vhost: access VhostUsrMsg via packed struct
authorDaniel Verkamp <daniel.verkamp@intel.com>
Fri, 26 May 2017 11:59:15 +0000 (13:59 +0200)
committerYuanhan Liu <yliu@fridaylinux.org>
Fri, 16 Jun 2017 12:04:25 +0000 (14:04 +0200)
Accessing fields of a packed struct through unaligned pointers is
undefined behavior. Instead of passing pointers to particular fields,
a pointer to the root struct should be used. This patch does exactly
that.

Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Acked-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
lib/librte_vhost/vhost_user.c

index 92db73a..49832d6 100644 (file)
@@ -197,11 +197,11 @@ vhost_user_set_features(struct virtio_net *dev, uint64_t features)
  */
 static int
 vhost_user_set_vring_num(struct virtio_net *dev,
-                        struct vhost_vring_state *state)
+                        VhostUserMsg *msg)
 {
-       struct vhost_virtqueue *vq = dev->virtqueue[state->index];
+       struct vhost_virtqueue *vq = dev->virtqueue[msg->payload.state.index];
 
-       vq->size = state->num;
+       vq->size = msg->payload.state.num;
 
        if (dev->dequeue_zero_copy) {
                vq->nr_zmbuf = 0;
@@ -334,7 +334,7 @@ qva_to_vva(struct virtio_net *dev, uint64_t qva)
  * This function then converts these to our address space.
  */
 static int
-vhost_user_set_vring_addr(struct virtio_net *dev, struct vhost_vring_addr *addr)
+vhost_user_set_vring_addr(struct virtio_net *dev, VhostUserMsg *msg)
 {
        struct vhost_virtqueue *vq;
 
@@ -342,11 +342,11 @@ vhost_user_set_vring_addr(struct virtio_net *dev, struct vhost_vring_addr *addr)
                return -1;
 
        /* addr->index refers to the queue index. The txq 1, rxq is 0. */
-       vq = dev->virtqueue[addr->index];
+       vq = dev->virtqueue[msg->payload.addr.index];
 
        /* The addresses are converted from QEMU virtual to Vhost virtual. */
        vq->desc = (struct vring_desc *)(uintptr_t)qva_to_vva(dev,
-                       addr->desc_user_addr);
+                       msg->payload.addr.desc_user_addr);
        if (vq->desc == 0) {
                RTE_LOG(ERR, VHOST_CONFIG,
                        "(%d) failed to find desc ring address.\n",
@@ -354,11 +354,11 @@ vhost_user_set_vring_addr(struct virtio_net *dev, struct vhost_vring_addr *addr)
                return -1;
        }
 
-       dev = numa_realloc(dev, addr->index);
-       vq = dev->virtqueue[addr->index];
+       dev = numa_realloc(dev, msg->payload.addr.index);
+       vq = dev->virtqueue[msg->payload.addr.index];
 
        vq->avail = (struct vring_avail *)(uintptr_t)qva_to_vva(dev,
-                       addr->avail_user_addr);
+                       msg->payload.addr.avail_user_addr);
        if (vq->avail == 0) {
                RTE_LOG(ERR, VHOST_CONFIG,
                        "(%d) failed to find avail ring address.\n",
@@ -367,7 +367,7 @@ vhost_user_set_vring_addr(struct virtio_net *dev, struct vhost_vring_addr *addr)
        }
 
        vq->used = (struct vring_used *)(uintptr_t)qva_to_vva(dev,
-                       addr->used_user_addr);
+                       msg->payload.addr.used_user_addr);
        if (vq->used == 0) {
                RTE_LOG(ERR, VHOST_CONFIG,
                        "(%d) failed to find used ring address.\n",
@@ -384,7 +384,7 @@ vhost_user_set_vring_addr(struct virtio_net *dev, struct vhost_vring_addr *addr)
                vq->last_avail_idx = vq->used->idx;
        }
 
-       vq->log_guest_addr = addr->log_guest_addr;
+       vq->log_guest_addr = msg->payload.addr.log_guest_addr;
 
