X-Git-Url: http://git.droids-corp.org/?a=blobdiff_plain;f=drivers%2Fnet%2Ftap%2Frte_eth_tap.c;h=7cfa0a85d375b19a140074bda12d30e1924b0e90;hb=98a7ea332ba3da0f74ec951595d36a616165b255;hp=efc44260bf314513f3ed4b3a834b72867a04fac3;hpb=e1dbde9a1f8e7702cf191fb489d37f52243114e2;p=dpdk.git diff --git a/drivers/net/tap/rte_eth_tap.c b/drivers/net/tap/rte_eth_tap.c index efc44260bf..7cfa0a85d3 100644 --- a/drivers/net/tap/rte_eth_tap.c +++ b/drivers/net/tap/rte_eth_tap.c @@ -1,7 +1,7 @@ /*- * BSD LICENSE * - * Copyright(c) 2016 Intel Corporation. All rights reserved. + * Copyright(c) 2016-2017 Intel Corporation. All rights reserved. * All rights reserved. * * Redistribution and use in source and binary forms, with or without @@ -31,48 +31,70 @@ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ +#include +#include +#include +#include #include #include +#include #include #include #include +#include +#include +#include #include #include #include #include +#include #include +#include +#include +#include +#include #include -#include #include -#include +#include #include #include +#include #include +#include +#include +#include +#include + /* Linux based path to the TUN device */ #define TUN_TAP_DEV_PATH "/dev/net/tun" #define DEFAULT_TAP_NAME "dtap" #define ETH_TAP_IFACE_ARG "iface" #define ETH_TAP_SPEED_ARG "speed" +#define ETH_TAP_REMOTE_ARG "remote" +#define ETH_TAP_MAC_ARG "mac" +#define ETH_TAP_MAC_FIXED "fixed" -#ifdef IFF_MULTI_QUEUE -#define RTE_PMD_TAP_MAX_QUEUES 16 -#else -#define RTE_PMD_TAP_MAX_QUEUES 1 -#endif +#define FLOWER_KERNEL_VERSION KERNEL_VERSION(4, 2, 0) +#define FLOWER_VLAN_KERNEL_VERSION KERNEL_VERSION(4, 9, 0) static struct rte_vdev_driver pmd_tap_drv; static const char *valid_arguments[] = { ETH_TAP_IFACE_ARG, ETH_TAP_SPEED_ARG, + ETH_TAP_REMOTE_ARG, + ETH_TAP_MAC_ARG, NULL }; static int tap_unit; +static volatile uint32_t tap_trigger; /* Rx trigger */ + static struct rte_eth_link pmd_link = { .link_speed = ETH_SPEED_NUM_10G, .link_duplex = ETH_LINK_FULL_DUPLEX, @@ -80,37 +102,21 @@ static struct rte_eth_link pmd_link = { .link_autoneg = ETH_LINK_SPEED_AUTONEG }; -struct pkt_stats { - uint64_t opackets; /* Number of output packets */ - uint64_t ipackets; /* Number of input packets */ - uint64_t obytes; /* Number of bytes on output */ - uint64_t ibytes; /* Number of bytes on input */ - uint64_t errs; /* Number of error packets */ -}; - -struct rx_queue { - struct rte_mempool *mp; /* Mempool for RX packets */ - uint16_t in_port; /* Port ID */ - int fd; - - struct pkt_stats stats; /* Stats for this RX queue */ -}; +static void +tap_trigger_cb(int sig __rte_unused) +{ + /* Valid trigger values are nonzero */ + tap_trigger = (tap_trigger + 1) | 0x80000000; +} -struct tx_queue { - int fd; - struct pkt_stats stats; /* Stats for this TX queue */ +/* Specifies on what netdevices the ioctl should be applied */ +enum ioctl_mode { + LOCAL_AND_REMOTE, + LOCAL_ONLY, + REMOTE_ONLY, }; -struct pmd_internals { - char name[RTE_ETH_NAME_MAX_LEN]; /* Internal Tap device name */ - uint16_t nb_queues; /* Number of queues supported */ - struct ether_addr eth_addr; /* Mac address of the device port */ - - int if_index; /* IF_INDEX for the port */ - - struct rx_queue rxq[RTE_PMD_TAP_MAX_QUEUES]; /* List of RX queues */ - struct tx_queue txq[RTE_PMD_TAP_MAX_QUEUES]; /* List of TX queues */ -}; +static int tap_intr_handle_set(struct rte_eth_dev *dev, int set); /* Tun/Tap allocation routine * @@ -118,7 +124,7 @@ struct pmd_internals { * supplied name. */ static int -tun_alloc(struct pmd_internals *pmd, uint16_t qid) +tun_alloc(struct pmd_internals *pmd) { struct ifreq ifr; #ifdef IFF_MULTI_QUEUE @@ -128,8 +134,12 @@ tun_alloc(struct pmd_internals *pmd, uint16_t qid) memset(&ifr, 0, sizeof(struct ifreq)); - ifr.ifr_flags = IFF_TAP | IFF_NO_PI; - strncpy(ifr.ifr_name, pmd->name, IFNAMSIZ); + /* + * 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 = IFF_TAP; + snprintf(ifr.ifr_name, IFNAMSIZ, "%s", pmd->name); RTE_LOG(DEBUG, PMD, "ifr_name '%s'\n", ifr.ifr_name); @@ -176,14 +186,41 @@ tun_alloc(struct pmd_internals *pmd, uint16_t qid) goto error; } - if (qid == 0) { - if (ioctl(fd, SIOCGIFHWADDR, &ifr) == -1) { - RTE_LOG(ERR, PMD, "ioctl failed (SIOCGIFHWADDR) (%s)\n", - ifr.ifr_name); - goto error; - } + /* Set up trigger to optimize empty Rx bursts */ + errno = 0; + do { + struct sigaction sa; + int flags = fcntl(fd, F_GETFL); - rte_memcpy(&pmd->eth_addr, ifr.ifr_hwaddr.sa_data, ETH_ALEN); + if (flags == -1 || sigaction(SIGIO, NULL, &sa) == -1) + break; + if (sa.sa_handler != tap_trigger_cb) { + /* + * Make sure SIGIO is not already taken. This is done + * as late as possible to leave the application a + * chance to set up its own signal handler first. + */ + if (sa.sa_handler != SIG_IGN && + sa.sa_handler != SIG_DFL) { + errno = EBUSY; + break; + } + sa = (struct sigaction){ + .sa_flags = SA_RESTART, + .sa_handler = tap_trigger_cb, + }; + if (sigaction(SIGIO, &sa, NULL) == -1) + break; + } + /* Enable SIGIO on file descriptor */ + fcntl(fd, F_SETFL, flags | O_ASYNC); + fcntl(fd, F_SETOWN, getpid()); + } while (0); + if (errno) { + /* Disable trigger globally in case of error */ + tap_trigger = 0; + RTE_LOG(WARNING, PMD, "Rx trigger disabled: %s\n", + strerror(errno)); } return fd; @@ -194,82 +231,261 @@ error: return -1; } +static void +tap_verify_csum(struct rte_mbuf *mbuf) +{ + uint32_t l2 = mbuf->packet_type & RTE_PTYPE_L2_MASK; + uint32_t l3 = mbuf->packet_type & RTE_PTYPE_L3_MASK; + uint32_t l4 = mbuf->packet_type & RTE_PTYPE_L4_MASK; + unsigned int l2_len = sizeof(struct ether_hdr); + unsigned int l3_len; + uint16_t cksum = 0; + void *l3_hdr; + void *l4_hdr; + + if (l2 == RTE_PTYPE_L2_ETHER_VLAN) + l2_len += 4; + else if (l2 == RTE_PTYPE_L2_ETHER_QINQ) + l2_len += 8; + /* Don't verify checksum for packets with discontinuous L2 header */ + if (unlikely(l2_len + sizeof(struct ipv4_hdr) > + rte_pktmbuf_data_len(mbuf))) + return; + l3_hdr = rte_pktmbuf_mtod_offset(mbuf, void *, l2_len); + if (l3 == RTE_PTYPE_L3_IPV4 || l3 == RTE_PTYPE_L3_IPV4_EXT) { + struct ipv4_hdr *iph = l3_hdr; + + /* ihl contains the number of 4-byte words in the header */ + l3_len = 4 * (iph->version_ihl & 0xf); + if (unlikely(l2_len + l3_len > rte_pktmbuf_data_len(mbuf))) + return; + + cksum = ~rte_raw_cksum(iph, l3_len); + mbuf->ol_flags |= cksum ? + PKT_RX_IP_CKSUM_BAD : + PKT_RX_IP_CKSUM_GOOD; + } else if (l3 == RTE_PTYPE_L3_IPV6) { + l3_len = sizeof(struct ipv6_hdr); + } else { + /* IPv6 extensions are not supported */ + return; + } + if (l4 == RTE_PTYPE_L4_UDP || l4 == RTE_PTYPE_L4_TCP) { + l4_hdr = rte_pktmbuf_mtod_offset(mbuf, void *, l2_len + l3_len); + /* Don't verify checksum for multi-segment packets. */ + if (mbuf->nb_segs > 1) + return; + if (l3 == RTE_PTYPE_L3_IPV4) + cksum = ~rte_ipv4_udptcp_cksum(l3_hdr, l4_hdr); + else if (l3 == RTE_PTYPE_L3_IPV6) + cksum = ~rte_ipv6_udptcp_cksum(l3_hdr, l4_hdr); + mbuf->ol_flags |= cksum ? + PKT_RX_L4_CKSUM_BAD : + PKT_RX_L4_CKSUM_GOOD; + } +} + /* Callback to handle the rx burst of packets to the correct interface and * file descriptor(s) in a multi-queue setup. */ static uint16_t pmd_rx_burst(void *queue, struct rte_mbuf **bufs, uint16_t nb_pkts) { - int len; - struct rte_mbuf *mbuf; struct rx_queue *rxq = queue; uint16_t num_rx; unsigned long num_rx_bytes = 0; + uint32_t trigger = tap_trigger; + if (trigger == rxq->trigger_seen) + return 0; + if (trigger) + rxq->trigger_seen = trigger; + rte_compiler_barrier(); for (num_rx = 0; num_rx < nb_pkts; ) { - /* allocate the next mbuf */ - mbuf = rte_pktmbuf_alloc(rxq->mp); - if (unlikely(!mbuf)) { - RTE_LOG(WARNING, PMD, "TAP unable to allocate mbuf\n"); + struct rte_mbuf *mbuf = rxq->pool; + struct rte_mbuf *seg = NULL; + struct rte_mbuf *new_tail = NULL; + uint16_t data_off = rte_pktmbuf_headroom(mbuf); + int len; + + len = readv(rxq->fd, *rxq->iovecs, + 1 + (rxq->rxmode->enable_scatter ? + rxq->nb_rx_desc : 1)); + if (len < (int)sizeof(struct tun_pi)) break; - } - len = read(rxq->fd, rte_pktmbuf_mtod(mbuf, char *), - rte_pktmbuf_tailroom(mbuf)); - if (len <= 0) { - rte_pktmbuf_free(mbuf); - break; + /* Packet couldn't fit in the provided mbuf */ + if (unlikely(rxq->pi.flags & TUN_PKT_STRIP)) { + rxq->stats.ierrors++; + continue; } - mbuf->data_len = len; + len -= sizeof(struct tun_pi); + mbuf->pkt_len = len; mbuf->port = rxq->in_port; + while (1) { + struct rte_mbuf *buf = rte_pktmbuf_alloc(rxq->mp); + + if (unlikely(!buf)) { + rxq->stats.rx_nombuf++; + /* No new buf has been allocated: do nothing */ + if (!new_tail || !seg) + goto end; + + seg->next = NULL; + rte_pktmbuf_free(mbuf); + + goto end; + } + seg = seg ? seg->next : mbuf; + if (rxq->pool == mbuf) + rxq->pool = buf; + if (new_tail) + new_tail->next = buf; + new_tail = buf; + new_tail->next = seg->next; + + /* iovecs[0] is reserved for packet info (pi) */ + (*rxq->iovecs)[mbuf->nb_segs].iov_len = + buf->buf_len - data_off; + (*rxq->iovecs)[mbuf->nb_segs].iov_base = + (char *)buf->buf_addr + data_off; + + seg->data_len = RTE_MIN(seg->buf_len - data_off, len); + seg->data_off = data_off; + + len -= seg->data_len; + if (len <= 0) + break; + mbuf->nb_segs++; + /* First segment has headroom, not the others */ + data_off = 0; + } + seg->next = NULL; + mbuf->packet_type = rte_net_get_ptype(mbuf, NULL, + RTE_PTYPE_ALL_MASK); + if (rxq->rxmode->hw_ip_checksum) + tap_verify_csum(mbuf); /* account for the receive frame */ bufs[num_rx++] = mbuf; num_rx_bytes += mbuf->pkt_len; } +end: rxq->stats.ipackets += num_rx; rxq->stats.ibytes += num_rx_bytes; return num_rx; } +static void +tap_tx_offload(char *packet, uint64_t ol_flags, unsigned int l2_len, + unsigned int l3_len) +{ + void *l3_hdr = packet + l2_len; + + if (ol_flags & (PKT_TX_IP_CKSUM | PKT_TX_IPV4)) { + struct ipv4_hdr *iph = l3_hdr; + uint16_t cksum; + + iph->hdr_checksum = 0; + cksum = rte_raw_cksum(iph, l3_len); + iph->hdr_checksum = (cksum == 0xffff) ? cksum : ~cksum; + } + if (ol_flags & PKT_TX_L4_MASK) { + uint16_t l4_len; + uint32_t cksum; + uint16_t *l4_cksum; + void *l4_hdr; + + l4_hdr = packet + l2_len + l3_len; + if ((ol_flags & PKT_TX_L4_MASK) == PKT_TX_UDP_CKSUM) + l4_cksum = &((struct udp_hdr *)l4_hdr)->dgram_cksum; + else if ((ol_flags & PKT_TX_L4_MASK) == PKT_TX_TCP_CKSUM) + l4_cksum = &((struct tcp_hdr *)l4_hdr)->cksum; + else + return; + *l4_cksum = 0; + if (ol_flags & PKT_TX_IPV4) { + struct ipv4_hdr *iph = l3_hdr; + + l4_len = rte_be_to_cpu_16(iph->total_length) - l3_len; + cksum = rte_ipv4_phdr_cksum(l3_hdr, 0); + } else { + struct ipv6_hdr *ip6h = l3_hdr; + + /* payload_len does not