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_app_win_max = 31;
29static int tcp_min_snd_mss_min = TCP_MIN_SND_MSS;
30static int tcp_min_snd_mss_max = 65535;
31static int ip_privileged_port_min;
32static int ip_privileged_port_max = 65535;
33static int ip_ttl_min = 1;
34static int ip_ttl_max = 255;
35static int tcp_syn_retries_min = 1;
36static int tcp_syn_retries_max = MAX_TCP_SYNCNT;
37static int tcp_syn_linear_timeouts_max = MAX_TCP_SYNCNT;
38static unsigned long ip_ping_group_range_min[] = { 0, 0 };
39static unsigned long ip_ping_group_range_max[] = { GID_T_MAX, GID_T_MAX };
40static u32 u32_max_div_HZ = UINT_MAX / HZ;
41static int one_day_secs = 24 * 3600;
42static u32 fib_multipath_hash_fields_all_mask __maybe_unused =
43 FIB_MULTIPATH_HASH_FIELD_ALL_MASK;
44static unsigned int tcp_child_ehash_entries_max = 16 * 1024 * 1024;
45static unsigned int udp_child_hash_entries_max = UDP_HTABLE_SIZE_MAX;
46static int tcp_plb_max_rounds = 31;
47static int tcp_plb_max_cong_thresh = 256;
48
49/* obsolete */
50static int sysctl_tcp_low_latency __read_mostly;
51
52/* Update system visible IP port range */
53static void set_local_port_range(struct net *net, unsigned int low, unsigned int high)
54{
55 bool same_parity = !((low ^ high) & 1);
56
57 if (same_parity && !net->ipv4.ip_local_ports.warned) {
58 net->ipv4.ip_local_ports.warned = true;
59 pr_err_ratelimited("ip_local_port_range: prefer different parity for start/end values.\n");
60 }
61 WRITE_ONCE(net->ipv4.ip_local_ports.range, high << 16 | low);
62}
63
64/* Validate changes from /proc interface. */
65static int ipv4_local_port_range(struct ctl_table *table, int write,
66 void *buffer, size_t *lenp, loff_t *ppos)
67{
68 struct net *net = table->data;
69 int ret;
70 int range[2];
71 struct ctl_table tmp = {
72 .data = &range,
73 .maxlen = sizeof(range),
74 .mode = table->mode,
75 .extra1 = &ip_local_port_range_min,
76 .extra2 = &ip_local_port_range_max,
77 };
78
79 inet_get_local_port_range(net, &range[0], &range[1]);
80
81 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
82
83 if (write && ret == 0) {
84 /* Ensure that the upper limit is not smaller than the lower,
85 * and that the lower does not encroach upon the privileged
86 * port limit.
87 */
88 if ((range[1] < range[0]) ||
89 (range[0] < READ_ONCE(net->ipv4.sysctl_ip_prot_sock)))
90 ret = -EINVAL;
91 else
92 set_local_port_range(net, range[0], range[1]);
93 }
94
95 return ret;
96}
97
98/* Validate changes from /proc interface. */
99static int ipv4_privileged_ports(struct ctl_table *table, int write,
100 void *buffer, size_t *lenp, loff_t *ppos)
101{
102 struct net *net = container_of(table->data, struct net,
103 ipv4.sysctl_ip_prot_sock);
104 int ret;
105 int pports;
106 int range[2];
107 struct ctl_table tmp = {
108 .data = &pports,
109 .maxlen = sizeof(pports),
110 .mode = table->mode,
111 .extra1 = &ip_privileged_port_min,
112 .extra2 = &ip_privileged_port_max,
113 };
114
115 pports = READ_ONCE(net->ipv4.sysctl_ip_prot_sock);
116
117 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
118
119 if (write && ret == 0) {
120 inet_get_local_port_range(net, &range[0], &range[1]);
121 /* Ensure that the local port range doesn't overlap with the
122 * privileged port range.
