1140,7 @@ static void handle_rx(struct vhost_net *net)
 	if (!sock)
 		goto out;
 
-	if (!vq_iotlb_prefetch(vq))
+	if (!vq_meta_prefetch(vq))
 		goto out;
 
 	vhost_disable_notify(&net->dev, vq);
diff --git a/drivers/vhost/vhost.c b/drivers/vhost/vhost.c
index 7335f2275ed3..bff4d586871d 100644
--- a/drivers/vhost/vhost.c
+++ b/drivers/vhost/vhost.c
@@ -1313,7 +1313,7 @@ static bool iotlb_access_ok(struct vhost_virtqueue *vq,
 	return true;
 }
 
-int vq_iotlb_prefetch(struct vhost_virtqueue *vq)
+int vq_meta_prefetch(struct vhost_virtqueue *vq)
 {
 	size_t s = vhost_has_feature(vq, VIRTIO_RING_F_EVENT_IDX) ? 2 : 0;
 	unsigned int num = vq->num;
@@ -1332,7 +1332,7 @@ int vq_iotlb_prefetch(struct vhost_virtqueue *vq)
 			       num * sizeof(*vq->used->ring) + s,
 			       VHOST_ADDR_USED);
 }
-EXPORT_SYMBOL_GPL(vq_iotlb_prefetch);
+EXPORT_SYMBOL_GPL(vq_meta_prefetch);
 
 /* Can we log writes? */
 /* Caller should have device mutex but not vq mutex */
diff --git a/drivers/vhost/vhost.h b/drivers/vhost/vhost.h
index 9490e7ddb340..7a7fc001265f 100644
--- a/drivers/vhost/vhost.h
+++ b/drivers/vhost/vhost.h
@@ -209,7 +209,7 @@ bool vhost_enable_notify(struct vhost_dev *, struct vhost_virtqueue *);
 int vhost_log_write(struct vhost_virtqueue *vq, struct vhost_log *log,
 		    unsigned int log_num, u64 len,
 		    struct iovec *iov, int count);
-int vq_iotlb_prefetch(struct vhost_virtqueue *vq);
+int vq_meta_prefetch(struct vhost_virtqueue *vq);
 
 struct vhost_msg_node *vhost_new_msg(struct vhost_virtqueue *vq, int type);
 void vhost_enqueue_msg(struct vhost_dev *dev,
-- 
2.18.1

[RFC PATCH V3 4/6] vhost: introduce helpers to get the size of metadata area

[Jason Wang]

To avoid code duplication since it will be used by kernel VA prefetching.

Signed-off-by: Jason Wang <jasowang@redhat.com>
---
 drivers/vhost/vhost.c | 51 ++++++++++++++++++++++++++++---------------
 1 file changed, 33 insertions(+), 18 deletions(-)

diff --git a/drivers/vhost/vhost.c b/drivers/vhost/vhost.c
index bff4d586871d..f3f86c3ed659 100644
--- a/drivers/vhost/vhost.c
+++ b/drivers/vhost/vhost.c
@@ -413,6 +413,32 @@ static void vhost_dev_free_iovecs(struct vhost_dev *dev)
 		vhost_vq_free_iovecs(dev->vqs[i]);
 }
 
+static size_t vhost_get_avail_size(struct vhost_virtqueue *vq,
+				   unsigned int num)
+{
+	size_t event __maybe_unused =
+	       vhost_has_feature(vq, VIRTIO_RING_F_EVENT_IDX) ? 2 : 0;
+
+	return sizeof(*vq->avail) +
+	       sizeof(*vq->avail->ring) * num + event;
+}
+
+static size_t vhost_get_used_size(struct vhost_virtqueue *vq,
+				  unsigned int num)
+{
+	size_t event __maybe_unused =
+	       vhost_has_feature(vq, VIRTIO_RING_F_EVENT_IDX) ? 2 : 0;
+
+	return sizeof(*vq->used) +
+	       sizeof(*vq->used->ring) * num + event;
+}
+
+static size_t vhost_get_desc_size(struct vhost_virtqueue *vq,
+				  unsigned int num)
+{
+	return sizeof(*vq->desc) * num;
+}
+
 void vhost_dev_init(struct vhost_dev *dev,
 		    struct vhost_virtqueue **vqs, int nvqs, int iov_limit)
 {
@@ -1257,13 +1283,9 @@ static bool vq_access_ok(struct vhost_virtqueue *vq, unsigned int num,
 			 struct vring_used __user *used)
 
 {
-	size_t s __maybe_unused = vhost_has_feature(vq, VIRTIO_RING_F_EVENT_IDX) ? 2 : 0;
-
-	return access_ok(desc, num * sizeof *desc) &&
-	       access_ok(avail,
-			 sizeof *avail + num * sizeof *avail->ring + s) &&
-	       access_ok(used,
-			sizeof *used + num * sizeof *used->ring + s);
+	return access_ok(desc, vhost_get_desc_size(vq, num)) &&
+	       access_ok(avail, vhost_get_avail_size(vq, num)) &&
+	       access_ok(used, vhost_get_used_size(vq, num));
 }
 
 static void vhost_vq_meta_update(struct vhost_virtqueue *vq,
@@ -1315,22 +1337,18 @@ static bool iotlb_access_ok(struct vhost_virtqueue *vq,
 
 int vq_meta_prefetch(struct vhost_virtqueue *vq)
 {
-	size_t s = vhost_has_feature(vq, VIRTIO_RING_F_EVENT_IDX) ? 2 : 0;
 	unsigned int num = vq->num;
 
