This is the start of the BETA21 branch.

It includes the --topology feature, and
TAP-Win32 driver changes to allow
non-admin access.



git-svn-id: http://svn.openvpn.net/projects/openvpn/branches/BETA21/openvpn@580 e7ae566f-a301-0410-adde-c780ea21d3b5

1 files changed, 2129 insertions, 0 deletions

diff --git a/multi.c b/multi.c
new file mode 100644
index 0000000..cf10b51
--- /dev/null
+++ b/multi.c
@@ -0,0 +1,2129 @@
+/*
+ *  OpenVPN -- An application to securely tunnel IP networks
+ *             over a single TCP/UDP port, with support for SSL/TLS-based
+ *             session authentication and key exchange,
+ *             packet encryption, packet authentication, and
+ *             packet compression.
+ *
+ *  Copyright (C) 2002-2005 OpenVPN Solutions LLC <info@openvpn.net>
+ *
+ *  This program is free software; you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License version 2
+ *  as published by the Free Software Foundation.
+ *
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License
+ *  along with this program (see the file COPYING included with this
+ *  distribution); if not, write to the Free Software Foundation, Inc.,
+ *  59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#ifdef WIN32
+#include "config-win32.h"
+#else
+#include "config.h"
+#endif
+
+#include "syshead.h"
+
+#if P2MP_SERVER
+
+#include "multi.h"
+#include "push.h"
+#include "misc.h"
+#include "otime.h"
+#include "gremlin.h"
+
+#include "memdbg.h"
+
+#include "forward-inline.h"
+
+/*#define MULTI_DEBUG_EVENT_LOOP*/
+
+#ifdef MULTI_DEBUG_EVENT_LOOP
+static const char *
+id (struct multi_instance *mi)
+{
+  if (mi)
+    return tls_common_name (mi->context.c2.tls_multi, false);
+  else
+    return "NULL";
+}
+#endif
+
+static bool
+learn_address_script (const struct multi_context *m,
+		      const struct multi_instance *mi,
+		      const char *op,
+		      const struct mroute_addr *addr)
+{
+  struct gc_arena gc = gc_new ();
+  struct env_set *es;
+  bool ret = true;
+
+  /* get environmental variable source */
+  if (mi && mi->context.c2.es)
+    es = mi->context.c2.es;
+  else
+    es = env_set_create (&gc);
+
+  if (plugin_defined (m->top.c1.plugins, OPENVPN_PLUGIN_LEARN_ADDRESS))
+    {
+      struct buffer cmd = alloc_buf_gc (256, &gc);
+
+      buf_printf (&cmd, "\"%s\" \"%s\"",
+		  op,
+		  mroute_addr_print (addr, &gc));
+      if (mi)
+	buf_printf (&cmd, " \"%s\"", tls_common_name (mi->context.c2.tls_multi, false));
+
+      if (plugin_call (m->top.c1.plugins, OPENVPN_PLUGIN_LEARN_ADDRESS, BSTR (&cmd), es))
+	{
+	  msg (M_WARN, "WARNING: learn-address plugin call failed");
+	  ret = false;
+	}
+    }
+
+  if (m->top.options.learn_address_script)
+    {
+      struct buffer cmd = alloc_buf_gc (256, &gc);
+
+      setenv_str (es, "script_type", "learn-address");
+
+      buf_printf (&cmd, "%s \"%s\" \"%s\"",
+		  m->top.options.learn_address_script,
+		  op,
+		  mroute_addr_print (addr, &gc));
+      if (mi)
+	buf_printf (&cmd, " \"%s\"", tls_common_name (mi->context.c2.tls_multi, false));
+
+      if (!system_check (BSTR (&cmd), es, S_SCRIPT, "WARNING: learn-address command failed"))
+	ret = false;
+    }
+
+  gc_free (&gc);
+  return ret;
+}
+
+void
+multi_ifconfig_pool_persist (struct multi_context *m, bool force)
+{
+ /* write pool data to file */
+  if (m->ifconfig_pool
+      && m->top.c1.ifconfig_pool_persist
+      && (force || ifconfig_pool_write_trigger (m->top.c1.ifconfig_pool_persist)))
+    {
+      ifconfig_pool_write (m->top.c1.ifconfig_pool_persist, m->ifconfig_pool);
+    }
+}
+
+static void
+multi_reap_range (const struct multi_context *m,
+		  int start_bucket,
+		  int end_bucket)
+{
+  struct gc_arena gc = gc_new ();
+  struct hash_iterator hi;
+  struct hash_element *he;
+
+  if (start_bucket < 0)
+    {
+      start_bucket = 0;
+      end_bucket = hash_n_buckets (m->vhash);
+    }
+
+  dmsg (D_MULTI_DEBUG, "MULTI: REAP range %d -> %d", start_bucket, end_bucket);
+  hash_iterator_init_range (m->vhash, &hi, true, start_bucket, end_bucket);
+  while ((he = hash_iterator_next (&hi)) != NULL)
+    {
+      struct multi_route *r = (struct multi_route *) he->value;
+      if (!multi_route_defined (m, r))
+	{
+	  dmsg (D_MULTI_DEBUG, "MULTI: REAP DEL %s",
+	       mroute_addr_print (&r->addr, &gc));
+	  learn_address_script (m, NULL, "delete", &r->addr);
+	  multi_route_del (r);
+	  hash_iterator_delete_element (&hi);
+	}
+    }
+  hash_iterator_free (&hi);
+  gc_free (&gc);
+}
+
+static void
+multi_reap_all (const struct multi_context *m)
+{
+  multi_reap_range (m, -1, 0);
+}
+
+static struct multi_reap *
+multi_reap_new (int buckets_per_pass)
+{
+  struct multi_reap *mr;
+  ALLOC_OBJ (mr, struct multi_reap);
+  mr->bucket_base = 0;
+  mr->buckets_per_pass = buckets_per_pass;
+  mr->last_call = now;
+  return mr;
+}
+
+void
+multi_reap_process_dowork (const struct multi_context *m)
+{
+  struct multi_reap *mr = m->reaper;
+  if (mr->bucket_base >= hash_n_buckets (m->vhash))
+    mr->bucket_base = 0;
+  multi_reap_range (m, mr->bucket_base, mr->bucket_base + mr->buckets_per_pass); 
+  mr->bucket_base += mr->buckets_per_pass;
+  mr->last_call = now;
+}
+
+static void
+multi_reap_free (struct multi_reap *mr)
+{
+  free (mr);
+}
+
+/*
+ * How many buckets in vhash to reap per pass.
+ */
+static int
+reap_buckets_per_pass (int n_buckets)
+{
+  return constrain_int (n_buckets / REAP_DIVISOR, REAP_MIN, REAP_MAX);
+}
+
+/*
+ * Main initialization function, init multi_context object.
+ */
+void
+multi_init (struct multi_context *m, struct context *t, bool tcp_mode, int thread_mode)
+{
+  int dev = DEV_TYPE_UNDEF;
+
+  msg (D_MULTI_LOW, "MULTI: multi_init called, r=%d v=%d",
+       t->options.real_hash_size,
+       t->options.virtual_hash_size);
+
+  /*
+   * Get tun/tap/null device type
+   */
+  dev = dev_type_enum (t->options.dev, t->options.dev_type);
+
+  /*
+   * Init our multi_context object.
+   */
+  CLEAR (*m);
+  
+  m->thread_mode = thread_mode;
+
+  /*
+   * Real address hash table (source port number is
+   * considered to be part of the address).  Used
+   * to determine which client sent an incoming packet
+   * which is seen on the TCP/UDP socket.
+   */
+  m->hash = hash_init (t->options.real_hash_size,
+		       mroute_addr_hash_function,
+		       mroute_addr_compare_function);
+
+  /*
+   * Virtual address hash table.  Used to determine
+   * which client to route a packet to. 
+   */
+  m->vhash = hash_init (t->options.virtual_hash_size,
+			mroute_addr_hash_function,
+			mroute_addr_compare_function);
+
+  /*
+   * This hash table is a clone of m->hash but with a
+   * bucket size of one so that it can be used
+   * for fast iteration through the list.
+   */
+  m->iter = hash_init (1,
+		       mroute_addr_hash_function,
+		       mroute_addr_compare_function);
+
+  /*
+   * This is our scheduler, for time-based wakeup
+   * events.
+   */
+  m->schedule = schedule_init ();
+
+  /*
+   * Limit frequency of incoming connections to control
+   * DoS.
+   */
+  m->new_connection_limiter = frequency_limit_init (t->options.cf_max,
+						    t->options.cf_per);
+
+  /*
+   * Allocate broadcast/multicast buffer list
+   */
+  m->mbuf = mbuf_init (t->options.n_bcast_buf);
+
+  /*
+   * Different status file format options are available
+   */
+  m->status_file_version = t->options.status_file_version;
+
+  /*
+   * Possibly allocate an ifconfig pool, do it
+   * differently based on whether a tun or tap style
+   * tunnel.
