xdpsock_user.c 18.0 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
static u32 opt_umem_flags;
static int opt_unaligned_chunks;
73
static int opt_mmap_flags;
74
static u32 opt_xdp_bind_flags;
75
static int opt_xsk_frame_size = XSK_UMEM__DEFAULT_FRAME_SIZE;
76 77
static int opt_timeout = 1000;
static bool opt_need_wakeup = true;
78
static __u32 prog_id;
79

80 81 82 83 84
struct xsk_umem_info {
	struct xsk_ring_prod fq;
	struct xsk_ring_cons cq;
	struct xsk_umem *umem;
	void *buffer;
85 86
};

87 88 89 90 91
struct xsk_socket_info {
	struct xsk_ring_cons rx;
	struct xsk_ring_prod tx;
	struct xsk_umem_info *umem;
	struct xsk_socket *xsk;
92 93 94 95
	unsigned long rx_npkts;
	unsigned long tx_npkts;
	unsigned long prev_rx_npkts;
	unsigned long prev_tx_npkts;
96
	u32 outstanding_tx;
97 98 99
};

static int num_socks;
100
struct xsk_socket_info *xsks[MAX_SOCKS];
101 102 103 104 105 106 107 108 109

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

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

110
static void print_benchmark(bool running)
111
{
112
	const char *bench_str = "INVALID";
113

114 115 116 117 118 119
	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";
120

121 122 123 124 125 126 127
	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("	");
128

129 130
	if (opt_poll)
		printf("poll() ");
131

132 133 134
	if (running) {
		printf("running...");
		fflush(stdout);
135 136 137
	}
}

138
static void dump_stats(void)
139
{
140 141 142
	unsigned long now = get_nsecs();
	long dt = now - prev_time;
	int i;
143

144
	prev_time = now;
145

146 147 148
	for (i = 0; i < num_socks && xsks[i]; i++) {
		char *fmt = "%-15s %'-11.0f %'-11lu\n";
		double rx_pps, tx_pps;
149

150 151 152 153
		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;
154

155 156 157
		printf("\n sock%d@", i);
		print_benchmark(false);
		printf("\n");
158

159 160 161 162
		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);
163

164 165
		xsks[i]->prev_rx_npkts = xsks[i]->rx_npkts;
		xsks[i]->prev_tx_npkts = xsks[i]->tx_npkts;
166 167 168
	}
}

169
static void *poller(void *arg)
170
{
171 172 173 174
	(void)arg;
	for (;;) {
		sleep(opt_interval);
		dump_stats();
175 176
	}

177
	return NULL;
178 179
}

180
static void remove_xdp_program(void)
181
{
182
	__u32 curr_prog_id = 0;
183

184 185 186
	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);
187
	}
188 189 190 191 192 193
	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");
194 195
}

196
static void int_exit(int sig)
197
{
198
	struct xsk_umem *umem = xsks[0]->umem->umem;
199

200
	(void)sig;
201

202 203 204 205
	dump_stats();
	xsk_socket__delete(xsks[0]->xsk);
	(void)xsk_umem__delete(umem);
	remove_xdp_program();
206

207
	exit(EXIT_SUCCESS);
208 209
}

210 211
static void __exit_with_error(int error, const char *file, const char *func,
			      int line)
212
{
213 214 215 216 217
	fprintf(stderr, "%s:%s:%i: errno: %d/\"%s\"\n", file, func,
		line, error, strerror(error));
	dump_stats();
	remove_xdp_program();
	exit(EXIT_FAILURE);
218 219
}

220 221
#define exit_with_error(error) __exit_with_error(error, __FILE__, __func__, \
						 __LINE__)
222

223 224 225 226 227
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";
228 229 230 231 232 233 234 235 236 237 238 239 240

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
241
static void hex_dump(void *pkt, size_t length, u64 addr)
242 243 244 245 246
{
	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
247 248
	char buf[32];
	int i = 0;
249

B
Björn Töpel 已提交
250 251 252 253
	if (!DEBUG_HEXDUMP)
		return;

	sprintf(buf, "addr=%llu", addr);
254
	printf("length = %zu\n", length);
B
Björn Töpel 已提交
255
	printf("%s | ", buf);
256 257 258 259 260 261 262 263 264 265 266 267 268 269
	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
270
				printf("%s | ", buf);
271 272 273 274 275
		}
	}
	printf("\n");
}

276
static size_t gen_eth_frame(struct xsk_umem_info *umem, u64 addr)
277
{
278 279
	memcpy(xsk_umem__get_data(umem->buffer, addr), pkt_data,
	       sizeof(pkt_data) - 1);
280 281 282
	return sizeof(pkt_data) - 1;
}