 	if (!vq->iotlb)
 		return 1;
 
 	return iotlb_access_ok(vq, VHOST_ACCESS_RO, (u64)(uintptr_t)vq->desc,
-			       num * sizeof(*vq->desc), VHOST_ADDR_DESC) &&
+			       vhost_get_desc_size(vq, num), VHOST_ADDR_DESC) &&
 	       iotlb_access_ok(vq, VHOST_ACCESS_RO, (u64)(uintptr_t)vq->avail,
-			       sizeof *vq->avail +
-			       num * sizeof(*vq->avail->ring) + s,
+			       vhost_get_avail_size(vq, num),
 			       VHOST_ADDR_AVAIL) &&
 	       iotlb_access_ok(vq, VHOST_ACCESS_WO, (u64)(uintptr_t)vq->used,
-			       sizeof *vq->used +
-			       num * sizeof(*vq->used->ring) + s,
-			       VHOST_ADDR_USED);
+			       vhost_get_used_size(vq, num), VHOST_ADDR_USED);
 }
 EXPORT_SYMBOL_GPL(vq_meta_prefetch);
 
@@ -1347,13 +1365,10 @@ EXPORT_SYMBOL_GPL(vhost_log_access_ok);
 static bool vq_log_access_ok(struct vhost_virtqueue *vq,
 			     void __user *log_base)
 {
-	size_t s = vhost_has_feature(vq, VIRTIO_RING_F_EVENT_IDX) ? 2 : 0;
-
 	return vq_memory_access_ok(log_base, vq->umem,
 				   vhost_has_feature(vq, VHOST_F_LOG_ALL)) &&
 		(!vq->log_used || log_access_ok(log_base, vq->log_addr,
-					sizeof *vq->used +
-					vq->num * sizeof *vq->used->ring + s));
+				  vhost_get_used_size(vq, vq->num)));
 }
 
 /* Can we start vq? */
-- 
2.18.1

[RFC PATCH V3 5/6] vhost: factor out setting vring addr and num

[Jason Wang]

Factoring vring address and num setting which needs special care for
accelerating vq metadata accessing.

Signed-off-by: Jason Wang <jasowang@redhat.com>
---
 drivers/vhost/vhost.c | 177 ++++++++++++++++++++++++------------------
 1 file changed, 103 insertions(+), 74 deletions(-)

diff --git a/drivers/vhost/vhost.c b/drivers/vhost/vhost.c
index f3f86c3ed659..c2362ed5839e 100644
--- a/drivers/vhost/vhost.c
+++ b/drivers/vhost/vhost.c
@@ -1468,6 +1468,104 @@ static long vhost_set_memory(struct vhost_dev *d, struct vhost_memory __user *m)
 	return -EFAULT;
 }
 
+static long vhost_vring_set_num(struct vhost_dev *d,
+				struct vhost_virtqueue *vq,
+				void __user *argp)
+{
+	struct vhost_vring_state s;
+
+	/* Resizing ring with an active backend?
+	 * You don't want to do that. */
+	if (vq->private_data)
+		return -EBUSY;
+
+	if (copy_from_user(&s, argp, sizeof s))
+		return -EFAULT;
+
+	if (!s.num || s.num > 0xffff || (s.num & (s.num - 1)))
+		return -EINVAL;
+	vq->num = s.num;
+
+	return 0;
+}
+
+static long vhost_vring_set_addr(struct vhost_dev *d,
+				 struct vhost_virtqueue *vq,
+				 void __user *argp)
+{
+	struct vhost_vring_addr a;
+
+	if (copy_from_user(&a, argp, sizeof a))
+		return -EFAULT;
+	if (a.flags & ~(0x1 << VHOST_VRING_F_LOG))
+		return -EOPNOTSUPP;
+
+	/* For 32bit, verify that the top 32bits of the user
+	   data are set to zero. */
+	if ((u64)(unsigned long)a.desc_user_addr != a.desc_user_addr ||
+	    (u64)(unsigned long)a.used_user_addr != a.used_user_addr ||
+	    (u64)(unsigned long)a.avail_user_addr != a.avail_user_addr)
+		return -EFAULT;
+
+	/* Make sure it's safe to cast pointers to vring types. */
+	BUILD_BUG_ON(__alignof__ *vq->avail > VRING_AVAIL_ALIGN_SIZE);
+	BUILD_BUG_ON(__alignof__ *vq->used > VRING_USED_ALIGN_SIZE);
+	if ((a.avail_user_addr & (VRING_AVAIL_ALIGN_SIZE - 1)) ||
+	    (a.used_user_addr & (VRING_USED_ALIGN_SIZE - 1)) ||
+	    (a.log_guest_addr & (VRING_USED_ALIGN_SIZE - 1)))
+		return -EINVAL;
+
+	/* We only verify access here if backend is configured.
+	 * If it is not, we don't as size might not have been setup.
+	 * We will verify when backend is configured. */
+	if (vq->private_data) {
+		if (!vq_access_ok(vq, vq->num,
+			(void __user *)(unsigned long)a.desc_user_addr,
+			(void __user *)(unsigned long)a.avail_user_addr,
+			(void __user *)(unsigned long)a.used_user_addr))
+			return -EINVAL;
+
+		/* Also validate log access for used ring if enabled. */
+		if ((a.flags & (0x1 << VHOST_VRING_F_LOG)) &&
+			!log_access_ok(vq->log_base, a.log_guest_addr,
+				sizeof *vq->used +
+				vq->num * sizeof *vq->used->ring))
+			return -EINVAL;
+	}
+
+	vq->log_used = !!(a.flags & (0x1 << VHOST_VRING_F_LOG));
+	vq->desc = (void __user *)(unsigned long)a.desc_user_addr;
+	vq->avail = (void __user *)(unsigned long)a.avail_user_addr;
+	vq->log_addr = a.log_guest_addr;
+	vq->used = (void __user *)(unsigned long)a.used_user_addr;
+
+	return 0;
+}
+
+static long vhost_vring_set_num_addr(struct vhost_dev *d,
+				     struct vhost_virtqueue *vq,
+				     unsigned int ioctl,
+				     void __user *argp)
+{
+	long r;
+
+	mutex_lock(&vq->mutex);
+
+	switch (ioctl) {
+	case VHOST_SET_VRING_NUM:
+		r = vhost_vring_set_num(d, vq, argp);
+		break;
+	case VHOST_SET_VRING_ADDR:
+		r = vhost_vring_set_addr(d, vq, argp);
+		break;
+	default:
+		BUG();
+	}
+
+	mutex_unlock(&vq->mutex);
+
+	return r;
+}
 long vhost_vring_ioctl(struct vhost_dev *d, unsigned int ioctl, void __user *argp)
 {
 	struct file *eventfp, *filep = NULL;
@@ -1477,7 +1575,6 @@ long vhost_vring_ioctl(struct vhost_dev *d, unsigned int ioctl, void __user *arg
 	struct vhost_virtqueue *vq;
 	struct vhost_vring_state s;
 	struct vhost_vring_file f;
-	struct vhost_vring_addr a;
 	u32 idx;
 	long r;
 
