Merge tag 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mst/vhost

Pull vhost infrastructure updates from Michael S. Tsirkin:
 "This reworks vhost core dropping unnecessary RCU uses in favor of VQ
  mutexes which are used on fast path anyway.  This fixes worst-case
  latency for users which change the memory mappings a lot.  Memory
  allocation for vhost-net now supports fallback on vmalloc (same as for
  vhost-scsi) this makes it possible to create the device on systems
  where memory is very fragmented, with slightly lower performance"

* tag 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mst/vhost:
  vhost: move memory pointer to VQs
  vhost: move acked_features to VQs
  vhost: replace rcu with mutex
  vhost-net: extend device allocation to vmalloc

drivers/vhost/net.c

@@ -17,6 +17,7 @@
 #include <linux/workqueue.h>
 #include <linux/file.h>
 #include <linux/slab.h>
+#include <linux/vmalloc.h>
 
 #include <linux/net.h>
 #include <linux/if_packet.h>
@@ -373,7 +374,7 @@ static void handle_tx(struct vhost_net *net)
 			      % UIO_MAXIOV == nvq->done_idx))
 			break;
 
-		head = vhost_get_vq_desc(&net->dev, vq, vq->iov,
+		head = vhost_get_vq_desc(vq, vq->iov,
 					 ARRAY_SIZE(vq->iov),
 					 &out, &in,
 					 NULL, NULL);
@@ -505,7 +506,7 @@ static int get_rx_bufs(struct vhost_virtqueue *vq,
 			r = -ENOBUFS;
 			goto err;
 		}
-		r = vhost_get_vq_desc(vq->dev, vq, vq->iov + seg,
+		r = vhost_get_vq_desc(vq, vq->iov + seg,
 				      ARRAY_SIZE(vq->iov) - seg, &out,
 				      &in, log, log_num);
 		if (unlikely(r < 0))
@@ -584,9 +585,9 @@ static void handle_rx(struct vhost_net *net)
 	vhost_hlen = nvq->vhost_hlen;
 	sock_hlen = nvq->sock_hlen;
 
-	vq_log = unlikely(vhost_has_feature(&net->dev, VHOST_F_LOG_ALL)) ?
+	vq_log = unlikely(vhost_has_feature(vq, VHOST_F_LOG_ALL)) ?
 		vq->log : NULL;
-	mergeable = vhost_has_feature(&net->dev, VIRTIO_NET_F_MRG_RXBUF);
+	mergeable = vhost_has_feature(vq, VIRTIO_NET_F_MRG_RXBUF);
 
 	while ((sock_len = peek_head_len(sock->sk))) {
 		sock_len += sock_hlen;
@@ -699,18 +700,30 @@ static void handle_rx_net(struct vhost_work *work)
 	handle_rx(net);
 }
 
+static void vhost_net_free(void *addr)
+{
+	if (is_vmalloc_addr(addr))
+		vfree(addr);
+	else
+		kfree(addr);
+}
+
 static int vhost_net_open(struct inode *inode, struct file *f)
 {
-	struct vhost_net *n = kmalloc(sizeof *n, GFP_KERNEL);
+	struct vhost_net *n;
 	struct vhost_dev *dev;
 	struct vhost_virtqueue **vqs;
 	int i;
 
+	n = kmalloc(sizeof *n, GFP_KERNEL | __GFP_NOWARN | __GFP_REPEAT);
+	if (!n) {
+		n = vmalloc(sizeof *n);
 		if (!n)
 			return -ENOMEM;
+	}
 	vqs = kmalloc(VHOST_NET_VQ_MAX * sizeof(*vqs), GFP_KERNEL);
 	if (!vqs) {
-		kfree(n);
+		vhost_net_free(n);
 		return -ENOMEM;
 	}
 
@@ -827,7 +840,7 @@ static int vhost_net_release(struct inode *inode, struct file *f)
 	 * since jobs can re-queue themselves. */
 	vhost_net_flush(n);
 	kfree(n->dev.vqs);
-	kfree(n);
+	vhost_net_free(n);
 	return 0;
 }
 
@@ -1038,15 +1051,13 @@ static int vhost_net_set_features(struct vhost_net *n, u64 features)
 		mutex_unlock(&n->dev.mutex);
 		return -EFAULT;
 	}
-	n->dev.acked_features = features;
-	smp_wmb();
 	for (i = 0; i < VHOST_NET_VQ_MAX; ++i) {
 		mutex_lock(&n->vqs[i].vq.mutex);
+		n->vqs[i].vq.acked_features = features;
 		n->vqs[i].vhost_hlen = vhost_hlen;
 		n->vqs[i].sock_hlen = sock_hlen;
 		mutex_unlock(&n->vqs[i].vq.mutex);
 	}
-	vhost_net_flush(n);
 	mutex_unlock(&n->dev.mutex);
 	return 0;
 }

drivers/vhost/scsi.c

@@ -606,7 +606,7 @@ tcm_vhost_do_evt_work(struct vhost_scsi *vs, struct tcm_vhost_evt *evt)
 
 again:
 	vhost_disable_notify(&vs->dev, vq);
-	head = vhost_get_vq_desc(&vs->dev, vq, vq->iov,
+	head = vhost_get_vq_desc(vq, vq->iov,
 			ARRAY_SIZE(vq->iov), &out, &in,
 			NULL, NULL);
 	if (head < 0) {
@@ -945,7 +945,7 @@ vhost_scsi_handle_vq(struct vhost_scsi *vs, struct vhost_virtqueue *vq)
 	vhost_disable_notify(&vs->dev, vq);
 
 	for (;;) {
-		head = vhost_get_vq_desc(&vs->dev, vq, vq->iov,
+		head = vhost_get_vq_desc(vq, vq->iov,
 					ARRAY_SIZE(vq->iov), &out, &in,
 					NULL, NULL);
 		pr_debug("vhost_get_vq_desc: head: %d, out: %u in: %u\n",
@@ -1373,6 +1373,9 @@ vhost_scsi_clear_endpoint(struct vhost_scsi *vs,
 
 static int vhost_scsi_set_features(struct vhost_scsi *vs, u64 features)
 {
+	struct vhost_virtqueue *vq;
+	int i;
+
 	if (features & ~VHOST_SCSI_FEATURES)
 		return -EOPNOTSUPP;
 
