bgpd: Adds support to mark up the BGP rib table entry with multipath

information based on the multipath list (mp_list) generated during
the best path calculation. Display "multipath" for paths that are
multipath and also on bestpath if the route is multipath. Flag a
best path with the BGP_INFO_MULTIPATH_CHG if the multipath
set has changed since the last update. This can be used to trigger
updates to zebra and peers.

The multipath markup is a lazily allocated bgp_info_mpath structure
that is added to the best path and any multipaths. The mpath structures
are linked together with the best path element at the head and the
other elements ordered by nexthop and then by peer address. This
markup scheme is updated by calling bgp_info_mpath_update() and passing
in a new mp_list the the current multipath set. There are additional
API's for walking the multipath set, querying the count of multipaths,
and for cleaning up the multipath markup information when freeing path
information.

* bgpd/bgp_mpath.c
  * bgp_info_mpath_new(): Allocation of new mpath element
  * bgp_info_mpath_free(): Release memory for mpath element
  * bgp_info_mpath_get(): Access mpath element of path. Allocate memory
    on-demand
  * bgp_info_mpath_enqueue(): Enqueue a path onto the multipath list
  * bgp_info_mpath_dequeue(): Remove a path from the multipath list
  * bgp_info_mpath_first(): Return first path on the multipath list
  * bgp_info_mpath_next(): Return next path on the multipath list
  * bgp_info_mpath_count(): Return the number of paths on the multipath list
  * bgp_info_mpath_count_set(): Set the number of paths on the multipath list
  * bgp_info_mpath_update(): Update multipath markup on bgp route table entry
    and flag any changes. Emit 'debug bgp event' output on any multipath
    change.
* bgpd/bgp_mpath.h
  * struct bgp_info_mpath: Information added to a bgp_info path to record
    multipath information
  * External declarations for new functions in bgp_mpath.c
* bgpd/bgp_route.c
  * bgp_info_free(): Free mpath memory when freeing path information
  * bgp_info_reap(): Dequeue path from multipath queue before deleting it
  * bgp_best_selection(): Calls bgp_info_mpath_update() with latest
    mp_list to mark-up rib table entry
  * bgp_vty_out_detail(): Add display of multipath flag for a path. Also
    display 'multipath' for bestpath if it is a multipath route
* bgpd/bgp_route.h
  * struct bgp_info: Add pointer to bgp_info_mpath information
  * Add flags to mark a path as multipath (BGP_INFO_MULTIPATH) and
    to mark bestpath if multipath information has changed
    (BGP_INFO_MULTIPATH_CHG)
* lib/memtypes.c
  * Add MTYPE_BGP_MPATH_INFO for allocating memory for bgp_info_mpath
* tests/bgp_mpath_test.c
  * Add test case for bgp_info_mpath_update() and supporting functions

1 files changed, 313 insertions, 0 deletions

diff --git a/bgpd/bgp_mpath.c b/bgpd/bgp_mpath.c
index 09b469516..af1c342c6 100644
--- a/bgpd/bgp_mpath.c
+++ b/bgpd/bgp_mpath.c
@@ -27,11 +27,13 @@
 #include "prefix.h"
 #include "linklist.h"
 #include "sockunion.h"
+#include "memory.h"
 
 #include "bgpd/bgpd.h"
 #include "bgpd/bgp_table.h"
 #include "bgpd/bgp_route.h"
 #include "bgpd/bgp_attr.h"
+#include "bgpd/bgp_debug.h"
 #include "bgpd/bgp_mpath.h"
 