123 */
124 if (range[0] < pports)
125 ret = -EINVAL;
126 else
127 WRITE_ONCE(net->ipv4.sysctl_ip_prot_sock, pports);
128 }
129
130 return ret;
131}
132
133static void inet_get_ping_group_range_table(struct ctl_table *table, kgid_t *low, kgid_t *high)
134{
135 kgid_t *data = table->data;
136 struct net *net =
137 container_of(table->data, struct net, ipv4.ping_group_range.range);
138 unsigned int seq;
139 do {
140 seq = read_seqbegin(&net->ipv4.ping_group_range.lock);
141
142 *low = data[0];
143 *high = data[1];
144 } while (read_seqretry(&net->ipv4.ping_group_range.lock, seq));
145}
146
147/* Update system visible IP port range */
148static void set_ping_group_range(struct ctl_table *table, kgid_t low, kgid_t high)
149{
150 kgid_t *data = table->data;
151 struct net *net =
152 container_of(table->data, struct net, ipv4.ping_group_range.range);
153 write_seqlock(&net->ipv4.ping_group_range.lock);
154 data[0] = low;
155 data[1] = high;
156 write_sequnlock(&net->ipv4.ping_group_range.lock);
157}
158
159/* Validate changes from /proc interface. */
160static int ipv4_ping_group_range(struct ctl_table *table, int write,
161 void *buffer, size_t *lenp, loff_t *ppos)
162{
163 struct user_namespace *user_ns = current_user_ns();
164 int ret;
165 unsigned long urange[2];
166 kgid_t low, high;
167 struct ctl_table tmp = {
168 .data = &urange,
169 .maxlen = sizeof(urange),
170 .mode = table->mode,
171 .extra1 = &ip_ping_group_range_min,
172 .extra2 = &ip_ping_group_range_max,
173 };
174
175 inet_get_ping_group_range_table(table, &low, &high);
176 urange[0] = from_kgid_munged(user_ns, low);
177 urange[1] = from_kgid_munged(user_ns, high);
178 ret = proc_doulongvec_minmax(&tmp, write, buffer, lenp, ppos);
179
180 if (write && ret == 0) {
181 low = make_kgid(user_ns, urange[0]);
182 high = make_kgid(user_ns, urange[1]);
183 if (!gid_valid(low) || !gid_valid(high))
184 return -EINVAL;
185 if (urange[1] < urange[0] || gid_lt(high, low)) {
186 low = make_kgid(&init_user_ns, 1);
187 high = make_kgid(&init_user_ns, 0);
188 }
189 set_ping_group_range(table, low, high);
190 }
191
192 return ret;
193}
194
195static int ipv4_fwd_update_priority(struct ctl_table *table, int write,
196 void *buffer, size_t *lenp, loff_t *ppos)
197{
198 struct net *net;
199 int ret;
200
201 net = container_of(table->data, struct net,
202 ipv4.sysctl_ip_fwd_update_priority);
203 ret = proc_dou8vec_minmax(table, write, buffer, lenp, ppos);
204 if (write && ret == 0)
205 call_netevent_notifiers(NETEVENT_IPV4_FWD_UPDATE_PRIORITY_UPDATE,
206 net);
207
208 return ret;
209}
210
211static int proc_tcp_congestion_control(struct ctl_table *ctl, int write,
212 void *buffer, size_t *lenp, loff_t *ppos)
213{
214 struct net *net = container_of(ctl->data, struct net,
215 ipv4.tcp_congestion_control);
216 char val[TCP_CA_NAME_MAX];
217 struct ctl_table tbl = {
218 .data = val,
