xdpsock_user.c 17.8 KB
Newer Older
1
// SPDX-License-Identifier: GPL-2.0
B
Björn Töpel 已提交
2
/* Copyright(c) 2017 - 2018 Intel Corporation. */
3

4
#include <asm/barrier.h>
5 6 7 8
#include <errno.h>
#include <getopt.h>
#include <libgen.h>
#include <linux/bpf.h>
9
#include <linux/compiler.h>
10 11 12
#include <linux/if_link.h>
#include <linux/if_xdp.h>
#include <linux/if_ether.h>
13 14
#include <locale.h>
#include <net/ethernet.h>
15
#include <net/if.h>
16 17
#include <poll.h>
#include <pthread.h>
18 19 20 21 22
#include <signal.h>
#include <stdbool.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
23
#include <sys/mman.h>
24 25
#include <sys/resource.h>
#include <sys/socket.h>
26
#include <sys/types.h>
27 28 29
#include <time.h>
#include <unistd.h>

30 31
#include "libbpf.h"
#include "xsk.h"
32
#include <bpf/bpf.h>
33 34 35 36 37 38 39 40 41 42 43 44 45

#ifndef SOL_XDP
#define SOL_XDP 283
#endif

#ifndef AF_XDP
#define AF_XDP 44
#endif

#ifndef PF_XDP
#define PF_XDP AF_XDP
#endif

46 47
#define NUM_FRAMES (4 * 1024)
#define BATCH_SIZE 64
48 49

#define DEBUG_HEXDUMP 0
50
#define MAX_SOCKS 8
51

B
Björn Töpel 已提交
52
typedef __u64 u64;
53 54 55 56 57 58 59 60 61 62 63
typedef __u32 u32;

static unsigned long prev_time;

enum benchmark_type {
	BENCH_RXDROP = 0,
	BENCH_TXONLY = 1,
	BENCH_L2FWD = 2,
};

static enum benchmark_type opt_bench = BENCH_RXDROP;
64
static u32 opt_xdp_flags = XDP_FLAGS_UPDATE_IF_NOEXIST;
65 66 67 68 69
static const char *opt_if = "";
static int opt_ifindex;
static int opt_queue;
static int opt_poll;
static int opt_interval = 1;
70
static u32 opt_xdp_bind_flags = XDP_USE_NEED_WAKEUP;
71 72 73
static u32 opt_umem_flags;
static int opt_unaligned_chunks;
static u32 opt_xdp_bind_flags;
74
static int opt_xsk_frame_size = XSK_UMEM__DEFAULT_FRAME_SIZE;
75 76
static int opt_timeout = 1000;
static bool opt_need_wakeup = true;
77
static __u32 prog_id;
78

79 80 81 82 83
struct xsk_umem_info {
	struct xsk_ring_prod fq;
	struct xsk_ring_cons cq;
	struct xsk_umem *umem;
	void *buffer;
84 85
};

86 87 88 89 90
struct xsk_socket_info {
	struct xsk_ring_cons rx;
	struct xsk_ring_prod tx;
	struct xsk_umem_info *umem;
	struct xsk_socket *xsk;
91 92 93 94
	unsigned long rx_npkts;
	unsigned long tx_npkts;
	unsigned long prev_rx_npkts;
	unsigned long prev_tx_npkts;
95
	u32 outstanding_tx;
96 97 98
};

static int num_socks;
99
struct xsk_socket_info *xsks[MAX_SOCKS];
100 101 102 103 104 105 106 107 108

static unsigned long get_nsecs(void)
{
	struct timespec ts;

	clock_gettime(CLOCK_MONOTONIC, &ts);
	return ts.tv_sec * 1000000000UL + ts.tv_nsec;
}

109
static void print_benchmark(bool running)
110
{
111
	const char *bench_str = "INVALID";
112

113 114 115 116 117 118
	if (opt_bench == BENCH_RXDROP)
		bench_str = "rxdrop";
	else if (opt_bench == BENCH_TXONLY)
		bench_str = "txonly";
	else if (opt_bench == BENCH_L2FWD)
		bench_str = "l2fwd";
119

120 121 122 123 124 125 126
	printf("%s:%d %s ", opt_if, opt_queue, bench_str);
	if (opt_xdp_flags & XDP_FLAGS_SKB_MODE)
		printf("xdp-skb ");
	else if (opt_xdp_flags & XDP_FLAGS_DRV_MODE)
		printf("xdp-drv ");
	else
		printf("	");
127

