[PATCH] testpmd: add hairpin-map parameter

[PATCH] testpmd: add hairpin-map parameter

[Gregory Etelson]

Testpmd hairpin implementation always sets the next valid port to
complete hairpin binding. That limits hairpin configuration options.

The new parameter allows explicit selection of Rx and Tx ports and
queues in hairpin configuration.
The new `hairpin-map` parameter is provided with 5 parameters,
separated by `:`

`--hairpin-map=Rx port id:Rx queue:Tx port id:Tx queue:queues number`

Testpmd operator can provide several `hairpin-map` parameters for
different hairpin maps.
Example:

dpdk-testpmd <EAL params> -- \
  <testpmd params> \
  --rxq=2 --txq=2 --hairpinq=2 --hairpin-mode=0x12 \
  --hairpin-map=0:2:1:2:1 \ # [1]
  --hairpin-map=0:3:2:2:3   # [2]

Hairpin map [1] binds Rx port 0, queue 2 with Tx port 1, queue 2.
Hairpin map [2] binds
  Rx port 0, queue 3 with Tx port 2, queue 2,
  Rx port 0, queue 4 with Tx port 2, queue 3,
  Rx port 0, queue 5 with Tx port 2, queue 4.

The new `hairpin-map` parameter is optional.
If omitted, testpmd will create "default" hairpin maps.

Signed-off-by: Gregory Etelson <getelson@nvidia.com>
---
 app/test-pmd/parameters.c             |  63 ++++++++
 app/test-pmd/testpmd.c                | 212 ++++++++++++++++++--------
 app/test-pmd/testpmd.h                |  18 +++
 doc/guides/testpmd_app_ug/run_app.rst |   3 +
 4 files changed, 230 insertions(+), 66 deletions(-)

diff --git a/app/test-pmd/parameters.c b/app/test-pmd/parameters.c
index a9ca58339d..675de81861 100644
--- a/app/test-pmd/parameters.c
+++ b/app/test-pmd/parameters.c
@@ -206,6 +206,12 @@ usage(char* progname)
 	printf("  --hairpin-mode=0xXX: bitmask set the hairpin port mode.\n"
 	       "    0x10 - explicit Tx rule, 0x02 - hairpin ports paired\n"
 	       "    0x01 - hairpin ports loop, 0x00 - hairpin port self\n");
+	printf("  --hairpin-map=rxpi:rxq:txpi:txq:n: hairpin map.\n"
+	       "    rxpi - Rx port index.\n"
+	       "    rxq  - Rx queue.\n"
+	       "    txpi - Tx port index.\n"
+	       "    txq  - Tx queue.\n"
+	       "    n    - hairpin queues number.\n");
 }
 
 #ifdef RTE_LIB_CMDLINE
@@ -588,6 +594,55 @@ parse_link_speed(int n)
 	return speed;
 }
 
+static __rte_always_inline
+char *parse_hairpin_map_entry(char *input, char **next)
+{
+	char *tail = strchr(input, ':');
+
+	if (!tail)
+		return NULL;
+	tail[0] = '\0';
+	*next = tail + 1;
+	return input;
+}
+
+static int
+parse_hairpin_map(char *hpmap)
+{
+	/*
+	 * Testpmd hairpin map format:
+	 * <Rx port id:First Rx queue:Tx port id:First Tx queue:queues number>
+	 */
+	char *head, *next = hpmap;
+	struct hairpin_map *map = calloc(1, sizeof(*map));
+
+	if (!map)
+		return -ENOMEM;
+
+	head = parse_hairpin_map_entry(next, &next);
+	if (!head)
+		goto err;
+	map->rx_port = atoi(head);
+	head = parse_hairpin_map_entry(next, &next);
+	if (!head)
+		goto err;
+	map->rxq_head = atoi(head);
+	head = parse_hairpin_map_entry(next, &next);
+	if (!head)
+		goto err;
+	map->tx_port = atoi(head);
+	head = parse_hairpin_map_entry(next, &next);
+	if (!head)
+		goto err;
+	map->txq_head = atoi(head);
+	map->qnum = atoi(next);
+	hairpin_add_multiport_map(map);
+	return 0;
+err:
+	free(map);
+	return -EINVAL;
+}
+
 void
 launch_args_parse(int argc, char** argv)
 {
@@ -663,6 +718,7 @@ launch_args_parse(int argc, char** argv)
 		{ "txd",			1, 0, 0 },
 		{ "hairpinq",			1, 0, 0 },
 		{ "hairpin-mode",		1, 0, 0 },
+		{ "hairpin-map",                1, 0, 0 },
 		{ "burst",			1, 0, 0 },
 		{ "flowgen-clones",		1, 0, 0 },
 		{ "flowgen-flows",		1, 0, 0 },
@@ -1111,6 +1167,13 @@ launch_args_parse(int argc, char** argv)
 				else
 					hairpin_mode = (uint32_t)n;
 			}
+			if (!strcmp(lgopts[opt_idx].name, "hairpin-map")) {
+				hairpin_multiport_mode = true;
+				ret = parse_hairpin_map(optarg);
+				if (ret)
+					rte_exit(EXIT_FAILURE, "invalid hairpin map\n");
+
+			}
 			if (!strcmp(lgopts[opt_idx].name, "burst")) {
 				n = atoi(optarg);
 				if (n == 0) {
diff --git a/app/test-pmd/testpmd.c b/app/test-pmd/testpmd.c
index 938ca035d4..2ba1727c51 100644
--- a/app/test-pmd/testpmd.c
+++ b/app/test-pmd/testpmd.c
@@ -434,6 +434,16 @@ uint8_t clear_ptypes = true;
 /* Hairpin ports configuration mode. */
 uint32_t hairpin_mode;
 
+bool hairpin_multiport_mode = false;
+
+static LIST_HEAD(,hairpin_map) hairpin_map_head = LIST_HEAD_INITIALIZER();
+
+void
+hairpin_add_multiport_map(struct hairpin_map *map)
+{
+	LIST_INSERT_HEAD(&hairpin_map_head, map, entry);
+}
+
 /* Pretty printing of ethdev events */
 static const char * const eth_event_desc[] = {
 	[RTE_ETH_EVENT_UNKNOWN] = "unknown",
@@ -2677,28 +2687,107 @@ port_is_started(portid_t port_id)
 #define HAIRPIN_MODE_TX_LOCKED_MEMORY RTE_BIT32(16)
 #define HAIRPIN_MODE_TX_RTE_MEMORY RTE_BIT32(17)
 
-
-/* Configure the Rx and Tx hairpin queues for the selected port. */
 static int
-setup_hairpin_queues(portid_t pi, portid_t p_pi, uint16_t cnt_pi)
+port_config_hairpin_rxq(portid_t pi, uint16_t peer_tx_port,
+			queueid_t rxq_head, queueid_t txq_head,
+			uint16_t qcount, uint32_t manual_bind)
 {
-	queueid_t qi;
+	int diag;
+	queueid_t i, qi;
+	uint32_t tx_explicit = !!(hairpin_mode & 0x10);
+	uint32_t force_mem = !!(hairpin_mode & HAIRPIN_MODE_RX_FORCE_MEMORY);
+	uint32_t locked_mem = !!(hairpin_mode & HAIRPIN_MODE_RX_LOCKED_MEMORY);
+	uint32_t rte_mem = !!(hairpin_mode & HAIRPIN_MODE_RX_RTE_MEMORY);
+	struct rte_port *port = &ports[pi];
 	struct rte_eth_hairpin_conf hairpin_conf = {
 		.peer_count = 1,
 	};
-	int i;
+
+	for (qi = rxq_head, i = 0; qi < rxq_head + qcount; qi++) {
+		hairpin_conf.peers[0].port = peer_tx_port;
+		hairpin_conf.peers[0].queue = i + txq_head;
+		hairpin_conf.manual_bind = manual_bind;
+		hairpin_conf.tx_explicit = tx_explicit;
+		hairpin_conf.force_memory = force_mem;
+		hairpin_conf.use_locked_device_memory = locked_mem;
+		hairpin_conf.use_rte_memory = rte_mem;
+		diag = rte_eth_rx_hairpin_queue_setup
+			(pi, qi, nb_rxd, &hairpin_conf);
+		i++;
+		if (diag == 0)
+			continue;
+
+		/* Fail to setup rx queue, return */
+		if (port->port_status == RTE_PORT_HANDLING)
+			port->port_status = RTE_PORT_STOPPED;
+		else
+			fprintf(stderr,
+				"Port %d can not be set back to stopped\n", pi);
+		fprintf(stderr,
+			"Port %d failed to configure hairpin on rxq %u.\n"
+			"Peer port: %u peer txq: %u\n",
+			pi, qi, peer_tx_port, i);
+		/* try to reconfigure queues next time */
+		port->need_reconfig_queues = 1;
+		return -1;
+	}
+	return 0;
+}
+
+static int
+port_config_hairpin_txq(portid_t pi, uint16_t peer_rx_port,
+			queueid_t rxq_head, queueid_t txq_head,
+			uint16_t qcount, uint32_t manual_bind)
+{
 	int diag;
+	queueid_t i, qi;
+	uint32_t tx_explicit = !!(hairpin_mode & 0x10);
+	uint32_t force_mem = !!(hairpin_mode & HAIRPIN_MODE_TX_FORCE_MEMORY);
+	uint32_t locked_mem = !!(hairpin_mode & HAIRPIN_MODE_TX_LOCKED_MEMORY);
+	uint32_t rte_mem = !!(hairpin_mode & HAIRPIN_MODE_TX_RTE_MEMORY);
 	struct rte_port *port = &ports[pi];
+	struct rte_eth_hairpin_conf hairpin_conf = {
+		.peer_count = 1,
+	};
+
+	for (qi = txq_head, i = 0; qi < txq_head + qcount; qi++) {
+		hairpin_conf.peers[0].port = peer_rx_port;
+		hairpin_conf.peers[0].queue = i + rxq_head;
+		hairpin_conf.manual_bind = manual_bind;
+		hairpin_conf.tx_explicit = tx_explicit;
+		hairpin_conf.force_memory = force_mem;
+		hairpin_conf.use_locked_device_memory = locked_mem;
+		hairpin_conf.use_rte_memory = rte_mem;
+		diag = rte_eth_tx_hairpin_queue_setup
+			(pi, qi, nb_txd, &hairpin_conf);
+		i++;
+		if (diag == 0)
+			continue;
+
+		/* Fail to setup rx queue, return */
+		if (port->port_status == RTE_PORT_HANDLING)
+			port->port_status = RTE_PORT_STOPPED;
+		else
+			fprintf(stderr,
+				"Port %d can not be set back to stopped\n", pi);
+		fprintf(stderr,
+			"Port %d failed to configure hairpin on txq %u.\n"
+			"Peer port: %u peer rxq: %u\n",
+			pi, qi, peer_rx_port, i);
+		/* try to reconfigure queues next time */
+		port->need_reconfig_queues = 1;
+		return -1;
+	}
+	return 0;
+}
+
+static int
+setup_legacy_hairpin_queus(portid_t pi, portid_t p_pi, uint16_t cnt_pi)
+{
+	int diag;
 	uint16_t peer_rx_port = pi;
 	uint16_t peer_tx_port = pi;
 	uint32_t manual = 1;
-	uint32_t tx_exp = hairpin_mode & 0x10;
-	uint32_t rx_force_memory = hairpin_mode & HAIRPIN_MODE_RX_FORCE_MEMORY;
-	uint32_t rx_locked_memory = hairpin_mode & HAIRPIN_MODE_RX_LOCKED_MEMORY;
-	uint32_t rx_rte_memory = hairpin_mode & HAIRPIN_MODE_RX_RTE_MEMORY;
-	uint32_t tx_force_memory = hairpin_mode & HAIRPIN_MODE_TX_FORCE_MEMORY;
-	uint32_t tx_locked_memory = hairpin_mode & HAIRPIN_MODE_TX_LOCKED_MEMORY;
-	uint32_t tx_rte_memory = hairpin_mode & HAIRPIN_MODE_TX_RTE_MEMORY;
 
 	if (!(hairpin_mode & 0xf)) {
 		peer_rx_port = pi;
@@ -2706,10 +2795,10 @@ setup_hairpin_queues(portid_t pi, portid_t p_pi, uint16_t cnt_pi)
 		manual = 0;
 	} else if (hairpin_mode & 0x1) {
 		peer_tx_port = rte_eth_find_next_owned_by(pi + 1,
-						       RTE_ETH_DEV_NO_OWNER);
+							  RTE_ETH_DEV_NO_OWNER);
 		if (peer_tx_port >= RTE_MAX_ETHPORTS)
 			peer_tx_port = rte_eth_find_next_owned_by(0,
-						RTE_ETH_DEV_NO_OWNER);
+								  RTE_ETH_DEV_NO_OWNER);
 		if (p_pi != RTE_MAX_ETHPORTS) {
 			peer_rx_port = p_pi;
 		} else {
@@ -2725,69 +2814,60 @@ setup_hairpin_queues(portid_t pi, portid_t p_pi, uint16_t cnt_pi)
 			peer_rx_port = p_pi;
 		} else {
 			peer_rx_port = rte_eth_find_next_owned_by(pi + 1,
-						RTE_ETH_DEV_NO_OWNER);
+								  RTE_ETH_DEV_NO_OWNER);
 			if (peer_rx_port >= RTE_MAX_ETHPORTS)
 				peer_rx_port = pi;
 		}
 		peer_tx_port = peer_rx_port;
 		manual = 1;
 	}
+	diag = port_config_hairpin_txq(pi, peer_rx_port, nb_rxq, nb_txq,
+				       nb_hairpinq, manual);
+	if (diag)
+		return diag;
+	diag = port_config_hairpin_rxq(pi, peer_tx_port, nb_rxq, nb_txq,
+				       nb_hairpinq, manual);
+	if (diag)
+		return diag;
+	return 0;
+}
 
-	for (qi = nb_txq, i = 0; qi < nb_hairpinq + nb_txq; qi++) {
-		hairpin_conf.peers[0].port = peer_rx_port;
-		hairpin_conf.peers[0].queue = i + nb_rxq;
-		hairpin_conf.manual_bind = !!manual;
-		hairpin_conf.tx_explicit = !!tx_exp;
-		hairpin_conf.force_memory = !!tx_force_memory;
-		hairpin_conf.use_locked_device_memory = !!tx_locked_memory;
-		hairpin_conf.use_rte_memory = !!tx_rte_memory;
-		diag = rte_eth_tx_hairpin_queue_setup
-			(pi, qi, nb_txd, &hairpin_conf);
-		i++;
-		if (diag == 0)
-			continue;
-
-		/* Fail to setup rx queue, return */
-		if (port->port_status == RTE_PORT_HANDLING)
-			port->port_status = RTE_PORT_STOPPED;
-		else
-			fprintf(stderr,
-				"Port %d can not be set back to stopped\n", pi);
-		fprintf(stderr, "Fail to configure port %d hairpin queues\n",
-			pi);
-		/* try to reconfigure queues next time */
-		port->need_reconfig_queues = 1;
-		return -1;
-	}
-	for (qi = nb_rxq, i = 0; qi < nb_hairpinq + nb_rxq; qi++) {
-		hairpin_conf.peers[0].port = peer_tx_port;
-		hairpin_conf.peers[0].queue = i + nb_txq;
-		hairpin_conf.manual_bind = !!manual;
-		hairpin_conf.tx_explicit = !!tx_exp;
-		hairpin_conf.force_memory = !!rx_force_memory;
-		hairpin_conf.use_locked_device_memory = !!rx_locked_memory;
-		hairpin_conf.use_rte_memory = !!rx_rte_memory;
-		diag = rte_eth_rx_hairpin_queue_setup
-			(pi, qi, nb_rxd, &hairpin_conf);
-		i++;
-		if (diag == 0)
-			continue;
-
-		/* Fail to setup rx queue, return */
-		if (port->port_status == RTE_PORT_HANDLING)
-			port->port_status = RTE_PORT_STOPPED;
-		else
-			fprintf(stderr,
-				"Port %d can not be set back to stopped\n", pi);
-		fprintf(stderr, "Fail to configure port %d hairpin queues\n",
-			pi);
-		/* try to reconfigure queues next time */
-		port->need_reconfig_queues = 1;
-		return -1;
+static int
+setup_mapped_harpin_queues(portid_t pi)
+{
+	int ret = 0;
+	struct hairpin_map *map;
+
+	LIST_FOREACH(map, &hairpin_map_head, entry) {
+		if (map->rx_port == pi) {
+			ret = port_config_hairpin_rxq(pi, map->tx_port,
+						      map->rxq_head,
+						      map->txq_head,
+						      map->qnum, true);
+			if (ret)
+				return ret;
+		}
+		if (map->tx_port == pi) {
+			ret = port_config_hairpin_txq(pi, map->rx_port,
+						      map->rxq_head,
+						      map->txq_head,
+						      map->qnum, true);
+			if (ret)
+				return ret;
+		}
 	}
 	return 0;
 }
 
+/* Configure the Rx and Tx hairpin queues for the selected port. */
+static int
+setup_hairpin_queues(portid_t pi, portid_t p_pi, uint16_t cnt_pi)
+{
+	return !hairpin_multiport_mode ?
+	       setup_legacy_hairpin_queus(pi, p_pi, cnt_pi) :
+	       setup_mapped_harpin_queues(pi);
+}
+
 /* Configure the Rx with optional split. */
 int
 rx_queue_setup(uint16_t port_id, uint16_t rx_queue_id,
diff --git a/app/test-pmd/testpmd.h b/app/test-pmd/testpmd.h
index f1df6a8faf..208e8e9514 100644
--- a/app/test-pmd/testpmd.h
+++ b/app/test-pmd/testpmd.h
@@ -125,6 +125,24 @@ enum noisy_fwd_mode {
 	NOISY_FWD_MODE_MAX,
 };
 
+struct hairpin_map {
+	LIST_ENTRY(hairpin_map) entry; /**< List entry. */
+	portid_t rx_port; /**< Hairpin Rx port ID. */
+	portid_t tx_port; /**< Hairpin Tx port ID. */
+	uint16_t rxq_head; /**< Hairpin Rx queue head. */
+	uint16_t txq_head; /**< Hairpin Tx queue head. */
+	uint16_t qnum; /**< Hairpin queues number. */
+};
+
+/**
+ * Command line arguments parser sets `hairpin_multiport_mode` to True
+ * if explicit hairpin map configuration mode was used.
+ */
+extern bool hairpin_multiport_mode;
+
+/** Hairpin maps list. */
+extern void hairpin_add_multiport_map(struct hairpin_map *map);
+
 /**
  * The data structure associated with RX and TX packet burst statistics
  * that are recorded for each forwarding stream.
diff --git a/doc/guides/testpmd_app_ug/run_app.rst b/doc/guides/testpmd_app_ug/run_app.rst
index 6e9c552e76..a202c98b4c 100644
--- a/doc/guides/testpmd_app_ug/run_app.rst
+++ b/doc/guides/testpmd_app_ug/run_app.rst
@@ -566,6 +566,9 @@ The command line options are:
 
     The default value is 0. Hairpin will use single port mode and implicit Tx flow mode.
 
+*   ``--hairpin-map=Rx port id:Rx queue:Tx port id:Tx queue:queues number``
+
+    Set explicit hairpin configuration.
 
 Testpmd Multi-Process Command-line Options
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-- 
2.39.2

[PATCH v2] testpmd: add hairpin-map parameter

[Gregory Etelson]

Testpmd hairpin implementation always sets the next valid port to
complete hairpin binding. That limits hairpin configuration options.

The new parameter allows explicit selection of Rx and Tx ports and
queues in hairpin configuration.
The new `hairpin-map` parameter is provided with 5 parameters,
separated by `:`

`--hairpin-map=Rx port id:Rx queue:Tx port id:Tx queue:queues number`

Testpmd operator can provide several `hairpin-map` parameters for
different hairpin maps.
Example:

dpdk-testpmd <EAL params> -- \
  <testpmd params> \
  --rxq=2 --txq=2 --hairpinq=2 --hairpin-mode=0x12 \
  --hairpin-map=0:2:1:2:1 \ # [1]
  --hairpin-map=0:3:2:2:3   # [2]

Hairpin map [1] binds Rx port 0, queue 2 with Tx port 1, queue 2.
Hairpin map [2] binds
  Rx port 0, queue 3 with Tx port 2, queue 2,
  Rx port 0, queue 4 with Tx port 2, queue 3,
  Rx port 0, queue 5 with Tx port 2, queue 4.

The new `hairpin-map` parameter is optional.
If omitted, testpmd will create "default" hairpin maps.

Signed-off-by: Gregory Etelson <getelson@nvidia.com>
---
v2: fix Windows Server 2019 compilation failure.
---
 app/test-pmd/parameters.c             |  63 ++++++++
 app/test-pmd/testpmd.c                | 212 ++++++++++++++++++--------
 app/test-pmd/testpmd.h                |  18 +++
 doc/guides/testpmd_app_ug/run_app.rst |   3 +
 4 files changed, 230 insertions(+), 66 deletions(-)

diff --git a/app/test-pmd/parameters.c b/app/test-pmd/parameters.c
index a9ca58339d..c6bdfdf06f 100644
--- a/app/test-pmd/parameters.c
+++ b/app/test-pmd/parameters.c
@@ -206,6 +206,12 @@ usage(char* progname)
 	printf("  --hairpin-mode=0xXX: bitmask set the hairpin port mode.\n"
 	       "    0x10 - explicit Tx rule, 0x02 - hairpin ports paired\n"
 	       "    0x01 - hairpin ports loop, 0x00 - hairpin port self\n");
+	printf("  --hairpin-map=rxpi:rxq:txpi:txq:n: hairpin map.\n"
+	       "    rxpi - Rx port index.\n"
+	       "    rxq  - Rx queue.\n"
+	       "    txpi - Tx port index.\n"
+	       "    txq  - Tx queue.\n"
+	       "    n    - hairpin queues number.\n");
 }
 
 #ifdef RTE_LIB_CMDLINE
@@ -588,6 +594,55 @@ parse_link_speed(int n)
 	return speed;
 }
 
+static __rte_always_inline
+char *parse_hairpin_map_entry(char *input, char **next)
+{
+	char *tail = strchr(input, ':');
+
+	if (!tail)
+		return NULL;
+	tail[0] = '\0';
+	*next = tail + 1;
+	return input;
+}
+
+static int
+parse_hairpin_map(const char *hpmap)
+{
+	/*
+	 * Testpmd hairpin map format:
+	 * <Rx port id:First Rx queue:Tx port id:First Tx queue:queues number>
+	 */
+	char *head, *next = (char *)(uintptr_t)hpmap;
+	struct hairpin_map *map = calloc(1, sizeof(*map));
+
+	if (!map)
+		return -ENOMEM;
+
+	head = parse_hairpin_map_entry(next, &next);
+	if (!head)
+		goto err;
+	map->rx_port = atoi(head);
+	head = parse_hairpin_map_entry(next, &next);
+	if (!head)
+		goto err;
+	map->rxq_head = atoi(head);
+	head = parse_hairpin_map_entry(next, &next);
+	if (!head)
+		goto err;
+	map->tx_port = atoi(head);
+	head = parse_hairpin_map_entry(next, &next);
+	if (!head)
+		goto err;
+	map->txq_head = atoi(head);
+	map->qnum = atoi(next);
+	hairpin_add_multiport_map(map);
+	return 0;
+err:
+	free(map);
+	return -EINVAL;
+}
+
 void
 launch_args_parse(int argc, char** argv)
 {
@@ -663,6 +718,7 @@ launch_args_parse(int argc, char** argv)
 		{ "txd",			1, 0, 0 },
 		{ "hairpinq",			1, 0, 0 },
 		{ "hairpin-mode",		1, 0, 0 },
+		{ "hairpin-map",                1, 0, 0 },
 		{ "burst",			1, 0, 0 },
 		{ "flowgen-clones",		1, 0, 0 },
 		{ "flowgen-flows",		1, 0, 0 },
@@ -1111,6 +1167,13 @@ launch_args_parse(int argc, char** argv)
 				else
 					hairpin_mode = (uint32_t)n;
 			}
+			if (!strcmp(lgopts[opt_idx].name, "hairpin-map")) {
+				hairpin_multiport_mode = true;
+				ret = parse_hairpin_map(optarg);
+				if (ret)
+					rte_exit(EXIT_FAILURE, "invalid hairpin map\n");
+
+			}
 			if (!strcmp(lgopts[opt_idx].name, "burst")) {
 				n = atoi(optarg);
 				if (n == 0) {
diff --git a/app/test-pmd/testpmd.c b/app/test-pmd/testpmd.c
index 938ca035d4..2c6975f22d 100644
--- a/app/test-pmd/testpmd.c
+++ b/app/test-pmd/testpmd.c
@@ -434,6 +434,16 @@ uint8_t clear_ptypes = true;
 /* Hairpin ports configuration mode. */
 uint32_t hairpin_mode;
 
+bool hairpin_multiport_mode = false;
+
+static LIST_HEAD(, hairpin_map) hairpin_map_head = LIST_HEAD_INITIALIZER();
+
+void
+hairpin_add_multiport_map(struct hairpin_map *map)
+{
+	LIST_INSERT_HEAD(&hairpin_map_head, map, entry);
+}
+
 /* Pretty printing of ethdev events */
 static const char * const eth_event_desc[] = {
 	[RTE_ETH_EVENT_UNKNOWN] = "unknown",
@@ -2677,28 +2687,107 @@ port_is_started(portid_t port_id)
 #define HAIRPIN_MODE_TX_LOCKED_MEMORY RTE_BIT32(16)
 #define HAIRPIN_MODE_TX_RTE_MEMORY RTE_BIT32(17)
 
