p2p: fix frequent weak_ptr exception on connection

When a handshake fails, it can fail due to timeout or destroyed
connection, in which case the connection will be, or already is,
closed, and we don't want to do it twice.
Additionally, when closing a connection directly from the top
level code, ensure the connection is gone from the m_connects
list so it won't be used again.

AFAICT this is now clean in netstat, /proc/PID/fd and print_cn.

This fixes a noisy (but harmless) exception.

2 files changed, 12 insertions, 5 deletions

diff --git a/contrib/epee/include/net/levin_protocol_handler_async.h b/contrib/epee/include/net/levin_protocol_handler_async.h
index 5774c0ba7..1341a4ae6 100644
--- a/contrib/epee/include/net/levin_protocol_handler_async.h
+++ b/contrib/epee/include/net/levin_protocol_handler_async.h
@@ -949,7 +949,12 @@ bool async_protocol_handler_config<t_connection_context>::close(boost::uuids::uu
 {
   CRITICAL_REGION_LOCAL(m_connects_lock);
   async_protocol_handler<t_connection_context>* aph = find_connection(connection_id);
-  return 0 != aph ? aph->close() : false;
+  if (!aph)
+    return false;
+  if (!aph->close())
+    return false;
+  m_connects.erase(connection_id);
+  return true;
 }
 //------------------------------------------------------------------------------------------
 template<class t_connection_context>
diff --git a/src/p2p/net_node.inl b/src/p2p/net_node.inl
index 9b6358dee..dcd16e5c1 100644
--- a/src/p2p/net_node.inl
+++ b/src/p2p/net_node.inl
@@ -1013,15 +1013,18 @@ namespace nodetool
 
     epee::simple_event ev;
     std::atomic<bool> hsh_result(false);
+    bool timeout = false;
 
     bool r = epee::net_utils::async_invoke_remote_command2<typename COMMAND_HANDSHAKE::response>(context_, COMMAND_HANDSHAKE::ID, arg, zone.m_net_server.get_config_object(),
-      [this, &pi, &ev, &hsh_result, &just_take_peerlist, &context_](int code, const typename COMMAND_HANDSHAKE::response& rsp, p2p_connection_context& context)
+      [this, &pi, &ev, &hsh_result, &just_take_peerlist, &context_, &timeout](int code, const typename COMMAND_HANDSHAKE::response& rsp, p2p_connection_context& context)
     {
       epee::misc_utils::auto_scope_leave_caller scope_exit_handler = epee::misc_utils::create_scope_leave_handler([&](){ev.raise();});
 
       if(code < 0)
       {
         LOG_WARNING_CC(context, "COMMAND_HANDSHAKE invoke failed. (" << code <<  ", " << epee::levin::get_err_descr(code) << ")");
+        if (code == LEVIN_ERROR_CONNECTION_TIMEDOUT || code == LEVIN_ERROR_CONNECTION_DESTROYED)
+          timeout = true;
         return;
       }
 
@@ -1079,7 +1082,8 @@ namespace nodetool
     if(!hsh_result)
     {
       LOG_WARNING_CC(context_, "COMMAND_HANDSHAKE Failed");
-      m_network_zones.at(context_.m_remote_address.get_zone()).m_net_server.get_config_object().close(context_.m_connection_id);
+      if (!timeout)
+        zone.m_net_server.get_config_object().close(context_.m_connection_id);
     }
     else if (!just_take_peerlist)
     {
@@ -1266,7 +1270,6 @@ namespace nodetool
       LOG_PRINT_CC_PRIORITY_NODE(is_priority, *con, "Failed to HANDSHAKE with peer "
         << na.str()
         /*<< ", try " << try_count*/);
-      zone.m_net_server.get_config_object().close(con->m_connection_id);
       record_addr_failed(na);
       return false;
     }
@@ -1330,7 +1333,6 @@ namespace nodetool
       bool is_priority = is_priority_node(na);
 
       LOG_PRINT_CC_PRIORITY_NODE(is_priority, *con, "Failed to HANDSHAKE with peer " << na.str());
-      zone.m_net_server.get_config_object().close(con->m_connection_id);
       record_addr_failed(na);
       return false;
     }