#include "daemon.h"
#include "dirs.h"
#include "dpif.h"
+#include "dp-packet.h"
#include "dynamic-string.h"
#include "flow.h"
#include "hash.h"
#include "list.h"
#include "netdev-provider.h"
#include "odp-netlink.h"
-#include "ofpbuf.h"
+#include "dp-packet.h"
#include "ovs-router.h"
#include "packets.h"
-#include "packet-dpif.h"
#include "poll-loop.h"
#include "route-table.h"
#include "shash.h"
sizeof(struct udp_header) + \
sizeof(struct vxlanhdr))
+#define GENEVE_BASE_HLEN (sizeof(struct eth_header) + \
+ sizeof(struct ip_header) + \
+ sizeof(struct udp_header) + \
+ sizeof(struct genevehdr))
+
#define DEFAULT_TTL 64
struct netdev_vport {
struct netdev_tunnel_config tnl_cfg;
struct smap_node *node;
- has_csum = strstr(type, "gre");
+ has_csum = strstr(type, "gre") || strstr(type, "geneve") ||
+ strstr(type, "vxlan");
ipsec_mech_set = false;
memset(&tnl_cfg, 0, sizeof tnl_cfg);
!strcmp(node->key, "in_key") ||
!strcmp(node->key, "out_key")) {
/* Handled separately below. */
+ } else if (!strcmp(node->key, "exts")) {
+ char *str = xstrdup(node->value);
+ char *ext, *save_ptr = NULL;
+
+ tnl_cfg.exts = 0;
+
+ ext = strtok_r(str, ",", &save_ptr);
+ while (ext) {
+ if (!strcmp(type, "vxlan") && !strcmp(ext, "gbp")) {
+ tnl_cfg.exts |= (1 << OVS_VXLAN_EXT_GBP);
+ } else {
+ VLOG_WARN("%s: unknown extension '%s'", name, ext);
+ }
+
+ ext = strtok_r(NULL, ",", &save_ptr);
+ }
+
+ free(str);
} else {
VLOG_WARN("%s: unknown %s argument '%s'", name, type, node->key);
}
&tnl_cfg.out_key_flow);
ovs_mutex_lock(&dev->mutex);
- dev->tnl_cfg = tnl_cfg;
- tunnel_check_status_change__(dev);
- netdev_change_seq_changed(dev_);
+ if (memcmp(&dev->tnl_cfg, &tnl_cfg, sizeof tnl_cfg)) {
+ dev->tnl_cfg = tnl_cfg;
+ tunnel_check_status_change__(dev);
+ netdev_change_seq_changed(dev_);
+ }
ovs_mutex_unlock(&dev->mutex);
return 0;
}
ovs_mutex_lock(&dev->mutex);
- free(dev->peer);
- dev->peer = xstrdup(peer);
- netdev_change_seq_changed(dev_);
+ if (!dev->peer || strcmp(dev->peer, peer)) {
+ free(dev->peer);
+ dev->peer = xstrdup(peer);
+ netdev_change_seq_changed(dev_);
+ }
ovs_mutex_unlock(&dev->mutex);
return 0;
}
static void *
-ip_extract_tnl_md(struct ofpbuf *packet, struct flow_tnl *tnl)
+ip_extract_tnl_md(struct dp_packet *packet, struct flow_tnl *tnl)
{
struct ip_header *nh;
void *l4;
- nh = ofpbuf_l3(packet);
- l4 = ofpbuf_l4(packet);
+ nh = dp_packet_l3(packet);
+ l4 = dp_packet_l4(packet);
if (!nh || !l4) {
return NULL;
tnl->ip_src = get_16aligned_be32(&nh->ip_src);
tnl->ip_dst = get_16aligned_be32(&nh->ip_dst);
tnl->ip_tos = nh->ip_tos;
+ tnl->ip_ttl = nh->ip_ttl;
return l4;
}
*
* Return pointer to the L4 header added to 'packet'. */
static void *
-push_ip_header(struct ofpbuf *packet,
+push_ip_header(struct dp_packet *packet,
