Revert "netifd: Link layer state support on device level"
[project/netifd.git] / system-linux.c
index 604c206..db78240 100644 (file)
 #include <linux/rtnetlink.h>
 #include <linux/sockios.h>
 #include <linux/ip.h>
+#include <linux/if_link.h>
 #include <linux/if_vlan.h>
 #include <linux/if_bridge.h>
 #include <linux/if_tunnel.h>
+#include <linux/ip6_tunnel.h>
 #include <linux/ethtool.h>
 #include <linux/fib_rules.h>
 
-#include <unistd.h>
+#ifndef RTN_FAILED_POLICY
+#define RTN_FAILED_POLICY 12
+#endif
+
 #include <string.h>
 #include <fcntl.h>
 #include <glob.h>
@@ -581,11 +586,9 @@ int system_bridge_addbr(struct device *bridge, struct bridge_config *cfg)
        system_set_dev_sysctl("/sys/devices/virtual/net/%s/bridge/multicast_snooping",
                bridge->ifname, cfg->igmp_snoop ? "1" : "0");
 
-       if (cfg->flags & BRIDGE_OPT_PRIORITY) {
-               args[0] = BRCTL_SET_BRIDGE_PRIORITY;
-               args[1] = cfg->priority;
-               system_bridge_if(bridge->ifname, NULL, SIOCDEVPRIVATE, &args);
-       }
+       args[0] = BRCTL_SET_BRIDGE_PRIORITY;
+       args[1] = cfg->priority;
+       system_bridge_if(bridge->ifname, NULL, SIOCDEVPRIVATE, &args);
 
        if (cfg->flags & BRIDGE_OPT_AGEING_TIME) {
                args[0] = BRCTL_SET_AGEING_TIME;
@@ -608,6 +611,93 @@ int system_bridge_addbr(struct device *bridge, struct bridge_config *cfg)
        return 0;
 }
 
+int system_macvlan_add(struct device *macvlan, struct device *dev, struct macvlan_config *cfg)
+{
+       struct nl_msg *msg;
+       struct nlattr *linkinfo, *data;
+       struct ifinfomsg iim = { .ifi_family = AF_INET };
+       int ifindex = system_if_resolve(dev);
+       int i, rv;
+       static const struct {
+               const char *name;
+               enum macvlan_mode val;
+       } modes[] = {
+               { "private", MACVLAN_MODE_PRIVATE },
+               { "vepa", MACVLAN_MODE_VEPA },
+               { "bridge", MACVLAN_MODE_BRIDGE },
+               { "passthru", MACVLAN_MODE_PASSTHRU },
+       };
+
+       if (ifindex == 0)
+               return -ENOENT;
+
+       msg = nlmsg_alloc_simple(RTM_NEWLINK, NLM_F_REQUEST | NLM_F_CREATE | NLM_F_EXCL);
+
+       if (!msg)
+               return -1;
+
+       nlmsg_append(msg, &iim, sizeof(iim), 0);
+
+       if (cfg->flags & MACVLAN_OPT_MACADDR)
+               nla_put(msg, IFLA_ADDRESS, sizeof(cfg->macaddr), cfg->macaddr);
+       nla_put(msg, IFLA_IFNAME, IFNAMSIZ, macvlan->ifname);
+       nla_put_u32(msg, IFLA_LINK, ifindex);
+
+       if (!(linkinfo = nla_nest_start(msg, IFLA_LINKINFO)))
+               goto nla_put_failure;
+
+       nla_put(msg, IFLA_INFO_KIND, strlen("macvlan"), "macvlan");
+
+       if (!(data = nla_nest_start(msg, IFLA_INFO_DATA)))
+               goto nla_put_failure;
+
+       if (cfg->mode) {
+               for (i = 0; i < ARRAY_SIZE(modes); i++) {
+                       if (strcmp(cfg->mode, modes[i].name) != 0)
+                               continue;
+
+                       nla_put_u32(msg, IFLA_MACVLAN_MODE, modes[i].val);
+                       break;
+               }
+       }
+
+       nla_nest_end(msg, data);
+       nla_nest_end(msg, linkinfo);
+
+       rv = system_rtnl_call(msg);
+       if (rv)
+               D(SYSTEM, "Error adding macvlan '%s' over '%s': %d\n", macvlan->ifname, dev->ifname, rv);
+
+       return rv;
+
+nla_put_failure:
+       nlmsg_free(msg);
+       return -ENOMEM;
+}
+
+int system_macvlan_del(struct device *macvlan)
+{
+       struct nl_msg *msg;
+       struct ifinfomsg iim;
+
+       iim.ifi_family = AF_INET;
+       iim.ifi_index  = 0;
+
+       msg = nlmsg_alloc_simple(RTM_DELLINK, 0);
+
+       if (!msg)
+               return -1;
+
+       nlmsg_append(msg, &iim, sizeof(iim), 0);
+
+       nla_put(msg, IFLA_INFO_KIND, strlen("macvlan"), "macvlan");
+       nla_put(msg, IFLA_IFNAME, sizeof(macvlan->ifname), macvlan->ifname);
+
+       system_rtnl_call(msg);
+
+       return 0;
+}
+
 static int system_vlan(struct device *dev, int id)
 {
        struct vlan_ioctl_args ifr = {
@@ -919,17 +1009,21 @@ static int system_addr(struct device *dev, struct device_addr *addr, int cmd)
                struct ifa_cacheinfo cinfo = {0xffffffffU, 0xffffffffU, 0, 0};
 
