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

Pull vhost updates from Michael Tsirkin:
 "virtio, vhost: optimizations, fixes

  Looks like a quiet cycle for vhost/virtio, just a couple of minor
  tweaks. Most notable is automatic interrupt affinity for blk and scsi.
  Hopefully other devices are not far behind"

* tag 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mst/vhost:
  virtio-console: avoid DMA from stack
  vhost: introduce O(1) vq metadata cache
  virtio_scsi: use virtio IRQ affinity
  virtio_blk: use virtio IRQ affinity
  blk-mq: provide a default queue mapping for virtio device
  virtio: provide a method to get the IRQ affinity mask for a virtqueue
  virtio: allow drivers to request IRQ affinity when creating VQs
  virtio_pci: simplify MSI-X setup
  virtio_pci: don't duplicate the msix_enable flag in struct pci_dev
  virtio_pci: use shared interrupts for virtqueues
  virtio_pci: remove struct virtio_pci_vq_info
  vhost: try avoiding avail index access when getting descriptor
  virtio_mmio: expose header to userspace

31 files changed, 456 insertions, 454 deletions

diff --git a/block/Kconfig b/block/Kconfig
index a2a92e57a87d..e9f780f815f5 100644
--- a/block/Kconfig
+++ b/block/Kconfig
@@ -189,4 +189,9 @@ config BLK_MQ_PCI
 	depends on BLOCK && PCI
 	default y
 
+config BLK_MQ_VIRTIO
+	bool
+	depends on BLOCK && VIRTIO
+	default y
+
 source block/Kconfig.iosched
diff --git a/block/Makefile b/block/Makefile
index 2ad7c304e3f5..081bb680789b 100644
--- a/block/Makefile
+++ b/block/Makefile
@@ -25,6 +25,7 @@ obj-$(CONFIG_BLOCK_COMPAT)	+= compat_ioctl.o
 obj-$(CONFIG_BLK_CMDLINE_PARSER)	+= cmdline-parser.o
 obj-$(CONFIG_BLK_DEV_INTEGRITY) += bio-integrity.o blk-integrity.o t10-pi.o
 obj-$(CONFIG_BLK_MQ_PCI)	+= blk-mq-pci.o
+obj-$(CONFIG_BLK_MQ_VIRTIO)	+= blk-mq-virtio.o
 obj-$(CONFIG_BLK_DEV_ZONED)	+= blk-zoned.o
 obj-$(CONFIG_BLK_WBT)		+= blk-wbt.o
 obj-$(CONFIG_BLK_DEBUG_FS)	+= blk-mq-debugfs.o
diff --git a/block/blk-mq-virtio.c b/block/blk-mq-virtio.c
new file mode 100644
index 000000000000..c3afbca11299
--- /dev/null
+++ b/block/blk-mq-virtio.c
@@ -0,0 +1,54 @@
+/*
+ * Copyright (c) 2016 Christoph Hellwig.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ */
+#include <linux/device.h>
+#include <linux/blk-mq.h>
+#include <linux/blk-mq-virtio.h>
+#include <linux/virtio_config.h>
+#include <linux/module.h>
+#include "blk-mq.h"
+
+/**
+ * blk_mq_virtio_map_queues - provide a default queue mapping for virtio device
+ * @set:	tagset to provide the mapping for
+ * @vdev:	virtio device associated with @set.
+ * @first_vec:	first interrupt vectors to use for queues (usually 0)
+ *
+ * This function assumes the virtio device @vdev has at least as many available
+ * interrupt vetors as @set has queues.  It will then queuery the vector
+ * corresponding to each queue for it's affinity mask and built queue mapping
+ * that maps a queue to the CPUs that have irq affinity for the corresponding
+ * vector.
+ */
+int blk_mq_virtio_map_queues(struct blk_mq_tag_set *set,
+		struct virtio_device *vdev, int first_vec)
+{
+	const struct cpumask *mask;
+	unsigned int queue, cpu;
+
+	if (!vdev->config->get_vq_affinity)
+		goto fallback;
+
+	for (queue = 0; queue < set->nr_hw_queues; queue++) {
+		mask = vdev->config->get_vq_affinity(vdev, first_vec + queue);
+		if (!mask)
+			goto fallback;
+
+		for_each_cpu(cpu, mask)
+			set->mq_map[cpu] = queue;
+	}
+
+	return 0;
+fallback:
+	return blk_mq_map_queues(set);
+}
+EXPORT_SYMBOL_GPL(blk_mq_virtio_map_queues);
diff --git a/drivers/block/virtio_blk.c b/drivers/block/virtio_blk.c
index 024b473524c0..1d4c9f8bc1e1 100644
--- a/drivers/block/virtio_blk.c
+++ b/drivers/block/virtio_blk.c
@@ -5,6 +5,7 @@
 #include <linux/hdreg.h>
 #include <linux/module.h>
 #include <linux/mutex.h>
+#include <linux/interrupt.h>
 #include <linux/virtio.h>
 #include <linux/virtio_blk.h>
 #include <linux/scatterlist.h>
@@ -12,6 +13,7 @@
 #include <scsi/scsi_cmnd.h>
 #include <linux/idr.h>
 #include <linux/blk-mq.h>
+#include <linux/blk-mq-virtio.h>
 #include <linux/numa.h>
 