+   */
+  if (t->options.ifconfig_pool_defined)
+    {
+      if (dev == DEV_TYPE_TAP || t->options.ifconfig_pool_linear)
+	{
+	  m->ifconfig_pool = ifconfig_pool_init (IFCONFIG_POOL_INDIV,
+						 t->options.ifconfig_pool_start,
+						 t->options.ifconfig_pool_end,
+						 t->options.duplicate_cn);
+	}
+      else if (dev == DEV_TYPE_TUN)
+	{
+	  m->ifconfig_pool = ifconfig_pool_init (IFCONFIG_POOL_30NET,
+						 t->options.ifconfig_pool_start,
+						 t->options.ifconfig_pool_end,
+						 t->options.duplicate_cn);
+	}
+      else
+	{
+	  ASSERT (0);
+	}
+
+      /* reload pool data from file */
+      if (t->c1.ifconfig_pool_persist)
+	ifconfig_pool_read (t->c1.ifconfig_pool_persist, m->ifconfig_pool);
+    }
+
+  /*
+   * Help us keep track of routing table.
+   */
+  m->route_helper = mroute_helper_init (MULTI_CACHE_ROUTE_TTL);
+
+  /*
+   * Initialize route and instance reaper.
+   */
+  m->reaper = multi_reap_new (reap_buckets_per_pass (t->options.virtual_hash_size));
+
+  /*
+   * Get local ifconfig address
+   */
+  CLEAR (m->local);
+  ASSERT (t->c1.tuntap);
+  mroute_extract_in_addr_t (&m->local, t->c1.tuntap->local);
+
+  /*
+   * Per-client limits
+   */
+  m->max_clients = t->options.max_clients;
+
+  /*
+   * Initialize multi-socket TCP I/O wait object
+   */
+  if (tcp_mode)
+    m->mtcp = multi_tcp_init (t->options.max_clients, &m->max_clients);
+  m->tcp_queue_limit = t->options.tcp_queue_limit;
+  
+  /*
+   * Allow client <-> client communication, without going through
+   * tun/tap interface and network stack?
+   */
+  m->enable_c2c = t->options.enable_c2c;
+}
+
+const char *
+multi_instance_string (const struct multi_instance *mi, bool null, struct gc_arena *gc)
+{
+  if (mi)
+    {
+      struct buffer out = alloc_buf_gc (256, gc);
+      const char *cn = tls_common_name (mi->context.c2.tls_multi, true);
+
+      if (cn)
+	buf_printf (&out, "%s/", cn);
+      buf_printf (&out, "%s", mroute_addr_print (&mi->real, gc));
+      return BSTR (&out);
+    }
+  else if (null)
+    return NULL;
+  else
+    return "UNDEF";
+}
+
+void
+generate_prefix (struct multi_instance *mi)
+{
+  mi->msg_prefix = multi_instance_string (mi, true, &mi->gc);
+  set_prefix (mi);
+}
+
+void
+ungenerate_prefix (struct multi_instance *mi)
+{
+  mi->msg_prefix = NULL;
+  set_prefix (mi);
+}
+
+/*
+ * Tell the route helper about deleted iroutes so
+ * that it can update its mask of currently used
+ * CIDR netlengths.
+ */
+static void
+multi_del_iroutes (struct multi_context *m,
+		   struct multi_instance *mi)
+{
+  const struct iroute *ir;
+  if (TUNNEL_TYPE (mi->context.c1.tuntap) == DEV_TYPE_TUN)
+    {
+      for (ir = mi->context.options.iroutes; ir != NULL; ir = ir->next)
+	mroute_helper_del_iroute (m->route_helper, ir);
+    }
+}
+
+static void
+multi_client_disconnect_setenv (struct multi_context *m,
+				struct multi_instance *mi)
+{
+  /* setenv client real IP address */
+  setenv_trusted (mi->context.c2.es, get_link_socket_info (&mi->context));
+
+  /* setenv stats */
+  setenv_int (mi->context.c2.es, "bytes_received", mi->context.c2.link_read_bytes);
+  setenv_int (mi->context.c2.es, "bytes_sent", mi->context.c2.link_write_bytes);
+
+}
+
+static void
+multi_client_disconnect_script (struct multi_context *m,
+				struct multi_instance *mi)
+{
+  if (mi->context.c2.context_auth == CAS_SUCCEEDED || mi->context.c2.context_auth == CAS_PARTIAL)
+    {
+      multi_client_disconnect_setenv (m, mi);
+
+      if (plugin_defined (m->top.c1.plugins, OPENVPN_PLUGIN_CLIENT_DISCONNECT))
+	{
+	  if (plugin_call (m->top.c1.plugins, OPENVPN_PLUGIN_CLIENT_DISCONNECT, NULL, mi->context.c2.es))
+	    msg (M_WARN, "WARNING: client-disconnect plugin call failed");
+	}
+
+      if (mi->context.options.client_disconnect_script)
+	{
+	  struct gc_arena gc = gc_new ();
+	  struct buffer cmd = alloc_buf_gc (256, &gc);
+
+	  setenv_str (mi->context.c2.es, "script_type", "client-disconnect");
+	  
+	  buf_printf (&cmd, "%s", mi->context.options.client_disconnect_script);
+
+	  system_check (BSTR (&cmd), mi->context.c2.es, S_SCRIPT, "client-disconnect command failed");
+	  
+	  gc_free (&gc);
+	}
+    }
+}
+
+void
+multi_close_instance (struct multi_context *m,
+		      struct multi_instance *mi,
+		      bool shutdown)
+{
+  perf_push (PERF_MULTI_CLOSE_INSTANCE);
+
+  ASSERT (!mi->halt);
+  mi->halt = true;
+
+  dmsg (D_MULTI_DEBUG, "MULTI: multi_close_instance called");
+
+  /* prevent dangling pointers */
+  if (m->pending == mi)
+    multi_set_pending (m, NULL);
+  if (m->earliest_wakeup == mi)
+    m->earliest_wakeup = NULL;
+
+  if (!shutdown)
+    {
+      if (mi->did_real_hash)
+	{
+	  ASSERT (hash_remove (m->hash, &mi->real));
+	}
+      if (mi->did_iter)
+	{
+	  ASSERT (hash_remove (m->iter, &mi->real));
+	}
+
+      schedule_remove_entry (m->schedule, (struct schedule_entry *) mi);
+
+      ifconfig_pool_release (m->ifconfig_pool, mi->vaddr_handle, false);
+      
+      if (mi->did_iroutes)
+        {
+          multi_del_iroutes (m, mi);
+          mi->did_iroutes = false;
+        }
+
+      if (m->mtcp)
+	multi_tcp_dereference_instance (m->mtcp, mi);
+
+      mbuf_dereference_instance (m->mbuf, mi);
+    }
+
+  multi_client_disconnect_script (m, mi);
+
+  if (mi->did_open_context)
+    close_context (&mi->context, SIGTERM, CC_GC_FREE);
+
+  multi_tcp_instance_specific_free (mi);
+
+  ungenerate_prefix (mi);
+
+  /*
+   * Don't actually delete the instance memory allocation yet,
+   * because virtual routes may still point to it.  Let the
+   * vhash reaper deal with it.
+   */
+  multi_instance_dec_refcount (mi);
+
+  perf_pop ();
+}
+
+/*
+ * Called on shutdown or restart.
+ */
+void
+multi_uninit (struct multi_context *m)
+{
+  if (m->thread_mode & MC_WORK_THREAD)
+    {
+      multi_top_free (m);
+      m->thread_mode = MC_UNDEF;
+    }
+  else if (m->thread_mode)
+    {
+      if (m->hash)
+	{
+	  struct hash_iterator hi;
+	  struct hash_element *he;
+
+	  hash_iterator_init (m->iter, &hi, true);
+	  while ((he = hash_iterator_next (&hi)))
+	    {
+	      struct multi_instance *mi = (struct multi_instance *) he->value;
+	      mi->did_iter = false;
+	      multi_close_instance (m, mi, true);
+	    }
+	  hash_iterator_free (&hi);
+
+	  multi_reap_all (m);
+
+	  hash_free (m->hash);
+	  hash_free (m->vhash);
+	  hash_free (m->iter);
+	  m->hash = NULL;
+
+	  schedule_free (m->schedule);
+	  mbuf_free (m->mbuf);
+	  ifconfig_pool_free (m->ifconfig_pool);
+	  frequency_limit_free (m->new_connection_limiter);
+	  multi_reap_free (m->reaper);
+	  mroute_helper_free (m->route_helper);
+	  multi_tcp_free (m->mtcp);
+	  m->thread_mode = MC_UNDEF;
+	}
+    }
+}
+
+/*
+ * Create a client instance object for a newly connected client.
