[PATCH v3 3/3] net/mlx5: add wait on time support in Tx datapath

[Viacheslav Ovsiienko <viacheslavo@nvidia.com>]

The hardware since ConnectX-7 supports waiting on
specified moment of time with new introduced wait
descriptor. A timestamp can be directly placed
into descriptor and pushed to sending queue.
Once hardware encounter the wait descriptor the
queue operation is suspended till specified moment
of time. This patch update the Tx datapath to handle
this new hardware wait capability.

PMD documentation and release notes updated accordingly.

Signed-off-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
---
 doc/guides/nics/mlx5.rst               |  5 ++
 doc/guides/rel_notes/release_22_03.rst |  6 +++
 drivers/net/mlx5/mlx5_tx.h             | 72 +++++++++++++++++++++++---
 3 files changed, 75 insertions(+), 8 deletions(-)

diff --git a/doc/guides/nics/mlx5.rst b/doc/guides/nics/mlx5.rst
index 968aac01d2..8956cd1dd8 100644
--- a/doc/guides/nics/mlx5.rst
+++ b/doc/guides/nics/mlx5.rst
@@ -829,6 +829,11 @@ for an additional list of options shared with other mlx5 drivers.
   By default (if the ``tx_pp`` is not specified) send scheduling on timestamps
   feature is disabled.
 
+  Starting since ConnectX-7 the capability to schedule traffic directly
+  on timestamp specified in descriptor is provided, no extra objects are
+  needed anymore and scheduling capability is advertised and handled
+  regardless tx_pp parameter presence.
+
 - ``tx_skew`` parameter [int]
 
   The parameter adjusts the send packet scheduling on timestamps and represents
diff --git a/doc/guides/rel_notes/release_22_03.rst b/doc/guides/rel_notes/release_22_03.rst
index 74965ebd56..acd56e0a80 100644
--- a/doc/guides/rel_notes/release_22_03.rst
+++ b/doc/guides/rel_notes/release_22_03.rst
@@ -118,6 +118,12 @@ New Features
   * Added PPPoL2TPv2oUDP FDIR distribute packets based on inner IP
     src/dst address and UDP/TCP src/dst port.
 
+* **Updated Mellanox mlx5 driver.**
+
+  Updated the Mellanox mlx5 driver with new features and improvements, including:
+
+  * Support ConnectX-7 capability to schedule traffic sending on timestamp
+
 * **Updated Wangxun ngbe driver.**
 
   * Added support for devices of custom PHY interfaces.
diff --git a/drivers/net/mlx5/mlx5_tx.h b/drivers/net/mlx5/mlx5_tx.h
index b50deb8b67..0adc3f4839 100644
--- a/drivers/net/mlx5/mlx5_tx.h
+++ b/drivers/net/mlx5/mlx5_tx.h
@@ -780,7 +780,7 @@ mlx5_tx_cseg_init(struct mlx5_txq_data *__rte_restrict txq,
  *   compile time and may be used for optimization.
  */
 static __rte_always_inline void
-mlx5_tx_wseg_init(struct mlx5_txq_data *restrict txq,
+mlx5_tx_qseg_init(struct mlx5_txq_data *restrict txq,
 		  struct mlx5_txq_local *restrict loc __rte_unused,
 		  struct mlx5_wqe *restrict wqe,
 		  unsigned int wci,
@@ -795,6 +795,43 @@ mlx5_tx_wseg_init(struct mlx5_txq_data *restrict txq,
 	qs->reserved1 = RTE_BE32(0);
 }
 
