svm_fifo rework to avoid contention on cursize

Problems Addressed:
- Contention of cursize by producer and consumer.
- Reduce the no of modulo operations.

Changes:
- Synchronization between producer and consumer changed from cursize
  to head and tail indexes
  Implications: reduces the usable size of fifo by 1.
- Using weaker memory ordering C++11 atomics to access head and tail
  based on producer and consumer role.
- Head and tail indexes are unsigned 32 bit integers. Additions and
  subtraction on them are implicit 32 bit Modulo operation.
- Adding weaker memory ordering variants of max_enq, max_deq, is_empty
  and is_full Using them appropriately in all places.

Perfomance improvement (iperf3 via Hoststack):

iperf3 Server: Marvell ThunderX2(AArch64) - iperf3 Client: Skylake(x86)
   ~6%(256 rxd/txd) - ~11%(2048 rxd/txd)

Change-Id: I1d484e000e437430fdd5a819657d1c6b62443018
Signed-off-by: Sirshak Das <sirshak.das@arm.com>
Reviewed-by: Honnappa Nagarahalli <honnappa.nagarahalli@arm.com>

1 files changed, 3 insertions, 2 deletions

diff --git a/src/vnet/udp/udp_input.c b/src/vnet/udp/udp_input.c
index db634cb2ddb..52c29d7ebfa 100644
--- a/src/vnet/udp/udp_input.c
+++ b/src/vnet/udp/udp_input.c
@@ -229,7 +229,7 @@ udp46_input_inline (vlib_main_t * vm, vlib_node_runtime_t * node,
 
 	  if (!uc0->is_connected)
 	    {
-	      if (svm_fifo_max_enqueue (s0->rx_fifo)
+	      if (svm_fifo_max_enqueue_prod (s0->rx_fifo)
 		  < b0->current_length + sizeof (session_dgram_hdr_t))
 		{
 		  error0 = UDP_ERROR_FIFO_FULL;
@@ -255,7 +255,8 @@ udp46_input_inline (vlib_main_t * vm, vlib_node_runtime_t * node,
 	    }
 	  else
 	    {
-	      if (svm_fifo_max_enqueue (s0->rx_fifo) < b0->current_length)
+	      if (svm_fifo_max_enqueue_prod (s0->rx_fifo) <
+		  b0->current_length)
 		{
 		  error0 = UDP_ERROR_FIFO_FULL;
 		  goto trace0;