tap_v2: multiple improvements

 - add support for assigning tap interface to the bridge
 - add support for assigning tap interface host side ip4 and ip6 address
 - host namespace can be specified as PID (pid:12345) or full path to file
 - automatically bring linux interface up

Change-Id: I1cf7c3cad9a740e430cc1b9c2bb0aad0ba4cc8d8
Signed-off-by: Damjan Marion <damarion@cisco.com>

5 files changed, 168 insertions, 60 deletions

diff --git a/src/vnet/devices/virtio/cli.c b/src/vnet/devices/virtio/cli.c
index 0c1b75f8c4e..efd14355627 100644
--- a/src/vnet/devices/virtio/cli.c
+++ b/src/vnet/devices/virtio/cli.c
@@ -22,6 +22,9 @@
 #include <vlib/vlib.h>
 #include <vlib/unix/unix.h>
 #include <vnet/ethernet/ethernet.h>
+#include <vnet/ip/ip4_packet.h>
+#include <vnet/ip/ip6_packet.h>
+#include <vnet/ip/format.h>
 #include <linux/virtio_net.h>
 #include <linux/vhost.h>
 #include <vnet/devices/virtio/virtio.h>
@@ -32,8 +35,8 @@ tap_create_command_fn (vlib_main_t * vm, unformat_input_t * input,
 		       vlib_cli_command_t * cmd)
 {
   unformat_input_t _line_input, *line_input = &_line_input;
-  int rv;
   tap_create_if_args_t args = { 0 };
+  int ip_addr_set = 0;
 
   /* Get a line of input. */
   if (!unformat_user (input, unformat_line_input, line_input))
@@ -43,8 +46,18 @@ tap_create_command_fn (vlib_main_t * vm, unformat_input_t * input,
     {
       if (unformat (line_input, "name %s", &args.name))
 	;
-      else if (unformat (line_input, "host-ns %s", &args.net_ns))
+      else if (unformat (line_input, "host-ns %s", &args.host_namespace))
+	;
+      else if (unformat (line_input, "host-bridge %s", &args.host_bridge))
 	;
+      else if (unformat (line_input, "host-ip4-addr %U/%d",
+			 unformat_ip4_address, &args.host_ip4_addr,
+			 &args.host_ip4_prefix_len))
+	ip_addr_set = 1;
+      else if (unformat (line_input, "host-ip6-addr %U/%d",
+			 unformat_ip6_address, &args.host_ip6_addr,
+			 &args.host_ip6_prefix_len))
+	ip_addr_set = 1;
       else if (unformat (line_input, "rx-ring-size %d", &args.rx_ring_sz))
 	;
       else if (unformat (line_input, "tx-ring-size %d", &args.tx_ring_sz))
@@ -58,34 +71,27 @@ tap_create_command_fn (vlib_main_t * vm, unformat_input_t * input,
     }
   unformat_free (line_input);
 
-  rv = tap_create_if (vm, &args);
+  if (ip_addr_set && args.host_bridge)
+    return clib_error_return (0, "Please specify either host ip address or "
+			      "host bridge");
+
+  tap_create_if (vm, &args);
 
   vec_free (args.name);
+  vec_free (args.host_namespace);
+  vec_free (args.host_bridge);
 
-  if (rv == VNET_API_ERROR_SYSCALL_ERROR_1)
-    return clib_error_return_unix (0, "open '/dev/vhost-net'");
-  else if (rv == VNET_API_ERROR_SYSCALL_ERROR_2)
-    return clib_error_return_unix (0, "open '/dev/net/tun'");
-  else if (rv == VNET_API_ERROR_UNSUPPORTED)
-    return clib_error_return (0, "vhost-net backend doesn't support needed"
-			      " features");
-  else if (rv == VNET_API_ERROR_NAMESPACE_CREATE)
-    return clib_error_return (0, "failed to create netlink namespace");
-  else if (rv == VNET_API_ERROR_VIRTIO_INIT)
-    return clib_error_return (0, "failed to init virtio ring");
-  else if (rv == VNET_API_ERROR_INVALID_REGISTRATION)
-    return clib_error_return (0, "failed to register interface");
-  else if (rv != 0)
-    return clib_error_return (0, "error on creating tap interface");
+  return args.error;
 