128 129
	if (opt_poll)
		printf("poll() ");
130

131 132 133
	if (running) {
		printf("running...");
		fflush(stdout);
134 135 136
	}
}

137
static void dump_stats(void)
138
{
139 140 141
	unsigned long now = get_nsecs();
	long dt = now - prev_time;
	int i;
142

143
	prev_time = now;
144

145 146 147
	for (i = 0; i < num_socks && xsks[i]; i++) {
		char *fmt = "%-15s %'-11.0f %'-11lu\n";
		double rx_pps, tx_pps;
148

149 150 151 152
		rx_pps = (xsks[i]->rx_npkts - xsks[i]->prev_rx_npkts) *
			 1000000000. / dt;
		tx_pps = (xsks[i]->tx_npkts - xsks[i]->prev_tx_npkts) *
			 1000000000. / dt;
153

154 155 156
		printf("\n sock%d@", i);
		print_benchmark(false);
		printf("\n");
157

158 159 160 161
		printf("%-15s %-11s %-11s %-11.2f\n", "", "pps", "pkts",
		       dt / 1000000000.);
		printf(fmt, "rx", rx_pps, xsks[i]->rx_npkts);
		printf(fmt, "tx", tx_pps, xsks[i]->tx_npkts);
162

163 164
		xsks[i]->prev_rx_npkts = xsks[i]->rx_npkts;
		xsks[i]->prev_tx_npkts = xsks[i]->tx_npkts;
165 166 167
	}
}

168
static void *poller(void *arg)
169
{
170 171 172 173
	(void)arg;
	for (;;) {
		sleep(opt_interval);
		dump_stats();
174 175
	}

176
	return NULL;
177 178
}

179
static void remove_xdp_program(void)
180
{
181
	__u32 curr_prog_id = 0;
182

183 184 185
	if (bpf_get_link_xdp_id(opt_ifindex, &curr_prog_id, opt_xdp_flags)) {
		printf("bpf_get_link_xdp_id failed\n");
		exit(EXIT_FAILURE);
186
	}
187 188 189 190 191 192
	if (prog_id == curr_prog_id)
		bpf_set_link_xdp_fd(opt_ifindex, -1, opt_xdp_flags);
	else if (!curr_prog_id)
		printf("couldn't find a prog id on a given interface\n");
	else
		printf("program on interface changed, not removing\n");
193 194
}

195
static void int_exit(int sig)
196
{
197
	struct xsk_umem *umem = xsks[0]->umem->umem;
198

199
	(void)sig;
200

201 202 203 204
	dump_stats();
	xsk_socket__delete(xsks[0]->xsk);
	(void)xsk_umem__delete(umem);
	remove_xdp_program();
205

206
	exit(EXIT_SUCCESS);
207 208
}

209 210
static void __exit_with_error(int error, const char *file, const char *func,
			      int line)
211
{
212 213 214 215 216
	fprintf(stderr, "%s:%s:%i: errno: %d/\"%s\"\n", file, func,
		line, error, strerror(error));
	dump_stats();
	remove_xdp_program();
	exit(EXIT_FAILURE);
217 218
}

219 220
#define exit_with_error(error) __exit_with_error(error, __FILE__, __func__, \
						 __LINE__)
221

222 223 224 225 226
static const char pkt_data[] =
	"\x3c\xfd\xfe\x9e\x7f\x71\xec\xb1\xd7\x98\x3a\xc0\x08\x00\x45\x00"
	"\x00\x2e\x00\x00\x00\x00\x40\x11\x88\x97\x05\x08\x07\x08\xc8\x14"
	"\x1e\x04\x10\x92\x10\x92\x00\x1a\x6d\xa3\x34\x33\x1f\x69\x40\x6b"
	"\x54\x59\xb6\x14\x2d\x11\x44\xbf\xaf\xd9\xbe\xaa";
227 228 229 230 231 232 233 234 235 236 237 238 239

static void swap_mac_addresses(void *data)
{
	struct ether_header *eth = (struct ether_header *)data;
	struct ether_addr *src_addr = (struct ether_addr *)&eth->ether_shost;
	struct ether_addr *dst_addr = (struct ether_addr *)&eth->ether_dhost;
	struct ether_addr tmp;

	tmp = *src_addr;
	*src_addr = *dst_addr;
	*dst_addr = tmp;
}

B
Björn Töpel 已提交
240
static void hex_dump(void *pkt, size_t length, u64 addr)
241 242 243 244 245
{
	const unsigned char *address = (unsigned char *)pkt;
	const unsigned char *line = address;
	size_t line_size = 32;
	unsigned char c;
B
Björn Töpel 已提交
246 247
	char buf[32];
	int i = 0;
248

