net/tap: fix protocol field
authorVipin Varghese <vipin.varghese@intel.com>
Tue, 22 May 2018 10:47:14 +0000 (16:17 +0530)
committerFerruh Yigit <ferruh.yigit@intel.com>
Tue, 22 May 2018 22:35:01 +0000 (00:35 +0200)
The TX function is shared between TAP and TUN PMD. Checking TUN-TAP
type field will ensure the TAP PMD will always have protocol field
as 0.

Fixes: 204d026a3922 ("net/tap: support tun")

Signed-off-by: Vipin Varghese <vipin.varghese@intel.com>
Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
drivers/net/tap/rte_eth_tap.c
drivers/net/tap/rte_eth_tap.h

index 01552fa..6bb7962 100644 (file)
@@ -68,7 +68,6 @@ static const char *valid_arguments[] = {
 static int tap_unit;
 static unsigned int tun_unit;
 
-static int tap_type;
 static char tuntap_name[8];
 
 static volatile uint32_t tap_trigger;  /* Rx trigger */
@@ -116,7 +115,8 @@ tun_alloc(struct pmd_internals *pmd)
         * Do not set IFF_NO_PI as packet information header will be needed
         * to check if a received packet has been truncated.
         */
-       ifr.ifr_flags = (tap_type) ? IFF_TAP : IFF_TUN | IFF_POINTOPOINT;
+       ifr.ifr_flags = (pmd->type == ETH_TUNTAP_TYPE_TAP) ?
+               IFF_TAP : IFF_TUN | IFF_POINTOPOINT;
        snprintf(ifr.ifr_name, IFNAMSIZ, "%s", pmd->name);
 
        TAP_LOG(DEBUG, "ifr_name '%s'", ifr.ifr_name);
@@ -472,20 +472,22 @@ pmd_tx_burst(void *queue, struct rte_mbuf **bufs, uint16_t nb_pkts)
                if (rte_pktmbuf_pkt_len(mbuf) > max_size)
                        break;
 
-               /*
-                * TUN and TAP are created with IFF_NO_PI disabled.
-                * For TUN PMD this mandatory as fields are used by
-                * Kernel tun.c to determine whether its IP or non IP
-                * packets.
-                *
-                * The logic fetches the first byte of data from mbuf.
-                * compares whether its v4 or v6. If none matches default
-                * value 0x00 is taken for protocol field.
-                */
-               char *buff_data = rte_pktmbuf_mtod(seg, void *);
-               j = (*buff_data & 0xf0);
-               pi.proto = (j == 0x40) ? 0x0008 :
-                               (j == 0x60) ? 0xdd86 : 0x00;
+               if (txq->type == ETH_TUNTAP_TYPE_TUN) {
+                       /*
+                        * TUN and TAP are created with IFF_NO_PI disabled.
+                        * For TUN PMD this mandatory as fields are used by
+                        * Kernel tun.c to determine whether its IP or non IP
+                        * packets.
+                        *
+                        * The logic fetches the first byte of data from mbuf
+                        * then compares whether its v4 or v6. If first byte
+                        * is 4 or 6, then protocol field is updated.
+                        */
+                       char *buff_data = rte_pktmbuf_mtod(seg, void *);
+                       j = (*buff_data & 0xf0);
+                       pi.proto = (j == 0x40) ? rte_cpu_to_be_16(ETHER_TYPE_IPv4) :
+                               (j == 0x60) ? rte_cpu_to_be_16(ETHER_TYPE_IPv6) : 0x00;
+               }
 
                iovecs[0].iov_base = &pi;
                iovecs[0].iov_len = sizeof(pi);
@@ -928,6 +930,12 @@ tap_mac_set(struct rte_eth_dev *dev, struct ether_addr *mac_addr)
        struct ifreq ifr;
        int ret;
 
+       if (pmd->type == ETH_TUNTAP_TYPE_TUN) {
+               TAP_LOG(ERR, "%s: can't MAC address for TUN",
+                       dev->device->name);
+               return -ENOTSUP;
+       }
+
        if (is_zero_ether_addr(mac_addr)) {
                TAP_LOG(ERR, "%s: can't set an empty MAC address",
                        dev->device->name);
@@ -1025,6 +1033,8 @@ tap_setup_queue(struct rte_eth_dev *dev,
        tx->mtu = &dev->data->mtu;
        rx->rxmode = &dev->data->dev_conf.rxmode;
 
+       tx->type = pmd->type;
+
        return *fd;
 }
 
@@ -1342,7 +1352,8 @@ static const struct eth_dev_ops ops = {
 
 static int
 eth_dev_tap_create(struct rte_vdev_device *vdev, char *tap_name,
-                  char *remote_iface, struct ether_addr *mac_addr)
+                  char *remote_iface, struct ether_addr *mac_addr,
+                  enum rte_tuntap_type type)
 {
        int numa_node = rte_socket_id();
        struct rte_eth_dev *dev;
@@ -1364,6 +1375,7 @@ eth_dev_tap_create(struct rte_vdev_device *vdev, char *tap_name,
        pmd = dev->data->dev_private;
        pmd->dev = dev;
        snprintf(pmd->name, sizeof(pmd->name), "%s", tap_name);
+       pmd->type = type;
 
        pmd->ioctl_sock = socket(AF_INET, SOCK_DGRAM, 0);
        if (pmd->ioctl_sock == -1) {
@@ -1400,7 +1412,7 @@ eth_dev_tap_create(struct rte_vdev_device *vdev, char *tap_name,
                pmd->txq[i].fd = -1;
        }
 
