DPDK patches and discussions
 help / color / mirror / Atom feed
From: Xueming Li <xuemingl@mellanox.com>
To: Wu Jingjing <jingjing.wu@intel.com>,
	Harry van Haaren <harry.van.haaren@intel.com>
Cc: Xueming Li <xuemingl@mellanox.com>, dev@dpdk.org
Subject: [dpdk-dev] [RFC v1 6/9] app/testpmd: add pktgen forwarding engine
Date: Fri,  8 Dec 2017 16:22:22 +0800	[thread overview]
Message-ID: <20171208082225.44913-7-xuemingl@mellanox.com> (raw)
In-Reply-To: <20171208082225.44913-1-xuemingl@mellanox.com>
In-Reply-To: <20171019140649.26668-2-xuemingl@mellanox.com>

Add new "pktgen" forwarding engine that send and receive packets
acoording to pktgen_task instruction.

Signed-off-by: Xueming Li <xuemingl@mellanox.com>
---
 app/test-pmd/Makefile  |   1 +
 app/test-pmd/cmdline.c |   3 +
 app/test-pmd/pktgen.c  | 605 +++++++++++++++++++++++++++++++++++++++++++++++++
 app/test-pmd/testpmd.c |   1 +
 app/test-pmd/testpmd.h |   1 +
 5 files changed, 611 insertions(+)
 create mode 100644 app/test-pmd/pktgen.c

diff --git a/app/test-pmd/Makefile b/app/test-pmd/Makefile
index d21308fcd..2610a8d49 100644
--- a/app/test-pmd/Makefile
+++ b/app/test-pmd/Makefile
@@ -55,6 +55,7 @@ SRCS-y += iofwd.c
 SRCS-y += macfwd.c
 SRCS-y += macswap.c
 SRCS-y += flowgen.c
+SRCS-y += pktgen.c
 SRCS-y += rxonly.c
 SRCS-y += txonly.c
 SRCS-y += csumonly.c
diff --git a/app/test-pmd/cmdline.c b/app/test-pmd/cmdline.c
index 6654a62cb..d0eb00c1f 100644
--- a/app/test-pmd/cmdline.c
+++ b/app/test-pmd/cmdline.c
@@ -15594,6 +15594,8 @@ cmdline_parse_inst_t cmd_py_run = {
 
 /* ******************************************************************************** */
 
+extern cmdline_parse_inst_t cmd_pktgen_cmd;
+
 /* list of instructions */
 cmdline_parse_ctx_t main_ctx[] = {
 	(cmdline_parse_inst_t *)&cmd_help_brief,
@@ -15603,6 +15605,7 @@ cmdline_parse_ctx_t main_ctx[] = {
 #ifdef RTE_LIBRTE_PYTHON
 	(cmdline_parse_inst_t *)&cmd_py_run,
 #endif
+	(cmdline_parse_inst_t *)&cmd_pktgen_cmd,
 	(cmdline_parse_inst_t *)&cmd_showport,
 	(cmdline_parse_inst_t *)&cmd_showqueue,
 	(cmdline_parse_inst_t *)&cmd_showportall,
diff --git a/app/test-pmd/pktgen.c b/app/test-pmd/pktgen.c
new file mode 100644
index 000000000..f4a1b00a9
--- /dev/null
+++ b/app/test-pmd/pktgen.c
@@ -0,0 +1,605 @@
+/*-
+ *   BSD LICENSE
+ *
+ *   Copyright(c) 2010-2014 Intel Corporation. All rights reserved.
+ *   All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions
+ *   are met:
+ *
+ *     * Redistributions of source code must retain the above copyright
+ *       notice, this list of conditions and the following disclaimer.
+ *     * Redistributions in binary form must reproduce the above copyright
+ *       notice, this list of conditions and the following disclaimer in
+ *       the documentation and/or other materials provided with the
+ *       distribution.
