1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24#define _GNU_SOURCE
25
26#include <arpa/inet.h>
27#include <errno.h>
28#include <error.h>
29#include <fcntl.h>
30#include <getopt.h>
31#include <linux/filter.h>
32#include <linux/if_ether.h>
33#include <linux/if_packet.h>
34#include <net/if.h>
35#include <netdb.h>
36#include <netinet/ip.h>
37#include <netinet/ip6.h>
38#include <netinet/tcp.h>
39#include <netinet/udp.h>
40#include <poll.h>
41#include <stdbool.h>
42#include <stddef.h>
43#include <stdint.h>
44#include <stdio.h>
45#include <stdlib.h>
46#include <string.h>
47#include <sys/mman.h>
48#include <sys/socket.h>
49#include <sys/stat.h>
50#include <sys/sysinfo.h>
51#include <sys/time.h>
52#include <sys/types.h>
53#include <unistd.h>
54
55#define TOEPLITZ_KEY_MIN_LEN 40
56#define TOEPLITZ_KEY_MAX_LEN 60
57
58#define TOEPLITZ_STR_LEN(K) (((K) * 3) - 1)
59#define TOEPLITZ_STR_MIN_LEN TOEPLITZ_STR_LEN(TOEPLITZ_KEY_MIN_LEN)
60#define TOEPLITZ_STR_MAX_LEN TOEPLITZ_STR_LEN(TOEPLITZ_KEY_MAX_LEN)
61
62#define FOUR_TUPLE_MAX_LEN ((sizeof(struct in6_addr) * 2) + (sizeof(uint16_t) * 2))
63
64#define RSS_MAX_CPUS (1 << 16)
65
66#define RPS_MAX_CPUS 16UL
67
68
69static uint16_t cfg_dport = 8000;
70static int cfg_family = AF_INET6;
71static char *cfg_ifname = "eth0";
72static int cfg_num_queues;
73static int cfg_num_rps_cpus;
74static bool cfg_sink;
75static int cfg_type = SOCK_STREAM;
76static int cfg_timeout_msec = 1000;
77static bool cfg_verbose;
78
79
80static int num_cpus;
81static int ring_block_nr;
82static int ring_block_sz;
83
84
85static int frames_received;
86static int frames_nohash;
87static int frames_error;
88
89#define log_verbose(args...) do { if (cfg_verbose) fprintf(stderr, args); } while (0)
90
91
92struct ring_state {
93 int fd;
94 char *mmap;
95 int idx;
96 int cpu;
97};
98
99static unsigned int rx_irq_cpus[RSS_MAX_CPUS];
100static int rps_silo_to_cpu[RPS_MAX_CPUS];
101static unsigned char toeplitz_key[TOEPLITZ_KEY_MAX_LEN];
102static struct ring_state rings[RSS_MAX_CPUS];
103
104static inline uint32_t toeplitz(const unsigned char *four_tuple,
105 const unsigned char *key)
106{
107 int i, bit, ret = 0;
108 uint32_t key32;
109
110 key32 = ntohl(*((uint32_t *)key));
111 key += 4;
112
113 for (i = 0; i < FOUR_TUPLE_MAX_LEN; i++) {
114 for (bit = 7; bit >= 0; bit--) {
115 if (four_tuple[i] & (1 << bit))
116 ret ^= key32;
117
118 key32 <<= 1;
119 key32 |= !!(key[0] & (1 << bit));
120 }
121 key++;
122 }
123
124 return ret;
125}
126
127
128static void verify_rss(uint32_t rx_hash, int cpu)
129{
130 int queue = rx_hash % cfg_num_queues;
131
132 log_verbose(" rxq %d (cpu %d)", queue, rx_irq_cpus[queue]);
133 if (rx_irq_cpus[queue] != cpu) {
134 log_verbose(". error: rss cpu mismatch (%d)", cpu);
135 frames_error++;
136 }
137}
138