 #define PART_BITS 4
@@ -426,6 +428,7 @@ static int init_vq(struct virtio_blk *vblk)
 	struct virtqueue **vqs;
 	unsigned short num_vqs;
 	struct virtio_device *vdev = vblk->vdev;
+	struct irq_affinity desc = { 0, };
 
 	err = virtio_cread_feature(vdev, VIRTIO_BLK_F_MQ,
 				   struct virtio_blk_config, num_queues,
@@ -452,7 +455,8 @@ static int init_vq(struct virtio_blk *vblk)
 	}
 
 	/* Discover virtqueues and write information to configuration.  */
-	err = vdev->config->find_vqs(vdev, num_vqs, vqs, callbacks, names);
+	err = vdev->config->find_vqs(vdev, num_vqs, vqs, callbacks, names,
+			&desc);
 	if (err)
 		goto out;
 
@@ -586,10 +590,18 @@ static int virtblk_init_request(void *data, struct request *rq,
 	return 0;
 }
 
+static int virtblk_map_queues(struct blk_mq_tag_set *set)
+{
+	struct virtio_blk *vblk = set->driver_data;
+
+	return blk_mq_virtio_map_queues(set, vblk->vdev, 0);
+}
+
 static struct blk_mq_ops virtio_mq_ops = {
 	.queue_rq	= virtio_queue_rq,
 	.complete	= virtblk_request_done,
 	.init_request	= virtblk_init_request,
+	.map_queues	= virtblk_map_queues,
 };
 
 static unsigned int virtblk_queue_depth;
diff --git a/drivers/char/virtio_console.c b/drivers/char/virtio_console.c
index 17857beb4892..e9b7e0b3cabe 100644
--- a/drivers/char/virtio_console.c
+++ b/drivers/char/virtio_console.c
@@ -1136,6 +1136,8 @@ static int put_chars(u32 vtermno, const char *buf, int count)
 {
 	struct port *port;
 	struct scatterlist sg[1];
+	void *data;
+	int ret;
 
 	if (unlikely(early_put_chars))
 		return early_put_chars(vtermno, buf, count);
@@ -1144,8 +1146,14 @@ static int put_chars(u32 vtermno, const char *buf, int count)
 	if (!port)
 		return -EPIPE;
 
-	sg_init_one(sg, buf, count);
-	return __send_to_port(port, sg, 1, count, (void *)buf, false);
+	data = kmemdup(buf, count, GFP_ATOMIC);
+	if (!data)
+		return -ENOMEM;
+
+	sg_init_one(sg, data, count);
+	ret = __send_to_port(port, sg, 1, count, data, false);
+	kfree(data);
+	return ret;
 }
 