+ *     * Neither the name of Intel Corporation nor the names of its
+ *       contributors may be used to endorse or promote products derived
+ *       from this software without specific prior written permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <stdio.h>
+#include <string.h>
+#include <errno.h>
+#include <stdint.h>
+#include <unistd.h>
+#include <inttypes.h>
+#include <fcntl.h>
+
+#include <sys/queue.h>
+#include <sys/stat.h>
+
+#include <rte_common.h>
+#include <rte_byteorder.h>
+#include <rte_log.h>
+#include <rte_debug.h>
+#include <rte_malloc.h>
+#include <rte_cycles.h>
+#include <rte_memory.h>
+#include <rte_memzone.h>
+#include <rte_launch.h>
+#include <rte_eal.h>
+#include <rte_per_lcore.h>
+#include <rte_lcore.h>
+#include <rte_atomic.h>
+#include <rte_branch_prediction.h>
+#include <rte_memory.h>
+#include <rte_memcpy.h>
+#include <rte_mempool.h>
+#include <rte_mbuf.h>
+#include <rte_interrupts.h>
+#include <rte_pci.h>
+#include <rte_ether.h>
+#include <rte_ethdev.h>
+#include <rte_flow.h>
+#include <rte_string_fns.h>
+#include <cmdline.h>
+#include <cmdline_rdline.h>
+#include <cmdline_parse.h>
+#include <cmdline_parse_num.h>
+#include <cmdline_parse_string.h>
+#include <cmdline_parse_ipaddr.h>
+#include <cmdline_parse_etheraddr.h>
+#include <cmdline_socket.h>
+#ifdef RTE_LIBRTE_PYTHON
+#include <rte_python.h>
+#endif
+
+#include "testpmd.h"
+
+#define PKTGEN_ROUND_END 0x1
+#define PKTGEN_TASK_END 0x0
+#define PKTGEN_TASK_START 0x3
+
+struct pktgen_task_stats{
+	uint8_t active;
+	uint16_t round;		/* number txrx */
+	uint64_t count;
+	uint64_t start; 	/* start tsc */
+	uint64_t end;		/* end tsc */
+};
+
+struct pktgen_task {
+	void *data;		/* rx: tx; tx: mbuf[] */
+	uint64_t count;
+	char *field;		/* field to match */
+	uint64_t val;		/* field value */
+	uint16_t round;		/* number txrx */
+	uint16_t cnt_mbufs;	/* number of templates */
+	uint8_t verbose:7;
+	uint8_t txrx:1; 	/* txrx type task */
+	struct pktgen_task_stats stats;
+} __rte_cache_aligned;
+
+static void *pktgen_tx_tasks;
+static void *pktgen_rx_tasks;
+static int pktgen_idle_mode = 2; /* 0-drop 1-loopback 2-forward 3-switch */
+static int pktgen_busy = 0;
+
+static inline struct rte_mbuf *
+pg_task_template_get(struct pktgen_task* task, uint32_t idx)
+{
+	struct rte_mbuf **mbufs = task->data;
+
+	return mbufs ? mbufs[idx % task->cnt_mbufs] : NULL;
+}
+
+static inline struct pktgen_task *
+task_tx(portid_t port, queueid_t queue)
+{
+	RTE_ASSERT(pktgen_tx_tasks);
+	struct pktgen_task (*tasks)[nb_txq] = pktgen_tx_tasks;
+	return &tasks[port][queue];
+}
+
+static inline struct pktgen_task *
