The core vhost-user protocol code should not do socket I/O, because the
details are transport-specific. Move code to send and receive
vhost-user messages into trans_af_unix.c.
The connection fd is a transport-specific feature. Therefore, it should
and eventually will be removed from the core vhost-user code. That is,
it will be removed from the vhost_user_msg_handler() and the message
handlers. We keep it for now, because vhost_user_set_mem_table() needs
it. In a later commit, we will refactor the map/unmap functionality and
after that we will be able to remove the connection fds from the core
vhost-user code.
Signed-off-by: Nikos Dragazis <ndraga...@arrikto.com>
Signed-off-by: Stefan Hajnoczi <stefa...@redhat.com>
---
lib/librte_vhost/trans_af_unix.c | 70 +++++++++++++++++++++++++++++++++---
lib/librte_vhost/vhost.h | 26 ++++++++++++++
lib/librte_vhost/vhost_user.c | 78 ++++++++--------------------------------
lib/librte_vhost/vhost_user.h | 7 +---
4 files changed, 108 insertions(+), 73 deletions(-)
diff --git a/lib/librte_vhost/trans_af_unix.c b/lib/librte_vhost/trans_af_unix.c
index 7e119b4..c0ba8df 100644
--- a/lib/librte_vhost/trans_af_unix.c
+++ b/lib/librte_vhost/trans_af_unix.c
@@ -50,7 +50,7 @@ static void vhost_user_read_cb(int connfd, void *dat, int
*remove);
* return bytes# of read on success or negative val on failure. Update fdnum
* with number of fds read.
*/
-int
+static int
read_fd_message(int sockfd, char *buf, int buflen, int *fds, int max_fds,
int *fd_num)
{
@@ -101,8 +101,8 @@ read_fd_message(int sockfd, char *buf, int buflen, int
*fds, int max_fds,
return ret;
}
-int
-send_fd_message(int sockfd, char *buf, int buflen, int *fds, int fd_num)
+static int
+send_fd_message(int sockfd, void *buf, int buflen, int *fds, int fd_num)
{
struct iovec iov;
struct msghdr msgh;
@@ -148,6 +148,23 @@ send_fd_message(int sockfd, char *buf, int buflen, int
*fds, int fd_num)
return ret;
}
+static int
+af_unix_send_reply(struct virtio_net *dev, struct VhostUserMsg *msg)
+{
+ struct vhost_user_connection *conn =
+ container_of(dev, struct vhost_user_connection, device);
+
+ return send_fd_message(conn->connfd, msg,
+ VHOST_USER_HDR_SIZE + msg->size, msg->fds,
msg->fd_num);
+}
+
+static int
+af_unix_send_slave_req(struct virtio_net *dev, struct VhostUserMsg *msg)
+{
+ return send_fd_message(dev->slave_req_fd, msg,
+ VHOST_USER_HDR_SIZE + msg->size, msg->fds,
msg->fd_num);
+}
+
static void
vhost_user_add_connection(int fd, struct vhost_user_socket *vsocket)
{
@@ -231,6 +248,36 @@ vhost_user_server_new_connection(int fd, void *dat, int
*remove __rte_unused)
vhost_user_add_connection(fd, vsocket);
}
+/* return bytes# of read on success or negative val on failure. */
+int
+read_vhost_message(int sockfd, struct VhostUserMsg *msg)
+{
+ int ret;
+
+ ret = read_fd_message(sockfd, (char *)msg, VHOST_USER_HDR_SIZE,
+ msg->fds, VHOST_MEMORY_MAX_NREGIONS, &msg->fd_num);
+ if (ret <= 0)
+ return ret;
+
+ if (msg->size) {
+ if (msg->size > sizeof(msg->payload)) {
+ RTE_LOG(ERR, VHOST_CONFIG,
+ "invalid msg size: %d\n", msg->size);
+ return -1;
+ }
+ ret = read(sockfd, &msg->payload, msg->size);
+ if (ret <= 0)
+ return ret;
+ if (ret != (int)msg->size) {
+ RTE_LOG(ERR, VHOST_CONFIG,
+ "read control message failed\n");
+ return -1;
+ }
+ }
+
+ return ret;
+}
+
static void
vhost_user_read_cb(int connfd, void *dat, int *remove)
{
@@ -238,10 +285,23 @@ vhost_user_read_cb(int connfd, void *dat, int *remove)
struct vhost_user_socket *vsocket = conn->vsocket;
struct af_unix_socket *af_vsocket =
container_of(vsocket, struct af_unix_socket, socket);
+ struct VhostUserMsg msg;
int ret;
- ret = vhost_user_msg_handler(conn->device.vid, connfd);
+ ret = read_vhost_message(connfd, &msg);
+ if (ret <= 0) {
+ if (ret < 0)
+ RTE_LOG(ERR, VHOST_CONFIG,
+ "vhost read message failed\n");
+ else if (ret == 0)
+ RTE_LOG(INFO, VHOST_CONFIG,
+ "vhost peer closed\n");
+ goto err;
+ }
+
+ ret = vhost_user_msg_handler(conn->device.vid, connfd, &msg);
if (ret < 0) {
+err:
close(connfd);
*remove = 1;
@@ -638,4 +698,6 @@ const struct vhost_transport_ops af_unix_trans_ops = {
.socket_cleanup = af_unix_socket_cleanup,
.socket_start = af_unix_socket_start,
.vring_call = af_unix_vring_call,
+ .send_reply = af_unix_send_reply,
+ .send_slave_req = af_unix_send_slave_req,
};
diff --git a/lib/librte_vhost/vhost.h b/lib/librte_vhost/vhost.h
index b9e4df1..b20773c 100644
--- a/lib/librte_vhost/vhost.h
+++ b/lib/librte_vhost/vhost.h
@@ -290,6 +290,7 @@ struct guest_page {
struct virtio_net;
struct vhost_user_socket;
+struct VhostUserMsg;
/**
* A structure containing function pointers for transport-specific operations.
