vhost: enable fds passing in vhost-user messages
authorMaxime Coquelin <maxime.coquelin@redhat.com>
Fri, 12 Oct 2018 12:40:37 +0000 (14:40 +0200)
committerFerruh Yigit <ferruh.yigit@intel.com>
Thu, 18 Oct 2018 08:24:39 +0000 (10:24 +0200)
Passing userfault fds to Qemu will be required for postcopy
live-migration feature.

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>
lib/librte_vhost/vhost_crypto.c
lib/librte_vhost/vhost_user.c

index c592b7a..9811a23 100644 (file)
@@ -447,6 +447,7 @@ vhost_crypto_msg_post_handler(int vid, void *msg)
        if (vmsg->request.master == VHOST_USER_CRYPTO_CREATE_SESS) {
                vhost_crypto_create_sess(vcrypto,
                                &vmsg->payload.crypto_session);
+               vmsg->fd_num = 0;
                ret = VH_RESULT_REPLY;
        } else if (vmsg->request.master == VHOST_USER_CRYPTO_CLOSE_SESS) {
                if (vhost_crypto_close_sess(vcrypto, vmsg->payload.u64))
index bce2395..99adced 100644 (file)
@@ -161,6 +161,7 @@ vhost_user_get_features(struct virtio_net **pdev, struct VhostUserMsg *msg,
 
        msg->payload.u64 = features;
        msg->size = sizeof(msg->payload.u64);
+       msg->fd_num = 0;
 
        return VH_RESULT_REPLY;
 }
@@ -179,6 +180,7 @@ vhost_user_get_queue_num(struct virtio_net **pdev, struct VhostUserMsg *msg,
 
        msg->payload.u64 = (uint64_t)queue_num;
        msg->size = sizeof(msg->payload.u64);
+       msg->fd_num = 0;
 
        return VH_RESULT_REPLY;
 }
@@ -1165,6 +1167,7 @@ vhost_user_get_vring_base(struct virtio_net **pdev,
        vq->batch_copy_elems = NULL;
 
        msg->size = sizeof(msg->payload.state);
+       msg->fd_num = 0;
 
        return VH_RESULT_REPLY;
 }
@@ -1224,6 +1227,7 @@ vhost_user_get_protocol_features(struct virtio_net **pdev,
 
        msg->payload.u64 = protocol_features;
        msg->size = sizeof(msg->payload.u64);
+       msg->fd_num = 0;
 
        return VH_RESULT_REPLY;
 }
@@ -1310,6 +1314,7 @@ vhost_user_set_log_base(struct virtio_net **pdev, struct VhostUserMsg *msg,
         * any payload in the reply.
         */
        msg->size = 0;
+       msg->fd_num = 0;
 
        return VH_RESULT_REPLY;
 }
@@ -1556,13 +1561,13 @@ read_vhost_message(int sockfd, struct VhostUserMsg *msg)
 }
 
 static int
-send_vhost_message(int sockfd, struct VhostUserMsg *msg, int *fds, int fd_num)
+send_vhost_message(int sockfd, struct VhostUserMsg *msg)
 {
        if (!msg)
                return 0;
 
        return send_fd_message(sockfd, (char *)msg,
-               VHOST_USER_HDR_SIZE + msg->size, fds, fd_num);
+               VHOST_USER_HDR_SIZE + msg->size, msg->fds, msg->fd_num);
 }
 
 static int
@@ -1576,19 +1581,18 @@ send_vhost_reply(int sockfd, struct VhostUserMsg *msg)
        msg->flags |= VHOST_USER_VERSION;
        msg->flags |= VHOST_USER_REPLY_MASK;
 
-       return send_vhost_message(sockfd, msg, NULL, 0);
+       return send_vhost_message(sockfd, msg);
 }
 
 static int
-send_vhost_slave_message(struct virtio_net *dev, struct VhostUserMsg *msg,
-                        int *fds, int fd_num)
+send_vhost_slave_message(struct virtio_net *dev, struct VhostUserMsg *msg)
 {
        int ret;
 
        if (msg->flags & VHOST_USER_NEED_REPLY)
                rte_spinlock_lock(&dev->slave_req_lock);
 
-       ret = send_vhost_message(dev->slave_req_fd, msg, fds, fd_num);
+       ret = send_vhost_message(dev->slave_req_fd, msg);
        if (ret < 0 && (msg->flags & VHOST_USER_NEED_REPLY))
                rte_spinlock_unlock(&dev->slave_req_lock);
 
@@ -1820,6 +1824,7 @@ skip_to_reply:
        if (msg.flags & VHOST_USER_NEED_REPLY) {
                msg.payload.u64 = ret == VH_RESULT_ERR;
                msg.size = sizeof(msg.payload.u64);
+               msg.fd_num = 0;
                send_vhost_reply(fd, &msg);
        } else if (ret == VH_RESULT_ERR) {
                RTE_LOG(ERR, VHOST_CONFIG,
@@ -1903,7 +1908,7 @@ vhost_user_iotlb_miss(struct virtio_net *dev, uint64_t iova, uint8_t perm)
                },
        };
 
-       ret = send_vhost_message(dev->slave_req_fd, &msg, NULL, 0);
+       ret = send_vhost_message(dev->slave_req_fd, &msg);
        if (ret < 0) {
                RTE_LOG(ERR, VHOST_CONFIG,
                                "Failed to send IOTLB miss message (%d)\n",
@@ -1919,8 +1924,6 @@ static int vhost_user_slave_set_vring_host_notifier(struct virtio_net *dev,
                                                    uint64_t offset,
                                                    uint64_t size)
 {
-       int *fdp = NULL;
-       size_t fd_num = 0;
        int ret;
        struct VhostUserMsg msg = {
                .request.slave = VHOST_USER_SLAVE_VRING_HOST_NOTIFIER_MSG,
@@ -1936,11 +1939,11 @@ static int vhost_user_slave_set_vring_host_notifier(struct virtio_net *dev,
        if (fd < 0)
                msg.payload.area.u64 |= VHOST_USER_VRING_NOFD_MASK;
        else {
-               fdp = &fd;
-               fd_num = 1;
+               msg.fds[0] = fd;
+               msg.fd_num = 1;
        }
 
-       ret = send_vhost_slave_message(dev, &msg, fdp, fd_num);
+       ret = send_vhost_slave_message(dev, &msg);
        if (ret < 0) {
                RTE_LOG(ERR, VHOST_CONFIG,
                        "Failed to set host notifier (%d)\n", ret);