                if (addr->preferred_until) {
-                       int preferred = addr->preferred_until - now;
+                       int64_t preferred = addr->preferred_until - now;
                        if (preferred < 0)
                                preferred = 0;
+                       else if (preferred > UINT32_MAX)
+                               preferred = UINT32_MAX;
 
                        cinfo.ifa_prefered = preferred;
                }
 
                if (addr->valid_until) {
-                       int valid = addr->valid_until - now;
+                       int64_t valid = addr->valid_until - now;
                        if (valid <= 0)
                                return -1;
+                       else if (valid > UINT32_MAX)
+                               valid = UINT32_MAX;
 
                        cinfo.ifa_valid = valid;
                }
@@ -967,11 +1061,13 @@ static int system_rt(struct device *dev, struct device_route *route, int cmd)
        unsigned char scope = (cmd == RTM_DELROUTE) ? RT_SCOPE_NOWHERE :
                        (have_gw) ? RT_SCOPE_UNIVERSE : RT_SCOPE_LINK;
 
-       unsigned int table = (route->flags & DEVROUTE_TABLE) ? route->table : RT_TABLE_MAIN;
+       unsigned int table = (route->flags & (DEVROUTE_TABLE | DEVROUTE_SRCTABLE))
+                       ? route->table : RT_TABLE_MAIN;
 
        struct rtmsg rtm = {
                .rtm_family = (alen == 4) ? AF_INET : AF_INET6,
                .rtm_dst_len = route->mask,
+               .rtm_src_len = route->sourcemask,
                .rtm_table = (table < 256) ? table : RT_TABLE_UNSPEC,
                .rtm_protocol = (route->flags & DEVADDR_KERNEL) ? RTPROT_KERNEL : RTPROT_STATIC,
                .rtm_scope = scope,
@@ -997,6 +1093,9 @@ static int system_rt(struct device *dev, struct device_route *route, int cmd)
        if (route->mask)
                nla_put(msg, RTA_DST, alen, &route->addr);
 
+       if (route->sourcemask)
+               nla_put(msg, RTA_SRC, alen, &route->source);
+
        if (route->metric > 0)
                nla_put_u32(msg, RTA_PRIORITY, route->metric);
 
