Overall tcp performance improvements (VPP-846)

- limit minimum rto per connection
- cleanup sack scoreboard
- switched svm fifo out-of-order data handling from absolute offsets to
  relative offsets.
- improve cwnd handling when using sacks
- add cc event debug stats
- improved uri tcp test client/server:  bugfixes and added half-duplex mode
- expanded builtin client/server
- updated uri socket client/server code to work in half-duplex
- ensure session node unsets fifo event for empty fifo
- fix session detach

Change-Id: Ia446972340e32a65e0694ee2844355167d0c170d
Signed-off-by: Florin Coras <fcoras@cisco.com>

1 files changed, 21 insertions, 5 deletions

diff --git a/src/svm/svm_fifo.h b/src/svm/svm_fifo.h
index 9cb93ff4..f32ef41d 100644
--- a/src/svm/svm_fifo.h
+++ b/src/svm/svm_fifo.h
@@ -127,21 +127,37 @@ format_function_t format_svm_fifo;
 always_inline ooo_segment_t *
 svm_fifo_newest_ooo_segment (svm_fifo_t * f)
 {
-  return f->ooo_segments + f->ooos_newest;
+  if (f->ooos_newest == OOO_SEGMENT_INVALID_INDEX)
+    return 0;
+  return pool_elt_at_index (f->ooo_segments, f->ooos_newest);
+}
+
+always_inline u32
+ooo_segment_distance_to_tail (svm_fifo_t * f, u32 a)
+{
+  /* Ambiguous. Assumption is that ooo segments don't touch tail */
+  if (a == f->tail && f->tail == f->head)
+    return f->nitems;
+
+  return ((f->nitems + a - f->tail) % f->nitems);
 }
 
 always_inline u32
 ooo_segment_offset (svm_fifo_t * f, ooo_segment_t * s)
 {
-//  return ((f->nitems + s->fifo_position - f->tail) % f->nitems);
-  return s->start;
+  return ooo_segment_distance_to_tail (f, s->start);
 }
 
 always_inline u32
 ooo_segment_end_offset (svm_fifo_t * f, ooo_segment_t * s)
 {
-//  return ((f->nitems + s->fifo_position + s->length - f->tail) % f->nitems);
-  return s->start + s->length;
+  return ooo_segment_distance_to_tail (f, s->start) + s->length;
+}
+
+always_inline u32
+ooo_segment_length (svm_fifo_t * f, ooo_segment_t * s)
+{
+  return s->length;
 }
 
 always_inline ooo_segment_t *