283
static struct xsk_umem_info *xsk_configure_umem(void *buffer, u64 size)
284
{
285
	struct xsk_umem_info *umem;
286 287 288 289 290
	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,
291
		.flags = opt_umem_flags
292
	};
293

294
	int ret;
295 296

	umem = calloc(1, sizeof(*umem));
297 298
	if (!umem)
		exit_with_error(errno);
299

300
	ret = xsk_umem__create(&umem->umem, buffer, size, &umem->fq, &umem->cq,
301
			       &cfg);
302

303 304
	if (ret)
		exit_with_error(-ret);
305

306
	umem->buffer = buffer;
307 308 309
	return umem;
}

310
static struct xsk_socket_info *xsk_configure_socket(struct xsk_umem_info *umem)
311
{
312 313 314 315 316
	struct xsk_socket_config cfg;
	struct xsk_socket_info *xsk;
	int ret;
	u32 idx;
	int i;
317 318

	xsk = calloc(1, sizeof(*xsk));
319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341
	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);
342 343 344
	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;
345 346
	xsk_ring_prod__submit(&xsk->umem->fq,
			      XSK_RING_PROD__DEFAULT_NUM_DESCS);
347 348 349 350 351 352 353 354 355 356 357 358 359 360

	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'},
361 362
	{"zero-copy", no_argument, 0, 'z'},
	{"copy", no_argument, 0, 'c'},
363
	{"frame-size", required_argument, 0, 'f'},
364
	{"no-need-wakeup", no_argument, 0, 'm'},
365
	{"unaligned", no_argument, 0, 'u'},
366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382
	{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"
383 384
		"  -z, --zero-copy      Force zero-copy mode.\n"
		"  -c, --copy           Force copy mode.\n"
385
		"  -f, --frame-size=n   Set the frame size (must be a power of two, default is %d).\n"
386
		"  -m, --no-need-wakeup Turn off use of driver need wakeup flag.\n"
387 388
		"  -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"
389
		"\n";
390
	fprintf(stderr, str, prog, XSK_UMEM__DEFAULT_FRAME_SIZE);
391 392 393 394 395 396 397 398 399 400
	exit(EXIT_FAILURE);
}

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

	opterr = 0;

	for (;;) {
401
		c = getopt_long(argc, argv, "Frtli:q:psSNn:czf:mu",
402
				long_options, &option_index);
403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426
		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;
427
			opt_xdp_bind_flags |= XDP_COPY;
428 429 430 431 432 433 434
			break;
		case 'N':
			opt_xdp_flags |= XDP_FLAGS_DRV_MODE;
			break;
		case 'n':
			opt_interval = atoi(optarg);
			break;
435 436 437 438 439 440
		case 'z':
			opt_xdp_bind_flags |= XDP_ZEROCOPY;
			break;
		case 'c':
			opt_xdp_bind_flags |= XDP_COPY;
			break;
441 442 443
		case 'u':
			opt_umem_flags |= XDP_UMEM_UNALIGNED_CHUNK_FLAG;
			opt_unaligned_chunks = 1;
444
			opt_mmap_flags = MAP_HUGETLB;
445
			break;
446 447 448
		case 'F':
			opt_xdp_flags &= ~XDP_FLAGS_UPDATE_IF_NOEXIST;
			break;
449 450
		case 'f':
			opt_xsk_frame_size = atoi(optarg);
451 452 453
		case 'm':
			opt_need_wakeup = false;
			opt_xdp_bind_flags &= ~XDP_USE_NEED_WAKEUP;
454
			break;
455

456 457 458 459 460 461 462 463 464 465 466
		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]));
	}
467

468 469
	if ((opt_xsk_frame_size & (opt_xsk_frame_size - 1)) &&
	    !opt_unaligned_chunks) {
470 471 472 473
		fprintf(stderr, "--frame-size=%d is not a power of two\n",
			opt_xsk_frame_size);
		usage(basename(argv[0]));
	}
474 475
}

476
static void kick_tx(struct xsk_socket_info *xsk)
477 478 479
{
	int ret;

480
	ret = sendto(xsk_socket__fd(xsk->xsk), NULL, 0, MSG_DONTWAIT, NULL, 0);
481
	if (ret >= 0 || errno == ENOBUFS || errno == EAGAIN || errno == EBUSY)
482
		return;
483
	exit_with_error(errno);
484 485
}

486 487
static inline void complete_tx_l2fwd(struct xsk_socket_info *xsk,
				     struct pollfd *fds)
488
{
489
	struct xsk_umem_info *umem = xsk->umem;
490
	u32 idx_cq = 0, idx_fq = 0;
491 492 493 494 495 496
	unsigned int rcvd;
	size_t ndescs;

	if (!xsk->outstanding_tx)
		return;