@@ -351,6 +352,31 @@ struct vhost_transport_ops {
* 0 on success, -1 on failure
*/
int (*vring_call)(struct virtio_net *dev, struct vhost_virtqueue *vq);
+
+ /**
+ * Send a reply to the master.
+ *
+ * @param dev
+ * vhost device
+ * @param reply
+ * reply message
+ * @return
+ * 0 on success, -1 on failure
+ */
+ int (*send_reply)(struct virtio_net *dev, struct VhostUserMsg *reply);
+
+ /**
+ * Send a slave request to the master.
+ *
+ * @param dev
+ * vhost device
+ * @param req
+ * request message
+ * @return
+ * 0 on success, -1 on failure
+ */
+ int (*send_slave_req)(struct virtio_net *dev,
+ struct VhostUserMsg *req);
};
/** The traditional AF_UNIX vhost-user protocol transport. */
diff --git a/lib/librte_vhost/vhost_user.c b/lib/librte_vhost/vhost_user.c
index c9e29ec..5c12435 100644
--- a/lib/librte_vhost/vhost_user.c
+++ b/lib/librte_vhost/vhost_user.c
@@ -80,8 +80,8 @@ static const char *vhost_message_str[VHOST_USER_MAX] = {
[VHOST_USER_POSTCOPY_END] = "VHOST_USER_POSTCOPY_END",
};
-static int send_vhost_reply(int sockfd, struct VhostUserMsg *msg);
-static int read_vhost_message(int sockfd, struct VhostUserMsg *msg);
+static int send_vhost_reply(struct virtio_net *dev, struct VhostUserMsg *msg);
+int read_vhost_message(int sockfd, struct VhostUserMsg *msg);
static uint64_t
get_blk_size(int fd)
@@ -1042,7 +1042,7 @@ vhost_user_set_mem_table(struct virtio_net **pdev, struct
VhostUserMsg *msg,
if (dev->postcopy_listening) {
/* Send the addresses back to qemu */
msg->fd_num = 0;
- send_vhost_reply(main_fd, msg);
+ send_vhost_reply(dev, msg);
/* Wait for qemu to acknolwedge it's got the addresses
* we've got to wait before we're allowed to generate faults.
@@ -1764,49 +1764,8 @@ static vhost_message_handler_t
vhost_message_handlers[VHOST_USER_MAX] = {
[VHOST_USER_POSTCOPY_END] = vhost_user_postcopy_end,
};
-
-/* return bytes# of read on success or negative val on failure. */
static int
-read_vhost_message(int sockfd, struct VhostUserMsg *msg)
-{
- int ret;
-
- ret = read_fd_message(sockfd, (char *)msg, VHOST_USER_HDR_SIZE,
- msg->fds, VHOST_MEMORY_MAX_NREGIONS, &msg->fd_num);
- if (ret <= 0)
- return ret;
-
- if (msg->size) {
- if (msg->size > sizeof(msg->payload)) {
- RTE_LOG(ERR, VHOST_CONFIG,
- "invalid msg size: %d\n", msg->size);
- return -1;
- }
- ret = read(sockfd, &msg->payload, msg->size);
- if (ret <= 0)
- return ret;
- if (ret != (int)msg->size) {
- RTE_LOG(ERR, VHOST_CONFIG,
- "read control message failed\n");
- return -1;
- }
- }
-
- return ret;
-}
-
-static int
-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, msg->fds, msg->fd_num);
-}
-
-static int
-send_vhost_reply(int sockfd, struct VhostUserMsg *msg)
+send_vhost_reply(struct virtio_net *dev, struct VhostUserMsg *msg)
{
if (!msg)
return 0;
@@ -1816,7 +1775,7 @@ 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);
+ return dev->trans_ops->send_reply(dev, msg);
}
static int
@@ -1827,7 +1786,7 @@ send_vhost_slave_message(struct virtio_net *dev, struct
VhostUserMsg *msg)
if (msg->flags & VHOST_USER_NEED_REPLY)
