Loading...
1/*
2 * sysctl_net_ipv4.c: sysctl interface to net IPV4 subsystem.
3 *
4 * Begun April 1, 1996, Mike Shaver.
5 * Added /proc/sys/net/ipv4 directory entry (empty =) ). [MS]
6 */
7
8#include <linux/mm.h>
9#include <linux/module.h>
10#include <linux/sysctl.h>
11#include <linux/igmp.h>
12#include <linux/inetdevice.h>
13#include <linux/seqlock.h>
14#include <linux/init.h>
15#include <linux/slab.h>
16#include <linux/nsproxy.h>
17#include <linux/swap.h>
18#include <net/snmp.h>
19#include <net/icmp.h>
20#include <net/ip.h>
21#include <net/route.h>
22#include <net/tcp.h>
23#include <net/udp.h>
24#include <net/cipso_ipv4.h>
25#include <net/inet_frag.h>
26#include <net/ping.h>
27
28static int zero;
29static int one = 1;
30static int four = 4;
31static int thousand = 1000;
32static int gso_max_segs = GSO_MAX_SEGS;
33static int tcp_retr1_max = 255;
34static int ip_local_port_range_min[] = { 1, 1 };
35static int ip_local_port_range_max[] = { 65535, 65535 };
36static int tcp_adv_win_scale_min = -31;
37static int tcp_adv_win_scale_max = 31;
38static int ip_ttl_min = 1;
39static int ip_ttl_max = 255;
40static int tcp_syn_retries_min = 1;
41static int tcp_syn_retries_max = MAX_TCP_SYNCNT;
42static int ip_ping_group_range_min[] = { 0, 0 };
43static int ip_ping_group_range_max[] = { GID_T_MAX, GID_T_MAX };
44
45/* Update system visible IP port range */
46static void set_local_port_range(struct net *net, int range[2])
47{
48 bool same_parity = !((range[0] ^ range[1]) & 1);
49
50 write_seqlock_bh(&net->ipv4.ip_local_ports.lock);
51 if (same_parity && !net->ipv4.ip_local_ports.warned) {
52 net->ipv4.ip_local_ports.warned = true;
53 pr_err_ratelimited("ip_local_port_range: prefer different parity for start/end values.\n");
54 }
55 net->ipv4.ip_local_ports.range[0] = range[0];
56 net->ipv4.ip_local_ports.range[1] = range[1];
57 write_sequnlock_bh(&net->ipv4.ip_local_ports.lock);
58}
59
60/* Validate changes from /proc interface. */
61static int ipv4_local_port_range(struct ctl_table *table, int write,
62 void __user *buffer,
63 size_t *lenp, loff_t *ppos)
64{
65 struct net *net =
66 container_of(table->data, struct net, ipv4.ip_local_ports.range);
67 int ret;
68 int range[2];
69 struct ctl_table tmp = {
70 .data = &range,
71 .maxlen = sizeof(range),
72 .mode = table->mode,
73 .extra1 = &ip_local_port_range_min,
74 .extra2 = &ip_local_port_range_max,
75 };
76
77 inet_get_local_port_range(net, &range[0], &range[1]);
78
79 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
80
81 if (write && ret == 0) {
82 if (range[1] < range[0])
83 ret = -EINVAL;
84 else
85 set_local_port_range(net, range);
86 }
87
88 return ret;
89}
90
91
92static void inet_get_ping_group_range_table(struct ctl_table *table, kgid_t *low, kgid_t *high)
93{
94 kgid_t *data = table->data;
95 struct net *net =
96 container_of(table->data, struct net, ipv4.ping_group_range.range);
97 unsigned int seq;
98 do {
99 seq = read_seqbegin(&net->ipv4.ip_local_ports.lock);
100
101 *low = data[0];
102 *high = data[1];
103 } while (read_seqretry(&net->ipv4.ip_local_ports.lock, seq));
104}
105
106/* Update system visible IP port range */
107static void set_ping_group_range(struct ctl_table *table, kgid_t low, kgid_t high)
108{
109 kgid_t *data = table->data;
110 struct net *net =
111 container_of(table->data, struct net, ipv4.ping_group_range.range);
112 write_seqlock(&net->ipv4.ip_local_ports.lock);
113 data[0] = low;
114 data[1] = high;
115 write_sequnlock(&net->ipv4.ip_local_ports.lock);
116}
117
118/* Validate changes from /proc interface. */
119static int ipv4_ping_group_range(struct ctl_table *table, int write,
120 void __user *buffer,
121 size_t *lenp, loff_t *ppos)
122{
123 struct user_namespace *user_ns = current_user_ns();
124 int ret;
125 gid_t urange[2];
126 kgid_t low, high;
127 struct ctl_table tmp = {
128 .data = &urange,
129 .maxlen = sizeof(urange),
130 .mode = table->mode,
131 .extra1 = &ip_ping_group_range_min,
132 .extra2 = &ip_ping_group_range_max,
133 };
134
135 inet_get_ping_group_range_table(table, &low, &high);
136 urange[0] = from_kgid_munged(user_ns, low);
137 urange[1] = from_kgid_munged(user_ns, high);
138 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
139
140 if (write && ret == 0) {
141 low = make_kgid(user_ns, urange[0]);
142 high = make_kgid(user_ns, urange[1]);
143 if (!gid_valid(low) || !gid_valid(high) ||
144 (urange[1] < urange[0]) || gid_lt(high, low)) {
145 low = make_kgid(&init_user_ns, 1);
146 high = make_kgid(&init_user_ns, 0);
147 }
148 set_ping_group_range(table, low, high);
149 }
150
151 return ret;
152}
153
154static int proc_tcp_congestion_control(struct ctl_table *ctl, int write,
155 void __user *buffer, size_t *lenp, loff_t *ppos)
156{
157 char val[TCP_CA_NAME_MAX];
158 struct ctl_table tbl = {
159 .data = val,
160 .maxlen = TCP_CA_NAME_MAX,
161 };
162 int ret;
163
164 tcp_get_default_congestion_control(val);
165
166 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
167 if (write && ret == 0)
168 ret = tcp_set_default_congestion_control(val);
169 return ret;
170}
171
172static int proc_tcp_available_congestion_control(struct ctl_table *ctl,
173 int write,
174 void __user *buffer, size_t *lenp,
175 loff_t *ppos)
176{
177 struct ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX, };
178 int ret;
179
180 tbl.data = kmalloc(tbl.maxlen, GFP_USER);
181 if (!tbl.data)
182 return -ENOMEM;
183 tcp_get_available_congestion_control(tbl.data, TCP_CA_BUF_MAX);
184 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
185 kfree(tbl.data);
186 return ret;
187}
188
189static int proc_allowed_congestion_control(struct ctl_table *ctl,
190 int write,
191 void __user *buffer, size_t *lenp,
192 loff_t *ppos)
193{
194 struct ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX };
195 int ret;
196
197 tbl.data = kmalloc(tbl.maxlen, GFP_USER);
198 if (!tbl.data)
199 return -ENOMEM;
200
201 tcp_get_allowed_congestion_control(tbl.data, tbl.maxlen);
202 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
203 if (write && ret == 0)
204 ret = tcp_set_allowed_congestion_control(tbl.data);
205 kfree(tbl.data);
206 return ret;
207}
208
209static int proc_tcp_fastopen_key(struct ctl_table *ctl, int write,
210 void __user *buffer, size_t *lenp,
211 loff_t *ppos)
212{
213 struct ctl_table tbl = { .maxlen = (TCP_FASTOPEN_KEY_LENGTH * 2 + 10) };
214 struct tcp_fastopen_context *ctxt;
215 int ret;
216 u32 user_key[4]; /* 16 bytes, matching TCP_FASTOPEN_KEY_LENGTH */
217
218 tbl.data = kmalloc(tbl.maxlen, GFP_KERNEL);
219 if (!tbl.data)
220 return -ENOMEM;
221
222 rcu_read_lock();
223 ctxt = rcu_dereference(tcp_fastopen_ctx);
224 if (ctxt)
225 memcpy(user_key, ctxt->key, TCP_FASTOPEN_KEY_LENGTH);
226 else
227 memset(user_key, 0, sizeof(user_key));
228 rcu_read_unlock();
229
230 snprintf(tbl.data, tbl.maxlen, "%08x-%08x-%08x-%08x",
231 user_key[0], user_key[1], user_key[2], user_key[3]);
232 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
233
234 if (write && ret == 0) {
235 if (sscanf(tbl.data, "%x-%x-%x-%x", user_key, user_key + 1,
236 user_key + 2, user_key + 3) != 4) {
237 ret = -EINVAL;
238 goto bad_key;
239 }
240 /* Generate a dummy secret but don't publish it. This
241 * is needed so we don't regenerate a new key on the
242 * first invocation of tcp_fastopen_cookie_gen
243 */
244 tcp_fastopen_init_key_once(false);
245 tcp_fastopen_reset_cipher(user_key, TCP_FASTOPEN_KEY_LENGTH);
246 }
247
248bad_key:
249 pr_debug("proc FO key set 0x%x-%x-%x-%x <- 0x%s: %u\n",
250 user_key[0], user_key[1], user_key[2], user_key[3],
251 (char *)tbl.data, ret);
252 kfree(tbl.data);
253 return ret;
254}
255
256static struct ctl_table ipv4_table[] = {
257 {
258 .procname = "tcp_timestamps",
259 .data = &sysctl_tcp_timestamps,
260 .maxlen = sizeof(int),
261 .mode = 0644,
262 .proc_handler = proc_dointvec
263 },
264 {
265 .procname = "tcp_window_scaling",
266 .data = &sysctl_tcp_window_scaling,
267 .maxlen = sizeof(int),
268 .mode = 0644,
269 .proc_handler = proc_dointvec
270 },
271 {
272 .procname = "tcp_sack",
273 .data = &sysctl_tcp_sack,
274 .maxlen = sizeof(int),
275 .mode = 0644,
276 .proc_handler = proc_dointvec
277 },
278 {
279 .procname = "tcp_retrans_collapse",
280 .data = &sysctl_tcp_retrans_collapse,
281 .maxlen = sizeof(int),
282 .mode = 0644,
283 .proc_handler = proc_dointvec
284 },
285 {
286 .procname = "tcp_max_orphans",
287 .data = &sysctl_tcp_max_orphans,
288 .maxlen = sizeof(int),
289 .mode = 0644,
290 .proc_handler = proc_dointvec
291 },
292 {
293 .procname = "tcp_max_tw_buckets",
294 .data = &tcp_death_row.sysctl_max_tw_buckets,
295 .maxlen = sizeof(int),
296 .mode = 0644,
297 .proc_handler = proc_dointvec
298 },
299 {
300 .procname = "tcp_fastopen",
301 .data = &sysctl_tcp_fastopen,
302 .maxlen = sizeof(int),
303 .mode = 0644,
304 .proc_handler = proc_dointvec,
305 },
306 {
307 .procname = "tcp_fastopen_key",
308 .mode = 0600,
309 .maxlen = ((TCP_FASTOPEN_KEY_LENGTH * 2) + 10),
310 .proc_handler = proc_tcp_fastopen_key,
311 },
312 {
313 .procname = "tcp_tw_recycle",
314 .data = &tcp_death_row.sysctl_tw_recycle,
315 .maxlen = sizeof(int),
316 .mode = 0644,
317 .proc_handler = proc_dointvec
318 },
