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

Pull more vhost fixes from Michael Tsirkin:
 "This fixes some minor issues in the patches that have been merged.

  We also finally drop the workaround disabling event_idx for scsi: it
  was always questionable, and now we know it's not needed.

  There's also a memory leak fix"

* tag 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mst/vhost:
  vhost-scsi: Enable VIRTIO_RING_F_EVENT_IDX
  vhost: drop virtio_net.h dependency
  vhost-net: Cleanup vhost_ubuf and vhost_zcopy
  vhost: Remove vhost_enable_zcopy in vhost.h
  vhost: Remove comments for hdr in vhost.h
  vhost: Move VHOST_NET_FEATURES to net.c
  vhost-net: Free ubuf when vhost_dev_set_owner fails
  vhost: Export vhost_dev_set_owner

diff --git a/drivers/vhost/net.c b/drivers/vhost/net.c
index a3645bd163d8ed0de10af23dbb72d30aab516fd3..2b51e2336aa20006e47cb0f1ac7ea146120413a3 100644 (file)
--- a/drivers/vhost/net.c
+++ b/drivers/vhost/net.c
@@ -58,13 +58,19 @@ MODULE_PARM_DESC(experimental_zcopytx, "Enable Zero Copy TX;"
 
 #define VHOST_DMA_IS_DONE(len) ((len) >= VHOST_DMA_DONE_LEN)
 
+enum {
+       VHOST_NET_FEATURES = VHOST_FEATURES |
+                        (1ULL << VHOST_NET_F_VIRTIO_NET_HDR) |
+                        (1ULL << VIRTIO_NET_F_MRG_RXBUF),
+};
+
 enum {
        VHOST_NET_VQ_RX = 0,
        VHOST_NET_VQ_TX = 1,
        VHOST_NET_VQ_MAX = 2,
 };
 
-struct vhost_ubuf_ref {
+struct vhost_net_ubuf_ref {
        struct kref kref;
        wait_queue_head_t wait;
        struct vhost_virtqueue *vq;
@@ -87,7 +93,7 @@ struct vhost_net_virtqueue {
        struct ubuf_info *ubuf_info;
        /* Reference counting for outstanding ubufs.
         * Protected by vq mutex. Writers must also take device mutex. */
-       struct vhost_ubuf_ref *ubufs;
+       struct vhost_net_ubuf_ref *ubufs;
 };
 
 struct vhost_net {
@@ -104,24 +110,25 @@ struct vhost_net {
        bool tx_flush;
 };
 
-static unsigned vhost_zcopy_mask __read_mostly;
+static unsigned vhost_net_zcopy_mask __read_mostly;
 
-void vhost_enable_zcopy(int vq)
+static void vhost_net_enable_zcopy(int vq)
 {
-       vhost_zcopy_mask |= 0x1 << vq;
+       vhost_net_zcopy_mask |= 0x1 << vq;
 }
 
-static void vhost_zerocopy_done_signal(struct kref *kref)
+static void vhost_net_zerocopy_done_signal(struct kref *kref)
 {
-       struct vhost_ubuf_ref *ubufs = container_of(kref, struct vhost_ubuf_ref,
-                                                   kref);
+       struct vhost_net_ubuf_ref *ubufs;
+
+       ubufs = container_of(kref, struct vhost_net_ubuf_ref, kref);
        wake_up(&ubufs->wait);
 }
 
-struct vhost_ubuf_ref *vhost_ubuf_alloc(struct vhost_virtqueue *vq,
-                                       bool zcopy)
+static struct vhost_net_ubuf_ref *
+vhost_net_ubuf_alloc(struct vhost_virtqueue *vq, bool zcopy)
 {
-       struct vhost_ubuf_ref *ubufs;
+       struct vhost_net_ubuf_ref *ubufs;
        /* No zero copy backend? Nothing to count. */
        if (!zcopy)
                return NULL;
@@ -134,25 +141,38 @@ struct vhost_ubuf_ref *vhost_ubuf_alloc(struct vhost_virtqueue *vq,
        return ubufs;
 }
 
-void vhost_ubuf_put(struct vhost_ubuf_ref *ubufs)
+static void vhost_net_ubuf_put(struct vhost_net_ubuf_ref *ubufs)
 {
-       kref_put(&ubufs->kref, vhost_zerocopy_done_signal);
+       kref_put(&ubufs->kref, vhost_net_zerocopy_done_signal);
 }
 