-
-/* Configure the Rx and Tx hairpin queues for the selected port. */
 static int
-setup_hairpin_queues(portid_t pi, portid_t p_pi, uint16_t cnt_pi)
+port_config_hairpin_rxq(portid_t pi, uint16_t peer_tx_port,
+			queueid_t rxq_head, queueid_t txq_head,
+			uint16_t qcount, uint32_t manual_bind)
 {
-	queueid_t qi;
+	int diag;
+	queueid_t i, qi;
+	uint32_t tx_explicit = !!(hairpin_mode & 0x10);
+	uint32_t force_mem = !!(hairpin_mode & HAIRPIN_MODE_RX_FORCE_MEMORY);
+	uint32_t locked_mem = !!(hairpin_mode & HAIRPIN_MODE_RX_LOCKED_MEMORY);
+	uint32_t rte_mem = !!(hairpin_mode & HAIRPIN_MODE_RX_RTE_MEMORY);
+	struct rte_port *port = &ports[pi];
 	struct rte_eth_hairpin_conf hairpin_conf = {
 		.peer_count = 1,
 	};
-	int i;
+
+	for (qi = rxq_head, i = 0; qi < rxq_head + qcount; qi++) {
+		hairpin_conf.peers[0].port = peer_tx_port;
+		hairpin_conf.peers[0].queue = i + txq_head;
+		hairpin_conf.manual_bind = manual_bind;
+		hairpin_conf.tx_explicit = tx_explicit;
+		hairpin_conf.force_memory = force_mem;
+		hairpin_conf.use_locked_device_memory = locked_mem;
+		hairpin_conf.use_rte_memory = rte_mem;
+		diag = rte_eth_rx_hairpin_queue_setup
+			(pi, qi, nb_rxd, &hairpin_conf);
+		i++;
+		if (diag == 0)
+			continue;
+
+		/* Fail to setup rx queue, return */
+		if (port->port_status == RTE_PORT_HANDLING)
+			port->port_status = RTE_PORT_STOPPED;
+		else
+			fprintf(stderr,
+				"Port %d can not be set back to stopped\n", pi);
+		fprintf(stderr,
+			"Port %d failed to configure hairpin on rxq %u.\n"
+			"Peer port: %u peer txq: %u\n",
+			pi, qi, peer_tx_port, i);
+		/* try to reconfigure queues next time */
+		port->need_reconfig_queues = 1;
+		return -1;
+	}
+	return 0;
+}
+
+static int
+port_config_hairpin_txq(portid_t pi, uint16_t peer_rx_port,
+			queueid_t rxq_head, queueid_t txq_head,
+			uint16_t qcount, uint32_t manual_bind)
+{
 	int diag;
+	queueid_t i, qi;
+	uint32_t tx_explicit = !!(hairpin_mode & 0x10);
+	uint32_t force_mem = !!(hairpin_mode & HAIRPIN_MODE_TX_FORCE_MEMORY);
+	uint32_t locked_mem = !!(hairpin_mode & HAIRPIN_MODE_TX_LOCKED_MEMORY);
+	uint32_t rte_mem = !!(hairpin_mode & HAIRPIN_MODE_TX_RTE_MEMORY);
 	struct rte_port *port = &ports[pi];
+	struct rte_eth_hairpin_conf hairpin_conf = {
+		.peer_count = 1,
+	};
+
+	for (qi = txq_head, i = 0; qi < txq_head + qcount; qi++) {
+		hairpin_conf.peers[0].port = peer_rx_port;
+		hairpin_conf.peers[0].queue = i + rxq_head;
+		hairpin_conf.manual_bind = manual_bind;
+		hairpin_conf.tx_explicit = tx_explicit;
+		hairpin_conf.force_memory = force_mem;
+		hairpin_conf.use_locked_device_memory = locked_mem;
+		hairpin_conf.use_rte_memory = rte_mem;
+		diag = rte_eth_tx_hairpin_queue_setup
+			(pi, qi, nb_txd, &hairpin_conf);
+		i++;
+		if (diag == 0)
+			continue;
+
+		/* Fail to setup rx queue, return */
+		if (port->port_status == RTE_PORT_HANDLING)
+			port->port_status = RTE_PORT_STOPPED;
+		else
+			fprintf(stderr,
+				"Port %d can not be set back to stopped\n", pi);
+		fprintf(stderr,
+			"Port %d failed to configure hairpin on txq %u.\n"
+			"Peer port: %u peer rxq: %u\n",
+			pi, qi, peer_rx_port, i);
+		/* try to reconfigure queues next time */
+		port->need_reconfig_queues = 1;
+		return -1;
+	}
+	return 0;
+}
+
+static int
+setup_legacy_hairpin_queus(portid_t pi, portid_t p_pi, uint16_t cnt_pi)
+{
+	int diag;
 	uint16_t peer_rx_port = pi;
 	uint16_t peer_tx_port = pi;
 	uint32_t manual = 1;
-	uint32_t tx_exp = hairpin_mode & 0x10;
-	uint32_t rx_force_memory = hairpin_mode & HAIRPIN_MODE_RX_FORCE_MEMORY;
-	uint32_t rx_locked_memory = hairpin_mode & HAIRPIN_MODE_RX_LOCKED_MEMORY;
-	uint32_t rx_rte_memory = hairpin_mode & HAIRPIN_MODE_RX_RTE_MEMORY;
-	uint32_t tx_force_memory = hairpin_mode & HAIRPIN_MODE_TX_FORCE_MEMORY;
-	uint32_t tx_locked_memory = hairpin_mode & HAIRPIN_MODE_TX_LOCKED_MEMORY;
-	uint32_t tx_rte_memory = hairpin_mode & HAIRPIN_MODE_TX_RTE_MEMORY;
 
 	if (!(hairpin_mode & 0xf)) {
 		peer_rx_port = pi;
@@ -2706,10 +2795,10 @@ setup_hairpin_queues(portid_t pi, portid_t p_pi, uint16_t cnt_pi)
 		manual = 0;
 	} else if (hairpin_mode & 0x1) {
 		peer_tx_port = rte_eth_find_next_owned_by(pi + 1,
-						       RTE_ETH_DEV_NO_OWNER);
+							  RTE_ETH_DEV_NO_OWNER);
 		if (peer_tx_port >= RTE_MAX_ETHPORTS)
 			peer_tx_port = rte_eth_find_next_owned_by(0,
-						RTE_ETH_DEV_NO_OWNER);
+								  RTE_ETH_DEV_NO_OWNER);
 		if (p_pi != RTE_MAX_ETHPORTS) {
 			peer_rx_port = p_pi;
 		} else {
@@ -2725,69 +2814,60 @@ setup_hairpin_queues(portid_t pi, portid_t p_pi, uint16_t cnt_pi)
 			peer_rx_port = p_pi;
 		} else {
 			peer_rx_port = rte_eth_find_next_owned_by(pi + 1,
-						RTE_ETH_DEV_NO_OWNER);
+								  RTE_ETH_DEV_NO_OWNER);
 			if (peer_rx_port >= RTE_MAX_ETHPORTS)
 				peer_rx_port = pi;
 		}
 		peer_tx_port = peer_rx_port;
 		manual = 1;
 	}
+	diag = port_config_hairpin_txq(pi, peer_rx_port, nb_rxq, nb_txq,
+				       nb_hairpinq, manual);
+	if (diag)
+		return diag;
+	diag = port_config_hairpin_rxq(pi, peer_tx_port, nb_rxq, nb_txq,
+				       nb_hairpinq, manual);
+	if (diag)
+		return diag;
+	return 0;
+}
 
-	for (qi = nb_txq, i = 0; qi < nb_hairpinq + nb_txq; qi++) {
-		hairpin_conf.peers[0].port = peer_rx_port;
-		hairpin_conf.peers[0].queue = i + nb_rxq;
-		hairpin_conf.manual_bind = !!manual;
-		hairpin_conf.tx_explicit = !!tx_exp;
-		hairpin_conf.force_memory = !!tx_force_memory;
-		hairpin_conf.use_locked_device_memory = !!tx_locked_memory;
-		hairpin_conf.use_rte_memory = !!tx_rte_memory;
-		diag = rte_eth_tx_hairpin_queue_setup
-			(pi, qi, nb_txd, &hairpin_conf);
-		i++;
-		if (diag == 0)
-			continue;
-
-		/* Fail to setup rx queue, return */
-		if (port->port_status == RTE_PORT_HANDLING)
-			port->port_status = RTE_PORT_STOPPED;
-		else
-			fprintf(stderr,
-				"Port %d can not be set back to stopped\n", pi);
-		fprintf(stderr, "Fail to configure port %d hairpin queues\n",
-			pi);
-		/* try to reconfigure queues next time */
-		port->need_reconfig_queues = 1;
-		return -1;
-	}
-	for (qi = nb_rxq, i = 0; qi < nb_hairpinq + nb_rxq; qi++) {
-		hairpin_conf.peers[0].port = peer_tx_port;
-		hairpin_conf.peers[0].queue = i + nb_txq;
-		hairpin_conf.manual_bind = !!manual;
-		hairpin_conf.tx_explicit = !!tx_exp;
-		hairpin_conf.force_memory = !!rx_force_memory;
-		hairpin_conf.use_locked_device_memory = !!rx_locked_memory;
-		hairpin_conf.use_rte_memory = !!rx_rte_memory;
-		diag = rte_eth_rx_hairpin_queue_setup
-			(pi, qi, nb_rxd, &hairpin_conf);
-		i++;
-		if (diag == 0)
-			continue;
-
-		/* Fail to setup rx queue, return */
-		if (port->port_status == RTE_PORT_HANDLING)
-			port->port_status = RTE_PORT_STOPPED;
-		else
-			fprintf(stderr,
-				"Port %d can not be set back to stopped\n", pi);
-		fprintf(stderr, "Fail to configure port %d hairpin queues\n",
-			pi);
-		/* try to reconfigure queues next time */
-		port->need_reconfig_queues = 1;
-		return -1;
+static int
+setup_mapped_harpin_queues(portid_t pi)
+{
+	int ret = 0;
+	struct hairpin_map *map;
+
+	LIST_FOREACH(map, &hairpin_map_head, entry) {
+		if (map->rx_port == pi) {
+			ret = port_config_hairpin_rxq(pi, map->tx_port,
+						      map->rxq_head,
+						      map->txq_head,
+						      map->qnum, true);
+			if (ret)
+				return ret;
+		}
+		if (map->tx_port == pi) {
+			ret = port_config_hairpin_txq(pi, map->rx_port,
+						      map->rxq_head,
+						      map->txq_head,
+						      map->qnum, true);
+			if (ret)
+				return ret;
+		}
 	}
 	return 0;
 }
 
+/* Configure the Rx and Tx hairpin queues for the selected port. */
+static int
+setup_hairpin_queues(portid_t pi, portid_t p_pi, uint16_t cnt_pi)
+{
+	return !hairpin_multiport_mode ?
+	       setup_legacy_hairpin_queus(pi, p_pi, cnt_pi) :
+	       setup_mapped_harpin_queues(pi);
+}
+
 /* Configure the Rx with optional split. */
 int
 rx_queue_setup(uint16_t port_id, uint16_t rx_queue_id,
diff --git a/app/test-pmd/testpmd.h b/app/test-pmd/testpmd.h
index f1df6a8faf..208e8e9514 100644
--- a/app/test-pmd/testpmd.h
+++ b/app/test-pmd/testpmd.h
@@ -125,6 +125,24 @@ enum noisy_fwd_mode {
 	NOISY_FWD_MODE_MAX,
 };
 
+struct hairpin_map {
+	LIST_ENTRY(hairpin_map) entry; /**< List entry. */
+	portid_t rx_port; /**< Hairpin Rx port ID. */
+	portid_t tx_port; /**< Hairpin Tx port ID. */
+	uint16_t rxq_head; /**< Hairpin Rx queue head. */
+	uint16_t txq_head; /**< Hairpin Tx queue head. */
+	uint16_t qnum; /**< Hairpin queues number. */
+};
+
+/**
+ * Command line arguments parser sets `hairpin_multiport_mode` to True
+ * if explicit hairpin map configuration mode was used.
+ */
+extern bool hairpin_multiport_mode;
+
+/** Hairpin maps list. */
+extern void hairpin_add_multiport_map(struct hairpin_map *map);
+
 /**
  * The data structure associated with RX and TX packet burst statistics
  * that are recorded for each forwarding stream.
diff --git a/doc/guides/testpmd_app_ug/run_app.rst b/doc/guides/testpmd_app_ug/run_app.rst
index 6e9c552e76..a202c98b4c 100644
--- a/doc/guides/testpmd_app_ug/run_app.rst
+++ b/doc/guides/testpmd_app_ug/run_app.rst
@@ -566,6 +566,9 @@ The command line options are:
 
     The default value is 0. Hairpin will use single port mode and implicit Tx flow mode.
 
+*   ``--hairpin-map=Rx port id:Rx queue:Tx port id:Tx queue:queues number``
+
+    Set explicit hairpin configuration.
 
 Testpmd Multi-Process Command-line Options
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-- 
2.39.2

[PATCH v3] testpmd: add hairpin-map parameter

[Gregory Etelson]

Hairpin offloads packet forwarding between ports.
Packet is expected on Rx port <rp>, Rx queue <rq> and is delivered
to Tx port <tp> Tx queue <tq>.

Testpmd implements a static hairpin configuration scheme.
The scheme implicitly matches next valid port for given <rp> or <tp>.
That approach can be used in a single or double port setups only.

The new parameter allows explicit selection of Rx and Tx ports and
queues in hairpin configuration.
The new `hairpin-map` parameter is provided with 5 parameters,
separated by `:`

`--hairpin-map=Rx port id:Rx queue:Tx port id:Tx queue:queues number`

Testpmd operator can provide several `hairpin-map` parameters for
different hairpin maps.
Example:

dpdk-testpmd <EAL params> -- \
  <testpmd params> \
  --rxq=2 --txq=2 --hairpinq=2 --hairpin-mode=0x12 \
  --hairpin-map=0:2:1:2:1 \ # [1]
  --hairpin-map=0:3:2:2:3   # [2]

Hairpin map [1] binds Rx port 0, queue 2 with Tx port 1, queue 2.
Hairpin map [2] binds
  Rx port 0, queue 3 with Tx port 2, queue 2,
  Rx port 0, queue 4 with Tx port 2, queue 3,
  Rx port 0, queue 5 with Tx port 2, queue 4.

The new `hairpin-map` parameter is optional.
If omitted, testpmd will create "default" hairpin maps.

Signed-off-by: Gregory Etelson <getelson@nvidia.com>
---
v2: Fix Windows Server 2019 compilation failure.
v3: Update the patch comment.
    Fix typo.
---
 app/test-pmd/parameters.c             |  63 ++++++++
 app/test-pmd/testpmd.c                | 212 ++++++++++++++++++--------
 app/test-pmd/testpmd.h                |  18 +++
 doc/guides/testpmd_app_ug/run_app.rst |   3 +
 4 files changed, 230 insertions(+), 66 deletions(-)

diff --git a/app/test-pmd/parameters.c b/app/test-pmd/parameters.c
index a9ca58339d..c6bdfdf06f 100644
--- a/app/test-pmd/parameters.c
+++ b/app/test-pmd/parameters.c
@@ -206,6 +206,12 @@ usage(char* progname)
 	printf("  --hairpin-mode=0xXX: bitmask set the hairpin port mode.\n"
 	       "    0x10 - explicit Tx rule, 0x02 - hairpin ports paired\n"
 	       "    0x01 - hairpin ports loop, 0x00 - hairpin port self\n");
+	printf("  --hairpin-map=rxpi:rxq:txpi:txq:n: hairpin map.\n"
+	       "    rxpi - Rx port index.\n"
+	       "    rxq  - Rx queue.\n"
+	       "    txpi - Tx port index.\n"
+	       "    txq  - Tx queue.\n"
+	       "    n    - hairpin queues number.\n");
 }
 
 #ifdef RTE_LIB_CMDLINE
@@ -588,6 +594,55 @@ parse_link_speed(int n)
 	return speed;
 }
 
+static __rte_always_inline
+char *parse_hairpin_map_entry(char *input, char **next)
+{
+	char *tail = strchr(input, ':');
+
+	if (!tail)
+		return NULL;
+	tail[0] = '\0';
+	*next = tail + 1;
+	return input;
+}
+
+static int
+parse_hairpin_map(const char *hpmap)
+{
+	/*
+	 * Testpmd hairpin map format:
+	 * <Rx port id:First Rx queue:Tx port id:First Tx queue:queues number>
+	 */
+	char *head, *next = (char *)(uintptr_t)hpmap;
+	struct hairpin_map *map = calloc(1, sizeof(*map));
+
+	if (!map)
+		return -ENOMEM;
+
+	head = parse_hairpin_map_entry(next, &next);
+	if (!head)
+		goto err;
+	map->rx_port = atoi(head);
+	head = parse_hairpin_map_entry(next, &next);
+	if (!head)
+		goto err;
+	map->rxq_head = atoi(head);
+	head = parse_hairpin_map_entry(next, &next);
+	if (!head)
+		goto err;
+	map->tx_port = atoi(head);
+	head = parse_hairpin_map_entry(next, &next);
+	if (!head)
+		goto err;
+	map->txq_head = atoi(head);
+	map->qnum = atoi(next);
+	hairpin_add_multiport_map(map);
+	return 0;
+err:
+	free(map);
+	return -EINVAL;
+}
+
 void
 launch_args_parse(int argc, char** argv)
 {
@@ -663,6 +718,7 @@ launch_args_parse(int argc, char** argv)
 		{ "txd",			1, 0, 0 },
 		{ "hairpinq",			1, 0, 0 },
 		{ "hairpin-mode",		1, 0, 0 },
+		{ "hairpin-map",                1, 0, 0 },
 		{ "burst",			1, 0, 0 },
 		{ "flowgen-clones",		1, 0, 0 },
 		{ "flowgen-flows",		1, 0, 0 },
@@ -1111,6 +1167,13 @@ launch_args_parse(int argc, char** argv)
 				else
 					hairpin_mode = (uint32_t)n;
 			}
+			if (!strcmp(lgopts[opt_idx].name, "hairpin-map")) {
+				hairpin_multiport_mode = true;
+				ret = parse_hairpin_map(optarg);
+				if (ret)
+					rte_exit(EXIT_FAILURE, "invalid hairpin map\n");
+
+			}
 			if (!strcmp(lgopts[opt_idx].name, "burst")) {
 				n = atoi(optarg);
 				if (n == 0) {
diff --git a/app/test-pmd/testpmd.c b/app/test-pmd/testpmd.c
index 938ca035d4..e01be907f1 100644
--- a/app/test-pmd/testpmd.c
+++ b/app/test-pmd/testpmd.c
@@ -434,6 +434,16 @@ uint8_t clear_ptypes = true;
 /* Hairpin ports configuration mode. */
 uint32_t hairpin_mode;
 
+bool hairpin_multiport_mode = false;
+
+static LIST_HEAD(, hairpin_map) hairpin_map_head = LIST_HEAD_INITIALIZER();
+
+void
+hairpin_add_multiport_map(struct hairpin_map *map)
+{
+	LIST_INSERT_HEAD(&hairpin_map_head, map, entry);
+}
+
 /* Pretty printing of ethdev events */
 static const char * const eth_event_desc[] = {
 	[RTE_ETH_EVENT_UNKNOWN] = "unknown",
@@ -2677,28 +2687,107 @@ port_is_started(portid_t port_id)
 #define HAIRPIN_MODE_TX_LOCKED_MEMORY RTE_BIT32(16)
 #define HAIRPIN_MODE_TX_RTE_MEMORY RTE_BIT32(17)
 
-
-/* Configure the Rx and Tx hairpin queues for the selected port. */
 static int
-setup_hairpin_queues(portid_t pi, portid_t p_pi, uint16_t cnt_pi)
+port_config_hairpin_rxq(portid_t pi, uint16_t peer_tx_port,
+			queueid_t rxq_head, queueid_t txq_head,
+			uint16_t qcount, uint32_t manual_bind)
 {
-	queueid_t qi;
+	int diag;
+	queueid_t i, qi;
+	uint32_t tx_explicit = !!(hairpin_mode & 0x10);
+	uint32_t force_mem = !!(hairpin_mode & HAIRPIN_MODE_RX_FORCE_MEMORY);
+	uint32_t locked_mem = !!(hairpin_mode & HAIRPIN_MODE_RX_LOCKED_MEMORY);
+	uint32_t rte_mem = !!(hairpin_mode & HAIRPIN_MODE_RX_RTE_MEMORY);
+	struct rte_port *port = &ports[pi];
 	struct rte_eth_hairpin_conf hairpin_conf = {
 		.peer_count = 1,
 	};
-	int i;
+
+	for (qi = rxq_head, i = 0; qi < rxq_head + qcount; qi++) {
+		hairpin_conf.peers[0].port = peer_tx_port;
+		hairpin_conf.peers[0].queue = i + txq_head;
+		hairpin_conf.manual_bind = manual_bind;
+		hairpin_conf.tx_explicit = tx_explicit;
+		hairpin_conf.force_memory = force_mem;
+		hairpin_conf.use_locked_device_memory = locked_mem;
+		hairpin_conf.use_rte_memory = rte_mem;
+		diag = rte_eth_rx_hairpin_queue_setup
+			(pi, qi, nb_rxd, &hairpin_conf);
+		i++;
+		if (diag == 0)
+			continue;
+
+		/* Fail to setup rx queue, return */
+		if (port->port_status == RTE_PORT_HANDLING)
+			port->port_status = RTE_PORT_STOPPED;
+		else
+			fprintf(stderr,
+				"Port %d can not be set back to stopped\n", pi);
+		fprintf(stderr,
+			"Port %d failed to configure hairpin on rxq %u.\n"
+			"Peer port: %u peer txq: %u\n",
+			pi, qi, peer_tx_port, i);
+		/* try to reconfigure queues next time */
+		port->need_reconfig_queues = 1;
+		return -1;
+	}
+	return 0;
+}
+
+static int
+port_config_hairpin_txq(portid_t pi, uint16_t peer_rx_port,
+			queueid_t rxq_head, queueid_t txq_head,
+			uint16_t qcount, uint32_t manual_bind)
+{
 	int diag;
+	queueid_t i, qi;
+	uint32_t tx_explicit = !!(hairpin_mode & 0x10);
+	uint32_t force_mem = !!(hairpin_mode & HAIRPIN_MODE_TX_FORCE_MEMORY);
+	uint32_t locked_mem = !!(hairpin_mode & HAIRPIN_MODE_TX_LOCKED_MEMORY);
+	uint32_t rte_mem = !!(hairpin_mode & HAIRPIN_MODE_TX_RTE_MEMORY);
 	struct rte_port *port = &ports[pi];
+	struct rte_eth_hairpin_conf hairpin_conf = {
+		.peer_count = 1,
+	};
+
+	for (qi = txq_head, i = 0; qi < txq_head + qcount; qi++) {
+		hairpin_conf.peers[0].port = peer_rx_port;
+		hairpin_conf.peers[0].queue = i + rxq_head;
+		hairpin_conf.manual_bind = manual_bind;
+		hairpin_conf.tx_explicit = tx_explicit;
+		hairpin_conf.force_memory = force_mem;
+		hairpin_conf.use_locked_device_memory = locked_mem;
+		hairpin_conf.use_rte_memory = rte_mem;
+		diag = rte_eth_tx_hairpin_queue_setup
+			(pi, qi, nb_txd, &hairpin_conf);
+		i++;
+		if (diag == 0)
+			continue;
+
+		/* Fail to setup rx queue, return */
+		if (port->port_status == RTE_PORT_HANDLING)
+			port->port_status = RTE_PORT_STOPPED;
+		else
+			fprintf(stderr,
+				"Port %d can not be set back to stopped\n", pi);
+		fprintf(stderr,
+			"Port %d failed to configure hairpin on txq %u.\n"
+			"Peer port: %u peer rxq: %u\n",
+			pi, qi, peer_rx_port, i);
+		/* try to reconfigure queues next time */
+		port->need_reconfig_queues = 1;
+		return -1;
+	}
+	return 0;
+}
+
+static int
+setup_legacy_hairpin_queues(portid_t pi, portid_t p_pi, uint16_t cnt_pi)
+{
+	int diag;
 	uint16_t peer_rx_port = pi;
 	uint16_t peer_tx_port = pi;
 	uint32_t manual = 1;
-	uint32_t tx_exp = hairpin_mode & 0x10;
-	uint32_t rx_force_memory = hairpin_mode & HAIRPIN_MODE_RX_FORCE_MEMORY;
-	uint32_t rx_locked_memory = hairpin_mode & HAIRPIN_MODE_RX_LOCKED_MEMORY;
-	uint32_t rx_rte_memory = hairpin_mode & HAIRPIN_MODE_RX_RTE_MEMORY;
-	uint32_t tx_force_memory = hairpin_mode & HAIRPIN_MODE_TX_FORCE_MEMORY;
-	uint32_t tx_locked_memory = hairpin_mode & HAIRPIN_MODE_TX_LOCKED_MEMORY;
-	uint32_t tx_rte_memory = hairpin_mode & HAIRPIN_MODE_TX_RTE_MEMORY;
 
 	if (!(hairpin_mode & 0xf)) {
 		peer_rx_port = pi;
@@ -2706,10 +2795,10 @@ setup_hairpin_queues(portid_t pi, portid_t p_pi, uint16_t cnt_pi)
 		manual = 0;
 	} else if (hairpin_mode & 0x1) {
 		peer_tx_port = rte_eth_find_next_owned_by(pi + 1,
-						       RTE_ETH_DEV_NO_OWNER);
+							  RTE_ETH_DEV_NO_OWNER);
 		if (peer_tx_port >= RTE_MAX_ETHPORTS)
 			peer_tx_port = rte_eth_find_next_owned_by(0,
-						RTE_ETH_DEV_NO_OWNER);
+								  RTE_ETH_DEV_NO_OWNER);
 		if (p_pi != RTE_MAX_ETHPORTS) {
 			peer_rx_port = p_pi;
 		} else {
@@ -2725,69 +2814,60 @@ setup_hairpin_queues(portid_t pi, portid_t p_pi, uint16_t cnt_pi)
 			peer_rx_port = p_pi;
 		} else {
 			peer_rx_port = rte_eth_find_next_owned_by(pi + 1,
-						RTE_ETH_DEV_NO_OWNER);
+								  RTE_ETH_DEV_NO_OWNER);
 			if (peer_rx_port >= RTE_MAX_ETHPORTS)
 				peer_rx_port = pi;
 		}
 		peer_tx_port = peer_rx_port;
 		manual = 1;
 	}
+	diag = port_config_hairpin_txq(pi, peer_rx_port, nb_rxq, nb_txq,
+				       nb_hairpinq, manual);
+	if (diag)
+		return diag;
+	diag = port_config_hairpin_rxq(pi, peer_tx_port, nb_rxq, nb_txq,
+				       nb_hairpinq, manual);
+	if (diag)
+		return diag;
+	return 0;
+}
 
