X-Git-Url: http://git.onelab.eu/?a=blobdiff_plain;f=ofproto%2Ftunnel.c;h=c23e2d7695db21ee5f3470dad49dc6d36f2343c9;hb=b0fb94a346e52f36aeef238dd5f9bef9a10c14ef;hp=d50cff8052c6ee28411f0bf3bf294629956115c0;hpb=74a99109015a68a469539d7fa04b8f6c13a437bf;p=sliver-openvswitch.git

diff --git a/ofproto/tunnel.c b/ofproto/tunnel.c
index d50cff805..c23e2d769 100644
--- a/ofproto/tunnel.c
+++ b/ofproto/tunnel.c
@@ -17,12 +17,11 @@
 
 #include <errno.h>
 
-#include "ofproto/ofproto-provider.h"
 #include "byte-order.h"
 #include "dynamic-string.h"
 #include "hash.h"
 #include "hmap.h"
-#include "netdev-vport.h"
+#include "netdev.h"
 #include "odp-util.h"
 #include "packets.h"
 #include "smap.h"
@@ -30,72 +29,79 @@
 #include "tunnel.h"
 #include "vlog.h"
 
-/* XXX:
- *
- * Ability to generate actions on input for ECN
- * Ability to generate metadata for packet-outs
- * IPsec using skb mark.
- * VXLAN.
- * Multicast group management (possibly).
- * Disallow netdevs with names like "gre64_system" to prevent collisions. */
-
 VLOG_DEFINE_THIS_MODULE(tunnel);
 
 struct tnl_match {
     ovs_be64 in_key;
     ovs_be32 ip_src;
     ovs_be32 ip_dst;
-    uint32_t odp_port;
-    bool in_key_present;
+    odp_port_t odp_port;
+    uint32_t skb_mark;
     bool in_key_flow;
+    bool ip_src_flow;
+    bool ip_dst_flow;
 };
 
 struct tnl_port {
+    struct hmap_node ofport_node;
     struct hmap_node match_node;
 
-    const struct ofport *ofport;
+    const struct ofport_dpif *ofport;
     unsigned int netdev_seq;
+    struct netdev *netdev;
+
     struct tnl_match match;
 };
 
-static struct hmap tnl_match_map = HMAP_INITIALIZER(&tnl_match_map);
+static struct ovs_rwlock rwlock = OVS_RWLOCK_INITIALIZER;
 
-/* Returned to callers when their ofport will never be used to receive or send
- * tunnel traffic. Alternatively, we could ask the caller to delete their
- * ofport, but this would be unclean in the reconfguration case.  For the first
- * time, an ofproto provider would have to call ofproto_port_del() on itself.*/
-static struct tnl_port void_tnl_port;
+static struct hmap tnl_match_map__ = HMAP_INITIALIZER(&tnl_match_map__);
+static struct hmap *tnl_match_map OVS_GUARDED_BY(rwlock) = &tnl_match_map__;
+
+static struct hmap ofport_map__ = HMAP_INITIALIZER(&ofport_map__);
+static struct hmap *ofport_map OVS_GUARDED_BY(rwlock) = &ofport_map__;
 
 static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(1, 5);
 static struct vlog_rate_limit dbg_rl = VLOG_RATE_LIMIT_INIT(60, 60);
 