@@ -1251,6 +1350,8 @@ bool system_resolve_iprule_action(const char *action, unsigned int *id)
                n = RTN_UNICAST;
        else if (!strcmp(action, "throw"))
                n = RTN_THROW;
+       else if (!strcmp(action, "failed_policy"))
+               n = RTN_FAILED_POLICY;
        else {
                n = strtoul(action, &e, 0);
                if (!e || *e || e == action || n > 255)
@@ -1279,14 +1380,6 @@ time_t system_get_rtime(void)
 #define IP_DF       0x4000
 #endif
 
-static void tunnel_parm_init(struct ip_tunnel_parm *p)
-{
-       memset(p, 0, sizeof(*p));
-       p->iph.version = 4;
-       p->iph.ihl = 5;
-       p->iph.frag_off = htons(IP_DF);
-}
-
 static int tunnel_ioctl(const char *name, int cmd, void *p)
 {
        struct ifreq ifr;
@@ -1299,10 +1392,7 @@ static int tunnel_ioctl(const char *name, int cmd, void *p)
 
 int system_del_ip_tunnel(const char *name)
 {
-       struct ip_tunnel_parm p;
-
-       tunnel_parm_init(&p);
-       return tunnel_ioctl(name, SIOCDELTUNNEL, &p);
+       return tunnel_ioctl(name, SIOCDELTUNNEL, NULL);
 }
 
 int system_update_ipv6_mtu(struct device *dev, int mtu)
@@ -1332,86 +1422,112 @@ out:
        return ret;
 }
 
-static int parse_ipaddr(struct blob_attr *attr, __be32 *addr)
-{
-       if (!attr)
-               return 1;
-
-       return inet_pton(AF_INET, blobmsg_data(attr), (void *) addr);
-}
-
-
 int system_add_ip_tunnel(const char *name, struct blob_attr *attr)
 {
        struct blob_attr *tb[__TUNNEL_ATTR_MAX];
        struct blob_attr *cur;
-       struct ip_tunnel_parm p;
-       const char *base, *str;
-       bool is_sit;
+       bool set_df = true;
+       const char *str;
 
        system_del_ip_tunnel(name);
 
-       tunnel_parm_init(&p);
-
        blobmsg_parse(tunnel_attr_list.params, __TUNNEL_ATTR_MAX, tb,
                blob_data(attr), blob_len(attr));
 
        if (!(cur = tb[TUNNEL_ATTR_TYPE]))
                return -EINVAL;
        str = blobmsg_data(cur);
-       is_sit = !strcmp(str, "sit");
-
-       if (is_sit) {
-               p.iph.protocol = IPPROTO_IPV6;
-               base = "sit0";
-       } else
-               return -EINVAL;
 
-       if (!parse_ipaddr(tb[TUNNEL_ATTR_LOCAL], &p.iph.saddr))
-               return -EINVAL;
-
-       if (!parse_ipaddr(tb[TUNNEL_ATTR_REMOTE], &p.iph.daddr))
-               return -EINVAL;
+       if ((cur = tb[TUNNEL_ATTR_DF]))
+               set_df = blobmsg_get_bool(cur);
 
+       unsigned int ttl = 0;
        if ((cur = tb[TUNNEL_ATTR_TTL])) {
-               unsigned int val = blobmsg_get_u32(cur);
+               ttl = blobmsg_get_u32(cur);
+               if (ttl > 255 || (!set_df && ttl))
+                       return -EINVAL;
+       }
 
-               if (val > 255)
+       unsigned int link = 0;
+       if ((cur = tb[TUNNEL_ATTR_LINK])) {
+               struct interface *iface = vlist_find(&interfaces, blobmsg_data(cur), iface, node);
+               if (!iface)
                        return -EINVAL;
 
-               p.iph.ttl = val;
+               if (iface->l3_dev.dev)
+                       link = iface->l3_dev.dev->ifindex;
        }
 