 /*
@@ -200,3 +202,314 @@ bgp_mp_list_add (struct list *mp_list, struct bgp_info *mpinfo)
   assert (mp_list && mpinfo);
   listnode_add_sort (mp_list, mpinfo);
 }
+
+/*
+ * bgp_info_mpath_new
+ *
+ * Allocate and zero memory for a new bgp_info_mpath element
+ */
+static struct bgp_info_mpath *
+bgp_info_mpath_new (void)
+{
+  struct bgp_info_mpath *new_mpath;
+  new_mpath = XCALLOC (MTYPE_BGP_MPATH_INFO, sizeof (struct bgp_info_mpath));
+  return new_mpath;
+}
+
+/*
+ * bgp_info_mpath_free
+ *
+ * Release resources for a bgp_info_mpath element and zero out pointer
+ */
+void
+bgp_info_mpath_free (struct bgp_info_mpath **mpath)
+{
+  if (mpath && *mpath)
+    {
+      XFREE (MTYPE_BGP_MPATH_INFO, *mpath);
+      *mpath = NULL;
+    }
+}
+
+/*
+ * bgp_info_mpath_get
+ *
+ * Fetch the mpath element for the given bgp_info. Used for
+ * doing lazy allocation.
+ */
+static struct bgp_info_mpath *
+bgp_info_mpath_get (struct bgp_info *binfo)
+{
+  struct bgp_info_mpath *mpath;
+  if (!binfo->mpath)
+    {
+      mpath = bgp_info_mpath_new();
+      if (!mpath)
+        return NULL;
+      binfo->mpath = mpath;
+      mpath->mp_info = binfo;
+    }
+  return binfo->mpath;
+}
+
+/*
+ * bgp_info_mpath_enqueue
+ *
+ * Enqueue a path onto the multipath list given the previous multipath
+ * list entry
+ */
+static void
+bgp_info_mpath_enqueue (struct bgp_info *prev_info, struct bgp_info *binfo)
+{
+  struct bgp_info_mpath *prev, *mpath;
+
+  prev = bgp_info_mpath_get (prev_info);
+  mpath = bgp_info_mpath_get (binfo);
+  if (!prev || !mpath)
+    return;
+
+  mpath->mp_next = prev->mp_next;
+  mpath->mp_prev = prev;
+  if (prev->mp_next)
+    prev->mp_next->mp_prev = mpath;
+  prev->mp_next = mpath;
+
+  SET_FLAG (binfo->flags, BGP_INFO_MULTIPATH);
+}
+
+/*
+ * bgp_info_mpath_dequeue
+ *
+ * Remove a path from the multipath list
+ */
+void
+bgp_info_mpath_dequeue (struct bgp_info *binfo)
+{
+  struct bgp_info_mpath *mpath = binfo->mpath;
+  if (!mpath)
+    return;
+  if (mpath->mp_prev)
+    mpath->mp_prev->mp_next = mpath->mp_next;
+  if (mpath->mp_next)
+    mpath->mp_next->mp_prev = mpath->mp_prev;
+  mpath->mp_next = mpath->mp_prev = NULL;
+  UNSET_FLAG (binfo->flags, BGP_INFO_MULTIPATH);
+}
+
+/*
+ * bgp_info_mpath_next
+ *
+ * Given a bgp_info, return the next multipath entry
+ */
+struct bgp_info *
+bgp_info_mpath_next (struct bgp_info *binfo)
+{
+  if (!binfo->mpath || !binfo->mpath->mp_next)
+    return NULL;
+  return binfo->mpath->mp_next->mp_info;
+}
+
+/*
+ * bgp_info_mpath_first
+ *
+ * Given bestpath bgp_info, return the first multipath entry.
+ */
+struct bgp_info *
+bgp_info_mpath_first (struct bgp_info *binfo)
+{
+  return bgp_info_mpath_next (binfo);
+}
+
+/*
+ * bgp_info_mpath_count
+ *
+ * Given the bestpath bgp_info, return the number of multipath entries
+ */
+u_int32_t
+bgp_info_mpath_count (struct bgp_info *binfo)
+{
+  if (!binfo->mpath)
+    return 0;
+  return binfo->mpath->mp_count;
+}
+
+/*
+ * bgp_info_mpath_count_set
+ *
+ * Sets the count of multipaths into bestpath's mpath element
+ */
+static void
+bgp_info_mpath_count_set (struct bgp_info *binfo, u_int32_t count)
+{
+  struct bgp_info_mpath *mpath;
+  if (!count && !binfo->mpath)
+    return;
+  mpath = bgp_info_mpath_get (binfo);
+  if (!mpath)
+    return;
+  mpath->mp_count = count;
+}
+
+/*
+ * bgp_info_mpath_update
+ *
+ * Compare and sync up the multipath list with the mp_list generated by
+ * bgp_best_selection
+ */
+void
+bgp_info_mpath_update (struct bgp_node *rn, struct bgp_info *new_best,
+                       struct bgp_info *old_best, struct list *mp_list,
+                       struct bgp_maxpaths_cfg *mpath_cfg)
+{
+  u_int16_t maxpaths, mpath_count, old_mpath_count;
+  struct listnode *mp_node, *mp_next_node;
+  struct bgp_info *cur_mpath, *new_mpath, *next_mpath, *prev_mpath;
+  int mpath_changed, debug;
+  char pfx_buf[INET_ADDRSTRLEN], nh_buf[2][INET_ADDRSTRLEN];
+
+  mpath_changed = 0;
+  maxpaths = BGP_DEFAULT_MAXPATHS;
+  mpath_count = 0;
+  cur_mpath = NULL;
+  old_mpath_count = 0;
+  prev_mpath = new_best;
+  mp_node = listhead (mp_list);
+  debug = BGP_DEBUG (events, EVENTS);
+
+  if (debug)
+    prefix2str (&rn->p, pfx_buf, sizeof (pfx_buf));
+
+  if (new_best)
+    {
+      mpath_count++;
+      if (new_best != old_best)
+        bgp_info_mpath_dequeue (new_best);
+      maxpaths = (peer_sort (new_best->peer) == BGP_PEER_IBGP) ?
+        mpath_cfg->maxpaths_ibgp : mpath_cfg->maxpaths_ebgp;
+    }
+
+  if (old_best)
+    {
+      cur_mpath = bgp_info_mpath_first (old_best);
+      old_mpath_count = bgp_info_mpath_count (old_best);
+      bgp_info_mpath_count_set (old_best, 0);
+      bgp_info_mpath_dequeue (old_best);
+    }
+
+  /*
+   * We perform an ordered walk through both lists in parallel.
+   * The reason for the ordered walk is that if there are paths
