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>
diff --git a/src/vnet/session/session.c b/src/vnet/session/session.c
index 90248b6..58ec1c0 100644
--- a/src/vnet/session/session.c
+++ b/src/vnet/session/session.c
@@ -411,7 +411,7 @@
{
int enqueued = 0, rv, in_order_off;
- ASSERT (svm_fifo_max_enqueue (s->rx_fifo)
+ ASSERT (svm_fifo_max_enqueue_prod (s->rx_fifo)
>= b->current_length + sizeof (*hdr));
svm_fifo_enqueue_nowait (s->rx_fifo, sizeof (session_dgram_hdr_t),
@@ -508,7 +508,7 @@
/* *INDENT-OFF* */
SESSION_EVT_DBG(SESSION_EVT_ENQ, s, ({
ed->data[0] = SESSION_IO_EVT_RX;
- ed->data[1] = svm_fifo_max_dequeue (s->rx_fifo);
+ ed->data[1] = svm_fifo_max_dequeue_prod (s->rx_fifo);
}));
/* *INDENT-ON* */
@@ -1112,7 +1112,7 @@
* point, either after sending everything or after a timeout, call delete
* notify. This will finally lead to the complete cleanup of the session.
*/
- if (svm_fifo_max_dequeue (s->tx_fifo))
+ if (svm_fifo_max_dequeue_cons (s->tx_fifo))
s->session_state = SESSION_STATE_CLOSED_WAITING;
else
s->session_state = SESSION_STATE_CLOSED;