@@ -1490,26 +1587,14 @@ long vhost_vring_ioctl(struct vhost_dev *d, unsigned int ioctl, void __user *arg
 	idx = array_index_nospec(idx, d->nvqs);
 	vq = d->vqs[idx];
 
+	if (ioctl == VHOST_SET_VRING_NUM ||
+	    ioctl == VHOST_SET_VRING_ADDR) {
+		return vhost_vring_set_num_addr(d, vq, ioctl, argp);
+	}
+
 	mutex_lock(&vq->mutex);
 
 	switch (ioctl) {
-	case VHOST_SET_VRING_NUM:
-		/* Resizing ring with an active backend?
-		 * You don't want to do that. */
-		if (vq->private_data) {
-			r = -EBUSY;
-			break;
-		}
-		if (copy_from_user(&s, argp, sizeof s)) {
-			r = -EFAULT;
-			break;
-		}
-		if (!s.num || s.num > 0xffff || (s.num & (s.num - 1))) {
-			r = -EINVAL;
-			break;
-		}
-		vq->num = s.num;
-		break;
 	case VHOST_SET_VRING_BASE:
 		/* Moving base with an active backend?
 		 * You don't want to do that. */
@@ -1535,62 +1620,6 @@ long vhost_vring_ioctl(struct vhost_dev *d, unsigned int ioctl, void __user *arg
 		if (copy_to_user(argp, &s, sizeof s))
 			r = -EFAULT;
 		break;
-	case VHOST_SET_VRING_ADDR:
-		if (copy_from_user(&a, argp, sizeof a)) {
-			r = -EFAULT;
-			break;
-		}
-		if (a.flags & ~(0x1 << VHOST_VRING_F_LOG)) {
-			r = -EOPNOTSUPP;
-			break;
-		}
-		/* For 32bit, verify that the top 32bits of the user
-		   data are set to zero. */
-		if ((u64)(unsigned long)a.desc_user_addr != a.desc_user_addr ||
-		    (u64)(unsigned long)a.used_user_addr != a.used_user_addr ||
-		    (u64)(unsigned long)a.avail_user_addr != a.avail_user_addr) {
-			r = -EFAULT;
-			break;
-		}
-
-		/* Make sure it's safe to cast pointers to vring types. */
-		BUILD_BUG_ON(__alignof__ *vq->avail > VRING_AVAIL_ALIGN_SIZE);
-		BUILD_BUG_ON(__alignof__ *vq->used > VRING_USED_ALIGN_SIZE);
-		if ((a.avail_user_addr & (VRING_AVAIL_ALIGN_SIZE - 1)) ||
-		    (a.used_user_addr & (VRING_USED_ALIGN_SIZE - 1)) ||
-		    (a.log_guest_addr & (VRING_USED_ALIGN_SIZE - 1))) {
-			r = -EINVAL;
-			break;
-		}
-
-		/* We only verify access here if backend is configured.
-		 * If it is not, we don't as size might not have been setup.
-		 * We will verify when backend is configured. */
-		if (vq->private_data) {
-			if (!vq_access_ok(vq, vq->num,
-				(void __user *)(unsigned long)a.desc_user_addr,
-				(void __user *)(unsigned long)a.avail_user_addr,
-				(void __user *)(unsigned long)a.used_user_addr)) {
-				r = -EINVAL;
-				break;
-			}
-
-			/* Also validate log access for used ring if enabled. */
-			if ((a.flags & (0x1 << VHOST_VRING_F_LOG)) &&
-			    !log_access_ok(vq->log_base, a.log_guest_addr,
-					   sizeof *vq->used +
-					   vq->num * sizeof *vq->used->ring)) {
-				r = -EINVAL;
-				break;
-			}
-		}
-
-		vq->log_used = !!(a.flags & (0x1 << VHOST_VRING_F_LOG));
-		vq->desc = (void __user *)(unsigned long)a.desc_user_addr;
-		vq->avail = (void __user *)(unsigned long)a.avail_user_addr;
-		vq->log_addr = a.log_guest_addr;
-		vq->used = (void __user *)(unsigned long)a.used_user_addr;
-		break;
 	case VHOST_SET_VRING_KICK:
 		if (copy_from_user(&f, argp, sizeof f)) {
 			r = -EFAULT;
-- 
2.18.1

[RFC PATCH V3 6/6] vhost: access vq metadata through kernel virtual address

[Jason Wang]

It was noticed that the copy_to/from_user() friends that was used to
access virtqueue metdata tends to be very expensive for dataplane
implementation like vhost since it involves lots of software checks,
speculation barriers, hardware feature toggling (e.g SMAP). The
extra cost will be more obvious when transferring small packets since
the time spent on metadata accessing become more significant.

