driver: net: ldpaa: Use DPMAC as net device

As per current implementation of DPAA2 ethernet driver DPNI is used as
net device. DPNI is tangible objects can be multiple connected to same physical lane.

Use DPMAC as net device where it represents physical lane.
Below modification done in driver
 - Use global DPNI object
 - Connect DPMAC to DPNI
 - Create and destroy DPMAC

Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>

drivers/net/fsl-mc/mc.c

@@ -659,6 +659,12 @@ unsigned long mc_get_dram_block_size(void)
 
 int fsl_mc_ldpaa_init(bd_t *bis)
 {
+	int i;
+
+	for (i = WRIOP1_DPMAC1; i < NUM_WRIOP_PORTS; i++)
+		if ((wriop_is_enabled_dpmac(i) == 1) &&
+		    (wriop_get_phy_address(i) != -1))
+			ldpaa_eth_init(i, wriop_get_enet_if(i));
 	return 0;
 }
 

drivers/net/ldpaa_eth/ldpaa_eth.c

@@ -12,6 +12,7 @@
 #include <hwconfig.h>
 #include <phy.h>
 #include <linux/compat.h>
+#include <fsl-mc/fsl_dpmac.h>
 
 #include "ldpaa_eth.h"
 
@@ -46,7 +47,7 @@ static void ldpaa_eth_rx(struct ldpaa_eth_priv *priv,
 		/* Read the frame annotation status word and check for errors */
 		fas = (struct ldpaa_fas *)
 				((uint8_t *)(fd_addr) +
-				priv->buf_layout.private_data_size);
+				dflt_dpni->buf_layout.private_data_size);
 		status = le32_to_cpu(fas->status);
 		if (status & LDPAA_ETH_RX_ERR_MASK) {
 			printf("Rx frame error(s): 0x%08x\n",
@@ -220,11 +221,31 @@ static int ldpaa_eth_open(struct eth_device *net_dev, bd_t *bd)
 {
 	struct ldpaa_eth_priv *priv = (struct ldpaa_eth_priv *)net_dev->priv;
 	struct dpni_queue_attr rx_queue_attr;
+	struct dpmac_link_state	dpmac_link_state = { 0 };
 	int err;
 
 	if (net_dev->state == ETH_STATE_ACTIVE)
 		return 0;
 
+	if (get_mc_boot_status() != 0) {
+		printf("ERROR (MC is not booted)\n");
+		return -ENODEV;
+	}
+
+	if (get_dpl_apply_status() == 0) {
+		printf("ERROR (DPL is deployed. No device available)\n");
+		return -ENODEV;
+	}
+	/* DPMAC initialization */
+	err = ldpaa_dpmac_setup(priv);
+	if (err < 0)
+		goto err_dpmac_setup;
+
+	/* DPMAC binding DPNI */
+	err = ldpaa_dpmac_bind(priv);
+	if (err)
+		goto err_dpamc_bind;
+
 	/* DPNI initialization */
 	err = ldpaa_dpni_setup(priv);
 	if (err < 0)
@@ -237,10 +258,10 @@ static int ldpaa_eth_open(struct eth_device *net_dev, bd_t *bd)
 	/* DPNI binding DPBP */
 	err = ldpaa_dpni_bind(priv);
 	if (err)
-		goto err_bind;
+		goto err_dpni_bind;
 
 	err = dpni_add_mac_addr(dflt_mc_io, MC_CMD_NO_FLAGS,
-				priv->dpni_handle, net_dev->enetaddr);
+				dflt_dpni->dpni_handle, net_dev->enetaddr);
 	if (err) {
 		printf("dpni_add_mac_addr() failed\n");
 		return err;
@@ -259,15 +280,24 @@ static int ldpaa_eth_open(struct eth_device *net_dev, bd_t *bd)
 	priv->phydev->duplex = DUPLEX_FULL;
 #endif
 
-	err = dpni_enable(dflt_mc_io, MC_CMD_NO_FLAGS, priv->dpni_handle);
+	err = dpni_enable(dflt_mc_io, MC_CMD_NO_FLAGS, dflt_dpni->dpni_handle);
 	if (err < 0) {
 		printf("dpni_enable() failed\n");
 		return err;
 	}
 