+task_rx(portid_t port, queueid_t queue)
+{
+	RTE_ASSERT(pktgen_rx_tasks);
+	struct pktgen_task (*tasks)[nb_rxq] = pktgen_rx_tasks;
+	return &tasks[port][queue];
+}
+
+/********************************************************/
+/* Forwarding thread functions                          */
+/********************************************************/
+
+static inline void
+pg_dump_mbuf_header(struct rte_mbuf *mbuf, struct fwd_stream* fs, int is_rx) {
+	char buf[256];
+
+	printf("%s P:%hu Q:%hu len:%hu ptype:0x%x ol_flags:0x%lx rss:0x%08x fdir:0x%x\n",
+			is_rx ? "RX" : "TX",
+			is_rx ? fs->rx_port : fs->tx_port,
+			is_rx ? fs->rx_queue : fs->tx_queue,
+			mbuf->data_len, mbuf->packet_type,
+			mbuf->ol_flags,
+			mbuf->hash.rss, mbuf->hash.fdir.hi
+			);
+	if (mbuf->packet_type) {
+		rte_get_ptype_name(mbuf->packet_type, buf, sizeof(buf));
+		printf("  ptype: %s\n", buf);
+	}
+	if (mbuf->tx_offload)
+		printf("  header len:%d/%d/%d/%d/%d tso len:%hu\n",
+				mbuf->outer_l2_len, mbuf->outer_l3_len,
+				mbuf->l2_len, mbuf->l3_len, mbuf->l4_len,
+				mbuf->tso_segsz);
+	else if (mbuf->ol_flags){
+		rte_get_rx_ol_flag_list(mbuf->ol_flags, buf, sizeof(buf));
+		printf("  ol_flags: %s\n", buf);
+	}
+}
+
+static inline void
+pg_debug(struct rte_mbuf *mbuf, uint8_t level, struct fwd_stream* fs, int is_rx)
+{
+	/* xxxx xxxx
+	 *   ||	  L- 1: summary 2: repr 3:show
+	 *   |L----- mbuf header
+	 *   L------ hex dump
+	 */
+	if (level & 0x10)
+		pg_dump_mbuf_header(mbuf, fs, is_rx);
+#ifdef RTE_LIBRTE_PYTHON
+	if (level)
+		rte_python_scapy_dump(mbuf, level);
+#else
+	(void) mbuf;
+#endif
+}
+
+static int
+pg_mbuf_field_expect(struct fwd_stream* fs, struct pktgen_task *task,
+		struct rte_mbuf *mbuf)
+{
+	#define OFF(field) offsetof(struct rte_mbuf, field)
+	unsigned int i;
+	uint64_t val;
+	static struct {
+		const char *name;
+		uint8_t offset;
+		uint8_t shift;
+		uint64_t mask;
+	} fields [] = {
+		{"port", 	OFF(port), 		0, UINT16_MAX},
+		{"ptype",	OFF(packet_type), 	0, UINT32_MAX},
+		{"rss", 	OFF(hash.rss),		0, UINT32_MAX},
+		{"fdir", 	OFF(hash.fdir.hi),	0, UINT32_MAX},
+		/* ignore rss bit */
+		{"ol_flags",	OFF(ol_flags), 		0, UINT64_MAX &
+							   ~PKT_RX_RSS_HASH},
+	};
+
+	RTE_ASSERT(task && mbuf);
+	if (!task->field || strlen(task->field) == 0)
+		return 0;
+	if (!strcmp(task->field, "queue")) {
+		if (fs->rx_queue != task->val) {
+			printf("Failed: queue expect: 0x%lu received:0x%hu\n",
+					task->val, fs->rx_queue);
+			return 1;
+		} else
+			return 0;
+	}
+	for (i = 0; i < RTE_DIM(fields); i++) {
+		if (strcmp(task->field, fields[i].name))
+			continue;