include ext headers */ + l4_len = rte_be_to_cpu_16(ip6h->payload_len) - + l3_len + sizeof(struct ipv6_hdr); + cksum = rte_ipv6_phdr_cksum(l3_hdr, 0); + } + cksum += rte_raw_cksum(l4_hdr, l4_len); + cksum = ((cksum & 0xffff0000) >> 16) + (cksum & 0xffff); + cksum = (~cksum) & 0xffff; + if (cksum == 0) + cksum = 0xffff; + *l4_cksum = cksum; + } +} + /* Callback to handle sending packets from the tap interface */ static uint16_t pmd_tx_burst(void *queue, struct rte_mbuf **bufs, uint16_t nb_pkts) { - struct rte_mbuf *mbuf; struct tx_queue *txq = queue; - struct pollfd pfd; uint16_t num_tx = 0; unsigned long num_tx_bytes = 0; - int i, n; + uint32_t max_size; + int i; if (unlikely(nb_pkts == 0)) return 0; - pfd.events = POLLOUT; - pfd.fd = txq->fd; + max_size = *txq->mtu + (ETHER_HDR_LEN + ETHER_CRC_LEN + 4); for (i = 0; i < nb_pkts; i++) { - n = poll(&pfd, 1, 0); - - if (n <= 0) + struct rte_mbuf *mbuf = bufs[num_tx]; + struct iovec iovecs[mbuf->nb_segs + 1]; + struct tun_pi pi = { .flags = 0 }; + struct rte_mbuf *seg = mbuf; + char m_copy[mbuf->data_len]; + int n; + int j; + + /* stats.errs will be incremented */ + if (rte_pktmbuf_pkt_len(mbuf) > max_size) break; - if (pfd.revents & POLLOUT) { - /* copy the tx frame data */ - mbuf = bufs[num_tx]; - n = write(pfd.fd, rte_pktmbuf_mtod(mbuf, void*), - rte_pktmbuf_pkt_len(mbuf)); - if (n <= 0) + iovecs[0].iov_base = π + iovecs[0].iov_len = sizeof(pi); + for (j = 1; j <= mbuf->nb_segs; j++) { + iovecs[j].iov_len = rte_pktmbuf_data_len(seg); + iovecs[j].iov_base = + rte_pktmbuf_mtod(seg, void *); + seg = seg->next; + } + if (mbuf->ol_flags & (PKT_TX_IP_CKSUM | PKT_TX_IPV4) || + (mbuf->ol_flags & PKT_TX_L4_MASK) == PKT_TX_UDP_CKSUM || + (mbuf->ol_flags & PKT_TX_L4_MASK) == PKT_TX_TCP_CKSUM) { + /* Support only packets with all data in the same seg */ + if (mbuf->nb_segs > 1) break; - - num_tx++; - num_tx_bytes += mbuf->pkt_len; - rte_pktmbuf_free(mbuf); + /* To change checksums, work on a copy of data. */ + rte_memcpy(m_copy, rte_pktmbuf_mtod(mbuf, void *), + rte_pktmbuf_data_len(mbuf)); + tap_tx_offload(m_copy, mbuf->ol_flags, + mbuf->l2_len, mbuf->l3_len); + iovecs[1].iov_base = m_copy; } + /* copy the tx frame data */ + n = writev(txq->fd, iovecs, mbuf->nb_segs + 1); + if (n <= 0) + break; + + num_tx++; + num_tx_bytes += mbuf->pkt_len; + rte_pktmbuf_free(mbuf); } txq->stats.opackets += num_tx; @@ -279,69 +495,101 @@ pmd_tx_burst(void *queue, struct rte_mbuf **bufs, uint16_t nb_pkts) return num_tx; } +static const char * +tap_ioctl_req2str(unsigned long request) +{ + switch (request) { + case SIOCSIFFLAGS: + return "SIOCSIFFLAGS"; + case SIOCGIFFLAGS: + return "SIOCGIFFLAGS"; + case SIOCGIFHWADDR: + return "SIOCGIFHWADDR"; + case SIOCSIFHWADDR: + return "SIOCSIFHWADDR"; + case SIOCSIFMTU: + return "SIOCSIFMTU"; + } + return "UNKNOWN"; +} + static int -tap_link_set_flags(struct pmd_internals *pmd, short flags, int add) +tap_ioctl(struct pmd_internals *pmd, unsigned long request, + struct ifreq *ifr, int set, enum ioctl_mode mode) { - struct ifreq ifr; - int err, s; + short req_flags = ifr->ifr_flags; + int remote = pmd->remote_if_index && + (mode == REMOTE_ONLY || mode == LOCAL_AND_REMOTE); + if (!pmd->remote_if_index && mode == REMOTE_ONLY) + return 0; /* - * An AF_INET/DGRAM socket is needed for - * SIOCGIFFLAGS/SIOCSIFFLAGS, using fd won't work. + * If there is a remote netdevice, apply ioctl on it, then apply it on + * the tap netdevice. */ - s = socket(AF_INET, SOCK_DGRAM, 0); - if (s < 0) { - RTE_LOG(ERR, PMD, - "Unable to get a socket to set flags: %s\n", - strerror(errno)); - return -1; - } - memset(&ifr, 0, sizeof(ifr)); - strncpy(ifr.ifr_name, pmd->name, IFNAMSIZ); - err = ioctl(s, SIOCGIFFLAGS, &ifr); - if (err < 0) { - RTE_LOG(WARNING, PMD, "Unable to get %s device flags: %s\n", - pmd->name, strerror(errno)); - close(s); - return -1; - } - if (add) - ifr.ifr_flags |= flags; - else - ifr.ifr_flags &= ~flags; - err = ioctl(s, SIOCSIFFLAGS, &ifr); - if (err < 0) { - RTE_LOG(WARNING, PMD, "Unable to %s flags 0x%x: %s\n", - add ? "set" : "unset", flags, strerror(errno)); - close(s); - return -1; +apply: + if (remote) + snprintf(ifr->ifr_name, IFNAMSIZ, "%s", pmd->remote_iface); + else if (mode == LOCAL_ONLY || mode == LOCAL_AND_REMOTE) + snprintf(ifr->ifr_name, IFNAMSIZ, "%s", pmd->name); + switch (request) { + case SIOCSIFFLAGS: + /* fetch current flags to leave other flags untouched */ + if (ioctl(pmd->ioctl_sock, SIOCGIFFLAGS, ifr) < 0) + goto error; + if (set) + ifr->ifr_flags |= req_flags; + else + ifr->ifr_flags &= ~req_flags; + break; + case SIOCGIFFLAGS: + case SIOCGIFHWADDR: + case SIOCSIFHWADDR: + case SIOCSIFMTU: + break; + default: + RTE_ASSERT(!"unsupported request type: must not happen"); } - close(s); - + if (ioctl(pmd->ioctl_sock, request, ifr) < 0) + goto error; + if (remote-- && mode == LOCAL_AND_REMOTE) + goto apply; return 0; + +error: + RTE_LOG(DEBUG, PMD, "%s: %s(%s) failed: %s(%d)\n", ifr->ifr_name, + __func__, tap_ioctl_req2str(request), strerror(errno), errno); + return -errno; } static int tap_link_set_down(struct rte_eth_dev *dev) { struct pmd_internals *pmd = dev->data->dev_private; + struct ifreq ifr = { .ifr_flags = IFF_UP }; dev->data->dev_link.link_status = ETH_LINK_DOWN; - return tap_link_set_flags(pmd, IFF_UP | IFF_NOARP, 0); + return tap_ioctl(pmd, SIOCSIFFLAGS, &ifr, 0, LOCAL_AND_REMOTE); } static int tap_link_set_up(struct rte_eth_dev *dev) { struct pmd_internals *pmd = dev->data->dev_private; + struct ifreq ifr = { .ifr_flags = IFF_UP }; dev->data->dev_link.link_status = ETH_LINK_UP; - return tap_link_set_flags(pmd, IFF_UP | IFF_NOARP, 1); + return tap_ioctl(pmd, SIOCSIFFLAGS, &ifr, 1, LOCAL_AND_REMOTE); } static int tap_dev_start(struct rte_eth_dev *dev) { + int err; + + err = tap_intr_handle_set(dev, 1); + if (err) + return err; return tap_link_set_up(dev); } @@ -350,6 +598,7 @@ tap_dev_start(struct rte_eth_dev *dev) static void tap_dev_stop(struct rte_eth_dev *dev) { + tap_intr_handle_set(dev, 0); tap_link_set_down(dev); } @@ -359,6 +608,40 @@ tap_dev_configure(struct rte_eth_dev *dev __rte_unused) return 0; } +static uint32_t +tap_dev_speed_capa(void) +{ + uint32_t speed = pmd_link.link_speed; + uint32_t capa = 0; + + if (speed >= ETH_SPEED_NUM_10M) + capa |= ETH_LINK_SPEED_10M; + if (speed >= ETH_SPEED_NUM_100M) + capa |= ETH_LINK_SPEED_100M; + if (speed >= ETH_SPEED_NUM_1G) + capa |= ETH_LINK_SPEED_1G; + if (speed >= ETH_SPEED_NUM_5G) + capa |= ETH_LINK_SPEED_2_5G; + if (speed >= ETH_SPEED_NUM_5G) + capa |= ETH_LINK_SPEED_5G; + if (speed >= ETH_SPEED_NUM_10G) + capa |= ETH_LINK_SPEED_10G; + if (speed >= ETH_SPEED_NUM_20G) + capa |= ETH_LINK_SPEED_20G; + if (speed >= ETH_SPEED_NUM_25G) + capa |= ETH_LINK_SPEED_25G; + if (speed >= ETH_SPEED_NUM_40G) + capa |= ETH_LINK_SPEED_40G; + if (speed >= ETH_SPEED_NUM_50G) + capa |= ETH_LINK_SPEED_50G; + if (speed >= ETH_SPEED_NUM_56G) + capa |= ETH_LINK_SPEED_56G; + if (speed >= ETH_SPEED_NUM_100G) + capa |= ETH_LINK_SPEED_100G; + + return capa; +} + static void tap_dev_info(struct rte_eth_dev *dev, struct rte_eth_dev_info *dev_info) { @@ -371,6 +654,14 @@ tap_dev_info(struct rte_eth_dev *dev, struct rte_eth_dev_info *dev_info) dev_info->max_tx_queues = internals->nb_queues; dev_info->min_rx_bufsize = 0; dev_info->pci_dev = NULL; + dev_info->speed_capa = tap_dev_speed_capa(); + dev_info->rx_offload_capa = (DEV_RX_OFFLOAD_IPV4_CKSUM | + DEV_RX_OFFLOAD_UDP_CKSUM | + DEV_RX_OFFLOAD_TCP_CKSUM); + dev_info->tx_offload_capa = + (DEV_TX_OFFLOAD_IPV4_CKSUM | + DEV_TX_OFFLOAD_UDP_CKSUM | + DEV_TX_OFFLOAD_TCP_CKSUM); } static void @@ -379,6 +670,7 @@ tap_stats_get(struct rte_eth_dev *dev, struct rte_eth_stats *tap_stats) unsigned int i, imax; unsigned long rx_total = 0, tx_total = 0, tx_err_total = 0; unsigned long rx_bytes_total = 0, tx_bytes_total = 0; + unsigned long rx_nombuf = 0, ierrors = 0; const struct pmd_internals *pmd = dev->data->dev_private; imax = (pmd->nb_queues < RTE_ETHDEV_QUEUE_STAT_CNTRS) ? @@ -389,6 +681,8 @@ tap_stats_get(struct rte_eth_dev *dev, struct rte_eth_stats *tap_stats) tap_stats->q_ibytes[i] = pmd->rxq[i].stats.ibytes; rx_total += tap_stats->q_ipackets[i]; rx_bytes_total += tap_stats->q_ibytes[i]; + rx_nombuf += pmd->rxq[i].stats.rx_nombuf; + ierrors += pmd->rxq[i].stats.ierrors; tap_stats->q_opackets[i] = pmd->txq[i].stats.opackets; tap_stats->q_errors[i] = pmd->txq[i].stats.errs; @@ -400,6 +694,8 @@ tap_stats_get(struct rte_eth_dev *dev, struct rte_eth_stats *tap_stats) tap_stats->ipackets = rx_total; tap_stats->ibytes = rx_bytes_total; + tap_stats->ierrors = ierrors; + tap_stats->rx_nombuf = rx_nombuf; tap_stats->opackets = tx_total; tap_stats->oerrors = tx_err_total; tap_stats->obytes = tx_bytes_total; @@ -414,6 +710,8 @@ tap_stats_reset(struct rte_eth_dev *dev) for (i = 0; i < pmd->nb_queues; i++) { pmd->rxq[i].stats.ipackets = 0; pmd->rxq[i].stats.ibytes = 0; + pmd->rxq[i].stats.ierrors = 0; + pmd->rxq[i].stats.rx_nombuf = 0; pmd->txq[i].stats.opackets = 0; pmd->txq[i].stats.errs = 0; @@ -422,12 +720,14 @@ tap_stats_reset(struct rte_eth_dev *dev) } static void -tap_dev_close(struct rte_eth_dev *dev __rte_unused) +tap_dev_close(struct rte_eth_dev *dev) { int i; struct pmd_internals *internals = dev->data->dev_private; tap_link_set_down(dev); + tap_flow_flush(dev, NULL); + tap_flow_implicit_flush(internals, NULL); for (i = 0; i < internals->nb_queues; i++) { if (internals->rxq[i].fd != -1) @@ -445,6 +745,10 @@ tap_rx_queue_release(void *queue) if (rxq && (rxq->fd > 0)) { close(rxq->fd); rxq->fd = -1; + rte_pktmbuf_free(rxq->pool); + rte_free(rxq->iovecs); + rxq->pool = NULL; + rxq->iovecs = NULL; } } @@ -460,9 +764,25 @@ tap_tx_queue_release(void *queue) } static int -tap_link_update(struct rte_eth_dev *dev __rte_unused, - int wait_to_complete __rte_unused) +tap_link_update(struct rte_eth_dev *dev, int wait_to_complete __rte_unused) { + struct rte_eth_link *dev_link = &dev->data->dev_link; + struct pmd_internals *pmd = dev->data->dev_private; + struct ifreq ifr = { .ifr_flags = 0 }; + + if (pmd->remote_if_index) { + tap_ioctl(pmd, SIOCGIFFLAGS, &ifr, 0, REMOTE_ONLY); + if (!(ifr.ifr_flags & IFF_UP) || + !