+	dpmac_link_state.rate = SPEED_1000;
+	dpmac_link_state.options = DPMAC_LINK_OPT_AUTONEG;
+	dpmac_link_state.up = 1;
+	err = dpmac_set_link_state(dflt_mc_io, MC_CMD_NO_FLAGS,
+				  priv->dpmac_handle, &dpmac_link_state);
+	if (err < 0) {
+		printf("dpmac_set_link_state() failed\n");
+		return err;
+	}
 	/* TODO: support multiple Rx flows */
-	err = dpni_get_rx_flow(dflt_mc_io, MC_CMD_NO_FLAGS, priv->dpni_handle,
-			       0, 0, &rx_queue_attr);
+	err = dpni_get_rx_flow(dflt_mc_io, MC_CMD_NO_FLAGS,
+			       dflt_dpni->dpni_handle, 0, 0, &rx_queue_attr);
 	if (err) {
 		printf("dpni_get_rx_flow() failed\n");
 		goto err_rx_flow;
@@ -275,7 +305,7 @@ static int ldpaa_eth_open(struct eth_device *net_dev, bd_t *bd)
 
 	priv->rx_dflt_fqid = rx_queue_attr.fqid;
 
-	err = dpni_get_qdid(dflt_mc_io, MC_CMD_NO_FLAGS, priv->dpni_handle,
+	err = dpni_get_qdid(dflt_mc_io, MC_CMD_NO_FLAGS, dflt_dpni->dpni_handle,
 			    &priv->tx_qdid);
 	if (err) {
 		printf("dpni_get_qdid() failed\n");
@@ -289,12 +319,14 @@ static int ldpaa_eth_open(struct eth_device *net_dev, bd_t *bd)
 
 err_qdid:
 err_rx_flow:
-	dpni_disable(dflt_mc_io, MC_CMD_NO_FLAGS, priv->dpni_handle);
-err_bind:
+	dpni_disable(dflt_mc_io, MC_CMD_NO_FLAGS, dflt_dpni->dpni_handle);
+err_dpni_bind:
 	ldpaa_dpbp_free();
 err_dpbp_setup:
-	dpni_close(dflt_mc_io, MC_CMD_NO_FLAGS, priv->dpni_handle);
+err_dpamc_bind:
+	dpni_close(dflt_mc_io, MC_CMD_NO_FLAGS, dflt_dpni->dpni_handle);
 err_dpni_setup:
+err_dpmac_setup:
 	return err;
 }
 
@@ -306,8 +338,18 @@ static void ldpaa_eth_stop(struct eth_device *net_dev)
 	if ((net_dev->state == ETH_STATE_PASSIVE) ||
 	    (net_dev->state == ETH_STATE_INIT))
 		return;
+
+	err = dprc_disconnect(dflt_mc_io, MC_CMD_NO_FLAGS,
+			      dflt_dprc_handle, &dpmac_endpoint);
+	if (err < 0)
+		printf("dprc_disconnect() failed dpmac_endpoint\n");
+
+	err = dpmac_destroy(dflt_mc_io, MC_CMD_NO_FLAGS, priv->dpmac_handle);
+	if (err < 0)
+		printf("dpmac_destroy() failed\n");
+
 	/* Stop Tx and Rx traffic */
-	err = dpni_disable(dflt_mc_io, MC_CMD_NO_FLAGS, priv->dpni_handle);
+	err = dpni_disable(dflt_mc_io, MC_CMD_NO_FLAGS, dflt_dpni->dpni_handle);
 	if (err < 0)
 		printf("dpni_disable() failed\n");
 
@@ -316,8 +358,8 @@ static void ldpaa_eth_stop(struct eth_device *net_dev)
 #endif
 
 	ldpaa_dpbp_free();
-	dpni_reset(dflt_mc_io, MC_CMD_NO_FLAGS, priv->dpni_handle);
-	dpni_close(dflt_mc_io, MC_CMD_NO_FLAGS, priv->dpni_handle);
+	dpni_reset(dflt_mc_io, MC_CMD_NO_FLAGS, dflt_dpni->dpni_handle);
+	dpni_close(dflt_mc_io, MC_CMD_NO_FLAGS, dflt_dpni->dpni_handle);
 }
 
 static void ldpaa_dpbp_drain_cnt(int count)
@@ -458,54 +500,96 @@ static void ldpaa_dpbp_free(void)
 	dpbp_close(dflt_mc_io, MC_CMD_NO_FLAGS, dflt_dpbp->dpbp_handle);
 }
 