+		val = *((uint64_t *)(void *)((char *)(void *)mbuf + fields[i].offset));
+		val &= fields[i].mask;
+		if ((val != (task->val & fields[i].mask))) {
+			printf("Failed: %s mask: 0x%lx expect: 0x%lx received: 0x%lx\n",
+				fields[i].name, fields[i].mask,
+				task->val & fields[i].mask, val);
+			return 1;
+		} else
+			return 0;
+	}
+	printf("Failed: unknown field '%s', valid names: queue,", task->field);
+	for (i = 0; i < RTE_DIM(fields); i++)
+		printf("%s%s", fields[i].name,
+		       i == RTE_DIM(fields) - 1 ? ",non(0)\n" : ",");
+	return 0;
+}
+
+static int
+pg_mbuf_expect(struct fwd_stream* fs, struct pktgen_task *task,
+		struct rte_mbuf *mbuf, uint32_t idx)
+{
+	int r = 0;
+	struct rte_mbuf *exp;
+
+	RTE_ASSERT(task && task->data && mbuf);
+	exp = pg_task_template_get(task->data, idx); /* tx->mbuf */
+	if (!exp) {
+		RTE_LOG(ERR, USER1, "packet tempalte not found, timeout?\n");
+		return -1;
+	}
+	r |= pg_mbuf_field_expect(fs, task, mbuf);
+	if (exp->data_len != mbuf->data_len) {
+		printf("Failed: packet length not same: %hu/%hu",
+				 mbuf->data_len, exp->data_len);
+		r |= 2;
+	} else if (memcmp(
+			rte_pktmbuf_mtod(mbuf, void *),
+			rte_pktmbuf_mtod(exp, void *),
+			mbuf->data_len)) {
+		printf("Failed: packet not same:\n");
+		r |= 4;
+#ifdef RTE_LIBRTE_PYTHON
+		rte_python_scapy_hexdiff(
+			rte_pktmbuf_mtod(exp, void *), exp->data_len,
+			rte_pktmbuf_mtod(mbuf, void *), mbuf->data_len);
+#endif
+	}
+	return r;
+}
+
+static inline void
+pg_mbuf_switch(struct rte_mbuf **pkts, uint16_t nb_to_tx)
+{
+	uint32_t i;
+	struct ether_hdr *eth;
+	struct ether_addr addr;
+
+	for (i = 0; i < nb_to_tx; i++) {
+		eth = rte_pktmbuf_mtod(pkts[i], struct ether_hdr *);
+		ether_addr_copy(&eth->d_addr, &addr);
+		ether_addr_copy(&eth->s_addr, &eth->d_addr);
+		ether_addr_copy(&addr, &eth->s_addr);
+	}
+}
+
+static inline uint16_t
+pg_tx_burst(struct fwd_stream* fs, int mode,
+		struct rte_mbuf **pkts_burst, uint16_t nb_to_tx)
+{
+	uint32_t retry;
+	uint16_t nb_tx, i;
+	portid_t port = mode == 1 ? fs->rx_port : fs->tx_port;
+	queueid_t queue = mode == 1 ? fs->rx_queue : fs->tx_queue;
+
+	if (unlikely(mode == 3))
+		pg_mbuf_switch(pkts_burst, nb_to_tx);
+	nb_tx = rte_eth_tx_burst(port, queue, pkts_burst,
+			nb_to_tx);
+	/* Retry if necessary */
+	if (unlikely(nb_tx < nb_to_tx) && fs->retry_enabled)
+	{
+		retry = 0;
+		while (nb_tx < nb_to_tx && retry++ < burst_tx_retry_num)
+		{
+			rte_delay_us(burst_tx_delay_time);
+			nb_tx += rte_eth_tx_burst(port, queue,
+					&pkts_burst[nb_tx], nb_to_tx - nb_tx);
+		}
+	}
+	/* Drop packets failed to send */
+	if (unlikely(nb_tx < nb_to_tx))
+	{
