[PATCH v4 11/20] net/txgbe: add RX desc merge mode for Amber-Lite NICs

[Zaiyu Wang]

Add RX desc merge mode for Amber-Lite NICs. When enabled,
the hardware batch-processes RX packets, significantly
enhancing performance. This feature is enabled by default
in the driver and can be configured via the rx_desc_merge
parameter in devargs.

Signed-off-by: Zaiyu Wang <zaiyuwang at trustnetic.com>
---
 drivers/net/txgbe/base/txgbe_regs.h |  4 ++++
 drivers/net/txgbe/base/txgbe_type.h |  3 +++
 drivers/net/txgbe/txgbe_ethdev.c    |  7 ++++++-
 drivers/net/txgbe/txgbe_rxtx.c      | 11 +++++++++++
 4 files changed, 24 insertions(+), 1 deletion(-)

diff --git a/drivers/net/txgbe/base/txgbe_regs.h b/drivers/net/txgbe/base/txgbe_regs.h
index 1a544bcd57..e050941992 100644
--- a/drivers/net/txgbe/base/txgbe_regs.h
+++ b/drivers/net/txgbe/base/txgbe_regs.h
@@ -1377,6 +1377,7 @@ enum txgbe_5tuple_protocol {
 #define   TXGBE_RXCFG_HDRLEN(v)         LS(HDRLEN(v), 12, 0xF)
 #define     TXGBE_RXCFG_HDRLEN_MASK     MS(12, 0xF)
 #define   TXGBE_RXCFG_WTHRESH(v)        LS(v, 16, 0x7)
+#define   TXGBE_RXCFG_DESC_MERGE        MS(19, 0x1)
 #define   TXGBE_RXCFG_ETAG              MS(22, 0x1)
 #define   TXGBE_RXCFG_RSCMAX_MASK       MS(23, 0x3)
 #define     TXGBE_RXCFG_RSCMAX_1        LS(0, 23, 0x3)
@@ -1671,6 +1672,9 @@ enum txgbe_5tuple_protocol {
 #define   TXGBE_RPUP2TC_UP_SHIFT        3
 #define   TXGBE_RPUP2TC_UP_MASK         0x7
 
+#define TXGBE_RDM_DCACHE_CTL             0x0120A8
+#define   TXGBE_RDM_DCACHE_CTL_EN        MS(0, 0x1)
+
 /* mac switcher */
 #define TXGBE_ETHADDRL                  0x016200
 #define   TXGBE_ETHADDRL_AD0(v)         LS(v, 0, 0xFF)
diff --git a/drivers/net/txgbe/base/txgbe_type.h b/drivers/net/txgbe/base/txgbe_type.h
index 844a2827bc..03e5bd489d 100644
--- a/drivers/net/txgbe/base/txgbe_type.h
+++ b/drivers/net/txgbe/base/txgbe_type.h
@@ -718,6 +718,7 @@ struct txgbe_phy_info {
 #define TXGBE_DEVARG_FFE_POST		"ffe_post"
 #define TXGBE_DEVARG_TX_HEAD_WB		"tx_headwb"
 #define TXGBE_DEVARG_TX_HEAD_WB_SIZE	"tx_headwb_size"
+#define TXGBE_DEVARG_RX_DESC_MERGE	"rx_desc_merge"
 
 static const char * const txgbe_valid_arguments[] = {
 	TXGBE_DEVARG_BP_AUTO,
@@ -730,6 +731,7 @@ static const char * const txgbe_valid_arguments[] = {
 	TXGBE_DEVARG_FFE_POST,
 	TXGBE_DEVARG_TX_HEAD_WB,
 	TXGBE_DEVARG_TX_HEAD_WB_SIZE,
+	TXGBE_DEVARG_RX_DESC_MERGE,
 	NULL
 };
 