@@ -1382,9 +1385,13 @@ static int vhost_scsi_set_features(struct vhost_scsi *vs, u64 features)
 		mutex_unlock(&vs->dev.mutex);
 		return -EFAULT;
 	}
-	vs->dev.acked_features = features;
-	smp_wmb();
-	vhost_scsi_flush(vs);
+
+	for (i = 0; i < VHOST_SCSI_MAX_VQ; i++) {
+		vq = &vs->vqs[i].vq;
+		mutex_lock(&vq->mutex);
+		vq->acked_features = features;
+		mutex_unlock(&vq->mutex);
+	}
 	mutex_unlock(&vs->dev.mutex);
 	return 0;
 }
@@ -1591,10 +1598,6 @@ tcm_vhost_do_plug(struct tcm_vhost_tpg *tpg,
 		return;
 
 	mutex_lock(&vs->dev.mutex);
-	if (!vhost_has_feature(&vs->dev, VIRTIO_SCSI_F_HOTPLUG)) {
-		mutex_unlock(&vs->dev.mutex);
-		return;
-	}
 
 	if (plug)
 		reason = VIRTIO_SCSI_EVT_RESET_RESCAN;
@@ -1603,6 +1606,7 @@ tcm_vhost_do_plug(struct tcm_vhost_tpg *tpg,
 
 	vq = &vs->vqs[VHOST_SCSI_VQ_EVT].vq;
 	mutex_lock(&vq->mutex);
+	if (vhost_has_feature(vq, VIRTIO_SCSI_F_HOTPLUG))
 		tcm_vhost_send_evt(vs, tpg, lun,
 				   VIRTIO_SCSI_T_TRANSPORT_RESET, reason);
 	mutex_unlock(&vq->mutex);

drivers/vhost/test.c

@@ -53,7 +53,7 @@ static void handle_vq(struct vhost_test *n)
 	vhost_disable_notify(&n->dev, vq);
 
 	for (;;) {
-		head = vhost_get_vq_desc(&n->dev, vq, vq->iov,
+		head = vhost_get_vq_desc(vq, vq->iov,
 					 ARRAY_SIZE(vq->iov),
 					 &out, &in,
 					 NULL, NULL);
@@ -241,15 +241,18 @@ static long vhost_test_reset_owner(struct vhost_test *n)
 
 static int vhost_test_set_features(struct vhost_test *n, u64 features)
 {
+	struct vhost_virtqueue *vq;
+
 	mutex_lock(&n->dev.mutex);
 	if ((features & (1 << VHOST_F_LOG_ALL)) &&
 	    !vhost_log_access_ok(&n->dev)) {
 		mutex_unlock(&n->dev.mutex);
 		return -EFAULT;
 	}
-	n->dev.acked_features = features;
-	smp_wmb();
-	vhost_test_flush(n);
+	vq = &n->vqs[VHOST_TEST_VQ];
+	mutex_lock(&vq->mutex);
+	vq->acked_features = features;
+	mutex_unlock(&vq->mutex);
 	mutex_unlock(&n->dev.mutex);
 	return 0;
 }

drivers/vhost/vhost.h

@@ -104,20 +104,18 @@ struct vhost_virtqueue {
 	struct iovec *indirect;
 	struct vring_used_elem *heads;
 	/* Protected by virtqueue mutex. */
+	struct vhost_memory *memory;
 	void *private_data;
+	unsigned acked_features;
 	/* Log write descriptors */
 	void __user *log_base;
 	struct vhost_log *log;
 };
 
 struct vhost_dev {
-	/* Readers use RCU to access memory table pointer
-	 * log base pointer and features.
-	 * Writers use mutex below.*/
-	struct vhost_memory __rcu *memory;
+	struct vhost_memory *memory;
 	struct mm_struct *mm;
 	struct mutex mutex;
-	unsigned acked_features;
 	struct vhost_virtqueue **vqs;
 	int nvqs;
 	struct file *log_file;
@@ -140,7 +138,7 @@ long vhost_vring_ioctl(struct vhost_dev *d, int ioctl, void __user *argp);
 int vhost_vq_access_ok(struct vhost_virtqueue *vq);
 int vhost_log_access_ok(struct vhost_dev *);
 
-int vhost_get_vq_desc(struct vhost_dev *, struct vhost_virtqueue *,
+int vhost_get_vq_desc(struct vhost_virtqueue *,
 		      struct iovec iov[], unsigned int iov_count,
 		      unsigned int *out_num, unsigned int *in_num,
 		      struct vhost_log *log, unsigned int *log_num);
@@ -174,13 +172,8 @@ enum {
 			 (1ULL << VHOST_F_LOG_ALL),
 };
 
-static inline int vhost_has_feature(struct vhost_dev *dev, int bit)
+static inline int vhost_has_feature(struct vhost_virtqueue *vq, int bit)
 {
-	unsigned acked_features;
-
-	/* TODO: check that we are running from vhost_worker or dev mutex is
-	 * held? */
-	acked_features = rcu_dereference_index_check(dev->acked_features, 1);
-	return acked_features & (1 << bit);
+	return vq->acked_features & (1 << bit);
 }
 #endif