Clean up multi-thread barrier-sync hold-down timer

Main thread: don't bother with the barrier sync hold-down timer if
none of the worker threads are busy.

Worker threads: avoid epoll_pwait (10ms timeout) when the
control-plane has been active in the last half-second.

Cherry-pick a recent dangling reference fix: pool_elt_at_index after
e.g. rx callback is required, in case the unix file pool expands.

DO NOT cherry-pick the patch into master!

Change-Id: I130dc72fd95c8a22716fa2f0f98ce35cdccc52d7
Signed-off-by: Dave Barach <dave@barachs.net>

1 files changed, 31 insertions, 5 deletions

diff --git a/src/vlib/unix/input.c b/src/vlib/unix/input.c
index 370e5dddaf6..76337f65865 100644
--- a/src/vlib/unix/input.c
+++ b/src/vlib/unix/input.c
@@ -145,9 +145,13 @@ linux_epoll_input_inline (vlib_main_t * vm, vlib_node_runtime_t * node,
     vlib_node_main_t *nm = &vm->node_main;
     u32 ticks_until_expiration;
     f64 timeout;
+    f64 now;
     int timeout_ms = 0, max_timeout_ms = 10;
     f64 vector_rate = vlib_last_vectors_per_main_loop (vm);
 
+    if (is_main == 0)
+      now = vlib_time_now (vm);
+
     /*
      * If we've been asked for a fixed-sleep between main loop polls,
      * do so right away.
@@ -194,8 +198,9 @@ linux_epoll_input_inline (vlib_main_t * vm, vlib_node_runtime_t * node,
 	  }
 	node->input_main_loops_per_call = 0;
       }
-    else if (is_main == 0 && vector_rate < 2 &&
-	     nm->input_node_counts_by_state[VLIB_NODE_STATE_POLLING] == 0)
+    else if (is_main == 0 && vector_rate < 2
+	     && (vlib_global_main.time_last_barrier_release + 0.5 < now)
+	     && nm->input_node_counts_by_state[VLIB_NODE_STATE_POLLING] == 0)
       {
 	timeout = 10e-3;
 	timeout_ms = max_timeout_ms;
@@ -223,12 +228,32 @@ linux_epoll_input_inline (vlib_main_t * vm, vlib_node_runtime_t * node,
 				      em->epoll_events,
 				      vec_len (em->epoll_events), timeout_ms);
 	  }
+
       }
     else
       {
+	/*
+	 * Worker thread, no epoll fd's, sleep for 100us at a time
+	 * and check for a barrier sync request
+	 */
 	if (timeout_ms)
-	  usleep (timeout_ms * 1000);
-	return 0;
+	  {
+	    struct timespec ts, tsrem;
+	    f64 limit = now + (f64) timeout_ms * 1e-3;
+
+	    while (vlib_time_now (vm) < limit)
+	      {
+		/* Sleep for 100us at a time */
+		ts.tv_sec = 0;
+		ts.tv_nsec = 1000 * 100;
+
+		while (nanosleep (&ts, &tsrem) < 0)
+		  ts = tsrem;
+		if (*vlib_worker_threads->wait_at_barrier)
+		  goto done;
+	      }
+	  }
+	goto done;
       }
   }
 
@@ -238,7 +263,7 @@ linux_epoll_input_inline (vlib_main_t * vm, vlib_node_runtime_t * node,
 	vlib_panic_with_error (vm, clib_error_return_unix (0, "epoll_wait"));
 
       /* non fatal error (e.g. EINTR). */
-      return 0;
+      goto done;
     }
 
   em->epoll_waits += 1;
@@ -318,6 +343,7 @@ linux_epoll_input_inline (vlib_main_t * vm, vlib_node_runtime_t * node,
 	}
     }
 
+done:
   return 0;
 }