B
Björn Töpel 已提交
249 250 251 252
	if (!DEBUG_HEXDUMP)
		return;

	sprintf(buf, "addr=%llu", addr);
253
	printf("length = %zu\n", length);
B
Björn Töpel 已提交
254
	printf("%s | ", buf);
255 256 257 258 259 260 261 262 263 264 265 266 267 268
	while (length-- > 0) {
		printf("%02X ", *address++);
		if (!(++i % line_size) || (length == 0 && i % line_size)) {
			if (length == 0) {
				while (i++ % line_size)
					printf("__ ");
			}
			printf(" | ");	/* right close */
			while (line < address) {
				c = *line++;
				printf("%c", (c < 33 || c == 255) ? 0x2E : c);
			}
			printf("\n");
			if (length > 0)
B
Björn Töpel 已提交
269
				printf("%s | ", buf);
270 271 272 273 274
		}
	}
	printf("\n");
}

275
static size_t gen_eth_frame(struct xsk_umem_info *umem, u64 addr)
276
{
277 278
	memcpy(xsk_umem__get_data(umem->buffer, addr), pkt_data,
	       sizeof(pkt_data) - 1);
279 280 281
	return sizeof(pkt_data) - 1;
}

282
static struct xsk_umem_info *xsk_configure_umem(void *buffer, u64 size)
283
{
284
	struct xsk_umem_info *umem;
285 286 287 288 289
	struct xsk_umem_config cfg = {
		.fill_size = XSK_RING_PROD__DEFAULT_NUM_DESCS,
		.comp_size = XSK_RING_CONS__DEFAULT_NUM_DESCS,
		.frame_size = opt_xsk_frame_size,
		.frame_headroom = XSK_UMEM__DEFAULT_FRAME_HEADROOM,
290
		.flags = opt_umem_flags
291
	};
292

293
	int ret;
294 295

	umem = calloc(1, sizeof(*umem));
296 297
	if (!umem)
		exit_with_error(errno);
298

299
	ret = xsk_umem__create(&umem->umem, buffer, size, &umem->fq, &umem->cq,
300
			       &cfg);
301

302 303
	if (ret)
		exit_with_error(-ret);
304

305
	umem->buffer = buffer;
306 307 308
	return umem;
}

309
static struct xsk_socket_info *xsk_configure_socket(struct xsk_umem_info *umem)
310
{
311 312 313 314 315
	struct xsk_socket_config cfg;
	struct xsk_socket_info *xsk;
	int ret;
	u32 idx;
	int i;
316 317

	xsk = calloc(1, sizeof(*xsk));
318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340
	if (!xsk)
		exit_with_error(errno);

	xsk->umem = umem;
	cfg.rx_size = XSK_RING_CONS__DEFAULT_NUM_DESCS;
	cfg.tx_size = XSK_RING_PROD__DEFAULT_NUM_DESCS;
	cfg.libbpf_flags = 0;
	cfg.xdp_flags = opt_xdp_flags;
	cfg.bind_flags = opt_xdp_bind_flags;
	ret = xsk_socket__create(&xsk->xsk, opt_if, opt_queue, umem->umem,
				 &xsk->rx, &xsk->tx, &cfg);
	if (ret)
		exit_with_error(-ret);

	ret = bpf_get_link_xdp_id(opt_ifindex, &prog_id, opt_xdp_flags);
	if (ret)
		exit_with_error(-ret);

	ret = xsk_ring_prod__reserve(&xsk->umem->fq,
				     XSK_RING_PROD__DEFAULT_NUM_DESCS,
				     &idx);
	if (ret != XSK_RING_PROD__DEFAULT_NUM_DESCS)
		exit_with_error(-ret);
341 342 343
	for (i = 0; i < XSK_RING_PROD__DEFAULT_NUM_DESCS; i++)
		*xsk_ring_prod__fill_addr(&xsk->umem->fq, idx++) =
			i * opt_xsk_frame_size;
344 345
	xsk_ring_prod__submit(&xsk->umem->fq,
			      XSK_RING_PROD__DEFAULT_NUM_DESCS);
346 347 348 349 350 351 352 353 354 355 356 357 358 359