+   * that were previously multipaths and are still multipaths, the walk
+   * should encounter them in both lists at the same time. Otherwise
+   * there will be paths that are in one list or another, and we
+   * will deal with these separately.
+   *
+   * Note that new_best might be somewhere in the mp_list, so we need
+   * to skip over it
+   */
+  while (mp_node || cur_mpath)
+    {
+      /*
+       * We can bail out of this loop if all existing paths on the
+       * multipath list have been visited (for cleanup purposes) and
+       * the maxpath requirement is fulfulled
+       */
+      if (!cur_mpath && (mpath_count >= maxpaths))
+        break;
+
+      mp_next_node = mp_node ? listnextnode (mp_node) : NULL;
+      next_mpath = cur_mpath ? bgp_info_mpath_next (cur_mpath) : NULL;
+
+      /*
+       * If equal, the path was a multipath and is still a multipath.
+       * Insert onto new multipath list if maxpaths allows.
+       */
+      if (mp_node && (listgetdata (mp_node) == cur_mpath))
+        {
+          list_delete_node (mp_list, mp_node);
+          bgp_info_mpath_dequeue (cur_mpath);
+          if ((mpath_count < maxpaths) &&
+              bgp_info_nexthop_cmp (prev_mpath, cur_mpath))
+            {
+              bgp_info_mpath_enqueue (prev_mpath, cur_mpath);
+              prev_mpath = cur_mpath;
+              mpath_count++;
+            }
+          else
+            {
+              mpath_changed = 1;
+              if (debug)
+                zlog_debug ("%s remove mpath nexthop %s peer %s", pfx_buf,
+                            inet_ntop (AF_INET, &cur_mpath->attr->nexthop,
+                                       nh_buf[0], sizeof (nh_buf[0])),
+                            sockunion2str (cur_mpath->peer->su_remote,
+                                           nh_buf[1], sizeof (nh_buf[1])));
+            }
+          mp_node = mp_next_node;
+          cur_mpath = next_mpath;
+          continue;
+        }
+
+      if (cur_mpath && (!mp_node ||
+                        (bgp_info_mpath_cmp (cur_mpath,
+                                             listgetdata (mp_node)) < 0)))
+        {
+          /*
+           * If here, we have an old multipath and either the mp_list
+           * is finished or the next mp_node points to a later
+           * multipath, so we need to purge this path from the
+           * multipath list
+           */
+          bgp_info_mpath_dequeue (cur_mpath);
+          mpath_changed = 1;
+          if (debug)
+            zlog_debug ("%s remove mpath nexthop %s peer %s", pfx_buf,
+                        inet_ntop (AF_INET, &cur_mpath->attr->nexthop,
+                                   nh_buf[0], sizeof (nh_buf[0])),
+                        sockunion2str (cur_mpath->peer->su_remote,
+                                       nh_buf[1], sizeof (nh_buf[1])));
+          cur_mpath = next_mpath;
+        }
+      else
+        {
+          /*
+           * If here, we have a path on the mp_list that was not previously
+           * a multipath (due to non-equivalance or maxpaths exceeded),
+           * or the matching multipath is sorted later in the multipath
+           * list. Before we enqueue the path on the new multipath list,
+           * make sure its not on the old_best multipath list or referenced
+           * via next_mpath:
+           * - If next_mpath points to this new path, update next_mpath to
+           *   point to the multipath after this one
+           * - Dequeue the path from the multipath list just to make sure
+           */
+          new_mpath = listgetdata (mp_node);
+          list_delete_node (mp_list, mp_node);
+          if (new_mpath == next_mpath)
+            next_mpath = bgp_info_mpath_next (new_mpath);
+          bgp_info_mpath_dequeue (new_mpath);
+          if ((mpath_count < maxpaths) && (new_mpath != new_best) &&
+              bgp_info_nexthop_cmp (prev_mpath, new_mpath))
+            {
+              bgp_info_mpath_enqueue (prev_mpath, new_mpath);
+              prev_mpath = new_mpath;
+              mpath_changed = 1;
+              mpath_count++;
+              if (debug)
+                zlog_debug ("%s add mpath nexthop %s peer %s", pfx_buf,
+                            inet_ntop (AF_INET, &new_mpath->attr->nexthop,
+                                       nh_buf[0], sizeof (nh_buf[0])),
+                            sockunion2str (new_mpath->peer->su_remote,
+                                           nh_buf[1], sizeof (nh_buf[1])));
+            }
+          mp_node = mp_next_node;
+        }
+    }
+
+  if (new_best)
+    {
+      bgp_info_mpath_count_set (new_best, mpath_count-1);
+      if (mpath_changed || (bgp_info_mpath_count (new_best) != old_mpath_count))
+        SET_FLAG (new_best->flags, BGP_INFO_MULTIPATH_CHG);
+    }
+}