-	for (qi = nb_txq, i = 0; qi < nb_hairpinq + nb_txq; qi++) {
-		hairpin_conf.peers[0].port = peer_rx_port;
-		hairpin_conf.peers[0].queue = i + nb_rxq;
-		hairpin_conf.manual_bind = !!manual;
-		hairpin_conf.tx_explicit = !!tx_exp;
-		hairpin_conf.force_memory = !!tx_force_memory;
-		hairpin_conf.use_locked_device_memory = !!tx_locked_memory;
-		hairpin_conf.use_rte_memory = !!tx_rte_memory;
-		diag = rte_eth_tx_hairpin_queue_setup
-			(pi, qi, nb_txd, &hairpin_conf);
-		i++;
-		if (diag == 0)
-			continue;
-
-		/* Fail to setup rx queue, return */
-		if (port->port_status == RTE_PORT_HANDLING)
-			port->port_status = RTE_PORT_STOPPED;
-		else
-			fprintf(stderr,
-				"Port %d can not be set back to stopped\n", pi);
-		fprintf(stderr, "Fail to configure port %d hairpin queues\n",
-			pi);
-		/* try to reconfigure queues next time */
-		port->need_reconfig_queues = 1;
-		return -1;
-	}
-	for (qi = nb_rxq, i = 0; qi < nb_hairpinq + nb_rxq; qi++) {
-		hairpin_conf.peers[0].port = peer_tx_port;
-		hairpin_conf.peers[0].queue = i + nb_txq;
-		hairpin_conf.manual_bind = !!manual;
-		hairpin_conf.tx_explicit = !!tx_exp;
-		hairpin_conf.force_memory = !!rx_force_memory;
-		hairpin_conf.use_locked_device_memory = !!rx_locked_memory;
-		hairpin_conf.use_rte_memory = !!rx_rte_memory;
-		diag = rte_eth_rx_hairpin_queue_setup
-			(pi, qi, nb_rxd, &hairpin_conf);
-		i++;
-		if (diag == 0)
-			continue;
-
-		/* Fail to setup rx queue, return */
-		if (port->port_status == RTE_PORT_HANDLING)
-			port->port_status = RTE_PORT_STOPPED;
-		else
-			fprintf(stderr,
-				"Port %d can not be set back to stopped\n", pi);
-		fprintf(stderr, "Fail to configure port %d hairpin queues\n",
-			pi);
-		/* try to reconfigure queues next time */
-		port->need_reconfig_queues = 1;
-		return -1;
+static int
+setup_mapped_harpin_queues(portid_t pi)
+{
+	int ret = 0;
+	struct hairpin_map *map;
+
+	LIST_FOREACH(map, &hairpin_map_head, entry) {
+		if (map->rx_port == pi) {
+			ret = port_config_hairpin_rxq(pi, map->tx_port,
+						      map->rxq_head,
+						      map->txq_head,
+						      map->qnum, true);
+			if (ret)
+				return ret;
+		}
+		if (map->tx_port == pi) {
+			ret = port_config_hairpin_txq(pi, map->rx_port,
+						      map->rxq_head,
+						      map->txq_head,
+						      map->qnum, true);
+			if (ret)
+				return ret;
+		}
 	}
 	return 0;
 }
 
+/* Configure the Rx and Tx hairpin queues for the selected port. */
+static int
+setup_hairpin_queues(portid_t pi, portid_t p_pi, uint16_t cnt_pi)
+{
+	return !hairpin_multiport_mode ?
+	       setup_legacy_hairpin_queues(pi, p_pi, cnt_pi) :
+	       setup_mapped_harpin_queues(pi);
+}
+
 /* Configure the Rx with optional split. */
 int
 rx_queue_setup(uint16_t port_id, uint16_t rx_queue_id,
diff --git a/app/test-pmd/testpmd.h b/app/test-pmd/testpmd.h
index f1df6a8faf..208e8e9514 100644
--- a/app/test-pmd/testpmd.h
+++ b/app/test-pmd/testpmd.h
@@ -125,6 +125,24 @@ enum noisy_fwd_mode {
 	NOISY_FWD_MODE_MAX,
 };
 
+struct hairpin_map {
+	LIST_ENTRY(hairpin_map) entry; /**< List entry. */
+	portid_t rx_port; /**< Hairpin Rx port ID. */
+	portid_t tx_port; /**< Hairpin Tx port ID. */
+	uint16_t rxq_head; /**< Hairpin Rx queue head. */
+	uint16_t txq_head; /**< Hairpin Tx queue head. */
+	uint16_t qnum; /**< Hairpin queues number. */
+};
+
+/**
+ * Command line arguments parser sets `hairpin_multiport_mode` to True
+ * if explicit hairpin map configuration mode was used.
+ */
+extern bool hairpin_multiport_mode;
+
+/** Hairpin maps list. */
+extern void hairpin_add_multiport_map(struct hairpin_map *map);
+
 /**
  * The data structure associated with RX and TX packet burst statistics
  * that are recorded for each forwarding stream.
diff --git a/doc/guides/testpmd_app_ug/run_app.rst b/doc/guides/testpmd_app_ug/run_app.rst
index 6e9c552e76..a202c98b4c 100644
--- a/doc/guides/testpmd_app_ug/run_app.rst
+++ b/doc/guides/testpmd_app_ug/run_app.rst
@@ -566,6 +566,9 @@ The command line options are:
 
     The default value is 0. Hairpin will use single port mode and implicit Tx flow mode.
 
+*   ``--hairpin-map=Rx port id:Rx queue:Tx port id:Tx queue:queues number``
+
+    Set explicit hairpin configuration.
 
 Testpmd Multi-Process Command-line Options
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-- 
2.39.2

RE: [PATCH v3] testpmd: add hairpin-map parameter

[Dariusz Sosnowski]

> Subject: [PATCH v3] testpmd: add hairpin-map parameter
> 
> Hairpin offloads packet forwarding between ports.
> Packet is expected on Rx port <rp>, Rx queue <rq> and is delivered to Tx port
> <tp> Tx queue <tq>.
> 
> Testpmd implements a static hairpin configuration scheme.
> The scheme implicitly matches next valid port for given <rp> or <tp>.
> That approach can be used in a single or double port setups only.
> 
> The new parameter allows explicit selection of Rx and Tx ports and queues in
> hairpin configuration.
> The new `hairpin-map` parameter is provided with 5 parameters, separated by
> `:`
> 
> `--hairpin-map=Rx port id:Rx queue:Tx port id:Tx queue:queues number`
> 
> Testpmd operator can provide several `hairpin-map` parameters for different
> hairpin maps.
> Example:
> 
> dpdk-testpmd <EAL params> -- \
>   <testpmd params> \
>   --rxq=2 --txq=2 --hairpinq=2 --hairpin-mode=0x12 \
>   --hairpin-map=0:2:1:2:1 \ # [1]
>   --hairpin-map=0:3:2:2:3   # [2]
> 
> Hairpin map [1] binds Rx port 0, queue 2 with Tx port 1, queue 2.
> Hairpin map [2] binds
>   Rx port 0, queue 3 with Tx port 2, queue 2,
>   Rx port 0, queue 4 with Tx port 2, queue 3,
>   Rx port 0, queue 5 with Tx port 2, queue 4.
> 
> The new `hairpin-map` parameter is optional.
> If omitted, testpmd will create "default" hairpin maps.
> 
> Signed-off-by: Gregory Etelson <getelson@nvidia.com>
Acked-by: Dariusz Sosnowski <dsosnowski@nvidia.com>

Best regards,
Dariusz Sosnowski

Re: [PATCH v3] testpmd: add hairpin-map parameter

[Ferruh Yigit]

On 9/28/2023 4:36 PM, Gregory Etelson wrote:
> Hairpin offloads packet forwarding between ports.
> Packet is expected on Rx port <rp>, Rx queue <rq> and is delivered
> to Tx port <tp> Tx queue <tq>.
> 
> Testpmd implements a static hairpin configuration scheme.
> The scheme implicitly matches next valid port for given <rp> or <tp>.
> That approach can be used in a single or double port setups only.
> 
> The new parameter allows explicit selection of Rx and Tx ports and
> queues in hairpin configuration.
> The new `hairpin-map` parameter is provided with 5 parameters,
> separated by `:`
> 
> `--hairpin-map=Rx port id:Rx queue:Tx port id:Tx queue:queues number`
> 
> Testpmd operator can provide several `hairpin-map` parameters for
> different hairpin maps.
> Example:
> 
> dpdk-testpmd <EAL params> -- \
>   <testpmd params> \
>   --rxq=2 --txq=2 --hairpinq=2 --hairpin-mode=0x12 \
>   --hairpin-map=0:2:1:2:1 \ # [1]
>   --hairpin-map=0:3:2:2:3   # [2]
> 
> Hairpin map [1] binds Rx port 0, queue 2 with Tx port 1, queue 2.
> Hairpin map [2] binds
>   Rx port 0, queue 3 with Tx port 2, queue 2,
>   Rx port 0, queue 4 with Tx port 2, queue 3,
>   Rx port 0, queue 5 with Tx port 2, queue 4.
> 
> The new `hairpin-map` parameter is optional.
> If omitted, testpmd will create "default" hairpin maps.
> 

Hi Gregory,

This patch is waiting for a while in the patchwork, not because there is
a problem with the implementation, but I have a feeling that it is
bloating the testpmd code with the hairpin feature which is not commonly
supported, and I don't know what to do.

Also there was no review on the code ...


As a brainstorm, what do you think move hairpin related code to its own
file, and have a kind of register logic to the commands, argument
parsing and usage() functions etc.

@David did something similar to move PMD specific testpmd code to the
drivers [1].

This can be initial steps for the more modular testpmd design and we can
introduce more features with this approach in the future, I think
hairpin support is a good instance to start this work, because of its
limited impact to users.


[1]
Commit 592ab76f9f0f ("app/testpmd: register driver specific commands")
Commit 94b3c1a72507 ("net/i40e: move testpmd commands")

> Signed-off-by: Gregory Etelson <getelson@nvidia.com>
> ---
> v2: Fix Windows Server 2019 compilation failure.
> v3: Update the patch comment.
>     Fix typo.
> ---
>  app/test-pmd/parameters.c             |  63 ++++++++
>  app/test-pmd/testpmd.c                | 212 ++++++++++++++++++--------
>  app/test-pmd/testpmd.h                |  18 +++
>  doc/guides/testpmd_app_ug/run_app.rst |   3 +
>  4 files changed, 230 insertions(+), 66 deletions(-)
> 
> diff --git a/app/test-pmd/parameters.c b/app/test-pmd/parameters.c
> index a9ca58339d..c6bdfdf06f 100644
> --- a/app/test-pmd/parameters.c
> +++ b/app/test-pmd/parameters.c
> @@ -206,6 +206,12 @@ usage(char* progname)
>  	printf("  --hairpin-mode=0xXX: bitmask set the hairpin port mode.\n"
>  	       "    0x10 - explicit Tx rule, 0x02 - hairpin ports paired\n"
>  	       "    0x01 - hairpin ports loop, 0x00 - hairpin port self\n");
> +	printf("  --hairpin-map=rxpi:rxq:txpi:txq:n: hairpin map.\n"
> +	       "    rxpi - Rx port index.\n"
> +	       "    rxq  - Rx queue.\n"
> +	       "    txpi - Tx port index.\n"
> +	       "    txq  - Tx queue.\n"
> +	       "    n    - hairpin queues number.\n");
>

'rxpi' or 'port index' are not common usage in DPDK, what about
'rx_port' & 'Rx port id' or similar?

>  }
>  
>  #ifdef RTE_LIB_CMDLINE
> @@ -588,6 +594,55 @@ parse_link_speed(int n)
>  	return speed;
>  }
>  
> +static __rte_always_inline
> +char *parse_hairpin_map_entry(char *input, char **next)
> +{
>

Why forcing this function to be inline, can make it static and left
decision to compiler.

<...>

> diff --git a/doc/guides/testpmd_app_ug/run_app.rst b/doc/guides/testpmd_app_ug/run_app.rst
> index 6e9c552e76..a202c98b4c 100644
> --- a/doc/guides/testpmd_app_ug/run_app.rst
> +++ b/doc/guides/testpmd_app_ug/run_app.rst
> @@ -566,6 +566,9 @@ The command line options are:
>  
>      The default value is 0. Hairpin will use single port mode and implicit Tx flow mode.
>  
> +*   ``--hairpin-map=Rx port id:Rx queue:Tx port id:Tx queue:queues number``
> +
> +    Set explicit hairpin configuration.
>

What 'queues number' does is not intuitive, sample in the commit log
makes it easier to understand.
Can you please either explain or provide sample to clarify what "queues
number" is for?

Re: [PATCH v3] testpmd: add hairpin-map parameter

[Gregory Etelson]

[-- Attachment #1: Type: text/plain, Size: 5733 bytes --]

Hello Ferruh,

As a brainstorm, what do you think move hairpin related code to its own
file, and have a kind of register logic to the commands, argument
parsing and usage() functions etc.

Please consider an option for the testpmd to dynamically add supported commands.
Testpmd can still provide some build-in core functionality, but custom test commands
can be added and removed on demand.
In addition, the custom command code does not have to be part of the testpmd.
That will reduce maintanance.

Regards,
Gregory

________________________________
From: Ferruh Yigit <ferruh.yigit@amd.com>
Sent: Friday, March 1, 2024 20:54
To: Gregory Etelson <getelson@nvidia.com>; dev@dpdk.org <dev@dpdk.org>
Cc: Maayan Kashani <mkashani@nvidia.com>; NBU-Contact-Thomas Monjalon (EXTERNAL) <thomas@monjalon.net>; Aman Singh <aman.deep.singh@intel.com>; Yuying Zhang <yuying.zhang@intel.com>; David Marchand <david.marchand@redhat.com>
Subject: Re: [PATCH v3] testpmd: add hairpin-map parameter

External email: Use caution opening links or attachments


On 9/28/2023 4:36 PM, Gregory Etelson wrote:
> Hairpin offloads packet forwarding between ports.
> Packet is expected on Rx port <rp>, Rx queue <rq> and is delivered
> to Tx port <tp> Tx queue <tq>.
>
> Testpmd implements a static hairpin configuration scheme.
> The scheme implicitly matches next valid port for given <rp> or <tp>.
> That approach can be used in a single or double port setups only.
>
> The new parameter allows explicit selection of Rx and Tx ports and
> queues in hairpin configuration.
> The new `hairpin-map` parameter is provided with 5 parameters,
> separated by `:`
>
> `--hairpin-map=Rx port id:Rx queue:Tx port id:Tx queue:queues number`
>
> Testpmd operator can provide several `hairpin-map` parameters for
> different hairpin maps.
> Example:
>
> dpdk-testpmd <EAL params> -- \
>   <testpmd params> \
>   --rxq=2 --txq=2 --hairpinq=2 --hairpin-mode=0x12 \
>   --hairpin-map=0:2:1:2:1 \ # [1]
>   --hairpin-map=0:3:2:2:3   # [2]
>
> Hairpin map [1] binds Rx port 0, queue 2 with Tx port 1, queue 2.
> Hairpin map [2] binds
>   Rx port 0, queue 3 with Tx port 2, queue 2,
>   Rx port 0, queue 4 with Tx port 2, queue 3,
>   Rx port 0, queue 5 with Tx port 2, queue 4.
>
> The new `hairpin-map` parameter is optional.
> If omitted, testpmd will create "default" hairpin maps.
>

Hi Gregory,

This patch is waiting for a while in the patchwork, not because there is
a problem with the implementation, but I have a feeling that it is
bloating the testpmd code with the hairpin feature which is not commonly
supported, and I don't know what to do.

Also there was no review on the code ...


As a brainstorm, what do you think move hairpin related code to its own
file, and have a kind of register logic to the commands, argument
parsing and usage() functions etc.

@David did something similar to move PMD specific testpmd code to the
drivers [1].

This can be initial steps for the more modular testpmd design and we can
introduce more features with this approach in the future, I think
hairpin support is a good instance to start this work, because of its
limited impact to users.


[1]
Commit 592ab76f9f0f ("app/testpmd: register driver specific commands")
Commit 94b3c1a72507 ("net/i40e: move testpmd commands")

> Signed-off-by: Gregory Etelson <getelson@nvidia.com>
> ---
> v2: Fix Windows Server 2019 compilation failure.
> v3: Update the patch comment.
>     Fix typo.
> ---
>  app/test-pmd/parameters.c             |  63 ++++++++
>  app/test-pmd/testpmd.c                | 212 ++++++++++++++++++--------
>  app/test-pmd/testpmd.h                |  18 +++
>  doc/guides/testpmd_app_ug/run_app.rst |   3 +
>  4 files changed, 230 insertions(+), 66 deletions(-)
>
> diff --git a/app/test-pmd/parameters.c b/app/test-pmd/parameters.c
> index a9ca58339d..c6bdfdf06f 100644
> --- a/app/test-pmd/parameters.c
> +++ b/app/test-pmd/parameters.c
> @@ -206,6 +206,12 @@ usage(char* progname)
>       printf("  --hairpin-mode=0xXX: bitmask set the hairpin port mode.\n"
>              "    0x10 - explicit Tx rule, 0x02 - hairpin ports paired\n"
>              "    0x01 - hairpin ports loop, 0x00 - hairpin port self\n");
> +     printf("  --hairpin-map=rxpi:rxq:txpi:txq:n: hairpin map.\n"
> +            "    rxpi - Rx port index.\n"
> +            "    rxq  - Rx queue.\n"
> +            "    txpi - Tx port index.\n"
> +            "    txq  - Tx queue.\n"
> +            "    n    - hairpin queues number.\n");
>

'rxpi' or 'port index' are not common usage in DPDK, what about
'rx_port' & 'Rx port id' or similar?

>  }
>
>  #ifdef RTE_LIB_CMDLINE
> @@ -588,6 +594,55 @@ parse_link_speed(int n)
>       return speed;
>  }
>
> +static __rte_always_inline
> +char *parse_hairpin_map_entry(char *input, char **next)
> +{
>

Why forcing this function to be inline, can make it static and left
decision to compiler.

<...>

> diff --git a/doc/guides/testpmd_app_ug/run_app.rst b/doc/guides/testpmd_app_ug/run_app.rst
> index 6e9c552e76..a202c98b4c 100644
> --- a/doc/guides/testpmd_app_ug/run_app.rst
> +++ b/doc/guides/testpmd_app_ug/run_app.rst
> @@ -566,6 +566,9 @@ The command line options are:
>
>      The default value is 0. Hairpin will use single port mode and implicit Tx flow mode.
>
> +*   ``--hairpin-map=Rx port id:Rx queue:Tx port id:Tx queue:queues number``
> +
> +    Set explicit hairpin configuration.
>

What 'queues number' does is not intuitive, sample in the commit log
makes it easier to understand.
Can you please either explain or provide sample to clarify what "queues
number" is for?


[-- Attachment #2: Type: text/html, Size: 11365 bytes --]

[PATCH v4 1/1] testpmd: add hairpin-map parameter

[Gregory Etelson]

Hairpin offloads packet forwarding between ports.
Packet is expected on Rx port <rp>, Rx queue <rq> and is forwarded
to Tx port <tp> Tx queue <tq>.

Testpmd implements a static hairpin configuration scheme.

The new parameter allows explicit selection of Rx and Tx ports and
queues in hairpin configuration.
The new `hairpin-map` parameter is provided with 5 parameters,
separated by `:`

`--hairpin-map=Rx port id:Rx queue:Tx port id:Tx queue:queues number`

Testpmd operator can provide several `hairpin-map` parameters for
different hairpin maps.
Example:

dpdk-testpmd <EAL params> -- \
  <testpmd params> \
  --rxq=2 --txq=2 --hairpinq=2 --hairpin-mode=0x12 \
  --hairpin-map=0:2:1:2:1 \ # [1]
  --hairpin-map=0:3:2:2:3   # [2]

Hairpin map [1] binds Rx port 0, queue 2 with Tx port 1, queue 2.
Hairpin map [2] binds
  Rx port 0, queue 3 with Tx port 2, queue 2,
  Rx port 0, queue 4 with Tx port 2, queue 3,
  Rx port 0, queue 5 with Tx port 2, queue 4.

The new `hairpin-map` parameter is optional.
If omitted, testpmd will create "default" hairpin maps.

Signed-off-by: Gregory Etelson <getelson@nvidia.com>
Acked-by: Dariusz Sosnowski <dsosnowski@nvidia.com>
---
 app/test-pmd/hairpin.c                | 385 ++++++++++++++++++++++++++
 app/test-pmd/meson.build              |   1 +
 app/test-pmd/parameters.c             |  10 +
 app/test-pmd/testpmd.c                | 217 +--------------
 app/test-pmd/testpmd.h                |  14 +
 doc/guides/testpmd_app_ug/run_app.rst |   3 +
 6 files changed, 416 insertions(+), 214 deletions(-)
 create mode 100644 app/test-pmd/hairpin.c