 /*
@@ -1939,7 +1947,7 @@ static int init_vqs(struct ports_device *portdev)
 	/* Find the queues. */
 	err = portdev->vdev->config->find_vqs(portdev->vdev, nr_queues, vqs,
 					      io_callbacks,
-					      (const char **)io_names);
+					      (const char **)io_names, NULL);
 	if (err)
 		goto free;
 
diff --git a/drivers/crypto/virtio/virtio_crypto_core.c b/drivers/crypto/virtio/virtio_crypto_core.c
index b5b153317376..21472e427f6f 100644
--- a/drivers/crypto/virtio/virtio_crypto_core.c
+++ b/drivers/crypto/virtio/virtio_crypto_core.c
@@ -120,7 +120,7 @@ static int virtcrypto_find_vqs(struct virtio_crypto *vi)
 	}
 
 	ret = vi->vdev->config->find_vqs(vi->vdev, total_vqs, vqs, callbacks,
-					 names);
+					 names, NULL);
 	if (ret)
 		goto err_find;
 
diff --git a/drivers/gpu/drm/virtio/virtgpu_kms.c b/drivers/gpu/drm/virtio/virtgpu_kms.c
index 30f989a0cafc..491866865c33 100644
--- a/drivers/gpu/drm/virtio/virtgpu_kms.c
+++ b/drivers/gpu/drm/virtio/virtgpu_kms.c
@@ -176,7 +176,7 @@ int virtio_gpu_driver_load(struct drm_device *dev, unsigned long flags)
 #endif
 
 	ret = vgdev->vdev->config->find_vqs(vgdev->vdev, 2, vqs,
-					    callbacks, names);
+					    callbacks, names, NULL);
 	if (ret) {
 		DRM_ERROR("failed to find virt queues\n");
 		goto err_vqs;
diff --git a/drivers/misc/mic/vop/vop_main.c b/drivers/misc/mic/vop/vop_main.c
index 1a2b67f3183d..c2e29d7f0de8 100644
--- a/drivers/misc/mic/vop/vop_main.c
+++ b/drivers/misc/mic/vop/vop_main.c
@@ -374,7 +374,7 @@ unmap:
 static int vop_find_vqs(struct virtio_device *dev, unsigned nvqs,
 			struct virtqueue *vqs[],
 			vq_callback_t *callbacks[],
-			const char * const names[])
+			const char * const names[], struct irq_affinity *desc)
 {
 	struct _vop_vdev *vdev = to_vopvdev(dev);
 	struct vop_device *vpdev = vdev->vpdev;
diff --git a/drivers/net/caif/caif_virtio.c b/drivers/net/caif/caif_virtio.c
index b306210b02b7..bc0eb47eccee 100644
--- a/drivers/net/caif/caif_virtio.c
+++ b/drivers/net/caif/caif_virtio.c
@@ -679,7 +679,8 @@ static int cfv_probe(struct virtio_device *vdev)
 		goto err;
 
 	/* Get the TX virtio ring. This is a "guest side vring". */
-	err = vdev->config->find_vqs(vdev, 1, &cfv->vq_tx, &vq_cbs, &names);
+	err = vdev->config->find_vqs(vdev, 1, &cfv->vq_tx, &vq_cbs, &names,
+			NULL);
 	if (err)
 		goto err;
 
diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index bf95016f442a..66fd3139be60 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -2080,7 +2080,7 @@ static int virtnet_find_vqs(struct virtnet_info *vi)
 	}
 
 	ret = vi->vdev->config->find_vqs(vi->vdev, total_vqs, vqs, callbacks,
-					 names);
+					 names, NULL);
 	if (ret)
 		goto err_find;
 
diff --git a/drivers/remoteproc/remoteproc_virtio.c b/drivers/remoteproc/remoteproc_virtio.c
index 364411fb7734..0142cc3f0c91 100644
--- a/drivers/remoteproc/remoteproc_virtio.c
+++ b/drivers/remoteproc/remoteproc_virtio.c
@@ -137,7 +137,8 @@ static void rproc_virtio_del_vqs(struct virtio_device *vdev)
 static int rproc_virtio_find_vqs(struct virtio_device *vdev, unsigned int nvqs,
 				 struct virtqueue *vqs[],
 				 vq_callback_t *callbacks[],
-				 const char * const names[])
+				 const char * const names[],
+				 struct irq_affinity *desc)
 {
 	int i, ret;
 
diff --git a/drivers/rpmsg/virtio_rpmsg_bus.c b/drivers/rpmsg/virtio_rpmsg_bus.c
index 3090b0d3072f..5e66e081027e 100644
--- a/drivers/rpmsg/virtio_rpmsg_bus.c
+++ b/drivers/rpmsg/virtio_rpmsg_bus.c
@@ -869,7 +869,7 @@ static int rpmsg_probe(struct virtio_device *vdev)
 	init_waitqueue_head(&vrp->sendq);
 