+		fs->fwd_dropped += (nb_to_tx - nb_tx);
+		i = nb_tx;
+		do {
+			rte_pktmbuf_free(pkts_burst[i]);
+		} while (++i < nb_to_tx);
+	}
+	return nb_tx;
+}
+
+static inline int
+pg_tx_fill(struct rte_mbuf **pkts_burst, uint64_t nb_to_tx,
+		struct pktgen_task *task, struct fwd_stream* fs)
+{
+	uint32_t i;
+	struct rte_mbuf *exp;
+
+	exp = pg_task_template_get(task, task->stats.count);
+	RTE_ASSERT(exp && exp->pool);
+	RTE_ASSERT(task->cnt_mbufs > 0);
+	if (rte_pktmbuf_alloc_bulk(exp->pool,
+			pkts_burst, nb_to_tx))
+		return -1;
+	for (i = 0; i < nb_to_tx; i++) {
+		exp = pg_task_template_get(task, task->stats.count + i);
+		rte_memcpy(rte_pktmbuf_mtod(pkts_burst[i], void *),
+				rte_pktmbuf_mtod(exp, void *),
+				exp->data_len);
+		pkts_burst[i]->pkt_len = exp->pkt_len;
+		pkts_burst[i]->data_len = exp->data_len;
+		pkts_burst[i]->ol_flags = exp->ol_flags;
+		pg_debug(pkts_burst[i], task->verbose, fs, 0);
+	}
+	return 0;
+}
+
+static inline int
+pg_start(struct pktgen_task *task, uint64_t start_tsc)
+{
+	/* even round end, has to check tx stats */
+	if (unlikely(task->stats.active == PKTGEN_TASK_END))
+		return -1;
+	if (!task->stats.start)
+		task->stats.start = start_tsc;
+	return 0;
+}
+
+static inline int
+pg_round_end(struct pktgen_task *task)
+{
+	/* if not txrx task, keep busy */
+	if (unlikely(task->txrx))
+		task->stats.active = PKTGEN_ROUND_END;
+	task->stats.round += 1;
+	return task->round == task->stats.round;
+}
+
+static inline void
+pg_end(struct pktgen_task *task)
+{
+	task->stats.active = PKTGEN_TASK_END;
+	task->stats.end = rte_rdtsc();
+}
+
+/* return -1 if nothing to do */
+static inline int
+pg_tx(struct fwd_stream* fs, struct pktgen_task *task,
+		uint64_t start_tsc)
+{
+	struct rte_mbuf *pkts_burst[nb_pkt_per_burst];
+	uint64_t nb_to_tx = 0;
+	uint64_t nb_tx;
+
+	if (pg_start(task, start_tsc))
+		return -1;
+	if (unlikely(task->stats.active != PKTGEN_TASK_START))
+		return -1;
+	if (task->count) {
+		nb_to_tx = task->count - task->stats.count;
+		if (nb_to_tx > nb_pkt_per_burst)
+			nb_to_tx = nb_pkt_per_burst;
+	} else
+			nb_to_tx = nb_pkt_per_burst;
+	if (likely((nb_to_tx && task->data))) {
+		if (unlikely((pg_tx_fill(pkts_burst, nb_to_tx, task, fs))))
+			return -1;
+		nb_tx = pg_tx_burst(fs, 2, pkts_burst, nb_to_tx);
+		fs->tx_packets += nb_tx;
+		task->stats.count += nb_tx;
+	}
+	if (task->stats.count == task->count) {
+		if (pg_round_end(task)) /* end of taks? */
+			pg_end(task);
+	}
+	return 0;
+}
+
+/* return -1 if nothing to do */
+static inline int
+pg_rx(struct fwd_stream* fs, struct pktgen_task *task, uint64_t start_tsc)
+{
+	struct rte_mbuf *pkts_burst[nb_pkt_per_burst];
