8. Flow Classify Sample Application

The Flow Classify sample application is based on the simple skeleton example of a forwarding application.

It is intended as a demonstration of the basic components of a DPDK forwarding application which uses the Flow Classify library API’s.

Please refer to the Flow Classification Library for more information.

8.1. Compiling the Application

To compile the sample application see Compiling the Sample Applications.

The application is located in the flow_classify sub-directory.

8.2. Running the Application

To run the example in a linux environment:

./<build_dir>/examples/dpdk-flow_classify -c 4 -n 4 -- /
--rule_ipv4="../ipv4_rules_file.txt"

Please refer to the DPDK Getting Started Guide, section Running Sample Applications for general information on running applications and the Environment Abstraction Layer (EAL) options.

8.3. Sample ipv4_rules_file.txt

#file format:
#src_ip/masklen dst_ip/masklen src_port : mask dst_port : mask proto/mask priority
#
2.2.2.3/24 2.2.2.7/24 32 : 0xffff 33 : 0xffff 17/0xff 0
9.9.9.3/24 9.9.9.7/24 32 : 0xffff 33 : 0xffff 17/0xff 1
9.9.9.3/24 9.9.9.7/24 32 : 0xffff 33 : 0xffff 6/0xff 2
9.9.8.3/24 9.9.8.7/24 32 : 0xffff 33 : 0xffff 6/0xff 3
6.7.8.9/24 2.3.4.5/24 32 : 0x0000 33 : 0x0000 132/0xff 4

8.4. Explanation

The following sections provide an explanation of the main components of the code.

All DPDK library functions used in the sample code are prefixed with rte_ and are explained in detail in the DPDK API Documentation.

8.4.1. ACL field definitions for the IPv4 5 tuple rule

The following field definitions are used when creating the ACL table during initialisation of the Flow Classify application


/* ACL field definitions for IPv4 5 tuple rule */
enum {
	PROTO_FIELD_IPV4,
	SRC_FIELD_IPV4,
	DST_FIELD_IPV4,
	SRCP_FIELD_IPV4,
	DSTP_FIELD_IPV4,
	NUM_FIELDS_IPV4
};

enum {
	PROTO_INPUT_IPV4,
	SRC_INPUT_IPV4,
	DST_INPUT_IPV4,
	SRCP_DESTP_INPUT_IPV4
};

static struct rte_acl_field_def ipv4_defs[NUM_FIELDS_IPV4] = {
	/* first input field - always one byte long. */
	{
		.type = RTE_ACL_FIELD_TYPE_BITMASK,
		.size = sizeof(uint8_t),
		.field_index = PROTO_FIELD_IPV4,
		.input_index = PROTO_INPUT_IPV4,
		.offset = sizeof(struct rte_ether_hdr) +
			offsetof(struct rte_ipv4_hdr, next_proto_id),
	},
	/* next input field (IPv4 source address) - 4 consecutive bytes. */
	{
		/* rte_flow uses a bit mask for IPv4 addresses */
		.type = RTE_ACL_FIELD_TYPE_BITMASK,
		.size = sizeof(uint32_t),
		.field_index = SRC_FIELD_IPV4,
		.input_index = SRC_INPUT_IPV4,
		.offset = sizeof(struct rte_ether_hdr) +
			offsetof(struct rte_ipv4_hdr, src_addr),
	},
	/* next input field (IPv4 destination address) - 4 consecutive bytes. */
	{
		/* rte_flow uses a bit mask for IPv4 addresses */
		.type = RTE_ACL_FIELD_TYPE_BITMASK,
		.size = sizeof(uint32_t),
		.field_index = DST_FIELD_IPV4,
		.input_index = DST_INPUT_IPV4,
		.offset = sizeof(struct rte_ether_hdr) +
			offsetof(struct rte_ipv4_hdr, dst_addr),
	},
	/*
	 * Next 2 fields (src & dst ports) form 4 consecutive bytes.
	 * They share the same input index.
	 */
	{
		/* rte_flow uses a bit mask for protocol ports */
		.type = RTE_ACL_FIELD_TYPE_BITMASK,
		.size = sizeof(uint16_t),
		.field_index = SRCP_FIELD_IPV4,
		.input_index = SRCP_DESTP_INPUT_IPV4,
		.offset = sizeof(struct rte_ether_hdr) +
			sizeof(struct rte_ipv4_hdr) +
			offsetof(struct rte_tcp_hdr, src_port),
	},
	{
		/* rte_flow uses a bit mask for protocol ports */
		.type = RTE_ACL_FIELD_TYPE_BITMASK,
		.size = sizeof(uint16_t),
		.field_index = DSTP_FIELD_IPV4,
		.input_index = SRCP_DESTP_INPUT_IPV4,
		.offset = sizeof(struct rte_ether_hdr) +
			sizeof(struct rte_ipv4_hdr) +
			offsetof(struct rte_tcp_hdr, dst_port),
	},
};

8.4.2. The Main Function

The main() function performs the initialization and calls the execution threads for each lcore.