 	/* We expect two virtqueues, rx and tx (and in this order) */
-	err = vdev->config->find_vqs(vdev, 2, vqs, vq_cbs, names);
+	err = vdev->config->find_vqs(vdev, 2, vqs, vq_cbs, names, NULL);
 	if (err)
 		goto free_vrp;
 
diff --git a/drivers/s390/virtio/kvm_virtio.c b/drivers/s390/virtio/kvm_virtio.c
index 5e5c11f37b24..2ce0b3eb2efe 100644
--- a/drivers/s390/virtio/kvm_virtio.c
+++ b/drivers/s390/virtio/kvm_virtio.c
@@ -255,7 +255,8 @@ static void kvm_del_vqs(struct virtio_device *vdev)
 static int kvm_find_vqs(struct virtio_device *vdev, unsigned nvqs,
 			struct virtqueue *vqs[],
 			vq_callback_t *callbacks[],
-			const char * const names[])
+			const char * const names[],
+			struct irq_affinity *desc)
 {
 	struct kvm_device *kdev = to_kvmdev(vdev);
 	int i;
diff --git a/drivers/s390/virtio/virtio_ccw.c b/drivers/s390/virtio/virtio_ccw.c
index 648373cde4a1..0ed209f3d8b0 100644
--- a/drivers/s390/virtio/virtio_ccw.c
+++ b/drivers/s390/virtio/virtio_ccw.c
@@ -628,7 +628,8 @@ out:
 static int virtio_ccw_find_vqs(struct virtio_device *vdev, unsigned nvqs,
 			       struct virtqueue *vqs[],
 			       vq_callback_t *callbacks[],
-			       const char * const names[])
+			       const char * const names[],
+			       struct irq_affinity *desc)
 {
 	struct virtio_ccw_device *vcdev = to_vc_device(vdev);
 	unsigned long *indicatorp = NULL;
diff --git a/drivers/scsi/virtio_scsi.c b/drivers/scsi/virtio_scsi.c
index c680d7641311..939c47df73fa 100644
--- a/drivers/scsi/virtio_scsi.c
+++ b/drivers/scsi/virtio_scsi.c
@@ -18,6 +18,7 @@
 #include <linux/module.h>
 #include <linux/slab.h>
 #include <linux/mempool.h>
+#include <linux/interrupt.h>
 #include <linux/virtio.h>
 #include <linux/virtio_ids.h>
 #include <linux/virtio_config.h>
@@ -29,6 +30,7 @@
 #include <scsi/scsi_cmnd.h>
 #include <scsi/scsi_tcq.h>
 #include <linux/seqlock.h>
+#include <linux/blk-mq-virtio.h>
 
 #define VIRTIO_SCSI_MEMPOOL_SZ 64
 #define VIRTIO_SCSI_EVENT_LEN 8
@@ -108,7 +110,6 @@ struct virtio_scsi {
 	bool affinity_hint_set;
 
 	struct hlist_node node;
-	struct hlist_node node_dead;
 
 	/* Protected by event_vq lock */
 	bool stop_events;
@@ -118,7 +119,6 @@ struct virtio_scsi {
 	struct virtio_scsi_vq req_vqs[];
 };
 
-static enum cpuhp_state virtioscsi_online;
 static struct kmem_cache *virtscsi_cmd_cache;
 static mempool_t *virtscsi_cmd_pool;
 
@@ -766,6 +766,13 @@ static void virtscsi_target_destroy(struct scsi_target *starget)
 	kfree(tgt);
 }
 
+static int virtscsi_map_queues(struct Scsi_Host *shost)
+{
+	struct virtio_scsi *vscsi = shost_priv(shost);
+
+	return blk_mq_virtio_map_queues(&shost->tag_set, vscsi->vdev, 2);
+}
+
 static struct scsi_host_template virtscsi_host_template_single = {
 	.module = THIS_MODULE,
 	.name = "Virtio SCSI HBA",
@@ -801,6 +808,7 @@ static struct scsi_host_template virtscsi_host_template_multi = {
 	.use_clustering = ENABLE_CLUSTERING,
 	.target_alloc = virtscsi_target_alloc,
 	.target_destroy = virtscsi_target_destroy,
+	.map_queues = virtscsi_map_queues,
 	.track_queue_depth = 1,
 };
 