+ */
+struct multi_instance *
+multi_create_instance (struct multi_context *m, const struct mroute_addr *real)
+{
+  struct gc_arena gc = gc_new ();
+  struct multi_instance *mi;
+
+  perf_push (PERF_MULTI_CREATE_INSTANCE);
+
+  msg (D_MULTI_LOW, "MULTI: multi_create_instance called");
+
+  ALLOC_OBJ_CLEAR (mi, struct multi_instance);
+
+  mutex_init (&mi->mutex);
+  mi->gc = gc_new ();
+  multi_instance_inc_refcount (mi);
+  mi->vaddr_handle = -1;
+  mi->created = now;
+  mroute_addr_init (&mi->real);
+
+  if (real)
+    {
+      mi->real = *real;
+      generate_prefix (mi);
+    }
+
+  inherit_context_child (&mi->context, &m->top);
+  if (IS_SIG (&mi->context))
+    goto err;
+  mi->did_open_context = true;
+
+  mi->context.c2.context_auth = CAS_PENDING;
+
+  if (hash_n_elements (m->hash) >= m->max_clients)
+    {
+      msg (D_MULTI_ERRORS, "MULTI: new incoming connection would exceed maximum number of clients (%d)", m->max_clients);
+      goto err;
+    }
+
+  if (!real) /* TCP mode? */
+    {
+      if (!multi_tcp_instance_specific_init (m, mi))
+	goto err;
+      generate_prefix (mi);
+    }
+
+  if (!hash_add (m->iter, &mi->real, mi, false))
+    {
+      msg (D_MULTI_LOW, "MULTI: unable to add real address [%s] to iterator hash table",
+	   mroute_addr_print (&mi->real, &gc));
+      goto err;
+    }
+  mi->did_iter = true;
+
+  mi->context.c2.push_reply_deferred = true;
+
+  if (!multi_process_post (m, mi, MPP_PRE_SELECT))
+    {
+      msg (D_MULTI_ERRORS, "MULTI: signal occurred during client instance initialization");
+      goto err;
+    }
+
+  perf_pop ();
+  gc_free (&gc);
+  return mi;
+
+ err:
+  multi_close_instance (m, mi, false);
+  perf_pop ();
+  gc_free (&gc);
+  return NULL;
+}
+
+/*
+ * Dump tables -- triggered by SIGUSR2.
+ * If status file is defined, write to file.
+ * If status file is NULL, write to syslog.
+ */
+void
+multi_print_status (struct multi_context *m, struct status_output *so, const int version)
+{
+  if (m->hash)
+    {
+      struct gc_arena gc_top = gc_new ();
+      struct hash_iterator hi;
+      const struct hash_element *he;
+
+      status_reset (so);
+
+      if (version == 1) /* WAS: m->status_file_version */
+	{
+	  /*
+	   * Status file version 1
+	   */
+	  status_printf (so, PACKAGE_NAME " CLIENT LIST");
+	  status_printf (so, "Updated,%s", time_string (0, 0, false, &gc_top));
+	  status_printf (so, "Common Name,Real Address,Bytes Received,Bytes Sent,Connected Since");
+	  hash_iterator_init (m->hash, &hi, true);
+	  while ((he = hash_iterator_next (&hi)))
+	    {
+	      struct gc_arena gc = gc_new ();
+	      const struct multi_instance *mi = (struct multi_instance *) he->value;
+
+	      if (!mi->halt)
+		{
+		  status_printf (so, "%s,%s," counter_format "," counter_format ",%s",
+				 tls_common_name (mi->context.c2.tls_multi, false),
+				 mroute_addr_print (&mi->real, &gc),
+				 mi->context.c2.link_read_bytes,
+				 mi->context.c2.link_write_bytes,
+				 time_string (mi->created, 0, false, &gc));
+		}
+	      gc_free (&gc);
+	    }
+	  hash_iterator_free (&hi);
+
+	  status_printf (so, "ROUTING TABLE");
+	  status_printf (so, "Virtual Address,Common Name,Real Address,Last Ref");
+	  hash_iterator_init (m->vhash, &hi, true);
+	  while ((he = hash_iterator_next (&hi)))
+	    {
+	      struct gc_arena gc = gc_new ();
+	      const struct multi_route *route = (struct multi_route *) he->value;
+
+	      if (multi_route_defined (m, route))
+		{
+		  const struct multi_instance *mi = route->instance;
+		  const struct mroute_addr *ma = &route->addr;
+		  char flags[2] = {0, 0};
+
+		  if (route->flags & MULTI_ROUTE_CACHE)
+		    flags[0] = 'C';
+		  status_printf (so, "%s%s,%s,%s,%s",
+				 mroute_addr_print (ma, &gc),
+				 flags,
+				 tls_common_name (mi->context.c2.tls_multi, false),
+				 mroute_addr_print (&mi->real, &gc),
+				 time_string (route->last_reference, 0, false, &gc));
+		}
+	      gc_free (&gc);
+	    }
+	  hash_iterator_free (&hi);
+
+	  status_printf (so, "GLOBAL STATS");
+	  if (m->mbuf)
+	    status_printf (so, "Max bcast/mcast queue length,%d",
+			   mbuf_maximum_queued (m->mbuf));
+
+	  status_printf (so, "END");
+	}
+      else if (version == 2)
+	{
+	  /*
+	   * Status file version 2
+	   */
+	  status_printf (so, "TITLE,%s", title_string);
+	  status_printf (so, "TIME,%s,%u", time_string (now, 0, false, &gc_top), (unsigned int)now);
+	  status_printf (so, "HEADER,CLIENT_LIST,Common Name,Real Address,Virtual Address,Bytes Received,Bytes Sent,Connected Since,Connected Since (time_t)");
+	  hash_iterator_init (m->hash, &hi, true);
+	  while ((he = hash_iterator_next (&hi)))
+	    {
+	      struct gc_arena gc = gc_new ();
+	      const struct multi_instance *mi = (struct multi_instance *) he->value;
+
+	      if (!mi->halt)
+		{
+		  status_printf (so, "CLIENT_LIST,%s,%s,%s," counter_format "," counter_format ",%s,%u",
+				 tls_common_name (mi->context.c2.tls_multi, false),
+				 mroute_addr_print (&mi->real, &gc),
+				 print_in_addr_t (mi->reporting_addr, IA_EMPTY_IF_UNDEF, &gc),
+				 mi->context.c2.link_read_bytes,
+				 mi->context.c2.link_write_bytes,
+				 time_string (mi->created, 0, false, &gc),
+				 (unsigned int)mi->created);
+		}
+	      gc_free (&gc);
+	    }
+	  hash_iterator_free (&hi);
+
+	  status_printf (so, "HEADER,ROUTING_TABLE,Virtual Address,Common Name,Real Address,Last Ref,Last Ref (time_t)");
+	  hash_iterator_init (m->vhash, &hi, true);
+	  while ((he = hash_iterator_next (&hi)))
+	    {
+	      struct gc_arena gc = gc_new ();
+	      const struct multi_route *route = (struct multi_route *) he->value;
+
+	      if (multi_route_defined (m, route))
+		{
+		  const struct multi_instance *mi = route->instance;
+		  const struct mroute_addr *ma = &route->addr;
+		  char flags[2] = {0, 0};
+
+		  if (route->flags & MULTI_ROUTE_CACHE)
+		    flags[0] = 'C';
+		  status_printf (so, "ROUTING_TABLE,%s%s,%s,%s,%s,%u",
+				 mroute_addr_print (ma, &gc),
+				 flags,
+				 tls_common_name (mi->context.c2.tls_multi, false),
+				 mroute_addr_print (&mi->real, &gc),
+				 time_string (route->last_reference, 0, false, &gc),
+				 (unsigned int)route->last_reference);
+		}
+	      gc_free (&gc);
+	    }
+	  hash_iterator_free (&hi);
+
+	  if (m->mbuf)
+	    status_printf (so, "GLOBAL_STATS,Max bcast/mcast queue length,%d",
+			   mbuf_maximum_queued (m->mbuf));
+
+	  status_printf (so, "END");
+	}
+      else
+	{
+	  status_printf (so, "ERROR: bad status format version number");
+	}
+      status_flush (so);
+      gc_free (&gc_top);
+    }
+}
+
+/*
+ * Learn a virtual address or route.
+ * The learn will fail if the learn address
+ * script/plugin fails.  In this case the
+ * return value may be != mi.
+ * Return the instance which owns this route,
+ * or NULL if none.