diff --git a/app/test-pmd/hairpin.c b/app/test-pmd/hairpin.c
new file mode 100644
index 0000000000..de54e8bac8
--- /dev/null
+++ b/app/test-pmd/hairpin.c
@@ -0,0 +1,385 @@
+/* SPDX-License-Identifier: BSD-3-Clause
+ * Copyright (c) 2022 NVIDIA Corporation & Affiliates
+ */
+
+#include <stdio.h>
+#include <string.h>
+#include <stdbool.h>
+#include <stdint.h>
+
+#include <sys/queue.h>
+
+#include "testpmd.h"
+
+/* Hairpin ports configuration mode. */
+uint32_t hairpin_mode;
+
+bool hairpin_multiport_mode = false;
+
+queueid_t nb_hairpinq; /**< Number of hairpin queues per port. */
+
+static LIST_HEAD(, hairpin_map) hairpin_map_head = LIST_HEAD_INITIALIZER();
+
+struct hairpin_map {
+	LIST_ENTRY(hairpin_map) entry; /**< List entry. */
+	portid_t rx_port; /**< Hairpin Rx port ID. */
+	portid_t tx_port; /**< Hairpin Tx port ID. */
+	uint16_t rxq_head; /**< Hairpin Rx queue head. */
+	uint16_t txq_head; /**< Hairpin Tx queue head. */
+	uint16_t qnum; /**< Hairpin queues number. */
+};
+
+void
+hairpin_add_multiport_map(struct hairpin_map *map)
+{
+	LIST_INSERT_HEAD(&hairpin_map_head, map, entry);
+}
+
+/*
+ * Get the allowed maximum number of hairpin queues.
+ * *pid return the port id which has minimal value of
+ * max_hairpin_queues in all ports.
+ */
+queueid_t
+get_allowed_max_nb_hairpinq(portid_t *pid)
+{
+	queueid_t allowed_max_hairpinq = RTE_MAX_QUEUES_PER_PORT;
+	portid_t pi;
+	struct rte_eth_hairpin_cap cap;
+
+	RTE_ETH_FOREACH_DEV(pi) {
+		if (rte_eth_dev_hairpin_capability_get(pi, &cap) != 0) {
+			*pid = pi;
+			return 0;
+		}
+		if (cap.max_nb_queues < allowed_max_hairpinq) {
+			allowed_max_hairpinq = cap.max_nb_queues;
+			*pid = pi;
+		}
+	}
+	return allowed_max_hairpinq;
+}
+
+/*
+ * Check input hairpin is valid or not.
+ * If input hairpin is not greater than any of maximum number
+ * of hairpin queues of all ports, it is valid.
+ * if valid, return 0, else return -1
+ */
+int
+check_nb_hairpinq(queueid_t hairpinq)
+{
+	queueid_t allowed_max_hairpinq;
+	portid_t pid = 0;
+
+	allowed_max_hairpinq = get_allowed_max_nb_hairpinq(&pid);
+	if (hairpinq > allowed_max_hairpinq) {
+		fprintf(stderr,
+			"Fail: input hairpin (%u) can't be greater than max_hairpin_queues (%u) of port %u\n",
+			hairpinq, allowed_max_hairpinq, pid);
+		return -1;
+	}
+	return 0;
+}
+
+#define HAIRPIN_MODE_RX_FORCE_MEMORY RTE_BIT32(8)
+#define HAIRPIN_MODE_TX_FORCE_MEMORY RTE_BIT32(9)
+
+#define HAIRPIN_MODE_RX_LOCKED_MEMORY RTE_BIT32(12)
+#define HAIRPIN_MODE_RX_RTE_MEMORY RTE_BIT32(13)
+
+#define HAIRPIN_MODE_TX_LOCKED_MEMORY RTE_BIT32(16)
+#define HAIRPIN_MODE_TX_RTE_MEMORY RTE_BIT32(17)
+
+static int
+port_config_hairpin_rxq(portid_t pi, uint16_t peer_tx_port,
+			queueid_t rxq_head, queueid_t txq_head,
+			uint16_t qcount, uint32_t manual_bind)
+{
+	int diag;
+	queueid_t i, qi;
+	uint32_t tx_explicit = !!(hairpin_mode & 0x10);
+	uint32_t force_mem = !!(hairpin_mode & HAIRPIN_MODE_RX_FORCE_MEMORY);
+	uint32_t locked_mem = !!(hairpin_mode & HAIRPIN_MODE_RX_LOCKED_MEMORY);
+	uint32_t rte_mem = !!(hairpin_mode & HAIRPIN_MODE_RX_RTE_MEMORY);
+	struct rte_port *port = &ports[pi];
+	struct rte_eth_hairpin_conf hairpin_conf = {
+		.peer_count = 1,
+	};
+
+	for (qi = rxq_head, i = 0; qi < rxq_head + qcount; qi++) {
+		hairpin_conf.peers[0].port = peer_tx_port;
+		hairpin_conf.peers[0].queue = i + txq_head;
+		hairpin_conf.manual_bind = manual_bind;
+		hairpin_conf.tx_explicit = tx_explicit;
+		hairpin_conf.force_memory = force_mem;
+		hairpin_conf.use_locked_device_memory = locked_mem;
+		hairpin_conf.use_rte_memory = rte_mem;
+		diag = rte_eth_rx_hairpin_queue_setup
+			(pi, qi, nb_rxd, &hairpin_conf);
+		i++;
+		if (diag == 0)
+			continue;
+
+		/* Fail to setup rx queue, return */
+		if (port->port_status == RTE_PORT_HANDLING)
+			port->port_status = RTE_PORT_STOPPED;
+		else
+			fprintf(stderr,
+				"Port %d can not be set back to stopped\n", pi);
+		fprintf(stderr,
+			"Port %d failed to configure hairpin on rxq %u.\n"
+			"Peer port: %u peer txq: %u\n",
+			pi, qi, peer_tx_port, i);
+		/* try to reconfigure queues next time */
+		port->need_reconfig_queues = 1;
+		return -1;
+	}
+	return 0;
+}
+
+static int
+port_config_hairpin_txq(portid_t pi, uint16_t peer_rx_port,
+			queueid_t rxq_head, queueid_t txq_head,
+			uint16_t qcount, uint32_t manual_bind)
+{
+	int diag;
+	queueid_t i, qi;
+	uint32_t tx_explicit = !!(hairpin_mode & 0x10);
+	uint32_t force_mem = !!(hairpin_mode & HAIRPIN_MODE_TX_FORCE_MEMORY);
+	uint32_t locked_mem = !!(hairpin_mode & HAIRPIN_MODE_TX_LOCKED_MEMORY);
+	uint32_t rte_mem = !!(hairpin_mode & HAIRPIN_MODE_TX_RTE_MEMORY);
+	struct rte_port *port = &ports[pi];
+	struct rte_eth_hairpin_conf hairpin_conf = {
+		.peer_count = 1,
+	};
+
+	for (qi = txq_head, i = 0; qi < txq_head + qcount; qi++) {
+		hairpin_conf.peers[0].port = peer_rx_port;
+		hairpin_conf.peers[0].queue = i + rxq_head;
+		hairpin_conf.manual_bind = manual_bind;
+		hairpin_conf.tx_explicit = tx_explicit;
+		hairpin_conf.force_memory = force_mem;
+		hairpin_conf.use_locked_device_memory = locked_mem;
+		hairpin_conf.use_rte_memory = rte_mem;
+		diag = rte_eth_tx_hairpin_queue_setup
+			(pi, qi, nb_txd, &hairpin_conf);
+		i++;
+		if (diag == 0)
+			continue;
+
+		/* Fail to setup rx queue, return */
+		if (port->port_status == RTE_PORT_HANDLING)
+			port->port_status = RTE_PORT_STOPPED;
+		else
+			fprintf(stderr,
+				"Port %d can not be set back to stopped\n", pi);
+		fprintf(stderr,
+			"Port %d failed to configure hairpin on txq %u.\n"
+			"Peer port: %u peer rxq: %u\n",
+			pi, qi, peer_rx_port, i);
+		/* try to reconfigure queues next time */
+		port->need_reconfig_queues = 1;
+		return -1;
+	}
+	return 0;
+}
+
+static int
+setup_legacy_hairpin_queues(portid_t pi, portid_t p_pi, uint16_t cnt_pi)
+{
+	int diag;
+	uint16_t peer_rx_port = pi;
+	uint16_t peer_tx_port = pi;
+	uint32_t manual = 1;
+
+	if (!(hairpin_mode & 0xf)) {
+		peer_rx_port = pi;
+		peer_tx_port = pi;
+		manual = 0;
+	} else if (hairpin_mode & 0x1) {
+		peer_tx_port = rte_eth_find_next_owned_by(pi + 1,
+							  RTE_ETH_DEV_NO_OWNER);
+		if (peer_tx_port >= RTE_MAX_ETHPORTS)
+			peer_tx_port = rte_eth_find_next_owned_by(0,
+								  RTE_ETH_DEV_NO_OWNER);
+		if (p_pi != RTE_MAX_ETHPORTS) {
+			peer_rx_port = p_pi;
+		} else {
+			uint16_t next_pi;
+
+			/* Last port will be the peer RX port of the first. */
+			RTE_ETH_FOREACH_DEV(next_pi)
+				peer_rx_port = next_pi;
+		}
+		manual = 1;
+	} else if (hairpin_mode & 0x2) {
+		if (cnt_pi & 0x1) {
+			peer_rx_port = p_pi;
+		} else {
+			peer_rx_port = rte_eth_find_next_owned_by(pi + 1,
+								  RTE_ETH_DEV_NO_OWNER);
+			if (peer_rx_port >= RTE_MAX_ETHPORTS)
+				peer_rx_port = pi;
+		}
+		peer_tx_port = peer_rx_port;
+		manual = 1;
+	}
+	diag = port_config_hairpin_txq(pi, peer_rx_port, nb_rxq, nb_txq,
+				       nb_hairpinq, manual);
+	if (diag)
+		return diag;
+	diag = port_config_hairpin_rxq(pi, peer_tx_port, nb_rxq, nb_txq,
+				       nb_hairpinq, manual);
+	if (diag)
+		return diag;
+	return 0;
+}
+
+static int
+setup_mapped_harpin_queues(portid_t pi)
+{
+	int ret = 0;
+	struct hairpin_map *map;
+
+	LIST_FOREACH(map, &hairpin_map_head, entry) {
+		if (map->rx_port == pi) {
+			ret = port_config_hairpin_rxq(pi, map->tx_port,
+						      map->rxq_head,
+						      map->txq_head,
+						      map->qnum, true);
+			if (ret)
+				return ret;
+		}
+		if (map->tx_port == pi) {
+			ret = port_config_hairpin_txq(pi, map->rx_port,
+						      map->rxq_head,
+						      map->txq_head,
+						      map->qnum, true);
+			if (ret)
+				return ret;
+		}
+	}
+	return 0;
+}
+
+/* Configure the Rx and Tx hairpin queues for the selected port. */
+int
+setup_hairpin_queues(portid_t pi, portid_t p_pi, uint16_t cnt_pi)
+{
+	return !hairpin_multiport_mode ?
+	       setup_legacy_hairpin_queues(pi, p_pi, cnt_pi) :
+	       setup_mapped_harpin_queues(pi);
+}
+
+int
+hairpin_bind(uint16_t cfg_pi, portid_t *pl, portid_t *peer_pl)
+{
+	uint16_t i;
+	portid_t pi;
+	int peer_pi;
+	int diag;
+	int j;
+
+	/* bind all started hairpin ports */
+	for (i = 0; i < cfg_pi; i++) {
+		pi = pl[i];
+		/* bind current Tx to all peer Rx */
+		peer_pi = rte_eth_hairpin_get_peer_ports(pi, peer_pl,
+							 RTE_MAX_ETHPORTS, 1);
+		if (peer_pi < 0)
+			return peer_pi;
+		for (j = 0; j < peer_pi; j++) {
+			if (!port_is_started(peer_pl[j]))
+				continue;
+			diag = rte_eth_hairpin_bind(pi, peer_pl[j]);
+			if (diag < 0) {
+				fprintf(stderr,
+					"Error during binding hairpin Tx port %u to %u: %s\n",
+					pi, peer_pl[j],
+					rte_strerror(-diag));
+				return -1;
+			}
+		}
+		/* bind all peer Tx to current Rx */
+		peer_pi = rte_eth_hairpin_get_peer_ports(pi, peer_pl,
+							 RTE_MAX_ETHPORTS, 0);
+		if (peer_pi < 0)
+			return peer_pi;
+		for (j = 0; j < peer_pi; j++) {
+			if (!port_is_started(peer_pl[j]))
+				continue;
+			diag = rte_eth_hairpin_bind(peer_pl[j], pi);
+			if (diag < 0) {
+				fprintf(stderr,
+					"Error during binding hairpin Tx port %u to %u: %s\n",
+					peer_pl[j], pi,
+					rte_strerror(-diag));
+				return -1;
+			}
+		}
+	}
+	return 0;
+}
+
+void
+hairpin_map_usage(void)
+{
+	printf("  --hairpin-map=rxpi:rxq:txpi:txq:n: hairpin map.\n"
+	       "    rxpi - Rx port index.\n"
+	       "    rxq  - Rx queue.\n"
+	       "    txpi - Tx port index.\n"
+	       "    txq  - Tx queue.\n"
+	       "    n    - hairpin queues number.\n");
+}
+
+static char
+*parse_hairpin_map_entry(char *input, char **next)
+{
+	char *tail = strchr(input, ':');
+
+	if (!tail)
+		return NULL;
+	tail[0] = '\0';
+	*next = tail + 1;
+	return input;
+}
+
+int
+parse_hairpin_map(const char *hpmap)
+{
+	/*
+	 * Testpmd hairpin map format:
+	 * <Rx port id:First Rx queue:Tx port id:First Tx queue:queues number>
+	 */
+	char *head, *next = (char *)(uintptr_t)hpmap;
+	struct hairpin_map *map = calloc(1, sizeof(*map));
+
+	if (!map)
+		return -ENOMEM;
+
+	head = parse_hairpin_map_entry(next, &next);
+	if (!head)
+		goto err;
+	map->rx_port = atoi(head);
+	head = parse_hairpin_map_entry(next, &next);
+	if (!head)
+		goto err;
+	map->rxq_head = atoi(head);
+	head = parse_hairpin_map_entry(next, &next);
+	if (!head)
+		goto err;
+	map->tx_port = atoi(head);
+	head = parse_hairpin_map_entry(next, &next);
+	if (!head)
+		goto err;
+	map->txq_head = atoi(head);
+	map->qnum = atoi(next);
+	hairpin_add_multiport_map(map);
+	return 0;
+err:
+	free(map);
+	return -EINVAL;
+}
+
+
diff --git a/app/test-pmd/meson.build b/app/test-pmd/meson.build
index 719f875be0..f1c36529b4 100644
--- a/app/test-pmd/meson.build
+++ b/app/test-pmd/meson.build
@@ -15,6 +15,7 @@ sources = files(
         'config.c',
         'csumonly.c',
         'flowgen.c',
+        'hairpin.c',
         'icmpecho.c',
         'ieee1588fwd.c',
         'iofwd.c',
diff --git a/app/test-pmd/parameters.c b/app/test-pmd/parameters.c
index 22364e09ab..7b31b94542 100644
--- a/app/test-pmd/parameters.c
+++ b/app/test-pmd/parameters.c
@@ -143,6 +143,8 @@ enum {
 	TESTPMD_OPT_HAIRPINQ_NUM,
 #define TESTPMD_OPT_HAIRPIN_MODE "hairpin-mode"
 	TESTPMD_OPT_HAIRPIN_MODE_NUM,
+#define TESTPMD_OPT_HAIRPIN_MAP "hairpin-map"
+	TESTPMD_OPT_HAIRPIN_MAP_NUM,
 #define TESTPMD_OPT_BURST "burst"
 	TESTPMD_OPT_BURST_NUM,
 #define TESTPMD_OPT_FLOWGEN_CLONES "flowgen-clones"
@@ -317,6 +319,7 @@ static const struct option long_options[] = {
 	REQUIRED_ARG(TESTPMD_OPT_TXD),
 	REQUIRED_ARG(TESTPMD_OPT_HAIRPINQ),
 	REQUIRED_ARG(TESTPMD_OPT_HAIRPIN_MODE),
+	REQUIRED_ARG(TESTPMD_OPT_HAIRPIN_MAP),
 	REQUIRED_ARG(TESTPMD_OPT_BURST),
 	REQUIRED_ARG(TESTPMD_OPT_FLOWGEN_CLONES),
 	REQUIRED_ARG(TESTPMD_OPT_FLOWGEN_FLOWS),
@@ -542,6 +545,7 @@ usage(char* progname)
 	printf("  --hairpin-mode=0xXX: bitmask set the hairpin port mode.\n"
 	       "    0x10 - explicit Tx rule, 0x02 - hairpin ports paired\n"
 	       "    0x01 - hairpin ports loop, 0x00 - hairpin port self\n");
+	hairpin_map_usage();
 }
 
 static int
@@ -1317,6 +1321,12 @@ launch_args_parse(int argc, char** argv)
 				hairpin_mode = (uint32_t)n;
 			break;
 		}
+		case TESTPMD_OPT_HAIRPIN_MAP_NUM:
+			hairpin_multiport_mode = true;
+			ret = parse_hairpin_map(optarg);
+			if (ret)
+				rte_exit(EXIT_FAILURE, "invalid hairpin map\n");
+			break;
 		case TESTPMD_OPT_BURST_NUM:
 			n = atoi(optarg);
 			if (n == 0) {
diff --git a/app/test-pmd/testpmd.c b/app/test-pmd/testpmd.c
index b1401136e4..f487769578 100644
--- a/app/test-pmd/testpmd.c
+++ b/app/test-pmd/testpmd.c
@@ -284,7 +284,6 @@ uint8_t dcb_config = 0;
 /*
  * Configurable number of RX/TX queues.
  */
-queueid_t nb_hairpinq; /**< Number of hairpin queues per port. */
 queueid_t nb_rxq = 1; /**< Number of RX queues per port. */
 queueid_t nb_txq = 1; /**< Number of TX queues per port. */
 
@@ -431,9 +430,6 @@ bool setup_on_probe_event = true;
 /* Clear ptypes on port initialization. */
 uint8_t clear_ptypes = true;
 
-/* Hairpin ports configuration mode. */
-uint32_t hairpin_mode;
-
 /* Pretty printing of ethdev events */
 static const char * const eth_event_desc[] = {
 	[RTE_ETH_EVENT_UNKNOWN] = "unknown",
@@ -1555,54 +1551,6 @@ check_nb_txd(queueid_t txd)
 	return 0;
 }
 
-
-/*
- * Get the allowed maximum number of hairpin queues.
- * *pid return the port id which has minimal value of
- * max_hairpin_queues in all ports.
- */
-queueid_t
-get_allowed_max_nb_hairpinq(portid_t *pid)
-{
-	queueid_t allowed_max_hairpinq = RTE_MAX_QUEUES_PER_PORT;
-	portid_t pi;
-	struct rte_eth_hairpin_cap cap;
-
-	RTE_ETH_FOREACH_DEV(pi) {
-		if (rte_eth_dev_hairpin_capability_get(pi, &cap) != 0) {
-			*pid = pi;
-			return 0;
-		}
-		if (cap.max_nb_queues < allowed_max_hairpinq) {
-			allowed_max_hairpinq = cap.max_nb_queues;
-			*pid = pi;
-		}
-	}
-	return allowed_max_hairpinq;
-}
-
-/*
- * Check input hairpin is valid or not.
- * If input hairpin is not greater than any of maximum number
- * of hairpin queues of all ports, it is valid.
- * if valid, return 0, else return -1
- */
-int
-check_nb_hairpinq(queueid_t hairpinq)
-{
-	queueid_t allowed_max_hairpinq;
-	portid_t pid = 0;
-
-	allowed_max_hairpinq = get_allowed_max_nb_hairpinq(&pid);
-	if (hairpinq > allowed_max_hairpinq) {
-		fprintf(stderr,
-			"Fail: input hairpin (%u) can't be greater than max_hairpin_queues (%u) of port %u\n",
-			hairpinq, allowed_max_hairpinq, pid);
-		return -1;
-	}
-	return 0;
-}
-
 static int
 get_eth_overhead(struct rte_eth_dev_info *dev_info)
 {
@@ -2684,126 +2632,6 @@ port_is_started(portid_t port_id)
 	return 1;
 }
 
-#define HAIRPIN_MODE_RX_FORCE_MEMORY RTE_BIT32(8)
-#define HAIRPIN_MODE_TX_FORCE_MEMORY RTE_BIT32(9)
-
-#define HAIRPIN_MODE_RX_LOCKED_MEMORY RTE_BIT32(12)
-#define HAIRPIN_MODE_RX_RTE_MEMORY RTE_BIT32(13)
-
-#define HAIRPIN_MODE_TX_LOCKED_MEMORY RTE_BIT32(16)
-#define HAIRPIN_MODE_TX_RTE_MEMORY RTE_BIT32(17)
-
-
-/* Configure the Rx and Tx hairpin queues for the selected port. */
-static int
-setup_hairpin_queues(portid_t pi, portid_t p_pi, uint16_t cnt_pi)
-{
-	queueid_t qi;
-	struct rte_eth_hairpin_conf hairpin_conf = {
-		.peer_count = 1,
-	};
-	int i;
-	int diag;
-	struct rte_port *port = &ports[pi];
-	uint16_t peer_rx_port = pi;
-	uint16_t peer_tx_port = pi;
-	uint32_t manual = 1;
-	uint32_t tx_exp = hairpin_mode & 0x10;
-	uint32_t rx_force_memory = hairpin_mode & HAIRPIN_MODE_RX_FORCE_MEMORY;
-	uint32_t rx_locked_memory = hairpin_mode & HAIRPIN_MODE_RX_LOCKED_MEMORY;
-	uint32_t rx_rte_memory = hairpin_mode & HAIRPIN_MODE_RX_RTE_MEMORY;
-	uint32_t tx_force_memory = hairpin_mode & HAIRPIN_MODE_TX_FORCE_MEMORY;
-	uint32_t tx_locked_memory = hairpin_mode & HAIRPIN_MODE_TX_LOCKED_MEMORY;
-	uint32_t tx_rte_memory = hairpin_mode & HAIRPIN_MODE_TX_RTE_MEMORY;
-
-	if (!(hairpin_mode & 0xf)) {
-		peer_rx_port = pi;
-		peer_tx_port = pi;
-		manual = 0;
-	} else if (hairpin_mode & 0x1) {
-		peer_tx_port = rte_eth_find_next_owned_by(pi + 1,
-						       RTE_ETH_DEV_NO_OWNER);
-		if (peer_tx_port >= RTE_MAX_ETHPORTS)
-			peer_tx_port = rte_eth_find_next_owned_by(0,
-						RTE_ETH_DEV_NO_OWNER);
-		if (p_pi != RTE_MAX_ETHPORTS) {
-			peer_rx_port = p_pi;
-		} else {
-			uint16_t next_pi;
-
-			/* Last port will be the peer RX port of the first. */
-			RTE_ETH_FOREACH_DEV(next_pi)
-				peer_rx_port = next_pi;
-		}
-		manual = 1;
-	} else if (hairpin_mode & 0x2) {
-		if (cnt_pi & 0x1) {
-			peer_rx_port = p_pi;
-		} else {
-			peer_rx_port = rte_eth_find_next_owned_by(pi + 1,
-						RTE_ETH_DEV_NO_OWNER);
-			if (peer_rx_port >= RTE_MAX_ETHPORTS)
-				peer_rx_port = pi;
-		}
-		peer_tx_port = peer_rx_port;
-		manual = 1;
-	}
-
-	for (qi = nb_txq, i = 0; qi < nb_hairpinq + nb_txq; qi++) {
-		hairpin_conf.peers[0].port = peer_rx_port;
-		hairpin_conf.peers[0].queue = i + nb_rxq;
-		hairpin_conf.manual_bind = !!manual;
-		hairpin_conf.tx_explicit = !!tx_exp;
-		hairpin_conf.force_memory = !!tx_force_memory;
-		hairpin_conf.use_locked_device_memory = !!tx_locked_memory;
-		hairpin_conf.use_rte_memory = !!tx_rte_memory;
-		diag = rte_eth_tx_hairpin_queue_setup
-			(pi, qi, nb_txd, &hairpin_conf);
-		i++;
-		if (diag == 0)
-			continue;
-
-		/* Fail to setup rx queue, return */
-		if (port->port_status == RTE_PORT_HANDLING)
-			port->port_status = RTE_PORT_STOPPED;
-		else
-			fprintf(stderr,
-				"Port %d can not be set back to stopped\n", pi);
-		fprintf(stderr, "Fail to configure port %d hairpin queues\n",
-			pi);
-		/* try to reconfigure queues next time */
-		port->need_reconfig_queues = 1;
-		return -1;
-	}
-	for (qi = nb_rxq, i = 0; qi < nb_hairpinq + nb_rxq; qi++) {
-		hairpin_conf.peers[0].port = peer_tx_port;
-		hairpin_conf.peers[0].queue = i + nb_txq;
-		hairpin_conf.manual_bind = !!manual;
-		hairpin_conf.tx_explicit = !!tx_exp;
-		hairpin_conf.force_memory = !!rx_force_memory;
-		hairpin_conf.use_locked_device_memory = !!rx_locked_memory;
-		hairpin_conf.use_rte_memory = !!rx_rte_memory;
-		diag = rte_eth_rx_hairpin_queue_setup
-			(pi, qi, nb_rxd, &hairpin_conf);
-		i++;
-		if (diag == 0)
-			continue;
-
-		/* Fail to setup rx queue, return */
-		if (port->port_status == RTE_PORT_HANDLING)
-			port->port_status = RTE_PORT_STOPPED;
-		else
-			fprintf(stderr,
-				"Port %d can not be set back to stopped\n", pi);
-		fprintf(stderr, "Fail to configure port %d hairpin queues\n",
-			pi);
-		/* try to reconfigure queues next time */
-		port->need_reconfig_queues = 1;
-		return -1;
-	}
-	return 0;
-}
-
 /* Configure the Rx with optional split. */
 int
 rx_queue_setup(uint16_t port_id, uint16_t rx_queue_id,
@@ -3043,7 +2871,6 @@ start_port(portid_t pid)
 	portid_t peer_pl[RTE_MAX_ETHPORTS];
 	uint16_t cnt_pi = 0;
 	uint16_t cfg_pi = 0;
-	int peer_pi;
 	queueid_t qi;
 	struct rte_port *port;
 	struct rte_eth_hairpin_cap cap;
@@ -3304,47 +3131,9 @@ start_port(portid_t pid)
 		fprintf(stderr, "Please stop the ports first\n");
 
 	if (hairpin_mode & 0xf) {
-		uint16_t i;
-		int j;
-
-		/* bind all started hairpin ports */
-		for (i = 0; i < cfg_pi; i++) {
-			pi = pl[i];
-			/* bind current Tx to all peer Rx */
-			peer_pi = rte_eth_hairpin_get_peer_ports(pi, peer_pl,
-							RTE_MAX_ETHPORTS, 1);
-			if (peer_pi < 0)
-				return peer_pi;
-			for (j = 0; j < peer_pi; j++) {
-				if (!port_is_started(peer_pl[j]))
-					continue;
-				diag = rte_eth_hairpin_bind(pi, peer_pl[j]);
-				if (diag < 0) {
-					fprintf(stderr,
-						"Error during binding hairpin Tx port %u to %u: %s\n",
-						pi, peer_pl[j],
-						rte_strerror(-diag));
-					return -1;
-				}
-			}
-			/* bind all peer Tx to current Rx */
-			peer_pi = rte_eth_hairpin_get_peer_ports(pi, peer_pl,
-							RTE_MAX_ETHPORTS, 0);
-			if (peer_pi < 0)
-				return peer_pi;
-			for (j = 0; j < peer_pi; j++) {
-				if (!port_is_started(peer_pl[j]))
-					continue;
-				diag = rte_eth_hairpin_bind(peer_pl[j], pi);
-				if (diag < 0) {
-					fprintf(stderr,
-						"Error during binding hairpin Tx port %u to %u: %s\n",
-						peer_pl[j], pi,
-						rte_strerror(-diag));
-					return -1;
-				}
-			}
-		}
+		diag = hairpin_bind(cfg_pi, pl, peer_pl);
+		if (diag < 0)
+			return -1;
 	}
 
 	fill_xstats_display_info_for_port(pid);
diff --git a/app/test-pmd/testpmd.h b/app/test-pmd/testpmd.h
index 9facd7f281..659c5b4fdc 100644
--- a/app/test-pmd/testpmd.h
+++ b/app/test-pmd/testpmd.h
@@ -126,6 +126,16 @@ enum noisy_fwd_mode {
 	NOISY_FWD_MODE_MAX,
 };
 
+/**
+ * Command line arguments parser sets `hairpin_multiport_mode` to True
+ * if explicit hairpin map configuration mode was used.
+ */
+extern bool hairpin_multiport_mode;
+
+/** Hairpin maps list. */
+struct hairpin_map;
+extern void hairpin_add_multiport_map(struct hairpin_map *map);
+
 /**
  * The data structure associated with RX and TX packet burst statistics
  * that are recorded for each forwarding stream.
@@ -1252,6 +1262,10 @@ extern int flow_parse(const char *src, void *result, unsigned int size,
 		      struct rte_flow_attr **attr,
 		      struct rte_flow_item **pattern,
 		      struct rte_flow_action **actions);
+int setup_hairpin_queues(portid_t pi, portid_t p_pi, uint16_t cnt_pi);
+int hairpin_bind(uint16_t cfg_pi, portid_t *pl, portid_t *peer_pl);
+void hairpin_map_usage(void);
+int parse_hairpin_map(const char *hpmap);
 
 uint64_t str_to_rsstypes(const char *str);
 const char *rsstypes_to_str(uint64_t rss_type);
diff --git a/doc/guides/testpmd_app_ug/run_app.rst b/doc/guides/testpmd_app_ug/run_app.rst
index 1a9b812a7f..48717707a7 100644
--- a/doc/guides/testpmd_app_ug/run_app.rst
+++ b/doc/guides/testpmd_app_ug/run_app.rst
@@ -571,6 +571,9 @@ The command line options are:
 
     The default value is 0. Hairpin will use single port mode and implicit Tx flow mode.
 
+*   ``--hairpin-map=Rx port id:Rx queue:Tx port id:Tx queue:queues number``
+
+    Set explicit hairpin configuration.
 
 Testpmd Multi-Process Command-line Options
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-- 
2.43.0

RE: [PATCH v4 1/1] testpmd: add hairpin-map parameter

[Singh, Aman Deep]

Hi Gregory,
Can you please resolve the compilation issues seen with this patch.

Hairpin offloads packet forwarding between ports.
Packet is expected on Rx port <rp>, Rx queue <rq> and is forwarded to Tx port <tp> Tx queue <tq>.

Testpmd implements a static hairpin configuration scheme.

The new parameter allows explicit selection of Rx and Tx ports and queues in hairpin configuration.
The new `hairpin-map` parameter is provided with 5 parameters, separated by `:`

`--hairpin-map=Rx port id:Rx queue:Tx port id:Tx queue:queues number`

Can we rephrase the parameter "queues number " to something like "total queue pairs"

Testpmd operator can provide several `hairpin-map` parameters for different hairpin maps.
Example:

dpdk-testpmd <EAL params> -- \
  <testpmd params> \
  --rxq=2 --txq=2 --hairpinq=2 --hairpin-mode=0x12 \
  --hairpin-map=0:2:1:2:1 \ # [1]
  --hairpin-map=0:3:2:2:3   # [2]

Hairpin map [1] binds Rx port 0, queue 2 with Tx port 1, queue 2.
Hairpin map [2] binds
  Rx port 0, queue 3 with Tx port 2, queue 2,
  Rx port 0, queue 4 with Tx port 2, queue 3,
  Rx port 0, queue 5 with Tx port 2, queue 4.