+	uint16_t nb_to_rx;
+	uint16_t nb_rx;
+	uint32_t i;
+	uint8_t verbose;
+	struct pktgen_task *tx_task;
+	int r;
+
+	if (pg_start(task, start_tsc))
+		return -1;
+	if (task->count) {
+		nb_to_rx = task->count - task->stats.count;
+		if (nb_to_rx > nb_pkt_per_burst)
+			nb_to_rx = nb_pkt_per_burst;
+	} else /* endless rx */
+		nb_to_rx = nb_pkt_per_burst;
+	if (nb_to_rx && task->stats.active == PKTGEN_TASK_START) {
+		nb_rx = rte_eth_rx_burst(fs->rx_port, fs->rx_queue,
+				pkts_burst, nb_pkt_per_burst);
+#ifdef RTE_TEST_PMD_RECORD_BURST_STATS
+		fs->rx_burst_stats.pkt_burst_spread[nb_rx]++;
+#endif
+		if (unlikely(nb_rx == 0))
+			return 0;
+		for (i = 0; i < nb_rx; i++) {
+			verbose = task->verbose;
+			if (task->data) {
+				r = pg_mbuf_expect(fs, task, pkts_burst[i], task->stats.count + i);
+				if (r < 0) /* task timeout */
+					return r;
+				else if (r) /* compare failed, simple dump */
+					verbose |= 0x10;
+			}
+			pg_debug(pkts_burst[i], verbose, fs, 1);
+			rte_pktmbuf_free(pkts_burst[i]);
+		}
+		fs->rx_packets += nb_rx;
+		task->stats.count += nb_rx;
+	}
+	if (task->count && task->stats.count >= task->count) {
+		tx_task = task->data;
+		if (task->stats.active == PKTGEN_TASK_START && pg_round_end(task))
+			pg_end(task);
+		else if (tx_task && tx_task->stats.active==PKTGEN_ROUND_END) {
+			/* has tx task, next round */
+			tx_task->stats.active =	PKTGEN_TASK_START;
+			tx_task->stats.count = 0;
+			task->stats.active = PKTGEN_TASK_START;
+			task->stats.count = 0;
+		}
+	}
+	return 0;
+}
+
+static void
+pg_idle_set(int mode)
+{
+	const char *names[] = {
+			"drop(0)",
+			"loopback(1)",
+			"io_forward(2)",
+			"mac_switch(3)"};
+	if (pktgen_idle_mode != mode)
+		printf("PktGen idle mode changed from %s to %s\n",
+		       names[pktgen_idle_mode], names[mode]);
+	pktgen_idle_mode = mode;
+}
+
+static inline int
+pg_rx_idle(struct fwd_stream* fs)
+{
+	struct rte_mbuf *pkts_burst[nb_pkt_per_burst];
+	uint16_t nb_rx;
+	uint32_t i;
+
+	nb_rx = rte_eth_rx_burst(fs->rx_port, fs->rx_queue,
+				pkts_burst, nb_pkt_per_burst);
+#ifdef RTE_TEST_PMD_RECORD_BURST_STATS
+	fs->rx_burst_stats.pkt_burst_spread[nb_rx]++;
+#endif
+	if (unlikely(nb_rx == 0))
+		return 0;
+	fs->rx_packets += nb_rx;
+	if (verbose_level & 0xff)
+		for (i = 0; i < nb_rx; i++)
+			pg_debug(pkts_burst[i], verbose_level & 0xff, fs, 1);
+	if (pktgen_idle_mode) /* no drop */
+		pg_tx_burst(fs, pktgen_idle_mode, pkts_burst, nb_rx);
+	else /* drop */
+		for (i = 0; i < nb_rx; i++)
+			rte_pktmbuf_free(pkts_burst[i]);
+	return 0;
+}
+
+/*
+ * TX and RX pacets according to traffic generator command.