319 {
320 .procname = "tcp_abort_on_overflow",
321 .data = &sysctl_tcp_abort_on_overflow,
322 .maxlen = sizeof(int),
323 .mode = 0644,
324 .proc_handler = proc_dointvec
325 },
326 {
327 .procname = "tcp_stdurg",
328 .data = &sysctl_tcp_stdurg,
329 .maxlen = sizeof(int),
330 .mode = 0644,
331 .proc_handler = proc_dointvec
332 },
333 {
334 .procname = "tcp_rfc1337",
335 .data = &sysctl_tcp_rfc1337,
336 .maxlen = sizeof(int),
337 .mode = 0644,
338 .proc_handler = proc_dointvec
339 },
340 {
341 .procname = "tcp_max_syn_backlog",
342 .data = &sysctl_max_syn_backlog,
343 .maxlen = sizeof(int),
344 .mode = 0644,
345 .proc_handler = proc_dointvec
346 },
347 {
348 .procname = "inet_peer_threshold",
349 .data = &inet_peer_threshold,
350 .maxlen = sizeof(int),
351 .mode = 0644,
352 .proc_handler = proc_dointvec
353 },
354 {
355 .procname = "inet_peer_minttl",
356 .data = &inet_peer_minttl,
357 .maxlen = sizeof(int),
358 .mode = 0644,
359 .proc_handler = proc_dointvec_jiffies,
360 },
361 {
362 .procname = "inet_peer_maxttl",
363 .data = &inet_peer_maxttl,
364 .maxlen = sizeof(int),
365 .mode = 0644,
366 .proc_handler = proc_dointvec_jiffies,
367 },
368 {
369 .procname = "tcp_fack",
370 .data = &sysctl_tcp_fack,
371 .maxlen = sizeof(int),
372 .mode = 0644,
373 .proc_handler = proc_dointvec
374 },
375 {
376 .procname = "tcp_recovery",
377 .data = &sysctl_tcp_recovery,
378 .maxlen = sizeof(int),
379 .mode = 0644,
380 .proc_handler = proc_dointvec,
381 },
382 {
383 .procname = "tcp_max_reordering",
384 .data = &sysctl_tcp_max_reordering,
385 .maxlen = sizeof(int),
386 .mode = 0644,
387 .proc_handler = proc_dointvec
388 },
389 {
390 .procname = "tcp_dsack",
391 .data = &sysctl_tcp_dsack,
392 .maxlen = sizeof(int),
393 .mode = 0644,
394 .proc_handler = proc_dointvec
395 },
396 {
397 .procname = "tcp_mem",
398 .maxlen = sizeof(sysctl_tcp_mem),
399 .data = &sysctl_tcp_mem,
400 .mode = 0644,
401 .proc_handler = proc_doulongvec_minmax,
402 },
403 {
404 .procname = "tcp_wmem",
405 .data = &sysctl_tcp_wmem,
406 .maxlen = sizeof(sysctl_tcp_wmem),
407 .mode = 0644,
408 .proc_handler = proc_dointvec_minmax,
409 .extra1 = &one,
410 },
411 {
412 .procname = "tcp_rmem",
413 .data = &sysctl_tcp_rmem,
414 .maxlen = sizeof(sysctl_tcp_rmem),
415 .mode = 0644,
416 .proc_handler = proc_dointvec_minmax,
417 .extra1 = &one,
418 },
419 {
420 .procname = "tcp_app_win",
421 .data = &sysctl_tcp_app_win,
422 .maxlen = sizeof(int),
423 .mode = 0644,
424 .proc_handler = proc_dointvec
425 },
426 {
427 .procname = "tcp_adv_win_scale",
428 .data = &sysctl_tcp_adv_win_scale,
429 .maxlen = sizeof(int),
430 .mode = 0644,
431 .proc_handler = proc_dointvec_minmax,
432 .extra1 = &tcp_adv_win_scale_min,
433 .extra2 = &tcp_adv_win_scale_max,
434 },
435 {
436 .procname = "tcp_tw_reuse",
437 .data = &sysctl_tcp_tw_reuse,
438 .maxlen = sizeof(int),
439 .mode = 0644,
440 .proc_handler = proc_dointvec
441 },
442 {
443 .procname = "tcp_frto",
444 .data = &sysctl_tcp_frto,
445 .maxlen = sizeof(int),
446 .mode = 0644,
447 .proc_handler = proc_dointvec
448 },
449 {
450 .procname = "tcp_min_rtt_wlen",
451 .data = &sysctl_tcp_min_rtt_wlen,
452 .maxlen = sizeof(int),
453 .mode = 0644,
454 .proc_handler = proc_dointvec
455 },
456 {
457 .procname = "tcp_low_latency",
458 .data = &sysctl_tcp_low_latency,
459 .maxlen = sizeof(int),
460 .mode = 0644,
461 .proc_handler = proc_dointvec
462 },
463 {
464 .procname = "tcp_no_metrics_save",
465 .data = &sysctl_tcp_nometrics_save,
466 .maxlen = sizeof(int),
467 .mode = 0644,
468 .proc_handler = proc_dointvec,
469 },
470 {
471 .procname = "tcp_moderate_rcvbuf",
472 .data = &sysctl_tcp_moderate_rcvbuf,
473 .maxlen = sizeof(int),
474 .mode = 0644,
475 .proc_handler = proc_dointvec,
476 },
477 {
478 .procname = "tcp_tso_win_divisor",
479 .data = &sysctl_tcp_tso_win_divisor,
480 .maxlen = sizeof(int),
481 .mode = 0644,
482 .proc_handler = proc_dointvec,
483 },
484 {
485 .procname = "tcp_congestion_control",
486 .mode = 0644,
487 .maxlen = TCP_CA_NAME_MAX,
488 .proc_handler = proc_tcp_congestion_control,
489 },
490 {
491 .procname = "tcp_workaround_signed_windows",
492 .data = &sysctl_tcp_workaround_signed_windows,
493 .maxlen = sizeof(int),
494 .mode = 0644,
495 .proc_handler = proc_dointvec
496 },
497 {
498 .procname = "tcp_limit_output_bytes",
499 .data = &sysctl_tcp_limit_output_bytes,
500 .maxlen = sizeof(int),
501 .mode = 0644,
502 .proc_handler = proc_dointvec
503 },
504 {
505 .procname = "tcp_challenge_ack_limit",
506 .data = &sysctl_tcp_challenge_ack_limit,
507 .maxlen = sizeof(int),
508 .mode = 0644,
509 .proc_handler = proc_dointvec
510 },
511 {
512 .procname = "tcp_slow_start_after_idle",
513 .data = &sysctl_tcp_slow_start_after_idle,
514 .maxlen = sizeof(int),
515 .mode = 0644,
516 .proc_handler = proc_dointvec
517 },
518#ifdef CONFIG_NETLABEL
519 {
520 .procname = "cipso_cache_enable",
521 .data = &cipso_v4_cache_enabled,
522 .maxlen = sizeof(int),
523 .mode = 0644,
524 .proc_handler = proc_dointvec,
525 },
526 {
527 .procname = "cipso_cache_bucket_size",
528 .data = &cipso_v4_cache_bucketsize,
529 .maxlen = sizeof(int),
530 .mode = 0644,
531 .proc_handler = proc_dointvec,
532 },
533 {
534 .procname = "cipso_rbm_optfmt",
535 .data = &cipso_v4_rbm_optfmt,
536 .maxlen = sizeof(int),
537 .mode = 0644,
538 .proc_handler = proc_dointvec,
539 },
540 {
541 .procname = "cipso_rbm_strictvalid",
542 .data = &cipso_v4_rbm_strictvalid,
543 .maxlen = sizeof(int),
544 .mode = 0644,
545 .proc_handler = proc_dointvec,
546 },
547#endif /* CONFIG_NETLABEL */
548 {
549 .procname = "tcp_available_congestion_control",
550 .maxlen = TCP_CA_BUF_MAX,
551 .mode = 0444,
552 .proc_handler = proc_tcp_available_congestion_control,
553 },
554 {
555 .procname = "tcp_allowed_congestion_control",
556 .maxlen = TCP_CA_BUF_MAX,
557 .mode = 0644,
558 .proc_handler = proc_allowed_congestion_control,
559 },
560 {
561 .procname = "tcp_thin_linear_timeouts",
562 .data = &sysctl_tcp_thin_linear_timeouts,
563 .maxlen = sizeof(int),
564 .mode = 0644,
565 .proc_handler = proc_dointvec
566 },
567 {
568 .procname = "tcp_thin_dupack",
569 .data = &sysctl_tcp_thin_dupack,
570 .maxlen = sizeof(int),
571 .mode = 0644,
572 .proc_handler = proc_dointvec
573 },
574 {
575 .procname = "tcp_early_retrans",
576 .data = &sysctl_tcp_early_retrans,
577 .maxlen = sizeof(int),
578 .mode = 0644,
579 .proc_handler = proc_dointvec_minmax,
580 .extra1 = &zero,
581 .extra2 = &four,
582 },
583 {
584 .procname = "tcp_min_tso_segs",
585 .data = &sysctl_tcp_min_tso_segs,
586 .maxlen = sizeof(int),
587 .mode = 0644,
588 .proc_handler = proc_dointvec_minmax,
589 .extra1 = &one,
590 .extra2 = &gso_max_segs,
591 },
592 {
593 .procname = "tcp_pacing_ss_ratio",
594 .data = &sysctl_tcp_pacing_ss_ratio,
595 .maxlen = sizeof(int),
596 .mode = 0644,
597 .proc_handler = proc_dointvec_minmax,
598 .extra1 = &zero,
599 .extra2 = &thousand,
600 },
601 {
602 .procname = "tcp_pacing_ca_ratio",
603 .data = &sysctl_tcp_pacing_ca_ratio,
604 .maxlen = sizeof(int),
605 .mode = 0644,
606 .proc_handler = proc_dointvec_minmax,
607 .extra1 = &zero,
608 .extra2 = &thousand,
609 },
610 {
611 .procname = "tcp_autocorking",
612 .data = &sysctl_tcp_autocorking,
613 .maxlen = sizeof(int),
614 .mode = 0644,
615 .proc_handler = proc_dointvec_minmax,
616 .extra1 = &zero,
617 .extra2 = &one,
618 },
619 {
620 .procname = "tcp_invalid_ratelimit",
621 .data = &sysctl_tcp_invalid_ratelimit,
622 .maxlen = sizeof(int),
623 .mode = 0644,
624 .proc_handler = proc_dointvec_ms_jiffies,
625 },
626 {
627 .procname = "icmp_msgs_per_sec",
628 .data = &sysctl_icmp_msgs_per_sec,
629 .maxlen = sizeof(int),
630 .mode = 0644,
631 .proc_handler = proc_dointvec_minmax,
632 .extra1 = &zero,
633 },
634 {
635 .procname = "icmp_msgs_burst",
636 .data = &sysctl_icmp_msgs_burst,
637 .maxlen = sizeof(int),
638 .mode = 0644,
639 .proc_handler = proc_dointvec_minmax,
640 .extra1 = &zero,
641 },
642 {
643 .procname = "udp_mem",
644 .data = &sysctl_udp_mem,
645 .maxlen = sizeof(sysctl_udp_mem),
646 .mode = 0644,
647 .proc_handler = proc_doulongvec_minmax,
648 },
649 {
650 .procname = "udp_rmem_min",
651 .data = &sysctl_udp_rmem_min,
652 .maxlen = sizeof(sysctl_udp_rmem_min),
653 .mode = 0644,
654 .proc_handler = proc_dointvec_minmax,
655 .extra1 = &one
656 },
657 {
658 .procname = "udp_wmem_min",
659 .data = &sysctl_udp_wmem_min,
660 .maxlen = sizeof(sysctl_udp_wmem_min),
661 .mode = 0644,
662 .proc_handler = proc_dointvec_minmax,
663 .extra1 = &one
664 },
665 { }
666};
667
668static struct ctl_table ipv4_net_table[] = {
669 {
670 .procname = "icmp_echo_ignore_all",
671 .data = &init_net.ipv4.sysctl_icmp_echo_ignore_all,
672 .maxlen = sizeof(int),
673 .mode = 0644,
674 .proc_handler = proc_dointvec
675 },
676 {
677 .procname = "icmp_echo_ignore_broadcasts",
678 .data = &init_net.ipv4.sysctl_icmp_echo_ignore_broadcasts,
679 .maxlen = sizeof(int),
680 .mode = 0644,
681 .proc_handler = proc_dointvec
682 },
683 {
684 .procname = "icmp_ignore_bogus_error_responses",
685 .data = &init_net.ipv4.sysctl_icmp_ignore_bogus_error_responses,
686 .maxlen = sizeof(int),
687 .mode = 0644,
688 .proc_handler = proc_dointvec
689 },
690 {