This patch tries to eliminate those overheads by accessing them
through direct mapping of those pages. Invalidation callbacks is
implemented for co-operation with general VM management (swap, KSM,
THP or NUMA balancing). We will try to get the direct mapping of vq
metadata before each round of packet processing if it doesn't
exist. If we fail, we will simplely fallback to copy_to/from_user()
friends.

This invalidation and direct mapping access are synchronized through
spinlock and RCU. All matedata accessing through direct map is
protected by RCU, and the setup or invalidation are done under
spinlock.

This method might does not work for high mem page which requires
temporary mapping so we just fallback to normal
copy_to/from_user() and may not for arch that has virtual tagged cache
since extra cache flushing is needed to eliminate the alias. This will
result complex logic and bad performance. For those archs, this patch
simply go for copy_to/from_user() friends. This is done by ruling out
kernel mapping codes through ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE.

Note that this is only done when device IOTLB is not enabled. We
could use similar method to optimize IOTLB in the future.

Tests shows at most about 23% improvement on TX PPS when using
virtio-user + vhost_net + xdp1 + TAP on 2.6GHz Broadwell:

        SMAP on | SMAP off
Before: 5.2Mpps | 7.1Mpps
After:  6.4Mpps | 8.2Mpps

Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: James Bottomley <James.Bottomley@hansenpartnership.com>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: David Miller <davem@davemloft.net>
Cc: Jerome Glisse <jglisse@redhat.com>
Cc: linux-mm@kvack.org
Cc: linux-arm-kernel@lists.infradead.org
Cc: linux-parisc@vger.kernel.org
Signed-off-by: Jason Wang <jasowang@redhat.com>
---
 drivers/vhost/vhost.c | 517 +++++++++++++++++++++++++++++++++++++++++-
 drivers/vhost/vhost.h |  32 +++
 2 files changed, 546 insertions(+), 3 deletions(-)

diff --git a/drivers/vhost/vhost.c b/drivers/vhost/vhost.c
index c2362ed5839e..a9fe04f0016a 100644
--- a/drivers/vhost/vhost.c
+++ b/drivers/vhost/vhost.c
@@ -299,6 +299,49 @@ static void vhost_vq_meta_reset(struct vhost_dev *d)
 		__vhost_vq_meta_reset(d->vqs[i]);
 }
 
+#if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE == 0
+static void vhost_map_unprefetch(struct vhost_map *map)
+{
+	kfree(map->pages);
+	map->pages = NULL;
+	map->npages = 0;
+	map->addr = NULL;
+}
+
+static void vhost_uninit_vq_maps(struct vhost_virtqueue *vq)
+{
+	struct vhost_map *map[VHOST_NUM_ADDRS];
+	int i;
+
+	spin_lock(&vq->mmu_lock);
+	for (i = 0; i < VHOST_NUM_ADDRS; i++) {
+		map[i] = rcu_dereference_protected(vq->maps[i],
+				  lockdep_is_held(&vq->mmu_lock));
+		if (map[i])
+			rcu_assign_pointer(vq->maps[i], NULL);
+	}
+	spin_unlock(&vq->mmu_lock);
+
+	synchronize_rcu();
+
+	for (i = 0; i < VHOST_NUM_ADDRS; i++)
+		if (map[i])
+			vhost_map_unprefetch(map[i]);
+
+}
+#endif
+
+static void vhost_reset_vq_maps(struct vhost_virtqueue *vq)
+{
+#if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE == 0
+	int i;
+
+	vhost_uninit_vq_maps(vq);
+	for (i = 0; i < VHOST_NUM_ADDRS; i++)
+		vq->uaddrs[i].size = 0;
+#endif
+}
+
 static void vhost_vq_reset(struct vhost_dev *dev,
 			   struct vhost_virtqueue *vq)
 {
@@ -327,7 +370,9 @@ static void vhost_vq_reset(struct vhost_dev *dev,
 	vq->busyloop_timeout = 0;
 	vq->umem = NULL;
 	vq->iotlb = NULL;
+	vq->invalidate_count = 0;
 	__vhost_vq_meta_reset(vq);
+	vhost_reset_vq_maps(vq);
 }
 
 static int vhost_worker(void *data)
@@ -439,6 +484,123 @@ static size_t vhost_get_desc_size(struct vhost_virtqueue *vq,
 	return sizeof(*vq->desc) * num;
 }
 