+ */
+static void
+pg_fwd(struct fwd_stream *fs)
+{
+	uint64_t start;
+#ifdef RTE_TEST_PMD_RECORD_CORE_CYCLES
+	uint64_t end_tsc;
+	uint64_t core_cycles;
+#endif
+
+	start = rte_rdtsc();
+	if (likely(pktgen_busy)) {
+		pg_tx(fs, task_tx(fs->tx_port, fs->tx_queue), start);
+		pg_rx(fs, task_rx(fs->rx_port, fs->rx_queue), start);
+	} else
+		pg_rx_idle(fs);
+#ifdef RTE_TEST_PMD_RECORD_CORE_CYCLES
+	end_tsc = rte_rdtsc();
+	core_cycles = (end_tsc - start_tsc);
+	fs->core_cycles += fs->core_cycles + core_cycles;
+#endif
+}
+
+static void
+pktgen_begin(portid_t pi __rte_unused)
+{
+	if (!pktgen_tx_tasks)
+		pktgen_tx_tasks = rte_malloc(NULL,
+			sizeof(struct pktgen_task) * nb_ports * nb_txq, 0);
+	if (!pktgen_tx_tasks)
+		RTE_LOG(ERR, USER1, "out of memory?\n");
+	if (!pktgen_rx_tasks)
+		pktgen_rx_tasks = rte_malloc(NULL,
+			sizeof(struct pktgen_task) * nb_ports * nb_rxq, 0);
+	if (!pktgen_rx_tasks)
+		RTE_LOG(ERR, USER1, "out of memory?\n");
+}
+
+static void
+pktgen_end(portid_t pi __rte_unused)
+{
+	if (pktgen_tx_tasks)
+		rte_free(pktgen_tx_tasks);
+	pktgen_tx_tasks = NULL;
+	if (pktgen_rx_tasks)
+		rte_free(pktgen_rx_tasks);
+	pktgen_rx_tasks = NULL;
+}
+
+struct fwd_engine pktgen_engine = {
+	.fwd_mode_name  = "pktgen",
+	.port_fwd_begin = pktgen_begin,
+	.port_fwd_end   = pktgen_end,
+	.packet_fwd     = pg_fwd,
+};
+
+/********************************************************/
+/* Control thread functions                             */
+/********************************************************/
+
+/* "pktgen loopback" command */
+struct cmd_pktgen_cmd_result {
+	cmdline_fixed_string_t pktgen;
+	cmdline_fixed_string_t cmd;
+	uint8_t mode;
+};
+
+cmdline_parse_token_string_t cmd_pktgen_cmd_pktgen =
+	TOKEN_STRING_INITIALIZER(struct cmd_pktgen_cmd_result, pktgen, "pktgen");
+cmdline_parse_token_string_t cmd_pktgen_cmd_cmd =
+	TOKEN_STRING_INITIALIZER(struct cmd_pktgen_cmd_result, cmd, "idle");
+cmdline_parse_token_string_t cmd_pktgen_cmd_mode =
+		TOKEN_NUM_INITIALIZER(struct cmd_pktgen_cmd_result, mode, UINT8);
+
+static void
+cmd_pktgen_cmd_parsed(
+	void *parsed_result,
+	__attribute__((unused)) struct cmdline *cl,
+	__attribute__((unused)) void *data)
+{
+	struct cmd_pktgen_cmd_result *res = parsed_result;
+
+	pktgen_idle_mode = res->mode;
+	printf("PktGen idle mode: %hhu\n", res->mode);
+}
+
+cmdline_parse_inst_t cmd_pktgen_cmd = {
+	.f = cmd_pktgen_cmd_parsed,
+	.data = NULL,
+	.help_str = "pktgen idle <mode>: 0-drop 1-loopback 2-forward 3-switch",
+	.tokens = {
+		(void *)&cmd_pktgen_cmd_pktgen,
+		(void *)&cmd_pktgen_cmd_cmd,
+		(void *)&cmd_pktgen_cmd_mode,
+		NULL,
+	},
+};
diff --git a/app/test-pmd/testpmd.c b/app/test-pmd/testpmd.c
index c3ab44849..2a50e522b 100644
--- a/app/test-pmd/testpmd.c
+++ b/app/test-pmd/testpmd.c
@@ -159,6 +159,7 @@ struct fwd_engine * fwd_engines[] = {
 	&mac_fwd_engine,
 	&mac_swap_engine,