-void vhost_ubuf_put_and_wait(struct vhost_ubuf_ref *ubufs)
+static void vhost_net_ubuf_put_and_wait(struct vhost_net_ubuf_ref *ubufs)
 {
-       kref_put(&ubufs->kref, vhost_zerocopy_done_signal);
+       kref_put(&ubufs->kref, vhost_net_zerocopy_done_signal);
        wait_event(ubufs->wait, !atomic_read(&ubufs->kref.refcount));
        kfree(ubufs);
 }
 
+static void vhost_net_clear_ubuf_info(struct vhost_net *n)
+{
+
+       bool zcopy;
+       int i;
+
+       for (i = 0; i < n->dev.nvqs; ++i) {
+               zcopy = vhost_net_zcopy_mask & (0x1 << i);
+               if (zcopy)
+                       kfree(n->vqs[i].ubuf_info);
+       }
+}
+
 int vhost_net_set_ubuf_info(struct vhost_net *n)
 {
        bool zcopy;
        int i;
 
        for (i = 0; i < n->dev.nvqs; ++i) {
-               zcopy = vhost_zcopy_mask & (0x1 << i);
+               zcopy = vhost_net_zcopy_mask & (0x1 << i);
                if (!zcopy)
                        continue;
                n->vqs[i].ubuf_info = kmalloc(sizeof(*n->vqs[i].ubuf_info) *
@@ -164,7 +184,7 @@ int vhost_net_set_ubuf_info(struct vhost_net *n)
 
 err:
        while (i--) {
-               zcopy = vhost_zcopy_mask & (0x1 << i);
+               zcopy = vhost_net_zcopy_mask & (0x1 << i);
                if (!zcopy)
                        continue;
                kfree(n->vqs[i].ubuf_info);
@@ -286,7 +306,7 @@ static int vhost_zerocopy_signal_used(struct vhost_net *net,
 
 static void vhost_zerocopy_callback(struct ubuf_info *ubuf, bool success)
 {
-       struct vhost_ubuf_ref *ubufs = ubuf->ctx;
+       struct vhost_net_ubuf_ref *ubufs = ubuf->ctx;
        struct vhost_virtqueue *vq = ubufs->vq;
        int cnt = atomic_read(&ubufs->kref.refcount);
 
@@ -303,7 +323,7 @@ static void vhost_zerocopy_callback(struct ubuf_info *ubuf, bool success)
        /* set len to mark this desc buffers done DMA */
        vq->heads[ubuf->desc].len = success ?
                VHOST_DMA_DONE_LEN : VHOST_DMA_FAILED_LEN;
-       vhost_ubuf_put(ubufs);
+       vhost_net_ubuf_put(ubufs);
 }
 
 /* Expects to be always run from workqueue - which acts as
@@ -326,7 +346,7 @@ static void handle_tx(struct vhost_net *net)
        int err;
        size_t hdr_size;
        struct socket *sock;
-       struct vhost_ubuf_ref *uninitialized_var(ubufs);
+       struct vhost_net_ubuf_ref *uninitialized_var(ubufs);
        bool zcopy, zcopy_used;
 
        /* TODO: check that we are running from vhost_worker? */
@@ -422,7 +442,7 @@ static void handle_tx(struct vhost_net *net)
                if (unlikely(err < 0)) {
                        if (zcopy_used) {
                                if (ubufs)
-                                       vhost_ubuf_put(ubufs);
+                                       vhost_net_ubuf_put(ubufs);
                                nvq->upend_idx = ((unsigned)nvq->upend_idx - 1)
                                        % UIO_MAXIOV;
                        }
@@ -776,7 +796,7 @@ static void vhost_net_flush(struct vhost_net *n)
                n->tx_flush = true;
                mutex_unlock(&n->vqs[VHOST_NET_VQ_TX].vq.mutex);
                /* Wait for all lower device DMAs done. */
-               vhost_ubuf_put_and_wait(n->vqs[VHOST_NET_VQ_TX].ubufs);
+               vhost_net_ubuf_put_and_wait(n->vqs[VHOST_NET_VQ_TX].ubufs);
                mutex_lock(&n->vqs[VHOST_NET_VQ_TX].vq.mutex);
                n->tx_flush = false;
                kref_init(&n->vqs[VHOST_NET_VQ_TX].ubufs->kref);
@@ -877,7 +897,7 @@ static long vhost_net_set_backend(struct vhost_net *n, unsigned index, int fd)
        struct socket *sock, *oldsock;
        struct vhost_virtqueue *vq;
        struct vhost_net_virtqueue *nvq;
-       struct vhost_ubuf_ref *ubufs, *oldubufs = NULL;
+       struct vhost_net_ubuf_ref *ubufs, *oldubufs = NULL;
        int r;
 