	return xsk;
}

static struct option long_options[] = {
	{"rxdrop", no_argument, 0, 'r'},
	{"txonly", no_argument, 0, 't'},
	{"l2fwd", no_argument, 0, 'l'},
	{"interface", required_argument, 0, 'i'},
	{"queue", required_argument, 0, 'q'},
	{"poll", no_argument, 0, 'p'},
	{"xdp-skb", no_argument, 0, 'S'},
	{"xdp-native", no_argument, 0, 'N'},
	{"interval", required_argument, 0, 'n'},
360 361
	{"zero-copy", no_argument, 0, 'z'},
	{"copy", no_argument, 0, 'c'},
362
	{"frame-size", required_argument, 0, 'f'},
363
	{"no-need-wakeup", no_argument, 0, 'm'},
364
	{"unaligned", no_argument, 0, 'u'},
365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381
	{0, 0, 0, 0}
};

static void usage(const char *prog)
{
	const char *str =
		"  Usage: %s [OPTIONS]\n"
		"  Options:\n"
		"  -r, --rxdrop		Discard all incoming packets (default)\n"
		"  -t, --txonly		Only send packets\n"
		"  -l, --l2fwd		MAC swap L2 forwarding\n"
		"  -i, --interface=n	Run on interface n\n"
		"  -q, --queue=n	Use queue n (default 0)\n"
		"  -p, --poll		Use poll syscall\n"
		"  -S, --xdp-skb=n	Use XDP skb-mod\n"
		"  -N, --xdp-native=n	Enfore XDP native mode\n"
		"  -n, --interval=n	Specify statistics update interval (default 1 sec).\n"
382 383
		"  -z, --zero-copy      Force zero-copy mode.\n"
		"  -c, --copy           Force copy mode.\n"
384
		"  -f, --frame-size=n   Set the frame size (must be a power of two, default is %d).\n"
385
		"  -m, --no-need-wakeup Turn off use of driver need wakeup flag.\n"
386 387
		"  -f, --frame-size=n   Set the frame size (must be a power of two in aligned mode, default is %d).\n"
		"  -u, --unaligned	Enable unaligned chunk placement\n"
388
		"\n";
389
	fprintf(stderr, str, prog, XSK_UMEM__DEFAULT_FRAME_SIZE);
390 391 392 393 394 395 396 397 398 399
	exit(EXIT_FAILURE);
}

static void parse_command_line(int argc, char **argv)
{
	int option_index, c;

	opterr = 0;

	for (;;) {
400
		c = getopt_long(argc, argv, "Frtli:q:psSNn:czf:mu",
401
				long_options, &option_index);
402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425
		if (c == -1)
			break;

		switch (c) {
		case 'r':
			opt_bench = BENCH_RXDROP;
			break;
		case 't':
			opt_bench = BENCH_TXONLY;
			break;
		case 'l':
			opt_bench = BENCH_L2FWD;
			break;
		case 'i':
			opt_if = optarg;
			break;
		case 'q':
			opt_queue = atoi(optarg);
			break;
		case 'p':
			opt_poll = 1;
			break;
		case 'S':
			opt_xdp_flags |= XDP_FLAGS_SKB_MODE;
426
			opt_xdp_bind_flags |= XDP_COPY;
427 428 429 430 431 432 433
			break;
		case 'N':
			opt_xdp_flags |= XDP_FLAGS_DRV_MODE;
			break;
		case 'n':
			opt_interval = atoi(optarg);
			break;
434 435 436 437 438 439
		case 'z':
			opt_xdp_bind_flags |= XDP_ZEROCOPY;
			break;
		case 'c':
			opt_xdp_bind_flags |= XDP_COPY;
			break;
440 441 442 443
		case 'u':
			opt_umem_flags |= XDP_UMEM_UNALIGNED_CHUNK_FLAG;
			opt_unaligned_chunks = 1;
			break;
444 445 446
		case 'F':
			opt_xdp_flags &= ~XDP_FLAGS_UPDATE_IF_NOEXIST;
			break;
447 448
		case 'f':
			opt_xsk_frame_size = atoi(optarg);
449 450 451
		case 'm':
			opt_need_wakeup = false;
			opt_xdp_bind_flags &= ~XDP_USE_NEED_WAKEUP;
452
			break;
453

