#include <rte_eth_tap.h>
#include <tap_flow.h>
+#include <tap_netlink.h>
#include <tap_tcmsgs.h>
/* Linux based path to the TUN device */
#define ETH_TAP_SPEED_ARG "speed"
#define ETH_TAP_REMOTE_ARG "remote"
-#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)
tap_trigger = (tap_trigger + 1) | 0x80000000;
}
+/* Specifies on what netdevices the ioctl should be applied */
+enum ioctl_mode {
+ LOCAL_AND_REMOTE,
+ LOCAL_ONLY,
+ REMOTE_ONLY,
+};
+
static int
tap_ioctl(struct pmd_internals *pmd, unsigned long request,
- struct ifreq *ifr, int set);
+ struct ifreq *ifr, int set, enum ioctl_mode mode);
+
+static int tap_intr_handle_set(struct rte_eth_dev *dev, int set);
/* Tun/Tap allocation routine
*
if (qid == 0) {
struct ifreq ifr;
- if (tap_ioctl(pmd, SIOCGIFHWADDR, &ifr, 0) < 0)
- goto error;
- rte_memcpy(&pmd->eth_addr, ifr.ifr_hwaddr.sa_data,
+ /*
+ * pmd->eth_addr contains the desired MAC, either from remote
+ * or from a random assignment. Sync it with the tap netdevice.
+ */
+ 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;
pmd->if_index = if_nametoindex(pmd->name);
if (!pmd->if_index) {
static int
tap_ioctl(struct pmd_internals *pmd, unsigned long request,
- struct ifreq *ifr, int set)
+ struct ifreq *ifr, int set, enum ioctl_mode mode)
{
short req_flags = ifr->ifr_flags;
- int remote = !!pmd->remote_if_index;
+ int remote = pmd->remote_if_index &&
+ (mode == REMOTE_ONLY || mode == LOCAL_AND_REMOTE);
+ if (!pmd->remote_if_index && mode == REMOTE_ONLY)
+ return 0;
/*
* If there is a remote netdevice, apply ioctl on it, then apply it on
* the tap netdevice.
apply:
if (remote)
snprintf(ifr->ifr_name, IFNAMSIZ, "%s", pmd->remote_iface);
- else
+ else if (mode == LOCAL_ONLY || mode == LOCAL_AND_REMOTE)
snprintf(ifr->ifr_name, IFNAMSIZ, "%s", pmd->name);
switch (request) {
case SIOCSIFFLAGS:
else
ifr->ifr_flags &= ~req_flags;
break;
+ case SIOCGIFFLAGS:
case SIOCGIFHWADDR:
- /* Set remote MAC on the tap netdevice */
- if (!remote && pmd->remote_if_index) {
- request = SIOCSIFHWADDR;
- goto apply;
- }
- break;
case SIOCSIFHWADDR:
case SIOCSIFMTU:
break;
}
if (ioctl(pmd->ioctl_sock, request, ifr) < 0)
goto error;
- if (remote--)
+ if (remote-- && mode == LOCAL_AND_REMOTE)
goto apply;
return 0;
struct ifreq ifr = { .ifr_flags = IFF_UP };
dev->data->dev_link.link_status = ETH_LINK_DOWN;
- return tap_ioctl(pmd, SIOCSIFFLAGS, &ifr, 0);
+ return tap_ioctl(pmd, SIOCSIFFLAGS, &ifr, 0, LOCAL_AND_REMOTE);
}
static int
struct ifreq ifr = { .ifr_flags = IFF_UP };
dev->data->dev_link.link_status = ETH_LINK_UP;
- return tap_ioctl(pmd, SIOCSIFFLAGS, &ifr, 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);
}
static void
tap_dev_stop(struct rte_eth_dev *dev)
{
+ tap_intr_handle_set(dev, 0);
tap_link_set_down(dev);
}
}
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;
}
struct ifreq ifr = { .ifr_flags = IFF_PROMISC };
dev->data->promiscuous = 1;
- tap_ioctl(pmd, SIOCSIFFLAGS, &ifr, 1);
+ tap_ioctl(pmd, SIOCSIFFLAGS, &ifr, 1, LOCAL_AND_REMOTE);
if (pmd->remote_if_index)
tap_flow_implicit_create(pmd, TAP_REMOTE_PROMISC);
}
struct ifreq ifr = { .ifr_flags = IFF_PROMISC };
dev->data->promiscuous = 0;
- tap_ioctl(pmd, SIOCSIFFLAGS, &ifr, 0);
+ tap_ioctl(pmd, SIOCSIFFLAGS, &ifr, 0, LOCAL_AND_REMOTE);
if (pmd->remote_if_index)
tap_flow_implicit_destroy(pmd, TAP_REMOTE_PROMISC);
}