691 .procname = "icmp_errors_use_inbound_ifaddr",
692 .data = &init_net.ipv4.sysctl_icmp_errors_use_inbound_ifaddr,
693 .maxlen = sizeof(int),
694 .mode = 0644,
695 .proc_handler = proc_dointvec
696 },
697 {
698 .procname = "icmp_ratelimit",
699 .data = &init_net.ipv4.sysctl_icmp_ratelimit,
700 .maxlen = sizeof(int),
701 .mode = 0644,
702 .proc_handler = proc_dointvec_ms_jiffies,
703 },
704 {
705 .procname = "icmp_ratemask",
706 .data = &init_net.ipv4.sysctl_icmp_ratemask,
707 .maxlen = sizeof(int),
708 .mode = 0644,
709 .proc_handler = proc_dointvec
710 },
711 {
712 .procname = "ping_group_range",
713 .data = &init_net.ipv4.ping_group_range.range,
714 .maxlen = sizeof(gid_t)*2,
715 .mode = 0644,
716 .proc_handler = ipv4_ping_group_range,
717 },
718 {
719 .procname = "tcp_ecn",
720 .data = &init_net.ipv4.sysctl_tcp_ecn,
721 .maxlen = sizeof(int),
722 .mode = 0644,
723 .proc_handler = proc_dointvec
724 },
725 {
726 .procname = "tcp_ecn_fallback",
727 .data = &init_net.ipv4.sysctl_tcp_ecn_fallback,
728 .maxlen = sizeof(int),
729 .mode = 0644,
730 .proc_handler = proc_dointvec
731 },
732 {
733 .procname = "ip_dynaddr",
734 .data = &init_net.ipv4.sysctl_ip_dynaddr,
735 .maxlen = sizeof(int),
736 .mode = 0644,
737 .proc_handler = proc_dointvec
738 },
739 {
740 .procname = "ip_early_demux",
741 .data = &init_net.ipv4.sysctl_ip_early_demux,
742 .maxlen = sizeof(int),
743 .mode = 0644,
744 .proc_handler = proc_dointvec
745 },
746 {
747 .procname = "ip_default_ttl",
748 .data = &init_net.ipv4.sysctl_ip_default_ttl,
749 .maxlen = sizeof(int),
750 .mode = 0644,
751 .proc_handler = proc_dointvec_minmax,
752 .extra1 = &ip_ttl_min,
753 .extra2 = &ip_ttl_max,
754 },
755 {
756 .procname = "ip_local_port_range",
757 .maxlen = sizeof(init_net.ipv4.ip_local_ports.range),
758 .data = &init_net.ipv4.ip_local_ports.range,
759 .mode = 0644,
760 .proc_handler = ipv4_local_port_range,
761 },
762 {
763 .procname = "ip_local_reserved_ports",
764 .data = &init_net.ipv4.sysctl_local_reserved_ports,
765 .maxlen = 65536,
766 .mode = 0644,
767 .proc_handler = proc_do_large_bitmap,
768 },
769 {
770 .procname = "ip_no_pmtu_disc",
771 .data = &init_net.ipv4.sysctl_ip_no_pmtu_disc,
772 .maxlen = sizeof(int),
773 .mode = 0644,
774 .proc_handler = proc_dointvec
775 },
776 {
777 .procname = "ip_forward_use_pmtu",
778 .data = &init_net.ipv4.sysctl_ip_fwd_use_pmtu,
779 .maxlen = sizeof(int),
780 .mode = 0644,
781 .proc_handler = proc_dointvec,
782 },
783 {
784 .procname = "ip_nonlocal_bind",
785 .data = &init_net.ipv4.sysctl_ip_nonlocal_bind,
786 .maxlen = sizeof(int),
787 .mode = 0644,
788 .proc_handler = proc_dointvec
789 },
790 {
791 .procname = "fwmark_reflect",
792 .data = &init_net.ipv4.sysctl_fwmark_reflect,
793 .maxlen = sizeof(int),
794 .mode = 0644,
795 .proc_handler = proc_dointvec,
796 },
797 {
798 .procname = "tcp_fwmark_accept",
799 .data = &init_net.ipv4.sysctl_tcp_fwmark_accept,
800 .maxlen = sizeof(int),
801 .mode = 0644,
802 .proc_handler = proc_dointvec,
803 },
804#ifdef CONFIG_NET_L3_MASTER_DEV
805 {
806 .procname = "tcp_l3mdev_accept",
807 .data = &init_net.ipv4.sysctl_tcp_l3mdev_accept,
808 .maxlen = sizeof(int),
809 .mode = 0644,
810 .proc_handler = proc_dointvec_minmax,
811 .extra1 = &zero,
812 .extra2 = &one,
813 },
814#endif
815 {
816 .procname = "tcp_mtu_probing",
817 .data = &init_net.ipv4.sysctl_tcp_mtu_probing,
818 .maxlen = sizeof(int),
819 .mode = 0644,
820 .proc_handler = proc_dointvec,
821 },
822 {
823 .procname = "tcp_base_mss",
824 .data = &init_net.ipv4.sysctl_tcp_base_mss,
825 .maxlen = sizeof(int),
826 .mode = 0644,
827 .proc_handler = proc_dointvec,
828 },
829 {
830 .procname = "tcp_probe_threshold",
831 .data = &init_net.ipv4.sysctl_tcp_probe_threshold,
832 .maxlen = sizeof(int),
833 .mode = 0644,
834 .proc_handler = proc_dointvec,
835 },
836 {
837 .procname = "tcp_probe_interval",
838 .data = &init_net.ipv4.sysctl_tcp_probe_interval,
839 .maxlen = sizeof(int),
840 .mode = 0644,
841 .proc_handler = proc_dointvec,
842 },
843 {
844 .procname = "igmp_link_local_mcast_reports",
845 .data = &init_net.ipv4.sysctl_igmp_llm_reports,
846 .maxlen = sizeof(int),
847 .mode = 0644,
848 .proc_handler = proc_dointvec
849 },
850 {
851 .procname = "igmp_max_memberships",
852 .data = &init_net.ipv4.sysctl_igmp_max_memberships,
853 .maxlen = sizeof(int),
854 .mode = 0644,
855 .proc_handler = proc_dointvec
856 },
857 {
858 .procname = "igmp_max_msf",
859 .data = &init_net.ipv4.sysctl_igmp_max_msf,
860 .maxlen = sizeof(int),
861 .mode = 0644,
862 .proc_handler = proc_dointvec
863 },
864#ifdef CONFIG_IP_MULTICAST
865 {
866 .procname = "igmp_qrv",
867 .data = &init_net.ipv4.sysctl_igmp_qrv,
868 .maxlen = sizeof(int),
869 .mode = 0644,
870 .proc_handler = proc_dointvec_minmax,
871 .extra1 = &one
872 },
873#endif
874 {
875 .procname = "tcp_keepalive_time",
876 .data = &init_net.ipv4.sysctl_tcp_keepalive_time,
877 .maxlen = sizeof(int),
878 .mode = 0644,
879 .proc_handler = proc_dointvec_jiffies,
880 },
881 {
882 .procname = "tcp_keepalive_probes",
883 .data = &init_net.ipv4.sysctl_tcp_keepalive_probes,
884 .maxlen = sizeof(int),
885 .mode = 0644,
886 .proc_handler = proc_dointvec
887 },
888 {
889 .procname = "tcp_keepalive_intvl",
890 .data = &init_net.ipv4.sysctl_tcp_keepalive_intvl,
891 .maxlen = sizeof(int),
892 .mode = 0644,
893 .proc_handler = proc_dointvec_jiffies,
894 },
895 {
896 .procname = "tcp_syn_retries",
897 .data = &init_net.ipv4.sysctl_tcp_syn_retries,
898 .maxlen = sizeof(int),
899 .mode = 0644,
900 .proc_handler = proc_dointvec_minmax,
901 .extra1 = &tcp_syn_retries_min,
902 .extra2 = &tcp_syn_retries_max
903 },
904 {
905 .procname = "tcp_synack_retries",
906 .data = &init_net.ipv4.sysctl_tcp_synack_retries,
907 .maxlen = sizeof(int),
908 .mode = 0644,
909 .proc_handler = proc_dointvec
910 },
911#ifdef CONFIG_SYN_COOKIES
912 {
913 .procname = "tcp_syncookies",
914 .data = &init_net.ipv4.sysctl_tcp_syncookies,
915 .maxlen = sizeof(int),
916 .mode = 0644,
917 .proc_handler = proc_dointvec
918 },
919#endif
920 {
921 .procname = "tcp_reordering",
922 .data = &init_net.ipv4.sysctl_tcp_reordering,
923 .maxlen = sizeof(int),
924 .mode = 0644,
925 .proc_handler = proc_dointvec
926 },
927 {
928 .procname = "tcp_retries1",
929 .data = &init_net.ipv4.sysctl_tcp_retries1,
930 .maxlen = sizeof(int),
931 .mode = 0644,
932 .proc_handler = proc_dointvec_minmax,
933 .extra2 = &tcp_retr1_max
934 },
935 {
936 .procname = "tcp_retries2",
937 .data = &init_net.ipv4.sysctl_tcp_retries2,
938 .maxlen = sizeof(int),
939 .mode = 0644,
940 .proc_handler = proc_dointvec
941 },
942 {
943 .procname = "tcp_orphan_retries",
944 .data = &init_net.ipv4.sysctl_tcp_orphan_retries,
945 .maxlen = sizeof(int),
946 .mode = 0644,
947 .proc_handler = proc_dointvec
948 },
949 {
950 .procname = "tcp_fin_timeout",
951 .data = &init_net.ipv4.sysctl_tcp_fin_timeout,
952 .maxlen = sizeof(int),
953 .mode = 0644,
954 .proc_handler = proc_dointvec_jiffies,
955 },
956 {
957 .procname = "tcp_notsent_lowat",
958 .data = &init_net.ipv4.sysctl_tcp_notsent_lowat,
959 .maxlen = sizeof(unsigned int),
960 .mode = 0644,
961 .proc_handler = proc_dointvec,
962 },
963 { }
964};
965
966static __net_init int ipv4_sysctl_init_net(struct net *net)
967{
968 struct ctl_table *table;
969
970 table = ipv4_net_table;
971 if (!net_eq(net, &init_net)) {
972 int i;
973
974 table = kmemdup(table, sizeof(ipv4_net_table), GFP_KERNEL);
975 if (!table)
976 goto err_alloc;
977
978 /* Update the variables to point into the current struct net */
979 for (i = 0; i < ARRAY_SIZE(ipv4_net_table) - 1; i++)
980 table[i].data += (void *)net - (void *)&init_net;
981 }
982
983 net->ipv4.ipv4_hdr = register_net_sysctl(net, "net/ipv4", table);
984 if (!net->ipv4.ipv4_hdr)
985 goto err_reg;
986
987 net->ipv4.sysctl_local_reserved_ports = kzalloc(65536 / 8, GFP_KERNEL);
988 if (!net->ipv4.sysctl_local_reserved_ports)
989 goto err_ports;
990
991 net->ipv4.sysctl_ip_default_ttl = IPDEFTTL;
992 net->ipv4.sysctl_ip_dynaddr = 0;
993 net->ipv4.sysctl_ip_early_demux = 1;
994
995 return 0;
996
997err_ports:
998 unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
999err_reg:
1000 if (!net_eq(net, &init_net))
1001 kfree(table);
1002err_alloc:
1003 return -ENOMEM;
1004}
1005
1006static __net_exit void ipv4_sysctl_exit_net(struct net *net)
1007{
1008 struct ctl_table *table;
1009
1010 kfree(net->ipv4.sysctl_local_reserved_ports);
1011 table = net->ipv4.ipv4_hdr->ctl_table_arg;
1012 unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
1013 kfree(table);
1014}
1015
1016static __net_initdata struct pernet_operations ipv4_sysctl_ops = {
1017 .init = ipv4_sysctl_init_net,
1018 .exit = ipv4_sysctl_exit_net,
1019};
1020
1021static __init int sysctl_ipv4_init(void)
1022{
1023 struct ctl_table_header *hdr;
1024
1025 hdr = register_net_sysctl(&init_net, "net/ipv4", ipv4_table);
1026 if (!hdr)
1027 return -ENOMEM;
1028
1029 if (register_pernet_subsys(&ipv4_sysctl_ops)) {
1030 unregister_net_sysctl_table(hdr);
1031 return -ENOMEM;
1032 }
1033
1034 return 0;
1035}
1036
1037__initcall(sysctl_ipv4_init);
1// SPDX-License-Identifier: GPL-2.0
2/*
3 * sysctl_net_ipv4.c: sysctl interface to net IPV4 subsystem.
