mainloop: Reactivate the linked connection event with a non empty list

Linked connections aren't woken up by libevent due to I/O but rather
artificially so we can, by chunks, empty the spooled object(s).

Commit 5719dfb48f87a54aeb5982ff03345303bc058ebb (in 0.3.4.1-alpha) made it
that the schedule_active_linked_connections_event would be only called once at
startup but this is wrong because then we would never go through again the
active linked connections.

Fortunately, everytime a new linked connection is created, the event is
activated and thus we would go through the active list again. On a busy relay,
this issue is mitigated by that but on a slower relays or bridge, a connection
could get stuck for a while until a new directory information request would
show up.

Fixes #28717, #28912

diff --git a/changes/ticket28912 b/changes/ticket28912
new file mode 100644 (file)
index 0000000..4119b77
--- /dev/null
+++ b/changes/ticket28912
@@ -0,0 +1,6 @@
+  o Major bugfixes (relay, directory):
+    - A connection serving directory information wouldn't get reactivated after
+      the first chunk of data was sent (usually 32KB). Tor now always activate
+      the main loop event that goes through these connections as long as at
+      least one connection is still active. Fixes bug 28912; bugfix on
+      0.3.4.1-alpha. Patch by "cypherpunks3".

diff --git a/src/or/main.c b/src/or/main.c
index bc01e07c3d1ac883d2aa1a42bcc23bf72eabe4bf..aaa31c0579f854d22ab040c561bd8865f5e7bdc4 100644 (file)
--- a/src/or/main.c
+++ b/src/or/main.c
@@ -404,6 +404,9 @@ connection_unlink(connection_t *conn)
   connection_free(conn);
 }
 
+/** Event that invokes schedule_active_linked_connections_cb. */
+static mainloop_event_t *schedule_active_linked_connections_event = NULL;
+
 /**
  * Callback: used to activate read events for all linked connections, so
  * libevent knows to call their read callbacks.  This callback run as a
@@ -420,10 +423,18 @@ schedule_active_linked_connections_cb(mainloop_event_t *event, void *arg)
    * so that libevent knows to run their callbacks. */
   SMARTLIST_FOREACH(active_linked_connection_lst, connection_t *, conn,
                     event_active(conn->read_event, EV_READ, 1));
-}
 
-/** Event that invokes schedule_active_linked_connections_cb. */
-static mainloop_event_t *schedule_active_linked_connections_event = NULL;
+  /* Reactivate the event if we still have connections in the active list.
+   *
+   * A linked connection doesn't get woken up by I/O but rather artificially
+   * by this event callback. It has directory data spooled in it and it is
+   * sent incrementally by small chunks unless spool_eagerly is true. For that
+   * to happen, we need to induce the activation of the read event so it can
+   * be flushed. */
+  if (smartlist_len(active_linked_connection_lst)) {
+    mainloop_event_activate(schedule_active_linked_connections_event);
+  }
+}
 
 /** Initialize the global connection list, closeable connection list,
  * and active connection list. */