The first task is to initialize the Environment Abstraction Layer (EAL). The argc and argv arguments are provided to the rte_eal_init() function. The value returned is the number of parsed arguments:

ret = rte_eal_init(argc, argv);
if (ret < 0)
	rte_exit(EXIT_FAILURE, "Error with EAL initialization\n");

It then parses the flow_classify application arguments

ret = parse_args(argc, argv);
if (ret < 0)
	rte_exit(EXIT_FAILURE, "Invalid flow_classify parameters\n");

The main() function also allocates a mempool to hold the mbufs (Message Buffers) used by the application:

mbuf_pool = rte_pktmbuf_pool_create("MBUF_POOL", NUM_MBUFS * nb_ports,
	MBUF_CACHE_SIZE, 0, RTE_MBUF_DEFAULT_BUF_SIZE, rte_socket_id());

mbufs are the packet buffer structure used by DPDK. They are explained in detail in the “Mbuf Library” section of the DPDK Programmer’s Guide.

The main() function also initializes all the ports using the user defined port_init() function which is explained in the next section:

RTE_ETH_FOREACH_DEV(portid)
	if (port_init(portid, mbuf_pool) != 0)
		rte_exit(EXIT_FAILURE, "Cannot init port %"PRIu8 "\n",
				portid);

The main() function creates the flow classifier object and adds an ACL table to the flow classifier.

struct flow_classifier {
	struct rte_flow_classifier *cls;
};

struct flow_classifier_acl {
	struct flow_classifier cls;
} __rte_cache_aligned;
size = RTE_CACHE_LINE_ROUNDUP(sizeof(struct flow_classifier_acl));
cls_app = rte_zmalloc(NULL, size, RTE_CACHE_LINE_SIZE);
if (cls_app == NULL)
	rte_exit(EXIT_FAILURE, "Cannot allocate classifier memory\n");

cls_params.name = "flow_classifier";
cls_params.socket_id = socket_id;

cls_app->cls = rte_flow_classifier_create(&cls_params);
if (cls_app->cls == NULL) {
	rte_free(cls_app);
	rte_exit(EXIT_FAILURE, "Cannot create classifier\n");
}

/* initialise ACL table params */
table_acl_params.name = "table_acl_ipv4_5tuple";
table_acl_params.n_rules = FLOW_CLASSIFY_MAX_RULE_NUM;
table_acl_params.n_rule_fields = RTE_DIM(ipv4_defs);
memcpy(table_acl_params.field_format, ipv4_defs, sizeof(ipv4_defs));

/* initialise table create params */
cls_table_params.ops = &rte_table_acl_ops;
cls_table_params.arg_create = &table_acl_params;
cls_table_params.type = RTE_FLOW_CLASSIFY_TABLE_ACL_IP4_5TUPLE;

ret = rte_flow_classify_table_create(cls_app->cls, &cls_table_params);
if (ret) {
	rte_flow_classifier_free(cls_app->cls);
	rte_free(cls_app);
	rte_exit(EXIT_FAILURE, "Failed to create classifier table\n");
}

It then reads the ipv4_rules_file.txt file and initialises the parameters for the rte_flow_classify_table_entry_add API. This API adds a rule to the ACL table.

if (add_rules(parm_config.rule_ipv4_name, cls_app)) {
	rte_flow_classifier_free(cls_app->cls);
	rte_free(cls_app);
	rte_exit(EXIT_FAILURE, "Failed to add rules\n");
}

Once the initialization is complete, the application is ready to launch a function on an lcore. In this example lcore_main() is called on a single lcore.

lcore_main(cls_app);

The lcore_main() function is explained below.

8.4.3. The Port Initialization Function

The main functional part of the port initialization used in the Basic Forwarding application is shown below:

static inline int
port_init(uint8_t port, struct rte_mempool *mbuf_pool)
{
	struct rte_eth_conf port_conf = port_conf_default;
	struct rte_ether_addr addr;
	const uint16_t rx_rings = 1, tx_rings = 1;
	int retval;
	uint16_t q;
	struct rte_eth_dev_info dev_info;
	struct rte_eth_txconf txconf;

	if (!rte_eth_dev_is_valid_port(port))
		return -1;

	retval = rte_eth_dev_info_get(port, &dev_info);
	if (retval != 0) {
		printf("Error during getting device (port %u) info: %s\n",
				port, strerror(-retval));
		return retval;
	}

	if (dev_info.tx_offload_capa & DEV_TX_OFFLOAD_MBUF_FAST_FREE)
		port_conf.txmode.offloads |=
			DEV_TX_OFFLOAD_MBUF_FAST_FREE;