4 *
5 * Begun April 1, 1996, Mike Shaver.
6 * Added /proc/sys/net/ipv4 directory entry (empty =) ). [MS]
7 */
8
9#include <linux/sysctl.h>
10#include <linux/seqlock.h>
11#include <linux/init.h>
12#include <linux/slab.h>
13#include <net/icmp.h>
14#include <net/ip.h>
15#include <net/ip_fib.h>
16#include <net/tcp.h>
17#include <net/udp.h>
18#include <net/cipso_ipv4.h>
19#include <net/ping.h>
20#include <net/protocol.h>
21#include <net/netevent.h>
22
23static int tcp_retr1_max = 255;
24static int ip_local_port_range_min[] = { 1, 1 };
25static int ip_local_port_range_max[] = { 65535, 65535 };
26static int tcp_adv_win_scale_min = -31;
27static int tcp_adv_win_scale_max = 31;
28static int tcp_min_snd_mss_min = TCP_MIN_SND_MSS;
29static int tcp_min_snd_mss_max = 65535;
30static int ip_privileged_port_min;
31static int ip_privileged_port_max = 65535;
32static int ip_ttl_min = 1;
33static int ip_ttl_max = 255;
34static int tcp_syn_retries_min = 1;
35static int tcp_syn_retries_max = MAX_TCP_SYNCNT;
36static int ip_ping_group_range_min[] = { 0, 0 };
37static int ip_ping_group_range_max[] = { GID_T_MAX, GID_T_MAX };
38static u32 u32_max_div_HZ = UINT_MAX / HZ;
39static int one_day_secs = 24 * 3600;
40static u32 fib_multipath_hash_fields_all_mask __maybe_unused =
41 FIB_MULTIPATH_HASH_FIELD_ALL_MASK;
42static unsigned int tcp_child_ehash_entries_max = 16 * 1024 * 1024;
43static unsigned int udp_child_hash_entries_max = UDP_HTABLE_SIZE_MAX;
44static int tcp_plb_max_rounds = 31;
45static int tcp_plb_max_cong_thresh = 256;
46
47/* obsolete */
48static int sysctl_tcp_low_latency __read_mostly;
49
50/* Update system visible IP port range */
51static void set_local_port_range(struct net *net, int range[2])
52{
53 bool same_parity = !((range[0] ^ range[1]) & 1);
54
55 write_seqlock_bh(&net->ipv4.ip_local_ports.lock);
56 if (same_parity && !net->ipv4.ip_local_ports.warned) {
57 net->ipv4.ip_local_ports.warned = true;
58 pr_err_ratelimited("ip_local_port_range: prefer different parity for start/end values.\n");
59 }
60 net->ipv4.ip_local_ports.range[0] = range[0];
61 net->ipv4.ip_local_ports.range[1] = range[1];
62 write_sequnlock_bh(&net->ipv4.ip_local_ports.lock);
63}
64
65/* Validate changes from /proc interface. */
66static int ipv4_local_port_range(struct ctl_table *table, int write,
67 void *buffer, size_t *lenp, loff_t *ppos)
68{
69 struct net *net =
70 container_of(table->data, struct net, ipv4.ip_local_ports.range);
71 int ret;
72 int range[2];
73 struct ctl_table tmp = {
74 .data = &range,
75 .maxlen = sizeof(range),
76 .mode = table->mode,
77 .extra1 = &ip_local_port_range_min,
78 .extra2 = &ip_local_port_range_max,
79 };
80
81 inet_get_local_port_range(net, &range[0], &range[1]);
82
83 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
84
85 if (write && ret == 0) {
86 /* Ensure that the upper limit is not smaller than the lower,
87 * and that the lower does not encroach upon the privileged
88 * port limit.
89 */
90 if ((range[1] < range[0]) ||
91 (range[0] < READ_ONCE(net->ipv4.sysctl_ip_prot_sock)))
92 ret = -EINVAL;
93 else
94 set_local_port_range(net, range);
95 }
96
97 return ret;
98}
99
100/* Validate changes from /proc interface. */
101static int ipv4_privileged_ports(struct ctl_table *table, int write,
102 void *buffer, size_t *lenp, loff_t *ppos)
103{
104 struct net *net = container_of(table->data, struct net,
105 ipv4.sysctl_ip_prot_sock);
106 int ret;
107 int pports;
108 int range[2];
109 struct ctl_table tmp = {
110 .data = &pports,
111 .maxlen = sizeof(pports),
112 .mode = table->mode,
113 .extra1 = &ip_privileged_port_min,
114 .extra2 = &ip_privileged_port_max,
115 };
116
117 pports = READ_ONCE(net->ipv4.sysctl_ip_prot_sock);
118
119 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
120
121 if (write && ret == 0) {
122 inet_get_local_port_range(net, &range[0], &range[1]);
123 /* Ensure that the local port range doesn't overlap with the
124 * privileged port range.
125 */
126 if (range[0] < pports)
127 ret = -EINVAL;
128 else
129 WRITE_ONCE(net->ipv4.sysctl_ip_prot_sock, pports);
130 }
131
132 return ret;
133}
134
135static void inet_get_ping_group_range_table(struct ctl_table *table, kgid_t *low, kgid_t *high)
136{
137 kgid_t *data = table->data;
138 struct net *net =
139 container_of(table->data, struct net, ipv4.ping_group_range.range);
140 unsigned int seq;
141 do {
142 seq = read_seqbegin(&net->ipv4.ping_group_range.lock);
143
144 *low = data[0];
145 *high = data[1];
146 } while (read_seqretry(&net->ipv4.ping_group_range.lock, seq));
147}
148
149/* Update system visible IP port range */
150static void set_ping_group_range(struct ctl_table *table, kgid_t low, kgid_t high)
151{
152 kgid_t *data = table->data;
153 struct net *net =
154 container_of(table->data, struct net, ipv4.ping_group_range.range);
155 write_seqlock(&net->ipv4.ping_group_range.lock);
156 data[0] = low;
157 data[1] = high;
158 write_sequnlock(&net->ipv4.ping_group_range.lock);
159}
160
161/* Validate changes from /proc interface. */
162static int ipv4_ping_group_range(struct ctl_table *table, int write,
163 void *buffer, size_t *lenp, loff_t *ppos)
164{
165 struct user_namespace *user_ns = current_user_ns();
166 int ret;
167 gid_t urange[2];
168 kgid_t low, high;
169 struct ctl_table tmp = {
170 .data = &urange,
171 .maxlen = sizeof(urange),
172 .mode = table->mode,
173 .extra1 = &ip_ping_group_range_min,
174 .extra2 = &ip_ping_group_range_max,
175 };
176
177 inet_get_ping_group_range_table(table, &low, &high);
178 urange[0] = from_kgid_munged(user_ns, low);
179 urange[1] = from_kgid_munged(user_ns, high);
180 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
181
182 if (write && ret == 0) {
183 low = make_kgid(user_ns, urange[0]);
184 high = make_kgid(user_ns, urange[1]);
185 if (!gid_valid(low) || !gid_valid(high))
186 return -EINVAL;
187 if (urange[1] < urange[0] || gid_lt(high, low)) {
188 low = make_kgid(&init_user_ns, 1);
189 high = make_kgid(&init_user_ns, 0);
190 }
191 set_ping_group_range(table, low, high);
192 }
193
194 return ret;
195}
196
197static int ipv4_fwd_update_priority(struct ctl_table *table, int write,
198 void *buffer, size_t *lenp, loff_t *ppos)
199{
200 struct net *net;
201 int ret;
202
203 net = container_of(table->data, struct net,
204 ipv4.sysctl_ip_fwd_update_priority);
205 ret = proc_dou8vec_minmax(table, write, buffer, lenp, ppos);
206 if (write && ret == 0)
207 call_netevent_notifiers(NETEVENT_IPV4_FWD_UPDATE_PRIORITY_UPDATE,
208 net);
209
210 return ret;
211}
212
213static int proc_tcp_congestion_control(struct ctl_table *ctl, int write,
214 void *buffer, size_t *lenp, loff_t *ppos)
215{
216 struct net *net = container_of(ctl->data, struct net,
217 ipv4.tcp_congestion_control);
218 char val[TCP_CA_NAME_MAX];
219 struct ctl_table tbl = {
220 .data = val,
221 .maxlen = TCP_CA_NAME_MAX,
222 };
223 int ret;
224
225 tcp_get_default_congestion_control(net, val);
226
227 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
228 if (write && ret == 0)
229 ret = tcp_set_default_congestion_control(net, val);
230 return ret;
231}
232
233static int proc_tcp_available_congestion_control(struct ctl_table *ctl,
234 int write, void *buffer,
235 size_t *lenp, loff_t *ppos)
236{
237 struct ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX, };
238 int ret;
239
240 tbl.data = kmalloc(tbl.maxlen, GFP_USER);
241 if (!tbl.data)
242 return -ENOMEM;
243 tcp_get_available_congestion_control(tbl.data, TCP_CA_BUF_MAX);
244 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
245 kfree(tbl.data);
246 return ret;
247}
248
249static int proc_allowed_congestion_control(struct ctl_table *ctl,
250 int write, void *buffer,
251 size_t *lenp, loff_t *ppos)
252{
253 struct ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX };
254 int ret;
255
256 tbl.data = kmalloc(tbl.maxlen, GFP_USER);
257 if (!tbl.data)
258 return -ENOMEM;
259
260 tcp_get_allowed_congestion_control(tbl.data, tbl.maxlen);
261 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
262 if (write && ret == 0)
263 ret = tcp_set_allowed_congestion_control(tbl.data);
264 kfree(tbl.data);
265 return ret;
266}
267
268static int sscanf_key(char *buf, __le32 *key)
269{
270 u32 user_key[4];
271 int i, ret = 0;
272
273 if (sscanf(buf, "%x-%x-%x-%x", user_key, user_key + 1,
274 user_key + 2, user_key + 3) != 4) {
275 ret = -EINVAL;
276 } else {
277 for (i = 0; i < ARRAY_SIZE(user_key); i++)
278 key[i] = cpu_to_le32(user_key[i]);
279 }
280 pr_debug("proc TFO key set 0x%x-%x-%x-%x <- 0x%s: %u\n",
281 user_key[0], user_key[1], user_key[2], user_key[3], buf, ret);
282
283 return ret;
284}
285
286static int proc_tcp_fastopen_key(struct ctl_table *table, int write,
287 void *buffer, size_t *lenp, loff_t *ppos)
288{
289 struct net *net = container_of(table->data, struct net,
290 ipv4.sysctl_tcp_fastopen);
291 /* maxlen to print the list of keys in hex (*2), with dashes
292 * separating doublewords and a comma in between keys.