139static void verify_rps(uint64_t rx_hash, int cpu)
140{
141 int silo = (rx_hash * cfg_num_rps_cpus) >> 32;
142
143 log_verbose(" silo %d (cpu %d)", silo, rps_silo_to_cpu[silo]);
144 if (rps_silo_to_cpu[silo] != cpu) {
145 log_verbose(". error: rps cpu mismatch (%d)", cpu);
146 frames_error++;
147 }
148}
149
150static void log_rxhash(int cpu, uint32_t rx_hash,
151 const char *addrs, int addr_len)
152{
153 char saddr[INET6_ADDRSTRLEN], daddr[INET6_ADDRSTRLEN];
154 uint16_t *ports;
155
156 if (!inet_ntop(cfg_family, addrs, saddr, sizeof(saddr)) ||
157 !inet_ntop(cfg_family, addrs + addr_len, daddr, sizeof(daddr)))
158 error(1, 0, "address parse error");
159
160 ports = (void *)addrs + (addr_len * 2);
161 log_verbose("cpu %d: rx_hash 0x%08x [saddr %s daddr %s sport %02hu dport %02hu]",
162 cpu, rx_hash, saddr, daddr,
163 ntohs(ports[0]), ntohs(ports[1]));
164}
165
166
167static void verify_rxhash(const char *pkt, uint32_t rx_hash, int cpu)
168{
169 unsigned char four_tuple[FOUR_TUPLE_MAX_LEN] = {0};
170 uint32_t rx_hash_sw;
171 const char *addrs;
172 int addr_len;
173
174 if (cfg_family == AF_INET) {
175 addr_len = sizeof(struct in_addr);
176 addrs = pkt + offsetof(struct iphdr, saddr);
177 } else {
178 addr_len = sizeof(struct in6_addr);
179 addrs = pkt + offsetof(struct ip6_hdr, ip6_src);
180 }
181
182 memcpy(four_tuple, addrs, (addr_len * 2) + (sizeof(uint16_t) * 2));
183 rx_hash_sw = toeplitz(four_tuple, toeplitz_key);
184
185 if (cfg_verbose)
186 log_rxhash(cpu, rx_hash, addrs, addr_len);
187
188 if (rx_hash != rx_hash_sw) {
189 log_verbose(" != expected 0x%x\n", rx_hash_sw);
190 frames_error++;
191 return;
192 }
193
194 log_verbose(" OK");
195 if (cfg_num_queues)
196 verify_rss(rx_hash, cpu);
197 else if (cfg_num_rps_cpus)
198 verify_rps(rx_hash, cpu);
199 log_verbose("\n");
200}
201
202static char *recv_frame(const struct ring_state *ring, char *frame)
203{
204 struct tpacket3_hdr *hdr = (void *)frame;
205
206 if (hdr->hv1.tp_rxhash)
207 verify_rxhash(frame + hdr->tp_net, hdr->hv1.tp_rxhash,
208 ring->cpu);
209 else
210 frames_nohash++;
211
212 return frame + hdr->tp_next_offset;
213}
214
215
216static void recv_block(struct ring_state *ring)
217{
218 struct tpacket_block_desc *block;
219 char *frame;
220 int i;
221
222 block = (void *)(ring->mmap + ring->idx * ring_block_sz);
223 if (!(block->hdr.bh1.block_status & TP_STATUS_USER))
224 return;
225
226 frame = (char *)block;
227 frame += block->hdr.bh1.offset_to_first_pkt;
228
229 for (i = 0; i < block->hdr.bh1.num_pkts; i++) {
230 frame = recv_frame(ring, frame);
231 frames_received++;
232 }
233
234 block->hdr.bh1.block_status = TP_STATUS_KERNEL;
235 ring->idx = (ring->idx + 1) % ring_block_nr;
236}
237
238
239static void process_rings(void)
240{
241 int i;
242
243 usleep(1000 * cfg_timeout_msec);