The new `hairpin-map` parameter is optional.
If omitted, testpmd will create "default" hairpin maps.

Signed-off-by: Gregory Etelson <getelson@nvidia.com>
Acked-by: Dariusz Sosnowski <dsosnowski@nvidia.com>
---
 app/test-pmd/hairpin.c                | 385 ++++++++++++++++++++++++++
 app/test-pmd/meson.build              |   1 +
 app/test-pmd/parameters.c             |  10 +
 app/test-pmd/testpmd.c                | 217 +--------------
 app/test-pmd/testpmd.h                |  14 +
 doc/guides/testpmd_app_ug/run_app.rst |   3 +
 6 files changed, 416 insertions(+), 214 deletions(-)  create mode 100644 app/test-pmd/hairpin.c

diff --git a/app/test-pmd/hairpin.c b/app/test-pmd/hairpin.c new file mode 100644 index 0000000000..de54e8bac8
--- /dev/null
+++ b/app/test-pmd/hairpin.c
@@ -0,0 +1,385 @@
+/* SPDX-License-Identifier: BSD-3-Clause
+ * Copyright (c) 2022 NVIDIA Corporation & Affiliates  */
+
+#include <stdio.h>
+#include <string.h>
+#include <stdbool.h>
+#include <stdint.h>
+
+#include <sys/queue.h>
+
+#include "testpmd.h"
+
+/* Hairpin ports configuration mode. */ uint32_t hairpin_mode;
+
+bool hairpin_multiport_mode = false;
+
+queueid_t nb_hairpinq; /**< Number of hairpin queues per port. */
+
+static LIST_HEAD(, hairpin_map) hairpin_map_head = 
+LIST_HEAD_INITIALIZER();
+
+struct hairpin_map {
+	LIST_ENTRY(hairpin_map) entry; /**< List entry. */
+	portid_t rx_port; /**< Hairpin Rx port ID. */
+	portid_t tx_port; /**< Hairpin Tx port ID. */
+	uint16_t rxq_head; /**< Hairpin Rx queue head. */
+	uint16_t txq_head; /**< Hairpin Tx queue head. */
+	uint16_t qnum; /**< Hairpin queues number. */ };
+
+void
+hairpin_add_multiport_map(struct hairpin_map *map) {
+	LIST_INSERT_HEAD(&hairpin_map_head, map, entry); }
+
+/*
+ * Get the allowed maximum number of hairpin queues.
+ * *pid return the port id which has minimal value of
+ * max_hairpin_queues in all ports.
+ */
+queueid_t
+get_allowed_max_nb_hairpinq(portid_t *pid) {
+	queueid_t allowed_max_hairpinq = RTE_MAX_QUEUES_PER_PORT;
+	portid_t pi;
+	struct rte_eth_hairpin_cap cap;
+
+	RTE_ETH_FOREACH_DEV(pi) {
+		if (rte_eth_dev_hairpin_capability_get(pi, &cap) != 0) {
+			*pid = pi;
+			return 0;
+		}
+		if (cap.max_nb_queues < allowed_max_hairpinq) {
+			allowed_max_hairpinq = cap.max_nb_queues;
+			*pid = pi;
+		}
+	}
+	return allowed_max_hairpinq;
+}
+
+/*
+ * Check input hairpin is valid or not.
+ * If input hairpin is not greater than any of maximum number
+ * of hairpin queues of all ports, it is valid.
+ * if valid, return 0, else return -1
+ */
+int
+check_nb_hairpinq(queueid_t hairpinq)
+{
+	queueid_t allowed_max_hairpinq;
+	portid_t pid = 0;
+
+	allowed_max_hairpinq = get_allowed_max_nb_hairpinq(&pid);
+	if (hairpinq > allowed_max_hairpinq) {
+		fprintf(stderr,
+			"Fail: input hairpin (%u) can't be greater than max_hairpin_queues (%u) of port %u\n",
+			hairpinq, allowed_max_hairpinq, pid);
+		return -1;
+	}
+	return 0;
+}
+
+#define HAIRPIN_MODE_RX_FORCE_MEMORY RTE_BIT32(8) #define 
+HAIRPIN_MODE_TX_FORCE_MEMORY RTE_BIT32(9)
+
+#define HAIRPIN_MODE_RX_LOCKED_MEMORY RTE_BIT32(12) #define 
+HAIRPIN_MODE_RX_RTE_MEMORY RTE_BIT32(13)
+
+#define HAIRPIN_MODE_TX_LOCKED_MEMORY RTE_BIT32(16) #define 
+HAIRPIN_MODE_TX_RTE_MEMORY RTE_BIT32(17)
+
+static int
+port_config_hairpin_rxq(portid_t pi, uint16_t peer_tx_port,
+			queueid_t rxq_head, queueid_t txq_head,
+			uint16_t qcount, uint32_t manual_bind) {
+	int diag;
+	queueid_t i, qi;
+	uint32_t tx_explicit = !!(hairpin_mode & 0x10);
+	uint32_t force_mem = !!(hairpin_mode & HAIRPIN_MODE_RX_FORCE_MEMORY);
+	uint32_t locked_mem = !!(hairpin_mode & HAIRPIN_MODE_RX_LOCKED_MEMORY);
+	uint32_t rte_mem = !!(hairpin_mode & HAIRPIN_MODE_RX_RTE_MEMORY);
+	struct rte_port *port = &ports[pi];
+	struct rte_eth_hairpin_conf hairpin_conf = {
+		.peer_count = 1,
+	};
+
+	for (qi = rxq_head, i = 0; qi < rxq_head + qcount; qi++) {
+		hairpin_conf.peers[0].port = peer_tx_port;
+		hairpin_conf.peers[0].queue = i + txq_head;
+		hairpin_conf.manual_bind = manual_bind;
+		hairpin_conf.tx_explicit = tx_explicit;
+		hairpin_conf.force_memory = force_mem;
+		hairpin_conf.use_locked_device_memory = locked_mem;
+		hairpin_conf.use_rte_memory = rte_mem;
+		diag = rte_eth_rx_hairpin_queue_setup
+			(pi, qi, nb_rxd, &hairpin_conf);
+		i++;
+		if (diag == 0)
+			continue;
+
+		/* Fail to setup rx queue, return */
+		if (port->port_status == RTE_PORT_HANDLING)
+			port->port_status = RTE_PORT_STOPPED;
+		else
+			fprintf(stderr,
+				"Port %d can not be set back to stopped\n", pi);
+		fprintf(stderr,
+			"Port %d failed to configure hairpin on rxq %u.\n"
+			"Peer port: %u peer txq: %u\n",
+			pi, qi, peer_tx_port, i);
+		/* try to reconfigure queues next time */
+		port->need_reconfig_queues = 1;
+		return -1;
+	}
+	return 0;
+}
+
+static int
+port_config_hairpin_txq(portid_t pi, uint16_t peer_rx_port,
+			queueid_t rxq_head, queueid_t txq_head,
+			uint16_t qcount, uint32_t manual_bind) {
+	int diag;
+	queueid_t i, qi;
+	uint32_t tx_explicit = !!(hairpin_mode & 0x10);
+	uint32_t force_mem = !!(hairpin_mode & HAIRPIN_MODE_TX_FORCE_MEMORY);
+	uint32_t locked_mem = !!(hairpin_mode & HAIRPIN_MODE_TX_LOCKED_MEMORY);
+	uint32_t rte_mem = !!(hairpin_mode & HAIRPIN_MODE_TX_RTE_MEMORY);
+	struct rte_port *port = &ports[pi];
+	struct rte_eth_hairpin_conf hairpin_conf = {
+		.peer_count = 1,
+	};
+
+	for (qi = txq_head, i = 0; qi < txq_head + qcount; qi++) {
+		hairpin_conf.peers[0].port = peer_rx_port;
+		hairpin_conf.peers[0].queue = i + rxq_head;
+		hairpin_conf.manual_bind = manual_bind;
+		hairpin_conf.tx_explicit = tx_explicit;
+		hairpin_conf.force_memory = force_mem;
+		hairpin_conf.use_locked_device_memory = locked_mem;
+		hairpin_conf.use_rte_memory = rte_mem;
+		diag = rte_eth_tx_hairpin_queue_setup
+			(pi, qi, nb_txd, &hairpin_conf);
+		i++;
+		if (diag == 0)
+			continue;
+
+		/* Fail to setup rx queue, return */
+		if (port->port_status == RTE_PORT_HANDLING)
+			port->port_status = RTE_PORT_STOPPED;
+		else
+			fprintf(stderr,
+				"Port %d can not be set back to stopped\n", pi);
+		fprintf(stderr,
+			"Port %d failed to configure hairpin on txq %u.\n"
+			"Peer port: %u peer rxq: %u\n",
+			pi, qi, peer_rx_port, i);
+		/* try to reconfigure queues next time */
+		port->need_reconfig_queues = 1;
+		return -1;
+	}
+	return 0;
+}
+
+static int
+setup_legacy_hairpin_queues(portid_t pi, portid_t p_pi, uint16_t 
+cnt_pi) {
+	int diag;
+	uint16_t peer_rx_port = pi;
+	uint16_t peer_tx_port = pi;
+	uint32_t manual = 1;
+
+	if (!(hairpin_mode & 0xf)) {
+		peer_rx_port = pi;
+		peer_tx_port = pi;
+		manual = 0;
+	} else if (hairpin_mode & 0x1) {
+		peer_tx_port = rte_eth_find_next_owned_by(pi + 1,
+							  RTE_ETH_DEV_NO_OWNER);
+		if (peer_tx_port >= RTE_MAX_ETHPORTS)
+			peer_tx_port = rte_eth_find_next_owned_by(0,
+								  RTE_ETH_DEV_NO_OWNER);
+		if (p_pi != RTE_MAX_ETHPORTS) {
+			peer_rx_port = p_pi;
+		} else {
+			uint16_t next_pi;
+
+			/* Last port will be the peer RX port of the first. */
+			RTE_ETH_FOREACH_DEV(next_pi)
+				peer_rx_port = next_pi;
+		}
+		manual = 1;
+	} else if (hairpin_mode & 0x2) {
+		if (cnt_pi & 0x1) {
+			peer_rx_port = p_pi;
+		} else {
+			peer_rx_port = rte_eth_find_next_owned_by(pi + 1,
+								  RTE_ETH_DEV_NO_OWNER);
+			if (peer_rx_port >= RTE_MAX_ETHPORTS)
+				peer_rx_port = pi;
+		}
+		peer_tx_port = peer_rx_port;
+		manual = 1;
+	}
+	diag = port_config_hairpin_txq(pi, peer_rx_port, nb_rxq, nb_txq,
+				       nb_hairpinq, manual);
+	if (diag)
+		return diag;
+	diag = port_config_hairpin_rxq(pi, peer_tx_port, nb_rxq, nb_txq,
+				       nb_hairpinq, manual);
+	if (diag)
+		return diag;
+	return 0;
+}
+
+static int
+setup_mapped_harpin_queues(portid_t pi) {
+	int ret = 0;
+	struct hairpin_map *map;
+
+	LIST_FOREACH(map, &hairpin_map_head, entry) {
+		if (map->rx_port == pi) {
+			ret = port_config_hairpin_rxq(pi, map->tx_port,
+						      map->rxq_head,
+						      map->txq_head,
+						      map->qnum, true);
+			if (ret)
+				return ret;
+		}
+		if (map->tx_port == pi) {
+			ret = port_config_hairpin_txq(pi, map->rx_port,
+						      map->rxq_head,
+						      map->txq_head,
+						      map->qnum, true);
+			if (ret)
+				return ret;
+		}
+	}
+	return 0;
+}
+
+/* Configure the Rx and Tx hairpin queues for the selected port. */ int 
+setup_hairpin_queues(portid_t pi, portid_t p_pi, uint16_t cnt_pi) {
+	return !hairpin_multiport_mode ?
+	       setup_legacy_hairpin_queues(pi, p_pi, cnt_pi) :
+	       setup_mapped_harpin_queues(pi); }
+
+int
+hairpin_bind(uint16_t cfg_pi, portid_t *pl, portid_t *peer_pl) {
+	uint16_t i;
+	portid_t pi;
+	int peer_pi;
+	int diag;
+	int j;
+
+	/* bind all started hairpin ports */
+	for (i = 0; i < cfg_pi; i++) {
+		pi = pl[i];
+		/* bind current Tx to all peer Rx */
+		peer_pi = rte_eth_hairpin_get_peer_ports(pi, peer_pl,
+							 RTE_MAX_ETHPORTS, 1);
+		if (peer_pi < 0)
+			return peer_pi;
+		for (j = 0; j < peer_pi; j++) {
+			if (!port_is_started(peer_pl[j]))
+				continue;
+			diag = rte_eth_hairpin_bind(pi, peer_pl[j]);
+			if (diag < 0) {
+				fprintf(stderr,
+					"Error during binding hairpin Tx port %u to %u: %s\n",
+					pi, peer_pl[j],
+					rte_strerror(-diag));
+				return -1;
+			}
+		}
+		/* bind all peer Tx to current Rx */
+		peer_pi = rte_eth_hairpin_get_peer_ports(pi, peer_pl,
+							 RTE_MAX_ETHPORTS, 0);
+		if (peer_pi < 0)
+			return peer_pi;
+		for (j = 0; j < peer_pi; j++) {
+			if (!port_is_started(peer_pl[j]))
+				continue;
+			diag = rte_eth_hairpin_bind(peer_pl[j], pi);
+			if (diag < 0) {
+				fprintf(stderr,
+					"Error during binding hairpin Tx port %u to %u: %s\n",
+					peer_pl[j], pi,
+					rte_strerror(-diag));
+				return -1;
+			}
+		}
+	}
+	return 0;
+}
+
+void
+hairpin_map_usage(void)
+{
+	printf("  --hairpin-map=rxpi:rxq:txpi:txq:n: hairpin map.\n"
+	       "    rxpi - Rx port index.\n"
+	       "    rxq  - Rx queue.\n"
+	       "    txpi - Tx port index.\n"
+	       "    txq  - Tx queue.\n"
+	       "    n    - hairpin queues number.\n");
+}
+
+static char
+*parse_hairpin_map_entry(char *input, char **next) {
+	char *tail = strchr(input, ':');
+
+	if (!tail)
+		return NULL;
+	tail[0] = '\0';
+	*next = tail + 1;
+	return input;
+}
+
+int
+parse_hairpin_map(const char *hpmap)
+{
+	/*
+	 * Testpmd hairpin map format:
+	 * <Rx port id:First Rx queue:Tx port id:First Tx queue:queues number>
+	 */
+	char *head, *next = (char *)(uintptr_t)hpmap;
+	struct hairpin_map *map = calloc(1, sizeof(*map));
Can use rte_calloc() here.
+
+	if (!map)
+		return -ENOMEM;
+
+	head = parse_hairpin_map_entry(next, &next);
+	if (!head)
+		goto err;
+	map->rx_port = atoi(head);
+	head = parse_hairpin_map_entry(next, &next);
+	if (!head)
+		goto err;
+	map->rxq_head = atoi(head);
+	head = parse_hairpin_map_entry(next, &next);
+	if (!head)
+		goto err;
+	map->tx_port = atoi(head);
+	head = parse_hairpin_map_entry(next, &next);
+	if (!head)
+		goto err;
+	map->txq_head = atoi(head);
+	map->qnum = atoi(next);
+	hairpin_add_multiport_map(map);
+	return 0;
+err:
+	free(map);
Can use rte_free()
+	return -EINVAL;
+}
+
+
diff --git a/app/test-pmd/meson.build b/app/test-pmd/meson.build index 719f875be0..f1c36529b4 100644
--- a/app/test-pmd/meson.build
+++ b/app/test-pmd/meson.build
@@ -15,6 +15,7 @@ sources = files(
         'config.c',
         'csumonly.c',
         'flowgen.c',
+        'hairpin.c',
         'icmpecho.c',
         'ieee1588fwd.c',
         'iofwd.c',
diff --git a/app/test-pmd/parameters.c b/app/test-pmd/parameters.c index 22364e09ab..7b31b94542 100644
--- a/app/test-pmd/parameters.c
+++ b/app/test-pmd/parameters.c
@@ -143,6 +143,8 @@ enum {
 	TESTPMD_OPT_HAIRPINQ_NUM,
 #define TESTPMD_OPT_HAIRPIN_MODE "hairpin-mode"
 	TESTPMD_OPT_HAIRPIN_MODE_NUM,
+#define TESTPMD_OPT_HAIRPIN_MAP "hairpin-map"
+	TESTPMD_OPT_HAIRPIN_MAP_NUM,
 #define TESTPMD_OPT_BURST "burst"
 	TESTPMD_OPT_BURST_NUM,
 #define TESTPMD_OPT_FLOWGEN_CLONES "flowgen-clones"
@@ -317,6 +319,7 @@ static const struct option long_options[] = {
 	REQUIRED_ARG(TESTPMD_OPT_TXD),
 	REQUIRED_ARG(TESTPMD_OPT_HAIRPINQ),
 	REQUIRED_ARG(TESTPMD_OPT_HAIRPIN_MODE),
+	REQUIRED_ARG(TESTPMD_OPT_HAIRPIN_MAP),
 	REQUIRED_ARG(TESTPMD_OPT_BURST),
 	REQUIRED_ARG(TESTPMD_OPT_FLOWGEN_CLONES),
 	REQUIRED_ARG(TESTPMD_OPT_FLOWGEN_FLOWS),
@@ -542,6 +545,7 @@ usage(char* progname)
 	printf("  --hairpin-mode=0xXX: bitmask set the hairpin port mode.\n"
 	       "    0x10 - explicit Tx rule, 0x02 - hairpin ports paired\n"
 	       "    0x01 - hairpin ports loop, 0x00 - hairpin port self\n");
+	hairpin_map_usage();
 }
 
 static int
@@ -1317,6 +1321,12 @@ launch_args_parse(int argc, char** argv)
 				hairpin_mode = (uint32_t)n;
 			break;
 		}
+		case TESTPMD_OPT_HAIRPIN_MAP_NUM:
+			hairpin_multiport_mode = true;
+			ret = parse_hairpin_map(optarg);
+			if (ret)
+				rte_exit(EXIT_FAILURE, "invalid hairpin map\n");
+			break;
 		case TESTPMD_OPT_BURST_NUM:
 			n = atoi(optarg);
 			if (n == 0) {
diff --git a/app/test-pmd/testpmd.c b/app/test-pmd/testpmd.c index b1401136e4..f487769578 100644
--- a/app/test-pmd/testpmd.c
+++ b/app/test-pmd/testpmd.c
@@ -284,7 +284,6 @@ uint8_t dcb_config = 0;
 /*
  * Configurable number of RX/TX queues.
  */
-queueid_t nb_hairpinq; /**< Number of hairpin queues per port. */  queueid_t nb_rxq = 1; /**< Number of RX queues per port. */  queueid_t nb_txq = 1; /**< Number of TX queues per port. */
 
@@ -431,9 +430,6 @@ bool setup_on_probe_event = true;
 /* Clear ptypes on port initialization. */  uint8_t clear_ptypes = true;
 
-/* Hairpin ports configuration mode. */ -uint32_t hairpin_mode;
-
 /* Pretty printing of ethdev events */
 static const char * const eth_event_desc[] = {
 	[RTE_ETH_EVENT_UNKNOWN] = "unknown",
@@ -1555,54 +1551,6 @@ check_nb_txd(queueid_t txd)
 	return 0;
 }
 
-
-/*
- * Get the allowed maximum number of hairpin queues.
- * *pid return the port id which has minimal value of
- * max_hairpin_queues in all ports.
- */
-queueid_t
-get_allowed_max_nb_hairpinq(portid_t *pid) -{
-	queueid_t allowed_max_hairpinq = RTE_MAX_QUEUES_PER_PORT;
-	portid_t pi;
-	struct rte_eth_hairpin_cap cap;
-
-	RTE_ETH_FOREACH_DEV(pi) {
-		if (rte_eth_dev_hairpin_capability_get(pi, &cap) != 0) {
-			*pid = pi;
-			return 0;
-		}
-		if (cap.max_nb_queues < allowed_max_hairpinq) {
-			allowed_max_hairpinq = cap.max_nb_queues;
-			*pid = pi;
-		}
-	}
-	return allowed_max_hairpinq;
-}
-
-/*
- * Check input hairpin is valid or not.
- * If input hairpin is not greater than any of maximum number
- * of hairpin queues of all ports, it is valid.
- * if valid, return 0, else return -1
- */
-int
-check_nb_hairpinq(queueid_t hairpinq)
-{
-	queueid_t allowed_max_hairpinq;
-	portid_t pid = 0;
-
-	allowed_max_hairpinq = get_allowed_max_nb_hairpinq(&pid);
-	if (hairpinq > allowed_max_hairpinq) {
-		fprintf(stderr,
-			"Fail: input hairpin (%u) can't be greater than max_hairpin_queues (%u) of port %u\n",
-			hairpinq, allowed_max_hairpinq, pid);
-		return -1;
-	}
-	return 0;
-}
-
 static int
 get_eth_overhead(struct rte_eth_dev_info *dev_info)  { @@ -2684,126 +2632,6 @@ port_is_started(portid_t port_id)
 	return 1;
 }
 
-#define HAIRPIN_MODE_RX_FORCE_MEMORY RTE_BIT32(8) -#define HAIRPIN_MODE_TX_FORCE_MEMORY RTE_BIT32(9)
-
-#define HAIRPIN_MODE_RX_LOCKED_MEMORY RTE_BIT32(12) -#define HAIRPIN_MODE_RX_RTE_MEMORY RTE_BIT32(13)
-
-#define HAIRPIN_MODE_TX_LOCKED_MEMORY RTE_BIT32(16) -#define HAIRPIN_MODE_TX_RTE_MEMORY RTE_BIT32(17)
-
-
-/* Configure the Rx and Tx hairpin queues for the selected port. */ -static int -setup_hairpin_queues(portid_t pi, portid_t p_pi, uint16_t cnt_pi) -{
-	queueid_t qi;
-	struct rte_eth_hairpin_conf hairpin_conf = {
-		.peer_count = 1,
-	};
-	int i;
-	int diag;
-	struct rte_port *port = &ports[pi];
-	uint16_t peer_rx_port = pi;
-	uint16_t peer_tx_port = pi;
-	uint32_t manual = 1;
-	uint32_t tx_exp = hairpin_mode & 0x10;
-	uint32_t rx_force_memory = hairpin_mode & HAIRPIN_MODE_RX_FORCE_MEMORY;
-	uint32_t rx_locked_memory = hairpin_mode & HAIRPIN_MODE_RX_LOCKED_MEMORY;
-	uint32_t rx_rte_memory = hairpin_mode & HAIRPIN_MODE_RX_RTE_MEMORY;
-	uint32_t tx_force_memory = hairpin_mode & HAIRPIN_MODE_TX_FORCE_MEMORY;
-	uint32_t tx_locked_memory = hairpin_mode & HAIRPIN_MODE_TX_LOCKED_MEMORY;
-	uint32_t tx_rte_memory = hairpin_mode & HAIRPIN_MODE_TX_RTE_MEMORY;
-
-	if (!(hairpin_mode & 0xf)) {
-		peer_rx_port = pi;
-		peer_tx_port = pi;
-		manual = 0;
-	} else if (hairpin_mode & 0x1) {
-		peer_tx_port = rte_eth_find_next_owned_by(pi + 1,
-						       RTE_ETH_DEV_NO_OWNER);
-		if (peer_tx_port >= RTE_MAX_ETHPORTS)
-			peer_tx_port = rte_eth_find_next_owned_by(0,
-						RTE_ETH_DEV_NO_OWNER);
-		if (p_pi != RTE_MAX_ETHPORTS) {
-			peer_rx_port = p_pi;
-		} else {
-			uint16_t next_pi;
-
-			/* Last port will be the peer RX port of the first. */
-			RTE_ETH_FOREACH_DEV(next_pi)
-				peer_rx_port = next_pi;
-		}
-		manual = 1;
-	} else if (hairpin_mode & 0x2) {
-		if (cnt_pi & 0x1) {
-			peer_rx_port = p_pi;
-		} else {
-			peer_rx_port = rte_eth_find_next_owned_by(pi + 1,
-						RTE_ETH_DEV_NO_OWNER);
-			if (peer_rx_port >= RTE_MAX_ETHPORTS)
-				peer_rx_port = pi;
-		}
-		peer_tx_port = peer_rx_port;
-		manual = 1;
-	}
-
-	for (qi = nb_txq, i = 0; qi < nb_hairpinq + nb_txq; qi++) {
-		hairpin_conf.peers[0].port = peer_rx_port;
-		hairpin_conf.peers[0].queue = i + nb_rxq;
-		hairpin_conf.manual_bind = !!manual;
-		hairpin_conf.tx_explicit = !!tx_exp;
-		hairpin_conf.force_memory = !!tx_force_memory;
-		hairpin_conf.use_locked_device_memory = !!tx_locked_memory;
-		hairpin_conf.use_rte_memory = !!tx_rte_memory;
-		diag = rte_eth_tx_hairpin_queue_setup
-			(pi, qi, nb_txd, &hairpin_conf);
-		i++;
-		if (diag == 0)
-			continue;
-
-		/* Fail to setup rx queue, return */
-		if (port->port_status == RTE_PORT_HANDLING)
-			port->port_status = RTE_PORT_STOPPED;
-		else
-			fprintf(stderr,
-				"Port %d can not be set back to stopped\n", pi);
-		fprintf(stderr, "Fail to configure port %d hairpin queues\n",
-			pi);
-		/* try to reconfigure queues next time */
-		port->need_reconfig_queues = 1;
-		return -1;
-	}
-	for (qi = nb_rxq, i = 0; qi < nb_hairpinq + nb_rxq; qi++) {
-		hairpin_conf.peers[0].port = peer_tx_port;
-		hairpin_conf.peers[0].queue = i + nb_txq;
-		hairpin_conf.manual_bind = !!manual;
-		hairpin_conf.tx_explicit = !!tx_exp;
-		hairpin_conf.force_memory = !!rx_force_memory;
-		hairpin_conf.use_locked_device_memory = !!rx_locked_memory;
-		hairpin_conf.use_rte_memory = !!rx_rte_memory;
-		diag = rte_eth_rx_hairpin_queue_setup
-			(pi, qi, nb_rxd, &hairpin_conf);
-		i++;
-		if (diag == 0)
-			continue;
-
-		/* Fail to setup rx queue, return */
-		if (port->port_status == RTE_PORT_HANDLING)
-			port->port_status = RTE_PORT_STOPPED;
-		else
-			fprintf(stderr,
-				"Port %d can not be set back to stopped\n", pi);
-		fprintf(stderr, "Fail to configure port %d hairpin queues\n",
-			pi);
-		/* try to reconfigure queues next time */
-		port->need_reconfig_queues = 1;
-		return -1;
-	}
-	return 0;
-}
-
 /* Configure the Rx with optional split. */  int  rx_queue_setup(uint16_t port_id, uint16_t rx_queue_id, @@ -3043,7 +2871,6 @@ start_port(portid_t pid)
 	portid_t peer_pl[RTE_MAX_ETHPORTS];
 	uint16_t cnt_pi = 0;
 	uint16_t cfg_pi = 0;
-	int peer_pi;
 	queueid_t qi;
 	struct rte_port *port;
 	struct rte_eth_hairpin_cap cap;
@@ -3304,47 +3131,9 @@ start_port(portid_t pid)
 		fprintf(stderr, "Please stop the ports first\n");
 
 	if (hairpin_mode & 0xf) {
-		uint16_t i;
-		int j;
-
-		/* bind all started hairpin ports */
-		for (i = 0; i < cfg_pi; i++) {
-			pi = pl[i];
-			/* bind current Tx to all peer Rx */
-			peer_pi = rte_eth_hairpin_get_peer_ports(pi, peer_pl,
-							RTE_MAX_ETHPORTS, 1);
-			if (peer_pi < 0)
-				return peer_pi;
-			for (j = 0; j < peer_pi; j++) {
-				if (!port_is_started(peer_pl[j]))
-					continue;
-				diag = rte_eth_hairpin_bind(pi, peer_pl[j]);
-				if (diag < 0) {
-					fprintf(stderr,
-						"Error during binding hairpin Tx port %u to %u: %s\n",
-						pi, peer_pl[j],
-						rte_strerror(-diag));
-					return -1;
-				}
-			}
-			/* bind all peer Tx to current Rx */
-			peer_pi = rte_eth_hairpin_get_peer_ports(pi, peer_pl,
-							RTE_MAX_ETHPORTS, 0);
-			if (peer_pi < 0)
-				return peer_pi;
-			for (j = 0; j < peer_pi; j++) {
-				if (!port_is_started(peer_pl[j]))
-					continue;
-				diag = rte_eth_hairpin_bind(peer_pl[j], pi);
-				if (diag < 0) {
-					fprintf(stderr,
-						"Error during binding hairpin Tx port %u to %u: %s\n",
-						peer_pl[j], pi,
-						rte_strerror(-diag));
-					return -1;
-				}
-			}
-		}
+		diag = hairpin_bind(cfg_pi, pl, peer_pl);
+		if (diag < 0)
+			return -1;
 	}
 
 	fill_xstats_display_info_for_port(pid);
diff --git a/app/test-pmd/testpmd.h b/app/test-pmd/testpmd.h index 9facd7f281..659c5b4fdc 100644
--- a/app/test-pmd/testpmd.h
+++ b/app/test-pmd/testpmd.h
@@ -126,6 +126,16 @@ enum noisy_fwd_mode {
 	NOISY_FWD_MODE_MAX,
 };
 