293 */
294 struct ctl_table tbl = { .maxlen = ((TCP_FASTOPEN_KEY_LENGTH *
295 2 * TCP_FASTOPEN_KEY_MAX) +
296 (TCP_FASTOPEN_KEY_MAX * 5)) };
297 u32 user_key[TCP_FASTOPEN_KEY_BUF_LENGTH / sizeof(u32)];
298 __le32 key[TCP_FASTOPEN_KEY_BUF_LENGTH / sizeof(__le32)];
299 char *backup_data;
300 int ret, i = 0, off = 0, n_keys;
301
302 tbl.data = kmalloc(tbl.maxlen, GFP_KERNEL);
303 if (!tbl.data)
304 return -ENOMEM;
305
306 n_keys = tcp_fastopen_get_cipher(net, NULL, (u64 *)key);
307 if (!n_keys) {
308 memset(&key[0], 0, TCP_FASTOPEN_KEY_LENGTH);
309 n_keys = 1;
310 }
311
312 for (i = 0; i < n_keys * 4; i++)
313 user_key[i] = le32_to_cpu(key[i]);
314
315 for (i = 0; i < n_keys; i++) {
316 off += snprintf(tbl.data + off, tbl.maxlen - off,
317 "%08x-%08x-%08x-%08x",
318 user_key[i * 4],
319 user_key[i * 4 + 1],
320 user_key[i * 4 + 2],
321 user_key[i * 4 + 3]);
322
323 if (WARN_ON_ONCE(off >= tbl.maxlen - 1))
324 break;
325
326 if (i + 1 < n_keys)
327 off += snprintf(tbl.data + off, tbl.maxlen - off, ",");
328 }
329
330 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
331
332 if (write && ret == 0) {
333 backup_data = strchr(tbl.data, ',');
334 if (backup_data) {
335 *backup_data = '\0';
336 backup_data++;
337 }
338 if (sscanf_key(tbl.data, key)) {
339 ret = -EINVAL;
340 goto bad_key;
341 }
342 if (backup_data) {
343 if (sscanf_key(backup_data, key + 4)) {
344 ret = -EINVAL;
345 goto bad_key;
346 }
347 }
348 tcp_fastopen_reset_cipher(net, NULL, key,
349 backup_data ? key + 4 : NULL);
350 }
351
352bad_key:
353 kfree(tbl.data);
354 return ret;
355}
356
357static int proc_tfo_blackhole_detect_timeout(struct ctl_table *table,
358 int write, void *buffer,
359 size_t *lenp, loff_t *ppos)
360{
361 struct net *net = container_of(table->data, struct net,
362 ipv4.sysctl_tcp_fastopen_blackhole_timeout);
363 int ret;
364
365 ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
366 if (write && ret == 0)
367 atomic_set(&net->ipv4.tfo_active_disable_times, 0);
368
369 return ret;
370}
371
372static int proc_tcp_available_ulp(struct ctl_table *ctl,
373 int write, void *buffer, size_t *lenp,
374 loff_t *ppos)
375{
376 struct ctl_table tbl = { .maxlen = TCP_ULP_BUF_MAX, };
377 int ret;
378
379 tbl.data = kmalloc(tbl.maxlen, GFP_USER);
380 if (!tbl.data)
381 return -ENOMEM;
382 tcp_get_available_ulp(tbl.data, TCP_ULP_BUF_MAX);
383 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
384 kfree(tbl.data);
385
386 return ret;
387}
388
389static int proc_tcp_ehash_entries(struct ctl_table *table, int write,
390 void *buffer, size_t *lenp, loff_t *ppos)
391{
392 struct net *net = container_of(table->data, struct net,
393 ipv4.sysctl_tcp_child_ehash_entries);
394 struct inet_hashinfo *hinfo = net->ipv4.tcp_death_row.hashinfo;
395 int tcp_ehash_entries;
396 struct ctl_table tbl;
397
398 tcp_ehash_entries = hinfo->ehash_mask + 1;
399
400 /* A negative number indicates that the child netns
401 * shares the global ehash.
402 */
403 if (!net_eq(net, &init_net) && !hinfo->pernet)
404 tcp_ehash_entries *= -1;
405
406 memset(&tbl, 0, sizeof(tbl));
407 tbl.data = &tcp_ehash_entries;
408 tbl.maxlen = sizeof(int);
409
410 return proc_dointvec(&tbl, write, buffer, lenp, ppos);
411}
412
413static int proc_udp_hash_entries(struct ctl_table *table, int write,
414 void *buffer, size_t *lenp, loff_t *ppos)
415{
416 struct net *net = container_of(table->data, struct net,
417 ipv4.sysctl_udp_child_hash_entries);
418 int udp_hash_entries;
419 struct ctl_table tbl;
420
421 udp_hash_entries = net->ipv4.udp_table->mask + 1;
422
423 /* A negative number indicates that the child netns
424 * shares the global udp_table.
425 */
426 if (!net_eq(net, &init_net) && net->ipv4.udp_table == &udp_table)
427 udp_hash_entries *= -1;
428
429 memset(&tbl, 0, sizeof(tbl));
430 tbl.data = &udp_hash_entries;
431 tbl.maxlen = sizeof(int);
432
433 return proc_dointvec(&tbl, write, buffer, lenp, ppos);
434}
435
436#ifdef CONFIG_IP_ROUTE_MULTIPATH
437static int proc_fib_multipath_hash_policy(struct ctl_table *table, int write,
438 void *buffer, size_t *lenp,
439 loff_t *ppos)
440{
441 struct net *net = container_of(table->data, struct net,
442 ipv4.sysctl_fib_multipath_hash_policy);
443 int ret;
444
445 ret = proc_dou8vec_minmax(table, write, buffer, lenp, ppos);
446 if (write && ret == 0)
447 call_netevent_notifiers(NETEVENT_IPV4_MPATH_HASH_UPDATE, net);
448
449 return ret;
450}
451
452static int proc_fib_multipath_hash_fields(struct ctl_table *table, int write,
453 void *buffer, size_t *lenp,
454 loff_t *ppos)
455{
456 struct net *net;
457 int ret;
458
459 net = container_of(table->data, struct net,
460 ipv4.sysctl_fib_multipath_hash_fields);
461 ret = proc_douintvec_minmax(table, write, buffer, lenp, ppos);
462 if (write && ret == 0)
463 call_netevent_notifiers(NETEVENT_IPV4_MPATH_HASH_UPDATE, net);
464
465 return ret;
466}
467#endif
468
469static struct ctl_table ipv4_table[] = {
470 {
471 .procname = "tcp_max_orphans",
472 .data = &sysctl_tcp_max_orphans,
473 .maxlen = sizeof(int),
474 .mode = 0644,
475 .proc_handler = proc_dointvec
476 },
477 {
478 .procname = "inet_peer_threshold",
479 .data = &inet_peer_threshold,
480 .maxlen = sizeof(int),
481 .mode = 0644,
482 .proc_handler = proc_dointvec
483 },
484 {
485 .procname = "inet_peer_minttl",
486 .data = &inet_peer_minttl,
487 .maxlen = sizeof(int),
488 .mode = 0644,
489 .proc_handler = proc_dointvec_jiffies,
490 },
491 {
492 .procname = "inet_peer_maxttl",
493 .data = &inet_peer_maxttl,
494 .maxlen = sizeof(int),
495 .mode = 0644,
496 .proc_handler = proc_dointvec_jiffies,
497 },
498 {
499 .procname = "tcp_mem",
500 .maxlen = sizeof(sysctl_tcp_mem),
501 .data = &sysctl_tcp_mem,
502 .mode = 0644,
503 .proc_handler = proc_doulongvec_minmax,
504 },
505 {
506 .procname = "tcp_low_latency",
507 .data = &sysctl_tcp_low_latency,
508 .maxlen = sizeof(int),
509 .mode = 0644,
510 .proc_handler = proc_dointvec
511 },
512#ifdef CONFIG_NETLABEL
513 {
514 .procname = "cipso_cache_enable",
515 .data = &cipso_v4_cache_enabled,
516 .maxlen = sizeof(int),
517 .mode = 0644,
518 .proc_handler = proc_dointvec,
519 },
520 {
521 .procname = "cipso_cache_bucket_size",
522 .data = &cipso_v4_cache_bucketsize,
523 .maxlen = sizeof(int),
524 .mode = 0644,
525 .proc_handler = proc_dointvec,
526 },
527 {
528 .procname = "cipso_rbm_optfmt",
529 .data = &cipso_v4_rbm_optfmt,
530 .maxlen = sizeof(int),
531 .mode = 0644,
532 .proc_handler = proc_dointvec,
533 },
534 {
535 .procname = "cipso_rbm_strictvalid",
536 .data = &cipso_v4_rbm_strictvalid,
537 .maxlen = sizeof(int),
538 .mode = 0644,
539 .proc_handler = proc_dointvec,
540 },
541#endif /* CONFIG_NETLABEL */
542 {
543 .procname = "tcp_available_ulp",
544 .maxlen = TCP_ULP_BUF_MAX,
545 .mode = 0444,
546 .proc_handler = proc_tcp_available_ulp,
547 },
548 {
549 .procname = "icmp_msgs_per_sec",
550 .data = &sysctl_icmp_msgs_per_sec,
551 .maxlen = sizeof(int),
552 .mode = 0644,
553 .proc_handler = proc_dointvec_minmax,
554 .extra1 = SYSCTL_ZERO,
555 },
556 {
557 .procname = "icmp_msgs_burst",
558 .data = &sysctl_icmp_msgs_burst,
559 .maxlen = sizeof(int),
560 .mode = 0644,
561 .proc_handler = proc_dointvec_minmax,
562 .extra1 = SYSCTL_ZERO,
563 },
564 {
565 .procname = "udp_mem",
566 .data = &sysctl_udp_mem,
567 .maxlen = sizeof(sysctl_udp_mem),
568 .mode = 0644,
569 .proc_handler = proc_doulongvec_minmax,
570 },
571 {
572 .procname = "fib_sync_mem",
573 .data = &sysctl_fib_sync_mem,
574 .maxlen = sizeof(sysctl_fib_sync_mem),
575 .mode = 0644,
576 .proc_handler = proc_douintvec_minmax,
577 .extra1 = &sysctl_fib_sync_mem_min,
578 .extra2 = &sysctl_fib_sync_mem_max,
579 },
580 { }
581};
582
583static struct ctl_table ipv4_net_table[] = {
584 {
585 .procname = "tcp_max_tw_buckets",
586 .data = &init_net.ipv4.tcp_death_row.sysctl_max_tw_buckets,
587 .maxlen = sizeof(int),
588 .mode = 0644,
589 .proc_handler = proc_dointvec
590 },
591 {
592 .procname = "icmp_echo_ignore_all",
593 .data = &init_net.ipv4.sysctl_icmp_echo_ignore_all,
594 .maxlen = sizeof(u8),
595 .mode = 0644,
596 .proc_handler = proc_dou8vec_minmax,
597 .extra1 = SYSCTL_ZERO,
598 .extra2 = SYSCTL_ONE
599 },
600 {
601 .procname = "icmp_echo_enable_probe",
602 .data = &init_net.ipv4.sysctl_icmp_echo_enable_probe,
603 .maxlen = sizeof(u8),
604 .mode = 0644,
605 .proc_handler = proc_dou8vec_minmax,
606 .extra1 = SYSCTL_ZERO,
607 .extra2 = SYSCTL_ONE
608 },
609 {
610 .procname = "icmp_echo_ignore_broadcasts",
611 .data = &init_net.ipv4.sysctl_icmp_echo_ignore_broadcasts,
612 .maxlen = sizeof(u8),
613 .mode = 0644,
614 .proc_handler = proc_dou8vec_minmax,
615 .extra1 = SYSCTL_ZERO,
616 .extra2 = SYSCTL_ONE
617 },
618 {
619 .procname = "icmp_ignore_bogus_error_responses",
620 .data = &init_net.ipv4.sysctl_icmp_ignore_bogus_error_responses,
621 .maxlen = sizeof(u8),
622 .mode = 0644,
623 .proc_handler = proc_dou8vec_minmax,
624 .extra1 = SYSCTL_ZERO,
625 .extra2 = SYSCTL_ONE
626 },
627 {
628 .procname = "icmp_errors_use_inbound_ifaddr",
629 .data = &init_net.ipv4.sysctl_icmp_errors_use_inbound_ifaddr,
630 .maxlen = sizeof(u8),
631 .mode = 0644,
632 .proc_handler = proc_dou8vec_minmax,
633 .extra1 = SYSCTL_ZERO,
634 .extra2 = SYSCTL_ONE
635 },
636 {
637 .procname = "icmp_ratelimit",
638 .data = &init_net.ipv4.sysctl_icmp_ratelimit,
639 .maxlen = sizeof(int),
640 .mode = 0644,
641 .proc_handler = proc_dointvec_ms_jiffies,
642 },
643 {
644 .procname = "icmp_ratemask",
645 .data = &init_net.ipv4.sysctl_icmp_ratemask,
646 .maxlen = sizeof(int),
647 .mode = 0644,
648 .proc_handler = proc_dointvec
649 },
650 {
651 .procname = "ping_group_range",
652 .data = &init_net.ipv4.ping_group_range.range,
653 .maxlen = sizeof(gid_t)*2,
654 .mode = 0644,
655 .proc_handler = ipv4_ping_group_range,
656 },
657#ifdef CONFIG_NET_L3_MASTER_DEV
658 {
659 .procname = "raw_l3mdev_accept",
660 .data = &init_net.ipv4.sysctl_raw_l3mdev_accept,
661 .maxlen = sizeof(u8),
662 .mode = 0644,
663 .proc_handler = proc_dou8vec_minmax,