struct ifreq ifr = { .ifr_flags = IFF_ALLMULTI };
dev->data->all_multicast = 1;
- tap_ioctl(pmd, SIOCSIFFLAGS, &ifr, 1);
+ tap_ioctl(pmd, SIOCSIFFLAGS, &ifr, 1, LOCAL_AND_REMOTE);
if (pmd->remote_if_index)
tap_flow_implicit_create(pmd, TAP_REMOTE_ALLMULTI);
}
struct ifreq ifr = { .ifr_flags = IFF_ALLMULTI };
dev->data->all_multicast = 0;
- tap_ioctl(pmd, SIOCSIFFLAGS, &ifr, 0);
+ tap_ioctl(pmd, SIOCSIFFLAGS, &ifr, 0, LOCAL_AND_REMOTE);
if (pmd->remote_if_index)
tap_flow_implicit_destroy(pmd, TAP_REMOTE_ALLMULTI);
}
dev->data->name);
return;
}
+ /* Check the actual current MAC address on the tap netdevice */
+ if (tap_ioctl(pmd, SIOCGIFHWADDR, &ifr, 0, LOCAL_ONLY) != 0) {
+ RTE_LOG(ERR, PMD,
+ "%s: couldn't check current tap MAC address\n",
+ dev->data->name);
+ return;
+ }
+ if (is_same_ether_addr((struct ether_addr *)&ifr.ifr_hwaddr.sa_data,
+ mac_addr))
+ return;
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) < 0)
+ if (tap_ioctl(pmd, SIOCSIFHWADDR, &ifr, 1, LOCAL_AND_REMOTE) < 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
struct ifreq ifr;
ifr.ifr_mtu = dev->data->mtu;
- if (tap_ioctl(pmd, SIOCSIFMTU, &ifr, 1) < 0) {
+ if (tap_ioctl(pmd, SIOCSIFMTU, &ifr, 1,
+ LOCAL_AND_REMOTE) < 0) {
close(fd);
return -1;
}
struct rte_mbuf **tmp = &rxq->pool;
struct iovec (*iovecs)[nb_rx_desc + 1];
int data_off = RTE_PKTMBUF_HEADROOM;
- uint16_t buf_size;
int ret = 0;
int fd;
int i;
}
rxq->iovecs = iovecs;
- /* 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) {
- RTE_LOG(WARNING, PMD,
- "%s: %d bytes will not fit in mbuf (%d bytes)\n",
- dev->data->name, ETH_FRAME_LEN, buf_size);
- ret = -ENOMEM;
- goto error;
- }
-
fd = rx_setup_queue(dev, internals, rx_queue_id);
if (fd == -1) {
ret = fd;
struct ifreq ifr = { .ifr_mtu = mtu };
int err = 0;
- err = tap_ioctl(pmd, SIOCSIFMTU, &ifr, 1);
+ err = tap_ioctl(pmd, SIOCSIFMTU, &ifr, 1, LOCAL_AND_REMOTE);
if (!err)
dev->data->mtu = mtu;
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(struct rte_intr_handle *intr_handle __rte_unused,
+ 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)
{
data->dev_private = pmd;
data->port_id = dev->data->port_id;
data->mtu = dev->data->mtu;
- data->dev_flags = RTE_ETH_DEV_DETACHABLE;
+ data->dev_flags = RTE_ETH_DEV_DETACHABLE | RTE_ETH_DEV_INTR_LSC;
data->kdrv = RTE_KDRV_NONE;
data->drv_name = pmd_tap_drv.driver.name;
data->numa_node = numa_node;
dev->rx_pkt_burst = pmd_rx_burst;
dev->tx_pkt_burst = pmd_tx_burst;
+ 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++) {
pmd->rxq[i].fd = -1;
* If no netlink socket can be created, then it will fail when
* creating/destroying flow rules.
*/
- pmd->nlsk_fd = nl_init();
+ pmd->nlsk_fd = nl_init(0);
if (strlen(remote_iface)) {
+ struct ifreq ifr;
+
pmd->remote_if_index = if_nametoindex(remote_iface);
snprintf(pmd->remote_iface, RTE_ETH_NAME_MAX_LEN,
"%s", remote_iface);
- if (!pmd->remote_if_index)
+ if (!pmd->remote_if_index) {
RTE_LOG(ERR, PMD, "Could not find %s ifindex: "
"remote interface will remain unconfigured\n",
remote_iface);
+ return 0;
+ }
+ if (tap_ioctl(pmd, SIOCGIFHWADDR, &ifr, 0, REMOTE_ONLY) < 0) {
+ RTE_LOG(ERR, PMD, "Could not get remote MAC address\n");
+ goto error_exit;
+ }
+ rte_memcpy(&pmd->eth_addr, ifr.ifr_hwaddr.sa_data,
+ ETHER_ADDR_LEN);
+ } else {
+ eth_random_addr((uint8_t *)&pmd->eth_addr);
}
return 0;