        mutex_lock(&n->dev.mutex);
@@ -908,7 +928,8 @@ static long vhost_net_set_backend(struct vhost_net *n, unsigned index, int fd)
        oldsock = rcu_dereference_protected(vq->private_data,
                                            lockdep_is_held(&vq->mutex));
        if (sock != oldsock) {
-               ubufs = vhost_ubuf_alloc(vq, sock && vhost_sock_zcopy(sock));
+               ubufs = vhost_net_ubuf_alloc(vq,
+                                            sock && vhost_sock_zcopy(sock));
                if (IS_ERR(ubufs)) {
                        r = PTR_ERR(ubufs);
                        goto err_ubufs;
@@ -934,7 +955,7 @@ static long vhost_net_set_backend(struct vhost_net *n, unsigned index, int fd)
        mutex_unlock(&vq->mutex);
 
        if (oldubufs) {
-               vhost_ubuf_put_and_wait(oldubufs);
+               vhost_net_ubuf_put_and_wait(oldubufs);
                mutex_lock(&vq->mutex);
                vhost_zerocopy_signal_used(n, vq);
                mutex_unlock(&vq->mutex);
@@ -952,7 +973,7 @@ err_used:
        rcu_assign_pointer(vq->private_data, oldsock);
        vhost_net_enable_vq(n, vq);
        if (ubufs)
-               vhost_ubuf_put_and_wait(ubufs);
+               vhost_net_ubuf_put_and_wait(ubufs);
 err_ubufs:
        fput(sock->file);
 err_vq:
@@ -1027,6 +1048,23 @@ static int vhost_net_set_features(struct vhost_net *n, u64 features)
        return 0;
 }
 
+static long vhost_net_set_owner(struct vhost_net *n)
+{
+       int r;
+
+       mutex_lock(&n->dev.mutex);
+       r = vhost_net_set_ubuf_info(n);
+       if (r)
+               goto out;
+       r = vhost_dev_set_owner(&n->dev);
+       if (r)
+               vhost_net_clear_ubuf_info(n);
+       vhost_net_flush(n);
+out:
+       mutex_unlock(&n->dev.mutex);
+       return r;
+}
+
 static long vhost_net_ioctl(struct file *f, unsigned int ioctl,
                            unsigned long arg)
 {
@@ -1055,19 +1093,15 @@ static long vhost_net_ioctl(struct file *f, unsigned int ioctl,
                return vhost_net_set_features(n, features);
        case VHOST_RESET_OWNER:
                return vhost_net_reset_owner(n);
+       case VHOST_SET_OWNER:
+               return vhost_net_set_owner(n);
        default:
                mutex_lock(&n->dev.mutex);
-               if (ioctl == VHOST_SET_OWNER) {
-                       r = vhost_net_set_ubuf_info(n);
-                       if (r)
-                               goto out;
-               }
                r = vhost_dev_ioctl(&n->dev, ioctl, argp);
                if (r == -ENOIOCTLCMD)
                        r = vhost_vring_ioctl(&n->dev, ioctl, argp);
                else
                        vhost_net_flush(n);
-out:
                mutex_unlock(&n->dev.mutex);
                return r;
        }
@@ -1101,7 +1135,7 @@ static struct miscdevice vhost_net_misc = {
 static int vhost_net_init(void)
 {
        if (experimental_zcopytx)
-               vhost_enable_zcopy(VHOST_NET_VQ_TX);
+               vhost_net_enable_zcopy(VHOST_NET_VQ_TX);
        return misc_register(&vhost_net_misc);
 }
 module_init(vhost_net_init);

diff --git a/drivers/vhost/scsi.c b/drivers/vhost/scsi.c
index 5179f7aa1b0bef8dee4dd813607bd2684ac29943..7014202972259f4d7046d5f4a110fe9a0cedef37 100644 (file)
--- a/drivers/vhost/scsi.c
+++ b/drivers/vhost/scsi.c
@@ -162,14 +162,8 @@ enum {
        VHOST_SCSI_VQ_IO = 2,
 };
 