+ */
+static struct multi_instance *
+multi_learn_addr (struct multi_context *m,
+		  struct multi_instance *mi,
+		  const struct mroute_addr *addr,
+		  const unsigned int flags)
+{
+  struct hash_element *he;
+  const uint32_t hv = hash_value (m->vhash, addr);
+  struct hash_bucket *bucket = hash_bucket (m->vhash, hv);
+  struct multi_route *oldroute = NULL;
+  struct multi_instance *owner = NULL;
+
+  hash_bucket_lock (bucket);
+
+  /* if route currently exists, get the instance which owns it */
+  he = hash_lookup_fast (m->vhash, bucket, addr, hv);
+  if (he)
+    oldroute = (struct multi_route *) he->value;
+  if (oldroute && multi_route_defined (m, oldroute))
+    owner = oldroute->instance;
+
+  /* do we need to add address to hash table? */
+  if ((!owner || owner != mi)
+      && mroute_learnable_address (addr)
+      && !mroute_addr_equal (addr, &m->local))
+    {
+      struct gc_arena gc = gc_new ();
+      struct multi_route *newroute;
+      bool learn_succeeded = false;
+
+      ALLOC_OBJ (newroute, struct multi_route);
+      newroute->addr = *addr;
+      newroute->instance = mi;
+      newroute->flags = flags;
+      newroute->last_reference = now;
+      newroute->cache_generation = 0;
+
+      /* The cache is invalidated when cache_generation is incremented */
+      if (flags & MULTI_ROUTE_CACHE)
+	newroute->cache_generation = m->route_helper->cache_generation;
+
+      if (oldroute) /* route already exists? */
+	{
+	  if (route_quota_test (m, mi) && learn_address_script (m, mi, "update", &newroute->addr))
+	    {
+	      learn_succeeded = true;
+	      owner = mi;
+	      multi_instance_inc_refcount (mi);
+	      route_quota_inc (mi);
+
+	      /* delete old route */
+	      multi_route_del (oldroute);
+
+	      /* modify hash table entry, replacing old route */
+	      he->key = &newroute->addr;
+	      he->value = newroute;
+	    }
+	}
+      else
+	{
+	  if (route_quota_test (m, mi) && learn_address_script (m, mi, "add", &newroute->addr))
+	    {
+	      learn_succeeded = true;
+	      owner = mi;
+	      multi_instance_inc_refcount (mi);
+	      route_quota_inc (mi);
+
+	      /* add new route */
+	      hash_add_fast (m->vhash, bucket, &newroute->addr, hv, newroute);
+	    }
+	}
+      
+      msg (D_MULTI_LOW, "MULTI: Learn%s: %s -> %s",
+	   learn_succeeded ? "" : " FAILED",
+	   mroute_addr_print (&newroute->addr, &gc),
+	   multi_instance_string (mi, false, &gc));
+
+      if (!learn_succeeded)
+	free (newroute);
+
+      gc_free (&gc);
+    }
+
+  hash_bucket_unlock (bucket);
+  return owner;
+}
+
+/*
+ * Get client instance based on virtual address.
+ */
+static struct multi_instance *
+multi_get_instance_by_virtual_addr (struct multi_context *m,
+				    const struct mroute_addr *addr,
+				    bool cidr_routing)
+{
+  struct multi_route *route;
+  struct multi_instance *ret = NULL;
+
+  /* check for local address */
+  if (mroute_addr_equal (addr, &m->local))
+    return NULL;
+
+  route = (struct multi_route *) hash_lookup (m->vhash, addr);
+
+  /* does host route (possible cached) exist? */
+  if (route && multi_route_defined (m, route))
+    {
+      struct multi_instance *mi = route->instance;
+      route->last_reference = now;
+      ret = mi;
+    }
+  else if (cidr_routing) /* do we need to regenerate a host route cache entry? */
+    {
+      struct mroute_helper *rh = m->route_helper;
+      struct mroute_addr tryaddr;
+      int i;
+
+      mroute_helper_lock (rh);
+
+      /* cycle through each CIDR length */
+      for (i = 0; i < rh->n_net_len; ++i)
+	{
+	  tryaddr = *addr;
+	  tryaddr.type |= MR_WITH_NETBITS;
+	  tryaddr.netbits = rh->net_len[i];
+	  mroute_addr_mask_host_bits (&tryaddr);
+
+	  /* look up a possible route with netbits netmask */
+	  route = (struct multi_route *) hash_lookup (m->vhash, &tryaddr);
+
+	  if (route && multi_route_defined (m, route))
+	    {
+	      /* found an applicable route, cache host route */
+	      struct multi_instance *mi = route->instance;
+	      multi_learn_addr (m, mi, addr, MULTI_ROUTE_CACHE|MULTI_ROUTE_AGEABLE);
+	      ret = mi;
+	      break;
+	    }
+	}
+      
+      mroute_helper_unlock (rh);
+    }
+  
+#ifdef ENABLE_DEBUG
+  if (check_debug_level (D_MULTI_DEBUG))
+    {
+      struct gc_arena gc = gc_new ();
+      const char *addr_text = mroute_addr_print (addr, &gc);
+      if (ret)
+	{
+	  dmsg (D_MULTI_DEBUG, "GET INST BY VIRT: %s -> %s via %s",
+	       addr_text,
+	       multi_instance_string (ret, false, &gc),
+	       mroute_addr_print (&route->addr, &gc));
+	}
+      else
+	{
+	  dmsg (D_MULTI_DEBUG, "GET INST BY VIRT: %s [failed]",
+	       addr_text);
+	}
+      gc_free (&gc);
+    }
+#endif
+
+  ASSERT (!(ret && ret->halt));
+  return ret;
+}
+
+/*
+ * Helper function to multi_learn_addr().
+ */
+static struct multi_instance *
+multi_learn_in_addr_t (struct multi_context *m,
+		       struct multi_instance *mi,
+		       in_addr_t a,
+		       int netbits) /* -1 if host route, otherwise # of network bits in address */
+{
+  struct sockaddr_in remote_si;
+  struct mroute_addr addr;
+
+  CLEAR (remote_si);
+  remote_si.sin_family = AF_INET;
+  remote_si.sin_addr.s_addr = htonl (a);
+  ASSERT (mroute_extract_sockaddr_in (&addr, &remote_si, false));
+
+  if (netbits >= 0)
+    {
+      addr.type |= MR_WITH_NETBITS;
+      addr.netbits = (uint8_t) netbits;
+    }
+  return multi_learn_addr (m, mi, &addr, 0);
+}
+
+/*
+ * A new client has connected, add routes (server -> client)
+ * to internal routing table.
+ */
+static void
+multi_add_iroutes (struct multi_context *m,
+		   struct multi_instance *mi)
+{
+  struct gc_arena gc = gc_new ();
+  const struct iroute *ir;
+  if (TUNNEL_TYPE (mi->context.c1.tuntap) == DEV_TYPE_TUN)
+    {
+      mi->did_iroutes = true;
+      for (ir = mi->context.options.iroutes; ir != NULL; ir = ir->next)
+	{
+	  if (ir->netbits >= 0)
+	    msg (D_MULTI_LOW, "MULTI: internal route %s/%d -> %s",
+		 print_in_addr_t (ir->network, 0, &gc),
+		 ir->netbits,
+		 multi_instance_string (mi, false, &gc));
+	  else
+	    msg (D_MULTI_LOW, "MULTI: internal route %s -> %s",
+		 print_in_addr_t (ir->network, 0, &gc),
+		 multi_instance_string (mi, false, &gc));
+
+	  mroute_helper_add_iroute (m->route_helper, ir);
+      
+	  multi_learn_in_addr_t (m, mi, ir->network, ir->netbits);
+	}
+    }
+  gc_free (&gc);
+}
+
+/*
+ * Given an instance (new_mi), delete all other instances which use the
+ * same common name.
+ */
+static void
+multi_delete_dup (struct multi_context *m, struct multi_instance *new_mi)
+{
+  if (new_mi)
+    {
+      const char *new_cn = tls_common_name (new_mi->context.c2.tls_multi, true);
+      if (new_cn)
+	{
+	  struct hash_iterator hi;
+	  struct hash_element *he;
+
+	  hash_iterator_init (m->iter, &hi, true);
+	  while ((he = hash_iterator_next (&hi)))
+	    {
+	      struct multi_instance *mi = (struct multi_instance *) he->value;
+	      if (mi != new_mi && !mi->halt)
+		{
+		  const char *cn = tls_common_name (mi->context.c2.tls_multi, true);
+		  if (cn && !strcmp (cn, new_cn))
+		    {
+		      mi->did_iter = false;
+		      multi_close_instance (m, mi, false);
+		      hash_iterator_delete_element (&hi);
+		    }
+		}
+	    }
+	  hash_iterator_free (&hi);
+	}
+    }
+}
+
+/*
+ * Select a virtual address for a new client instance.
+ * Use an --ifconfig-push directive, if given (static IP).
+ * Otherwise use an --ifconfig-pool address (dynamic IP). 
+ */
+static void
+multi_select_virtual_addr (struct multi_context *m, struct multi_instance *mi)
+{
+  struct gc_arena gc = gc_new ();
+
+  /*
+   * If ifconfig addresses were set by dynamic config file,
+   * release pool addresses, otherwise keep them.