454 455 456 457 458 459 460 461 462 463 464
		default:
			usage(basename(argv[0]));
		}
	}

	opt_ifindex = if_nametoindex(opt_if);
	if (!opt_ifindex) {
		fprintf(stderr, "ERROR: interface \"%s\" does not exist\n",
			opt_if);
		usage(basename(argv[0]));
	}
465

466 467
	if ((opt_xsk_frame_size & (opt_xsk_frame_size - 1)) &&
	    !opt_unaligned_chunks) {
468 469 470 471
		fprintf(stderr, "--frame-size=%d is not a power of two\n",
			opt_xsk_frame_size);
		usage(basename(argv[0]));
	}
472 473
}

474
static void kick_tx(struct xsk_socket_info *xsk)
475 476 477
{
	int ret;

478
	ret = sendto(xsk_socket__fd(xsk->xsk), NULL, 0, MSG_DONTWAIT, NULL, 0);
479
	if (ret >= 0 || errno == ENOBUFS || errno == EAGAIN || errno == EBUSY)
480
		return;
481
	exit_with_error(errno);
482 483
}

484 485
static inline void complete_tx_l2fwd(struct xsk_socket_info *xsk,
				     struct pollfd *fds)
486
{
487
	struct xsk_umem_info *umem = xsk->umem;
488
	u32 idx_cq = 0, idx_fq = 0;
489 490 491 492 493 494
	unsigned int rcvd;
	size_t ndescs;

	if (!xsk->outstanding_tx)
		return;

495 496 497
	if (!opt_need_wakeup || xsk_ring_prod__needs_wakeup(&xsk->tx))
		kick_tx(xsk);

498
	ndescs = (xsk->outstanding_tx > BATCH_SIZE) ? BATCH_SIZE :
499
		xsk->outstanding_tx;
500 501

	/* re-add completed Tx buffers */
502
	rcvd = xsk_ring_cons__peek(&umem->cq, ndescs, &idx_cq);
503
	if (rcvd > 0) {
504 505 506
		unsigned int i;
		int ret;

507
		ret = xsk_ring_prod__reserve(&umem->fq, rcvd, &idx_fq);
508 509 510
		while (ret != rcvd) {
			if (ret < 0)
				exit_with_error(-ret);
511
			if (xsk_ring_prod__needs_wakeup(&umem->fq))
512
				ret = poll(fds, num_socks, opt_timeout);
513
			ret = xsk_ring_prod__reserve(&umem->fq, rcvd, &idx_fq);
514
		}
515

516
		for (i = 0; i < rcvd; i++)
517 518
			*xsk_ring_prod__fill_addr(&umem->fq, idx_fq++) =
				*xsk_ring_cons__comp_addr(&umem->cq, idx_cq++);
519 520 521

		xsk_ring_prod__submit(&xsk->umem->fq, rcvd);
		xsk_ring_cons__release(&xsk->umem->cq, rcvd);
522 523 524 525 526
		xsk->outstanding_tx -= rcvd;
		xsk->tx_npkts += rcvd;
	}
}

527
static inline void complete_tx_only(struct xsk_socket_info *xsk)
528 529
{
	unsigned int rcvd;
530
	u32 idx;
531 532 533 534

	if (!xsk->outstanding_tx)
		return;

535 536
	if (!opt_need_wakeup || xsk_ring_prod__needs_wakeup(&xsk->tx))
		kick_tx(xsk);
537

538
	rcvd = xsk_ring_cons__peek(&xsk->umem->cq, BATCH_SIZE, &idx);
539
	if (rcvd > 0) {
540
		xsk_ring_cons__release(&xsk->umem->cq, rcvd);
541 542 543 544 545
		xsk->outstanding_tx -= rcvd;
		xsk->tx_npkts += rcvd;
	}
}

546
static void rx_drop(struct xsk_socket_info *xsk, struct pollfd *fds)
547 548
{
	unsigned int rcvd, i;
549
	u32 idx_rx = 0, idx_fq = 0;
550
	int ret;
551

552
	rcvd = xsk_ring_cons__peek(&xsk->rx, BATCH_SIZE, &idx_rx);
553 554 555
	if (!rcvd) {
		if (xsk_ring_prod__needs_wakeup(&xsk->umem->fq))
			ret = poll(fds, num_socks, opt_timeout);
556
		return;
557
	}
558