+/**
+ * Command line arguments parser sets `hairpin_multiport_mode` to True
+ * if explicit hairpin map configuration mode was used.
+ */
+extern bool hairpin_multiport_mode;
+
+/** Hairpin maps list. */
+struct hairpin_map;
+extern void hairpin_add_multiport_map(struct hairpin_map *map);
+
 /**
  * The data structure associated with RX and TX packet burst statistics
  * that are recorded for each forwarding stream.
@@ -1252,6 +1262,10 @@ extern int flow_parse(const char *src, void *result, unsigned int size,
 		      struct rte_flow_attr **attr,
 		      struct rte_flow_item **pattern,
 		      struct rte_flow_action **actions);
+int setup_hairpin_queues(portid_t pi, portid_t p_pi, uint16_t cnt_pi); 
+int hairpin_bind(uint16_t cfg_pi, portid_t *pl, portid_t *peer_pl); 
+void hairpin_map_usage(void); int parse_hairpin_map(const char *hpmap);
 
 uint64_t str_to_rsstypes(const char *str);  const char *rsstypes_to_str(uint64_t rss_type); diff --git a/doc/guides/testpmd_app_ug/run_app.rst b/doc/guides/testpmd_app_ug/run_app.rst
index 1a9b812a7f..48717707a7 100644
--- a/doc/guides/testpmd_app_ug/run_app.rst
+++ b/doc/guides/testpmd_app_ug/run_app.rst
@@ -571,6 +571,9 @@ The command line options are:
 
     The default value is 0. Hairpin will use single port mode and implicit Tx flow mode.
 
+*   ``--hairpin-map=Rx port id:Rx queue:Tx port id:Tx queue:queues number``
+
+    Set explicit hairpin configuration.
 
 Testpmd Multi-Process Command-line Options  ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
--
2.43.0

[PATCH v5 1/1] testpmd: add hairpin-map parameter

[Gregory Etelson]

Hairpin offloads packet forwarding between ports.
Packet is expected on Rx port <rp>, Rx queue <rq> and is forwarded
to Tx port <tp> Tx queue <tq>.

Testpmd implements a static hairpin configuration scheme.

The new parameter allows explicit selection of Rx and Tx ports and
queues in hairpin configuration.
The new `hairpin-map` parameter is provided with 5 parameters,
separated by `:`

`--hairpin-map=Rx port id:Rx queue:Tx port id:Tx queue:queues number`

Testpmd operator can provide several `hairpin-map` parameters for
different hairpin maps.
Example:

dpdk-testpmd <EAL params> -- \
  <testpmd params> \
  --rxq=2 --txq=2 --hairpinq=2 --hairpin-mode=0x12 \
  --hairpin-map=0:2:1:2:1 \ # [1]
  --hairpin-map=0:3:2:2:3   # [2]

Hairpin map [1] binds Rx port 0, queue 2 with Tx port 1, queue 2.
Hairpin map [2] binds
  Rx port 0, queue 3 with Tx port 2, queue 2,
  Rx port 0, queue 4 with Tx port 2, queue 3,
  Rx port 0, queue 5 with Tx port 2, queue 4.

The new `hairpin-map` parameter is optional.
If omitted, testpmd will create "default" hairpin maps.

Signed-off-by: Gregory Etelson <getelson@nvidia.com>
Acked-by: Dariusz Sosnowski <dsosnowski@nvidia.com>
---
 app/test-pmd/hairpin.c                | 386 ++++++++++++++++++++++++++
 app/test-pmd/meson.build              |   1 +
 app/test-pmd/parameters.c             |  10 +
 app/test-pmd/testpmd.c                | 217 +--------------
 app/test-pmd/testpmd.h                |  14 +
 doc/guides/testpmd_app_ug/run_app.rst |   3 +
 6 files changed, 417 insertions(+), 214 deletions(-)
 create mode 100644 app/test-pmd/hairpin.c

diff --git a/app/test-pmd/hairpin.c b/app/test-pmd/hairpin.c
new file mode 100644
index 0000000000..445d543851
--- /dev/null
+++ b/app/test-pmd/hairpin.c
@@ -0,0 +1,386 @@
+/* SPDX-License-Identifier: BSD-3-Clause
+ * Copyright (c) 2022 NVIDIA Corporation & Affiliates
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <stdbool.h>
+#include <stdint.h>
+
+#include <sys/queue.h>
+
+#include "testpmd.h"
+
+/* Hairpin ports configuration mode. */
+uint32_t hairpin_mode;
+
+bool hairpin_multiport_mode = false;
+
+queueid_t nb_hairpinq; /**< Number of hairpin queues per port. */
+
+static LIST_HEAD(, hairpin_map) hairpin_map_head = LIST_HEAD_INITIALIZER();
+
+struct hairpin_map {
+	LIST_ENTRY(hairpin_map) entry; /**< List entry. */
+	portid_t rx_port; /**< Hairpin Rx port ID. */
+	portid_t tx_port; /**< Hairpin Tx port ID. */
+	uint16_t rxq_head; /**< Hairpin Rx queue head. */
+	uint16_t txq_head; /**< Hairpin Tx queue head. */
+	uint16_t qnum; /**< Hairpin queues number. */
+};
+
+void
+hairpin_add_multiport_map(struct hairpin_map *map)
+{
+	LIST_INSERT_HEAD(&hairpin_map_head, map, entry);
+}
+
+/*
+ * Get the allowed maximum number of hairpin queues.
+ * *pid return the port id which has minimal value of
+ * max_hairpin_queues in all ports.
+ */
+queueid_t
+get_allowed_max_nb_hairpinq(portid_t *pid)
+{
+	queueid_t allowed_max_hairpinq = RTE_MAX_QUEUES_PER_PORT;
+	portid_t pi;
+	struct rte_eth_hairpin_cap cap;
+
+	RTE_ETH_FOREACH_DEV(pi) {
+		if (rte_eth_dev_hairpin_capability_get(pi, &cap) != 0) {
+			*pid = pi;
+			return 0;
+		}
+		if (cap.max_nb_queues < allowed_max_hairpinq) {
+			allowed_max_hairpinq = cap.max_nb_queues;
+			*pid = pi;
+		}
+	}
+	return allowed_max_hairpinq;
+}
+
+/*
+ * Check input hairpin is valid or not.
+ * If input hairpin is not greater than any of maximum number
+ * of hairpin queues of all ports, it is valid.
+ * if valid, return 0, else return -1
+ */
+int
+check_nb_hairpinq(queueid_t hairpinq)
+{
+	queueid_t allowed_max_hairpinq;
+	portid_t pid = 0;
+
+	allowed_max_hairpinq = get_allowed_max_nb_hairpinq(&pid);
+	if (hairpinq > allowed_max_hairpinq) {
+		fprintf(stderr,
+			"Fail: input hairpin (%u) can't be greater than max_hairpin_queues (%u) of port %u\n",
+			hairpinq, allowed_max_hairpinq, pid);
+		return -1;
+	}
+	return 0;
+}
+
+#define HAIRPIN_MODE_RX_FORCE_MEMORY RTE_BIT32(8)
+#define HAIRPIN_MODE_TX_FORCE_MEMORY RTE_BIT32(9)
+
+#define HAIRPIN_MODE_RX_LOCKED_MEMORY RTE_BIT32(12)
+#define HAIRPIN_MODE_RX_RTE_MEMORY RTE_BIT32(13)
+
+#define HAIRPIN_MODE_TX_LOCKED_MEMORY RTE_BIT32(16)
+#define HAIRPIN_MODE_TX_RTE_MEMORY RTE_BIT32(17)
+
+static int
+port_config_hairpin_rxq(portid_t pi, uint16_t peer_tx_port,
+			queueid_t rxq_head, queueid_t txq_head,
+			uint16_t qcount, uint32_t manual_bind)
+{
+	int diag;
+	queueid_t i, qi;
+	uint32_t tx_explicit = !!(hairpin_mode & 0x10);
+	uint32_t force_mem = !!(hairpin_mode & HAIRPIN_MODE_RX_FORCE_MEMORY);
+	uint32_t locked_mem = !!(hairpin_mode & HAIRPIN_MODE_RX_LOCKED_MEMORY);
+	uint32_t rte_mem = !!(hairpin_mode & HAIRPIN_MODE_RX_RTE_MEMORY);
+	struct rte_port *port = &ports[pi];
+	struct rte_eth_hairpin_conf hairpin_conf = {
+		.peer_count = 1,
+	};
+
+	for (qi = rxq_head, i = 0; qi < rxq_head + qcount; qi++) {
+		hairpin_conf.peers[0].port = peer_tx_port;
+		hairpin_conf.peers[0].queue = i + txq_head;
+		hairpin_conf.manual_bind = manual_bind;
+		hairpin_conf.tx_explicit = tx_explicit;
+		hairpin_conf.force_memory = force_mem;
+		hairpin_conf.use_locked_device_memory = locked_mem;
+		hairpin_conf.use_rte_memory = rte_mem;
+		diag = rte_eth_rx_hairpin_queue_setup
+			(pi, qi, nb_rxd, &hairpin_conf);
+		i++;
+		if (diag == 0)
+			continue;
+
+		/* Fail to setup rx queue, return */
+		if (port->port_status == RTE_PORT_HANDLING)
+			port->port_status = RTE_PORT_STOPPED;
+		else
+			fprintf(stderr,
+				"Port %d can not be set back to stopped\n", pi);
+		fprintf(stderr,
+			"Port %d failed to configure hairpin on rxq %u.\n"
+			"Peer port: %u peer txq: %u\n",
+			pi, qi, peer_tx_port, i);
+		/* try to reconfigure queues next time */
+		port->need_reconfig_queues = 1;
+		return -1;
+	}
+	return 0;
+}
+
+static int
+port_config_hairpin_txq(portid_t pi, uint16_t peer_rx_port,
+			queueid_t rxq_head, queueid_t txq_head,
+			uint16_t qcount, uint32_t manual_bind)
+{
+	int diag;
+	queueid_t i, qi;
+	uint32_t tx_explicit = !!(hairpin_mode & 0x10);
+	uint32_t force_mem = !!(hairpin_mode & HAIRPIN_MODE_TX_FORCE_MEMORY);
+	uint32_t locked_mem = !!(hairpin_mode & HAIRPIN_MODE_TX_LOCKED_MEMORY);
+	uint32_t rte_mem = !!(hairpin_mode & HAIRPIN_MODE_TX_RTE_MEMORY);
+	struct rte_port *port = &ports[pi];
+	struct rte_eth_hairpin_conf hairpin_conf = {
+		.peer_count = 1,
+	};
+
+	for (qi = txq_head, i = 0; qi < txq_head + qcount; qi++) {
+		hairpin_conf.peers[0].port = peer_rx_port;
+		hairpin_conf.peers[0].queue = i + rxq_head;
+		hairpin_conf.manual_bind = manual_bind;
+		hairpin_conf.tx_explicit = tx_explicit;
+		hairpin_conf.force_memory = force_mem;
+		hairpin_conf.use_locked_device_memory = locked_mem;
+		hairpin_conf.use_rte_memory = rte_mem;
+		diag = rte_eth_tx_hairpin_queue_setup
+			(pi, qi, nb_txd, &hairpin_conf);
+		i++;
+		if (diag == 0)
+			continue;
+
+		/* Fail to setup rx queue, return */
+		if (port->port_status == RTE_PORT_HANDLING)
+			port->port_status = RTE_PORT_STOPPED;
+		else
+			fprintf(stderr,
+				"Port %d can not be set back to stopped\n", pi);
+		fprintf(stderr,
+			"Port %d failed to configure hairpin on txq %u.\n"
+			"Peer port: %u peer rxq: %u\n",
+			pi, qi, peer_rx_port, i);
+		/* try to reconfigure queues next time */
+		port->need_reconfig_queues = 1;
+		return -1;
+	}
+	return 0;
+}
+
+static int
+setup_legacy_hairpin_queues(portid_t pi, portid_t p_pi, uint16_t cnt_pi)
+{
+	int diag;
+	uint16_t peer_rx_port = pi;
+	uint16_t peer_tx_port = pi;
+	uint32_t manual = 1;
+
+	if (!(hairpin_mode & 0xf)) {
+		peer_rx_port = pi;
+		peer_tx_port = pi;
+		manual = 0;
+	} else if (hairpin_mode & 0x1) {
+		peer_tx_port = rte_eth_find_next_owned_by(pi + 1,
+							  RTE_ETH_DEV_NO_OWNER);
+		if (peer_tx_port >= RTE_MAX_ETHPORTS)
+			peer_tx_port = rte_eth_find_next_owned_by(0,
+								  RTE_ETH_DEV_NO_OWNER);
+		if (p_pi != RTE_MAX_ETHPORTS) {
+			peer_rx_port = p_pi;
+		} else {
+			uint16_t next_pi;
+
+			/* Last port will be the peer RX port of the first. */
+			RTE_ETH_FOREACH_DEV(next_pi)
+				peer_rx_port = next_pi;
+		}
+		manual = 1;
+	} else if (hairpin_mode & 0x2) {
+		if (cnt_pi & 0x1) {
+			peer_rx_port = p_pi;
+		} else {
+			peer_rx_port = rte_eth_find_next_owned_by(pi + 1,
+								  RTE_ETH_DEV_NO_OWNER);
+			if (peer_rx_port >= RTE_MAX_ETHPORTS)
+				peer_rx_port = pi;
+		}
+		peer_tx_port = peer_rx_port;
+		manual = 1;
+	}
+	diag = port_config_hairpin_txq(pi, peer_rx_port, nb_rxq, nb_txq,
+				       nb_hairpinq, manual);
+	if (diag)
+		return diag;
+	diag = port_config_hairpin_rxq(pi, peer_tx_port, nb_rxq, nb_txq,
+				       nb_hairpinq, manual);
+	if (diag)
+		return diag;
+	return 0;
+}
+
+static int
+setup_mapped_harpin_queues(portid_t pi)
+{
+	int ret = 0;
+	struct hairpin_map *map;
+
+	LIST_FOREACH(map, &hairpin_map_head, entry) {
+		if (map->rx_port == pi) {
+			ret = port_config_hairpin_rxq(pi, map->tx_port,
+						      map->rxq_head,
+						      map->txq_head,
+						      map->qnum, true);
+			if (ret)
+				return ret;
+		}
+		if (map->tx_port == pi) {
+			ret = port_config_hairpin_txq(pi, map->rx_port,
+						      map->rxq_head,
+						      map->txq_head,
+						      map->qnum, true);
+			if (ret)
+				return ret;
+		}
+	}
+	return 0;
+}
+
+/* Configure the Rx and Tx hairpin queues for the selected port. */
+int
+setup_hairpin_queues(portid_t pi, portid_t p_pi, uint16_t cnt_pi)
+{
+	return !hairpin_multiport_mode ?
+	       setup_legacy_hairpin_queues(pi, p_pi, cnt_pi) :
+	       setup_mapped_harpin_queues(pi);
+}
+
+int
+hairpin_bind(uint16_t cfg_pi, portid_t *pl, portid_t *peer_pl)
+{
+	uint16_t i;
+	portid_t pi;
+	int peer_pi;
+	int diag;
+	int j;
+
+	/* bind all started hairpin ports */
+	for (i = 0; i < cfg_pi; i++) {
+		pi = pl[i];
+		/* bind current Tx to all peer Rx */
+		peer_pi = rte_eth_hairpin_get_peer_ports(pi, peer_pl,
+							 RTE_MAX_ETHPORTS, 1);
+		if (peer_pi < 0)
+			return peer_pi;
+		for (j = 0; j < peer_pi; j++) {
+			if (!port_is_started(peer_pl[j]))
+				continue;
+			diag = rte_eth_hairpin_bind(pi, peer_pl[j]);
+			if (diag < 0) {
+				fprintf(stderr,
+					"Error during binding hairpin Tx port %u to %u: %s\n",
+					pi, peer_pl[j],
+					rte_strerror(-diag));
+				return -1;
+			}
+		}
+		/* bind all peer Tx to current Rx */
+		peer_pi = rte_eth_hairpin_get_peer_ports(pi, peer_pl,
+							 RTE_MAX_ETHPORTS, 0);
+		if (peer_pi < 0)
+			return peer_pi;
+		for (j = 0; j < peer_pi; j++) {
+			if (!port_is_started(peer_pl[j]))
+				continue;
+			diag = rte_eth_hairpin_bind(peer_pl[j], pi);
+			if (diag < 0) {
+				fprintf(stderr,
+					"Error during binding hairpin Tx port %u to %u: %s\n",
+					peer_pl[j], pi,
+					rte_strerror(-diag));
+				return -1;
+			}
+		}
+	}
+	return 0;
+}
+
+void
+hairpin_map_usage(void)
+{
+	printf("  --hairpin-map=rxpi:rxq:txpi:txq:n: hairpin map.\n"
+	       "    rxpi - Rx port index.\n"
+	       "    rxq  - Rx queue.\n"
+	       "    txpi - Tx port index.\n"
+	       "    txq  - Tx queue.\n"
+	       "    n    - hairpin queues number.\n");
+}
+
+static char
+*parse_hairpin_map_entry(char *input, char **next)
+{
+	char *tail = strchr(input, ':');
+
+	if (!tail)
+		return NULL;
+	tail[0] = '\0';
+	*next = tail + 1;
+	return input;
+}
+
+int
+parse_hairpin_map(const char *hpmap)
+{
+	/*
+	 * Testpmd hairpin map format:
+	 * <Rx port id:First Rx queue:Tx port id:First Tx queue:queues number>
+	 */
+	char *head, *next = (char *)(uintptr_t)hpmap;
+	struct hairpin_map *map = calloc(1, sizeof(*map));
+
+	if (!map)
+		return -ENOMEM;
+
+	head = parse_hairpin_map_entry(next, &next);
+	if (!head)
+		goto err;
+	map->rx_port = atoi(head);
+	head = parse_hairpin_map_entry(next, &next);
+	if (!head)
+		goto err;
+	map->rxq_head = atoi(head);
+	head = parse_hairpin_map_entry(next, &next);
+	if (!head)
+		goto err;
+	map->tx_port = atoi(head);
+	head = parse_hairpin_map_entry(next, &next);
+	if (!head)
+		goto err;
+	map->txq_head = atoi(head);
+	map->qnum = atoi(next);
+	hairpin_add_multiport_map(map);
+	return 0;
+err:
+	free(map);
+	return -EINVAL;
+}
+
+
diff --git a/app/test-pmd/meson.build b/app/test-pmd/meson.build
index 719f875be0..f1c36529b4 100644
--- a/app/test-pmd/meson.build
+++ b/app/test-pmd/meson.build
@@ -15,6 +15,7 @@ sources = files(
         'config.c',
         'csumonly.c',
         'flowgen.c',
+        'hairpin.c',
         'icmpecho.c',
         'ieee1588fwd.c',
         'iofwd.c',
diff --git a/app/test-pmd/parameters.c b/app/test-pmd/parameters.c
index 22364e09ab..7b31b94542 100644
--- a/app/test-pmd/parameters.c
+++ b/app/test-pmd/parameters.c
@@ -143,6 +143,8 @@ enum {
 	TESTPMD_OPT_HAIRPINQ_NUM,
 #define TESTPMD_OPT_HAIRPIN_MODE "hairpin-mode"
 	TESTPMD_OPT_HAIRPIN_MODE_NUM,
+#define TESTPMD_OPT_HAIRPIN_MAP "hairpin-map"
+	TESTPMD_OPT_HAIRPIN_MAP_NUM,
 #define TESTPMD_OPT_BURST "burst"
 	TESTPMD_OPT_BURST_NUM,
 #define TESTPMD_OPT_FLOWGEN_CLONES "flowgen-clones"
@@ -317,6 +319,7 @@ static const struct option long_options[] = {
 	REQUIRED_ARG(TESTPMD_OPT_TXD),
 	REQUIRED_ARG(TESTPMD_OPT_HAIRPINQ),
 	REQUIRED_ARG(TESTPMD_OPT_HAIRPIN_MODE),
+	REQUIRED_ARG(TESTPMD_OPT_HAIRPIN_MAP),
 	REQUIRED_ARG(TESTPMD_OPT_BURST),
 	REQUIRED_ARG(TESTPMD_OPT_FLOWGEN_CLONES),
 	REQUIRED_ARG(TESTPMD_OPT_FLOWGEN_FLOWS),
@@ -542,6 +545,7 @@ usage(char* progname)
 	printf("  --hairpin-mode=0xXX: bitmask set the hairpin port mode.\n"
 	       "    0x10 - explicit Tx rule, 0x02 - hairpin ports paired\n"
 	       "    0x01 - hairpin ports loop, 0x00 - hairpin port self\n");
+	hairpin_map_usage();
 }
 
 static int
@@ -1317,6 +1321,12 @@ launch_args_parse(int argc, char** argv)
 				hairpin_mode = (uint32_t)n;
 			break;
 		}
+		case TESTPMD_OPT_HAIRPIN_MAP_NUM:
+			hairpin_multiport_mode = true;
+			ret = parse_hairpin_map(optarg);
+			if (ret)
+				rte_exit(EXIT_FAILURE, "invalid hairpin map\n");
+			break;
 		case TESTPMD_OPT_BURST_NUM:
 			n = atoi(optarg);
 			if (n == 0) {
diff --git a/app/test-pmd/testpmd.c b/app/test-pmd/testpmd.c
index b1401136e4..f487769578 100644
--- a/app/test-pmd/testpmd.c
+++ b/app/test-pmd/testpmd.c
@@ -284,7 +284,6 @@ uint8_t dcb_config = 0;
 /*
  * Configurable number of RX/TX queues.
  */
-queueid_t nb_hairpinq; /**< Number of hairpin queues per port. */
 queueid_t nb_rxq = 1; /**< Number of RX queues per port. */
 queueid_t nb_txq = 1; /**< Number of TX queues per port. */
 
@@ -431,9 +430,6 @@ bool setup_on_probe_event = true;
 /* Clear ptypes on port initialization. */
 uint8_t clear_ptypes = true;
 
-/* Hairpin ports configuration mode. */
-uint32_t hairpin_mode;
-
 /* Pretty printing of ethdev events */
 static const char * const eth_event_desc[] = {
 	[RTE_ETH_EVENT_UNKNOWN] = "unknown",
@@ -1555,54 +1551,6 @@ check_nb_txd(queueid_t txd)
 	return 0;
 }
 