-static struct tnl_port *tnl_find(struct tnl_match *);
-static struct tnl_port *tnl_find_exact(struct tnl_match *);
+static struct tnl_port *tnl_find(struct tnl_match *) OVS_REQ_RDLOCK(rwlock);
+static struct tnl_port *tnl_find_exact(struct tnl_match *)
+    OVS_REQ_RDLOCK(rwlock);
+static struct tnl_port *tnl_find_ofport(const struct ofport_dpif *)
+    OVS_REQ_RDLOCK(rwlock);
+
 static uint32_t tnl_hash(struct tnl_match *);
 static void tnl_match_fmt(const struct tnl_match *, struct ds *);
-static char *tnl_port_fmt(const struct tnl_port *);
-static void tnl_port_mod_log(const struct tnl_port *, const char *action);
-static const char *tnl_port_get_name(const struct tnl_port *);
-
-static struct tnl_port *
-tnl_port_add__(const struct ofport *ofport, uint32_t odp_port,
-               bool warn)
+static char *tnl_port_fmt(const struct tnl_port *) OVS_REQ_RDLOCK(rwlock);
+static void tnl_port_mod_log(const struct tnl_port *, const char *action)
+    OVS_REQ_RDLOCK(rwlock);
+static const char *tnl_port_get_name(const struct tnl_port *)
+    OVS_REQ_RDLOCK(rwlock);
+static void tnl_port_del__(const struct ofport_dpif *) OVS_REQ_WRLOCK(rwlock);
+
+static bool
+tnl_port_add__(const struct ofport_dpif *ofport, const struct netdev *netdev,
+               odp_port_t odp_port, bool warn)
+    OVS_REQ_WRLOCK(rwlock)
 {
     const struct netdev_tunnel_config *cfg;
     struct tnl_port *existing_port;
     struct tnl_port *tnl_port;
 
-    cfg = netdev_get_tunnel_config(ofport->netdev);
+    cfg = netdev_get_tunnel_config(netdev);
     ovs_assert(cfg);
 
     tnl_port = xzalloc(sizeof *tnl_port);
     tnl_port->ofport = ofport;
-    tnl_port->netdev_seq = netdev_change_seq(tnl_port->ofport->netdev);
+    tnl_port->netdev = netdev_ref(netdev);
+    tnl_port->netdev_seq = netdev_change_seq(tnl_port->netdev);
 
     tnl_port->match.in_key = cfg->in_key;
     tnl_port->match.ip_src = cfg->ip_src;
     tnl_port->match.ip_dst = cfg->ip_dst;
-    tnl_port->match.in_key_present = cfg->in_key_present;
+    tnl_port->match.ip_src_flow = cfg->ip_src_flow;
+    tnl_port->match.ip_dst_flow = cfg->ip_dst_flow;
+    tnl_port->match.skb_mark = cfg->ipsec ? IPSEC_MARK : 0;
     tnl_port->match.in_key_flow = cfg->in_key_flow;
     tnl_port->match.odp_port = odp_port;
 
@@ -110,109 +116,121 @@ tnl_port_add__(const struct ofport *ofport, uint32_t odp_port,
             ds_destroy(&ds);
             free(tnl_port);
         }
-        return &void_tnl_port;
+        return false;
     }
 
-    hmap_insert(&tnl_match_map, &tnl_port->match_node,
+    hmap_insert(ofport_map, &tnl_port->ofport_node, hash_pointer(ofport, 0));
+    hmap_insert(tnl_match_map, &tnl_port->match_node,
                 tnl_hash(&tnl_port->match));
     tnl_port_mod_log(tnl_port, "adding");
-    return tnl_port;
+    return true;
 }
 
 /* Adds 'ofport' to the module with datapath port number 'odp_port'. 'ofport's
  * must be added before they can be used by the module. 'ofport' must be a
  * tunnel. */
-struct tnl_port *
-tnl_port_add(const struct ofport *ofport, uint32_t odp_port)
+void
+tnl_port_add(const struct ofport_dpif *ofport, const struct netdev *netdev,
+             odp_port_t odp_port) OVS_EXCLUDED(rwlock)
 {
-    return tnl_port_add__(ofport, odp_port, true);
+    ovs_rwlock_wrlock(&rwlock);
+    tnl_port_add__(ofport, netdev, odp_port, true);
+    ovs_rwlock_unlock(&rwlock);
 }
 
-/* Checks if the tnl_port pointed to by 'tnl_portp' needs reconfiguration due
- * to changes in its netdev_tunnel_config.  If it does, updates 'tnl_portp' to
- * point to a new tnl_port and returns true.  Otherwise, returns false.
- * 'ofport' and 'odp_port' should be the same as would be passed to
+/* Checks if the tunnel represented by 'ofport' reconfiguration due to changes
+ * in its netdev_tunnel_config.  If it does, returns true. Otherwise, returns
+ * false.  'ofport' and 'odp_port' should be the same as would be passed to
  * tnl_port_add(). */
 bool
-tnl_port_reconfigure(const struct ofport *ofport, uint32_t odp_port,
-                     struct tnl_port **tnl_portp)
+tnl_port_reconfigure(const struct ofport_dpif *ofport,
+                     const struct netdev *netdev, odp_port_t odp_port)
+    OVS_EXCLUDED(rwlock)
 {
-    struct tnl_port *tnl_port = *tnl_portp;
+    struct tnl_port *tnl_port;
+    bool changed = false;
 