(ifr.ifr_flags & IFF_RUNNING)) { + dev_link->link_status = ETH_LINK_DOWN; + return 0; + } + } + tap_ioctl(pmd, SIOCGIFFLAGS, &ifr, 0, LOCAL_ONLY); + dev_link->link_status = + ((ifr.ifr_flags & IFF_UP) && (ifr.ifr_flags & IFF_RUNNING) ? + ETH_LINK_UP : + ETH_LINK_DOWN); return 0; } @@ -470,36 +790,92 @@ static void tap_promisc_enable(struct rte_eth_dev *dev) { struct pmd_internals *pmd = dev->data->dev_private; + struct ifreq ifr = { .ifr_flags = IFF_PROMISC }; dev->data->promiscuous = 1; - tap_link_set_flags(pmd, IFF_PROMISC, 1); + tap_ioctl(pmd, SIOCSIFFLAGS, &ifr, 1, LOCAL_AND_REMOTE); + if (pmd->remote_if_index) + tap_flow_implicit_create(pmd, TAP_REMOTE_PROMISC); } static void tap_promisc_disable(struct rte_eth_dev *dev) { struct pmd_internals *pmd = dev->data->dev_private; + struct ifreq ifr = { .ifr_flags = IFF_PROMISC }; dev->data->promiscuous = 0; - tap_link_set_flags(pmd, IFF_PROMISC, 0); + tap_ioctl(pmd, SIOCSIFFLAGS, &ifr, 0, LOCAL_AND_REMOTE); + if (pmd->remote_if_index) + tap_flow_implicit_destroy(pmd, TAP_REMOTE_PROMISC); } static void tap_allmulti_enable(struct rte_eth_dev *dev) { struct pmd_internals *pmd = dev->data->dev_private; + struct ifreq ifr = { .ifr_flags = IFF_ALLMULTI }; dev->data->all_multicast = 1; - tap_link_set_flags(pmd, IFF_ALLMULTI, 1); + tap_ioctl(pmd, SIOCSIFFLAGS, &ifr, 1, LOCAL_AND_REMOTE); + if (pmd->remote_if_index) + tap_flow_implicit_create(pmd, TAP_REMOTE_ALLMULTI); } static void tap_allmulti_disable(struct rte_eth_dev *dev) { struct pmd_internals *pmd = dev->data->dev_private; + struct ifreq ifr = { .ifr_flags = IFF_ALLMULTI }; dev->data->all_multicast = 0; - tap_link_set_flags(pmd, IFF_ALLMULTI, 0); + tap_ioctl(pmd, SIOCSIFFLAGS, &ifr, 0, LOCAL_AND_REMOTE); + if (pmd->remote_if_index) + tap_flow_implicit_destroy(pmd, TAP_REMOTE_ALLMULTI); +} + +static void +tap_mac_set(struct rte_eth_dev *dev, struct ether_addr *mac_addr) +{ + struct pmd_internals *pmd = dev->data->dev_private; + enum ioctl_mode mode = LOCAL_ONLY; + struct ifreq ifr; + + if (is_zero_ether_addr(mac_addr)) { + RTE_LOG(ERR, PMD, "%s: can't set an empty MAC address\n", + dev->data->name); + return; + } + /* Check the actual current MAC address on the tap netdevice */ + if (tap_ioctl(pmd, SIOCGIFHWADDR, &ifr, 0, LOCAL_ONLY) < 0) + return; + if (is_same_ether_addr((struct ether_addr *)&ifr.ifr_hwaddr.sa_data, + mac_addr)) + return; + /* Check the current MAC address on the remote */ + if (tap_ioctl(pmd, SIOCGIFHWADDR, &ifr, 0, REMOTE_ONLY) < 0) + return; + if (!is_same_ether_addr((struct ether_addr *)&ifr.ifr_hwaddr.sa_data, + mac_addr)) + mode = LOCAL_AND_REMOTE; + ifr.ifr_hwaddr.sa_family = AF_LOCAL; + rte_memcpy(ifr.ifr_hwaddr.sa_data, mac_addr, ETHER_ADDR_LEN); + if (tap_ioctl(pmd, SIOCSIFHWADDR, &ifr, 1, mode) < 0) + return; + rte_memcpy(&pmd->eth_addr, mac_addr, ETHER_ADDR_LEN); + if (pmd->remote_if_index) { + /* Replace MAC redirection rule after a MAC change */ + if (tap_flow_implicit_destroy(pmd, TAP_REMOTE_LOCAL_MAC) < 0) { + RTE_LOG(ERR, PMD, + "%s: Couldn't delete MAC redirection rule\n", + dev->data->name); + return; + } + if (tap_flow_implicit_create(pmd, TAP_REMOTE_LOCAL_MAC) < 0) + RTE_LOG(ERR, PMD, + "%s: Couldn't add MAC redirection rule\n", + dev->data->name); + } } static int @@ -510,60 +886,45 @@ tap_setup_queue(struct rte_eth_dev *dev, struct pmd_internals *pmd = dev->data->dev_private; struct rx_queue *rx = &internals->rxq[qid]; struct tx_queue *tx = &internals->txq[qid]; - int fd; + int fd = rx->fd == -1 ? tx->fd : rx->fd; - fd = rx->fd; - if (fd < 0) { - fd = tx->fd; + if (fd == -1) { + RTE_LOG(INFO, PMD, "Add queue to TAP %s for qid %d\n", + pmd->name, qid); + fd = tun_alloc(pmd); if (fd < 0) { - RTE_LOG(INFO, PMD, "Add queue to TAP %s for qid %d\n", - pmd->name, qid); - fd = tun_alloc(pmd, qid); - if (fd < 0) { - RTE_LOG(ERR, PMD, "tun_alloc(%s, %d) failed\n", - pmd->name, qid); - return -1; - } + RTE_LOG(ERR, PMD, "%s: tun_alloc() failed.\n", + pmd->name); + return -1; } } rx->fd = fd; tx->fd = fd; + tx->mtu = &dev->data->mtu; + rx->rxmode = &dev->data->dev_conf.rxmode; return fd; } -static int -rx_setup_queue(struct rte_eth_dev *dev, - struct pmd_internals *internals, - uint16_t qid) -{ - dev->data->rx_queues[qid] = &internals->rxq[qid]; - - return tap_setup_queue(dev, internals, qid); -} - -static int -tx_setup_queue(struct rte_eth_dev *dev, - struct pmd_internals *internals, - uint16_t qid) -{ - dev->data->tx_queues[qid] = &internals->txq[qid]; - - return tap_setup_queue(dev, internals, qid); -} - static int tap_rx_queue_setup(struct rte_eth_dev *dev, uint16_t rx_queue_id, - uint16_t nb_rx_desc __rte_unused, - unsigned int socket_id __rte_unused, + uint16_t nb_rx_desc, + unsigned int socket_id, const struct rte_eth_rxconf *rx_conf __rte_unused, struct rte_mempool *mp) { struct pmd_internals *internals = dev->data->dev_private; - uint16_t buf_size; + struct rx_queue *rxq = &internals->rxq[rx_queue_id]; + struct rte_mbuf **tmp = &rxq->pool; + long iov_max = sysconf(_SC_IOV_MAX); + uint16_t nb_desc = RTE_MIN(nb_rx_desc, iov_max - 1); + struct iovec (*iovecs)[nb_desc + 1]; + int data_off = RTE_PKTMBUF_HEADROOM; + int ret = 0; int fd; + int i; if ((rx_queue_id >= internals->nb_queues) || !mp) { RTE_LOG(WARNING, PMD, @@ -572,28 +933,57 @@ tap_rx_queue_setup(struct rte_eth_dev *dev, return -1; } - internals->rxq[rx_queue_id].mp = mp; - internals->rxq[rx_queue_id].in_port = dev->data->port_id; - - /* Now get the space available for data in the mbuf */ - buf_size = (uint16_t)(rte_pktmbuf_data_room_size(mp) - - RTE_PKTMBUF_HEADROOM); - - if (buf_size < ETH_FRAME_LEN) { + rxq->mp = mp; + rxq->trigger_seen = 1; /* force initial burst */ + rxq->in_port = dev->data->port_id; + rxq->nb_rx_desc = nb_desc; + iovecs = rte_zmalloc_socket(dev->data->name, sizeof(*iovecs), 0, + socket_id); + if (!iovecs) { RTE_LOG(WARNING, PMD, - "%s: %d bytes will not fit in mbuf (%d bytes)\n", - dev->data->name, ETH_FRAME_LEN, buf_size); + "%s: Couldn't allocate %d RX descriptors\n", + dev->data->name, nb_desc); return -ENOMEM; } + rxq->iovecs = iovecs; - fd = rx_setup_queue(dev, internals, rx_queue_id); - if (fd == -1) - return -1; + dev->data->rx_queues[rx_queue_id] = rxq; + fd = tap_setup_queue(dev, internals, rx_queue_id); + if (fd == -1) { + ret = fd; + goto error; + } + + (*rxq->iovecs)[0].iov_len = sizeof(struct tun_pi); + (*rxq->iovecs)[0].iov_base = &rxq->pi; + + for (i = 1; i <= nb_desc; i++) { + *tmp = rte_pktmbuf_alloc(rxq->mp); + if (!