219 .maxlen = TCP_CA_NAME_MAX,
220 };
221 int ret;
222
223 tcp_get_default_congestion_control(net, val);
224
225 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
226 if (write && ret == 0)
227 ret = tcp_set_default_congestion_control(net, val);
228 return ret;
229}
230
231static int proc_tcp_available_congestion_control(struct ctl_table *ctl,
232 int write, void *buffer,
233 size_t *lenp, loff_t *ppos)
234{
235 struct ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX, };
236 int ret;
237
238 tbl.data = kmalloc(tbl.maxlen, GFP_USER);
239 if (!tbl.data)
240 return -ENOMEM;
241 tcp_get_available_congestion_control(tbl.data, TCP_CA_BUF_MAX);
242 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
243 kfree(tbl.data);
244 return ret;
245}
246
247static int proc_allowed_congestion_control(struct ctl_table *ctl,
248 int write, void *buffer,
249 size_t *lenp, loff_t *ppos)
250{
251 struct ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX };
252 int ret;
253
254 tbl.data = kmalloc(tbl.maxlen, GFP_USER);
255 if (!tbl.data)
256 return -ENOMEM;
257
258 tcp_get_allowed_congestion_control(tbl.data, tbl.maxlen);
259 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
260 if (write && ret == 0)
261 ret = tcp_set_allowed_congestion_control(tbl.data);
262 kfree(tbl.data);
263 return ret;
264}
265
266static int sscanf_key(char *buf, __le32 *key)
267{
268 u32 user_key[4];
269 int i, ret = 0;
270
271 if (sscanf(buf, "%x-%x-%x-%x", user_key, user_key + 1,
272 user_key + 2, user_key + 3) != 4) {
273 ret = -EINVAL;
274 } else {
275 for (i = 0; i < ARRAY_SIZE(user_key); i++)
276 key[i] = cpu_to_le32(user_key[i]);
277 }
278 pr_debug("proc TFO key set 0x%x-%x-%x-%x <- 0x%s: %u\n",
279 user_key[0], user_key[1], user_key[2], user_key[3], buf, ret);
280
281 return ret;
282}
283
284static int proc_tcp_fastopen_key(struct ctl_table *table, int write,
285 void *buffer, size_t *lenp, loff_t *ppos)
286{
287 struct net *net = container_of(table->data, struct net,
288 ipv4.sysctl_tcp_fastopen);
289 /* maxlen to print the list of keys in hex (*2), with dashes
290 * separating doublewords and a comma in between keys.
291 */
292 struct ctl_table tbl = { .maxlen = ((TCP_FASTOPEN_KEY_LENGTH *
293 2 * TCP_FASTOPEN_KEY_MAX) +
294 (TCP_FASTOPEN_KEY_MAX * 5)) };
295 u32 user_key[TCP_FASTOPEN_KEY_BUF_LENGTH / sizeof(u32)];
296 __le32 key[TCP_FASTOPEN_KEY_BUF_LENGTH / sizeof(__le32)];
297 char *backup_data;
298 int ret, i = 0, off = 0, n_keys;
299
300 tbl.data = kmalloc(tbl.maxlen, GFP_KERNEL);
301 if (!tbl.data)
302 return -ENOMEM;
303
304 n_keys = tcp_fastopen_get_cipher(net, NULL, (u64 *)key);
305 if (!n_keys) {
306 memset(&key[0], 0, TCP_FASTOPEN_KEY_LENGTH);
307 n_keys = 1;
308 }
309
310 for (i = 0; i < n_keys * 4; i++)
311 user_key[i] = le32_to_cpu(key[i]);
312
313 for (i = 0; i < n_keys; i++) {
314 off += snprintf(tbl.data + off, tbl.maxlen - off,
315 "%08x-%08x-%08x-%08x",