-  return 0;
 }
 
 /* *INDENT-OFF* */
 VLIB_CLI_COMMAND (tap_create_command, static) = {
   .path = "create tap",
-  .short_help = "create tap {name <if-name>} [hw-addr <mac-address>]"
-    "[rx-ring-size <size>] [tx-ring-size <size>] [host-ns <netns>]",
+  .short_help = "create tap {name <if-name>} [hw-addr <mac-address>] "
+    "[rx-ring-size <size>] [tx-ring-size <size>] [host-ns <netns>] "
+    "[host-bridge <bridge-name>] [host-ip4-addr <ip4addr/mask>] "
+    "[host-ip6-addr <ip6-addr]",
   .function = tap_create_command_fn,
 };
 /* *INDENT-ON* */
diff --git a/src/vnet/devices/virtio/tap.c b/src/vnet/devices/virtio/tap.c
index 34339c95737..cc28df3be7f 100644
--- a/src/vnet/devices/virtio/tap.c
+++ b/src/vnet/devices/virtio/tap.c
@@ -31,6 +31,8 @@
 #include <vlib/vlib.h>
 #include <vlib/unix/unix.h>
 #include <vnet/ethernet/ethernet.h>
+#include <vnet/ip/ip4_packet.h>
+#include <vnet/ip/ip6_packet.h>
 #include <vnet/devices/netlink.h>
 #include <vnet/devices/virtio/virtio.h>
 #include <vnet/devices/virtio/tap.h>
@@ -47,23 +49,23 @@ virtio_eth_flag_change (vnet_main_t * vnm, vnet_hw_interface_t * hi,
 			u32 flags)
 {
   /* nothing for now */
+  //TODO On MTU change call vnet_netlink_set_if_mtu
   return 0;
 }
 
-int
+void
 tap_create_if (vlib_main_t * vm, tap_create_if_args_t * args)
 {
   vnet_main_t *vnm = vnet_get_main ();
   virtio_main_t *vim = &virtio_main;
   vnet_sw_interface_t *sw;
   vnet_hw_interface_t *hw;
-  int i;
-  clib_error_t *err = 0;
+  int i, fd;
   struct ifreq ifr;
   size_t hdrsz;
   struct vhost_memory *vhost_mem = 0;
   virtio_if_t *vif = 0;
-  int rv = 0;
+  clib_error_t *err = 0;
 
   memset (&ifr, 0, sizeof (ifr));
   pool_get (vim->interfaces, vif);
@@ -72,7 +74,8 @@ tap_create_if (vlib_main_t * vm, tap_create_if_args_t * args)
 
   if ((vif->fd = open ("/dev/vhost-net", O_RDWR | O_NONBLOCK)) < 0)
     {
-      rv = VNET_API_ERROR_SYSCALL_ERROR_1;
+      args->rv = VNET_API_ERROR_SYSCALL_ERROR_1;
+      args->error = clib_error_return_unix (0, "open '/dev/vhost-net'");
       goto error;
     }
 
@@ -80,19 +83,25 @@ tap_create_if (vlib_main_t * vm, tap_create_if_args_t * args)
 
   if ((vif->remote_features & (1ULL << VIRTIO_NET_F_MRG_RXBUF)) == 0)
     {
-      rv = VNET_API_ERROR_UNSUPPORTED;
+      args->rv = VNET_API_ERROR_UNSUPPORTED;
+      args->error = clib_error_return (0, "vhost-net backend doesn't support "
+				       "VIRTIO_NET_F_MRG_RXBUF feature");
       goto error;
     }
 
   if ((vif->remote_features & (1ULL << VIRTIO_RING_F_INDIRECT_DESC)) == 0)
     {
-      rv = VNET_API_ERROR_UNSUPPORTED;
+      args->rv = VNET_API_ERROR_UNSUPPORTED;
+      args->error = clib_error_return (0, "vhost-net backend doesn't support "
+				       "VIRTIO_RING_F_INDIRECT_DESC feature");
       goto error;
     }
 
   if ((vif->remote_features & (1ULL << VIRTIO_F_VERSION_1)) == 0)
     {
-      rv = VNET_API_ERROR_UNSUPPORTED;
+      args->rv = VNET_API_ERROR_UNSUPPORTED;
+      args->error = clib_error_return (0, "vhost-net backend doesn't support "
+				       "VIRTIO_F_VERSION_1 features");
       goto error;
     }
 