const void *header, int size, int *ip_tot_size)
{
struct eth_header *eth;
struct ip_header *ip;
- eth = ofpbuf_push_uninit(packet, size);
- *ip_tot_size = ofpbuf_size(packet) - sizeof (struct eth_header);
+ eth = dp_packet_push_uninit(packet, size);
+ *ip_tot_size = dp_packet_size(packet) - sizeof (struct eth_header);
memcpy(eth, header, size);
ip = ip_hdr(eth);
return ip + 1;
}
+static void *
+udp_extract_tnl_md(struct dp_packet *packet, struct flow_tnl *tnl)
+{
+ struct udp_header *udp;
+
+ udp = ip_extract_tnl_md(packet, tnl);
+ if (!udp) {
+ return NULL;
+ }
+
+ tnl->tp_src = udp->udp_src;
+ tnl->tp_dst = udp->udp_dst;
+
+ return udp + 1;
+}
+
+static ovs_be16
+get_src_port(struct dp_packet *packet)
+{
+ uint32_t hash;
+
+ hash = dp_packet_get_dp_hash(packet);
+
+ return htons((((uint64_t) hash * (tnl_udp_port_max - tnl_udp_port_min)) >> 32) +
+ tnl_udp_port_min);
+}
+
+static void *
+push_udp_header(struct dp_packet *packet, const void *header, int size)
+{
+ struct udp_header *udp;
+ int ip_tot_size;
+
+ udp = push_ip_header(packet, header, size, &ip_tot_size);
+
+ /* set udp src port */
+ udp->udp_src = get_src_port(packet);
+ udp->udp_len = htons(ip_tot_size - sizeof (struct ip_header));
+ /* udp_csum is zero */
+
+ return udp + 1;
+}
+
+static void *
+udp_build_header(struct netdev_tunnel_config *tnl_cfg,
+ struct ovs_action_push_tnl *data)
+{
+ struct ip_header *ip;
+ struct udp_header *udp;
+
+ ip = ip_hdr(data->header);
+ ip->ip_proto = IPPROTO_UDP;
+
+ udp = (struct udp_header *) (ip + 1);
+ udp->udp_dst = tnl_cfg->dst_port;
+
+ return udp + 1;
+}
+
static int
gre_header_len(ovs_be16 flags)
{
}
static int
-parse_gre_header(struct ofpbuf *packet,
+parse_gre_header(struct dp_packet *packet,
struct flow_tnl *tnl)
{
const struct gre_base_hdr *greh;
return -EINVAL;
}
+ if (greh->protocol != htons(ETH_TYPE_TEB)) {
+ return -EINVAL;
+ }
+
hlen = gre_header_len(greh->flags);
- if (hlen > ofpbuf_size(packet)) {
+ if (hlen > dp_packet_size(packet)) {
return -EINVAL;
}
if (greh->flags & htons(GRE_CSUM)) {
ovs_be16 pkt_csum;
- pkt_csum = csum(greh, ofpbuf_size(packet) -
+ pkt_csum = csum(greh, dp_packet_size(packet) -
((const unsigned char *)greh -
- (const unsigned char *)ofpbuf_l2(packet)));
+ (const unsigned char *)dp_packet_l2(packet)));
if (pkt_csum) {
return -EINVAL;
}
}
static void
-gre_extract_md(struct dpif_packet *dpif_pkt)
+gre_extract_md(struct dp_packet *packet)
{
- struct ofpbuf *packet = &dpif_pkt->ofpbuf;
- struct pkt_metadata *md = &dpif_pkt->md;
+ struct pkt_metadata *md = &packet->md;
struct flow_tnl *tnl = &md->tunnel;
int hlen = sizeof(struct eth_header) +
sizeof(struct ip_header) + 4;
memset(md, 0, sizeof *md);
- if (hlen > ofpbuf_size(packet)) {
+ if (hlen > dp_packet_size(packet)) {
return;
}
