rtpjitterbuffer: fix timer-reuse bug

When doing rtx, the jitterbuffer will always add an rtx-timer for the next
sequence number.

In the case of the packet corresponding to that sequence number arriving,
that same timer will be reused, and simply moved on to wait for the
following sequence number etc.

Once an rtx-timer expires (after all retries), it will be rescheduled as
a lost-timer instead for the same sequence number.

Now, if this particular sequence-number now arrives (after the timer has
become a lost-timer), the reuse mechanism *should* now set a new
rtx-timer for the next sequence number, but the bug is that it does
not change the timer-type, and hence schedules a lost-timer for that
following sequence number, with the result that you will have a very
early lost-event for a packet that might still arrive, and you will
never be able to send any rtx for this packet.

Found by Erlend Graff - erlend@pexip.com

https://bugzilla.gnome.org/show_bug.cgi?id=773891

diff --git a/gst/rtpmanager/gstrtpjitterbuffer.c b/gst/rtpmanager/gstrtpjitterbuffer.c
index c85400c..a6e0b3a 100644 (file)
--- a/gst/rtpmanager/gstrtpjitterbuffer.c
+++ b/gst/rtpmanager/gstrtpjitterbuffer.c
@@ -2372,6 +2372,7 @@ update_timers (GstRtpJitterBuffer * jitterbuffer, guint16 seqnum,
         GST_TIME_ARGS (priv->packet_spacing), GST_TIME_ARGS (priv->avg_jitter));
 
     if (timer) {
+      timer->type = TIMER_TYPE_EXPECTED;
       reschedule_timer (jitterbuffer, timer, priv->next_in_seqnum, expected,
           delay, TRUE);
     } else {

diff --git a/tests/check/elements/rtpjitterbuffer.c b/tests/check/elements/rtpjitterbuffer.c
index 66b570c..095fb16 100644 (file)
--- a/tests/check/elements/rtpjitterbuffer.c
+++ b/tests/check/elements/rtpjitterbuffer.c
@@ -1683,6 +1683,11 @@ GST_START_TEST (test_rtx_original_buffer_does_not_update_rtx_stats)
   fail_unless_equals_int (6, get_rtp_seq_num (buffer));
   gst_buffer_unref (buffer);
 
+  /* due to the advance in time, we will now also have sent
+     an rtx-request for 7 */
+  verify_rtx_event (gst_harness_pull_upstream_event (h),
+      7, 7 * PCMU_BUF_DURATION, 10, PCMU_BUF_DURATION);
+
   /* The original buffer does not count in the RTX stats. */
   fail_unless (verify_jb_stats (h->element,
           gst_structure_new ("application/x-rtp-jitterbuffer-stats",
@@ -1690,7 +1695,7 @@ GST_START_TEST (test_rtx_original_buffer_does_not_update_rtx_stats)
               "num-lost", G_TYPE_UINT64, (guint64) 0,
               "num-late", G_TYPE_UINT64, (guint64) 0,
               "num-duplicates", G_TYPE_UINT64, (guint64) 0,
-              "rtx-count", G_TYPE_UINT64, (guint64) 1,
+              "rtx-count", G_TYPE_UINT64, (guint64) 2,
               "rtx-success-count", G_TYPE_UINT64, (guint64) 0,
               "rtx-per-packet", G_TYPE_DOUBLE, 0.0,
               "rtx-rtt", G_TYPE_UINT64, (guint64) 0, NULL)));
@@ -1706,7 +1711,7 @@ GST_START_TEST (test_rtx_original_buffer_does_not_update_rtx_stats)
               "num-lost", G_TYPE_UINT64, (guint64) 0,
               "num-late", G_TYPE_UINT64, (guint64) 0,
               "num-duplicates", G_TYPE_UINT64, (guint64) 1,
-              "rtx-count", G_TYPE_UINT64, (guint64) 1,
+              "rtx-count", G_TYPE_UINT64, (guint64) 2,
               "rtx-success-count", G_TYPE_UINT64, (guint64) 0,
               "rtx-per-packet", G_TYPE_DOUBLE, 1.0,
               "rtx-rtt", G_TYPE_UINT64, (guint64) (now - last_rtx_request),
@@ -2252,6 +2257,58 @@ arrival(dts):  3     5     5
 
 GST_END_TEST;
 
+GST_START_TEST (test_rtx_timer_reuse)
+{
+  GstHarness *h = gst_harness_new ("rtpjitterbuffer");
+  GstTestClock *testclock;
+  gint latency_ms = 5 * PCMU_BUF_MS;
+  gint num_init_buffers = latency_ms / PCMU_BUF_MS + 1;
+
+  testclock = gst_harness_get_testclock (h);
+  gst_harness_set_src_caps (h, generate_caps ());
+  g_object_set (h->element, "do-retransmission", TRUE, "latency", latency_ms,
+      "do-lost", TRUE, "rtx-max-retries", 1, NULL);
+
+  /* Push/pull buffers and advance time past buffer 0's timeout (in order to
+   * simplify the test) */
+  for (gint i = 0; i < num_init_buffers; i++) {
+    gst_test_clock_set_time (testclock, i * PCMU_BUF_DURATION);
+    fail_unless_equals_int (GST_FLOW_OK,
+        gst_harness_push (h, generate_test_buffer (i)));
+    gst_harness_wait_for_clock_id_waits (h, 1, 60);
+  }
+
+  gst_harness_crank_single_clock_wait (h);
+  fail_unless_equals_int64 (latency_ms * GST_MSECOND,
+      gst_clock_get_time (GST_CLOCK (testclock)));
+
+  for (gint i = 0; i < num_init_buffers; i++)
+    gst_buffer_unref (gst_harness_pull (h));
+
+  /* drop reconfigure event */
+  gst_event_unref (gst_harness_pull_upstream_event (h));
+
+  /* crank to timeout the only rtx-request, and the timer will
+   * now reschedule as a lost-timer internally */
+  gst_harness_crank_single_clock_wait (h);
+  verify_rtx_event (gst_harness_pull_upstream_event (h),
+      6, 6 * PCMU_BUF_DURATION, 10, PCMU_BUF_DURATION);
+
+  /* but now buffer 6 arrives, and this should now reuse the lost-timer
+   * for 6, as an expected-timer for 7 */
+  fail_unless_equals_int (GST_FLOW_OK,
+      gst_harness_push (h, generate_test_buffer (6)));
+
+  /* now crank to timeout the expected-timer for 7 and verify */
+  gst_harness_crank_single_clock_wait (h);
+  verify_rtx_event (gst_harness_pull_upstream_event (h),
+      7, 7 * PCMU_BUF_DURATION, 10, PCMU_BUF_DURATION);
+
+  gst_object_unref (testclock);
+  gst_harness_teardown (h);
+}
+
+GST_END_TEST;
 
 GST_START_TEST (test_gap_exceeds_latency)
 {
@@ -2701,6 +2758,7 @@ rtpjitterbuffer_suite (void)
   tcase_add_test (tc_chain, test_rtx_no_request_if_time_past_retry_period);
   tcase_add_test (tc_chain, test_rtx_same_delay_and_retry_timeout);
   tcase_add_test (tc_chain, test_rtx_with_backwards_rtptime);
+  tcase_add_test (tc_chain, test_rtx_timer_reuse);
 
   tcase_add_test (tc_chain, test_gap_exceeds_latency);
   tcase_add_test (tc_chain, test_deadline_ts_offset);