@@ -104,7 +113,8 @@ tap_create_if (vlib_main_t * vm, tap_create_if_args_t * args)
 
   if ((vif->tap_fd = open ("/dev/net/tun", O_RDWR | O_NONBLOCK)) < 0)
     {
-      rv = VNET_API_ERROR_SYSCALL_ERROR_2;
+      args->rv = VNET_API_ERROR_SYSCALL_ERROR_2;
+      args->error = clib_error_return_unix (0, "open '/dev/net/tun'");
       goto error;
     }
 
@@ -120,13 +130,49 @@ tap_create_if (vlib_main_t * vm, tap_create_if_args_t * args)
   _IOCTL (vif->tap_fd, TUNSETVNETHDRSZ, &hdrsz);
   _IOCTL (vif->fd, VHOST_SET_OWNER, 0);
 
-  if (args->net_ns)
+  if (args->host_bridge)
     {
-      err = vnet_netlink_set_if_namespace (vif->ifindex,
-					   (char *) args->net_ns);
-      if (err)
+      int master_ifindex = if_nametoindex ((char *) args->host_bridge);
+      args->error = vnet_netlink_set_if_master (vif->ifindex, master_ifindex);
+      if (args->error)
 	{
-	  rv = VNET_API_ERROR_NAMESPACE_CREATE;
+	  args->rv = VNET_API_ERROR_NETLINK_ERROR;
+	  goto error;
+	}
+    }
+
+  if (args->host_namespace)
+    {
+      args->error = vnet_netlink_set_if_namespace (vif->ifindex,
+						   (char *)
+						   args->host_namespace);
+      if (args->error)
+	{
+	  args->rv = VNET_API_ERROR_NETLINK_ERROR;
+	  goto error;
+	}
+    }
+
+  if (args->host_ip4_prefix_len)
+    {
+      args->error = vnet_netlink_add_ip4_addr (vif->ifindex,
+					       &args->host_ip4_addr,
+					       args->host_ip4_prefix_len);
+      if (args->error)
+	{
+	  args->rv = VNET_API_ERROR_NETLINK_ERROR;
+	  goto error;
+	}
+    }
+
+  if (args->host_ip6_prefix_len)
+    {
+      args->error = vnet_netlink_add_ip6_addr (vif->ifindex,
+					       &args->host_ip6_addr,
+					       args->host_ip6_prefix_len);
+      if (args->error)
+	{
+	  args->rv = VNET_API_ERROR_NETLINK_ERROR;
 	  goto error;
 	}
     }
@@ -139,18 +185,29 @@ tap_create_if (vlib_main_t * vm, tap_create_if_args_t * args)
   vhost_mem->regions[0].memory_size = (1ULL << 47) - 4096;
   _IOCTL (vif->fd, VHOST_SET_MEM_TABLE, vhost_mem);
 
-  if ((err = virtio_vring_init (vm, vif, 0, args->rx_ring_sz)))
+  if ((args->error = virtio_vring_init (vm, vif, 0, args->rx_ring_sz)))
     {
-      rv = VNET_API_ERROR_VIRTIO_INIT;
+      args->rv = VNET_API_ERROR_INIT_FAILED;
       goto error;
     }
 
-  if ((err = virtio_vring_init (vm, vif, 1, args->tx_ring_sz)))
+  if ((args->error = virtio_vring_init (vm, vif, 1, args->tx_ring_sz)))
     {
-      rv = VNET_API_ERROR_VIRTIO_INIT;
+      args->rv = VNET_API_ERROR_INIT_FAILED;
       goto error;
     }
 