-    if (tnl_port == &void_tnl_port) {
-        *tnl_portp = tnl_port_add__(ofport, odp_port, false);
-        return *tnl_portp != &void_tnl_port;
-    } else if (tnl_port->ofport != ofport
+    ovs_rwlock_wrlock(&rwlock);
+    tnl_port = tnl_find_ofport(ofport);
+    if (!tnl_port) {
+        changed = tnl_port_add__(ofport, netdev, odp_port, false);
+    } else if (tnl_port->netdev != netdev
                || tnl_port->match.odp_port != odp_port
-               || tnl_port->netdev_seq != netdev_change_seq(ofport->netdev)) {
+               || tnl_port->netdev_seq != netdev_change_seq(netdev)) {
         VLOG_DBG("reconfiguring %s", tnl_port_get_name(tnl_port));
-        tnl_port_del(tnl_port);
-        *tnl_portp = tnl_port_add(ofport, odp_port);
-        return true;
+        tnl_port_del__(ofport);
+        tnl_port_add__(ofport, netdev, odp_port, true);
+        changed = true;
     }
-    return false;
+    ovs_rwlock_unlock(&rwlock);
+    return changed;
 }
 
-/* Removes 'tnl_port' from the module. */
-void
-tnl_port_del(struct tnl_port *tnl_port)
+static void
+tnl_port_del__(const struct ofport_dpif *ofport) OVS_REQ_WRLOCK(rwlock)
 {
-    if (tnl_port && tnl_port != &void_tnl_port) {
+    struct tnl_port *tnl_port;
+
+    if (!ofport) {
+        return;
+    }
+
+    tnl_port = tnl_find_ofport(ofport);
+    if (tnl_port) {
         tnl_port_mod_log(tnl_port, "removing");
-        hmap_remove(&tnl_match_map, &tnl_port->match_node);
+        hmap_remove(tnl_match_map, &tnl_port->match_node);
+        hmap_remove(ofport_map, &tnl_port->ofport_node);
+        netdev_close(tnl_port->netdev);
         free(tnl_port);
     }
 }
 
-/* Transforms 'flow' so that it appears to have been received by a tunnel
- * OpenFlow port controlled by this module instead of the datapath port it
- * actually came in on.  Sets 'flow''s in_port to the appropriate OpenFlow port
- * number.  Returns the 'ofport' corresponding to the new in_port.
+/* Removes 'ofport' from the module. */
+void
+tnl_port_del(const struct ofport_dpif *ofport) OVS_EXCLUDED(rwlock)
+{
+    ovs_rwlock_wrlock(&rwlock);
+    tnl_port_del__(ofport);
+    ovs_rwlock_unlock(&rwlock);
+}
+
+/* Looks in the table of tunnels for a tunnel matching the metadata in 'flow'.
+ * Returns the 'ofport' corresponding to the new in_port, or a null pointer if
+ * none is found.
  *
  * Callers should verify that 'flow' needs to be received by calling
- * tnl_port_should_receive() before this function.
- *
- * Leaves 'flow' untouched and returns null if unsuccessful. */
-const struct ofport *
-tnl_port_receive(struct flow *flow)
+ * tnl_port_should_receive() before this function. */
+const struct ofport_dpif *
+tnl_port_receive(const struct flow *flow) OVS_EXCLUDED(rwlock)
 {
     char *pre_flow_str = NULL;
+    const struct ofport_dpif *ofport;
     struct tnl_port *tnl_port;
     struct tnl_match match;
 
     memset(&match, 0, sizeof match);
-    match.odp_port = flow->in_port;
+    match.odp_port = flow->in_port.odp_port;
     match.ip_src = flow->tunnel.ip_dst;
     match.ip_dst = flow->tunnel.ip_src;
     match.in_key = flow->tunnel.tun_id;
-    match.in_key_present = flow->tunnel.flags & FLOW_TNL_F_KEY;
+    match.skb_mark = flow->skb_mark;
 