@@ -782,6 +784,7 @@ struct txgbe_devargs {
 	u16 sgmii;
 	u16 tx_headwb;
 	u16 tx_headwb_size;
+	u16 rx_desc_merge;
 };
 
 struct txgbe_hw {
diff --git a/drivers/net/txgbe/txgbe_ethdev.c b/drivers/net/txgbe/txgbe_ethdev.c
index 9c14e4b8ed..13d1d6924d 100644
--- a/drivers/net/txgbe/txgbe_ethdev.c
+++ b/drivers/net/txgbe/txgbe_ethdev.c
@@ -516,6 +516,7 @@ txgbe_parse_devargs(struct txgbe_hw *hw, struct rte_devargs *devargs)
 	/* New devargs for amberlite config */
 	u16 tx_headwb = 1;
 	u16 tx_headwb_size = 16;
+	u16 rx_desc_merge = 1;
 
 	if (devargs == NULL)
 		goto null;
@@ -544,6 +545,8 @@ txgbe_parse_devargs(struct txgbe_hw *hw, struct rte_devargs *devargs)
 			   &txgbe_handle_devarg, &tx_headwb);
 	rte_kvargs_process(kvlist, TXGBE_DEVARG_TX_HEAD_WB_SIZE,
 			   &txgbe_handle_devarg, &tx_headwb_size);
+	rte_kvargs_process(kvlist, TXGBE_DEVARG_RX_DESC_MERGE,
+			   &txgbe_handle_devarg, &rx_desc_merge);
 	rte_kvargs_free(kvlist);
 
 null:
@@ -553,6 +556,7 @@ txgbe_parse_devargs(struct txgbe_hw *hw, struct rte_devargs *devargs)
 	hw->devarg.sgmii = sgmii;
 	hw->devarg.tx_headwb = tx_headwb;
 	hw->devarg.tx_headwb_size = tx_headwb_size;
+	hw->devarg.rx_desc_merge = rx_desc_merge;
 	hw->phy.ffe_set = ffe_set;
 	hw->phy.ffe_main = ffe_main;
 	hw->phy.ffe_pre = ffe_pre;
@@ -5729,7 +5733,8 @@ RTE_PMD_REGISTER_PARAM_STRING(net_txgbe,
 			      TXGBE_DEVARG_FFE_PRE "=<uint16>"
 			      TXGBE_DEVARG_FFE_POST "=<uint16>"
 			      TXGBE_DEVARG_TX_HEAD_WB "=<0|1>"
-			      TXGBE_DEVARG_TX_HEAD_WB_SIZE "=<1|16>");
+			      TXGBE_DEVARG_TX_HEAD_WB_SIZE "=<1|16>"
+			      TXGBE_DEVARG_RX_DESC_MERGE "=<0|1>");
 
 RTE_LOG_REGISTER_SUFFIX(txgbe_logtype_init, init, NOTICE);
 RTE_LOG_REGISTER_SUFFIX(txgbe_logtype_driver, driver, NOTICE);
diff --git a/drivers/net/txgbe/txgbe_rxtx.c b/drivers/net/txgbe/txgbe_rxtx.c
index b15b6788e6..957aa5c6eb 100644
--- a/drivers/net/txgbe/txgbe_rxtx.c
+++ b/drivers/net/txgbe/txgbe_rxtx.c
@@ -4667,6 +4667,17 @@ txgbe_dev_rx_init(struct rte_eth_dev *dev)
 		buf_size = ROUND_DOWN(buf_size, 0x1 << 10);
 		srrctl |= TXGBE_RXCFG_PKTLEN(buf_size);
 
+		if ((hw->mac.type == txgbe_mac_aml ||
+		     hw->mac.type == txgbe_mac_aml40) && hw->devarg.rx_desc_merge == 1) {
+			srrctl |= TXGBE_RXCFG_DESC_MERGE;
+
+			wr32(hw, TXGBE_RDM_DCACHE_CTL, TXGBE_RDM_DCACHE_CTL_EN);
+			wr32m(hw, TXGBE_RDM_RSC_CTL, TXGBE_RDM_RSC_CTL_FREE_CTL,
+							 TXGBE_RDM_RSC_CTL_FREE_CTL);
+			wr32m(hw, TXGBE_RDM_RSC_CTL, TXGBE_RDM_RSC_CTL_FREE_CNT_DIS,
+							 ~TXGBE_RDM_RSC_CTL_FREE_CNT_DIS);
+		}
+
 		wr32(hw, TXGBE_RXCFG(rxq->reg_idx), srrctl);
 
 		/* It adds dual VLAN length for supporting dual VLAN */
-- 
2.21.0.windows.1