+  /* set host side up */
+  if ((fd = socket (AF_INET, SOCK_STREAM, 0)) > 0)
+    {
+      memset (&ifr, 0, sizeof (struct ifreq));
+      strncpy (ifr.ifr_name, (char *) args->name, sizeof (ifr.ifr_name) - 1);
+      _IOCTL (fd, SIOCGIFFLAGS, (void *) &ifr);
+      ifr.ifr_flags |= IFF_UP | IFF_RUNNING;
+      _IOCTL (fd, SIOCSIFFLAGS, (void *) &ifr);
+      close (fd);
+    }
+
   if (!args->hw_addr_set)
     {
       f64 now = vlib_time_now (vm);
@@ -164,14 +221,17 @@ tap_create_if (vlib_main_t * vm, tap_create_if_args_t * args)
     }
   vif->name = args->name;
   args->name = 0;
-  vif->net_ns = args->net_ns;
-  args->net_ns = 0;
-  err = ethernet_register_interface (vnm, virtio_device_class.index,
-				     vif->dev_instance, args->hw_addr,
-				     &vif->hw_if_index,
-				     virtio_eth_flag_change);
-  if (err)
-    rv = VNET_API_ERROR_INVALID_REGISTRATION;
+  vif->net_ns = args->host_namespace;
+  args->host_namespace = 0;
+  args->error = ethernet_register_interface (vnm, virtio_device_class.index,
+					     vif->dev_instance, args->hw_addr,
+					     &vif->hw_if_index,
+					     virtio_eth_flag_change);
+  if (args->error)
+    {
+      args->rv = VNET_API_ERROR_INVALID_REGISTRATION;
+      goto error;
+    }
 
   sw = vnet_get_hw_sw_interface (vnm, vif->hw_if_index);
   vif->sw_if_index = sw->sw_if_index;
@@ -191,6 +251,12 @@ tap_create_if (vlib_main_t * vm, tap_create_if_args_t * args)
   goto done;
 
 error:
+  if (err)
+    {
+      ASSERT (args->error == 0);
+      args->error = err;
+      args->rv = VNET_API_ERROR_SYSCALL_ERROR_3;
+    }
   if (vif->tap_fd != -1)
     close (vif->tap_fd);
   if (vif->fd != -1)
@@ -202,8 +268,6 @@ error:
 done:
   if (vhost_mem)
     clib_mem_free (vhost_mem);
-
-  return rv;
 }
 
 int
diff --git a/src/vnet/devices/virtio/tap.h b/src/vnet/devices/virtio/tap.h
index 58dcb5bda51..0e0f8cb4056 100644
--- a/src/vnet/devices/virtio/tap.h
+++ b/src/vnet/devices/virtio/tap.h
@@ -25,13 +25,20 @@
 typedef struct
 {
   u8 *name;
-  u8 *net_ns;
   u8 hw_addr_set;
   u8 hw_addr[6];
   u16 rx_ring_sz;
   u16 tx_ring_sz;
+  u8 *host_namespace;
+  u8 *host_bridge;
+  ip4_address_t host_ip4_addr;
+  u32 host_ip4_prefix_len;
+  ip6_address_t host_ip6_addr;
+  u32 host_ip6_prefix_len;
   /* return */
   u32 sw_if_index;
+  int rv;
+  clib_error_t *error;
 } tap_create_if_args_t;
 
 /** TAP interface details struct */
@@ -41,7 +48,7 @@ typedef struct
   u8 dev_name[64];
 } tap_interface_details_t;
 
-int tap_create_if (vlib_main_t * vm, tap_create_if_args_t * args);
+void tap_create_if (vlib_main_t * vm, tap_create_if_args_t * args);
 int tap_delete_if (vlib_main_t * vm, u32 sw_if_index);
 int tap_dump_ifs (tap_interface_details_t ** out_tapids);
 