316 user_key[i * 4],
317 user_key[i * 4 + 1],
318 user_key[i * 4 + 2],
319 user_key[i * 4 + 3]);
320
321 if (WARN_ON_ONCE(off >= tbl.maxlen - 1))
322 break;
323
324 if (i + 1 < n_keys)
325 off += snprintf(tbl.data + off, tbl.maxlen - off, ",");
326 }
327
328 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
329
330 if (write && ret == 0) {
331 backup_data = strchr(tbl.data, ',');
332 if (backup_data) {
333 *backup_data = '\0';
334 backup_data++;
335 }
336 if (sscanf_key(tbl.data, key)) {
337 ret = -EINVAL;
338 goto bad_key;
339 }
340 if (backup_data) {
341 if (sscanf_key(backup_data, key + 4)) {
342 ret = -EINVAL;
343 goto bad_key;
344 }
345 }
346 tcp_fastopen_reset_cipher(net, NULL, key,
347 backup_data ? key + 4 : NULL);
348 }
349
350bad_key:
351 kfree(tbl.data);
352 return ret;
353}
354
355static int proc_tfo_blackhole_detect_timeout(struct ctl_table *table,
356 int write, void *buffer,
357 size_t *lenp, loff_t *ppos)
358{
359 struct net *net = container_of(table->data, struct net,
360 ipv4.sysctl_tcp_fastopen_blackhole_timeout);
361 int ret;
362
363 ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
364 if (write && ret == 0)
365 atomic_set(&net->ipv4.tfo_active_disable_times, 0);
366
367 return ret;
368}
369
370static int proc_tcp_available_ulp(struct ctl_table *ctl,
371 int write, void *buffer, size_t *lenp,
372 loff_t *ppos)
373{
374 struct ctl_table tbl = { .maxlen = TCP_ULP_BUF_MAX, };
375 int ret;
376
377 tbl.data = kmalloc(tbl.maxlen, GFP_USER);
378 if (!tbl.data)
379 return -ENOMEM;
380 tcp_get_available_ulp(tbl.data, TCP_ULP_BUF_MAX);
381 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
382 kfree(tbl.data);
383
384 return ret;
385}
386
387static int proc_tcp_ehash_entries(struct ctl_table *table, int write,
388 void *buffer, size_t *lenp, loff_t *ppos)
389{
390 struct net *net = container_of(table->data, struct net,
391 ipv4.sysctl_tcp_child_ehash_entries);
392 struct inet_hashinfo *hinfo = net->ipv4.tcp_death_row.hashinfo;
393 int tcp_ehash_entries;
394 struct ctl_table tbl;
395
396 tcp_ehash_entries = hinfo->ehash_mask + 1;
397
398 /* A negative number indicates that the child netns
399 * shares the global ehash.
400 */
401 if (!net_eq(net, &init_net) && !hinfo->pernet)
402 tcp_ehash_entries *= -1;
403
404 memset(&tbl, 0, sizeof(tbl));
405 tbl.data = &tcp_ehash_entries;
406 tbl.maxlen = sizeof(int);
407
408 return proc_dointvec(&tbl, write, buffer, lenp, ppos);
409}
410
411static int proc_udp_hash_entries(struct ctl_table *table, int write,
412 void *buffer, size_t *lenp, loff_t *ppos)
413{
414 struct net *net = container_of(table->data, struct net,
415 ipv4.sysctl_udp_child_hash_entries);
416 int udp_hash_entries;
417 struct ctl_table tbl;
418
419 udp_hash_entries = net->ipv4.udp_table->mask + 1;
420
421 /* A negative number indicates that the child netns
422 * shares the global udp_table.