	/* Configure the Ethernet device. */
	retval = rte_eth_dev_configure(port, rx_rings, tx_rings, &port_conf);
	if (retval != 0)
		return retval;

	/* Allocate and set up 1 RX queue per Ethernet port. */
	for (q = 0; q < rx_rings; q++) {
		retval = rte_eth_rx_queue_setup(port, q, RX_RING_SIZE,
				rte_eth_dev_socket_id(port), NULL, mbuf_pool);
		if (retval < 0)
			return retval;
	}

	txconf = dev_info.default_txconf;
	txconf.offloads = port_conf.txmode.offloads;
	/* Allocate and set up 1 TX queue per Ethernet port. */
	for (q = 0; q < tx_rings; q++) {
		retval = rte_eth_tx_queue_setup(port, q, TX_RING_SIZE,
				rte_eth_dev_socket_id(port), &txconf);
		if (retval < 0)
			return retval;
	}

	/* Start the Ethernet port. 8< */
	retval = rte_eth_dev_start(port);
	/* >8 End of starting the Ethernet port. */
	if (retval < 0)
		return retval;

	/* Display the port MAC address. */
	retval = rte_eth_macaddr_get(port, &addr);
	if (retval != 0)
		return retval;

	printf("Port %u MAC: %02" PRIx8 " %02" PRIx8 " %02" PRIx8
			   " %02" PRIx8 " %02" PRIx8 " %02" PRIx8 "\n",
			port, RTE_ETHER_ADDR_BYTES(&addr));

	/* Enable RX in promiscuous mode for the Ethernet device. */
	retval = rte_eth_promiscuous_enable(port);
	if (retval != 0)
		return retval;

	return 0;
}

The Ethernet ports are configured with default settings using the rte_eth_dev_configure() function and the port_conf_default struct.

static const struct rte_eth_conf port_conf_default = {
	.rxmode = {
		.max_rx_pkt_len = RTE_ETHER_MAX_LEN,
	},
};

For this example the ports are set up with 1 RX and 1 TX queue using the rte_eth_rx_queue_setup() and rte_eth_tx_queue_setup() functions.

The Ethernet port is then started:

retval = rte_eth_dev_start(port);

Finally the RX port is set in promiscuous mode:

retval = rte_eth_promiscuous_enable(port);

8.4.4. The Add Rules function

The add_rules function reads the ipv4_rules_file.txt file and calls the add_classify_rule function which calls the rte_flow_classify_table_entry_add API.

static int
add_rules(const char *rule_path, struct flow_classifier *cls_app)
{
	FILE *fh;
	char buff[LINE_MAX];
	unsigned int i = 0;
	unsigned int total_num = 0;
	struct rte_eth_ntuple_filter ntuple_filter;
	int ret;

	fh = fopen(rule_path, "rb");
	if (fh == NULL)
		rte_exit(EXIT_FAILURE, "%s: fopen %s failed\n", __func__,
			rule_path);

	ret = fseek(fh, 0, SEEK_SET);
	if (ret)
		rte_exit(EXIT_FAILURE, "%s: fseek %d failed\n", __func__,
			ret);

	i = 0;
	while (fgets(buff, LINE_MAX, fh) != NULL) {
		i++;

		if (is_bypass_line(buff))
			continue;

		if (total_num >= FLOW_CLASSIFY_MAX_RULE_NUM - 1) {
			printf("\nINFO: classify rule capacity %d reached\n",
				total_num);
			break;
		}

		if (parse_ipv4_5tuple_rule(buff, &ntuple_filter) != 0)
			rte_exit(EXIT_FAILURE,
				"%s Line %u: parse rules error\n",
				rule_path, i);

		if (add_classify_rule(&ntuple_filter, cls_app) != 0)
			rte_exit(EXIT_FAILURE, "add rule error\n");

		total_num++;
	}

	fclose(fh);
	return 0;
}

8.4.5. The Lcore Main function

As we saw above the main() function calls an application function on the available lcores. The lcore_main function calls the rte_flow_classifier_query API. For the Basic Forwarding application the lcore_main function looks like the following:

static int num_classify_rules;
static struct rte_flow_classify_rule *rules[MAX_NUM_CLASSIFY];
static struct rte_flow_classify_ipv4_5tuple_stats ntuple_stats;
static struct rte_flow_classify_stats classify_stats = {
		.stats = (void **)&ntuple_stats
};
static __rte_noreturn void
lcore_main(struct flow_classifier *cls_app)
{
	uint16_t port;
	int ret;
	int i = 0;

	ret = rte_flow_classify_table_entry_delete(cls_app->cls,
			rules[7]);
	if (ret)
		printf("table_entry_delete failed [7] %d\n\n", ret);
	else
		printf("table_entry_delete succeeded [7]\n\n");