+   */
+  if (mi->context.options.push_ifconfig_defined)
+    {
+      /* ifconfig addresses were set statically,
+	 release dynamic allocation */
+      if (mi->vaddr_handle >= 0)
+	{
+	  ifconfig_pool_release (m->ifconfig_pool, mi->vaddr_handle, true);
+	  mi->vaddr_handle = -1;
+	}
+
+      mi->context.c2.push_ifconfig_defined = true;
+      mi->context.c2.push_ifconfig_local = mi->context.options.push_ifconfig_local;
+      mi->context.c2.push_ifconfig_remote_netmask = mi->context.options.push_ifconfig_remote_netmask;
+    }
+  else if (m->ifconfig_pool && mi->vaddr_handle < 0) /* otherwise, choose a pool address */
+    {
+      in_addr_t local=0, remote=0;
+      const char *cn = NULL;
+
+      if (!mi->context.options.duplicate_cn)
+	cn = tls_common_name (mi->context.c2.tls_multi, true);
+
+      mi->vaddr_handle = ifconfig_pool_acquire (m->ifconfig_pool, &local, &remote, cn);
+      if (mi->vaddr_handle >= 0)
+	{
+	  /* use pool ifconfig address(es) */
+	  mi->context.c2.push_ifconfig_local = remote;
+	  if (TUNNEL_TYPE (mi->context.c1.tuntap) == DEV_TYPE_TUN)
+	    {
+	      if (mi->context.options.ifconfig_pool_linear)		    
+		mi->context.c2.push_ifconfig_remote_netmask = mi->context.c1.tuntap->local;
+	      else
+		mi->context.c2.push_ifconfig_remote_netmask = local;
+	      mi->context.c2.push_ifconfig_defined = true;
+	    }
+	  else if (TUNNEL_TYPE (mi->context.c1.tuntap) == DEV_TYPE_TAP)
+	    {
+	      mi->context.c2.push_ifconfig_remote_netmask = mi->context.options.ifconfig_pool_netmask;
+	      if (!mi->context.c2.push_ifconfig_remote_netmask)
+		mi->context.c2.push_ifconfig_remote_netmask = mi->context.c1.tuntap->remote_netmask;
+	      if (mi->context.c2.push_ifconfig_remote_netmask)
+		mi->context.c2.push_ifconfig_defined = true;
+	      else
+		msg (D_MULTI_ERRORS, "MULTI: no --ifconfig-pool netmask parameter is available to push to %s",
+		     multi_instance_string (mi, false, &gc));
+	    }
+	}
+      else
+	{
+	  msg (D_MULTI_ERRORS, "MULTI: no free --ifconfig-pool addresses are available");
+	}
+    }
+  gc_free (&gc);
+}
+
+/*
+ * Set virtual address environmental variables.
+ */
+static void
+multi_set_virtual_addr_env (struct multi_context *m, struct multi_instance *mi)
+{
+  setenv_del (mi->context.c2.es, "ifconfig_pool_local_ip");
+  setenv_del (mi->context.c2.es, "ifconfig_pool_remote_ip");
+  setenv_del (mi->context.c2.es, "ifconfig_pool_netmask");
+
+  if (mi->context.c2.push_ifconfig_defined)
+    {
+      setenv_in_addr_t (mi->context.c2.es,
+			"ifconfig_pool_remote_ip",
+			mi->context.c2.push_ifconfig_local,
+			SA_SET_IF_NONZERO);
+
+      if (TUNNEL_TYPE (mi->context.c1.tuntap) == DEV_TYPE_TUN)
+	{
+	  setenv_in_addr_t (mi->context.c2.es,
+			    "ifconfig_pool_local_ip",
+			    mi->context.c2.push_ifconfig_remote_netmask,
+			    SA_SET_IF_NONZERO);
+	}
+      else if (TUNNEL_TYPE (mi->context.c1.tuntap) == DEV_TYPE_TAP)
+	{
+	  setenv_in_addr_t (mi->context.c2.es,
+			    "ifconfig_pool_netmask",
+			    mi->context.c2.push_ifconfig_remote_netmask,
+			    SA_SET_IF_NONZERO);
+	}
+    }
+}
+
+/*
+ * Called after client-connect script or plug-in is called
+ */
+static void
+multi_client_connect_post (struct multi_context *m,
+			   struct multi_instance *mi,
+			   const char *dc_file,
+			   unsigned int option_permissions_mask,
+			   unsigned int *option_types_found)
+{
+  /* Did script generate a dynamic config file? */
+  if (test_file (dc_file))
+    {
+      options_server_import (&mi->context.options,
+			     dc_file,
+			     D_IMPORT_ERRORS|M_OPTERR,
+			     option_permissions_mask,
+			     option_types_found,
+			     mi->context.c2.es);
+
+      if (!delete_file (dc_file))
+	msg (D_MULTI_ERRORS, "MULTI: problem deleting temporary file: %s",
+	     dc_file);
+
+      /*
+       * If the --client-connect script generates a config file
+       * with an --ifconfig-push directive, it will override any
+       * --ifconfig-push directive from the --client-config-dir
+       * directory or any --ifconfig-pool dynamic address.
+       */
+      multi_select_virtual_addr (m, mi);
+      multi_set_virtual_addr_env (m, mi);
+    }
+}
+
+/*
+ * Called as soon as the SSL/TLS connection authenticates.
+ *
+ * Instance-specific directives to be processed:
+ *
+ *   iroute start-ip end-ip
+ *   ifconfig-push local remote-netmask
+ *   push
+ */
+static void
+multi_connection_established (struct multi_context *m, struct multi_instance *mi)
+{
+  if (tls_authenticated (mi->context.c2.tls_multi))
+    {
+      struct gc_arena gc = gc_new ();
+      unsigned int option_types_found = 0;
+      const unsigned int option_permissions_mask = OPT_P_INSTANCE|OPT_P_INHERIT|OPT_P_PUSH|OPT_P_TIMER|OPT_P_CONFIG|OPT_P_ECHO;
+      int cc_succeeded = true; /* client connect script status */
+      int cc_succeeded_count = 0;
+
+      ASSERT (mi->context.c1.tuntap);
+
+      /* lock down the common name so it can't change during future TLS renegotiations */
+      tls_lock_common_name (mi->context.c2.tls_multi);
+
+      /* generate a msg() prefix for this client instance */
+      generate_prefix (mi);
+
+      /* delete instances of previous clients with same common-name */
+      if (!mi->context.options.duplicate_cn)
+	multi_delete_dup (m, mi);
+
+      /* reset pool handle to null */
+      mi->vaddr_handle = -1;
+
+      /*
+       * Try to source a dynamic config file from the
+       * --client-config-dir directory.
+       */
+      if (mi->context.options.client_config_dir)
+	{
+	  const char *ccd_file;
+	  
+	  ccd_file = gen_path (mi->context.options.client_config_dir,
+			       tls_common_name (mi->context.c2.tls_multi, false),
+			       &gc);
+
+	  /* try common-name file */
+	  if (test_file (ccd_file))
+	    {
+	      options_server_import (&mi->context.options,
+				     ccd_file,
+				     D_IMPORT_ERRORS|M_OPTERR,
+				     option_permissions_mask,
+				     &option_types_found,
+				     mi->context.c2.es);
+	    }
+	  else /* try default file */
+	    {
+	      ccd_file = gen_path (mi->context.options.client_config_dir,
+				   CCD_DEFAULT,
+				   &gc);
+
+	      if (test_file (ccd_file))
+		{
+		  options_server_import (&mi->context.options,
+					 ccd_file,
+					 D_IMPORT_ERRORS|M_OPTERR,
+					 option_permissions_mask,
+					 &option_types_found,
+					 mi->context.c2.es);
+		}
+	    }
+	}
+
+      /*
+       * Select a virtual address from either --ifconfig-push in --client-config-dir file
+       * or --ifconfig-pool.
+       */
+      multi_select_virtual_addr (m, mi);
+
+      /* setenv incoming cert common name for script */
+      setenv_str (mi->context.c2.es, "common_name", tls_common_name (mi->context.c2.tls_multi, true));
+
+      /* setenv client real IP address */
+      setenv_trusted (mi->context.c2.es, get_link_socket_info (&mi->context));
+
+      /* setenv client virtual IP address */
+      multi_set_virtual_addr_env (m, mi);
+
+      /*
+       * Call client-connect plug-in.
+       */
+      if (plugin_defined (m->top.c1.plugins, OPENVPN_PLUGIN_CLIENT_CONNECT))
+	{
+	  const char *dc_file = create_temp_filename (mi->context.options.tmp_dir, &gc);
+
+	  delete_file (dc_file);
+
+	  if (plugin_call (m->top.c1.plugins, OPENVPN_PLUGIN_CLIENT_CONNECT, dc_file, mi->context.c2.es))
+	    {
+	      msg (M_WARN, "WARNING: client-connect plugin call failed");
+	      cc_succeeded = false;
+	    }
+	  else
+	    {
+	      multi_client_connect_post (m, mi, dc_file, option_permissions_mask, &option_types_found);
+	      ++cc_succeeded_count;
+	    }
+	}
+
+      /*
+       * Run --client-connect script.
+       */
+      if (mi->context.options.client_connect_script && cc_succeeded)
+	{
+	  struct buffer cmd = alloc_buf_gc (256, &gc);
+	  const char *dc_file = NULL;
+
+	  setenv_str (mi->context.c2.es, "script_type", "client-connect");
+
+	  dc_file = create_temp_filename (mi->context.options.tmp_dir, &gc);
+
+	  delete_file (dc_file);
+
+	  buf_printf (&cmd, "%s %s",
+		      mi->context.options.client_connect_script,
+		      dc_file);
+
+	  if (system_check (BSTR (&cmd), mi->context.c2.es, S_SCRIPT, "client-connect command failed"))
+	    {
+	      multi_client_connect_post (m, mi, dc_file, option_permissions_mask, &option_types_found);
+	      ++cc_succeeded_count;
+	    }
+	  else
+	    cc_succeeded = false;
+	}
+
+      /*
+       * Check for "disable" directive in client-config-dir file
+       * or config file generated by --client-connect script.