+static int ldpaa_dpmac_setup(struct ldpaa_eth_priv *priv)
+{
+	int err = 0;
+	struct dpmac_cfg dpmac_cfg;
+
+	dpmac_cfg.mac_id = priv->dpmac_id;
+	err = dpmac_create(dflt_mc_io, MC_CMD_NO_FLAGS, &dpmac_cfg,
+			  &priv->dpmac_handle);
+	if (err)
+		printf("dpmac_create() failed\n");
+	return err;
+}
+
+static int ldpaa_dpmac_bind(struct ldpaa_eth_priv *priv)
+{
+	int err = 0;
+	struct dprc_connection_cfg dprc_connection_cfg = {
+		/* If both rates are zero the connection */
+		/* will be configured in "best effort" mode. */
+		.committed_rate = 0,
+		.max_rate = 0
+	};
+
+	memset(&dpmac_endpoint, 0, sizeof(struct dprc_endpoint));
+	sprintf(dpmac_endpoint.type, "dpmac");
+	dpmac_endpoint.id = priv->dpmac_id;
+
+	memset(&dpni_endpoint, 0, sizeof(struct dprc_endpoint));
+	sprintf(dpni_endpoint.type, "dpni");
+	dpni_endpoint.id = dflt_dpni->dpni_id;
+
+	err = dprc_connect(dflt_mc_io, MC_CMD_NO_FLAGS,
+			     dflt_dprc_handle,
+			     &dpmac_endpoint,
+			     &dpni_endpoint,
+			     &dprc_connection_cfg);
+	return err;
+}
+
 static int ldpaa_dpni_setup(struct ldpaa_eth_priv *priv)
 {
 	int err;
 
 	/* and get a handle for the DPNI this interface is associate with */
-	err = dpni_open(dflt_mc_io, MC_CMD_NO_FLAGS, priv->dpni_id,
-			&priv->dpni_handle);
+	err = dpni_open(dflt_mc_io, MC_CMD_NO_FLAGS, dflt_dpni->dpni_id,
+			&dflt_dpni->dpni_handle);
 	if (err) {
 		printf("dpni_open() failed\n");
 		goto err_open;
 	}
 
 	err = dpni_get_attributes(dflt_mc_io, MC_CMD_NO_FLAGS,
-				  priv->dpni_handle, &priv->dpni_attrs);
+				  dflt_dpni->dpni_handle,
+				  &dflt_dpni->dpni_attrs);
 	if (err) {
 		printf("dpni_get_attributes() failed (err=%d)\n", err);
 		goto err_get_attr;
 	}
 
 	/* Configure our buffers' layout */
-	priv->buf_layout.options = DPNI_BUF_LAYOUT_OPT_PARSER_RESULT |
+	dflt_dpni->buf_layout.options = DPNI_BUF_LAYOUT_OPT_PARSER_RESULT |
 				   DPNI_BUF_LAYOUT_OPT_FRAME_STATUS |
 				   DPNI_BUF_LAYOUT_OPT_PRIVATE_DATA_SIZE;
-	priv->buf_layout.pass_parser_result = true;
-	priv->buf_layout.pass_frame_status = true;
-	priv->buf_layout.private_data_size = LDPAA_ETH_SWA_SIZE;
+	dflt_dpni->buf_layout.pass_parser_result = true;
+	dflt_dpni->buf_layout.pass_frame_status = true;
+	dflt_dpni->buf_layout.private_data_size = LDPAA_ETH_SWA_SIZE;
 	/* ...rx, ... */
 	err = dpni_set_rx_buffer_layout(dflt_mc_io, MC_CMD_NO_FLAGS,
-					priv->dpni_handle, &priv->buf_layout);
+					dflt_dpni->dpni_handle,
+					&dflt_dpni->buf_layout);
 	if (err) {
 		printf("dpni_set_rx_buffer_layout() failed");
 		goto err_buf_layout;
 	}
 