423 */
424 if (!net_eq(net, &init_net) && net->ipv4.udp_table == &udp_table)
425 udp_hash_entries *= -1;
426
427 memset(&tbl, 0, sizeof(tbl));
428 tbl.data = &udp_hash_entries;
429 tbl.maxlen = sizeof(int);
430
431 return proc_dointvec(&tbl, write, buffer, lenp, ppos);
432}
433
434#ifdef CONFIG_IP_ROUTE_MULTIPATH
435static int proc_fib_multipath_hash_policy(struct ctl_table *table, int write,
436 void *buffer, size_t *lenp,
437 loff_t *ppos)
438{
439 struct net *net = container_of(table->data, struct net,
440 ipv4.sysctl_fib_multipath_hash_policy);
441 int ret;
442
443 ret = proc_dou8vec_minmax(table, write, buffer, lenp, ppos);
444 if (write && ret == 0)
445 call_netevent_notifiers(NETEVENT_IPV4_MPATH_HASH_UPDATE, net);
446
447 return ret;
448}
449
450static int proc_fib_multipath_hash_fields(struct ctl_table *table, int write,
451 void *buffer, size_t *lenp,
452 loff_t *ppos)
453{
454 struct net *net;
455 int ret;
456
457 net = container_of(table->data, struct net,
458 ipv4.sysctl_fib_multipath_hash_fields);
459 ret = proc_douintvec_minmax(table, write, buffer, lenp, ppos);
460 if (write && ret == 0)
461 call_netevent_notifiers(NETEVENT_IPV4_MPATH_HASH_UPDATE, net);
462
463 return ret;
464}
465#endif
466
467static struct ctl_table ipv4_table[] = {
468 {
469 .procname = "tcp_max_orphans",
470 .data = &sysctl_tcp_max_orphans,
471 .maxlen = sizeof(int),
472 .mode = 0644,
473 .proc_handler = proc_dointvec
474 },
475 {
476 .procname = "inet_peer_threshold",
477 .data = &inet_peer_threshold,
478 .maxlen = sizeof(int),
479 .mode = 0644,
480 .proc_handler = proc_dointvec
481 },
482 {
483 .procname = "inet_peer_minttl",
484 .data = &inet_peer_minttl,
485 .maxlen = sizeof(int),
486 .mode = 0644,
487 .proc_handler = proc_dointvec_jiffies,
488 },
489 {
490 .procname = "inet_peer_maxttl",
491 .data = &inet_peer_maxttl,
492 .maxlen = sizeof(int),
493 .mode = 0644,
494 .proc_handler = proc_dointvec_jiffies,
495 },
496 {
497 .procname = "tcp_mem",
498 .maxlen = sizeof(sysctl_tcp_mem),
499 .data = &sysctl_tcp_mem,
500 .mode = 0644,
501 .proc_handler = proc_doulongvec_minmax,
502 },
503 {
504 .procname = "tcp_low_latency",
505 .data = &sysctl_tcp_low_latency,
506 .maxlen = sizeof(int),
507 .mode = 0644,
508 .proc_handler = proc_dointvec
509 },
510#ifdef CONFIG_NETLABEL
511 {
512 .procname = "cipso_cache_enable",
513 .data = &cipso_v4_cache_enabled,
514 .maxlen = sizeof(int),
515 .mode = 0644,
516 .proc_handler = proc_dointvec,
517 },
518 {
519 .procname = "cipso_cache_bucket_size",
520 .data = &cipso_v4_cache_bucketsize,
521 .maxlen = sizeof(int),
522 .mode = 0644,
523 .proc_handler = proc_dointvec,
524 },
525 {
526 .procname = "cipso_rbm_optfmt",
527 .data = &cipso_v4_rbm_optfmt,
528 .maxlen = sizeof(int),
529 .mode = 0644,