-/*
- * VIRTIO_RING_F_EVENT_IDX seems broken. Not sure the bug is in
- * kernel but disabling it helps.
- * TODO: debug and remove the workaround.
- */
 enum {
-       VHOST_SCSI_FEATURES = (VHOST_FEATURES & (~VIRTIO_RING_F_EVENT_IDX)) |
-                             (1ULL << VIRTIO_SCSI_F_HOTPLUG)
+       VHOST_SCSI_FEATURES = VHOST_FEATURES | (1ULL << VIRTIO_SCSI_F_HOTPLUG)
 };
 
 #define VHOST_SCSI_MAX_TARGET  256

diff --git a/drivers/vhost/vhost.c b/drivers/vhost/vhost.c
index 749b5ab5bfbbd1f280b31cc341d9aae5cc64eef4..beee7f5787e6c814b2420f745be2555ca6a71af5 100644 (file)
--- a/drivers/vhost/vhost.c
+++ b/drivers/vhost/vhost.c
@@ -13,7 +13,7 @@
 
 #include <linux/eventfd.h>
 #include <linux/vhost.h>
-#include <linux/virtio_net.h>
+#include <linux/socket.h> /* memcpy_fromiovec */
 #include <linux/mm.h>
 #include <linux/mmu_context.h>
 #include <linux/miscdevice.h>
@@ -344,7 +344,7 @@ static int vhost_attach_cgroups(struct vhost_dev *dev)
 }
 
 /* Caller should have device mutex */
-static long vhost_dev_set_owner(struct vhost_dev *dev)
+long vhost_dev_set_owner(struct vhost_dev *dev)
 {
        struct task_struct *worker;
        int err;

diff --git a/drivers/vhost/vhost.h b/drivers/vhost/vhost.h
index b58f4ae82cb8e1e912f83c88eeec1e82eb950d5d..a7ad63592987f33ed4fd773691b7f5f604220d51 100644 (file)
--- a/drivers/vhost/vhost.h
+++ b/drivers/vhost/vhost.h
@@ -99,9 +99,6 @@ struct vhost_virtqueue {
        u64 log_addr;
 
        struct iovec iov[UIO_MAXIOV];
-       /* hdr is used to store the virtio header.
-        * Since each iovec has >= 1 byte length, we never need more than
-        * header length entries to store the header. */
        struct iovec *indirect;
        struct vring_used_elem *heads;
        /* We use a kind of RCU to access private pointer.
@@ -135,6 +132,7 @@ struct vhost_dev {
 };
 
 long vhost_dev_init(struct vhost_dev *, struct vhost_virtqueue **vqs, int nvqs);
+long vhost_dev_set_owner(struct vhost_dev *dev);
 long vhost_dev_check_owner(struct vhost_dev *);
 struct vhost_memory *vhost_dev_reset_owner_prepare(void);
 void vhost_dev_reset_owner(struct vhost_dev *, struct vhost_memory *);
@@ -177,9 +175,6 @@ enum {
                         (1ULL << VIRTIO_RING_F_INDIRECT_DESC) |
                         (1ULL << VIRTIO_RING_F_EVENT_IDX) |
                         (1ULL << VHOST_F_LOG_ALL),
-       VHOST_NET_FEATURES = VHOST_FEATURES |
-                        (1ULL << VHOST_NET_F_VIRTIO_NET_HDR) |
-                        (1ULL << VIRTIO_NET_F_MRG_RXBUF),
 };
 
 static inline int vhost_has_feature(struct vhost_dev *dev, int bit)
@@ -191,7 +186,4 @@ static inline int vhost_has_feature(struct vhost_dev *dev, int bit)
        acked_features = rcu_dereference_index_check(dev->acked_features, 1);
        return acked_features & (1 << bit);
 }
-
-void vhost_enable_zcopy(int vq);
-
 #endif