reset_tnl_md(md);
}
- ofpbuf_reset_packet(packet, hlen);
+ dp_packet_reset_packet(packet, hlen);
}
static int
netdev_gre_pop_header(struct netdev *netdev_ OVS_UNUSED,
- struct dpif_packet **pkt, int cnt)
+ struct dp_packet **pkt, int cnt)
{
int i;
}
static void
-netdev_gre_push_header__(struct ofpbuf *packet,
+netdev_gre_push_header__(struct dp_packet *packet,
const void *header, int size)
{
struct gre_base_hdr *greh;
greh = push_ip_header(packet, header, size, &ip_tot_size);
if (greh->flags & htons(GRE_CSUM)) {
- ovs_16aligned_be32 *options = (ovs_16aligned_be32 *) (greh + 1);
-
- put_16aligned_be32(options,
- (OVS_FORCE ovs_be32) csum(greh, ip_tot_size - sizeof (struct ip_header)));
+ ovs_be16 *csum_opt = (ovs_be16 *) (greh + 1);
+ *csum_opt = csum(greh, ip_tot_size - sizeof (struct ip_header));
}
}
static int
netdev_gre_push_header(const struct netdev *netdev OVS_UNUSED,
- struct dpif_packet **packets, int cnt,
+ struct dp_packet **packets, int cnt,
const struct ovs_action_push_tnl *data)
{
int i;
for (i = 0; i < cnt; i++) {
- netdev_gre_push_header__(&packets[i]->ofpbuf,
- data->header, data->header_len);
+ netdev_gre_push_header__(packets[i], data->header, data->header_len);
packets[i]->md = PKT_METADATA_INITIALIZER(u32_to_odp(data->out_port));
}
return 0;
static int
netdev_gre_build_header(const struct netdev *netdev,
- struct ovs_action_push_tnl *data)
+ struct ovs_action_push_tnl *data,
+ const struct flow *tnl_flow)
{
struct netdev_vport *dev = netdev_vport_cast(netdev);
struct netdev_tunnel_config *tnl_cfg;
greh->flags = 0;
options = (ovs_16aligned_be32 *) (greh + 1);
- if (tnl_cfg->csum) {
+ if (tnl_flow->tunnel.flags & FLOW_TNL_F_CSUM) {
greh->flags |= htons(GRE_CSUM);
put_16aligned_be32(options, 0);
options++;
if (tnl_cfg->out_key_present) {
greh->flags |= htons(GRE_KEY);
put_16aligned_be32(options, (OVS_FORCE ovs_be32)
- ((OVS_FORCE uint64_t) tnl_cfg->out_key >> 32));
+ ((OVS_FORCE uint64_t) tnl_flow->tunnel.tun_id >> 32));
options++;
}
}
static void
-vxlan_extract_md(struct dpif_packet *dpif_pkt)
+vxlan_extract_md(struct dp_packet *packet)
{
- struct ofpbuf *packet = &dpif_pkt->ofpbuf;
- struct pkt_metadata *md = &dpif_pkt->md;
+ struct pkt_metadata *md = &packet->md;
struct flow_tnl *tnl = &md->tunnel;
- struct udp_header *udp;
struct vxlanhdr *vxh;
memset(md, 0, sizeof *md);
- if (VXLAN_HLEN > ofpbuf_size(packet)) {
+ if (VXLAN_HLEN > dp_packet_size(packet)) {
return;
}
- udp = ip_extract_tnl_md(packet, tnl);
- if (!udp) {
+ vxh = udp_extract_tnl_md(packet, tnl);
+ if (!vxh) {
return;
}
- vxh = (struct vxlanhdr *) (udp + 1);
if (get_16aligned_be32(&vxh->vx_flags) != htonl(VXLAN_FLAGS) ||
(get_16aligned_be32(&vxh->vx_vni) & htonl(0xff))) {
reset_tnl_md(md);
return;
}
- tnl->tp_src = udp->udp_src;