-       if (tap_type) {
+       if (pmd->type == ETH_TUNTAP_TYPE_TAP) {
                if (is_zero_ether_addr(mac_addr))
                        eth_random_addr((uint8_t *)&pmd->eth_addr);
                else
@@ -1423,7 +1435,7 @@ eth_dev_tap_create(struct rte_vdev_device *vdev, char *tap_name,
        if (tap_ioctl(pmd, SIOCSIFMTU, &ifr, 1, LOCAL_AND_REMOTE) < 0)
                goto error_exit;
 
-       if (tap_type) {
+       if (pmd->type == ETH_TUNTAP_TYPE_TAP) {
                memset(&ifr, 0, sizeof(struct ifreq));
                ifr.ifr_hwaddr.sa_family = AF_LOCAL;
                rte_memcpy(ifr.ifr_hwaddr.sa_data, &pmd->eth_addr,
@@ -1642,7 +1654,6 @@ rte_pmd_tun_probe(struct rte_vdev_device *dev)
        char tun_name[RTE_ETH_NAME_MAX_LEN];
        char remote_iface[RTE_ETH_NAME_MAX_LEN];
 
-       tap_type = 0;
        strcpy(tuntap_name, "TUN");
 
        name = rte_vdev_device_name(dev);
@@ -1673,7 +1684,8 @@ rte_pmd_tun_probe(struct rte_vdev_device *dev)
        TAP_LOG(NOTICE, "Initializing pmd_tun for %s as %s",
                name, tun_name);
 
-       ret = eth_dev_tap_create(dev, tun_name, remote_iface, 0);
+       ret = eth_dev_tap_create(dev, tun_name, remote_iface, 0,
+               ETH_TUNTAP_TYPE_TUN);
 
 leave:
        if (ret == -1) {
@@ -1700,7 +1712,6 @@ rte_pmd_tap_probe(struct rte_vdev_device *dev)
        struct ether_addr user_mac = { .addr_bytes = {0} };
        struct rte_eth_dev *eth_dev;
 
-       tap_type = 1;
        strcpy(tuntap_name, "TAP");
 
        name = rte_vdev_device_name(dev);
@@ -1762,7 +1773,8 @@ rte_pmd_tap_probe(struct rte_vdev_device *dev)
        TAP_LOG(NOTICE, "Initializing pmd_tap for %s as %s",
                name, tap_name);
 
-       ret = eth_dev_tap_create(dev, tap_name, remote_iface, &user_mac);
+       ret = eth_dev_tap_create(dev, tap_name, remote_iface, &user_mac,
+               ETH_TUNTAP_TYPE_TAP);
 
 leave:
        if (ret == -1) {
@@ -1784,15 +1796,17 @@ rte_pmd_tap_remove(struct rte_vdev_device *dev)
        struct pmd_internals *internals;
        int i;
 
-       TAP_LOG(DEBUG, "Closing TUN/TAP Ethernet device on numa %u",
-               rte_socket_id());
-
        /* find the ethdev entry */
        eth_dev = rte_eth_dev_allocated(rte_vdev_device_name(dev));
        if (!eth_dev)
                return 0;
 
        internals = eth_dev->data->dev_private;
+
+       TAP_LOG(DEBUG, "Closing %s Ethernet device on numa %u",
+               (internals->type == ETH_TUNTAP_TYPE_TAP) ? "TAP" : "TUN",
+               rte_socket_id());
+
        if (internals->nlsk_fd) {
                tap_flow_flush(eth_dev, NULL);
                tap_flow_implicit_flush(internals, NULL);
index c87a0b8..7b21d0d 100644 (file)
 #define RTE_PMD_TAP_MAX_QUEUES 1
 #endif
 
+enum rte_tuntap_type {
+       ETH_TUNTAP_TYPE_UNKNOWN,
+       ETH_TUNTAP_TYPE_TUN,
+       ETH_TUNTAP_TYPE_TAP,
+       ETH_TUNTAP_TYPE_MAX,
+};
+
 struct pkt_stats {
        uint64_t opackets;              /* Number of output packets */
        uint64_t ipackets;              /* Number of input packets */
@@ -48,6 +55,7 @@ struct rx_queue {
 
 struct tx_queue {
        int fd;
+       int type;                       /* Type field - TUN|TAP */
        uint16_t *mtu;                  /* Pointer to MTU from dev_data */
        uint16_t csum:1;                /* Enable checksum offloading */
        struct pkt_stats stats;         /* Stats for this TX queue */
@@ -57,6 +65,7 @@ struct pmd_internals {
        struct rte_eth_dev *dev;          /* Ethernet device. */
        char remote_iface[RTE_ETH_NAME_MAX_LEN]; /* Remote netdevice name */
        char name[RTE_ETH_NAME_MAX_LEN];  /* Internal Tap device name */
+       int type;                         /* Type field - TUN|TAP */
        struct ether_addr eth_addr;       /* Mac address of the device port */
        struct ifreq remote_initial_flags;   /* Remote netdevice flags on init */
        int remote_if_index;              /* remote netdevice IF_INDEX */