+       */
+      if (mi->context.options.disable)
+	{
+	  msg (D_MULTI_ERRORS, "MULTI: client has been rejected due to 'disable' directive");
+	  cc_succeeded = false;
+	}
+
+      if (cc_succeeded)
+	{
+	  /*
+	   * Process sourced options.
+	   */
+	  do_deferred_options (&mi->context, option_types_found);
+
+	  /*
+	   * make sure we got ifconfig settings from somewhere
+	   */
+	  if (!mi->context.c2.push_ifconfig_defined)
+	    {
+	      msg (D_MULTI_ERRORS, "MULTI: no dynamic or static remote --ifconfig address is available for %s",
+		   multi_instance_string (mi, false, &gc));
+	    }
+
+	  /*
+	   * For routed tunnels, set up internal route to endpoint
+	   * plus add all iroute routes.
+	   */
+	  if (TUNNEL_TYPE (mi->context.c1.tuntap) == DEV_TYPE_TUN)
+	    {
+	      if (mi->context.c2.push_ifconfig_defined)
+		{
+		  multi_learn_in_addr_t (m, mi, mi->context.c2.push_ifconfig_local, -1);
+		  msg (D_MULTI_LOW, "MULTI: primary virtual IP for %s: %s",
+		       multi_instance_string (mi, false, &gc),
+		       print_in_addr_t (mi->context.c2.push_ifconfig_local, 0, &gc));
+		}
+
+	      /* add routes locally, pointing to new client, if
+		 --iroute options have been specified */
+	      multi_add_iroutes (m, mi);
+
+	      /*
+	       * iroutes represent subnets which are "owned" by a particular
+	       * client.  Therefore, do not actually push a route to a client
+	       * if it matches one of the client's iroutes.
+	       */
+	      remove_iroutes_from_push_route_list (&mi->context.options);
+	    }
+	  else if (mi->context.options.iroutes)
+	    {
+	      msg (D_MULTI_ERRORS, "MULTI: --iroute options rejected for %s -- iroute only works with tun-style tunnels",
+		   multi_instance_string (mi, false, &gc));
+	    }
+
+	  /* set our client's VPN endpoint for status reporting purposes */
+	  mi->reporting_addr = mi->context.c2.push_ifconfig_local;
+
+	  /* set context-level authentication flag */
+	  mi->context.c2.context_auth = CAS_SUCCEEDED;
+	}
+      else
+	{
+	  /* set context-level authentication flag */
+	  mi->context.c2.context_auth = cc_succeeded_count ? CAS_PARTIAL : CAS_FAILED;
+	}
+
+      /* set flag so we don't get called again */
+      mi->connection_established_flag = true;
+
+      gc_free (&gc);
+    }
+
+  /*
+   * Reply now to client's PUSH_REQUEST query
+   */
+  mi->context.c2.push_reply_deferred = false;
+}
+
+/*
+ * Add a mbuf buffer to a particular
+ * instance.
+ */
+void
+multi_add_mbuf (struct multi_context *m,
+		struct multi_instance *mi,
+		struct mbuf_buffer *mb)
+{
+  if (multi_output_queue_ready (m, mi))
+    {
+      struct mbuf_item item;
+      item.buffer = mb;
+      item.instance = mi;
+      mbuf_add_item (m->mbuf, &item);
+    }
+  else
+    {
+      msg (D_MULTI_DROPPED, "MULTI: packet dropped due to output saturation (multi_add_mbuf)");
+    }
+}
+
+/*
+ * Add a packet to a client instance output queue.
+ */
+static inline void
+multi_unicast (struct multi_context *m,
+	       const struct buffer *buf,
+	       struct multi_instance *mi)
+{
+  struct mbuf_buffer *mb;
+
+  if (BLEN (buf) > 0)
+    {
+      mb = mbuf_alloc_buf (buf);
+      mb->flags = MF_UNICAST;
+      multi_add_mbuf (m, mi, mb);
+      mbuf_free_buf (mb);
+    }
+}
+
+/*
+ * Broadcast a packet to all clients.
+ */
+void
+multi_bcast (struct multi_context *m,
+	     const struct buffer *buf,
+	     struct multi_instance *omit)
+{
+  struct hash_iterator hi;
+  struct hash_element *he;
+  struct multi_instance *mi;
+  struct mbuf_buffer *mb;
+
+  if (BLEN (buf) > 0)
+    {
+      perf_push (PERF_MULTI_BCAST);
+#ifdef MULTI_DEBUG_EVENT_LOOP
+      printf ("BCAST len=%d\n", BLEN (buf));
+#endif
+      mb = mbuf_alloc_buf (buf);
+      hash_iterator_init (m->iter, &hi, true);
+
+      while ((he = hash_iterator_next (&hi)))
+	{
+	  mi = (struct multi_instance *) he->value;
+	  if (mi != omit && !mi->halt)
+	    multi_add_mbuf (m, mi, mb);
+	}
+
+      hash_iterator_free (&hi);
+      mbuf_free_buf (mb);
+      perf_pop ();
+    }
+}
+
+/*
+ * Given a time delta, indicating that we wish to be
+ * awoken by the scheduler at time now + delta, figure
+ * a sigma parameter (in microseconds) that represents
+ * a sort of fuzz factor around delta, so that we're
+ * really telling the scheduler to wake us up any time
+ * between now + delta - sigma and now + delta + sigma.
+ *
+ * The sigma parameter helps the scheduler to run more efficiently.
+ * Sigma should be no larger than TV_WITHIN_SIGMA_MAX_USEC
+ */
+static inline unsigned int
+compute_wakeup_sigma (const struct timeval *delta)
+{
+  if (delta->tv_sec < 1)
+    {
+      /* if < 1 sec, fuzz = # of microseconds / 8 */
+      return delta->tv_usec >> 3;
+    }
+  else
+    {
+      /* if < 10 minutes, fuzz = 13.1% of timeout */
+      if (delta->tv_sec < 600)
+	return delta->tv_sec << 17;
+      else
+	return 120000000; /* if >= 10 minutes, fuzz = 2 minutes */
+    }
+}
+
+/*
+ * Figure instance-specific timers, convert
+ * earliest to absolute time in mi->wakeup,
+ * call scheduler with our future wakeup time.
+ *
+ * Also close context on signal.
+ */
+bool
+multi_process_post (struct multi_context *m, struct multi_instance *mi, const unsigned int flags)
+{
+  bool ret = true;
+
+  if (!IS_SIG (&mi->context) && ((flags & MPP_PRE_SELECT) || ((flags & MPP_CONDITIONAL_PRE_SELECT) && !ANY_OUT (&mi->context))))
+    {
+      /* figure timeouts and fetch possible outgoing
+	 to_link packets (such as ping or TLS control) */
+      pre_select (&mi->context);
+
+      if (!IS_SIG (&mi->context))
+	{
+	  /* calculate an absolute wakeup time */
+	  ASSERT (!gettimeofday (&mi->wakeup, NULL));
+	  tv_add (&mi->wakeup, &mi->context.c2.timeval);
+
+	  /* tell scheduler to wake us up at some point in the future */
+	  schedule_add_entry (m->schedule,
+			      (struct schedule_entry *) mi,
+			      &mi->wakeup,
+			      compute_wakeup_sigma (&mi->context.c2.timeval));
+
+	  /* connection is "established" when SSL/TLS key negotiation succeeds
+	     and (if specified) auth user/pass succeeds */
+	  if (!mi->connection_established_flag && CONNECTION_ESTABLISHED (&mi->context))
+	    multi_connection_established (m, mi);
+	}
+    }
+
+  if (IS_SIG (&mi->context))
+    {
+      if (flags & MPP_CLOSE_ON_SIGNAL)
+	{
+	  multi_close_instance_on_signal (m, mi);
+	  ret = false;
+	}
+    }
+  else
+    {
+      /* continue to pend on output? */
+      multi_set_pending (m, ANY_OUT (&mi->context) ? mi : NULL);
+
+#ifdef MULTI_DEBUG_EVENT_LOOP
+      printf ("POST %s[%d] to=%d lo=%d/%d w=%d/%d\n",
+	      id(mi),
+	      (int) (mi == m->pending),
+	      mi ? mi->context.c2.to_tun.len : -1,
+	      mi ? mi->context.c2.to_link.len : -1,
+	      (mi && mi->context.c2.fragment) ? mi->context.c2.fragment->outgoing.len : -1,
+	      (int)mi->context.c2.timeval.tv_sec,
+	      (int)mi->context.c2.timeval.tv_usec);
+#endif
+    }
+
+  if ((flags & MPP_RECORD_TOUCH) && m->mpp_touched)
+    *m->mpp_touched = mi;
+
+  return ret;
+}
+
+/*
+ * Process packets in the TCP/UDP socket -> TUN/TAP interface direction,
+ * i.e. client -> server direction.