-
-/*
- * Get the allowed maximum number of hairpin queues.
- * *pid return the port id which has minimal value of
- * max_hairpin_queues in all ports.
- */
-queueid_t
-get_allowed_max_nb_hairpinq(portid_t *pid)
-{
-	queueid_t allowed_max_hairpinq = RTE_MAX_QUEUES_PER_PORT;
-	portid_t pi;
-	struct rte_eth_hairpin_cap cap;
-
-	RTE_ETH_FOREACH_DEV(pi) {
-		if (rte_eth_dev_hairpin_capability_get(pi, &cap) != 0) {
-			*pid = pi;
-			return 0;
-		}
-		if (cap.max_nb_queues < allowed_max_hairpinq) {
-			allowed_max_hairpinq = cap.max_nb_queues;
-			*pid = pi;
-		}
-	}
-	return allowed_max_hairpinq;
-}
-
-/*
- * Check input hairpin is valid or not.
- * If input hairpin is not greater than any of maximum number
- * of hairpin queues of all ports, it is valid.
- * if valid, return 0, else return -1
- */
-int
-check_nb_hairpinq(queueid_t hairpinq)
-{
-	queueid_t allowed_max_hairpinq;
-	portid_t pid = 0;
-
-	allowed_max_hairpinq = get_allowed_max_nb_hairpinq(&pid);
-	if (hairpinq > allowed_max_hairpinq) {
-		fprintf(stderr,
-			"Fail: input hairpin (%u) can't be greater than max_hairpin_queues (%u) of port %u\n",
-			hairpinq, allowed_max_hairpinq, pid);
-		return -1;
-	}
-	return 0;
-}
-
 static int
 get_eth_overhead(struct rte_eth_dev_info *dev_info)
 {
@@ -2684,126 +2632,6 @@ port_is_started(portid_t port_id)
 	return 1;
 }
 
-#define HAIRPIN_MODE_RX_FORCE_MEMORY RTE_BIT32(8)
-#define HAIRPIN_MODE_TX_FORCE_MEMORY RTE_BIT32(9)
-
-#define HAIRPIN_MODE_RX_LOCKED_MEMORY RTE_BIT32(12)
-#define HAIRPIN_MODE_RX_RTE_MEMORY RTE_BIT32(13)
-
-#define HAIRPIN_MODE_TX_LOCKED_MEMORY RTE_BIT32(16)
-#define HAIRPIN_MODE_TX_RTE_MEMORY RTE_BIT32(17)
-
-
-/* Configure the Rx and Tx hairpin queues for the selected port. */
-static int
-setup_hairpin_queues(portid_t pi, portid_t p_pi, uint16_t cnt_pi)
-{
-	queueid_t qi;
-	struct rte_eth_hairpin_conf hairpin_conf = {
-		.peer_count = 1,
-	};
-	int i;
-	int diag;
-	struct rte_port *port = &ports[pi];
-	uint16_t peer_rx_port = pi;
-	uint16_t peer_tx_port = pi;
-	uint32_t manual = 1;
-	uint32_t tx_exp = hairpin_mode & 0x10;
-	uint32_t rx_force_memory = hairpin_mode & HAIRPIN_MODE_RX_FORCE_MEMORY;
-	uint32_t rx_locked_memory = hairpin_mode & HAIRPIN_MODE_RX_LOCKED_MEMORY;
-	uint32_t rx_rte_memory = hairpin_mode & HAIRPIN_MODE_RX_RTE_MEMORY;
-	uint32_t tx_force_memory = hairpin_mode & HAIRPIN_MODE_TX_FORCE_MEMORY;
-	uint32_t tx_locked_memory = hairpin_mode & HAIRPIN_MODE_TX_LOCKED_MEMORY;
-	uint32_t tx_rte_memory = hairpin_mode & HAIRPIN_MODE_TX_RTE_MEMORY;
-
-	if (!(hairpin_mode & 0xf)) {
-		peer_rx_port = pi;
-		peer_tx_port = pi;
-		manual = 0;
-	} else if (hairpin_mode & 0x1) {
-		peer_tx_port = rte_eth_find_next_owned_by(pi + 1,
-						       RTE_ETH_DEV_NO_OWNER);
-		if (peer_tx_port >= RTE_MAX_ETHPORTS)
-			peer_tx_port = rte_eth_find_next_owned_by(0,
-						RTE_ETH_DEV_NO_OWNER);
-		if (p_pi != RTE_MAX_ETHPORTS) {
-			peer_rx_port = p_pi;
-		} else {
-			uint16_t next_pi;
-
-			/* Last port will be the peer RX port of the first. */
-			RTE_ETH_FOREACH_DEV(next_pi)
-				peer_rx_port = next_pi;
-		}
-		manual = 1;
-	} else if (hairpin_mode & 0x2) {
-		if (cnt_pi & 0x1) {
-			peer_rx_port = p_pi;
-		} else {
-			peer_rx_port = rte_eth_find_next_owned_by(pi + 1,
-						RTE_ETH_DEV_NO_OWNER);
-			if (peer_rx_port >= RTE_MAX_ETHPORTS)
-				peer_rx_port = pi;
-		}
-		peer_tx_port = peer_rx_port;
-		manual = 1;
-	}
-
-	for (qi = nb_txq, i = 0; qi < nb_hairpinq + nb_txq; qi++) {
-		hairpin_conf.peers[0].port = peer_rx_port;
-		hairpin_conf.peers[0].queue = i + nb_rxq;
-		hairpin_conf.manual_bind = !!manual;
-		hairpin_conf.tx_explicit = !!tx_exp;
-		hairpin_conf.force_memory = !!tx_force_memory;
-		hairpin_conf.use_locked_device_memory = !!tx_locked_memory;
-		hairpin_conf.use_rte_memory = !!tx_rte_memory;
-		diag = rte_eth_tx_hairpin_queue_setup
-			(pi, qi, nb_txd, &hairpin_conf);
-		i++;
-		if (diag == 0)
-			continue;
-
-		/* Fail to setup rx queue, return */
-		if (port->port_status == RTE_PORT_HANDLING)
-			port->port_status = RTE_PORT_STOPPED;
-		else
-			fprintf(stderr,
-				"Port %d can not be set back to stopped\n", pi);
-		fprintf(stderr, "Fail to configure port %d hairpin queues\n",
-			pi);
-		/* try to reconfigure queues next time */
-		port->need_reconfig_queues = 1;
-		return -1;
-	}
-	for (qi = nb_rxq, i = 0; qi < nb_hairpinq + nb_rxq; qi++) {
-		hairpin_conf.peers[0].port = peer_tx_port;
-		hairpin_conf.peers[0].queue = i + nb_txq;
-		hairpin_conf.manual_bind = !!manual;
-		hairpin_conf.tx_explicit = !!tx_exp;
-		hairpin_conf.force_memory = !!rx_force_memory;
-		hairpin_conf.use_locked_device_memory = !!rx_locked_memory;
-		hairpin_conf.use_rte_memory = !!rx_rte_memory;
-		diag = rte_eth_rx_hairpin_queue_setup
-			(pi, qi, nb_rxd, &hairpin_conf);
-		i++;
-		if (diag == 0)
-			continue;
-
-		/* Fail to setup rx queue, return */
-		if (port->port_status == RTE_PORT_HANDLING)
-			port->port_status = RTE_PORT_STOPPED;
-		else
-			fprintf(stderr,
-				"Port %d can not be set back to stopped\n", pi);
-		fprintf(stderr, "Fail to configure port %d hairpin queues\n",
-			pi);
-		/* try to reconfigure queues next time */
-		port->need_reconfig_queues = 1;
-		return -1;
-	}
-	return 0;
-}
-
 /* Configure the Rx with optional split. */
 int
 rx_queue_setup(uint16_t port_id, uint16_t rx_queue_id,
@@ -3043,7 +2871,6 @@ start_port(portid_t pid)
 	portid_t peer_pl[RTE_MAX_ETHPORTS];
 	uint16_t cnt_pi = 0;
 	uint16_t cfg_pi = 0;
-	int peer_pi;
 	queueid_t qi;
 	struct rte_port *port;
 	struct rte_eth_hairpin_cap cap;
@@ -3304,47 +3131,9 @@ start_port(portid_t pid)
 		fprintf(stderr, "Please stop the ports first\n");
 
 	if (hairpin_mode & 0xf) {
-		uint16_t i;
-		int j;
-
-		/* bind all started hairpin ports */
-		for (i = 0; i < cfg_pi; i++) {
-			pi = pl[i];
-			/* bind current Tx to all peer Rx */
-			peer_pi = rte_eth_hairpin_get_peer_ports(pi, peer_pl,
-							RTE_MAX_ETHPORTS, 1);
-			if (peer_pi < 0)
-				return peer_pi;
-			for (j = 0; j < peer_pi; j++) {
-				if (!port_is_started(peer_pl[j]))
-					continue;
-				diag = rte_eth_hairpin_bind(pi, peer_pl[j]);
-				if (diag < 0) {
-					fprintf(stderr,
-						"Error during binding hairpin Tx port %u to %u: %s\n",
-						pi, peer_pl[j],
-						rte_strerror(-diag));
-					return -1;
-				}
-			}
-			/* bind all peer Tx to current Rx */
-			peer_pi = rte_eth_hairpin_get_peer_ports(pi, peer_pl,
-							RTE_MAX_ETHPORTS, 0);
-			if (peer_pi < 0)
-				return peer_pi;
-			for (j = 0; j < peer_pi; j++) {
-				if (!port_is_started(peer_pl[j]))
-					continue;
-				diag = rte_eth_hairpin_bind(peer_pl[j], pi);
-				if (diag < 0) {
-					fprintf(stderr,
-						"Error during binding hairpin Tx port %u to %u: %s\n",
-						peer_pl[j], pi,
-						rte_strerror(-diag));
-					return -1;
-				}
-			}
-		}
+		diag = hairpin_bind(cfg_pi, pl, peer_pl);
+		if (diag < 0)
+			return -1;
 	}
 
 	fill_xstats_display_info_for_port(pid);
diff --git a/app/test-pmd/testpmd.h b/app/test-pmd/testpmd.h
index 9facd7f281..659c5b4fdc 100644
--- a/app/test-pmd/testpmd.h
+++ b/app/test-pmd/testpmd.h
@@ -126,6 +126,16 @@ enum noisy_fwd_mode {
 	NOISY_FWD_MODE_MAX,
 };
 
+/**
+ * Command line arguments parser sets `hairpin_multiport_mode` to True
+ * if explicit hairpin map configuration mode was used.
+ */
+extern bool hairpin_multiport_mode;
+
+/** Hairpin maps list. */
+struct hairpin_map;
+extern void hairpin_add_multiport_map(struct hairpin_map *map);
+
 /**
  * The data structure associated with RX and TX packet burst statistics
  * that are recorded for each forwarding stream.
@@ -1252,6 +1262,10 @@ extern int flow_parse(const char *src, void *result, unsigned int size,
 		      struct rte_flow_attr **attr,
 		      struct rte_flow_item **pattern,
 		      struct rte_flow_action **actions);
+int setup_hairpin_queues(portid_t pi, portid_t p_pi, uint16_t cnt_pi);
+int hairpin_bind(uint16_t cfg_pi, portid_t *pl, portid_t *peer_pl);
+void hairpin_map_usage(void);
+int parse_hairpin_map(const char *hpmap);
 
 uint64_t str_to_rsstypes(const char *str);
 const char *rsstypes_to_str(uint64_t rss_type);
diff --git a/doc/guides/testpmd_app_ug/run_app.rst b/doc/guides/testpmd_app_ug/run_app.rst
index 1a9b812a7f..48717707a7 100644
--- a/doc/guides/testpmd_app_ug/run_app.rst
+++ b/doc/guides/testpmd_app_ug/run_app.rst
@@ -571,6 +571,9 @@ The command line options are:
 
     The default value is 0. Hairpin will use single port mode and implicit Tx flow mode.
 
+*   ``--hairpin-map=Rx port id:Rx queue:Tx port id:Tx queue:queues number``
+
+    Set explicit hairpin configuration.
 
 Testpmd Multi-Process Command-line Options
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-- 
2.43.0

RE: [PATCH v4 1/1] testpmd: add hairpin-map parameter

[Etelson, Gregory]

On Fri, 20 Sep 2024, Singh, Aman Deep wrote:

> External email: Use caution opening links or attachments
>
>
> Hi Gregory,
> Can you please resolve the compilation issues seen with this patch.
>

Hello,

I've posted v5.

Regards,
Gregory

Re: [PATCH v5 1/1] testpmd: add hairpin-map parameter

[Ferruh Yigit]

On 9/22/2024 7:50 AM, Gregory Etelson wrote:
> Hairpin offloads packet forwarding between ports.
> Packet is expected on Rx port <rp>, Rx queue <rq> and is forwarded
> to Tx port <tp> Tx queue <tq>.
> 
> Testpmd implements a static hairpin configuration scheme.
> 
> The new parameter allows explicit selection of Rx and Tx ports and
> queues in hairpin configuration.
> The new `hairpin-map` parameter is provided with 5 parameters,
> separated by `:`
> 
> `--hairpin-map=Rx port id:Rx queue:Tx port id:Tx queue:queues number`
> 
> Testpmd operator can provide several `hairpin-map` parameters for
> different hairpin maps.
> Example:
> 
> dpdk-testpmd <EAL params> -- \
>   <testpmd params> \
>   --rxq=2 --txq=2 --hairpinq=2 --hairpin-mode=0x12 \
>   --hairpin-map=0:2:1:2:1 \ # [1]
>   --hairpin-map=0:3:2:2:3   # [2]
> 
> Hairpin map [1] binds Rx port 0, queue 2 with Tx port 1, queue 2.
> Hairpin map [2] binds
>   Rx port 0, queue 3 with Tx port 2, queue 2,
>   Rx port 0, queue 4 with Tx port 2, queue 3,
>   Rx port 0, queue 5 with Tx port 2, queue 4.
> 
> The new `hairpin-map` parameter is optional.
> If omitted, testpmd will create "default" hairpin maps.
> 
> Signed-off-by: Gregory Etelson <getelson@nvidia.com>
> Acked-by: Dariusz Sosnowski <dsosnowski@nvidia.com>
>

Hi Aman, reminder of this patch.

[PATCH v5 1/1] testpmd: add hairpin-map parameter

[Gregory Etelson]

Hello,

Is this patch scheduled to be merged into DPDK-24.11 ?
What can I do to speed up it's integration ?

Regards,
Gregory

Re: [PATCH v5 1/1] testpmd: add hairpin-map parameter

[Ferruh Yigit]

On 9/22/2024 7:50 AM, Gregory Etelson wrote:
> Hairpin offloads packet forwarding between ports.
> Packet is expected on Rx port <rp>, Rx queue <rq> and is forwarded
> to Tx port <tp> Tx queue <tq>.
> 
> Testpmd implements a static hairpin configuration scheme.
> 
> The new parameter allows explicit selection of Rx and Tx ports and
> queues in hairpin configuration.
> The new `hairpin-map` parameter is provided with 5 parameters,
> separated by `:`
> 
> `--hairpin-map=Rx port id:Rx queue:Tx port id:Tx queue:queues number`
> 
> Testpmd operator can provide several `hairpin-map` parameters for
> different hairpin maps.
> Example:
> 
> dpdk-testpmd <EAL params> -- \
>   <testpmd params> \
>   --rxq=2 --txq=2 --hairpinq=2 --hairpin-mode=0x12 \
>   --hairpin-map=0:2:1:2:1 \ # [1]
>   --hairpin-map=0:3:2:2:3   # [2]
> 
> Hairpin map [1] binds Rx port 0, queue 2 with Tx port 1, queue 2.
> Hairpin map [2] binds
>   Rx port 0, queue 3 with Tx port 2, queue 2,
>   Rx port 0, queue 4 with Tx port 2, queue 3,
>   Rx port 0, queue 5 with Tx port 2, queue 4.
> 
> The new `hairpin-map` parameter is optional.
> If omitted, testpmd will create "default" hairpin maps.
> 
> Signed-off-by: Gregory Etelson <getelson@nvidia.com>
> Acked-by: Dariusz Sosnowski <dsosnowski@nvidia.com>
>

+Praveen
cpfl is only other driver that implements hairpin, @Praveen can you
please help reviewing this patch, and confirm change works fine with
your changes.

@Stephen, if you have any free cycles, can you please help on this patch
too?

Thanks,
ferruh

Re: [PATCH v5 1/1] testpmd: add hairpin-map parameter

[Stephen Hemminger]

On Tue, 29 Oct 2024 21:51:40 +0000
Ferruh Yigit <ferruh.yigit@amd.com> wrote:

> On 9/22/2024 7:50 AM, Gregory Etelson wrote:
> > Hairpin offloads packet forwarding between ports.
> > Packet is expected on Rx port <rp>, Rx queue <rq> and is forwarded
> > to Tx port <tp> Tx queue <tq>.
> > 
> > Testpmd implements a static hairpin configuration scheme.
> > 
> > The new parameter allows explicit selection of Rx and Tx ports and
> > queues in hairpin configuration.
> > The new `hairpin-map` parameter is provided with 5 parameters,
> > separated by `:`
> > 
> > `--hairpin-map=Rx port id:Rx queue:Tx port id:Tx queue:queues number`
> > 
> > Testpmd operator can provide several `hairpin-map` parameters for
> > different hairpin maps.
> > Example:
> > 
> > dpdk-testpmd <EAL params> -- \
> >   <testpmd params> \
> >   --rxq=2 --txq=2 --hairpinq=2 --hairpin-mode=0x12 \
> >   --hairpin-map=0:2:1:2:1 \ # [1]
> >   --hairpin-map=0:3:2:2:3   # [2]
> > 
> > Hairpin map [1] binds Rx port 0, queue 2 with Tx port 1, queue 2.
> > Hairpin map [2] binds
> >   Rx port 0, queue 3 with Tx port 2, queue 2,
> >   Rx port 0, queue 4 with Tx port 2, queue 3,
> >   Rx port 0, queue 5 with Tx port 2, queue 4.
> > 
> > The new `hairpin-map` parameter is optional.
> > If omitted, testpmd will create "default" hairpin maps.
> > 
> > Signed-off-by: Gregory Etelson <getelson@nvidia.com>
> > Acked-by: Dariusz Sosnowski <dsosnowski@nvidia.com>
> >  
> 
> +Praveen
> cpfl is only other driver that implements hairpin, @Praveen can you
> please help reviewing this patch, and confirm change works fine with
> your changes.
> 
> @Stephen, if you have any free cycles, can you please help on this patch
> too?
> 
> Thanks,
> ferruh

Ok, can do semantic review but do not have any working HW with hairpin capability.

[PATCH v6] testpmd: add hairpin map parameter

[Gregory Etelson]

Hairpin offloads packet forwarding between ports.
Packet is expected on Rx port <rp>, Rx queue <rq> and is forwarded
to Tx port <tp> Tx queue <tq>.

Testpmd implements a static hairpin configuration scheme.

The new parameter allows explicit selection of Rx and Tx ports and
queues in hairpin configuration.
The new `hairpin-map` parameter is provided with 5 parameters,
separated by `:`

`--hairpin-map=Rx port id:Rx queue:Tx port id:Tx queue:queues number`

Testpmd operator can provide several `hairpin-map` parameters for
different hairpin maps.
Example:

dpdk-testpmd <EAL params> -- \
  <testpmd params> \
  --rxq=2 --txq=2 --hairpinq=2 --hairpin-mode=0x12 \
  --hairpin-map=0:2:1:2:1 \ # [1]
  --hairpin-map=0:3:2:2:3   # [2]

Hairpin map [1] binds Rx port 0, queue 2 with Tx port 1, queue 2.
Hairpin map [2] binds
  Rx port 0, queue 3 with Tx port 2, queue 2,
  Rx port 0, queue 4 with Tx port 2, queue 3,
  Rx port 0, queue 5 with Tx port 2, queue 4.

The new `hairpin-map` parameter is optional.
If omitted, testpmd will create "default" hairpin maps.

Signed-off-by: Gregory Etelson <getelson@nvidia.com>
Acked-by: Dariusz Sosnowski <dsosnowski@nvidia.com>
---
 app/test-pmd/hairpin.c                | 360 ++++++++++++++++++++++++++
 app/test-pmd/meson.build              |   1 +
 app/test-pmd/parameters.c             |  10 +
 app/test-pmd/testpmd.c                | 217 +---------------
 app/test-pmd/testpmd.h                |  14 +
 doc/guides/testpmd_app_ug/run_app.rst |   3 +
 6 files changed, 391 insertions(+), 214 deletions(-)
 create mode 100644 app/test-pmd/hairpin.c