+    ovs_rwlock_rdlock(&rwlock);
     tnl_port = tnl_find(&match);
+    ofport = tnl_port ? tnl_port->ofport : NULL;
     if (!tnl_port) {
         struct ds ds = DS_EMPTY_INITIALIZER;
 
         tnl_match_fmt(&match, &ds);
         VLOG_WARN_RL(&rl, "receive tunnel port not found (%s)", ds_cstr(&ds));
         ds_destroy(&ds);
-        return NULL;
-    }
-
-    if (is_ip_any(flow)
-        && ((flow->tunnel.ip_tos & IP_ECN_MASK) == IP_ECN_CE)
-        && (flow->nw_tos & IP_ECN_MASK) == IP_ECN_NOT_ECT) {
-        VLOG_WARN_RL(&rl, "dropping tunnel packet marked ECN CE but is not ECN"
-                     " capable");
-        return NULL;
+        goto out;
     }
 
     if (!VLOG_DROP_DBG(&dbg_rl)) {
         pre_flow_str = flow_to_string(flow);
     }
 
-    flow->in_port = tnl_port->ofport->ofp_port;
-    memset(&flow->tunnel, 0, sizeof flow->tunnel);
-    flow->tunnel.tun_id = match.in_key;
-
     if (pre_flow_str) {
         char *post_flow_str = flow_to_string(flow);
         char *tnl_str = tnl_port_fmt(tnl_port);
@@ -225,49 +243,70 @@ tnl_port_receive(struct flow *flow)
         free(pre_flow_str);
         free(post_flow_str);
     }
-    return tnl_port->ofport;
+
+out:
+    ovs_rwlock_unlock(&rwlock);
+    return ofport;
 }
 
 /* Given that 'flow' should be output to the ofport corresponding to
  * 'tnl_port', updates 'flow''s tunnel headers and returns the actual datapath
- * port that the output should happen on.  May return OVSP_NONE if the output
+ * port that the output should happen on.  May return ODPP_NONE if the output
  * shouldn't occur. */
-uint32_t
-tnl_port_send(const struct tnl_port *tnl_port, struct flow *flow)
+odp_port_t
+tnl_port_send(const struct ofport_dpif *ofport, struct flow *flow,
+              struct flow_wildcards *wc) OVS_EXCLUDED(rwlock)
 {
     const struct netdev_tunnel_config *cfg;
+    struct tnl_port *tnl_port;
     char *pre_flow_str = NULL;
+    odp_port_t out_port;
 
-    if (tnl_port == &void_tnl_port) {
-        return OVSP_NONE;
+    ovs_rwlock_rdlock(&rwlock);
+    tnl_port = tnl_find_ofport(ofport);
+    out_port = tnl_port ? tnl_port->match.odp_port : ODPP_NONE;
+    if (!tnl_port) {
+        goto out;
     }
 
-    cfg = netdev_get_tunnel_config(tnl_port->ofport->netdev);
+    cfg = netdev_get_tunnel_config(tnl_port->netdev);
     ovs_assert(cfg);
 
     if (!VLOG_DROP_DBG(&dbg_rl)) {
         pre_flow_str = flow_to_string(flow);
     }
 
-    flow->tunnel.ip_src = tnl_port->match.ip_src;
-    flow->tunnel.ip_dst = tnl_port->match.ip_dst;
+    if (!cfg->ip_src_flow) {
+        flow->tunnel.ip_src = tnl_port->match.ip_src;
+    }
+    if (!cfg->ip_dst_flow) {
+        flow->tunnel.ip_dst = tnl_port->match.ip_dst;
+    }
+    flow->skb_mark = tnl_port->match.skb_mark;
 
     if (!cfg->out_key_flow) {
         flow->tunnel.tun_id = cfg->out_key;
     }
 
     if (cfg->ttl_inherit && is_ip_any(flow)) {
+        wc->masks.nw_ttl = 0xff;
         flow->tunnel.ip_ttl = flow->nw_ttl;
     } else {
         flow->tunnel.ip_ttl = cfg->ttl;
     }
 
     if (cfg->tos_inherit && is_ip_any(flow)) {
+        wc->masks.nw_tos = 0xff;
         flow->tunnel.ip_tos = flow->nw_tos & IP_DSCP_MASK;
     } else {
         flow->tunnel.ip_tos = cfg->tos;
     }
 