 	/* ... tx, ... */
-	priv->buf_layout.options &= ~DPNI_BUF_LAYOUT_OPT_PARSER_RESULT;
+	dflt_dpni->buf_layout.options &= ~DPNI_BUF_LAYOUT_OPT_PARSER_RESULT;
 	err = dpni_set_tx_buffer_layout(dflt_mc_io, MC_CMD_NO_FLAGS,
-					priv->dpni_handle, &priv->buf_layout);
+					dflt_dpni->dpni_handle,
+					&dflt_dpni->buf_layout);
 	if (err) {
 		printf("dpni_set_tx_buffer_layout() failed");
 		goto err_buf_layout;
 	}
 
 	/* ... tx-confirm. */
-	priv->buf_layout.options &= ~DPNI_BUF_LAYOUT_OPT_PRIVATE_DATA_SIZE;
+	dflt_dpni->buf_layout.options &= ~DPNI_BUF_LAYOUT_OPT_PRIVATE_DATA_SIZE;
 	err = dpni_set_tx_conf_buffer_layout(dflt_mc_io, MC_CMD_NO_FLAGS,
-					     priv->dpni_handle,
-					     &priv->buf_layout);
+					     dflt_dpni->dpni_handle,
+					     &dflt_dpni->buf_layout);
 	if (err) {
 		printf("dpni_set_tx_conf_buffer_layout() failed");
 		goto err_buf_layout;
@@ -515,7 +599,8 @@ static int ldpaa_dpni_setup(struct ldpaa_eth_priv *priv)
 	 * required tx data offset.
 	 */
 	err = dpni_get_tx_data_offset(dflt_mc_io, MC_CMD_NO_FLAGS,
-				      priv->dpni_handle, &priv->tx_data_offset);
+				      dflt_dpni->dpni_handle,
+				      &priv->tx_data_offset);
 	if (err) {
 		printf("dpni_get_tx_data_offset() failed\n");
 		goto err_data_offset;
@@ -533,7 +618,7 @@ static int ldpaa_dpni_setup(struct ldpaa_eth_priv *priv)
 err_data_offset:
 err_buf_layout:
 err_get_attr:
-	dpni_close(dflt_mc_io, MC_CMD_NO_FLAGS, priv->dpni_handle);
+	dpni_close(dflt_mc_io, MC_CMD_NO_FLAGS, dflt_dpni->dpni_handle);
 err_open:
 	return err;
 }
@@ -547,8 +632,8 @@ static int ldpaa_dpni_bind(struct ldpaa_eth_priv *priv)
 	pools_params.num_dpbp = 1;
 	pools_params.pools[0].dpbp_id = (uint16_t)dflt_dpbp->dpbp_attr.id;
 	pools_params.pools[0].buffer_size = LDPAA_ETH_RX_BUFFER_SIZE;
-	err = dpni_set_pools(dflt_mc_io, MC_CMD_NO_FLAGS, priv->dpni_handle,
-			     &pools_params);
+	err = dpni_set_pools(dflt_mc_io, MC_CMD_NO_FLAGS,
+			     dflt_dpni->dpni_handle, &pools_params);
 	if (err) {
 		printf("dpni_set_pools() failed\n");
 		return err;
@@ -560,8 +645,9 @@ static int ldpaa_dpni_bind(struct ldpaa_eth_priv *priv)
 	dflt_tx_flow.options = DPNI_TX_FLOW_OPT_ONLY_TX_ERROR;
 	dflt_tx_flow.conf_err_cfg.use_default_queue = 0;
 	dflt_tx_flow.conf_err_cfg.errors_only = 1;
-	err = dpni_set_tx_flow(dflt_mc_io, MC_CMD_NO_FLAGS, priv->dpni_handle,
-			       &priv->tx_flow_id, &dflt_tx_flow);
+	err = dpni_set_tx_flow(dflt_mc_io, MC_CMD_NO_FLAGS,
+			       dflt_dpni->dpni_handle, &priv->tx_flow_id,
+			       &dflt_tx_flow);
 	if (err) {
 		printf("dpni_set_tx_flow() failed\n");
 		return err;
@@ -570,12 +656,14 @@ static int ldpaa_dpni_bind(struct ldpaa_eth_priv *priv)
 	return 0;
 }
 