664 .extra1 = SYSCTL_ZERO,
665 .extra2 = SYSCTL_ONE,
666 },
667#endif
668 {
669 .procname = "tcp_ecn",
670 .data = &init_net.ipv4.sysctl_tcp_ecn,
671 .maxlen = sizeof(u8),
672 .mode = 0644,
673 .proc_handler = proc_dou8vec_minmax,
674 .extra1 = SYSCTL_ZERO,
675 .extra2 = SYSCTL_TWO,
676 },
677 {
678 .procname = "tcp_ecn_fallback",
679 .data = &init_net.ipv4.sysctl_tcp_ecn_fallback,
680 .maxlen = sizeof(u8),
681 .mode = 0644,
682 .proc_handler = proc_dou8vec_minmax,
683 .extra1 = SYSCTL_ZERO,
684 .extra2 = SYSCTL_ONE,
685 },
686 {
687 .procname = "ip_dynaddr",
688 .data = &init_net.ipv4.sysctl_ip_dynaddr,
689 .maxlen = sizeof(u8),
690 .mode = 0644,
691 .proc_handler = proc_dou8vec_minmax,
692 },
693 {
694 .procname = "ip_early_demux",
695 .data = &init_net.ipv4.sysctl_ip_early_demux,
696 .maxlen = sizeof(u8),
697 .mode = 0644,
698 .proc_handler = proc_dou8vec_minmax,
699 },
700 {
701 .procname = "udp_early_demux",
702 .data = &init_net.ipv4.sysctl_udp_early_demux,
703 .maxlen = sizeof(u8),
704 .mode = 0644,
705 .proc_handler = proc_dou8vec_minmax,
706 },
707 {
708 .procname = "tcp_early_demux",
709 .data = &init_net.ipv4.sysctl_tcp_early_demux,
710 .maxlen = sizeof(u8),
711 .mode = 0644,
712 .proc_handler = proc_dou8vec_minmax,
713 },
714 {
715 .procname = "nexthop_compat_mode",
716 .data = &init_net.ipv4.sysctl_nexthop_compat_mode,
717 .maxlen = sizeof(u8),
718 .mode = 0644,
719 .proc_handler = proc_dou8vec_minmax,
720 .extra1 = SYSCTL_ZERO,
721 .extra2 = SYSCTL_ONE,
722 },
723 {
724 .procname = "ip_default_ttl",
725 .data = &init_net.ipv4.sysctl_ip_default_ttl,
726 .maxlen = sizeof(u8),
727 .mode = 0644,
728 .proc_handler = proc_dou8vec_minmax,
729 .extra1 = &ip_ttl_min,
730 .extra2 = &ip_ttl_max,
731 },
732 {
733 .procname = "ip_local_port_range",
734 .maxlen = sizeof(init_net.ipv4.ip_local_ports.range),
735 .data = &init_net.ipv4.ip_local_ports.range,
736 .mode = 0644,
737 .proc_handler = ipv4_local_port_range,
738 },
739 {
740 .procname = "ip_local_reserved_ports",
741 .data = &init_net.ipv4.sysctl_local_reserved_ports,
742 .maxlen = 65536,
743 .mode = 0644,
744 .proc_handler = proc_do_large_bitmap,
745 },
746 {
747 .procname = "ip_no_pmtu_disc",
748 .data = &init_net.ipv4.sysctl_ip_no_pmtu_disc,
749 .maxlen = sizeof(u8),
750 .mode = 0644,
751 .proc_handler = proc_dou8vec_minmax,
752 },
753 {
754 .procname = "ip_forward_use_pmtu",
755 .data = &init_net.ipv4.sysctl_ip_fwd_use_pmtu,
756 .maxlen = sizeof(u8),
757 .mode = 0644,
758 .proc_handler = proc_dou8vec_minmax,
759 },
760 {
761 .procname = "ip_forward_update_priority",
762 .data = &init_net.ipv4.sysctl_ip_fwd_update_priority,
763 .maxlen = sizeof(u8),
764 .mode = 0644,
765 .proc_handler = ipv4_fwd_update_priority,
766 .extra1 = SYSCTL_ZERO,
767 .extra2 = SYSCTL_ONE,
768 },
769 {
770 .procname = "ip_nonlocal_bind",
771 .data = &init_net.ipv4.sysctl_ip_nonlocal_bind,
772 .maxlen = sizeof(u8),
773 .mode = 0644,
774 .proc_handler = proc_dou8vec_minmax,
775 },
776 {
777 .procname = "ip_autobind_reuse",
778 .data = &init_net.ipv4.sysctl_ip_autobind_reuse,
779 .maxlen = sizeof(u8),
780 .mode = 0644,
781 .proc_handler = proc_dou8vec_minmax,
782 .extra1 = SYSCTL_ZERO,
783 .extra2 = SYSCTL_ONE,
784 },
785 {
786 .procname = "fwmark_reflect",
787 .data = &init_net.ipv4.sysctl_fwmark_reflect,
788 .maxlen = sizeof(u8),
789 .mode = 0644,
790 .proc_handler = proc_dou8vec_minmax,
791 },
792 {
793 .procname = "tcp_fwmark_accept",
794 .data = &init_net.ipv4.sysctl_tcp_fwmark_accept,
795 .maxlen = sizeof(u8),
796 .mode = 0644,
797 .proc_handler = proc_dou8vec_minmax,
798 },
799#ifdef CONFIG_NET_L3_MASTER_DEV
800 {
801 .procname = "tcp_l3mdev_accept",
802 .data = &init_net.ipv4.sysctl_tcp_l3mdev_accept,
803 .maxlen = sizeof(u8),
804 .mode = 0644,
805 .proc_handler = proc_dou8vec_minmax,
806 .extra1 = SYSCTL_ZERO,
807 .extra2 = SYSCTL_ONE,
808 },
809#endif
810 {
811 .procname = "tcp_mtu_probing",
812 .data = &init_net.ipv4.sysctl_tcp_mtu_probing,
813 .maxlen = sizeof(u8),
814 .mode = 0644,
815 .proc_handler = proc_dou8vec_minmax,
816 },
817 {
818 .procname = "tcp_base_mss",
819 .data = &init_net.ipv4.sysctl_tcp_base_mss,
820 .maxlen = sizeof(int),
821 .mode = 0644,
822 .proc_handler = proc_dointvec,
823 },
824 {
825 .procname = "tcp_min_snd_mss",
826 .data = &init_net.ipv4.sysctl_tcp_min_snd_mss,
827 .maxlen = sizeof(int),
828 .mode = 0644,
829 .proc_handler = proc_dointvec_minmax,
830 .extra1 = &tcp_min_snd_mss_min,
831 .extra2 = &tcp_min_snd_mss_max,
832 },
833 {
834 .procname = "tcp_mtu_probe_floor",
835 .data = &init_net.ipv4.sysctl_tcp_mtu_probe_floor,
836 .maxlen = sizeof(int),
837 .mode = 0644,
838 .proc_handler = proc_dointvec_minmax,
839 .extra1 = &tcp_min_snd_mss_min,
840 .extra2 = &tcp_min_snd_mss_max,
841 },
842 {
843 .procname = "tcp_probe_threshold",
844 .data = &init_net.ipv4.sysctl_tcp_probe_threshold,
845 .maxlen = sizeof(int),
846 .mode = 0644,
847 .proc_handler = proc_dointvec,
848 },
849 {
850 .procname = "tcp_probe_interval",
851 .data = &init_net.ipv4.sysctl_tcp_probe_interval,
852 .maxlen = sizeof(u32),
853 .mode = 0644,
854 .proc_handler = proc_douintvec_minmax,
855 .extra2 = &u32_max_div_HZ,
856 },
857 {
858 .procname = "igmp_link_local_mcast_reports",
859 .data = &init_net.ipv4.sysctl_igmp_llm_reports,
860 .maxlen = sizeof(u8),
861 .mode = 0644,
862 .proc_handler = proc_dou8vec_minmax,
863 },
864 {
865 .procname = "igmp_max_memberships",
866 .data = &init_net.ipv4.sysctl_igmp_max_memberships,
867 .maxlen = sizeof(int),
868 .mode = 0644,
869 .proc_handler = proc_dointvec
870 },
871 {
872 .procname = "igmp_max_msf",
873 .data = &init_net.ipv4.sysctl_igmp_max_msf,
874 .maxlen = sizeof(int),
875 .mode = 0644,
876 .proc_handler = proc_dointvec
877 },
878#ifdef CONFIG_IP_MULTICAST
879 {
880 .procname = "igmp_qrv",
881 .data = &init_net.ipv4.sysctl_igmp_qrv,
882 .maxlen = sizeof(int),
883 .mode = 0644,
884 .proc_handler = proc_dointvec_minmax,
885 .extra1 = SYSCTL_ONE
886 },
887#endif
888 {
889 .procname = "tcp_congestion_control",
890 .data = &init_net.ipv4.tcp_congestion_control,
891 .mode = 0644,
892 .maxlen = TCP_CA_NAME_MAX,
893 .proc_handler = proc_tcp_congestion_control,
894 },
895 {
896 .procname = "tcp_available_congestion_control",
897 .maxlen = TCP_CA_BUF_MAX,
898 .mode = 0444,
899 .proc_handler = proc_tcp_available_congestion_control,
900 },
901 {
902 .procname = "tcp_allowed_congestion_control",
903 .maxlen = TCP_CA_BUF_MAX,
904 .mode = 0644,
905 .proc_handler = proc_allowed_congestion_control,
906 },
907 {
908 .procname = "tcp_keepalive_time",
909 .data = &init_net.ipv4.sysctl_tcp_keepalive_time,
910 .maxlen = sizeof(int),
911 .mode = 0644,
912 .proc_handler = proc_dointvec_jiffies,
913 },
914 {
915 .procname = "tcp_keepalive_probes",
916 .data = &init_net.ipv4.sysctl_tcp_keepalive_probes,
917 .maxlen = sizeof(u8),
918 .mode = 0644,
919 .proc_handler = proc_dou8vec_minmax,
920 },
921 {
922 .procname = "tcp_keepalive_intvl",
923 .data = &init_net.ipv4.sysctl_tcp_keepalive_intvl,
924 .maxlen = sizeof(int),
925 .mode = 0644,
926 .proc_handler = proc_dointvec_jiffies,
927 },
928 {
929 .procname = "tcp_syn_retries",
930 .data = &init_net.ipv4.sysctl_tcp_syn_retries,
931 .maxlen = sizeof(u8),
932 .mode = 0644,
933 .proc_handler = proc_dou8vec_minmax,
934 .extra1 = &tcp_syn_retries_min,
935 .extra2 = &tcp_syn_retries_max
936 },
937 {
938 .procname = "tcp_synack_retries",
939 .data = &init_net.ipv4.sysctl_tcp_synack_retries,
940 .maxlen = sizeof(u8),
941 .mode = 0644,
942 .proc_handler = proc_dou8vec_minmax,
943 },
944#ifdef CONFIG_SYN_COOKIES
945 {
946 .procname = "tcp_syncookies",
947 .data = &init_net.ipv4.sysctl_tcp_syncookies,
948 .maxlen = sizeof(u8),
949 .mode = 0644,
950 .proc_handler = proc_dou8vec_minmax,
951 },
952#endif
953 {
954 .procname = "tcp_migrate_req",
955 .data = &init_net.ipv4.sysctl_tcp_migrate_req,
956 .maxlen = sizeof(u8),
957 .mode = 0644,
958 .proc_handler = proc_dou8vec_minmax,
959 .extra1 = SYSCTL_ZERO,
960 .extra2 = SYSCTL_ONE
961 },
962 {
963 .procname = "tcp_reordering",
964 .data = &init_net.ipv4.sysctl_tcp_reordering,
965 .maxlen = sizeof(int),
966 .mode = 0644,
967 .proc_handler = proc_dointvec
968 },
969 {
970 .procname = "tcp_retries1",
971 .data = &init_net.ipv4.sysctl_tcp_retries1,
972 .maxlen = sizeof(u8),
973 .mode = 0644,
974 .proc_handler = proc_dou8vec_minmax,
975 .extra2 = &tcp_retr1_max
976 },
977 {
978 .procname = "tcp_retries2",
979 .data = &init_net.ipv4.sysctl_tcp_retries2,
980 .maxlen = sizeof(u8),
981 .mode = 0644,
982 .proc_handler = proc_dou8vec_minmax,
983 },
984 {
985 .procname = "tcp_orphan_retries",
986 .data = &init_net.ipv4.sysctl_tcp_orphan_retries,
987 .maxlen = sizeof(u8),
988 .mode = 0644,
989 .proc_handler = proc_dou8vec_minmax,
990 },
991 {
992 .procname = "tcp_fin_timeout",
993 .data = &init_net.ipv4.sysctl_tcp_fin_timeout,
994 .maxlen = sizeof(int),
995 .mode = 0644,
996 .proc_handler = proc_dointvec_jiffies,
997 },
998 {
999 .procname = "tcp_notsent_lowat",
1000 .data = &init_net.ipv4.sysctl_tcp_notsent_lowat,
1001 .maxlen = sizeof(unsigned int),
1002 .mode = 0644,
1003 .proc_handler = proc_douintvec,
1004 },
1005 {
1006 .procname = "tcp_tw_reuse",
1007 .data = &init_net.ipv4.sysctl_tcp_tw_reuse,
1008 .maxlen = sizeof(u8),
1009 .mode = 0644,
1010 .proc_handler = proc_dou8vec_minmax,
1011 .extra1 = SYSCTL_ZERO,
1012 .extra2 = SYSCTL_TWO,
1013 },
1014 {
1015 .procname = "tcp_max_syn_backlog",
1016 .data = &init_net.ipv4.sysctl_max_syn_backlog,
1017 .maxlen = sizeof(int),
1018 .mode = 0644,
1019 .proc_handler = proc_dointvec
1020 },
1021 {
1022 .procname = "tcp_fastopen",
1023 .data = &init_net.ipv4.sysctl_tcp_fastopen,
1024 .maxlen = sizeof(int),
1025 .mode = 0644,
1026 .proc_handler = proc_dointvec,
1027 },
1028 {
1029 .procname = "tcp_fastopen_key",
1030 .mode = 0600,
1031 .data = &init_net.ipv4.sysctl_tcp_fastopen,
1032 /* maxlen to print the list of keys in hex (*2), with dashes
1033 * separating doublewords and a comma in between keys.