+    /* ECN fields are always inherited. */
+    if (is_ip_any(flow)) {
+        wc->masks.nw_tos |= IP_ECN_MASK;
+    }
+
     if ((flow->nw_tos & IP_ECN_MASK) == IP_ECN_CE) {
         flow->tunnel.ip_tos |= IP_ECN_ECT_0;
     } else {
@@ -291,7 +330,9 @@ tnl_port_send(const struct tnl_port *tnl_port, struct flow *flow)
         free(post_flow_str);
     }
 
-    return tnl_port->match.odp_port;
+out:
+    ovs_rwlock_unlock(&rwlock);
+    return out_port;
 }
 
 static uint32_t
@@ -302,12 +343,26 @@ tnl_hash(struct tnl_match *match)
 }
 
 static struct tnl_port *
-tnl_find_exact(struct tnl_match *match)
+tnl_find_ofport(const struct ofport_dpif *ofport) OVS_REQ_RDLOCK(rwlock)
+{
+    struct tnl_port *tnl_port;
+
+    HMAP_FOR_EACH_IN_BUCKET (tnl_port, ofport_node, hash_pointer(ofport, 0),
+                             ofport_map) {
+        if (tnl_port->ofport == ofport) {
+            return tnl_port;
+        }
+    }
+    return NULL;
+}
+
+static struct tnl_port *
+tnl_find_exact(struct tnl_match *match) OVS_REQ_RDLOCK(rwlock)
 {
     struct tnl_port *tnl_port;
 
     HMAP_FOR_EACH_WITH_HASH (tnl_port, match_node, tnl_hash(match),
-                             &tnl_match_map) {
+                             tnl_match_map) {
         if (!memcmp(match, &tnl_port->match, sizeof *match)) {
             return tnl_port;
         }
@@ -316,18 +371,15 @@ tnl_find_exact(struct tnl_match *match)
 }
 
 static struct tnl_port *
-tnl_find(struct tnl_match *match_)
+tnl_find(struct tnl_match *match_) OVS_REQ_RDLOCK(rwlock)
 {
     struct tnl_match match = *match_;
-    bool is_multicast = ip_is_multicast(match.ip_src);
     struct tnl_port *tnl_port;
 
     /* remote_ip, local_ip, in_key */
-    if (!is_multicast) {
-        tnl_port = tnl_find_exact(&match);
-        if (tnl_port) {
-            return tnl_port;
-        }
+    tnl_port = tnl_find_exact(&match);
+    if (tnl_port) {
+        return tnl_port;
     }
 
     /* remote_ip, in_key */
@@ -339,69 +391,65 @@ tnl_find(struct tnl_match *match_)
     match.ip_src = match_->ip_src;
 
     /* remote_ip, local_ip */
-    if (!is_multicast) {
-        match.in_key = 0;
-        match.in_key_flow = true;
-        tnl_port = tnl_find_exact(&match);
-        if (tnl_port) {
-            return tnl_port;
-        }
-        match.in_key = match_->in_key;
-        match.in_key_flow = false;
+    match.in_key = 0;
+    match.in_key_flow = true;
+    tnl_port = tnl_find_exact(&match);
+    if (tnl_port) {
+        return tnl_port;
     }
 
     /* remote_ip */
     match.ip_src = 0;
-    match.in_key = 0;
-    match.in_key_flow = true;
     tnl_port = tnl_find_exact(&match);
     if (tnl_port) {
         return tnl_port;
     }
-    match.ip_src = match_->ip_src;
-    match.in_key = match_->in_key;
-    match.in_key_flow = false;
 