-static int ldpaa_eth_netdev_init(struct eth_device *net_dev)
+static int ldpaa_eth_netdev_init(struct eth_device *net_dev,
+				 phy_interface_t enet_if)
 {
 	int err;
 	struct ldpaa_eth_priv *priv = (struct ldpaa_eth_priv *)net_dev->priv;
 
-	sprintf(net_dev->name, "DPNI%d", priv->dpni_id);
+	sprintf(net_dev->name, "DPMAC%d@%s", priv->dpmac_id,
+		phy_interface_strings[enet_if]);
 
 	net_dev->iobase = 0;
 	net_dev->init = ldpaa_eth_open;
@@ -601,7 +689,7 @@ static int ldpaa_eth_netdev_init(struct eth_device *net_dev)
 	return 0;
 }
 
-int ldpaa_eth_init(int id)
+int ldpaa_eth_init(int dpmac_id, phy_interface_t enet_if)
 {
 	struct eth_device		*net_dev = NULL;
 	struct ldpaa_eth_priv		*priv = NULL;
@@ -626,9 +714,10 @@ int ldpaa_eth_init(int id)
 
 	net_dev->priv = (void *)priv;
 	priv->net_dev = (struct eth_device *)net_dev;
-	priv->dpni_id = id;
+	priv->dpmac_id = dpmac_id;
+	debug("%s dpmac_id=%d\n", __func__, dpmac_id);
 
-	err = ldpaa_eth_netdev_init(net_dev);
+	err = ldpaa_eth_netdev_init(net_dev, enet_if);
 	if (err)
 		goto err_netdev_init;
 

drivers/net/ldpaa_eth/ldpaa_eth.h

@@ -117,13 +117,9 @@ struct ldpaa_fas {
 
 struct ldpaa_eth_priv {
 	struct eth_device *net_dev;
-	int dpni_id;
-	uint16_t dpni_handle;
-	struct dpni_attr dpni_attrs;
-	/* Insofar as the MC is concerned, we're using one layout on all 3 types
-	 * of buffers (Rx, Tx, Tx-Conf).
-	 */
-	struct dpni_buffer_layout buf_layout;
+	int dpmac_id;
+	uint16_t dpmac_handle;
+
 	uint16_t tx_data_offset;
 
 	uint32_t rx_dflt_fqid;
@@ -134,9 +130,14 @@ struct ldpaa_eth_priv {
 	struct phy_device *phydev;
 };
 
+struct dprc_endpoint dpmac_endpoint;
+struct dprc_endpoint dpni_endpoint;
+
 extern struct fsl_mc_io *dflt_mc_io;
 extern struct fsl_dpbp_obj *dflt_dpbp;
 extern struct fsl_dpio_obj *dflt_dpio;
+extern struct fsl_dpni_obj *dflt_dpni;
+extern uint16_t dflt_dprc_handle;
 
 static void ldpaa_dpbp_drain_cnt(int count);
 static void ldpaa_dpbp_drain(void);
@@ -145,4 +146,6 @@ static void ldpaa_dpbp_free(void);
 static int ldpaa_dpni_setup(struct ldpaa_eth_priv *priv);
 static int ldpaa_dpbp_setup(void);
 static int ldpaa_dpni_bind(struct ldpaa_eth_priv *priv);
+static int ldpaa_dpmac_setup(struct ldpaa_eth_priv *priv);
+static int ldpaa_dpmac_bind(struct ldpaa_eth_priv *priv);
 #endif	/* __LDPAA_H */

include/fsl-mc/fsl_mc_private.h

@@ -13,6 +13,7 @@
 #include <linux/compat.h>
 #include <linux/types.h>
 #include <linux/stringify.h>
+#include <phy.h>
 
 #include <fsl-mc/fsl_mc_sys.h>
 #include <fsl-mc/fsl_mc_cmd.h>
@@ -64,6 +65,6 @@ struct fsl_dpni_obj {
 extern struct fsl_dpni_obj *dflt_dpni;
 
 int mc_init(u64 mc_fw_addr, u64 mc_dpc_addr);
-int ldpaa_eth_init(int dpmac_id);
+int ldpaa_eth_init(int dpmac_id, phy_interface_t enet_if);
 int mc_apply_dpl(u64 mc_dpl_addr);
 #endif /* _FSL_MC_PRIVATE_H_ */