530 .proc_handler = proc_dointvec,
531 },
532 {
533 .procname = "cipso_rbm_strictvalid",
534 .data = &cipso_v4_rbm_strictvalid,
535 .maxlen = sizeof(int),
536 .mode = 0644,
537 .proc_handler = proc_dointvec,
538 },
539#endif /* CONFIG_NETLABEL */
540 {
541 .procname = "tcp_available_ulp",
542 .maxlen = TCP_ULP_BUF_MAX,
543 .mode = 0444,
544 .proc_handler = proc_tcp_available_ulp,
545 },
546 {
547 .procname = "icmp_msgs_per_sec",
548 .data = &sysctl_icmp_msgs_per_sec,
549 .maxlen = sizeof(int),
550 .mode = 0644,
551 .proc_handler = proc_dointvec_minmax,
552 .extra1 = SYSCTL_ZERO,
553 },
554 {
555 .procname = "icmp_msgs_burst",
556 .data = &sysctl_icmp_msgs_burst,
557 .maxlen = sizeof(int),
558 .mode = 0644,
559 .proc_handler = proc_dointvec_minmax,
560 .extra1 = SYSCTL_ZERO,
561 },
562 {
563 .procname = "udp_mem",
564 .data = &sysctl_udp_mem,
565 .maxlen = sizeof(sysctl_udp_mem),
566 .mode = 0644,
567 .proc_handler = proc_doulongvec_minmax,
568 },
569 {
570 .procname = "fib_sync_mem",
571 .data = &sysctl_fib_sync_mem,
572 .maxlen = sizeof(sysctl_fib_sync_mem),
573 .mode = 0644,
574 .proc_handler = proc_douintvec_minmax,
575 .extra1 = &sysctl_fib_sync_mem_min,
576 .extra2 = &sysctl_fib_sync_mem_max,
577 },
578 { }
579};
580
581static struct ctl_table ipv4_net_table[] = {
582 {
583 .procname = "tcp_max_tw_buckets",
584 .data = &init_net.ipv4.tcp_death_row.sysctl_max_tw_buckets,
585 .maxlen = sizeof(int),
586 .mode = 0644,
587 .proc_handler = proc_dointvec
588 },
589 {
590 .procname = "icmp_echo_ignore_all",
591 .data = &init_net.ipv4.sysctl_icmp_echo_ignore_all,
592 .maxlen = sizeof(u8),
593 .mode = 0644,
594 .proc_handler = proc_dou8vec_minmax,
595 .extra1 = SYSCTL_ZERO,
596 .extra2 = SYSCTL_ONE
597 },
598 {
599 .procname = "icmp_echo_enable_probe",
600 .data = &init_net.ipv4.sysctl_icmp_echo_enable_probe,
601 .maxlen = sizeof(u8),
602 .mode = 0644,
603 .proc_handler = proc_dou8vec_minmax,
604 .extra1 = SYSCTL_ZERO,
605 .extra2 = SYSCTL_ONE
606 },
607 {
608 .procname = "icmp_echo_ignore_broadcasts",
609 .data = &init_net.ipv4.sysctl_icmp_echo_ignore_broadcasts,
610 .maxlen = sizeof(u8),
611 .mode = 0644,
612 .proc_handler = proc_dou8vec_minmax,
613 .extra1 = SYSCTL_ZERO,
614 .extra2 = SYSCTL_ONE
615 },
616 {
617 .procname = "icmp_ignore_bogus_error_responses",
618 .data = &init_net.ipv4.sysctl_icmp_ignore_bogus_error_responses,
619 .maxlen = sizeof(u8),
620 .mode = 0644,
621 .proc_handler = proc_dou8vec_minmax,
622 .extra1 = SYSCTL_ZERO,
623 .extra2 = SYSCTL_ONE
624 },
625 {
626 .procname = "icmp_errors_use_inbound_ifaddr",
627 .data = &init_net.ipv4.sysctl_icmp_errors_use_inbound_ifaddr,
628 .maxlen = sizeof(u8),
629 .mode = 0644,
630 .proc_handler = proc_dou8vec_minmax,
631 .extra1 = SYSCTL_ZERO,
632 .extra2 = SYSCTL_ONE
633 },
634 {