@@ -817,80 +825,6 @@ static struct scsi_host_template virtscsi_host_template_multi = {
 		virtio_cwrite(vdev, struct virtio_scsi_config, fld, &__val); \
 	} while(0)
 
-static void __virtscsi_set_affinity(struct virtio_scsi *vscsi, bool affinity)
-{
-	int i;
-	int cpu;
-
-	/* In multiqueue mode, when the number of cpu is equal
-	 * to the number of request queues, we let the qeueues
-	 * to be private to one cpu by setting the affinity hint
-	 * to eliminate the contention.
-	 */
-	if ((vscsi->num_queues == 1 ||
-	     vscsi->num_queues != num_online_cpus()) && affinity) {
-		if (vscsi->affinity_hint_set)
-			affinity = false;
-		else
-			return;
-	}
-
-	if (affinity) {
-		i = 0;
-		for_each_online_cpu(cpu) {
-			virtqueue_set_affinity(vscsi->req_vqs[i].vq, cpu);
-			i++;
-		}
-
-		vscsi->affinity_hint_set = true;
-	} else {
-		for (i = 0; i < vscsi->num_queues; i++) {
-			if (!vscsi->req_vqs[i].vq)
-				continue;
-
-			virtqueue_set_affinity(vscsi->req_vqs[i].vq, -1);
-		}
-
-		vscsi->affinity_hint_set = false;
-	}
-}
-
-static void virtscsi_set_affinity(struct virtio_scsi *vscsi, bool affinity)
-{
-	get_online_cpus();
-	__virtscsi_set_affinity(vscsi, affinity);
-	put_online_cpus();
-}
-
-static int virtscsi_cpu_online(unsigned int cpu, struct hlist_node *node)
-{
-	struct virtio_scsi *vscsi = hlist_entry_safe(node, struct virtio_scsi,
-						     node);
-	__virtscsi_set_affinity(vscsi, true);
-	return 0;
-}
-
-static int virtscsi_cpu_notif_add(struct virtio_scsi *vi)
-{
-	int ret;
-
-	ret = cpuhp_state_add_instance(virtioscsi_online, &vi->node);
-	if (ret)
-		return ret;
-
-	ret = cpuhp_state_add_instance(CPUHP_VIRT_SCSI_DEAD, &vi->node_dead);
-	if (ret)
-		cpuhp_state_remove_instance(virtioscsi_online, &vi->node);
-	return ret;
-}
-
-static void virtscsi_cpu_notif_remove(struct virtio_scsi *vi)
-{
-	cpuhp_state_remove_instance_nocalls(virtioscsi_online, &vi->node);
-	cpuhp_state_remove_instance_nocalls(CPUHP_VIRT_SCSI_DEAD,
-					    &vi->node_dead);
-}
-
 static void virtscsi_init_vq(struct virtio_scsi_vq *virtscsi_vq,
 			     struct virtqueue *vq)
 {
@@ -900,14 +834,8 @@ static void virtscsi_init_vq(struct virtio_scsi_vq *virtscsi_vq,
 
 static void virtscsi_remove_vqs(struct virtio_device *vdev)
 {
-	struct Scsi_Host *sh = virtio_scsi_host(vdev);
-	struct virtio_scsi *vscsi = shost_priv(sh);
-
-	virtscsi_set_affinity(vscsi, false);
-
 	/* Stop all the virtqueues. */
 	vdev->config->reset(vdev);
-
 	vdev->config->del_vqs(vdev);
 }
 
@@ -920,6 +848,7 @@ static int virtscsi_init(struct virtio_device *vdev,
 	vq_callback_t **callbacks;
 	const char **names;
 	struct virtqueue **vqs;
+	struct irq_affinity desc = { .pre_vectors = 2 };
 
 	num_vqs = vscsi->num_queues + VIRTIO_SCSI_VQ_BASE;
 	vqs = kmalloc(num_vqs * sizeof(struct virtqueue *), GFP_KERNEL);
@@ -941,7 +870,8 @@ static int virtscsi_init(struct virtio_device *vdev,
 	}
 
 	/* Discover virtqueues and write information to configuration.  */
-	err = vdev->config->find_vqs(vdev, num_vqs, vqs, callbacks, names);
+	err = vdev->config->find_vqs(vdev, num_vqs, vqs, callbacks, names,
+			&desc);
 	if (err)
 		goto out;
 