1034 */
1035 .maxlen = ((TCP_FASTOPEN_KEY_LENGTH *
1036 2 * TCP_FASTOPEN_KEY_MAX) +
1037 (TCP_FASTOPEN_KEY_MAX * 5)),
1038 .proc_handler = proc_tcp_fastopen_key,
1039 },
1040 {
1041 .procname = "tcp_fastopen_blackhole_timeout_sec",
1042 .data = &init_net.ipv4.sysctl_tcp_fastopen_blackhole_timeout,
1043 .maxlen = sizeof(int),
1044 .mode = 0644,
1045 .proc_handler = proc_tfo_blackhole_detect_timeout,
1046 .extra1 = SYSCTL_ZERO,
1047 },
1048#ifdef CONFIG_IP_ROUTE_MULTIPATH
1049 {
1050 .procname = "fib_multipath_use_neigh",
1051 .data = &init_net.ipv4.sysctl_fib_multipath_use_neigh,
1052 .maxlen = sizeof(u8),
1053 .mode = 0644,
1054 .proc_handler = proc_dou8vec_minmax,
1055 .extra1 = SYSCTL_ZERO,
1056 .extra2 = SYSCTL_ONE,
1057 },
1058 {
1059 .procname = "fib_multipath_hash_policy",
1060 .data = &init_net.ipv4.sysctl_fib_multipath_hash_policy,
1061 .maxlen = sizeof(u8),
1062 .mode = 0644,
1063 .proc_handler = proc_fib_multipath_hash_policy,
1064 .extra1 = SYSCTL_ZERO,
1065 .extra2 = SYSCTL_THREE,
1066 },
1067 {
1068 .procname = "fib_multipath_hash_fields",
1069 .data = &init_net.ipv4.sysctl_fib_multipath_hash_fields,
1070 .maxlen = sizeof(u32),
1071 .mode = 0644,
1072 .proc_handler = proc_fib_multipath_hash_fields,
1073 .extra1 = SYSCTL_ONE,
1074 .extra2 = &fib_multipath_hash_fields_all_mask,
1075 },
1076#endif
1077 {
1078 .procname = "ip_unprivileged_port_start",
1079 .maxlen = sizeof(int),
1080 .data = &init_net.ipv4.sysctl_ip_prot_sock,
1081 .mode = 0644,
1082 .proc_handler = ipv4_privileged_ports,
1083 },
1084#ifdef CONFIG_NET_L3_MASTER_DEV
1085 {
1086 .procname = "udp_l3mdev_accept",
1087 .data = &init_net.ipv4.sysctl_udp_l3mdev_accept,
1088 .maxlen = sizeof(u8),
1089 .mode = 0644,
1090 .proc_handler = proc_dou8vec_minmax,
1091 .extra1 = SYSCTL_ZERO,
1092 .extra2 = SYSCTL_ONE,
1093 },
1094#endif
1095 {
1096 .procname = "tcp_sack",
1097 .data = &init_net.ipv4.sysctl_tcp_sack,
1098 .maxlen = sizeof(u8),
1099 .mode = 0644,
1100 .proc_handler = proc_dou8vec_minmax,
1101 },
1102 {
1103 .procname = "tcp_window_scaling",
1104 .data = &init_net.ipv4.sysctl_tcp_window_scaling,
1105 .maxlen = sizeof(u8),
1106 .mode = 0644,
1107 .proc_handler = proc_dou8vec_minmax,
1108 },
1109 {
1110 .procname = "tcp_timestamps",
1111 .data = &init_net.ipv4.sysctl_tcp_timestamps,
1112 .maxlen = sizeof(u8),
1113 .mode = 0644,
1114 .proc_handler = proc_dou8vec_minmax,
1115 },
1116 {
1117 .procname = "tcp_early_retrans",
1118 .data = &init_net.ipv4.sysctl_tcp_early_retrans,
1119 .maxlen = sizeof(u8),
1120 .mode = 0644,
1121 .proc_handler = proc_dou8vec_minmax,
1122 .extra1 = SYSCTL_ZERO,
1123 .extra2 = SYSCTL_FOUR,
1124 },
1125 {
1126 .procname = "tcp_recovery",
1127 .data = &init_net.ipv4.sysctl_tcp_recovery,
1128 .maxlen = sizeof(u8),
1129 .mode = 0644,
1130 .proc_handler = proc_dou8vec_minmax,
1131 },
1132 {
1133 .procname = "tcp_thin_linear_timeouts",
1134 .data = &init_net.ipv4.sysctl_tcp_thin_linear_timeouts,
1135 .maxlen = sizeof(u8),
1136 .mode = 0644,
1137 .proc_handler = proc_dou8vec_minmax,
1138 },
1139 {
1140 .procname = "tcp_slow_start_after_idle",
1141 .data = &init_net.ipv4.sysctl_tcp_slow_start_after_idle,
1142 .maxlen = sizeof(u8),
1143 .mode = 0644,
1144 .proc_handler = proc_dou8vec_minmax,
1145 },
1146 {
1147 .procname = "tcp_retrans_collapse",
1148 .data = &init_net.ipv4.sysctl_tcp_retrans_collapse,
1149 .maxlen = sizeof(u8),
1150 .mode = 0644,
1151 .proc_handler = proc_dou8vec_minmax,
1152 },
1153 {
1154 .procname = "tcp_stdurg",
1155 .data = &init_net.ipv4.sysctl_tcp_stdurg,
1156 .maxlen = sizeof(u8),
1157 .mode = 0644,
1158 .proc_handler = proc_dou8vec_minmax,
1159 },
1160 {
1161 .procname = "tcp_rfc1337",
1162 .data = &init_net.ipv4.sysctl_tcp_rfc1337,
1163 .maxlen = sizeof(u8),
1164 .mode = 0644,
1165 .proc_handler = proc_dou8vec_minmax,
1166 },
1167 {
1168 .procname = "tcp_abort_on_overflow",
1169 .data = &init_net.ipv4.sysctl_tcp_abort_on_overflow,
1170 .maxlen = sizeof(u8),
1171 .mode = 0644,
1172 .proc_handler = proc_dou8vec_minmax,
1173 },
1174 {
1175 .procname = "tcp_fack",
1176 .data = &init_net.ipv4.sysctl_tcp_fack,
1177 .maxlen = sizeof(u8),
1178 .mode = 0644,
1179 .proc_handler = proc_dou8vec_minmax,
1180 },
1181 {
1182 .procname = "tcp_max_reordering",
1183 .data = &init_net.ipv4.sysctl_tcp_max_reordering,
1184 .maxlen = sizeof(int),
1185 .mode = 0644,
1186 .proc_handler = proc_dointvec
1187 },
1188 {
1189 .procname = "tcp_dsack",
1190 .data = &init_net.ipv4.sysctl_tcp_dsack,
1191 .maxlen = sizeof(u8),
1192 .mode = 0644,
1193 .proc_handler = proc_dou8vec_minmax,
1194 },
1195 {
1196 .procname = "tcp_app_win",
1197 .data = &init_net.ipv4.sysctl_tcp_app_win,
1198 .maxlen = sizeof(u8),
1199 .mode = 0644,
1200 .proc_handler = proc_dou8vec_minmax,
1201 },
1202 {
1203 .procname = "tcp_adv_win_scale",
1204 .data = &init_net.ipv4.sysctl_tcp_adv_win_scale,
1205 .maxlen = sizeof(int),
1206 .mode = 0644,
1207 .proc_handler = proc_dointvec_minmax,
1208 .extra1 = &tcp_adv_win_scale_min,
1209 .extra2 = &tcp_adv_win_scale_max,
1210 },
1211 {
1212 .procname = "tcp_frto",
1213 .data = &init_net.ipv4.sysctl_tcp_frto,
1214 .maxlen = sizeof(u8),
1215 .mode = 0644,
1216 .proc_handler = proc_dou8vec_minmax,
1217 },
1218 {
1219 .procname = "tcp_no_metrics_save",
1220 .data = &init_net.ipv4.sysctl_tcp_nometrics_save,
1221 .maxlen = sizeof(u8),
1222 .mode = 0644,
1223 .proc_handler = proc_dou8vec_minmax,
1224 },
1225 {
1226 .procname = "tcp_no_ssthresh_metrics_save",
1227 .data = &init_net.ipv4.sysctl_tcp_no_ssthresh_metrics_save,
1228 .maxlen = sizeof(u8),
1229 .mode = 0644,
1230 .proc_handler = proc_dou8vec_minmax,
1231 .extra1 = SYSCTL_ZERO,
1232 .extra2 = SYSCTL_ONE,
1233 },
1234 {
1235 .procname = "tcp_moderate_rcvbuf",
1236 .data = &init_net.ipv4.sysctl_tcp_moderate_rcvbuf,
1237 .maxlen = sizeof(u8),
1238 .mode = 0644,
1239 .proc_handler = proc_dou8vec_minmax,
1240 },
1241 {
1242 .procname = "tcp_tso_win_divisor",
1243 .data = &init_net.ipv4.sysctl_tcp_tso_win_divisor,
1244 .maxlen = sizeof(u8),
1245 .mode = 0644,
1246 .proc_handler = proc_dou8vec_minmax,
1247 },
1248 {
1249 .procname = "tcp_workaround_signed_windows",
1250 .data = &init_net.ipv4.sysctl_tcp_workaround_signed_windows,
1251 .maxlen = sizeof(u8),
1252 .mode = 0644,
1253 .proc_handler = proc_dou8vec_minmax,
1254 },
1255 {
1256 .procname = "tcp_limit_output_bytes",
1257 .data = &init_net.ipv4.sysctl_tcp_limit_output_bytes,
1258 .maxlen = sizeof(int),
1259 .mode = 0644,
1260 .proc_handler = proc_dointvec
1261 },
1262 {
1263 .procname = "tcp_challenge_ack_limit",
1264 .data = &init_net.ipv4.sysctl_tcp_challenge_ack_limit,
1265 .maxlen = sizeof(int),
1266 .mode = 0644,
1267 .proc_handler = proc_dointvec
1268 },
1269 {
1270 .procname = "tcp_min_tso_segs",
1271 .data = &init_net.ipv4.sysctl_tcp_min_tso_segs,
1272 .maxlen = sizeof(u8),
1273 .mode = 0644,
1274 .proc_handler = proc_dou8vec_minmax,
1275 .extra1 = SYSCTL_ONE,
1276 },
1277 {
1278 .procname = "tcp_tso_rtt_log",
1279 .data = &init_net.ipv4.sysctl_tcp_tso_rtt_log,
1280 .maxlen = sizeof(u8),
1281 .mode = 0644,
1282 .proc_handler = proc_dou8vec_minmax,
1283 },
1284 {
1285 .procname = "tcp_min_rtt_wlen",
1286 .data = &init_net.ipv4.sysctl_tcp_min_rtt_wlen,
1287 .maxlen = sizeof(int),
1288 .mode = 0644,
1289 .proc_handler = proc_dointvec_minmax,
1290 .extra1 = SYSCTL_ZERO,