-    if (is_multicast) {
-        match.ip_src = 0;
-        match.ip_dst = match_->ip_src;
-
-        /* multicast remote_ip, in_key */
-        tnl_port = tnl_find_exact(&match);
-        if (tnl_port) {
-            return tnl_port;
-        }
+    /* Flow-based remote */
+    match.ip_dst = 0;
+    match.ip_dst_flow = true;
+    tnl_port = tnl_find_exact(&match);
+    if (tnl_port) {
+        return tnl_port;
+    }
 
-        /* multicast remote_ip */
-        match.in_key = 0;
-        match.in_key_flow = true;
-        tnl_port = tnl_find_exact(&match);
-        if (tnl_port) {
-            return tnl_port;
-        }
+    /* Flow-based everything */
+    match.ip_src = 0;
+    match.ip_src_flow = true;
+    tnl_port = tnl_find_exact(&match);
+    if (tnl_port) {
+        return tnl_port;
     }
+
     return NULL;
 }
 
 static void
 tnl_match_fmt(const struct tnl_match *match, struct ds *ds)
+    OVS_REQ_RDLOCK(rwlock)
 {
-    ds_put_format(ds, IP_FMT"->"IP_FMT, IP_ARGS(match->ip_src),
-                  IP_ARGS(match->ip_dst));
+    if (!match->ip_dst_flow) {
+        ds_put_format(ds, IP_FMT"->"IP_FMT, IP_ARGS(match->ip_src),
+                      IP_ARGS(match->ip_dst));
+    } else if (!match->ip_src_flow) {
+        ds_put_format(ds, IP_FMT"->flow", IP_ARGS(match->ip_src));
+    } else {
+        ds_put_cstr(ds, "flow->flow");
+    }
 
-    if (match->in_key_present) {
-        if (match->in_key_flow) {
-            ds_put_cstr(ds, ", key=flow");
-        } else {
-            ds_put_format(ds, ", key=%#"PRIx64, ntohll(match->in_key));
-        }
+    if (match->in_key_flow) {
+        ds_put_cstr(ds, ", key=flow");
+    } else {
+        ds_put_format(ds, ", key=%#"PRIx64, ntohll(match->in_key));
     }
 
     ds_put_format(ds, ", dp port=%"PRIu32, match->odp_port);
+    ds_put_format(ds, ", skb mark=%"PRIu32, match->skb_mark);
 }
 
 static void
 tnl_port_mod_log(const struct tnl_port *tnl_port, const char *action)
+    OVS_REQ_RDLOCK(rwlock)
 {
     if (VLOG_IS_DBG_ENABLED()) {
         struct ds ds = DS_EMPTY_INITIALIZER;
@@ -414,15 +462,15 @@ tnl_port_mod_log(const struct tnl_port *tnl_port, const char *action)
 }
 
 static char *
-tnl_port_fmt(const struct tnl_port *tnl_port)
+tnl_port_fmt(const struct tnl_port *tnl_port) OVS_REQ_RDLOCK(rwlock)
 {
     const struct netdev_tunnel_config *cfg =
-        netdev_get_tunnel_config(tnl_port->ofport->netdev);
+        netdev_get_tunnel_config(tnl_port->netdev);
     struct ds ds = DS_EMPTY_INITIALIZER;
 
     ds_put_format(&ds, "port %"PRIu32": %s (%s: ", tnl_port->match.odp_port,
                   tnl_port_get_name(tnl_port),
-                  netdev_get_type(tnl_port->ofport->netdev));
+                  netdev_get_type(tnl_port->netdev));
     tnl_match_fmt(&tnl_port->match, &ds);
 
     if (cfg->out_key != cfg->in_key ||
@@ -464,7 +512,7 @@ tnl_port_fmt(const struct tnl_port *tnl_port)
 }
 
 static const char *
-tnl_port_get_name(const struct tnl_port *tnl_port)
+tnl_port_get_name(const struct tnl_port *tnl_port) OVS_REQ_RDLOCK(rwlock)
 {
-    return netdev_get_name(tnl_port->ofport->netdev);
+    return netdev_get_name(tnl_port->netdev);
 }