@@ -1007,10 +937,6 @@ static int virtscsi_probe(struct virtio_device *vdev)
 	if (err)
 		goto virtscsi_init_failed;
 
-	err = virtscsi_cpu_notif_add(vscsi);
-	if (err)
-		goto scsi_add_host_failed;
-
 	cmd_per_lun = virtscsi_config_get(vdev, cmd_per_lun) ?: 1;
 	shost->cmd_per_lun = min_t(u32, cmd_per_lun, shost->can_queue);
 	shost->max_sectors = virtscsi_config_get(vdev, max_sectors) ?: 0xFFFF;
@@ -1065,9 +991,6 @@ static void virtscsi_remove(struct virtio_device *vdev)
 		virtscsi_cancel_event_work(vscsi);
 
 	scsi_remove_host(shost);
-
-	virtscsi_cpu_notif_remove(vscsi);
-
 	virtscsi_remove_vqs(vdev);
 	scsi_host_put(shost);
 }
@@ -1075,10 +998,6 @@ static void virtscsi_remove(struct virtio_device *vdev)
 #ifdef CONFIG_PM_SLEEP
 static int virtscsi_freeze(struct virtio_device *vdev)
 {
-	struct Scsi_Host *sh = virtio_scsi_host(vdev);
-	struct virtio_scsi *vscsi = shost_priv(sh);
-
-	virtscsi_cpu_notif_remove(vscsi);
 	virtscsi_remove_vqs(vdev);
 	return 0;
 }
@@ -1093,11 +1012,6 @@ static int virtscsi_restore(struct virtio_device *vdev)
 	if (err)
 		return err;
 
-	err = virtscsi_cpu_notif_add(vscsi);
-	if (err) {
-		vdev->config->del_vqs(vdev);
-		return err;
-	}
 	virtio_device_ready(vdev);
 
 	if (virtio_has_feature(vdev, VIRTIO_SCSI_F_HOTPLUG))
@@ -1152,16 +1066,6 @@ static int __init init(void)
 		pr_err("mempool_create() for virtscsi_cmd_pool failed\n");
 		goto error;
 	}
-	ret = cpuhp_setup_state_multi(CPUHP_AP_ONLINE_DYN,
-				      "scsi/virtio:online",
-				      virtscsi_cpu_online, NULL);
-	if (ret < 0)
-		goto error;
-	virtioscsi_online = ret;
-	ret = cpuhp_setup_state_multi(CPUHP_VIRT_SCSI_DEAD, "scsi/virtio:dead",
-				      NULL, virtscsi_cpu_online);
-	if (ret)
-		goto error;
 	ret = register_virtio_driver(&virtio_scsi_driver);
 	if (ret < 0)
 		goto error;
@@ -1177,17 +1081,12 @@ error:
 		kmem_cache_destroy(virtscsi_cmd_cache);
 		virtscsi_cmd_cache = NULL;
 	}
-	if (virtioscsi_online)
-		cpuhp_remove_multi_state(virtioscsi_online);
-	cpuhp_remove_multi_state(CPUHP_VIRT_SCSI_DEAD);
 	return ret;
 }
 
 static void __exit fini(void)
 {
 	unregister_virtio_driver(&virtio_scsi_driver);
-	cpuhp_remove_multi_state(virtioscsi_online);
-	cpuhp_remove_multi_state(CPUHP_VIRT_SCSI_DEAD);
 	mempool_destroy(virtscsi_cmd_pool);
 	kmem_cache_destroy(virtscsi_cmd_cache);
 }
diff --git a/drivers/vhost/vhost.c b/drivers/vhost/vhost.c
index 4269e621e254..9469364eefd7 100644
--- a/drivers/vhost/vhost.c
+++ b/drivers/vhost/vhost.c
@@ -282,6 +282,22 @@ void vhost_poll_queue(struct vhost_poll *poll)
 }
 EXPORT_SYMBOL_GPL(vhost_poll_queue);
 