*tmp) { + RTE_LOG(WARNING, PMD, + "%s: couldn't allocate memory for queue %d\n", + dev->data->name, rx_queue_id); + ret = -ENOMEM; + goto error; + } + (*rxq->iovecs)[i].iov_len = (*tmp)->buf_len - data_off; + (*rxq->iovecs)[i].iov_base = + (char *)(*tmp)->buf_addr + data_off; + data_off = 0; + tmp = &(*tmp)->next; + } RTE_LOG(DEBUG, PMD, " RX TAP device name %s, qid %d on fd %d\n", internals->name, rx_queue_id, internals->rxq[rx_queue_id].fd); return 0; + +error: + rte_pktmbuf_free(rxq->pool); + rxq->pool = NULL; + rte_free(rxq->iovecs); + rxq->iovecs = NULL; + return ret; } static int @@ -609,7 +999,8 @@ tap_tx_queue_setup(struct rte_eth_dev *dev, if (tx_queue_id >= internals->nb_queues) return -1; - ret = tx_setup_queue(dev, internals, tx_queue_id); + dev->data->tx_queues[tx_queue_id] = &internals->txq[tx_queue_id]; + ret = tap_setup_queue(dev, internals, tx_queue_id); if (ret == -1) return -1; @@ -619,6 +1010,128 @@ tap_tx_queue_setup(struct rte_eth_dev *dev, return 0; } +static int +tap_mtu_set(struct rte_eth_dev *dev, uint16_t mtu) +{ + struct pmd_internals *pmd = dev->data->dev_private; + struct ifreq ifr = { .ifr_mtu = mtu }; + int err = 0; + + err = tap_ioctl(pmd, SIOCSIFMTU, &ifr, 1, LOCAL_AND_REMOTE); + if (!err) + dev->data->mtu = mtu; + + return err; +} + +static int +tap_set_mc_addr_list(struct rte_eth_dev *dev __rte_unused, + struct ether_addr *mc_addr_set __rte_unused, + uint32_t nb_mc_addr __rte_unused) +{ + /* + * Nothing to do actually: the tap has no filtering whatsoever, every + * packet is received. + */ + return 0; +} + +static int +tap_nl_msg_handler(struct nlmsghdr *nh, void *arg) +{ + struct rte_eth_dev *dev = arg; + struct pmd_internals *pmd = dev->data->dev_private; + struct ifinfomsg *info = NLMSG_DATA(nh); + + if (nh->nlmsg_type != RTM_NEWLINK || + (info->ifi_index != pmd->if_index && + info->ifi_index != pmd->remote_if_index)) + return 0; + return tap_link_update(dev, 0); +} + +static void +tap_dev_intr_handler(void *cb_arg) +{ + struct rte_eth_dev *dev = cb_arg; + struct pmd_internals *pmd = dev->data->dev_private; + + nl_recv(pmd->intr_handle.fd, tap_nl_msg_handler, dev); +} + +static int +tap_intr_handle_set(struct rte_eth_dev *dev, int set) +{ + struct pmd_internals *pmd = dev->data->dev_private; + + /* In any case, disable interrupt if the conf is no longer there. */ + if (!dev->data->dev_conf.intr_conf.lsc) { + if (pmd->intr_handle.fd != -1) + nl_final(pmd->intr_handle.fd); + rte_intr_callback_unregister( + &pmd->intr_handle, tap_dev_intr_handler, dev); + return 0; + } + if (set) { + pmd->intr_handle.fd = nl_init(RTMGRP_LINK); + if (unlikely(pmd->intr_handle.fd == -1)) + return -EBADF; + return rte_intr_callback_register( + &pmd->intr_handle, tap_dev_intr_handler, dev); + } + nl_final(pmd->intr_handle.fd); + return rte_intr_callback_unregister(&pmd->intr_handle, + tap_dev_intr_handler, dev); +} + +static const uint32_t* +tap_dev_supported_ptypes_get(struct rte_eth_dev *dev __rte_unused) +{ + static const uint32_t ptypes[] = { + RTE_PTYPE_INNER_L2_ETHER, + RTE_PTYPE_INNER_L2_ETHER_VLAN, + RTE_PTYPE_INNER_L2_ETHER_QINQ, + RTE_PTYPE_INNER_L3_IPV4, + RTE_PTYPE_INNER_L3_IPV4_EXT, + RTE_PTYPE_INNER_L3_IPV6, + RTE_PTYPE_INNER_L3_IPV6_EXT, + RTE_PTYPE_INNER_L4_FRAG, + RTE_PTYPE_INNER_L4_UDP, + RTE_PTYPE_INNER_L4_TCP, + RTE_PTYPE_INNER_L4_SCTP, + RTE_PTYPE_L2_ETHER, + RTE_PTYPE_L2_ETHER_VLAN, + RTE_PTYPE_L2_ETHER_QINQ, + RTE_PTYPE_L3_IPV4, + RTE_PTYPE_L3_IPV4_EXT, + RTE_PTYPE_L3_IPV6_EXT, + RTE_PTYPE_L3_IPV6, + RTE_PTYPE_L4_FRAG, + RTE_PTYPE_L4_UDP, + RTE_PTYPE_L4_TCP, + RTE_PTYPE_L4_SCTP, + }; + + return ptypes; +} + +static int +tap_flow_ctrl_get(struct rte_eth_dev *dev __rte_unused, + struct rte_eth_fc_conf *fc_conf) +{ + fc_conf->mode = RTE_FC_NONE; + return 0; +} + +static int +tap_flow_ctrl_set(struct rte_eth_dev *dev __rte_unused, + struct rte_eth_fc_conf *fc_conf) +{ + if (fc_conf->mode != RTE_FC_NONE) + return -ENOTSUP; + return 0; +} + static const struct eth_dev_ops ops = { .dev_start = tap_dev_start, .dev_stop = tap_dev_stop, @@ -629,6 +1142,8 @@ static const struct eth_dev_ops ops = { .tx_queue_setup = tap_tx_queue_setup, .rx_queue_release = tap_rx_queue_release, .tx_queue_release = tap_tx_queue_release, + .flow_ctrl_get = tap_flow_ctrl_get, + .flow_ctrl_set = tap_flow_ctrl_set, .link_update = tap_link_update, .dev_set_link_up = tap_link_set_up, .dev_set_link_down = tap_link_set_down, @@ -636,17 +1151,42 @@ static const struct eth_dev_ops ops = { .promiscuous_disable = tap_promisc_disable, .allmulticast_enable = tap_allmulti_enable, .allmulticast_disable = tap_allmulti_disable, + .mac_addr_set = tap_mac_set, + .mtu_set = tap_mtu_set, + .set_mc_addr_list = tap_set_mc_addr_list, .stats_get = tap_stats_get, .stats_reset = tap_stats_reset, + .dev_supported_ptypes_get = tap_dev_supported_ptypes_get, + .filter_ctrl = tap_dev_filter_ctrl, }; static int -eth_dev_tap_create(const char *name, char *tap_name) +tap_kernel_support(struct pmd_internals *pmd) +{ + struct utsname utsname; + int ver[3]; + + if (uname(&utsname) == -1 || + sscanf(utsname.release, "%d.