rte_spinlock_lock(&dev->slave_req_lock);
- ret = send_vhost_message(dev->slave_req_fd, msg);
+ ret = dev->trans_ops->send_slave_req(dev, msg);
if (ret < 0 && (msg->flags & VHOST_USER_NEED_REPLY))
rte_spinlock_unlock(&dev->slave_req_lock);
@@ -1908,10 +1867,10 @@ vhost_user_unlock_all_queue_pairs(struct virtio_net
*dev)
}
int
-vhost_user_msg_handler(int vid, int fd)
+vhost_user_msg_handler(int vid, int fd, const struct VhostUserMsg *msg_)
{
+ struct VhostUserMsg msg = *msg_; /* copy so we can build the reply */
struct virtio_net *dev;
- struct VhostUserMsg msg;
struct rte_vdpa_device *vdpa_dev;
int did = -1;
int ret;
@@ -1933,15 +1892,8 @@ vhost_user_msg_handler(int vid, int fd)
}
}
- ret = read_vhost_message(fd, &msg);
- if (ret <= 0) {
- if (ret < 0)
- RTE_LOG(ERR, VHOST_CONFIG,
- "vhost read message failed\n");
- else
- RTE_LOG(INFO, VHOST_CONFIG,
- "vhost peer closed\n");
-
+ if (msg.request.master >= VHOST_USER_MAX) {
+ RTE_LOG(ERR, VHOST_CONFIG, "vhost read incorrect message\n");
return -1;
}
@@ -2004,7 +1956,7 @@ vhost_user_msg_handler(int vid, int fd)
(void *)&msg);
switch (ret) {
case RTE_VHOST_MSG_RESULT_REPLY:
- send_vhost_reply(fd, &msg);
+ send_vhost_reply(dev, &msg);
/* Fall-through */
case RTE_VHOST_MSG_RESULT_ERR:
case RTE_VHOST_MSG_RESULT_OK:
@@ -2038,7 +1990,7 @@ vhost_user_msg_handler(int vid, int fd)
RTE_LOG(DEBUG, VHOST_CONFIG,
"Processing %s succeeded and needs reply.\n",
vhost_message_str[request]);
- send_vhost_reply(fd, &msg);
+ send_vhost_reply(dev, &msg);
handled = true;
break;
default:
@@ -2053,7 +2005,7 @@ vhost_user_msg_handler(int vid, int fd)
(void *)&msg);
switch (ret) {
case RTE_VHOST_MSG_RESULT_REPLY:
- send_vhost_reply(fd, &msg);
+ send_vhost_reply(dev, &msg);
/* Fall-through */
case RTE_VHOST_MSG_RESULT_ERR:
case RTE_VHOST_MSG_RESULT_OK:
@@ -2083,7 +2035,7 @@ vhost_user_msg_handler(int vid, int fd)
msg.payload.u64 = ret == RTE_VHOST_MSG_RESULT_ERR;
msg.size = sizeof(msg.payload.u64);
msg.fd_num = 0;
- send_vhost_reply(fd, &msg);
+ send_vhost_reply(dev, &msg);
} else if (ret == RTE_VHOST_MSG_RESULT_ERR) {
RTE_LOG(ERR, VHOST_CONFIG,
"vhost message handling failed.\n");
@@ -2161,7 +2113,7 @@ vhost_user_iotlb_miss(struct virtio_net *dev, uint64_t
iova, uint8_t perm)
},
};
- ret = send_vhost_message(dev->slave_req_fd, &msg);
+ ret = send_vhost_slave_req(dev, &msg);
if (ret < 0) {
RTE_LOG(ERR, VHOST_CONFIG,
"Failed to send IOTLB miss message (%d)\n",
diff --git a/lib/librte_vhost/vhost_user.h b/lib/librte_vhost/vhost_user.h
index 2a650fe..0169bd2 100644
--- a/lib/librte_vhost/vhost_user.h
+++ b/lib/librte_vhost/vhost_user.h
@@ -146,12 +146,7 @@ typedef struct VhostUserMsg {
/* vhost_user.c */
-int vhost_user_msg_handler(int vid, int fd);
+int vhost_user_msg_handler(int vid, int fd, const struct VhostUserMsg *msg);
int vhost_user_iotlb_miss(struct virtio_net *dev, uint64_t iova, uint8_t perm);
-/* socket.c */
-int read_fd_message(int sockfd, char *buf, int buflen, int *fds, int max_fds,
- int *fd_num);
-int send_fd_message(int sockfd, char *buf, int buflen, int *fds, int fd_num);
-
#endif
--
2.7.4