 	&flow_gen_engine,
+	&pktgen_engine,
 	&rx_only_engine,
 	&tx_only_engine,
 	&csum_fwd_engine,
diff --git a/app/test-pmd/testpmd.h b/app/test-pmd/testpmd.h
index 1639d27e7..ec83c212d 100644
--- a/app/test-pmd/testpmd.h
+++ b/app/test-pmd/testpmd.h
@@ -296,6 +296,7 @@ extern struct fwd_engine io_fwd_engine;
 extern struct fwd_engine mac_fwd_engine;
 extern struct fwd_engine mac_swap_engine;
 extern struct fwd_engine flow_gen_engine;
+extern struct fwd_engine pktgen_engine;
 extern struct fwd_engine rx_only_engine;
 extern struct fwd_engine tx_only_engine;
 extern struct fwd_engine csum_fwd_engine;
-- 
2.13.3

  parent reply	other threads:[~2017-12-08  8:23 UTC|newest]

Thread overview: 26+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-10-19 14:06 [dpdk-dev] [RFC PATCH 0/2] using scapy to generate packet templates Xueming Li
2017-10-19 14:06 ` [dpdk-dev] [RFC PATCH 1/2] app/testpmd: add packet template Xueming Li
2017-12-05  4:43   ` [dpdk-dev] [RFC v1 00/11] scapy/python extension Xueming Li
2017-12-05  4:45   ` Xueming Li
2017-12-05  4:48   ` Xueming Li
2017-12-05  4:55   ` Xueming Li
2017-12-05  6:14     ` Xueming(Steven) Li
2017-12-05  5:00   ` Xueming Li
2017-12-05  5:03   ` [dpdk-dev] [RFC v1 00/11] scappy/pythoon extension Xueming Li
2017-12-05  5:04   ` [dpdk-dev] [RFC v1 00/11] scapy/python extension Xueming Li
2017-12-10 23:16     ` Wiles, Keith
2019-01-10 13:06     ` Eelco Chaudron
2019-01-16 13:24       ` Xueming(Steven) Li
2017-12-08  8:22   ` [dpdk-dev] [RFC v1 0/9] " Xueming Li
2017-12-08  8:22   ` [dpdk-dev] [RFC v1 1/9] lib/cmdline: add echo support in batch loading from file Xueming Li
2017-12-08  8:22   ` [dpdk-dev] [RFC v1 2/9] app/testpmd: support command echo in CLI batch loading Xueming Li
2017-12-08  8:22   ` [dpdk-dev] [RFC v1 3/9] test: update batch loading test Xueming Li
2017-12-08  8:22   ` [dpdk-dev] [RFC v1 4/9] lib/python: add embedded python lib Xueming Li
2017-12-08  8:22   ` [dpdk-dev] [RFC v1 5/9] app/testpmd: add python command Xueming Li
2017-12-08  8:22   ` Xueming Li [this message]
2017-12-08  8:22   ` [dpdk-dev] [RFC v1 7/9] app/testpmd: add pktgen engine scapy commands Xueming Li
2017-12-08  8:22   ` [dpdk-dev] [RFC v1 8/9] test/expect: add expect test scripts Xueming Li
2017-12-08  8:22   ` [dpdk-dev] [RFC v1 9/9] doc/scapy: add scapy how-to guide Xueming Li
2017-10-19 14:06 ` [dpdk-dev] [RFC PATCH 2/2] app/testpmd: add scapy command as pkt template Xueming Li
2017-10-19 15:21 ` [dpdk-dev] [RFC PATCH 0/2] using scapy to generate packet templates Van Haaren, Harry
2017-10-21 16:04   ` Xueming(Steven) Li

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20171208082225.44913-7-xuemingl@mellanox.com \
    --to=xuemingl@mellanox.com \
    --cc=dev@dpdk.org \
    --cc=harry.van.haaren@intel.com \
    --cc=jingjing.wu@intel.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).