%d.%d", + &ver[0], &ver[1], &ver[2]) != 3) + return 0; + if (KERNEL_VERSION(ver[0], ver[1], ver[2]) >= FLOWER_KERNEL_VERSION) + pmd->flower_support = 1; + if (KERNEL_VERSION(ver[0], ver[1], ver[2]) >= + FLOWER_VLAN_KERNEL_VERSION) + pmd->flower_vlan_support = 1; + return 1; +} + +static int +eth_dev_tap_create(struct rte_vdev_device *vdev, char *tap_name, + char *remote_iface, int fixed_mac_type) { int numa_node = rte_socket_id(); - struct rte_eth_dev *dev = NULL; - struct pmd_internals *pmd = NULL; - struct rte_eth_dev_data *data = NULL; + struct rte_eth_dev *dev; + struct pmd_internals *pmd; + struct rte_eth_dev_data *data; + struct ifreq ifr; int i; RTE_LOG(DEBUG, PMD, " TAP device on numa %u\n", rte_socket_id()); @@ -657,28 +1197,28 @@ eth_dev_tap_create(const char *name, char *tap_name) goto error_exit; } - pmd = rte_zmalloc_socket(tap_name, sizeof(*pmd), 0, numa_node); - if (!pmd) { - RTE_LOG(ERR, PMD, "TAP Unable to allocate internal struct\n"); - goto error_exit; - } - - dev = rte_eth_dev_allocate(tap_name); + dev = rte_eth_vdev_allocate(vdev, sizeof(*pmd)); if (!dev) { RTE_LOG(ERR, PMD, "TAP Unable to allocate device struct\n"); goto error_exit; } + pmd = dev->data->dev_private; snprintf(pmd->name, sizeof(pmd->name), "%s", tap_name); - pmd->nb_queues = RTE_PMD_TAP_MAX_QUEUES; + pmd->ioctl_sock = socket(AF_INET, SOCK_DGRAM, 0); + if (pmd->ioctl_sock == -1) { + RTE_LOG(ERR, PMD, + "TAP Unable to get a socket for management: %s\n", + strerror(errno)); + goto error_exit; + } + /* Setup some default values */ + rte_memcpy(data, dev->data, sizeof(*data)); data->dev_private = pmd; - data->port_id = dev->data->port_id; - data->dev_flags = RTE_ETH_DEV_DETACHABLE; - data->kdrv = RTE_KDRV_NONE; - data->drv_name = pmd_tap_drv.driver.name; + data->dev_flags = RTE_ETH_DEV_DETACHABLE | RTE_ETH_DEV_INTR_LSC; data->numa_node = numa_node; data->dev_link = pmd_link; @@ -688,10 +1228,11 @@ eth_dev_tap_create(const char *name, char *tap_name) dev->data = data; dev->dev_ops = &ops; - dev->driver = NULL; dev->rx_pkt_burst = pmd_rx_burst; dev->tx_pkt_burst = pmd_tx_burst; - snprintf(dev->data->name, sizeof(dev->data->name), "%s", name); + + pmd->intr_handle.type = RTE_INTR_HANDLE_EXT; + pmd->intr_handle.fd = -1; /* Presetup the fds to -1 as being not valid */ for (i = 0; i < RTE_PMD_TAP_MAX_QUEUES; i++) { @@ -699,16 +1240,146 @@ eth_dev_tap_create(const char *name, char *tap_name) pmd->txq[i].fd = -1; } + if (fixed_mac_type) { + /* fixed mac = 00:64:74:61:70: */ + static int iface_idx; + char mac[ETHER_ADDR_LEN] = "\0dtap"; + + mac[ETHER_ADDR_LEN - 1] = iface_idx++; + rte_memcpy(&pmd->eth_addr, mac, ETHER_ADDR_LEN); + } else { + eth_random_addr((uint8_t *)&pmd->eth_addr); + } + + /* Immediately create the netdevice (this will create the 1st queue). */ + if (tap_setup_queue(dev, pmd, 0) == -1) + goto error_exit; + + ifr.ifr_mtu = dev->data->mtu; + if (tap_ioctl(pmd, SIOCSIFMTU, &ifr, 1, LOCAL_AND_REMOTE) < 0) + goto error_exit; + + memset(&ifr, 0, sizeof(struct ifreq)); + ifr.ifr_hwaddr.sa_family = AF_LOCAL; + rte_memcpy(ifr.ifr_hwaddr.sa_data, &pmd->eth_addr, ETHER_ADDR_LEN); + if (tap_ioctl(pmd, SIOCSIFHWADDR, &ifr, 0, LOCAL_ONLY) < 0) + goto error_exit; + + tap_kernel_support(pmd); + if (!pmd->flower_support) { + if (remote_iface[0]) { + RTE_LOG(ERR, PMD, + "%s: kernel does not support TC rules, required for remote feature.", + pmd->name); + goto error_exit; + } else { + RTE_LOG(INFO, PMD, + "%s: kernel too old for Flow API support.\n", + pmd->name); + return 0; + } + } + + /* + * Set up everything related to rte_flow: + * - netlink socket + * - tap / remote if_index + * - mandatory QDISCs + * - rte_flow actual/implicit lists + * - implicit rules + */ + pmd->nlsk_fd = nl_init(0); + if (pmd->nlsk_fd == -1) { + RTE_LOG(WARNING, PMD, "%s: failed to create netlink socket.", + pmd->name); + goto disable_rte_flow; + } + pmd->if_index = if_nametoindex(pmd->name); + if (!pmd->if_index) { + RTE_LOG(ERR, PMD, "%s: failed to get if_index.", pmd->name); + goto disable_rte_flow; + } + if (qdisc_create_multiq(pmd->nlsk_fd, pmd->if_index) < 0) { + RTE_LOG(ERR, PMD, "%s: failed to create multiq qdisc.", + pmd->name); + goto disable_rte_flow; + } + if (qdisc_create_ingress(pmd->nlsk_fd, pmd->if_index) < 0) { + RTE_LOG(ERR, PMD, "%s: failed to create ingress qdisc.", + pmd->name); + goto disable_rte_flow; + } + LIST_INIT(&pmd->flows); + + if (strlen(remote_iface)) { + pmd->remote_if_index = if_nametoindex(remote_iface); + if (!pmd->remote_if_index) { + RTE_LOG(ERR, PMD, "%s: failed to get %s if_index.", + pmd->name, remote_iface); + goto error_remote; + } + snprintf(pmd->remote_iface, RTE_ETH_NAME_MAX_LEN, + "%s", remote_iface); + if (tap_ioctl(pmd, SIOCGIFHWADDR, &ifr, 0, REMOTE_ONLY) < 