-       strncpy(p.name, name, sizeof(p.name));
-       if (tunnel_ioctl(base, SIOCADDTUNNEL, &p) < 0)
-               return -1;
-
-#ifdef SIOCADD6RD
-       cur = tb[TUNNEL_ATTR_6RD_PREFIX];
-       if (cur && is_sit) {
-               unsigned int mask;
-               struct ip_tunnel_6rd p6;
+       if (!strcmp(str, "sit")) {
+               struct ip_tunnel_parm p = {
+                       .link = link,
+                       .iph = {
+                               .version = 4,
+                               .ihl = 5,
+                               .frag_off = set_df ? htons(IP_DF) : 0,
+                               .protocol = IPPROTO_IPV6,
+                               .ttl = ttl
+                       }
+               };
 
-               memset(&p6, 0, sizeof(p6));
+               if ((cur = tb[TUNNEL_ATTR_LOCAL]) &&
+                               inet_pton(AF_INET, blobmsg_data(cur), &p.iph.saddr) < 1)
+                       return -EINVAL;
 
-               if (!parse_ip_and_netmask(AF_INET6, blobmsg_data(cur),
-                                       &p6.prefix, &mask) || mask > 128)
+               if ((cur = tb[TUNNEL_ATTR_REMOTE]) &&
+                               inet_pton(AF_INET, blobmsg_data(cur), &p.iph.daddr) < 1)
                        return -EINVAL;
-               p6.prefixlen = mask;
 
-               if ((cur = tb[TUNNEL_ATTR_6RD_RELAY_PREFIX])) {
-                       if (!parse_ip_and_netmask(AF_INET, blobmsg_data(cur),
-                                               &p6.relay_prefix, &mask) || mask > 32)
+               strncpy(p.name, name, sizeof(p.name));
+               if (tunnel_ioctl("sit0", SIOCADDTUNNEL, &p) < 0)
+                       return -1;
+
+#ifdef SIOCADD6RD
+               if ((cur = tb[TUNNEL_ATTR_6RD_PREFIX])) {
+                       unsigned int mask;
+                       struct ip_tunnel_6rd p6;
+
+                       memset(&p6, 0, sizeof(p6));
+
+                       if (!parse_ip_and_netmask(AF_INET6, blobmsg_data(cur),
+                                               &p6.prefix, &mask) || mask > 128)
                                return -EINVAL;
-                       p6.relay_prefixlen = mask;
-               }
+                       p6.prefixlen = mask;
 
-               if (tunnel_ioctl(name, SIOCADD6RD, &p6) < 0) {
-                       system_del_ip_tunnel(name);
-                       return -1;
+                       if ((cur = tb[TUNNEL_ATTR_6RD_RELAY_PREFIX])) {
+                               if (!parse_ip_and_netmask(AF_INET, blobmsg_data(cur),
+                                                       &p6.relay_prefix, &mask) || mask > 32)
+                                       return -EINVAL;
+                               p6.relay_prefixlen = mask;
+                       }
+
+                       if (tunnel_ioctl(name, SIOCADD6RD, &p6) < 0) {
+                               system_del_ip_tunnel(name);
+                               return -1;
+                       }
                }
-       }
 #endif
+       } else if (!strcmp(str, "ipip6")) {
+               struct ip6_tnl_parm p = {
+                       .link = link,
+                       .proto = IPPROTO_IPIP,
+                       .hop_limit = (ttl) ? ttl : 64,
+                       .encap_limit = 4,
+               };
+
+               if ((cur = tb[TUNNEL_ATTR_LOCAL]) &&
+                               inet_pton(AF_INET6, blobmsg_data(cur), &p.laddr) < 1)
+                       return -EINVAL;
+
+               if ((cur = tb[TUNNEL_ATTR_REMOTE]) &&
+                               inet_pton(AF_INET6, blobmsg_data(cur), &p.raddr) < 1)
+                       return -EINVAL;
+
+               strncpy(p.name, name, sizeof(p.name));
+               if (tunnel_ioctl("ip6tnl0", SIOCADDTUNNEL, &p) < 0)
+                       return -1;
+       } else
+               return -EINVAL;
 
        return 0;
 }