559 560 561 562
	ret = xsk_ring_prod__reserve(&xsk->umem->fq, rcvd, &idx_fq);
	while (ret != rcvd) {
		if (ret < 0)
			exit_with_error(-ret);
563 564
		if (xsk_ring_prod__needs_wakeup(&xsk->umem->fq))
			ret = poll(fds, num_socks, opt_timeout);
565 566 567
		ret = xsk_ring_prod__reserve(&xsk->umem->fq, rcvd, &idx_fq);
	}

568
	for (i = 0; i < rcvd; i++) {
569 570
		u64 addr = xsk_ring_cons__rx_desc(&xsk->rx, idx_rx)->addr;
		u32 len = xsk_ring_cons__rx_desc(&xsk->rx, idx_rx++)->len;
571 572 573
		u64 orig = xsk_umem__extract_addr(addr);

		addr = xsk_umem__add_offset_to_addr(addr);
574
		char *pkt = xsk_umem__get_data(xsk->umem->buffer, addr);
575

576
		hex_dump(pkt, len, addr);
577
		*xsk_ring_prod__fill_addr(&xsk->umem->fq, idx_fq++) = orig;
578 579
	}

580 581
	xsk_ring_prod__submit(&xsk->umem->fq, rcvd);
	xsk_ring_cons__release(&xsk->rx, rcvd);
582 583 584 585 586 587
	xsk->rx_npkts += rcvd;
}

static void rx_drop_all(void)
{
	struct pollfd fds[MAX_SOCKS + 1];
588
	int i, ret;
589 590 591 592

	memset(fds, 0, sizeof(fds));

	for (i = 0; i < num_socks; i++) {
593
		fds[i].fd = xsk_socket__fd(xsks[i]->xsk);
594 595 596 597 598
		fds[i].events = POLLIN;
	}

	for (;;) {
		if (opt_poll) {
599
			ret = poll(fds, num_socks, opt_timeout);
600 601 602 603 604
			if (ret <= 0)
				continue;
		}

		for (i = 0; i < num_socks; i++)
605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626
			rx_drop(xsks[i], fds);
	}
}

static void tx_only(struct xsk_socket_info *xsk, u32 frame_nb)
{
	u32 idx;

	if (xsk_ring_prod__reserve(&xsk->tx, BATCH_SIZE, &idx) == BATCH_SIZE) {
		unsigned int i;

		for (i = 0; i < BATCH_SIZE; i++) {
			xsk_ring_prod__tx_desc(&xsk->tx, idx + i)->addr	=
				(frame_nb + i) << XSK_UMEM__DEFAULT_FRAME_SHIFT;
			xsk_ring_prod__tx_desc(&xsk->tx, idx + i)->len =
				sizeof(pkt_data) - 1;
		}

		xsk_ring_prod__submit(&xsk->tx, BATCH_SIZE);
		xsk->outstanding_tx += BATCH_SIZE;
		frame_nb += BATCH_SIZE;
		frame_nb %= NUM_FRAMES;
627
	}
628 629

	complete_tx_only(xsk);
630 631
}

632
static void tx_only_all(void)
633
{
634 635 636
	struct pollfd fds[MAX_SOCKS];
	u32 frame_nb[MAX_SOCKS] = {};
	int i, ret;
637 638

	memset(fds, 0, sizeof(fds));
639 640 641 642
	for (i = 0; i < num_socks; i++) {
		fds[0].fd = xsk_socket__fd(xsks[i]->xsk);
		fds[0].events = POLLOUT;
	}
643 644 645

	for (;;) {
		if (opt_poll) {
646
			ret = poll(fds, num_socks, opt_timeout);
647 648 649
			if (ret <= 0)
				continue;

650
			if (!(fds[0].revents & POLLOUT))
651 652 653
				continue;
		}

654 655
		for (i = 0; i < num_socks; i++)
			tx_only(xsks[i], frame_nb[i]);
656 657 658
	}
}

659
static void l2fwd(struct xsk_socket_info *xsk, struct pollfd *fds)
660
{
661 662 663
	unsigned int rcvd, i;
	u32 idx_rx = 0, idx_tx = 0;
	int ret;
664

665
	complete_tx_l2fwd(xsk, fds);
666

667 668 669 670 671 672
	rcvd = xsk_ring_cons__peek(&xsk->rx, BATCH_SIZE, &idx_rx);
	if (!rcvd) {
		if (xsk_ring_prod__needs_wakeup(&xsk->umem->fq))
			ret = poll(fds, num_socks, opt_timeout);
		return;
	}
673