+#if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE == 0
+static bool vhost_map_range_overlap(struct vhost_uaddr *uaddr,
+				     unsigned long start,
+				     unsigned long end)
+{
+	if (unlikely(!uaddr->size))
+		return false;
+
+	return !(end < uaddr->uaddr || start > uaddr->uaddr - 1 + uaddr->size);
+}
+
+static void vhost_invalidate_vq_start(struct vhost_virtqueue *vq,
+				      int index,
+				      unsigned long start,
+				      unsigned long end)
+{
+	struct vhost_uaddr *uaddr = &vq->uaddrs[index];
+	struct vhost_map *map;
+	int i;
+
+	if (!vhost_map_range_overlap(uaddr, start, end))
+		return;
+
+	spin_lock(&vq->mmu_lock);
+	++vq->invalidate_count;
+
+	map = rcu_dereference_protected(vq->maps[index],
+					lockdep_is_held(&vq->mmu_lock));
+	if (map) {
+		if (uaddr->write) {
+			for (i = 0; i < map->npages; i++)
+				set_page_dirty(map->pages[i]);
+		}
+		rcu_assign_pointer(vq->maps[index], NULL);
+	}
+	spin_unlock(&vq->mmu_lock);
+
+	if (map) {
+		synchronize_rcu();
+		vhost_map_unprefetch(map);
+	}
+}
+
+static void vhost_invalidate_vq_end(struct vhost_virtqueue *vq,
+				    int index,
+				    unsigned long start,
+				    unsigned long end)
+{
+	if (!vhost_map_range_overlap(&vq->uaddrs[index], start, end))
+		return;
+
+	spin_lock(&vq->mmu_lock);
+	--vq->invalidate_count;
+	spin_unlock(&vq->mmu_lock);
+}
+
+static int vhost_invalidate_range_start(struct mmu_notifier *mn,
+					const struct mmu_notifier_range *range)
+{
+	struct vhost_dev *dev = container_of(mn, struct vhost_dev,
+					     mmu_notifier);
+	int i, j;
+
+	if (!range->blockable)
+		return -EAGAIN;
+
+	for (i = 0; i < dev->nvqs; i++) {
+		struct vhost_virtqueue *vq = dev->vqs[i];
+
+		for (j = 0; j < VHOST_NUM_ADDRS; j++)
+			vhost_invalidate_vq_start(vq, j,
+						  range->start,
+						  range->end);
+	}
+
+	return 0;
+}
+
+static void vhost_invalidate_range_end(struct mmu_notifier *mn,
+				       const struct mmu_notifier_range *range)
+{
+	struct vhost_dev *dev = container_of(mn, struct vhost_dev,
+					     mmu_notifier);
+	int i, j;
+
+	for (i = 0; i < dev->nvqs; i++) {
+		struct vhost_virtqueue *vq = dev->vqs[i];
+
+		for (j = 0; j < VHOST_NUM_ADDRS; j++)
+			vhost_invalidate_vq_end(vq, j,
+						range->start,
+						range->end);
+	}
+}
+
+static const struct mmu_notifier_ops vhost_mmu_notifier_ops = {
+	.invalidate_range_start = vhost_invalidate_range_start,
+	.invalidate_range_end = vhost_invalidate_range_end,
+};
+#endif
+
+static void vhost_init_maps(struct vhost_dev *dev)
+{
+#if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE == 0
+	struct vhost_virtqueue *vq;
+	int i, j;
+
+	dev->mmu_notifier.ops = &vhost_mmu_notifier_ops;
+
+	for (i = 0; i < dev->nvqs; ++i) {
+		vq = dev->vqs[i];
+		for (j = 0; j < VHOST_NUM_ADDRS; j++)
+			RCU_INIT_POINTER(vq->maps[j], NULL);
+	}
+#endif
+}
+
 void vhost_dev_init(struct vhost_dev *dev,
 		    struct vhost_virtqueue **vqs, int nvqs, int iov_limit)
 {
@@ -459,7 +621,7 @@ void vhost_dev_init(struct vhost_dev *dev,
 	INIT_LIST_HEAD(&dev->read_list);
 	INIT_LIST_HEAD(&dev->pending_list);
 	spin_lock_init(&dev->iotlb_lock);
-
+	vhost_init_maps(dev);
 
 	for (i = 0; i < dev->nvqs; ++i) {
 		vq = dev->vqs[i];
@@ -468,6 +630,7 @@ void vhost_dev_init(struct vhost_dev *dev,
 		vq->heads = NULL;
 		vq->dev = dev;
 		mutex_init(&vq->mutex);
+		spin_lock_init(&vq->mmu_lock);
 		vhost_vq_reset(dev, vq);
 		if (vq->handle_kick)
 			vhost_poll_init(&vq->poll, vq->handle_kick,
@@ -547,7 +710,18 @@ long vhost_dev_set_owner(struct vhost_dev *dev)
 	if (err)
 		goto err_cgroup;
 
+#if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE == 0
+	err = mmu_notifier_register(&dev->mmu_notifier, dev->mm);
+	if (err)
+		goto err_mmu_notifier;
+#endif
+
 	return 0;
+
+#if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE == 0
+err_mmu_notifier:
+	vhost_dev_free_iovecs(dev);
+#endif
 err_cgroup:
 	kthread_stop(worker);
 	dev->worker = NULL;
@@ -638,6 +812,107 @@ static void vhost_clear_msg(struct vhost_dev *dev)
 	spin_unlock(&dev->iotlb_lock);
 }
 