244
245 for (i = 0; i < num_cpus; i++)
246 recv_block(&rings[i]);
247
248 fprintf(stderr, "count: pass=%u nohash=%u fail=%u\n",
249 frames_received - frames_nohash - frames_error,
250 frames_nohash, frames_error);
251}
252
253static char *setup_ring(int fd)
254{
255 struct tpacket_req3 req3 = {0};
256 void *ring;
257
258 req3.tp_retire_blk_tov = cfg_timeout_msec;
259 req3.tp_feature_req_word = TP_FT_REQ_FILL_RXHASH;
260
261 req3.tp_frame_size = 2048;
262 req3.tp_frame_nr = 1 << 10;
263 req3.tp_block_nr = 2;
264
265 req3.tp_block_size = req3.tp_frame_size * req3.tp_frame_nr;
266 req3.tp_block_size /= req3.tp_block_nr;
267
268 if (setsockopt(fd, SOL_PACKET, PACKET_RX_RING, &req3, sizeof(req3)))
269 error(1, errno, "setsockopt PACKET_RX_RING");
270
271 ring_block_sz = req3.tp_block_size;
272 ring_block_nr = req3.tp_block_nr;
273
274 ring = mmap(0, req3.tp_block_size * req3.tp_block_nr,
275 PROT_READ | PROT_WRITE,
276 MAP_SHARED | MAP_LOCKED | MAP_POPULATE, fd, 0);
277 if (ring == MAP_FAILED)
278 error(1, 0, "mmap failed");
279
280 return ring;
281}
282
283static void __set_filter(int fd, int off_proto, uint8_t proto, int off_dport)
284{
285 struct sock_filter filter[] = {
286 BPF_STMT(BPF_LD + BPF_B + BPF_ABS, SKF_AD_OFF + SKF_AD_PKTTYPE),
287 BPF_JUMP(BPF_JMP + BPF_JEQ + BPF_K, PACKET_HOST, 0, 4),
288 BPF_STMT(BPF_LD + BPF_B + BPF_ABS, off_proto),
289 BPF_JUMP(BPF_JMP + BPF_JEQ + BPF_K, proto, 0, 2),
290 BPF_STMT(BPF_LD + BPF_H + BPF_ABS, off_dport),
291 BPF_JUMP(BPF_JMP + BPF_JEQ + BPF_K, cfg_dport, 1, 0),
292 BPF_STMT(BPF_RET + BPF_K, 0),
293 BPF_STMT(BPF_RET + BPF_K, 0xFFFF),
294 };
295 struct sock_fprog prog = {};
296
297 prog.filter = filter;
298 prog.len = sizeof(filter) / sizeof(struct sock_filter);
299 if (setsockopt(fd, SOL_SOCKET, SO_ATTACH_FILTER, &prog, sizeof(prog)))
300 error(1, errno, "setsockopt filter");
301}
302
303
304static void set_filter(int fd)
305{
306 const int off_dport = offsetof(struct tcphdr, dest);
307 uint8_t proto;
308
309 proto = cfg_type == SOCK_STREAM ? IPPROTO_TCP : IPPROTO_UDP;
310 if (cfg_family == AF_INET)
311 __set_filter(fd, offsetof(struct iphdr, protocol), proto,
312 sizeof(struct iphdr) + off_dport);
313 else
314 __set_filter(fd, offsetof(struct ip6_hdr, ip6_nxt), proto,
315 sizeof(struct ip6_hdr) + off_dport);
316}
317
318
319static void set_filter_null(int fd)
320{
321 struct sock_filter filter[] = {
322 BPF_STMT(BPF_RET + BPF_K, 0),
323 };
324 struct sock_fprog prog = {};
325
326 prog.filter = filter;
327 prog.len = sizeof(filter) / sizeof(struct sock_filter);
328 if (setsockopt(fd, SOL_SOCKET, SO_ATTACH_FILTER, &prog, sizeof(prog)))
329 error(1, errno, "setsockopt filter");
330}
331
332static int create_ring(char **ring)
333{
334 struct fanout_args args = {
335 .id = 1,
336 .type_flags = PACKET_FANOUT_CPU,