- tnl->tp_dst = udp->udp_dst;
tnl->tun_id = htonll(ntohl(get_16aligned_be32(&vxh->vx_vni)) >> 8);
+ tnl->flags |= FLOW_TNL_F_KEY;
- ofpbuf_reset_packet(packet, VXLAN_HLEN);
+ dp_packet_reset_packet(packet, VXLAN_HLEN);
}
static int
netdev_vxlan_pop_header(struct netdev *netdev_ OVS_UNUSED,
- struct dpif_packet **pkt, int cnt)
+ struct dp_packet **pkt, int cnt)
{
int i;
static int
netdev_vxlan_build_header(const struct netdev *netdev,
- struct ovs_action_push_tnl *data)
+ struct ovs_action_push_tnl *data,
+ const struct flow *tnl_flow)
{
struct netdev_vport *dev = netdev_vport_cast(netdev);
struct netdev_tunnel_config *tnl_cfg;
- struct ip_header *ip;
- struct udp_header *udp;
struct vxlanhdr *vxh;
/* XXX: RCUfy tnl_cfg. */
ovs_mutex_lock(&dev->mutex);
tnl_cfg = &dev->tnl_cfg;
- ip = ip_hdr(data->header);
- ip->ip_proto = IPPROTO_UDP;
-
- udp = (struct udp_header *) (ip + 1);
- udp->udp_dst = tnl_cfg->dst_port;
+ vxh = udp_build_header(tnl_cfg, data);
- vxh = (struct vxlanhdr *) (udp + 1);
put_16aligned_be32(&vxh->vx_flags, htonl(VXLAN_FLAGS));
- put_16aligned_be32(&vxh->vx_vni, htonl(ntohll(tnl_cfg->out_key) << 8));
+ put_16aligned_be32(&vxh->vx_vni, htonl(ntohll(tnl_flow->tunnel.tun_id) << 8));
ovs_mutex_unlock(&dev->mutex);
data->header_len = VXLAN_HLEN;
return 0;
}
-static ovs_be16
-get_src_port(struct dpif_packet *packet)
+static int
+netdev_vxlan_push_header(const struct netdev *netdev OVS_UNUSED,
+ struct dp_packet **packets, int cnt,
+ const struct ovs_action_push_tnl *data)
{
- uint32_t hash;
-
- hash = dpif_packet_get_dp_hash(packet);
+ int i;
- return htons((((uint64_t) hash * (tnl_udp_port_max - tnl_udp_port_min)) >> 32) +
- tnl_udp_port_min);
+ for (i = 0; i < cnt; i++) {
+ push_udp_header(packets[i], data->header, VXLAN_HLEN);
+ packets[i]->md = PKT_METADATA_INITIALIZER(u32_to_odp(data->out_port));
+ }
+ return 0;
}
static void
-netdev_vxlan_push_header__(struct dpif_packet *packet,
- const void *header, int size)
+geneve_extract_md(struct dp_packet *packet)
{
- struct udp_header *udp;
- int ip_tot_size;
+ struct pkt_metadata *md = &packet->md;
+ struct flow_tnl *tnl = &md->tunnel;
+ struct genevehdr *gnh;
+ unsigned int hlen;
- udp = push_ip_header(&packet->ofpbuf, header, size, &ip_tot_size);
+ memset(md, 0, sizeof *md);
+ if (GENEVE_BASE_HLEN > dp_packet_size(packet)) {
+ VLOG_WARN_RL(&err_rl, "geneve packet too small: min header=%u packet size=%u\n",
+ (unsigned int)GENEVE_BASE_HLEN, dp_packet_size(packet));
+ return;
+ }
- /* set udp src port */
- udp->udp_src = get_src_port(packet);
- udp->udp_len = htons(ip_tot_size - sizeof (struct ip_header));
- /* udp_csum is zero */
+ gnh = udp_extract_tnl_md(packet, tnl);
+ if (!gnh) {
+ return;
+ }
+
+ hlen = GENEVE_BASE_HLEN + gnh->opt_len * 4;
+ if (hlen > dp_packet_size(packet)) {