+#if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE == 0
+static void vhost_setup_uaddr(struct vhost_virtqueue *vq,
+			      int index, unsigned long uaddr,
+			      size_t size, bool write)
+{
+	struct vhost_uaddr *addr = &vq->uaddrs[index];
+
+	addr->uaddr = uaddr;
+	addr->size = size;
+	addr->write = write;
+}
+
+static void vhost_setup_vq_uaddr(struct vhost_virtqueue *vq)
+{
+	vhost_setup_uaddr(vq, VHOST_ADDR_DESC,
+			  (unsigned long)vq->desc,
+			  vhost_get_desc_size(vq, vq->num),
+			  false);
+	vhost_setup_uaddr(vq, VHOST_ADDR_AVAIL,
+			  (unsigned long)vq->avail,
+			  vhost_get_avail_size(vq, vq->num),
+			  false);
+	vhost_setup_uaddr(vq, VHOST_ADDR_USED,
+			  (unsigned long)vq->used,
+			  vhost_get_used_size(vq, vq->num),
+			  true);
+}
+
+static int vhost_map_prefetch(struct vhost_virtqueue *vq,
+			       int index)
+{
+	struct vhost_map *map;
+	struct vhost_uaddr *uaddr = &vq->uaddrs[index];
+	struct page **pages;
+	int npages = DIV_ROUND_UP(uaddr->size, PAGE_SIZE);
+	int npinned;
+	void *vaddr, *v;
+	int err;
+	int i;
+
+	spin_lock(&vq->mmu_lock);
+
+	err = -EFAULT;
+	if (vq->invalidate_count)
+		goto err;
+
+	err = -ENOMEM;
+	map = kmalloc(sizeof(*map), GFP_ATOMIC);
+	if (!map)
+		goto err;
+
+	pages = kmalloc_array(npages, sizeof(struct page *), GFP_ATOMIC);
+	if (!pages)
+		goto err_pages;
+
+	err = EFAULT;
+	npinned = __get_user_pages_fast(uaddr->uaddr, npages,
+					uaddr->write, pages);
+	if (npinned > 0)
+		release_pages(pages, npinned);
+	if (npinned != npages)
+		goto err_gup;
+
+	for (i = 0; i < npinned; i++)
+		if (PageHighMem(pages[i]))
+			goto err_gup;
+
+	vaddr = v = page_address(pages[0]);
+
+	/* For simplicity, fallback to userspace address if VA is not
+	 * contigious.
+	 */
+	for (i = 1; i < npinned; i++) {
+		v += PAGE_SIZE;
+		if (v != page_address(pages[i]))
+			goto err_gup;
+	}
+
+	map->addr = vaddr + (uaddr->uaddr & (PAGE_SIZE - 1));
+	map->npages = npages;
+	map->pages = pages;
+
+	rcu_assign_pointer(vq->maps[index], map);
+	/* No need for a synchronize_rcu(). This function should be
+	 * called by dev->worker so we are serialized with all
+	 * readers.
+	 */
+	spin_unlock(&vq->mmu_lock);
+
+	return 0;
+
+err_gup:
+	kfree(pages);
+err_pages:
+	kfree(map);
+err:
+	spin_unlock(&vq->mmu_lock);
+	return err;
+}
+#endif
+
 void vhost_dev_cleanup(struct vhost_dev *dev)
 {
 	int i;
@@ -667,8 +942,16 @@ void vhost_dev_cleanup(struct vhost_dev *dev)
 		kthread_stop(dev->worker);
 		dev->worker = NULL;
 	}
-	if (dev->mm)
+	if (dev->mm) {
+#if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE == 0
+		mmu_notifier_unregister(&dev->mmu_notifier, dev->mm);
+#endif
 		mmput(dev->mm);
+	}
+#if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE == 0
+	for (i = 0; i < dev->nvqs; i++)
+		vhost_uninit_vq_maps(dev->vqs[i]);
+#endif
 	dev->mm = NULL;
 }
 EXPORT_SYMBOL_GPL(vhost_dev_cleanup);
@@ -897,6 +1180,26 @@ static inline void __user *__vhost_get_user(struct vhost_virtqueue *vq,
 
 static inline int vhost_put_avail_event(struct vhost_virtqueue *vq)
 {
+#if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE == 0
+	struct vhost_map *map;
+	struct vring_used *used;
+
+	if (!vq->iotlb) {
+		rcu_read_lock();
+
+		map = rcu_dereference(vq->maps[VHOST_ADDR_USED]);
+		if (likely(map)) {
+			used = map->addr;
+			*((__virtio16 *)&used->ring[vq->num]) =
+				cpu_to_vhost16(vq, vq->avail_idx);
+			rcu_read_unlock();
+			return 0;
+		}
+
+		rcu_read_unlock();
+	}
+#endif
+
 	return vhost_put_user(vq, cpu_to_vhost16(vq, vq->avail_idx),
 			      vhost_avail_event(vq));
 }
@@ -905,6 +1208,27 @@ static inline int vhost_put_used(struct vhost_virtqueue *vq,
 				 struct vring_used_elem *head, int idx,
 				 int count)
 {
+#if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE == 0
+	struct vhost_map *map;
+	struct vring_used *used;
+	size_t size;
+
+	if (!vq->iotlb) {
+		rcu_read_lock();
+
+		map = rcu_dereference(vq->maps[VHOST_ADDR_USED]);
+		if (likely(map)) {
+			used = map->addr;
+			size = count * sizeof(*head);
+			memcpy(used->ring + idx, head, size);
+			rcu_read_unlock();
+			return 0;
+		}
+
+		rcu_read_unlock();
+	}
+#endif
+
 	return vhost_copy_to_user(vq, vq->used->ring + idx, head,
 				  count * sizeof(*head));
 }
@@ -912,6 +1236,25 @@ static inline int vhost_put_used(struct vhost_virtqueue *vq,
 static inline int vhost_put_used_flags(struct vhost_virtqueue *vq)
 