337 .max_num_members = RSS_MAX_CPUS
338 };
339 struct sockaddr_ll ll = { 0 };
340 int fd, val;
341
342 fd = socket(PF_PACKET, SOCK_DGRAM, 0);
343 if (fd == -1)
344 error(1, errno, "socket creation failed");
345
346 val = TPACKET_V3;
347 if (setsockopt(fd, SOL_PACKET, PACKET_VERSION, &val, sizeof(val)))
348 error(1, errno, "setsockopt PACKET_VERSION");
349 *ring = setup_ring(fd);
350
351
352
353
354 set_filter_null(fd);
355
356 ll.sll_family = AF_PACKET;
357 ll.sll_ifindex = if_nametoindex(cfg_ifname);
358 ll.sll_protocol = cfg_family == AF_INET ? htons(ETH_P_IP) :
359 htons(ETH_P_IPV6);
360 if (bind(fd, (void *)&ll, sizeof(ll)))
361 error(1, errno, "bind");
362
363
364 if (setsockopt(fd, SOL_PACKET, PACKET_FANOUT, &args, sizeof(args))) {
365
366
367
368
369
370 if (cfg_num_queues || num_cpus > 256 ||
371 setsockopt(fd, SOL_PACKET, PACKET_FANOUT,
372 &args, sizeof(uint32_t)))
373 error(1, errno, "setsockopt PACKET_FANOUT cpu");
374 }
375
376 return fd;
377}
378
379
380static int setup_sink(void)
381{
382 int fd, val;
383
384 fd = socket(cfg_family, cfg_type, 0);
385 if (fd == -1)
386 error(1, errno, "socket %d.%d", cfg_family, cfg_type);
387
388 val = 1 << 20;
389 if (setsockopt(fd, SOL_SOCKET, SO_RCVBUFFORCE, &val, sizeof(val)))
390 error(1, errno, "setsockopt rcvbuf");
391
392 return fd;
393}
394
395static void setup_rings(void)
396{
397 int i;
398
399 for (i = 0; i < num_cpus; i++) {
400 rings[i].cpu = i;
401 rings[i].fd = create_ring(&rings[i].mmap);
402 }
403
404
405 for (i = 0; i < num_cpus; i++)
406 set_filter(rings[i].fd);
407}
408
409static void cleanup_rings(void)
410{
411 int i;
412
413 for (i = 0; i < num_cpus; i++) {
414 if (munmap(rings[i].mmap, ring_block_nr * ring_block_sz))
415 error(1, errno, "munmap");
416 if (close(rings[i].fd))
417 error(1, errno, "close");
418 }
419}
420
421static void parse_cpulist(const char *arg)
422{
423 do {
424 rx_irq_cpus[cfg_num_queues++] = strtol(arg, NULL, 10);
425
426 arg = strchr(arg, ',');
427 if (!arg)
428 break;
429 arg++;
430 } while (1);
431}
432
433static void show_cpulist(void)
434{
435 int i;
436
437 for (i = 0; i < cfg_num_queues; i++)
438 fprintf(stderr, "rxq %d: cpu %d\n", i, rx_irq_cpus[i]);
439}
440
441static void show_silos(void)
442{
443 int i;
444
445 for (i = 0; i < cfg_num_rps_cpus; i++)
446 fprintf(stderr, "silo %d: cpu %d\n", i, rps_silo_to_cpu[i]);
447}
448
449static void parse_toeplitz_key(const char *str, int slen, unsigned char *key)
450{
451 int i, ret, off;
452
453 if (slen < TOEPLITZ_STR_MIN_LEN ||
454 slen > TOEPLITZ_STR_MAX_LEN + 1)
455 error(1, 0, "invalid toeplitz key");
456
457 for (i = 0, off = 0; off < slen; i++, off += 3) {
458 ret = sscanf(str + off, "%hhx", &key[i]);
459 if (ret != 1)
460 error(1, 0, "key parse error at %d off %d len %d",
461 i, off, slen);