674 675 676 677 678 679
	ret = xsk_ring_prod__reserve(&xsk->tx, rcvd, &idx_tx);
	while (ret != rcvd) {
		if (ret < 0)
			exit_with_error(-ret);
		if (xsk_ring_prod__needs_wakeup(&xsk->tx))
			kick_tx(xsk);
680
		ret = xsk_ring_prod__reserve(&xsk->tx, rcvd, &idx_tx);
681 682 683 684 685
	}

	for (i = 0; i < rcvd; i++) {
		u64 addr = xsk_ring_cons__rx_desc(&xsk->rx, idx_rx)->addr;
		u32 len = xsk_ring_cons__rx_desc(&xsk->rx, idx_rx++)->len;
686 687 688
		u64 orig = xsk_umem__extract_addr(addr);

		addr = xsk_umem__add_offset_to_addr(addr);
689 690 691
		char *pkt = xsk_umem__get_data(xsk->umem->buffer, addr);

		swap_mac_addresses(pkt);
692

693
		hex_dump(pkt, len, addr);
694
		xsk_ring_prod__tx_desc(&xsk->tx, idx_tx)->addr = orig;
695 696
		xsk_ring_prod__tx_desc(&xsk->tx, idx_tx++)->len = len;
	}
697

698 699
	xsk_ring_prod__submit(&xsk->tx, rcvd);
	xsk_ring_cons__release(&xsk->rx, rcvd);
700

701 702 703 704 705 706 707 708 709 710 711 712 713 714 715
	xsk->rx_npkts += rcvd;
	xsk->outstanding_tx += rcvd;
}

static void l2fwd_all(void)
{
	struct pollfd fds[MAX_SOCKS];
	int i, ret;

	memset(fds, 0, sizeof(fds));

	for (i = 0; i < num_socks; i++) {
		fds[i].fd = xsk_socket__fd(xsks[i]->xsk);
		fds[i].events = POLLOUT | POLLIN;
	}
716

717 718 719 720 721 722
	for (;;) {
		if (opt_poll) {
			ret = poll(fds, num_socks, opt_timeout);
			if (ret <= 0)
				continue;
		}
723

724 725
		for (i = 0; i < num_socks; i++)
			l2fwd(xsks[i], fds);
726 727 728 729 730 731
	}
}

int main(int argc, char **argv)
{
	struct rlimit r = {RLIM_INFINITY, RLIM_INFINITY};
732
	struct xsk_umem_info *umem;
733
	pthread_t pt;
734 735
	void *bufs;
	int ret;
736 737 738 739 740 741 742 743 744

	parse_command_line(argc, argv);

	if (setrlimit(RLIMIT_MEMLOCK, &r)) {
		fprintf(stderr, "ERROR: setrlimit(RLIMIT_MEMLOCK) \"%s\"\n",
			strerror(errno));
		exit(EXIT_FAILURE);
	}

745
	ret = posix_memalign(&bufs, getpagesize(), /* PAGE_SIZE aligned */
746
			     NUM_FRAMES * opt_xsk_frame_size);
747 748
	if (ret)
		exit_with_error(ret);
749

750
       /* Create sockets... */
751
	umem = xsk_configure_umem(bufs, NUM_FRAMES * opt_xsk_frame_size);
752
	xsks[num_socks++] = xsk_configure_socket(umem);
753

754 755
	if (opt_bench == BENCH_TXONLY) {
		int i;
756

757 758
		for (i = 0; i < NUM_FRAMES; i++)
			(void)gen_eth_frame(umem, i * opt_xsk_frame_size);
759 760 761 762 763 764 765 766 767
	}

	signal(SIGINT, int_exit);
	signal(SIGTERM, int_exit);
	signal(SIGABRT, int_exit);

	setlocale(LC_ALL, "");

	ret = pthread_create(&pt, NULL, poller, NULL);
768 769
	if (ret)
		exit_with_error(ret);
770 771 772 773 774 775

	prev_time = get_nsecs();

	if (opt_bench == BENCH_RXDROP)
		rx_drop_all();
	else if (opt_bench == BENCH_TXONLY)
776
		tx_only_all();
777
	else
778
		l2fwd_all();
779 780 781

	return 0;
}