 {
+#if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE == 0
+	struct vhost_map *map;
+	struct vring_used *used;
+
+	if (!vq->iotlb) {
+		rcu_read_lock();
+
+		map = rcu_dereference(vq->maps[VHOST_ADDR_USED]);
+		if (likely(map)) {
+			used = map->addr;
+			used->flags = cpu_to_vhost16(vq, vq->used_flags);
+			rcu_read_unlock();
+			return 0;
+		}
+
+		rcu_read_unlock();
+	}
+#endif
+
 	return vhost_put_user(vq, cpu_to_vhost16(vq, vq->used_flags),
 			      &vq->used->flags);
 }
@@ -919,6 +1262,25 @@ static inline int vhost_put_used_flags(struct vhost_virtqueue *vq)
 static inline int vhost_put_used_idx(struct vhost_virtqueue *vq)
 
 {
+#if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE == 0
+	struct vhost_map *map;
+	struct vring_used *used;
+
+	if (!vq->iotlb) {
+		rcu_read_lock();
+
+		map = rcu_dereference(vq->maps[VHOST_ADDR_USED]);
+		if (likely(map)) {
+			used = map->addr;
+			used->idx = cpu_to_vhost16(vq, vq->last_used_idx);
+			rcu_read_unlock();
+			return 0;
+		}
+
+		rcu_read_unlock();
+	}
+#endif
+
 	return vhost_put_user(vq, cpu_to_vhost16(vq, vq->last_used_idx),
 			      &vq->used->idx);
 }
@@ -964,12 +1326,50 @@ static void vhost_dev_unlock_vqs(struct vhost_dev *d)
 static inline int vhost_get_avail_idx(struct vhost_virtqueue *vq,
 				      __virtio16 *idx)
 {
+#if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE == 0
+	struct vhost_map *map;
+	struct vring_avail *avail;
+
+	if (!vq->iotlb) {
+		rcu_read_lock();
+
+		map = rcu_dereference(vq->maps[VHOST_ADDR_AVAIL]);
+		if (likely(map)) {
+			avail = map->addr;
+			*idx = avail->idx;
+			rcu_read_unlock();
+			return 0;
+		}
+
+		rcu_read_unlock();
+	}
+#endif
+
 	return vhost_get_avail(vq, *idx, &vq->avail->idx);
 }
 
 static inline int vhost_get_avail_head(struct vhost_virtqueue *vq,
 				       __virtio16 *head, int idx)
 {
+#if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE == 0
+	struct vhost_map *map;
+	struct vring_avail *avail;
+
+	if (!vq->iotlb) {
+		rcu_read_lock();
+
+		map = rcu_dereference(vq->maps[VHOST_ADDR_AVAIL]);
+		if (likely(map)) {
+			avail = map->addr;
+			*head = avail->ring[idx & (vq->num - 1)];
+			rcu_read_unlock();
+			return 0;
+		}
+
+		rcu_read_unlock();
+	}
+#endif
+
 	return vhost_get_avail(vq, *head,
 			       &vq->avail->ring[idx & (vq->num - 1)]);
 }
@@ -977,24 +1377,98 @@ static inline int vhost_get_avail_head(struct vhost_virtqueue *vq,
 static inline int vhost_get_avail_flags(struct vhost_virtqueue *vq,
 					__virtio16 *flags)
 {
+#if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE == 0
+	struct vhost_map *map;
+	struct vring_avail *avail;
+
+	if (!vq->iotlb) {
+		rcu_read_lock();
+
+		map = rcu_dereference(vq->maps[VHOST_ADDR_AVAIL]);
+		if (likely(map)) {
+			avail = map->addr;
+			*flags = avail->flags;
+			rcu_read_unlock();
+			return 0;
+		}
+
+		rcu_read_unlock();
+	}
+#endif
+
 	return vhost_get_avail(vq, *flags, &vq->avail->flags);
 }
 
 static inline int vhost_get_used_event(struct vhost_virtqueue *vq,
 				       __virtio16 *event)
 {
+#if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE == 0
+	struct vhost_map *map;
+	struct vring_avail *avail;
+
+	if (!vq->iotlb) {
+		rcu_read_lock();
+		map = rcu_dereference(vq->maps[VHOST_ADDR_AVAIL]);
+		if (likely(map)) {
+			avail = map->addr;
+			*event = (__virtio16)avail->ring[vq->num];
+			rcu_read_unlock();
+			return 0;
+		}
+		rcu_read_unlock();
+	}
+#endif
+
 	return vhost_get_avail(vq, *event, vhost_used_event(vq));
 }
 
 static inline int vhost_get_used_idx(struct vhost_virtqueue *vq,
 				     __virtio16 *idx)
 {
+#if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE == 0
+	struct vhost_map *map;
+	struct vring_used *used;
+
+	if (!vq->iotlb) {
+		rcu_read_lock();
+
+		map = rcu_dereference(vq->maps[VHOST_ADDR_USED]);
+		if (likely(map)) {
+			used = map->addr;
+			*idx = used->idx;
+			rcu_read_unlock();
+			return 0;
+		}
+
+		rcu_read_unlock();
+	}
+#endif
+
 	return vhost_get_used(vq, *idx, &vq->used->idx);
 }
 
 static inline int vhost_get_desc(struct vhost_virtqueue *vq,
 				 struct vring_desc *desc, int idx)
 {
+#if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE == 0
+	struct vhost_map *map;
+	struct vring_desc *d;
+
+	if (!vq->iotlb) {
+		rcu_read_lock();
+
+		map = rcu_dereference(vq->maps[VHOST_ADDR_DESC]);
+		if (likely(map)) {
+			d = map->addr;
+			*desc = *(d + idx);
+			rcu_read_unlock();
+			return 0;
+		}
+
+		rcu_read_unlock();
+	}
+#endif
+
 	return vhost_copy_from_user(vq, desc, vq->desc + idx, sizeof(*desc));
 }
 