+static void __vhost_vq_meta_reset(struct vhost_virtqueue *vq)
+{
+	int j;
+
+	for (j = 0; j < VHOST_NUM_ADDRS; j++)
+		vq->meta_iotlb[j] = NULL;
+}
+
+static void vhost_vq_meta_reset(struct vhost_dev *d)
+{
+	int i;
+
+	for (i = 0; i < d->nvqs; ++i)
+		__vhost_vq_meta_reset(d->vqs[i]);
+}
+
 static void vhost_vq_reset(struct vhost_dev *dev,
 			   struct vhost_virtqueue *vq)
 {
@@ -312,6 +328,7 @@ static void vhost_vq_reset(struct vhost_dev *dev,
 	vq->busyloop_timeout = 0;
 	vq->umem = NULL;
 	vq->iotlb = NULL;
+	__vhost_vq_meta_reset(vq);
 }
 
 static int vhost_worker(void *data)
@@ -691,6 +708,18 @@ static int vq_memory_access_ok(void __user *log_base, struct vhost_umem *umem,
 	return 1;
 }
 
+static inline void __user *vhost_vq_meta_fetch(struct vhost_virtqueue *vq,
+					       u64 addr, unsigned int size,
+					       int type)
+{
+	const struct vhost_umem_node *node = vq->meta_iotlb[type];
+
+	if (!node)
+		return NULL;
+
+	return (void *)(uintptr_t)(node->userspace_addr + addr - node->start);
+}
+
 /* Can we switch to this memory table? */
 /* Caller should have device mutex but not vq mutex */
 static int memory_access_ok(struct vhost_dev *d, struct vhost_umem *umem,
@@ -733,8 +762,14 @@ static int vhost_copy_to_user(struct vhost_virtqueue *vq, void __user *to,
 		 * could be access through iotlb. So -EAGAIN should
 		 * not happen in this case.
 		 */
-		/* TODO: more fast path */
 		struct iov_iter t;
+		void __user *uaddr = vhost_vq_meta_fetch(vq,
+				     (u64)(uintptr_t)to, size,
+				     VHOST_ADDR_DESC);
+
+		if (uaddr)
+			return __copy_to_user(uaddr, from, size);
+
 		ret = translate_desc(vq, (u64)(uintptr_t)to, size, vq->iotlb_iov,
 				     ARRAY_SIZE(vq->iotlb_iov),
 				     VHOST_ACCESS_WO);
@@ -762,8 +797,14 @@ static int vhost_copy_from_user(struct vhost_virtqueue *vq, void *to,
 		 * could be access through iotlb. So -EAGAIN should
 		 * not happen in this case.
 		 */
-		/* TODO: more fast path */
+		void __user *uaddr = vhost_vq_meta_fetch(vq,
+				     (u64)(uintptr_t)from, size,
+				     VHOST_ADDR_DESC);
 		struct iov_iter f;
+
+		if (uaddr)
+			return __copy_from_user(to, uaddr, size);
+
 		ret = translate_desc(vq, (u64)(uintptr_t)from, size, vq->iotlb_iov,
 				     ARRAY_SIZE(vq->iotlb_iov),
 				     VHOST_ACCESS_RO);
@@ -783,17 +824,12 @@ out:
 	return ret;
 }
 
-static void __user *__vhost_get_user(struct vhost_virtqueue *vq,
-				     void __user *addr, unsigned size)
+static void __user *__vhost_get_user_slow(struct vhost_virtqueue *vq,
+					  void __user *addr, unsigned int size,
+					  int type)
 {
 	int ret;
 
-	/* This function should be called after iotlb
-	 * prefetch, which means we're sure that vq
-	 * could be access through iotlb. So -EAGAIN should
-	 * not happen in this case.
-	 */
-	/* TODO: more fast path */
 	ret = translate_desc(vq, (u64)(uintptr_t)addr, size, vq->iotlb_iov,
 			     ARRAY_SIZE(vq->iotlb_iov),
 			     VHOST_ACCESS_RO);
@@ -814,14 +850,32 @@ static void __user *__vhost_get_user(struct vhost_virtqueue *vq,
 	return vq->iotlb_iov[0].iov_base;
 }
 