+ VLOG_WARN_RL(&err_rl, "geneve packet too small: header len=%u packet size=%u\n",
+ hlen, dp_packet_size(packet));
+ reset_tnl_md(md);
+ return;
+ }
+
+ if (gnh->ver != 0) {
+ VLOG_WARN_RL(&err_rl, "unknown geneve version: %"PRIu8"\n", gnh->ver);
+ reset_tnl_md(md);
+ return;
+ }
+
+ if (gnh->opt_len && gnh->critical) {
+ VLOG_WARN_RL(&err_rl, "unknown geneve critical options: %"PRIu8" bytes\n",
+ gnh->opt_len * 4);
+ reset_tnl_md(md);
+ return;
+ }
+
+ if (gnh->proto_type != htons(ETH_TYPE_TEB)) {
+ VLOG_WARN_RL(&err_rl, "unknown geneve encapsulated protocol: %#x\n",
+ ntohs(gnh->proto_type));
+ reset_tnl_md(md);
+ return;
+ }
+
+ tnl->flags |= gnh->oam ? FLOW_TNL_F_OAM : 0;
+ tnl->tun_id = htonll(ntohl(get_16aligned_be32(&gnh->vni)) >> 8);
+ tnl->flags |= FLOW_TNL_F_KEY;
+
+ dp_packet_reset_packet(packet, hlen);
}
static int
-netdev_vxlan_push_header(const struct netdev *netdev OVS_UNUSED,
- struct dpif_packet **packets, int cnt,
- const struct ovs_action_push_tnl *data)
+netdev_geneve_pop_header(struct netdev *netdev_ OVS_UNUSED,
+ struct dp_packet **pkt, int cnt)
+{
+ int i;
+
+ for (i = 0; i < cnt; i++) {
+ geneve_extract_md(pkt[i]);
+ }
+ return 0;
+}
+
+static int
+netdev_geneve_build_header(const struct netdev *netdev,
+ struct ovs_action_push_tnl *data,
+ const struct flow *tnl_flow)
+{
+ struct netdev_vport *dev = netdev_vport_cast(netdev);
+ struct netdev_tunnel_config *tnl_cfg;
+ struct genevehdr *gnh;
+
+ /* XXX: RCUfy tnl_cfg. */
+ ovs_mutex_lock(&dev->mutex);
+ tnl_cfg = &dev->tnl_cfg;
+
+ gnh = udp_build_header(tnl_cfg, data);
+
+ gnh->oam = !!(tnl_flow->tunnel.flags & FLOW_TNL_F_OAM);
+ gnh->proto_type = htons(ETH_TYPE_TEB);
+ put_16aligned_be32(&gnh->vni, htonl(ntohll(tnl_flow->tunnel.tun_id) << 8));
+
+ ovs_mutex_unlock(&dev->mutex);
+ data->header_len = GENEVE_BASE_HLEN;
+ data->tnl_type = OVS_VPORT_TYPE_GENEVE;
+ return 0;
+}
+
+static int
+netdev_geneve_push_header(const struct netdev *netdev OVS_UNUSED,
+ struct dp_packet **packets, int cnt,
+ const struct ovs_action_push_tnl *data)
{
int i;
for (i = 0; i < cnt; i++) {
- netdev_vxlan_push_header__(packets[i],
- data->header, VXLAN_HLEN);
+ push_udp_header(packets[i], data->header, data->header_len);
packets[i]->md = PKT_METADATA_INITIALIZER(u32_to_odp(data->out_port));
}
return 0;
/* The name of the dpif_port should be short enough to accomodate adding
* a port number to the end if one is necessary. */
static const struct vport_class vport_classes[] = {
- TUNNEL_CLASS("geneve", "genev_sys", NULL, NULL, NULL),
+ TUNNEL_CLASS("geneve", "genev_sys", netdev_geneve_build_header,
+ netdev_geneve_push_header,
+ netdev_geneve_pop_header),
TUNNEL_CLASS("gre", "gre_sys", netdev_gre_build_header,
netdev_gre_push_header,
netdev_gre_pop_header),