        LOG_DEBUG(VHOST_CONFIG, "(%d) mapped address desc: %p\n",
                        dev->vid, vq->desc);
@@ -403,10 +403,12 @@ vhost_user_set_vring_addr(struct virtio_net *dev, struct vhost_vring_addr *addr)
  */
 static int
 vhost_user_set_vring_base(struct virtio_net *dev,
-                         struct vhost_vring_state *state)
+                         VhostUserMsg *msg)
 {
-       dev->virtqueue[state->index]->last_used_idx  = state->num;
-       dev->virtqueue[state->index]->last_avail_idx = state->num;
+       dev->virtqueue[msg->payload.state.index]->last_used_idx  =
+                       msg->payload.state.num;
+       dev->virtqueue[msg->payload.state.index]->last_avail_idx =
+                       msg->payload.state.num;
 
        return 0;
 }
@@ -704,9 +706,9 @@ free_zmbufs(struct vhost_virtqueue *vq)
  */
 static int
 vhost_user_get_vring_base(struct virtio_net *dev,
-                         struct vhost_vring_state *state)
+                         VhostUserMsg *msg)
 {
-       struct vhost_virtqueue *vq = dev->virtqueue[state->index];
+       struct vhost_virtqueue *vq = dev->virtqueue[msg->payload.state.index];
 
        /* We have to stop the queue (virtio) if it is running. */
        if (dev->flags & VIRTIO_DEV_RUNNING) {
@@ -717,10 +719,11 @@ vhost_user_get_vring_base(struct virtio_net *dev,
        dev->flags &= ~VIRTIO_DEV_READY;
 
        /* Here we are safe to get the last used index */
-       state->num = vq->last_used_idx;
+       msg->payload.state.num = vq->last_used_idx;
 
        RTE_LOG(INFO, VHOST_CONFIG,
-               "vring base idx:%d file:%d\n", state->index, state->num);
+               "vring base idx:%d file:%d\n", msg->payload.state.index,
+               msg->payload.state.num);
        /*
         * Based on current qemu vhost-user implementation, this message is
         * sent and only sent in vhost_vring_stop.
@@ -745,18 +748,19 @@ vhost_user_get_vring_base(struct virtio_net *dev,
  */
 static int
 vhost_user_set_vring_enable(struct virtio_net *dev,
-                           struct vhost_vring_state *state)
+                           VhostUserMsg *msg)
 {
-       int enable = (int)state->num;
+       int enable = (int)msg->payload.state.num;
 
        RTE_LOG(INFO, VHOST_CONFIG,
                "set queue enable: %d to qp idx: %d\n",
-               enable, state->index);
+               enable, msg->payload.state.index);
 
        if (dev->notify_ops->vring_state_changed)
-               dev->notify_ops->vring_state_changed(dev->vid, state->index, enable);
+               dev->notify_ops->vring_state_changed(dev->vid,
+                               msg->payload.state.index, enable);
 
-       dev->virtqueue[state->index]->enabled = enable;
+       dev->virtqueue[msg->payload.state.index]->enabled = enable;
 
        return 0;
 }
@@ -1045,17 +1049,17 @@ vhost_user_msg_handler(int vid, int fd)
                break;
 
        case VHOST_USER_SET_VRING_NUM:
-               vhost_user_set_vring_num(dev, &msg.payload.state);
+               vhost_user_set_vring_num(dev, &msg);
                break;
        case VHOST_USER_SET_VRING_ADDR:
-               vhost_user_set_vring_addr(dev, &msg.payload.addr);
+               vhost_user_set_vring_addr(dev, &msg);
                break;
        case VHOST_USER_SET_VRING_BASE:
-               vhost_user_set_vring_base(dev, &msg.payload.state);
+               vhost_user_set_vring_base(dev, &msg);
                break;
 
        case VHOST_USER_GET_VRING_BASE:
-               vhost_user_get_vring_base(dev, &msg.payload.state);
+               vhost_user_get_vring_base(dev, &msg);
                msg.size = sizeof(msg.payload.state);
                send_vhost_message(fd, &msg);
                break;
@@ -1080,7 +1084,7 @@ vhost_user_msg_handler(int vid, int fd)
                break;
 
        case VHOST_USER_SET_VRING_ENABLE:
-               vhost_user_set_vring_enable(dev, &msg.payload.state);
+               vhost_user_set_vring_enable(dev, &msg);
                break;
        case VHOST_USER_SEND_RARP:
                vhost_user_send_rarp(dev, &msg);