635 .procname = "icmp_ratelimit",
636 .data = &init_net.ipv4.sysctl_icmp_ratelimit,
637 .maxlen = sizeof(int),
638 .mode = 0644,
639 .proc_handler = proc_dointvec_ms_jiffies,
640 },
641 {
642 .procname = "icmp_ratemask",
643 .data = &init_net.ipv4.sysctl_icmp_ratemask,
644 .maxlen = sizeof(int),
645 .mode = 0644,
646 .proc_handler = proc_dointvec
647 },
648 {
649 .procname = "ping_group_range",
650 .data = &init_net.ipv4.ping_group_range.range,
651 .maxlen = sizeof(gid_t)*2,
652 .mode = 0644,
653 .proc_handler = ipv4_ping_group_range,
654 },
655#ifdef CONFIG_NET_L3_MASTER_DEV
656 {
657 .procname = "raw_l3mdev_accept",
658 .data = &init_net.ipv4.sysctl_raw_l3mdev_accept,
659 .maxlen = sizeof(u8),
660 .mode = 0644,
661 .proc_handler = proc_dou8vec_minmax,
662 .extra1 = SYSCTL_ZERO,
663 .extra2 = SYSCTL_ONE,
664 },
665#endif
666 {
667 .procname = "tcp_ecn",
668 .data = &init_net.ipv4.sysctl_tcp_ecn,
669 .maxlen = sizeof(u8),
670 .mode = 0644,
671 .proc_handler = proc_dou8vec_minmax,
672 .extra1 = SYSCTL_ZERO,
673 .extra2 = SYSCTL_TWO,
674 },
675 {
676 .procname = "tcp_ecn_fallback",
677 .data = &init_net.ipv4.sysctl_tcp_ecn_fallback,
678 .maxlen = sizeof(u8),
679 .mode = 0644,
680 .proc_handler = proc_dou8vec_minmax,
681 .extra1 = SYSCTL_ZERO,
682 .extra2 = SYSCTL_ONE,
683 },
684 {
685 .procname = "ip_dynaddr",
686 .data = &init_net.ipv4.sysctl_ip_dynaddr,
687 .maxlen = sizeof(u8),
688 .mode = 0644,
689 .proc_handler = proc_dou8vec_minmax,
690 },
691 {
692 .procname = "ip_early_demux",
693 .data = &init_net.ipv4.sysctl_ip_early_demux,
694 .maxlen = sizeof(u8),
695 .mode = 0644,
696 .proc_handler = proc_dou8vec_minmax,
697 },
698 {
699 .procname = "udp_early_demux",
700 .data = &init_net.ipv4.sysctl_udp_early_demux,
701 .maxlen = sizeof(u8),
702 .mode = 0644,
703 .proc_handler = proc_dou8vec_minmax,
704 },
705 {
706 .procname = "tcp_early_demux",
707 .data = &init_net.ipv4.sysctl_tcp_early_demux,
708 .maxlen = sizeof(u8),
709 .mode = 0644,
710 .proc_handler = proc_dou8vec_minmax,
711 },
712 {
713 .procname = "nexthop_compat_mode",
714 .data = &init_net.ipv4.sysctl_nexthop_compat_mode,
715 .maxlen = sizeof(u8),
716 .mode = 0644,
717 .proc_handler = proc_dou8vec_minmax,
718 .extra1 = SYSCTL_ZERO,
719 .extra2 = SYSCTL_ONE,
720 },
721 {
722 .procname = "ip_default_ttl",
723 .data = &init_net.ipv4.sysctl_ip_default_ttl,
724 .maxlen = sizeof(u8),
725 .mode = 0644,
726 .proc_handler = proc_dou8vec_minmax,
727 .extra1 = &ip_ttl_min,
728 .extra2 = &ip_ttl_max,
729 },
730 {
731 .procname = "ip_local_port_range",
732 .maxlen = 0,
733 .data = &init_net,
734 .mode = 0644,
735 .proc_handler = ipv4_local_port_range,
736 },
737 {
738 .procname = "ip_local_reserved_ports",
739 .data = &init_net.ipv4.sysctl_local_reserved_ports,
740 .maxlen = 65536,