462 }
463}
464
465static void parse_rps_bitmap(const char *arg)
466{
467 unsigned long bitmap;
468 int i;
469
470 bitmap = strtoul(arg, NULL, 0);
471
472 if (bitmap & ~(RPS_MAX_CPUS - 1))
473 error(1, 0, "rps bitmap 0x%lx out of bounds 0..%lu",
474 bitmap, RPS_MAX_CPUS - 1);
475
476 for (i = 0; i < RPS_MAX_CPUS; i++)
477 if (bitmap & 1UL << i)
478 rps_silo_to_cpu[cfg_num_rps_cpus++] = i;
479}
480
481static void parse_opts(int argc, char **argv)
482{
483 static struct option long_options[] = {
484 {"dport", required_argument, 0, 'd'},
485 {"cpus", required_argument, 0, 'C'},
486 {"key", required_argument, 0, 'k'},
487 {"iface", required_argument, 0, 'i'},
488 {"ipv4", no_argument, 0, '4'},
489 {"ipv6", no_argument, 0, '6'},
490 {"sink", no_argument, 0, 's'},
491 {"tcp", no_argument, 0, 't'},
492 {"timeout", required_argument, 0, 'T'},
493 {"udp", no_argument, 0, 'u'},
494 {"verbose", no_argument, 0, 'v'},
495 {"rps", required_argument, 0, 'r'},
496 {0, 0, 0, 0}
497 };
498 bool have_toeplitz = false;
499 int index, c;
500
501 while ((c = getopt_long(argc, argv, "46C:d:i:k:r:stT:uv", long_options, &index)) != -1) {
502 switch (c) {
503 case '4':
504 cfg_family = AF_INET;
505 break;
506 case '6':
507 cfg_family = AF_INET6;
508 break;
509 case 'C':
510 parse_cpulist(optarg);
511 break;
512 case 'd':
513 cfg_dport = strtol(optarg, NULL, 0);
514 break;
515 case 'i':
516 cfg_ifname = optarg;
517 break;
518 case 'k':
519 parse_toeplitz_key(optarg, strlen(optarg),
520 toeplitz_key);
521 have_toeplitz = true;
522 break;
523 case 'r':
524 parse_rps_bitmap(optarg);
525 break;
526 case 's':
527 cfg_sink = true;
528 break;
529 case 't':
530 cfg_type = SOCK_STREAM;
531 break;
532 case 'T':
533 cfg_timeout_msec = strtol(optarg, NULL, 0);
534 break;
535 case 'u':
536 cfg_type = SOCK_DGRAM;
537 break;
538 case 'v':
539 cfg_verbose = true;
540 break;
541
542 default:
543 error(1, 0, "unknown option %c", optopt);
544 break;
545 }
546 }
547
548 if (!have_toeplitz)
549 error(1, 0, "Must supply rss key ('-k')");
550
551 num_cpus = get_nprocs();
552 if (num_cpus > RSS_MAX_CPUS)
553 error(1, 0, "increase RSS_MAX_CPUS");
554
555 if (cfg_num_queues && cfg_num_rps_cpus)
556 error(1, 0,
557 "Can't supply both RSS cpus ('-C') and RPS map ('-r')");
558 if (cfg_verbose) {
559 show_cpulist();
560 show_silos();
561 }
562}
563
564int main(int argc, char **argv)
565{
566 const int min_tests = 10;
567 int fd_sink = -1;
568
569 parse_opts(argc, argv);
570
571 if (cfg_sink)
572 fd_sink = setup_sink();
573
574 setup_rings();
575 process_rings();
576 cleanup_rings();
577
578 if (cfg_sink && close(fd_sink))
579 error(1, errno, "close sink");
580
581 if (frames_received - frames_nohash < min_tests)
582 error(1, 0, "too few frames for verification");
583
584 return frames_error;
585}
586