diff --git a/src/vnet/devices/virtio/tapv2.api b/src/vnet/devices/virtio/tapv2.api
index e1592cf7d61..03788607fe5 100644
--- a/src/vnet/devices/virtio/tapv2.api
+++ b/src/vnet/devices/virtio/tapv2.api
@@ -27,10 +27,18 @@ vl_api_version 1.0.0
     @param use_random_mac - let the system generate a unique mac address
     @param tap_name - name to associate with the new interface
     @param mac_address - mac addr to assign to the interface if use_radom not set
-    @param net_ns_set - net_ns is entered
-    @param net_ns - netlink name space
     @param tx_ring_sz - the number of entries of TX ring
     @param rx_ring_sz - the number of entries of RX ring
+    @param host_namespace_set - host namespece should be set
+    @param host_namespace - host namespace to attach interface to
+    @param host_bridge_set - host bridge should be set
+    @param host_bridge - host bridge to attach interface to
+    @param host_ip4_addr_set - host IPv4 ip address should be set
+    @param host_ip4_addr - host IPv4 ip address
+    @param host_ip4_prefix_len - host IPv4 ip address prefix length
+    @param host_ip6_addr_set - host IPv6 ip address should be set
+    @param host_ip6_addr - host IPv6 ip address
+    @param host_ip6_prefix_len - host IPv6 ip address prefix length
 */
 define tap_create_v2
 {
@@ -39,10 +47,18 @@ define tap_create_v2
   u8 use_random_mac;
   u8 tap_name[64];
   u8 mac_address[6];
-  u8 net_ns_set;
-  u8 net_ns[64];
   u16 tx_ring_sz; /* optional, default is 256 entries, must be power of 2 */
   u16 rx_ring_sz; /* optional, default is 256 entries, must be power of 2 */
+  u8 host_namespace_set;
+  u8 host_namespace[64];
+  u8 host_bridge_set;
+  u8 host_bridge[64];
+  u8 host_ip4_addr_set;
+  u8 host_ip4_addr[4];
+  u8 host_ip4_prefix_len;
+  u8 host_ip6_addr_set;
+  u8 host_ip6_addr[16];
+  u8 host_ip6_prefix_len;
 };
 
 /** \brief Reply for tap create reply
diff --git a/src/vnet/devices/virtio/tapv2_api.c b/src/vnet/devices/virtio/tapv2_api.c
index 1c559e6c28e..7a6adca083f 100644
--- a/src/vnet/devices/virtio/tapv2_api.c
+++ b/src/vnet/devices/virtio/tapv2_api.c
@@ -53,7 +53,6 @@ static void
 vl_api_tap_create_v2_t_handler (vl_api_tap_create_v2_t * mp)
 {
   vlib_main_t *vm = vlib_get_main ();
-  int rv;
   vl_api_tap_create_v2_reply_t *rmp;
   unix_shared_memory_queue_t *q;
   tap_create_if_args_t _a, *ap = &_a;
@@ -69,10 +68,26 @@ vl_api_tap_create_v2_t_handler (vl_api_tap_create_v2_t * mp)
   ap->rx_ring_sz = mp->rx_ring_sz;
   ap->tx_ring_sz = mp->tx_ring_sz;
   ap->sw_if_index = (u32) ~ 0;
-  if (mp->net_ns_set)
-    ap->net_ns = mp->net_ns;
 
-  rv = tap_create_if (vm, ap);
+  if (mp->host_namespace_set)
+    ap->host_namespace = mp->host_namespace;
+
+  if (mp->host_bridge_set)
+    ap->host_bridge = mp->host_bridge;
+
+  if (mp->host_ip4_addr_set)
+    {
+      clib_memcpy (&ap->host_ip4_addr.as_u8, mp->host_ip4_addr, 4);
+      ap->host_ip4_prefix_len = mp->host_ip4_prefix_len;
+    }
+
+  if (mp->host_ip6_addr_set)
+    {
+      clib_memcpy (&ap->host_ip6_addr, mp->host_ip6_addr, 16);
+      ap->host_ip6_prefix_len = mp->host_ip6_prefix_len;
+    }
+
+  tap_create_if (vm, ap);
 
   q = vl_api_client_index_to_input_queue (mp->client_index);
   if (!q)
@@ -81,7 +96,7 @@ vl_api_tap_create_v2_t_handler (vl_api_tap_create_v2_t * mp)
   rmp = vl_msg_api_alloc (sizeof (*rmp));
   rmp->_vl_msg_id = ntohs (VL_API_TAP_CREATE_V2_REPLY);
   rmp->context = mp->context;
-  rmp->retval = ntohl (rv);
+  rmp->retval = ntohl (ap->rv);
   rmp->sw_if_index = ntohl (ap->sw_if_index);
 
   vl_msg_api_send_shmem (q, (u8 *) & rmp);