497 498 499
	if (!opt_need_wakeup || xsk_ring_prod__needs_wakeup(&xsk->tx))
		kick_tx(xsk);

500
	ndescs = (xsk->outstanding_tx > BATCH_SIZE) ? BATCH_SIZE :
501
		xsk->outstanding_tx;
502 503

	/* re-add completed Tx buffers */
504
	rcvd = xsk_ring_cons__peek(&umem->cq, ndescs, &idx_cq);
505
	if (rcvd > 0) {
506 507 508
		unsigned int i;
		int ret;

509
		ret = xsk_ring_prod__reserve(&umem->fq, rcvd, &idx_fq);
510 511 512
		while (ret != rcvd) {
			if (ret < 0)
				exit_with_error(-ret);
513
			if (xsk_ring_prod__needs_wakeup(&umem->fq))
514
				ret = poll(fds, num_socks, opt_timeout);
515
			ret = xsk_ring_prod__reserve(&umem->fq, rcvd, &idx_fq);
516
		}
517

518
		for (i = 0; i < rcvd; i++)
519 520
			*xsk_ring_prod__fill_addr(&umem->fq, idx_fq++) =
				*xsk_ring_cons__comp_addr(&umem->cq, idx_cq++);
521 522 523

		xsk_ring_prod__submit(&xsk->umem->fq, rcvd);
		xsk_ring_cons__release(&xsk->umem->cq, rcvd);
524 525 526 527 528
		xsk->outstanding_tx -= rcvd;
		xsk->tx_npkts += rcvd;
	}
}

529
static inline void complete_tx_only(struct xsk_socket_info *xsk)
530 531
{
	unsigned int rcvd;
532
	u32 idx;
533 534 535 536

	if (!xsk->outstanding_tx)
		return;

537 538
	if (!opt_need_wakeup || xsk_ring_prod__needs_wakeup(&xsk->tx))
		kick_tx(xsk);
539

540
	rcvd = xsk_ring_cons__peek(&xsk->umem->cq, BATCH_SIZE, &idx);
541
	if (rcvd > 0) {
542
		xsk_ring_cons__release(&xsk->umem->cq, rcvd);
543 544 545 546 547
		xsk->outstanding_tx -= rcvd;
		xsk->tx_npkts += rcvd;
	}
}

548
static void rx_drop(struct xsk_socket_info *xsk, struct pollfd *fds)
549 550
{
	unsigned int rcvd, i;
551
	u32 idx_rx = 0, idx_fq = 0;
552
	int ret;
553

554
	rcvd = xsk_ring_cons__peek(&xsk->rx, BATCH_SIZE, &idx_rx);
555 556 557
	if (!rcvd) {
		if (xsk_ring_prod__needs_wakeup(&xsk->umem->fq))
			ret = poll(fds, num_socks, opt_timeout);
558
		return;
559
	}
560

561 562 563 564
	ret = xsk_ring_prod__reserve(&xsk->umem->fq, rcvd, &idx_fq);
	while (ret != rcvd) {
		if (ret < 0)
			exit_with_error(-ret);
565 566
		if (xsk_ring_prod__needs_wakeup(&xsk->umem->fq))
			ret = poll(fds, num_socks, opt_timeout);
567 568 569
		ret = xsk_ring_prod__reserve(&xsk->umem->fq, rcvd, &idx_fq);
	}

570
	for (i = 0; i < rcvd; i++) {
571 572
		u64 addr = xsk_ring_cons__rx_desc(&xsk->rx, idx_rx)->addr;
		u32 len = xsk_ring_cons__rx_desc(&xsk->rx, idx_rx++)->len;
573 574 575
		u64 orig = xsk_umem__extract_addr(addr);

		addr = xsk_umem__add_offset_to_addr(addr);
576
		char *pkt = xsk_umem__get_data(xsk->umem->buffer, addr);
577

578
		hex_dump(pkt, len, addr);
579
		*xsk_ring_prod__fill_addr(&xsk->umem->fq, idx_fq++) = orig;
580 581
	}

582 583
	xsk_ring_prod__submit(&xsk->umem->fq, rcvd);
	xsk_ring_cons__release(&xsk->rx, rcvd);
584 585 586 587 588 589
	xsk->rx_npkts += rcvd;
}

static void rx_drop_all(void)
{
	struct pollfd fds[MAX_SOCKS + 1];
590
	int i, ret;
591 592 593 594