+ */
+bool
+multi_process_incoming_link (struct multi_context *m, struct multi_instance *instance, const unsigned int mpp_flags)
+{
+  struct gc_arena gc = gc_new ();
+
+  struct context *c;
+  struct mroute_addr src, dest;
+  unsigned int mroute_flags;
+  struct multi_instance *mi;
+  bool ret = true;
+
+  ASSERT (!m->pending);
+
+  if (!instance)
+    {
+#ifdef MULTI_DEBUG_EVENT_LOOP
+      printf ("TCP/UDP -> TUN [%d]\n", BLEN (&m->top.c2.buf));
+#endif
+      multi_set_pending (m, multi_get_create_instance_udp (m));
+    }
+  else
+    multi_set_pending (m, instance);
+
+  if (m->pending)
+    {
+      set_prefix (m->pending);
+
+      /* get instance context */
+      c = &m->pending->context;
+
+      if (!instance)
+	{
+	  /* transfer packet pointer from top-level context buffer to instance */
+	  c->c2.buf = m->top.c2.buf;
+
+	  /* transfer from-addr from top-level context buffer to instance */
+	  c->c2.from = m->top.c2.from;
+	}
+
+      if (BLEN (&c->c2.buf) > 0)
+	{
+	  /* decrypt in instance context */
+	  process_incoming_link (c);
+
+	  if (TUNNEL_TYPE (m->top.c1.tuntap) == DEV_TYPE_TUN)
+	    {
+	      /* extract packet source and dest addresses */
+	      mroute_flags = mroute_extract_addr_from_packet (&src,
+							      &dest,
+							      &c->c2.to_tun,
+							      DEV_TYPE_TUN);
+
+	      /* drop packet if extract failed */
+	      if (!(mroute_flags & MROUTE_EXTRACT_SUCCEEDED))
+		{
+		  c->c2.to_tun.len = 0;
+		}
+	      /* make sure that source address is associated with this client */
+	      else if (multi_get_instance_by_virtual_addr (m, &src, true) != m->pending)
+		{
+		  msg (D_MULTI_DROPPED, "MULTI: bad source address from client [%s], packet dropped",
+		       mroute_addr_print (&src, &gc));
+		  c->c2.to_tun.len = 0;
+		}
+	      /* client-to-client communication enabled? */
+	      else if (m->enable_c2c)
+		{
+		  /* multicast? */
+		  if (mroute_flags & MROUTE_EXTRACT_MCAST)
+		    {
+		      /* for now, treat multicast as broadcast */
+		      multi_bcast (m, &c->c2.to_tun, m->pending);
+		    }
+		  else /* possible client to client routing */
+		    {
+		      ASSERT (!(mroute_flags & MROUTE_EXTRACT_BCAST));
+		      mi = multi_get_instance_by_virtual_addr (m, &dest, true);
+		  
+		      /* if dest addr is a known client, route to it */
+		      if (mi)
+			{
+			  multi_unicast (m, &c->c2.to_tun, mi);
+			  register_activity (c);
+			  c->c2.to_tun.len = 0;
+			}
+		    }
+		}
+	    }
+	  else if (TUNNEL_TYPE (m->top.c1.tuntap) == DEV_TYPE_TAP)
+	    {
+	      /* extract packet source and dest addresses */
+	      mroute_flags = mroute_extract_addr_from_packet (&src,
+							      &dest,
+							      &c->c2.to_tun,
+							      DEV_TYPE_TAP);
+
+	      if (mroute_flags & MROUTE_EXTRACT_SUCCEEDED)
+		{
+		  if (multi_learn_addr (m, m->pending, &src, 0) == m->pending)
+		    {
+		      /* check for broadcast */
+		      if (m->enable_c2c)
+			{
+			  if (mroute_flags & (MROUTE_EXTRACT_BCAST|MROUTE_EXTRACT_MCAST))
+			    {
+			      multi_bcast (m, &c->c2.to_tun, m->pending);
+			    }
+			  else /* try client-to-client routing */
+			    {
+			      mi = multi_get_instance_by_virtual_addr (m, &dest, false);
+
+			      /* if dest addr is a known client, route to it */
+			      if (mi)
+				{
+				  multi_unicast (m, &c->c2.to_tun, mi);
+				  register_activity (c);
+				  c->c2.to_tun.len = 0;
+				}
+			    }
+			}
+		    }
+		  else
+		    {
+		      msg (D_MULTI_DROPPED, "MULTI: bad source address from client [%s], packet dropped",
+			   mroute_addr_print (&src, &gc));
+		      c->c2.to_tun.len = 0;
+		    }
+		}
+	      else
+		{
+		  c->c2.to_tun.len = 0;
+		}
+	    }
+	}
+
+      /* postprocess and set wakeup */
+      ret = multi_process_post (m, m->pending, mpp_flags);
+
+      clear_prefix ();
+    }
+
+  gc_free (&gc);
+  return ret;
+}
+
+/*
+ * Process packets in the TUN/TAP interface -> TCP/UDP socket direction,
+ * i.e. server -> client direction.
+ */
+bool
+multi_process_incoming_tun (struct multi_context *m, const unsigned int mpp_flags)
+{
+  struct gc_arena gc = gc_new ();
+  bool ret = true;
+
+  if (BLEN (&m->top.c2.buf) > 0)
+    {
+      unsigned int mroute_flags;
+      struct mroute_addr src, dest;
+      const int dev_type = TUNNEL_TYPE (m->top.c1.tuntap);
+
+#ifdef MULTI_DEBUG_EVENT_LOOP
+      printf ("TUN -> TCP/UDP [%d]\n", BLEN (&m->top.c2.buf));
+#endif
+
+      ASSERT (!m->pending);
+
+      /* 
+       * Route an incoming tun/tap packet to
+       * the appropriate multi_instance object.
+       */
+
+      mroute_flags = mroute_extract_addr_from_packet (&src,
+						      &dest,
+						      &m->top.c2.buf,
+						      dev_type);
+
+      if (mroute_flags & MROUTE_EXTRACT_SUCCEEDED)
+	{
+	  struct context *c;
+
+	  /* broadcast or multicast dest addr? */
+	  if (mroute_flags & (MROUTE_EXTRACT_BCAST|MROUTE_EXTRACT_MCAST))
+	    {
+	      /* for now, treat multicast as broadcast */
+	      multi_bcast (m, &m->top.c2.buf, NULL);
+	    }
+	  else
+	    {
+	      multi_set_pending (m, multi_get_instance_by_virtual_addr (m, &dest, dev_type == DEV_TYPE_TUN));
+
+	      if (m->pending)
+		{
+		  /* get instance context */
+		  c = &m->pending->context;
+		  
+		  set_prefix (m->pending);
+
+		  if (multi_output_queue_ready (m, m->pending))
+		    {
+		      /* transfer packet pointer from top-level context buffer to instance */
+		      c->c2.buf = m->top.c2.buf;
+		    }
+		  else
+		    {
+		      /* drop packet */
+		      msg (D_MULTI_DROPPED, "MULTI: packet dropped due to output saturation (multi_process_incoming_tun)");
+		      buf_clear (&c->c2.buf);
+		    }
+	      
+		  /* encrypt in instance context */
+		  process_incoming_tun (c);
+
+		  /* postprocess and set wakeup */
+		  ret = multi_process_post (m, m->pending, mpp_flags);
+
+		  clear_prefix ();
+		}
+	    }
+	}
+    }
+  gc_free (&gc);
+  return ret;
+}
+
+/*
+ * Process a possible client-to-client/bcast/mcast message in the
+ * queue.
+ */
+struct multi_instance *
+multi_get_queue (struct mbuf_set *ms)
+{
+  struct mbuf_item item;
+
+  if (mbuf_extract_item (ms, &item, true)) /* cleartext IP packet */
+    {
+      unsigned int pipv4_flags = PIPV4_PASSTOS;
+
+      set_prefix (item.instance);
+      item.instance->context.c2.buf = item.buffer->buf;
+      if (item.buffer->flags & MF_UNICAST) /* --mssfix doesn't make sense for broadcast or multicast */
+	pipv4_flags |= PIPV4_MSSFIX;
+      process_ipv4_header (&item.instance->context, pipv4_flags, &item.instance->context.c2.buf);
+      encrypt_sign (&item.instance->context, true);
+      mbuf_free_buf (item.buffer);
+
+      dmsg (D_MULTI_DEBUG, "MULTI: C2C/MCAST/BCAST");
+
+      clear_prefix ();
+      return item.instance;
+    }
+  else
+    {
+      return NULL;
+    }
+}
+
+/*
+ * Called when an I/O wait times out.  Usually means that a particular
+ * client instance object needs timer-based service.