0) { + RTE_LOG(ERR, PMD, "%s: failed to get %s MAC address.", + pmd->name, pmd->remote_iface); + goto error_remote; + } + rte_memcpy(&pmd->eth_addr, ifr.ifr_hwaddr.sa_data, + ETHER_ADDR_LEN); + /* The desired MAC is already in ifreq after SIOCGIFHWADDR. */ + if (tap_ioctl(pmd, SIOCSIFHWADDR, &ifr, 0, LOCAL_ONLY) < 0) { + RTE_LOG(ERR, PMD, "%s: failed to get %s MAC address.", + pmd->name, remote_iface); + goto error_remote; + } + + /* + * Flush usually returns negative value because it tries to + * delete every QDISC (and on a running device, one QDISC at + * least is needed). Ignore negative return value. + */ + qdisc_flush(pmd->nlsk_fd, pmd->remote_if_index); + if (qdisc_create_ingress(pmd->nlsk_fd, + pmd->remote_if_index) < 0) { + RTE_LOG(ERR, PMD, "%s: failed to create ingress qdisc.", + pmd->remote_iface); + goto error_remote; + } + LIST_INIT(&pmd->implicit_flows); + if (tap_flow_implicit_create(pmd, TAP_REMOTE_TX) < 0 || + tap_flow_implicit_create(pmd, TAP_REMOTE_LOCAL_MAC) < 0 || + tap_flow_implicit_create(pmd, TAP_REMOTE_BROADCAST) < 0 || + tap_flow_implicit_create(pmd, TAP_REMOTE_BROADCASTV6) < 0) { + RTE_LOG(ERR, PMD, + "%s: failed to create implicit rules.", + pmd->name); + goto error_remote; + } + } + return 0; -error_exit: - RTE_LOG(DEBUG, PMD, "TAP Unable to initialize %s\n", name); +disable_rte_flow: + RTE_LOG(ERR, PMD, " Disabling rte flow support: %s(%d)\n", + strerror(errno), errno); + if (strlen(remote_iface)) { + RTE_LOG(ERR, PMD, "Remote feature requires flow support.\n"); + goto error_exit; + } + pmd->flower_support = 0; + return 0; - rte_free(data); - rte_free(pmd); +error_remote: + RTE_LOG(ERR, PMD, " Can't set up remote feature: %s(%d)\n", + strerror(errno), errno); + tap_flow_implicit_flush(pmd, NULL); - rte_eth_dev_release_port(dev); +error_exit: + RTE_LOG(ERR, PMD, "TAP Unable to initialize %s\n", + rte_vdev_device_name(vdev)); + rte_free(data); return -EINVAL; } @@ -738,22 +1409,53 @@ set_interface_speed(const char *key __rte_unused, return 0; } +static int +set_remote_iface(const char *key __rte_unused, + const char *value, + void *extra_args) +{ + char *name = (char *)extra_args; + + if (value) + snprintf(name, RTE_ETH_NAME_MAX_LEN, "%s", value); + + return 0; +} + +static int +set_mac_type(const char *key __rte_unused, + const char *value, + void *extra_args) +{ + if (value && + !strncasecmp(ETH_TAP_MAC_FIXED, value, strlen(ETH_TAP_MAC_FIXED))) + *(int *)extra_args = 1; + return 0; +} + /* Open a TAP interface device. */ static int -rte_pmd_tap_probe(const char *name, const char *params) +rte_pmd_tap_probe(struct rte_vdev_device *dev) { + const char *name, *params; int ret; struct rte_kvargs *kvlist = NULL; int speed; char tap_name[RTE_ETH_NAME_MAX_LEN]; + char remote_iface[RTE_ETH_NAME_MAX_LEN]; + int fixed_mac_type = 0; + + name = rte_vdev_device_name(dev); + params = rte_vdev_device_args(dev); speed = ETH_SPEED_NUM_10G; snprintf(tap_name, sizeof(tap_name), "%s%d", DEFAULT_TAP_NAME, tap_unit++); + memset(remote_iface, 0, RTE_ETH_NAME_MAX_LEN); if (params && (params[0] != '\0')) { - RTE_LOG(DEBUG, PMD, "paramaters (%s)\n", params); + RTE_LOG(DEBUG, PMD, "parameters (%s)\n", params); kvlist = rte_kvargs_parse(params, valid_arguments); if (kvlist) { @@ -774,6 +1476,24 @@ rte_pmd_tap_probe(const char *name, const char *params) if (ret == -1) goto leave; } + + if (rte_kvargs_count(kvlist, ETH_TAP_REMOTE_ARG) == 1) { + ret = rte_kvargs_process(kvlist, + ETH_TAP_REMOTE_ARG, + &set_remote_iface, + remote_iface); + if (ret == -1) + goto leave; + } + + if (rte_kvargs_count(kvlist, ETH_TAP_MAC_ARG) == 1) { + ret = rte_kvargs_process(kvlist, + ETH_TAP_MAC_ARG, + &set_mac_type, + &fixed_mac_type); + if (ret == -1) + goto leave; + } } } pmd_link.link_speed = speed; @@ -781,7 +1501,7 @@ rte_pmd_tap_probe(const char *name, const char *params) RTE_LOG(NOTICE, PMD, "Initializing pmd_tap for %s as %s\n", name, tap_name); - ret = eth_dev_tap_create(name, tap_name); + ret = eth_dev_tap_create(dev, tap_name, remote_iface, fixed_mac_type); leave: if (ret == -1) { @@ -797,7 +1517,7 @@ leave: /* detach a TAP device. */ static int -rte_pmd_tap_remove(const char *name) +rte_pmd_tap_remove(struct rte_vdev_device *dev) { struct rte_eth_dev *eth_dev = NULL; struct pmd_internals *internals; @@ -807,15 +1527,21 @@ rte_pmd_tap_remove(const char *name) rte_socket_id()); /* find the ethdev entry */ - eth_dev = rte_eth_dev_allocated(name); + eth_dev = rte_eth_dev_allocated(rte_vdev_device_name(dev)); if (!eth_dev) return 0; internals = eth_dev->data->dev_private; + if (internals->flower_support && internals->nlsk_fd) { + tap_flow_flush(eth_dev, NULL); + tap_flow_implicit_flush(internals, NULL); + nl_final(internals->nlsk_fd); + } for (i = 0; i < internals->nb_queues; i++) if (internals->rxq[i].fd != -1) close(internals->rxq[i].fd); + close(internals->ioctl_sock); rte_free(eth_dev->data->dev_private); rte_free(eth_dev->data); @@ -830,4 +1556,8 @@ static struct rte_vdev_driver pmd_tap_drv = { }; RTE_PMD_REGISTER_VDEV(net_tap, pmd_tap_drv); RTE_PMD_REGISTER_ALIAS(net_tap, eth_tap); -RTE_PMD_REGISTER_PARAM_STRING(net_tap, "iface=,speed=N"); +RTE_PMD_REGISTER_PARAM_STRING(net_tap, + ETH_TAP_IFACE_ARG "= " + ETH_TAP_SPEED_ARG "= " + ETH_TAP_MAC_ARG "=" ETH_TAP_MAC_FIXED " " + ETH_TAP_REMOTE_ARG "=");