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

	for (i = 0; i < num_socks; i++) {
595
		fds[i].fd = xsk_socket__fd(xsks[i]->xsk);
596 597 598 599 600
		fds[i].events = POLLIN;
	}

	for (;;) {
		if (opt_poll) {
601
			ret = poll(fds, num_socks, opt_timeout);
602 603 604 605 606
			if (ret <= 0)
				continue;
		}

		for (i = 0; i < num_socks; i++)
607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628
			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;
629
	}
630 631

	complete_tx_only(xsk);
632 633
}

634
static void tx_only_all(void)
635
{
636 637 638
	struct pollfd fds[MAX_SOCKS];
	u32 frame_nb[MAX_SOCKS] = {};
	int i, ret;
639 640

	memset(fds, 0, sizeof(fds));
641 642 643 644
	for (i = 0; i < num_socks; i++) {
		fds[0].fd = xsk_socket__fd(xsks[i]->xsk);
		fds[0].events = POLLOUT;
	}
645 646 647

	for (;;) {
		if (opt_poll) {
648
			ret = poll(fds, num_socks, opt_timeout);
649 650 651
			if (ret <= 0)
				continue;

652
			if (!(fds[0].revents & POLLOUT))
653 654 655
				continue;
		}

656 657
		for (i = 0; i < num_socks; i++)
			tx_only(xsks[i], frame_nb[i]);
658 659 660
	}
}

661
static void l2fwd(struct xsk_socket_info *xsk, struct pollfd *fds)
662
{
663 664 665
	unsigned int rcvd, i;
	u32 idx_rx = 0, idx_tx = 0;
	int ret;
666

667
	complete_tx_l2fwd(xsk, fds);
668

669 670 671 672 673 674
	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;
	}
675

676 677 678 679 680 681
	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);
682
		ret = xsk_ring_prod__reserve(&xsk->tx, rcvd, &idx_tx);
683 684 685 686 687
	}

	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;
688
		u64 orig = addr;
689 690

		addr = xsk_umem__add_offset_to_addr(addr);
691 692 693
		char *pkt = xsk_umem__get_data(xsk->umem->buffer, addr);

		swap_mac_addresses(pkt);
694

695
		hex_dump(pkt, len, addr);
696
		xsk_ring_prod__tx_desc(&xsk->tx, idx_tx)->addr = orig;
697 698
		xsk_ring_prod__tx_desc(&xsk->tx, idx_tx++)->len = len;
	}
699

700 701
	xsk_ring_prod__submit(&xsk->tx, rcvd);
	xsk_ring_cons__release(&xsk->rx, rcvd);
702

703 704 705 706 707 708 709 710 711 712 713 714 715 716 717
	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;
	}
718

719 720 721 722 723 724
	for (;;) {
		if (opt_poll) {
			ret = poll(fds, num_socks, opt_timeout);
			if (ret <= 0)
				continue;
		}
725

726 727
		for (i = 0; i < num_socks; i++)
			l2fwd(xsks[i], fds);
728 729 730 731 732 733
	}
}

int main(int argc, char **argv)
{
	struct rlimit r = {RLIM_INFINITY, RLIM_INFINITY};
734
	struct xsk_umem_info *umem;
735
	pthread_t pt;
736 737
	void *bufs;
	int ret;
738 739 740 741 742 743 744 745 746

	parse_command_line(argc, argv);

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

747 748 749 750 751 752 753 754
	/* Reserve memory for the umem. Use hugepages if unaligned chunk mode */
	bufs = mmap(NULL, NUM_FRAMES * opt_xsk_frame_size,
		    PROT_READ | PROT_WRITE,
		    MAP_PRIVATE | MAP_ANONYMOUS | opt_mmap_flags, -1, 0);
	if (bufs == MAP_FAILED) {
		printf("ERROR: mmap failed\n");
		exit(EXIT_FAILURE);
	}
755
       /* Create sockets... */
756
	umem = xsk_configure_umem(bufs, NUM_FRAMES * opt_xsk_frame_size);
757
	xsks[num_socks++] = xsk_configure_socket(umem);
758

759 760
	if (opt_bench == BENCH_TXONLY) {
		int i;
761

762 763
		for (i = 0; i < NUM_FRAMES; i++)
			(void)gen_eth_frame(umem, i * opt_xsk_frame_size);
764 765 766 767 768 769 770 771 772
	}

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

	setlocale(LC_ALL, "");

	ret = pthread_create(&pt, NULL, poller, NULL);
773 774
	if (ret)
		exit_with_error(ret);
775 776 777 778 779 780

	prev_time = get_nsecs();

	if (opt_bench == BENCH_RXDROP)
		rx_drop_all();
	else if (opt_bench == BENCH_TXONLY)
781
		tx_only_all();
782
	else
783
		l2fwd_all();
784 785 786

	return 0;
}