+/**
+ * Build the Wait on Time Segment with specified timestamp value.
+ *
+ * @param txq
+ *   Pointer to TX queue structure.
+ * @param loc
+ *   Pointer to burst routine local context.
+ * @param wqe
+ *   Pointer to WQE to fill with built Control Segment.
+ * @param ts
+ *   Timesatmp value to wait.
+ * @param olx
+ *   Configured Tx offloads mask. It is fully defined at
+ *   compile time and may be used for optimization.
+ */
+static __rte_always_inline void
+mlx5_tx_wseg_init(struct mlx5_txq_data *restrict txq,
+		  struct mlx5_txq_local *restrict loc __rte_unused,
+		  struct mlx5_wqe *restrict wqe,
+		  uint64_t ts,
+		  unsigned int olx __rte_unused)
+{
+	struct mlx5_wqe_wseg *ws;
+
+	ws = RTE_PTR_ADD(wqe, MLX5_WSEG_SIZE);
+	ws->operation = rte_cpu_to_be_32(MLX5_WAIT_COND_CYCLIC_BIGGER);
+	ws->lkey = RTE_BE32(0);
+	ws->va_high = RTE_BE32(0);
+	ws->va_low = RTE_BE32(0);
+	if (txq->rt_timestamp) {
+		ts = ts % (uint64_t)NS_PER_S
+		   | (ts / (uint64_t)NS_PER_S) << 32;
+	}
+	ws->value = rte_cpu_to_be_64(ts);
+	ws->mask = txq->rt_timemask;
+}
+
 /**
  * Build the Ethernet Segment without inlined data.
  * Supports Software Parser, Checksums and VLAN insertion Tx offload features.
@@ -1626,9 +1663,9 @@ mlx5_tx_schedule_send(struct mlx5_txq_data *restrict txq,
 {
 	if (MLX5_TXOFF_CONFIG(TXPP) &&
 	    loc->mbuf->ol_flags & txq->ts_mask) {
+		struct mlx5_dev_ctx_shared *sh;
 		struct mlx5_wqe *wqe;
 		uint64_t ts;
-		int32_t wci;
 
 		/*
 		 * Estimate the required space quickly and roughly.
@@ -1640,13 +1677,32 @@ mlx5_tx_schedule_send(struct mlx5_txq_data *restrict txq,
 			return MLX5_TXCMP_CODE_EXIT;
 		/* Convert the timestamp into completion to wait. */
 		ts = *RTE_MBUF_DYNFIELD(loc->mbuf, txq->ts_offset, uint64_t *);
-		wci = mlx5_txpp_convert_tx_ts(txq->sh, ts);
-		if (unlikely(wci < 0))
-			return MLX5_TXCMP_CODE_SINGLE;
-		/* Build the WAIT WQE with specified completion. */
 		wqe = txq->wqes + (txq->wqe_ci & txq->wqe_m);
-		mlx5_tx_cseg_init(txq, loc, wqe, 2, MLX5_OPCODE_WAIT, olx);
-		mlx5_tx_wseg_init(txq, loc, wqe, wci, olx);
+		sh = txq->sh;
+		if (txq->wait_on_time) {
+			/* The wait on time capability should be used. */
+			ts -= sh->txpp.skew;
+			mlx5_tx_cseg_init(txq, loc, wqe,
+					  1 + sizeof(struct mlx5_wqe_wseg) /
+					      MLX5_WSEG_SIZE,
+					  MLX5_OPCODE_WAIT |
+					  MLX5_OPC_MOD_WAIT_TIME << 24, olx);
+			mlx5_tx_wseg_init(txq, loc, wqe, ts, olx);
+		} else {
+			/* Legacy cross-channel operation should be used. */
+			int32_t wci;
+
+			wci = mlx5_txpp_convert_tx_ts(sh, ts);
+			if (unlikely(wci < 0))
+				return MLX5_TXCMP_CODE_SINGLE;
+			/* Build the WAIT WQE with specified completion. */
+			mlx5_tx_cseg_init(txq, loc, wqe,
+					  1 + sizeof(struct mlx5_wqe_qseg) /
+					      MLX5_WSEG_SIZE,
+					  MLX5_OPCODE_WAIT |
+					  MLX5_OPC_MOD_WAIT_CQ_PI << 24, olx);
+			mlx5_tx_qseg_init(txq, loc, wqe, wci, olx);
+		}
 		++txq->wqe_ci;
 		--loc->wqe_free;
 		return MLX5_TXCMP_CODE_MULTI;
-- 
2.18.1