@@ -1335,12 +1809,32 @@ static bool iotlb_access_ok(struct vhost_virtqueue *vq,
 	return true;
 }
 
+#if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE == 0
+static void vhost_vq_map_prefetch(struct vhost_virtqueue *vq)
+{
+	struct vhost_map __rcu *map;
+	int i;
+
+	for (i = 0; i < VHOST_NUM_ADDRS; i++) {
+		rcu_read_lock();
+		map = rcu_dereference(vq->maps[i]);
+		rcu_read_unlock();
+		if (unlikely(!map))
+			vhost_map_prefetch(vq, i);
+	}
+}
+#endif
+
 int vq_meta_prefetch(struct vhost_virtqueue *vq)
 {
 	unsigned int num = vq->num;
 
-	if (!vq->iotlb)
+	if (!vq->iotlb) {
+#if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE == 0
+		vhost_vq_map_prefetch(vq);
+#endif
 		return 1;
+	}
 
 	return iotlb_access_ok(vq, VHOST_ACCESS_RO, (u64)(uintptr_t)vq->desc,
 			       vhost_get_desc_size(vq, num), VHOST_ADDR_DESC) &&
@@ -1551,6 +2045,16 @@ static long vhost_vring_set_num_addr(struct vhost_dev *d,
 
 	mutex_lock(&vq->mutex);
 
+#if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE == 0
+	/* Unregister MMU notifer to allow invalidation callback
+	 * can access vq->uaddrs[] without holding a lock.
+	 */
+	if (d->mm)
+		mmu_notifier_unregister(&d->mmu_notifier, d->mm);
+
+	vhost_uninit_vq_maps(vq);
+#endif
+
 	switch (ioctl) {
 	case VHOST_SET_VRING_NUM:
 		r = vhost_vring_set_num(d, vq, argp);
@@ -1562,6 +2066,13 @@ static long vhost_vring_set_num_addr(struct vhost_dev *d,
 		BUG();
 	}
 
+#if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE == 0
+	vhost_setup_vq_uaddr(vq);
+
+	if (d->mm)
+		mmu_notifier_register(&d->mmu_notifier, d->mm);
+#endif
+
 	mutex_unlock(&vq->mutex);
 
 	return r;
diff --git a/drivers/vhost/vhost.h b/drivers/vhost/vhost.h
index 7a7fc001265f..f33208b55875 100644
--- a/drivers/vhost/vhost.h
+++ b/drivers/vhost/vhost.h
@@ -12,6 +12,9 @@
 #include <linux/virtio_config.h>
 #include <linux/virtio_ring.h>
 #include <linux/atomic.h>
+#include <linux/pagemap.h>
+#include <linux/mmu_notifier.h>
+#include <asm/cacheflush.h>
 
 struct vhost_work;
 typedef void (*vhost_work_fn_t)(struct vhost_work *work);
@@ -80,6 +83,18 @@ enum vhost_uaddr_type {
 	VHOST_NUM_ADDRS = 3,
 };
 
+struct vhost_map {
+	int npages;
+	void *addr;
+	struct page **pages;
+};
+
+struct vhost_uaddr {
+	unsigned long uaddr;
+	size_t size;
+	bool write;
+};
+
 /* The virtqueue structure describes a queue attached to a device. */
 struct vhost_virtqueue {
 	struct vhost_dev *dev;
@@ -90,7 +105,21 @@ struct vhost_virtqueue {
 	struct vring_desc __user *desc;
 	struct vring_avail __user *avail;
 	struct vring_used __user *used;
+
+#if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE == 0
+	/* read by memory accessors, modified by meta data
+	 * prefetching, MMU notifier and vring ioctl().
+	 * Synchonrized through mmu_lock (writers) and RCU (writers
+	 * and readers).
+	 */
+	struct vhost_map __rcu *maps[VHOST_NUM_ADDRS];
+	/* read by MMU notifier, write by vring ioctl(), synchronized
+	 * through register/unregister MMU notifier.
+	 */
+	struct vhost_uaddr uaddrs[VHOST_NUM_ADDRS];
+#endif
 	const struct vhost_umem_node *meta_iotlb[VHOST_NUM_ADDRS];
+
 	struct file *kick;
 	struct eventfd_ctx *call_ctx;
 	struct eventfd_ctx *error_ctx;
@@ -145,6 +174,8 @@ struct vhost_virtqueue {
 	bool user_be;
 #endif
 	u32 busyloop_timeout;
+	spinlock_t mmu_lock;
+	int invalidate_count;
 };
 
 struct vhost_msg_node {
@@ -158,6 +189,7 @@ struct vhost_msg_node {
 
 struct vhost_dev {
 	struct mm_struct *mm;
+	struct mmu_notifier mmu_notifier;
 	struct mutex mutex;
 	struct vhost_virtqueue **vqs;
 	int nvqs;
-- 
2.18.1

Re: [RFC PATCH V3 0/6] vhost: accelerate metadata access

[Jason Wang]

On 2019/4/23 下午1:54, Jason Wang wrote:
> This series tries to access virtqueue metadata through kernel virtual
> address instead of copy_user() friends since they had too much
> overheads like checks, spec barriers or even hardware feature
> toggling. This is done through setup kernel address through direct
> mapping and co-opreate VM management with MMU notifiers.
>
> Test shows about 23% improvement on TX PPS. TCP_STREAM doesn't see
> obvious improvement.


Ping. Comments are more than welcomed.

Thanks