-#define vhost_put_user(vq, x, ptr) \
+/* This function should be called after iotlb
+ * prefetch, which means we're sure that vq
+ * could be access through iotlb. So -EAGAIN should
+ * not happen in this case.
+ */
+static inline void __user *__vhost_get_user(struct vhost_virtqueue *vq,
+					    void *addr, unsigned int size,
+					    int type)
+{
+	void __user *uaddr = vhost_vq_meta_fetch(vq,
+			     (u64)(uintptr_t)addr, size, type);
+	if (uaddr)
+		return uaddr;
+
+	return __vhost_get_user_slow(vq, addr, size, type);
+}
+
+#define vhost_put_user(vq, x, ptr)		\
 ({ \
 	int ret = -EFAULT; \
 	if (!vq->iotlb) { \
 		ret = __put_user(x, ptr); \
 	} else { \
 		__typeof__(ptr) to = \
-			(__typeof__(ptr)) __vhost_get_user(vq, ptr, sizeof(*ptr)); \
+			(__typeof__(ptr)) __vhost_get_user(vq, ptr,	\
+					  sizeof(*ptr), VHOST_ADDR_USED); \
 		if (to != NULL) \
 			ret = __put_user(x, to); \
 		else \
@@ -830,14 +884,16 @@ static void __user *__vhost_get_user(struct vhost_virtqueue *vq,
 	ret; \
 })
 
-#define vhost_get_user(vq, x, ptr) \
+#define vhost_get_user(vq, x, ptr, type)		\
 ({ \
 	int ret; \
 	if (!vq->iotlb) { \
 		ret = __get_user(x, ptr); \
 	} else { \
 		__typeof__(ptr) from = \
-			(__typeof__(ptr)) __vhost_get_user(vq, ptr, sizeof(*ptr)); \
+			(__typeof__(ptr)) __vhost_get_user(vq, ptr, \
+							   sizeof(*ptr), \
+							   type); \
 		if (from != NULL) \
 			ret = __get_user(x, from); \
 		else \
@@ -846,6 +902,12 @@ static void __user *__vhost_get_user(struct vhost_virtqueue *vq,
 	ret; \
 })
 
+#define vhost_get_avail(vq, x, ptr) \
+	vhost_get_user(vq, x, ptr, VHOST_ADDR_AVAIL)
+
+#define vhost_get_used(vq, x, ptr) \
+	vhost_get_user(vq, x, ptr, VHOST_ADDR_USED)
+
 static void vhost_dev_lock_vqs(struct vhost_dev *d)
 {
 	int i = 0;
@@ -951,6 +1013,7 @@ static int vhost_process_iotlb_msg(struct vhost_dev *dev,
 			ret = -EFAULT;
 			break;
 		}
+		vhost_vq_meta_reset(dev);
 		if (vhost_new_umem_range(dev->iotlb, msg->iova, msg->size,
 					 msg->iova + msg->size - 1,
 					 msg->uaddr, msg->perm)) {
@@ -960,6 +1023,7 @@ static int vhost_process_iotlb_msg(struct vhost_dev *dev,
 		vhost_iotlb_notify_vq(dev, msg);
 		break;
 	case VHOST_IOTLB_INVALIDATE:
+		vhost_vq_meta_reset(dev);
 		vhost_del_umem_range(dev->iotlb, msg->iova,
 				     msg->iova + msg->size - 1);
 		break;
@@ -1103,12 +1167,26 @@ static int vq_access_ok(struct vhost_virtqueue *vq, unsigned int num,
 			sizeof *used + num * sizeof *used->ring + s);
 }
 
+static void vhost_vq_meta_update(struct vhost_virtqueue *vq,
+				 const struct vhost_umem_node *node,
+				 int type)
+{
+	int access = (type == VHOST_ADDR_USED) ?
+		     VHOST_ACCESS_WO : VHOST_ACCESS_RO;
+
+	if (likely(node->perm & access))
+		vq->meta_iotlb[type] = node;
+}
+
 static int iotlb_access_ok(struct vhost_virtqueue *vq,
-			   int access, u64 addr, u64 len)
+			   int access, u64 addr, u64 len, int type)
 {
 	const struct vhost_umem_node *node;
 	struct vhost_umem *umem = vq->iotlb;
-	u64 s = 0, size;
+	u64 s = 0, size, orig_addr = addr;
+
+	if (vhost_vq_meta_fetch(vq, addr, len, type))
+		return true;
 
 	while (len > s) {
 		node = vhost_umem_interval_tree_iter_first(&umem->umem_tree,
@@ -1125,6 +1203,1