1291 .extra2 = &one_day_secs
1292 },
1293 {
1294 .procname = "tcp_autocorking",
1295 .data = &init_net.ipv4.sysctl_tcp_autocorking,
1296 .maxlen = sizeof(u8),
1297 .mode = 0644,
1298 .proc_handler = proc_dou8vec_minmax,
1299 .extra1 = SYSCTL_ZERO,
1300 .extra2 = SYSCTL_ONE,
1301 },
1302 {
1303 .procname = "tcp_invalid_ratelimit",
1304 .data = &init_net.ipv4.sysctl_tcp_invalid_ratelimit,
1305 .maxlen = sizeof(int),
1306 .mode = 0644,
1307 .proc_handler = proc_dointvec_ms_jiffies,
1308 },
1309 {
1310 .procname = "tcp_pacing_ss_ratio",
1311 .data = &init_net.ipv4.sysctl_tcp_pacing_ss_ratio,
1312 .maxlen = sizeof(int),
1313 .mode = 0644,
1314 .proc_handler = proc_dointvec_minmax,
1315 .extra1 = SYSCTL_ZERO,
1316 .extra2 = SYSCTL_ONE_THOUSAND,
1317 },
1318 {
1319 .procname = "tcp_pacing_ca_ratio",
1320 .data = &init_net.ipv4.sysctl_tcp_pacing_ca_ratio,
1321 .maxlen = sizeof(int),
1322 .mode = 0644,
1323 .proc_handler = proc_dointvec_minmax,
1324 .extra1 = SYSCTL_ZERO,
1325 .extra2 = SYSCTL_ONE_THOUSAND,
1326 },
1327 {
1328 .procname = "tcp_wmem",
1329 .data = &init_net.ipv4.sysctl_tcp_wmem,
1330 .maxlen = sizeof(init_net.ipv4.sysctl_tcp_wmem),
1331 .mode = 0644,
1332 .proc_handler = proc_dointvec_minmax,
1333 .extra1 = SYSCTL_ONE,
1334 },
1335 {
1336 .procname = "tcp_rmem",
1337 .data = &init_net.ipv4.sysctl_tcp_rmem,
1338 .maxlen = sizeof(init_net.ipv4.sysctl_tcp_rmem),
1339 .mode = 0644,
1340 .proc_handler = proc_dointvec_minmax,
1341 .extra1 = SYSCTL_ONE,
1342 },
1343 {
1344 .procname = "tcp_comp_sack_delay_ns",
1345 .data = &init_net.ipv4.sysctl_tcp_comp_sack_delay_ns,
1346 .maxlen = sizeof(unsigned long),
1347 .mode = 0644,
1348 .proc_handler = proc_doulongvec_minmax,
1349 },
1350 {
1351 .procname = "tcp_comp_sack_slack_ns",
1352 .data = &init_net.ipv4.sysctl_tcp_comp_sack_slack_ns,
1353 .maxlen = sizeof(unsigned long),
1354 .mode = 0644,
1355 .proc_handler = proc_doulongvec_minmax,
1356 },
1357 {
1358 .procname = "tcp_comp_sack_nr",
1359 .data = &init_net.ipv4.sysctl_tcp_comp_sack_nr,
1360 .maxlen = sizeof(u8),
1361 .mode = 0644,
1362 .proc_handler = proc_dou8vec_minmax,
1363 .extra1 = SYSCTL_ZERO,
1364 },
1365 {
1366 .procname = "tcp_reflect_tos",
1367 .data = &init_net.ipv4.sysctl_tcp_reflect_tos,
1368 .maxlen = sizeof(u8),
1369 .mode = 0644,
1370 .proc_handler = proc_dou8vec_minmax,
1371 .extra1 = SYSCTL_ZERO,
1372 .extra2 = SYSCTL_ONE,
1373 },
1374 {
1375 .procname = "tcp_ehash_entries",
1376 .data = &init_net.ipv4.sysctl_tcp_child_ehash_entries,
1377 .mode = 0444,
1378 .proc_handler = proc_tcp_ehash_entries,
1379 },
1380 {
1381 .procname = "tcp_child_ehash_entries",
1382 .data = &init_net.ipv4.sysctl_tcp_child_ehash_entries,
1383 .maxlen = sizeof(unsigned int),
1384 .mode = 0644,
1385 .proc_handler = proc_douintvec_minmax,
1386 .extra1 = SYSCTL_ZERO,
1387 .extra2 = &tcp_child_ehash_entries_max,
1388 },
1389 {
1390 .procname = "udp_hash_entries",
1391 .data = &init_net.ipv4.sysctl_udp_child_hash_entries,
1392 .mode = 0444,
1393 .proc_handler = proc_udp_hash_entries,
1394 },
1395 {
1396 .procname = "udp_child_hash_entries",
1397 .data = &init_net.ipv4.sysctl_udp_child_hash_entries,
1398 .maxlen = sizeof(unsigned int),
1399 .mode = 0644,
1400 .proc_handler = proc_douintvec_minmax,
1401 .extra1 = SYSCTL_ZERO,
1402 .extra2 = &udp_child_hash_entries_max,
1403 },
1404 {
1405 .procname = "udp_rmem_min",
1406 .data = &init_net.ipv4.sysctl_udp_rmem_min,
1407 .maxlen = sizeof(init_net.ipv4.sysctl_udp_rmem_min),
1408 .mode = 0644,
1409 .proc_handler = proc_dointvec_minmax,
1410 .extra1 = SYSCTL_ONE
1411 },
1412 {
1413 .procname = "udp_wmem_min",
1414 .data = &init_net.ipv4.sysctl_udp_wmem_min,
1415 .maxlen = sizeof(init_net.ipv4.sysctl_udp_wmem_min),
1416 .mode = 0644,
1417 .proc_handler = proc_dointvec_minmax,
1418 .extra1 = SYSCTL_ONE
1419 },
1420 {
1421 .procname = "fib_notify_on_flag_change",
1422 .data = &init_net.ipv4.sysctl_fib_notify_on_flag_change,
1423 .maxlen = sizeof(u8),
1424 .mode = 0644,
1425 .proc_handler = proc_dou8vec_minmax,
1426 .extra1 = SYSCTL_ZERO,
1427 .extra2 = SYSCTL_TWO,
1428 },
1429 {
1430 .procname = "tcp_plb_enabled",
1431 .data = &init_net.ipv4.sysctl_tcp_plb_enabled,
1432 .maxlen = sizeof(u8),
1433 .mode = 0644,
1434 .proc_handler = proc_dou8vec_minmax,
1435 .extra1 = SYSCTL_ZERO,
1436 .extra2 = SYSCTL_ONE,
1437 },
1438 {
1439 .procname = "tcp_plb_idle_rehash_rounds",
1440 .data = &init_net.ipv4.sysctl_tcp_plb_idle_rehash_rounds,
1441 .maxlen = sizeof(u8),
1442 .mode = 0644,
1443 .proc_handler = proc_dou8vec_minmax,
1444 .extra2 = &tcp_plb_max_rounds,
1445 },
1446 {
1447 .procname = "tcp_plb_rehash_rounds",
1448 .data = &init_net.ipv4.sysctl_tcp_plb_rehash_rounds,
1449 .maxlen = sizeof(u8),
1450 .mode = 0644,
1451 .proc_handler = proc_dou8vec_minmax,
1452 .extra2 = &tcp_plb_max_rounds,
1453 },
1454 {
1455 .procname = "tcp_plb_suspend_rto_sec",
1456 .data = &init_net.ipv4.sysctl_tcp_plb_suspend_rto_sec,
1457 .maxlen = sizeof(u8),
1458 .mode = 0644,
1459 .proc_handler = proc_dou8vec_minmax,
1460 },
1461 {
1462 .procname = "tcp_plb_cong_thresh",
1463 .data = &init_net.ipv4.sysctl_tcp_plb_cong_thresh,
1464 .maxlen = sizeof(int),
1465 .mode = 0644,
1466 .proc_handler = proc_dointvec_minmax,
1467 .extra1 = SYSCTL_ZERO,
1468 .extra2 = &tcp_plb_max_cong_thresh,
1469 },
1470 { }
1471};
1472
1473static __net_init int ipv4_sysctl_init_net(struct net *net)
1474{
1475 struct ctl_table *table;
1476
1477 table = ipv4_net_table;
1478 if (!net_eq(net, &init_net)) {
1479 int i;
1480
1481 table = kmemdup(table, sizeof(ipv4_net_table), GFP_KERNEL);
1482 if (!table)
1483 goto err_alloc;
1484
1485 for (i = 0; i < ARRAY_SIZE(ipv4_net_table) - 1; i++) {
1486 if (table[i].data) {
1487 /* Update the variables to point into
1488 * the current struct net
1489 */
1490 table[i].data += (void *)net - (void *)&init_net;
1491 } else {
1492 /* Entries without data pointer are global;
1493 * Make them read-only in non-init_net ns
1494 */
1495 table[i].mode &= ~0222;
1496 }
1497 }
1498 }
1499
1500 net->ipv4.ipv4_hdr = register_net_sysctl(net, "net/ipv4", table);
1501 if (!net->ipv4.ipv4_hdr)
1502 goto err_reg;
1503
1504 net->ipv4.sysctl_local_reserved_ports = kzalloc(65536 / 8, GFP_KERNEL);
1505 if (!net->ipv4.sysctl_local_reserved_ports)
1506 goto err_ports;
1507
1508 return 0;
1509
1510err_ports:
1511 unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
1512err_reg:
1513 if (!net_eq(net, &init_net))
1514 kfree(table);
1515err_alloc:
1516 return -ENOMEM;
1517}
1518
1519static __net_exit void ipv4_sysctl_exit_net(struct net *net)
1520{
1521 struct ctl_table *table;
1522
1523 kfree(net->ipv4.sysctl_local_reserved_ports);
1524 table = net->ipv4.ipv4_hdr->ctl_table_arg;
1525 unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
1526 kfree(table);
1527}
1528
1529static __net_initdata struct pernet_operations ipv4_sysctl_ops = {
1530 .init = ipv4_sysctl_init_net,
1531 .exit = ipv4_sysctl_exit_net,
1532};
1533
1534static __init int sysctl_ipv4_init(void)
1535{
1536 struct ctl_table_header *hdr;
1537
1538 hdr = register_net_sysctl(&init_net, "net/ipv4", ipv4_table);
1539 if (!hdr)
1540 return -ENOMEM;
1541
1542 if (register_pernet_subsys(&ipv4_sysctl_ops)) {
1543 unregister_net_sysctl_table(hdr);
1544 return -ENOMEM;
1545 }
1546
1547 return 0;
1548}
1549
1550__initcall(sysctl_ipv4_init);