	/*
	 * Check that the port is on the same NUMA node as the polling thread
	 * for best performance.
	 */
	RTE_ETH_FOREACH_DEV(port)
		if (rte_eth_dev_socket_id(port) >= 0 &&
			rte_eth_dev_socket_id(port) != (int)rte_socket_id()) {
			printf("\n\n");
			printf("WARNING: port %u is on remote NUMA node\n",
			       port);
			printf("to polling thread.\n");
			printf("Performance will not be optimal.\n");
		}
	printf("\nCore %u forwarding packets. ", rte_lcore_id());
	printf("[Ctrl+C to quit]\n");

	/* Run until the application is quit or killed. 8< */
	for (;;) {
		/*
		 * Receive packets on a port, classify them and forward them
		 * on the paired port.
		 * The mapping is 0 -> 1, 1 -> 0, 2 -> 3, 3 -> 2, etc.
		 */
		RTE_ETH_FOREACH_DEV(port) {
			/* Get burst of RX packets, from first port of pair. */
			struct rte_mbuf *bufs[BURST_SIZE];
			const uint16_t nb_rx = rte_eth_rx_burst(port, 0,
					bufs, BURST_SIZE);

			if (unlikely(nb_rx == 0))
				continue;

			for (i = 0; i < MAX_NUM_CLASSIFY; i++) {
				if (rules[i]) {
					ret = rte_flow_classifier_query(
						cls_app->cls,
						bufs, nb_rx, rules[i],
						&classify_stats);
					if (ret)
						printf(
							"rule [%d] query failed ret [%d]\n\n",
							i, ret);
					else {
						printf(
						"rule[%d] count=%"PRIu64"\n",
						i, ntuple_stats.counter1);

						printf("proto = %d\n",
						ntuple_stats.ipv4_5tuple.proto);
					}
				}
			}

			/* Send burst of TX packets, to second port of pair. */
			const uint16_t nb_tx = rte_eth_tx_burst(port ^ 1, 0,
					bufs, nb_rx);

			/* Free any unsent packets. */
			if (unlikely(nb_tx < nb_rx)) {
				uint16_t buf;

				for (buf = nb_tx; buf < nb_rx; buf++)
					rte_pktmbuf_free(bufs[buf]);
			}
		}
	}
	/* >8 End of main loop. */
}

The main work of the application is done within the loop:

for (;;) {
	/*
	 * Receive packets on a port, classify them and forward them
	 * on the paired port.
	 * The mapping is 0 -> 1, 1 -> 0, 2 -> 3, 3 -> 2, etc.
	 */
	RTE_ETH_FOREACH_DEV(port) {
		/* Get burst of RX packets, from first port of pair. */
		struct rte_mbuf *bufs[BURST_SIZE];
		const uint16_t nb_rx = rte_eth_rx_burst(port, 0,
				bufs, BURST_SIZE);

		if (unlikely(nb_rx == 0))
			continue;

		for (i = 0; i < MAX_NUM_CLASSIFY; i++) {
			if (rules[i]) {
				ret = rte_flow_classifier_query(
					cls_app->cls,
					bufs, nb_rx, rules[i],
					&classify_stats);
				if (ret)
					printf(
						"rule [%d] query failed ret [%d]\n\n",
						i, ret);
				else {
					printf(
					"rule[%d] count=%"PRIu64"\n",
					i, ntuple_stats.counter1);

					printf("proto = %d\n",
					ntuple_stats.ipv4_5tuple.proto);
				}
			}
		}

		/* Send burst of TX packets, to second port of pair. */
		const uint16_t nb_tx = rte_eth_tx_burst(port ^ 1, 0,
				bufs, nb_rx);

		/* Free any unsent packets. */
		if (unlikely(nb_tx < nb_rx)) {
			uint16_t buf;

			for (buf = nb_tx; buf < nb_rx; buf++)
				rte_pktmbuf_free(bufs[buf]);
		}
	}
}

Packets are received in bursts on the RX ports and transmitted in bursts on the TX ports. The ports are grouped in pairs with a simple mapping scheme using the an XOR on the port number:

0 -> 1
1 -> 0

2 -> 3
3 -> 2

etc.

The rte_eth_tx_burst() function frees the memory buffers of packets that are transmitted. If packets fail to transmit, (nb_tx < nb_rx), then they must be freed explicitly using rte_pktmbuf_free().

The forwarding loop can be interrupted and the application closed using Ctrl-C.