diff --git a/app/test-pmd/hairpin.c b/app/test-pmd/hairpin.c
new file mode 100644
index 0000000000..6d42fdc088
--- /dev/null
+++ b/app/test-pmd/hairpin.c
@@ -0,0 +1,360 @@
+/* SPDX-License-Identifier: BSD-3-Clause
+ * Copyright (c) 2022 NVIDIA Corporation & Affiliates
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <stdbool.h>
+#include <stdint.h>
+
+#include <sys/queue.h>
+
+#include "testpmd.h"
+
+/* Hairpin ports configuration mode. */
+uint32_t hairpin_mode;
+
+bool hairpin_multiport_mode = false;
+
+queueid_t nb_hairpinq; /**< Number of hairpin queues per port. */
+
+static LIST_HEAD(, hairpin_map) hairpin_map_head = LIST_HEAD_INITIALIZER();
+
+struct hairpin_map {
+	LIST_ENTRY(hairpin_map) entry; /**< List entry. */
+	portid_t rx_port; /**< Hairpin Rx port ID. */
+	portid_t tx_port; /**< Hairpin Tx port ID. */
+	uint16_t rxq_head; /**< Hairpin Rx queue head. */
+	uint16_t txq_head; /**< Hairpin Tx queue head. */
+	uint16_t qnum; /**< Hairpin queues number. */
+};
+
+void
+hairpin_add_multiport_map(struct hairpin_map *map)
+{
+	LIST_INSERT_HEAD(&hairpin_map_head, map, entry);
+}
+
+/*
+ * Get the allowed maximum number of hairpin queues.
+ * *pid return the port id which has minimal value of
+ * max_hairpin_queues in all ports.
+ */
+queueid_t
+get_allowed_max_nb_hairpinq(portid_t *pid)
+{
+	queueid_t allowed_max_hairpinq = RTE_MAX_QUEUES_PER_PORT;
+	portid_t pi;
+	struct rte_eth_hairpin_cap cap;
+
+	RTE_ETH_FOREACH_DEV(pi) {
+		if (rte_eth_dev_hairpin_capability_get(pi, &cap) != 0) {
+			*pid = pi;
+			return 0;
+		}
+		if (cap.max_nb_queues < allowed_max_hairpinq) {
+			allowed_max_hairpinq = cap.max_nb_queues;
+			*pid = pi;
+		}
+	}
+	return allowed_max_hairpinq;
+}
+
+/*
+ * Check input hairpin is valid or not.
+ * If input hairpin is not greater than any of maximum number
+ * of hairpin queues of all ports, it is valid.
+ * if valid, return 0, else return -1
+ */
+int
+check_nb_hairpinq(queueid_t hairpinq)
+{
+	queueid_t allowed_max_hairpinq;
+	portid_t pid = 0;
+
+	allowed_max_hairpinq = get_allowed_max_nb_hairpinq(&pid);
+	if (hairpinq > allowed_max_hairpinq) {
+		fprintf(stderr,
+			"Fail: input hairpin (%u) can't be greater than max_hairpin_queues (%u) of port %u\n",
+			hairpinq, allowed_max_hairpinq, pid);
+		return -1;
+	}
+	return 0;
+}
+
+#define HAIRPIN_MODE_RX_FORCE_MEMORY RTE_BIT32(8)
+#define HAIRPIN_MODE_TX_FORCE_MEMORY RTE_BIT32(9)
+
+#define HAIRPIN_MODE_RX_LOCKED_MEMORY RTE_BIT32(12)
+#define HAIRPIN_MODE_RX_RTE_MEMORY RTE_BIT32(13)
+
+#define HAIRPIN_MODE_TX_LOCKED_MEMORY RTE_BIT32(16)
+#define HAIRPIN_MODE_TX_RTE_MEMORY RTE_BIT32(17)
+
+static int
+port_config_hairpin_rxq(portid_t pi, uint16_t peer_tx_port,
+			queueid_t rxq_head, queueid_t txq_head,
+			uint16_t qcount, uint32_t manual_bind)
+{
+	int diag;
+	queueid_t i, qi;
+	uint32_t tx_explicit = !!(hairpin_mode & 0x10);
+	uint32_t force_mem = !!(hairpin_mode & HAIRPIN_MODE_RX_FORCE_MEMORY);
+	uint32_t locked_mem = !!(hairpin_mode & HAIRPIN_MODE_RX_LOCKED_MEMORY);
+	uint32_t rte_mem = !!(hairpin_mode & HAIRPIN_MODE_RX_RTE_MEMORY);
+	struct rte_port *port = &ports[pi];
+	struct rte_eth_hairpin_conf hairpin_conf = {
+		.peer_count = 1,
+	};
+
+	for (qi = rxq_head, i = 0; qi < rxq_head + qcount; qi++) {
+		hairpin_conf.peers[0].port = peer_tx_port;
+		hairpin_conf.peers[0].queue = i + txq_head;
+		hairpin_conf.manual_bind = manual_bind;
+		hairpin_conf.tx_explicit = tx_explicit;
+		hairpin_conf.force_memory = force_mem;
+		hairpin_conf.use_locked_device_memory = locked_mem;
+		hairpin_conf.use_rte_memory = rte_mem;
+		diag = rte_eth_rx_hairpin_queue_setup
+			(pi, qi, nb_rxd, &hairpin_conf);
+		i++;
+		if (diag == 0)
+			continue;
+
+		/* Fail to setup rx queue, return */
+		if (port->port_status == RTE_PORT_HANDLING)
+			port->port_status = RTE_PORT_STOPPED;
+		else
+			fprintf(stderr,
+				"Port %d can not be set back to stopped\n", pi);
+		fprintf(stderr,
+			"Port %u failed to configure hairpin on rxq %u.\n"
+			"Peer port: %u peer txq: %u\n",
+			pi, qi, peer_tx_port, i);
+		/* try to reconfigure queues next time */
+		port->need_reconfig_queues = 1;
+		return -1;
+	}
+	return 0;
+}
+
+static int
+port_config_hairpin_txq(portid_t pi, uint16_t peer_rx_port,
+			queueid_t rxq_head, queueid_t txq_head,
+			uint16_t qcount, uint32_t manual_bind)
+{
+	int diag;
+	queueid_t i, qi;
+	uint32_t tx_explicit = !!(hairpin_mode & 0x10);
+	uint32_t force_mem = !!(hairpin_mode & HAIRPIN_MODE_TX_FORCE_MEMORY);
+	uint32_t locked_mem = !!(hairpin_mode & HAIRPIN_MODE_TX_LOCKED_MEMORY);
+	uint32_t rte_mem = !!(hairpin_mode & HAIRPIN_MODE_TX_RTE_MEMORY);
+	struct rte_port *port = &ports[pi];
+	struct rte_eth_hairpin_conf hairpin_conf = {
+		.peer_count = 1,
+	};
+
+	for (qi = txq_head, i = 0; qi < txq_head + qcount; qi++) {
+		hairpin_conf.peers[0].port = peer_rx_port;
+		hairpin_conf.peers[0].queue = i + rxq_head;
+		hairpin_conf.manual_bind = manual_bind;
+		hairpin_conf.tx_explicit = tx_explicit;
+		hairpin_conf.force_memory = force_mem;
+		hairpin_conf.use_locked_device_memory = locked_mem;
+		hairpin_conf.use_rte_memory = rte_mem;
+		diag = rte_eth_tx_hairpin_queue_setup
+			(pi, qi, nb_txd, &hairpin_conf);
+		i++;
+		if (diag == 0)
+			continue;
+
+		/* Fail to setup rx queue, return */
+		if (port->port_status == RTE_PORT_HANDLING)
+			port->port_status = RTE_PORT_STOPPED;
+		else
+			fprintf(stderr,
+				"Port %d can not be set back to stopped\n", pi);
+		fprintf(stderr,
+			"Port %d failed to configure hairpin on txq %u.\n"
+			"Peer port: %u peer rxq: %u\n",
+			pi, qi, peer_rx_port, i);
+		/* try to reconfigure queues next time */
+		port->need_reconfig_queues = 1;
+		return -1;
+	}
+	return 0;
+}
+
+static int
+setup_legacy_hairpin_queues(portid_t pi, portid_t p_pi, uint16_t cnt_pi)
+{
+	int diag;
+	uint16_t peer_rx_port = pi;
+	uint16_t peer_tx_port = pi;
+	uint32_t manual = 1;
+
+	if (!(hairpin_mode & 0xf)) {
+		peer_rx_port = pi;
+		peer_tx_port = pi;
+		manual = 0;
+	} else if (hairpin_mode & 0x1) {
+		peer_tx_port = rte_eth_find_next_owned_by(pi + 1,
+							  RTE_ETH_DEV_NO_OWNER);
+		if (peer_tx_port >= RTE_MAX_ETHPORTS)
+			peer_tx_port = rte_eth_find_next_owned_by(0,
+								  RTE_ETH_DEV_NO_OWNER);
+		if (p_pi != RTE_MAX_ETHPORTS) {
+			peer_rx_port = p_pi;
+		} else {
+			uint16_t next_pi;
+
+			/* Last port will be the peer RX port of the first. */
+			RTE_ETH_FOREACH_DEV(next_pi)
+				peer_rx_port = next_pi;
+		}
+		manual = 1;
+	} else if (hairpin_mode & 0x2) {
+		if (cnt_pi & 0x1) {
+			peer_rx_port = p_pi;
+		} else {
+			peer_rx_port = rte_eth_find_next_owned_by(pi + 1,
+								  RTE_ETH_DEV_NO_OWNER);
+			if (peer_rx_port >= RTE_MAX_ETHPORTS)
+				peer_rx_port = pi;
+		}
+		peer_tx_port = peer_rx_port;
+		manual = 1;
+	}
+	diag = port_config_hairpin_txq(pi, peer_rx_port, nb_rxq, nb_txq,
+				       nb_hairpinq, manual);
+	if (diag)
+		return diag;
+	diag = port_config_hairpin_rxq(pi, peer_tx_port, nb_rxq, nb_txq,
+				       nb_hairpinq, manual);
+	if (diag)
+		return diag;
+	return 0;
+}
+
+static int
+setup_mapped_harpin_queues(portid_t pi)
+{
+	int ret = 0;
+	struct hairpin_map *map;
+
+	LIST_FOREACH(map, &hairpin_map_head, entry) {
+		if (map->rx_port == pi) {
+			ret = port_config_hairpin_rxq(pi, map->tx_port,
+						      map->rxq_head,
+						      map->txq_head,
+						      map->qnum, true);
+			if (ret)
+				return ret;
+		}
+		if (map->tx_port == pi) {
+			ret = port_config_hairpin_txq(pi, map->rx_port,
+						      map->rxq_head,
+						      map->txq_head,
+						      map->qnum, true);
+			if (ret)
+				return ret;
+		}
+	}
+	return 0;
+}
+
+/* Configure the Rx and Tx hairpin queues for the selected port. */
+int
+setup_hairpin_queues(portid_t pi, portid_t p_pi, uint16_t cnt_pi)
+{
+	if (hairpin_multiport_mode)
+		return setup_mapped_harpin_queues(pi);
+
+	return setup_legacy_hairpin_queues(pi, p_pi, cnt_pi);
+}
+
+int
+hairpin_bind(uint16_t cfg_pi, portid_t *pl, portid_t *peer_pl)
+{
+	uint16_t i;
+	portid_t pi;
+	int peer_pi;
+	int diag;
+	int j;
+
+	/* bind all started hairpin ports */
+	for (i = 0; i < cfg_pi; i++) {
+		pi = pl[i];
+		/* bind current Tx to all peer Rx */
+		peer_pi = rte_eth_hairpin_get_peer_ports(pi, peer_pl,
+							 RTE_MAX_ETHPORTS, 1);
+		if (peer_pi < 0)
+			return peer_pi;
+		for (j = 0; j < peer_pi; j++) {
+			if (!port_is_started(peer_pl[j]))
+				continue;
+			diag = rte_eth_hairpin_bind(pi, peer_pl[j]);
+			if (diag < 0) {
+				fprintf(stderr,
+					"Error during binding hairpin Tx port %u to %u: %s\n",
+					pi, peer_pl[j],
+					rte_strerror(-diag));
+				return -1;
+			}
+		}
+		/* bind all peer Tx to current Rx */
+		peer_pi = rte_eth_hairpin_get_peer_ports(pi, peer_pl,
+							 RTE_MAX_ETHPORTS, 0);
+		if (peer_pi < 0)
+			return peer_pi;
+		for (j = 0; j < peer_pi; j++) {
+			if (!port_is_started(peer_pl[j]))
+				continue;
+			diag = rte_eth_hairpin_bind(peer_pl[j], pi);
+			if (diag < 0) {
+				fprintf(stderr,
+					"Error during binding hairpin Tx port %u to %u: %s\n",
+					peer_pl[j], pi,
+					rte_strerror(-diag));
+				return -1;
+			}
+		}
+	}
+	return 0;
+}
+
+void
+hairpin_map_usage(void)
+{
+	printf("  --hairpin-map=rxpi:rxq:txpi:txq:n: hairpin map.\n"
+	       "    rxpi - Rx port index.\n"
+	       "    rxq  - Rx queue.\n"
+	       "    txpi - Tx port index.\n"
+	       "    txq  - Tx queue.\n"
+	       "    n    - hairpin queues number.\n");
+}
+
+int
+parse_hairpin_map(const char *hpmap)
+{
+	/*
+	 * Testpmd hairpin map format:
+	 * <Rx port id:First Rx queue:Tx port id:First Tx queue:queues number>
+	 */
+	int ret;
+	struct hairpin_map *map = calloc(1, sizeof(*map));
+
+	if (!map)
+		return -ENOMEM;
+
+	ret = sscanf(hpmap, "%hu:%hu:%hu:%hu:%hu",
+		     &map->rx_port, &map->rxq_head,
+		     &map->tx_port, &map->txq_head, &map->qnum);
+	if (ret != 5) {
+		free(map);
+		return -EINVAL;
+	}
+	hairpin_add_multiport_map(map);
+	return 0;
+}
diff --git a/app/test-pmd/meson.build b/app/test-pmd/meson.build
index 719f875be0..f1c36529b4 100644
--- a/app/test-pmd/meson.build
+++ b/app/test-pmd/meson.build
@@ -15,6 +15,7 @@ sources = files(
         'config.c',
         'csumonly.c',
         'flowgen.c',
+        'hairpin.c',
         'icmpecho.c',
         'ieee1588fwd.c',
         'iofwd.c',
diff --git a/app/test-pmd/parameters.c b/app/test-pmd/parameters.c
index 22364e09ab..7b31b94542 100644
--- a/app/test-pmd/parameters.c
+++ b/app/test-pmd/parameters.c
@@ -143,6 +143,8 @@ enum {
 	TESTPMD_OPT_HAIRPINQ_NUM,
 #define TESTPMD_OPT_HAIRPIN_MODE "hairpin-mode"
 	TESTPMD_OPT_HAIRPIN_MODE_NUM,
+#define TESTPMD_OPT_HAIRPIN_MAP "hairpin-map"
+	TESTPMD_OPT_HAIRPIN_MAP_NUM,
 #define TESTPMD_OPT_BURST "burst"
 	TESTPMD_OPT_BURST_NUM,
 #define TESTPMD_OPT_FLOWGEN_CLONES "flowgen-clones"
@@ -317,6 +319,7 @@ static const struct option long_options[] = {
 	REQUIRED_ARG(TESTPMD_OPT_TXD),
 	REQUIRED_ARG(TESTPMD_OPT_HAIRPINQ),
 	REQUIRED_ARG(TESTPMD_OPT_HAIRPIN_MODE),
+	REQUIRED_ARG(TESTPMD_OPT_HAIRPIN_MAP),
 	REQUIRED_ARG(TESTPMD_OPT_BURST),
 	REQUIRED_ARG(TESTPMD_OPT_FLOWGEN_CLONES),
 	REQUIRED_ARG(TESTPMD_OPT_FLOWGEN_FLOWS),
@@ -542,6 +545,7 @@ usage(char* progname)
 	printf("  --hairpin-mode=0xXX: bitmask set the hairpin port mode.\n"
 	       "    0x10 - explicit Tx rule, 0x02 - hairpin ports paired\n"
 	       "    0x01 - hairpin ports loop, 0x00 - hairpin port self\n");
+	hairpin_map_usage();
 }
 
 static int
@@ -1317,6 +1321,12 @@ launch_args_parse(int argc, char** argv)
 				hairpin_mode = (uint32_t)n;
 			break;
 		}
+		case TESTPMD_OPT_HAIRPIN_MAP_NUM:
+			hairpin_multiport_mode = true;
+			ret = parse_hairpin_map(optarg);
+			if (ret)
+				rte_exit(EXIT_FAILURE, "invalid hairpin map\n");
+			break;
 		case TESTPMD_OPT_BURST_NUM:
 			n = atoi(optarg);
 			if (n == 0) {
diff --git a/app/test-pmd/testpmd.c b/app/test-pmd/testpmd.c
index b1401136e4..f487769578 100644
--- a/app/test-pmd/testpmd.c
+++ b/app/test-pmd/testpmd.c
@@ -284,7 +284,6 @@ uint8_t dcb_config = 0;
 /*
  * Configurable number of RX/TX queues.
  */
-queueid_t nb_hairpinq; /**< Number of hairpin queues per port. */
 queueid_t nb_rxq = 1; /**< Number of RX queues per port. */
 queueid_t nb_txq = 1; /**< Number of TX queues per port. */
 
@@ -431,9 +430,6 @@ bool setup_on_probe_event = true;
 /* Clear ptypes on port initialization. */
 uint8_t clear_ptypes = true;
 
-/* Hairpin ports configuration mode. */
-uint32_t hairpin_mode;
-
 /* Pretty printing of ethdev events */
 static const char * const eth_event_desc[] = {
 	[RTE_ETH_EVENT_UNKNOWN] = "unknown",
@@ -1555,54 +1551,6 @@ check_nb_txd(queueid_t txd)
 	return 0;
 }
 
-
-/*
- * Get the allowed maximum number of hairpin queues.
- * *pid return the port id which has minimal value of
- * max_hairpin_queues in all ports.
- */
-queueid_t
-get_allowed_max_nb_hairpinq(portid_t *pid)
-{
-	queueid_t allowed_max_hairpinq = RTE_MAX_QUEUES_PER_PORT;
-	portid_t pi;
-	struct rte_eth_hairpin_cap cap;
-
-	RTE_ETH_FOREACH_DEV(pi) {
-		if (rte_eth_dev_hairpin_capability_get(pi, &cap) != 0) {
-			*pid = pi;
-			return 0;
-		}
-		if (cap.max_nb_queues < allowed_max_hairpinq) {
-			allowed_max_hairpinq = cap.max_nb_queues;
-			*pid = pi;
-		}
-	}
-	return allowed_max_hairpinq;
-}
-
-/*
- * Check input hairpin is valid or not.
- * If input hairpin is not greater than any of maximum number
- * of hairpin queues of all ports, it is valid.
- * if valid, return 0, else return -1
- */
-int
-check_nb_hairpinq(queueid_t hairpinq)
-{
-	queueid_t allowed_max_hairpinq;
-	portid_t pid = 0;
-
-	allowed_max_hairpinq = get_allowed_max_nb_hairpinq(&pid);
-	if (hairpinq > allowed_max_hairpinq) {
-		fprintf(stderr,
-			"Fail: input hairpin (%u) can't be greater than max_hairpin_queues (%u) of port %u\n",
-			hairpinq, allowed_max_hairpinq, pid);
-		return -1;
-	}
-	return 0;
-}
-
 static int
 get_eth_overhead(struct rte_eth_dev_info *dev_info)
 {
@@ -2684,126 +2632,6 @@ port_is_started(portid_t port_id)
 	return 1;
 }
 
-#define HAIRPIN_MODE_RX_FORCE_MEMORY RTE_BIT32(8)
-#define HAIRPIN_MODE_TX_FORCE_MEMORY RTE_BIT32(9)
-
-#define HAIRPIN_MODE_RX_LOCKED_MEMORY RTE_BIT32(12)
-#define HAIRPIN_MODE_RX_RTE_MEMORY RTE_BIT32(13)
-
-#define HAIRPIN_MODE_TX_LOCKED_MEMORY RTE_BIT32(16)
-#define HAIRPIN_MODE_TX_RTE_MEMORY RTE_BIT32(17)
-
-
-/* Configure the Rx and Tx hairpin queues for the selected port. */
-static int
-setup_hairpin_queues(portid_t pi, portid_t p_pi, uint16_t cnt_pi)
-{
-	queueid_t qi;
-	struct rte_eth_hairpin_conf hairpin_conf = {
-		.peer_count = 1,
-	};
-	int i;
-	int diag;
-	struct rte_port *port = &ports[pi];
-	uint16_t peer_rx_port = pi;
-	uint16_t peer_tx_port = pi;
-	uint32_t manual = 1;
-	uint32_t tx_exp = hairpin_mode & 0x10;
-	uint32_t rx_force_memory = hairpin_mode & HAIRPIN_MODE_RX_FORCE_MEMORY;
-	uint32_t rx_locked_memory = hairpin_mode & HAIRPIN_MODE_RX_LOCKED_MEMORY;
-	uint32_t rx_rte_memory = hairpin_mode & HAIRPIN_MODE_RX_RTE_MEMORY;
-	uint32_t tx_force_memory = hairpin_mode & HAIRPIN_MODE_TX_FORCE_MEMORY;
-	uint32_t tx_locked_memory = hairpin_mode & HAIRPIN_MODE_TX_LOCKED_MEMORY;
-	uint32_t tx_rte_memory = hairpin_mode & HAIRPIN_MODE_TX_RTE_MEMORY;
-
-	if (!(hairpin_mode & 0xf)) {
-		peer_rx_port = pi;
-		peer_tx_port = pi;
-		manual = 0;
-	} else if (hairpin_mode & 0x1) {
-		peer_tx_port = rte_eth_find_next_owned_by(pi + 1,
-						       RTE_ETH_DEV_NO_OWNER);
-		if (peer_tx_port >= RTE_MAX_ETHPORTS)
-			peer_tx_port = rte_eth_find_next_owned_by(0,
-						RTE_ETH_DEV_NO_OWNER);
-		if (p_pi != RTE_MAX_ETHPORTS) {
-			peer_rx_port = p_pi;
-		} else {
-			uint16_t next_pi;
-
-			/* Last port will be the peer RX port of the first. */
-			RTE_ETH_FOREACH_DEV(next_pi)
-				peer_rx_port = next_pi;
-		}
-		manual = 1;
-	} else if (hairpin_mode & 0x2) {
-		if (cnt_pi & 0x1) {
-			peer_rx_port = p_pi;
-		} else {
-			peer_rx_port = rte_eth_find_next_owned_by(pi + 1,
-						RTE_ETH_DEV_NO_OWNER);
-			if (peer_rx_port >= RTE_MAX_ETHPORTS)
-				peer_rx_port = pi;
-		}
-		peer_tx_port = peer_rx_port;
-		manual = 1;
-	}
-
-	for (qi = nb_txq, i = 0; qi < nb_hairpinq + nb_txq; qi++) {
-		hairpin_conf.peers[0].port = peer_rx_port;
-		hairpin_conf.peers[0].queue = i + nb_rxq;
-		hairpin_conf.manual_bind = !!manual;
-		hairpin_conf.tx_explicit = !!tx_exp;
-		hairpin_conf.force_memory = !!tx_force_memory;
-		hairpin_conf.use_locked_device_memory = !!tx_locked_memory;
-		hairpin_conf.use_rte_memory = !!tx_rte_memory;
-		diag = rte_eth_tx_hairpin_queue_setup
-			(pi, qi, nb_txd, &hairpin_conf);
-		i++;
-		if (diag == 0)
-			continue;
-
-		/* Fail to setup rx queue, return */
-		if (port->port_status == RTE_PORT_HANDLING)
-			port->port_status = RTE_PORT_STOPPED;
-		else
-			fprintf(stderr,
-				"Port %d can not be set back to stopped\n", pi);
-		fprintf(stderr, "Fail to configure port %d hairpin queues\n",
-			pi);
-		/* try to reconfigure queues next time */
-		port->need_reconfig_queues = 1;
-		return -1;
-	}
-	for (qi = nb_rxq, i = 0; qi < nb_hairpinq + nb_rxq; qi++) {
-		hairpin_conf.peers[0].port = peer_tx_port;
-		hairpin_conf.peers[0].queue = i + nb_txq;
-		hairpin_conf.manual_bind = !!manual;
-		hairpin_conf.tx_explicit = !!tx_exp;
-		hairpin_conf.force_memory = !!rx_force_memory;
-		hairpin_conf.use_locked_device_memory = !!rx_locked_memory;
-		hairpin_conf.use_rte_memory = !!rx_rte_memory;
-		diag = rte_eth_rx_hairpin_queue_setup
-			(pi, qi, nb_rxd, &hairpin_conf);
-		i++;
-		if (diag == 0)
-			continue;
-
-		/* Fail to setup rx queue, return */
-		if (port->port_status == RTE_PORT_HANDLING)
-			port->port_status = RTE_PORT_STOPPED;
-		else
-			fprintf(stderr,
-				"Port %d can not be set back to stopped\n", pi);
-		fprintf(stderr, "Fail to configure port %d hairpin queues\n",
-			pi);
-		/* try to reconfigure queues next time */
-		port->need_reconfig_queues = 1;
-		return -1;
-	}
-	return 0;
-}
-
 /* Configure the Rx with optional split. */
 int
 rx_queue_setup(uint16_t port_id, uint16_t rx_queue_id,
@@ -3043,7 +2871,6 @@ start_port(portid_t pid)
 	portid_t peer_pl[RTE_MAX_ETHPORTS];
 	uint16_t cnt_pi = 0;
 	uint16_t cfg_pi = 0;
-	int peer_pi;
 	queueid_t qi;
 	struct rte_port *port;
 	struct rte_eth_hairpin_cap cap;
@@ -3304,47 +3131,9 @@ start_port(portid_t pid)
 		fprintf(stderr, "Please stop the ports first\n");
 
 	if (hairpin_mode & 0xf) {
-		uint16_t i;
-		int j;
-
-		/* bind all started hairpin ports */
-		for (i = 0; i < cfg_pi; i++) {
-			pi = pl[i];
-			/* bind current Tx to all peer Rx */
-			peer_pi = rte_eth_hairpin_get_peer_ports(pi, peer_pl,
-							RTE_MAX_ETHPORTS, 1);
-			if (peer_pi < 0)
-				return peer_pi;
-			for (j = 0; j < peer_pi; j++) {
-				if (!port_is_started(peer_pl[j]))
-					continue;
-				diag = rte_eth_hairpin_bind(pi, peer_pl[j]);
-				if (diag < 0) {
-					fprintf(stderr,
-						"Error during binding hairpin Tx port %u to %u: %s\n",
-						pi, peer_pl[j],
-						rte_strerror(-diag));
-					return -1;
-				}
-			}
-			/* bind all peer Tx to current Rx */
-			peer_pi = rte_eth_hairpin_get_peer_ports(pi, peer_pl,
-							RTE_MAX_ETHPORTS, 0);
-			if (peer_pi < 0)
-				return peer_pi;
-			for (j = 0; j < peer_pi; j++) {
-				if (!port_is_started(peer_pl[j]))
-					continue;
-				diag = rte_eth_hairpin_bind(peer_pl[j], pi);
-				if (diag < 0) {
-					fprintf(stderr,
-						"Error during binding hairpin Tx port %u to %u: %s\n",
-						peer_pl[j], pi,
-						rte_strerror(-diag));
-					return -1;
-				}
-			}
-		}
+		diag = hairpin_bind(cfg_pi, pl, peer_pl);
+		if (diag < 0)
+			return -1;
 	}
 
 	fill_xstats_display_info_for_port(pid);
diff --git a/app/test-pmd/testpmd.h b/app/test-pmd/testpmd.h
index 131ea53f84..314482e69c 100644
--- a/app/test-pmd/testpmd.h
+++ b/app/test-pmd/testpmd.h
@@ -126,6 +126,16 @@ enum noisy_fwd_mode {
 	NOISY_FWD_MODE_MAX,
 };
 
+/**
+ * Command line arguments parser sets `hairpin_multiport_mode` to True
+ * if explicit hairpin map configuration mode was used.
+ */
+extern bool hairpin_multiport_mode;
+
+/** Hairpin maps list. */
+struct hairpin_map;
+extern void hairpin_add_multiport_map(struct hairpin_map *map);
+
 /**
  * The data structure associated with RX and TX packet burst statistics
  * that are recorded for each forwarding stream.
@@ -1255,6 +1265,10 @@ extern int flow_parse(const char *src, void *result, unsigned int size,
 		      struct rte_flow_attr **attr,
 		      struct rte_flow_item **pattern,
 		      struct rte_flow_action **actions);
+int setup_hairpin_queues(portid_t pi, portid_t p_pi, uint16_t cnt_pi);
+int hairpin_bind(uint16_t cfg_pi, portid_t *pl, portid_t *peer_pl);
+void hairpin_map_usage(void);
+int parse_hairpin_map(const char *hpmap);
 
 uint64_t str_to_rsstypes(const char *str);
 const char *rsstypes_to_str(uint64_t rss_type);
diff --git a/doc/guides/testpmd_app_ug/run_app.rst b/doc/guides/testpmd_app_ug/run_app.rst
index 1a9b812a7f..48717707a7 100644
--- a/doc/guides/testpmd_app_ug/run_app.rst
+++ b/doc/guides/testpmd_app_ug/run_app.rst
@@ -571,6 +571,9 @@ The command line options are:
 
     The default value is 0. Hairpin will use single port mode and implicit Tx flow mode.
 
+*   ``--hairpin-map=Rx port id:Rx queue:Tx port id:Tx queue:queues number``
+
+    Set explicit hairpin configuration.
 
 Testpmd Multi-Process Command-line Options
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-- 
2.43.0

Re: [PATCH v6] testpmd: add hairpin map parameter

[Stephen Hemminger]

On Wed, 30 Oct 2024 09:37:24 +0200
Gregory Etelson <getelson@nvidia.com> wrote:

> +static int
> +port_config_hairpin_rxq(portid_t pi, uint16_t peer_tx_port,
> +			queueid_t rxq_head, queueid_t txq_head,
> +			uint16_t qcount, uint32_t manual_bind)
> +{
> +	int diag;
> +	queueid_t i, qi;
> +	uint32_t tx_explicit = !!(hairpin_mode & 0x10);
> +	uint32_t force_mem = !!(hairpin_mode & HAIRPIN_MODE_RX_FORCE_MEMORY);
> +	uint32_t locked_mem = !!(hairpin_mode & HAIRPIN_MODE_RX_LOCKED_MEMORY);
> +	uint32_t rte_mem = !!(hairpin_mode & HAIRPIN_MODE_RX_RTE_MEMORY);
> +	struct rte_port *port = &ports[pi];
> +	struct rte_eth_hairpin_conf hairpin_conf = {
> +		.peer_count = 1,
> +	};
> +
> +	for (qi = rxq_head, i = 0; qi < rxq_head + qcount; qi++) {
> +		hairpin_conf.peers[0].port = peer_tx_port;
> +		hairpin_conf.peers[0].queue = i + txq_head;
> +		hairpin_conf.manual_bind = manual_bind;
> +		hairpin_conf.tx_explicit = tx_explicit;
> +		hairpin_conf.force_memory = force_mem;
> +		hairpin_conf.use_locked_device_memory = locked_mem;
> +		hairpin_conf.use_rte_memory = rte_mem;

Did you consider that most of the hairpin_conf values could just be set directly,
avoiding intermediate variables.


> +		diag = rte_eth_rx_hairpin_queue_setup
> +			(pi, qi, nb_rxd, &hairpin_conf);

Unnecessary line break.

Suggestion:

static int
port_config_hairpin_rxq(portid_t pi, uint16_t peer_tx_port,
			queueid_t rxq_head, queueid_t txq_head,
			uint16_t qcount, uint32_t manual_bind)
{
	int diag;
	queueid_t i, qi;
	struct rte_port *port = &ports[pi];
	struct rte_eth_hairpin_conf hairpin_conf = {
		.peer_count = 1,
		.manual_bind = manual_bind,
		.tx_explicit = !!(hairpin_mode & 0x10),
		.force_memory = !!(hairpin_mode & HAIRPIN_MODE_RX_FORCE_MEMORY),
		.use_locked_device_memory = !!(hairpin_mode & HAIRPIN_MODE_RX_LOCKED_MEMORY),
		.use_rte_memory = !!(hairpin_mode & HAIRPIN_MODE_RX_RTE_MEMORY),
	};

	for (qi = rxq_head, i = 0; qi < rxq_head + qcount; qi++) {
		hairpin_conf.peers[0].port = peer_tx_port;
		hairpin_conf.peers[0].queue = i + txq_head;

		diag = rte_eth_rx_hairpin_queue_setup(pi, qi, nb_rxd, &hairpin_conf);
		i++;
		if (diag == 0)
			continue;