+ */
+bool
+multi_process_timeout (struct multi_context *m, const unsigned int mpp_flags)
+{
+  bool ret = true;
+
+#ifdef MULTI_DEBUG_EVENT_LOOP
+  printf ("%s -> TIMEOUT\n", id(m->earliest_wakeup));
+#endif
+
+  /* instance marked for wakeup? */
+  if (m->earliest_wakeup)
+    {
+      set_prefix (m->earliest_wakeup);
+      ret = multi_process_post (m, m->earliest_wakeup, mpp_flags);
+      m->earliest_wakeup = NULL;
+      clear_prefix ();
+    }
+  return ret;
+}
+
+/*
+ * Drop a TUN/TAP outgoing packet..
+ */
+void
+multi_process_drop_outgoing_tun (struct multi_context *m, const unsigned int mpp_flags)
+{
+  struct multi_instance *mi = m->pending;
+
+  ASSERT (mi);
+
+  set_prefix (mi);
+
+  msg (D_MULTI_ERRORS, "MULTI: Outgoing TUN queue full, dropped packet len=%d",
+       mi->context.c2.to_tun.len);
+
+  buf_reset (&mi->context.c2.to_tun);
+
+  multi_process_post (m, mi, mpp_flags);
+  clear_prefix ();
+}
+
+/*
+ * Per-client route quota management
+ */
+
+void
+route_quota_exceeded (const struct multi_context *m, const struct multi_instance *mi)
+{
+  struct gc_arena gc = gc_new ();
+  msg (D_ROUTE_QUOTA, "MULTI ROUTE: route quota (%d) exceeded for %s (see --max-routes-per-client option)",
+	mi->context.options.max_routes_per_client,
+	multi_instance_string (mi, false, &gc));
+  gc_free (&gc);
+}
+
+#ifdef ENABLE_DEBUG
+/*
+ * Flood clients with random packets
+ */
+static void
+gremlin_flood_clients (struct multi_context *m)
+{
+  const int level = GREMLIN_PACKET_FLOOD_LEVEL (m->top.options.gremlin);
+  if (level)
+    {
+      struct gc_arena gc = gc_new ();
+      struct buffer buf = alloc_buf_gc (BUF_SIZE (&m->top.c2.frame), &gc);
+      struct packet_flood_parms parm = get_packet_flood_parms (level);
+      int i;
+
+      ASSERT (buf_init (&buf, FRAME_HEADROOM (&m->top.c2.frame)));
+      parm.packet_size = min_int (parm.packet_size, MAX_RW_SIZE_TUN (&m->top.c2.frame));
+
+      msg (D_GREMLIN, "GREMLIN_FLOOD_CLIENTS: flooding clients with %d packets of size %d",
+	   parm.n_packets,
+	   parm.packet_size);
+
+      for (i = 0; i < parm.packet_size; ++i)
+	ASSERT (buf_write_u8 (&buf, get_random () & 0xFF));
+
+      for (i = 0; i < parm.n_packets; ++i)
+	multi_bcast (m, &buf, NULL);
+
+      gc_free (&gc);
+    }
+}
+#endif
+
+/*
+ * Process timers in the top-level context
+ */
+void
+multi_process_per_second_timers_dowork (struct multi_context *m)
+{
+  /* possibly reap instances/routes in vhash */
+  multi_reap_process (m);
+
+  /* possibly print to status log */
+  if (m->top.c1.status_output)
+    {
+      if (status_trigger (m->top.c1.status_output))
+	multi_print_status (m, m->top.c1.status_output, m->status_file_version);
+    }
+
+  /* possibly flush ifconfig-pool file */
+  multi_ifconfig_pool_persist (m, false);
+
+#ifdef ENABLE_DEBUG
+  gremlin_flood_clients (m);
+#endif
+}
+
+void
+multi_top_init (struct multi_context *m, const struct context *top, const bool alloc_buffers)
+{
+  inherit_context_top (&m->top, top);
+  m->top.c2.buffers = NULL;
+  if (alloc_buffers)
+    m->top.c2.buffers = init_context_buffers (&top->c2.frame);
+}
+
+void
+multi_top_free (struct multi_context *m)
+{
+  close_context (&m->top, -1, CC_GC_FREE);
+  free_context_buffers (m->top.c2.buffers);
+}
+
+/*
+ * Return true if event loop should break,
+ * false if it should continue.
+ */
+bool
+multi_process_signal (struct multi_context *m)
+{
+  if (m->top.sig->signal_received == SIGUSR2)
+    {
+      struct status_output *so = status_open (NULL, 0, M_INFO, NULL, 0);
+      multi_print_status (m, so, m->status_file_version);
+      status_close (so);
+      m->top.sig->signal_received = 0;
+      return false;
+    }
+  return true;
+}
+
+/*
+ * Called when an instance should be closed due to the
+ * reception of a soft signal.
+ */
+void
+multi_close_instance_on_signal (struct multi_context *m, struct multi_instance *mi)
+{
+  remap_signal (&mi->context);
+  set_prefix (mi);
+  print_signal (mi->context.sig, "client-instance", D_MULTI_LOW);
+  clear_prefix ();
+  multi_close_instance (m, mi, false);
+}
+
+static void
+multi_signal_instance (struct multi_context *m, struct multi_instance *mi, const int sig)
+{
+  mi->context.sig->signal_received = sig;
+  multi_close_instance_on_signal (m, mi);
+}
+
+/*
+ * Management subsystem callbacks
+ */
+
+#ifdef ENABLE_MANAGEMENT
+
+static void
+management_callback_status (void *arg, const int version, struct status_output *so)
+{
+  struct multi_context *m = (struct multi_context *) arg;
+
+  if (!version)
+    multi_print_status (m, so, m->status_file_version);
+  else
+    multi_print_status (m, so, version);
+}
+
+static int
+management_callback_kill_by_cn (void *arg, const char *del_cn)
+{
+  struct multi_context *m = (struct multi_context *) arg;
+  struct hash_iterator hi;
+  struct hash_element *he;
+  int count = 0;
+
+  hash_iterator_init (m->iter, &hi, true);
+  while ((he = hash_iterator_next (&hi)))
+    {
+      struct multi_instance *mi = (struct multi_instance *) he->value;
+      if (!mi->halt)
+	{
+	  const char *cn = tls_common_name (mi->context.c2.tls_multi, false);
+	  if (cn && !strcmp (cn, del_cn))
+	    {
+	      multi_signal_instance (m, mi, SIGTERM);
+	      ++count;
+	    }
+	}
+    }
+  hash_iterator_free (&hi);
+  return count;
+}
+
+static int
+management_callback_kill_by_addr (void *arg, const in_addr_t addr, const int port)
+{
+  struct multi_context *m = (struct multi_context *) arg;
+  struct hash_iterator hi;
+  struct hash_element *he;
+  struct sockaddr_in saddr;
+  struct mroute_addr maddr;
+  int count = 0;
+
+  CLEAR (saddr);
+  saddr.sin_family = AF_INET;
+  saddr.sin_addr.s_addr = htonl (addr);
+  saddr.sin_port = htons (port);
+  if (mroute_extract_sockaddr_in (&maddr, &saddr, true))
+    {
+      hash_iterator_init (m->iter, &hi, true);
+      while ((he = hash_iterator_next (&hi)))
+	{
+	  struct multi_instance *mi = (struct multi_instance *) he->value;
+	  if (!mi->halt && mroute_addr_equal (&maddr, &mi->real))
+	    {
+	      multi_signal_instance (m, mi, SIGTERM);
+	      ++count;
+	    }
+	}
+      hash_iterator_free (&hi);
+    }
+  return count;
+}
+
+static void
+management_delete_event (void *arg, event_t event)
+{
+  struct multi_context *m = (struct multi_context *) arg;
+  if (m->mtcp)
+    multi_tcp_delete_event (m->mtcp, event);
+}
+
+#endif
+
+void
+init_management_callback_multi (struct multi_context *m)
+{
+#ifdef ENABLE_MANAGEMENT
+  if (management)
+    {
+      struct management_callback cb;
+      CLEAR (cb);
+      cb.arg = m;
+      cb.status = management_callback_status;
+      cb.show_net = management_show_net_callback;
+      cb.kill_by_cn = management_callback_kill_by_cn;
+      cb.kill_by_addr = management_callback_kill_by_addr;
+      cb.delete_event = management_delete_event;
+      management_set_callback (management, &cb);
+    }
+#endif
+}
+
+void
+uninit_management_callback_multi (struct multi_context *m)
+{
+  uninit_management_callback ();
+}
+
+/*
+ * Top level event loop.
+ */
+void
+tunnel_server (struct context *top)
+{
+  ASSERT (top->options.mode == MODE_SERVER);
+
+  switch (top->options.proto) {
+  case PROTO_UDPv4:
+    tunnel_server_udp (top);
+    break;
+  case PROTO_TCPv4_SERVER:
+    tunnel_server_tcp (top);
+    break;
+  default:
+    ASSERT (0);
+  }
+}
+
+#else
+static void dummy(void) {}
+#endif /* P2MP_SERVER */