741 .mode = 0644,
742 .proc_handler = proc_do_large_bitmap,
743 },
744 {
745 .procname = "ip_no_pmtu_disc",
746 .data = &init_net.ipv4.sysctl_ip_no_pmtu_disc,
747 .maxlen = sizeof(u8),
748 .mode = 0644,
749 .proc_handler = proc_dou8vec_minmax,
750 },
751 {
752 .procname = "ip_forward_use_pmtu",
753 .data = &init_net.ipv4.sysctl_ip_fwd_use_pmtu,
754 .maxlen = sizeof(u8),
755 .mode = 0644,
756 .proc_handler = proc_dou8vec_minmax,
757 },
758 {
759 .procname = "ip_forward_update_priority",
760 .data = &init_net.ipv4.sysctl_ip_fwd_update_priority,
761 .maxlen = sizeof(u8),
762 .mode = 0644,
763 .proc_handler = ipv4_fwd_update_priority,
764 .extra1 = SYSCTL_ZERO,
765 .extra2 = SYSCTL_ONE,
766 },
767 {
768 .procname = "ip_nonlocal_bind",
769 .data = &init_net.ipv4.sysctl_ip_nonlocal_bind,
770 .maxlen = sizeof(u8),
771 .mode = 0644,
772 .proc_handler = proc_dou8vec_minmax,
773 },
774 {
775 .procname = "ip_autobind_reuse",
776 .data = &init_net.ipv4.sysctl_ip_autobind_reuse,
777 .maxlen = sizeof(u8),
778 .mode = 0644,
779 .proc_handler = proc_dou8vec_minmax,
780 .extra1 = SYSCTL_ZERO,
781 .extra2 = SYSCTL_ONE,
782 },
783 {
784 .procname = "fwmark_reflect",
785 .data = &init_net.ipv4.sysctl_fwmark_reflect,
786 .maxlen = sizeof(u8),
787 .mode = 0644,
788 .proc_handler = proc_dou8vec_minmax,
789 },
790 {
791 .procname = "tcp_fwmark_accept",
792 .data = &init_net.ipv4.sysctl_tcp_fwmark_accept,
793 .maxlen = sizeof(u8),
794 .mode = 0644,
795 .proc_handler = proc_dou8vec_minmax,
796 },
797#ifdef CONFIG_NET_L3_MASTER_DEV
798 {
799 .procname = "tcp_l3mdev_accept",
800 .data = &init_net.ipv4.sysctl_tcp_l3mdev_accept,
801 .maxlen = sizeof(u8),
802 .mode = 0644,
803 .proc_handler = proc_dou8vec_minmax,
804 .extra1 = SYSCTL_ZERO,
805 .extra2 = SYSCTL_ONE,
806 },
807#endif
808 {
809 .procname = "tcp_mtu_probing",
810 .data = &init_net.ipv4.sysctl_tcp_mtu_probing,
811 .maxlen = sizeof(u8),
812 .mode = 0644,
813 .proc_handler = proc_dou8vec_minmax,
814 },
815 {
816 .procname = "tcp_base_mss",
817 .data = &init_net.ipv4.sysctl_tcp_base_mss,
818 .maxlen = sizeof(int),
819 .mode = 0644,
820 .proc_handler = proc_dointvec,
821 },
822 {
823 .procname = "tcp_min_snd_mss",
824 .data = &init_net.ipv4.sysctl_tcp_min_snd_mss,
825 .maxlen = sizeof(int),
826 .mode = 0644,
827 .proc_handler = proc_dointvec_minmax,
828 .extra1 = &tcp_min_snd_mss_min,
829 .extra2 = &tcp_min_snd_mss_max,
830 },
831 {
832 .procname = "tcp_mtu_probe_floor",
833 .data = &init_net.ipv4.sysctl_tcp_mtu_probe_floor,
834 .maxlen = sizeof(int),
835 .mode = 0644,
836 .proc_handler = proc_dointvec_minmax,
837 .extra1 = &tcp_min_snd_mss_min,
838 .extra2 = &tcp_min_snd_mss_max,
839 },
840 {
841 .procname = "tcp_probe_threshold",
842 .data = &init_net.ipv4.sysctl_tcp_probe_threshold,
843 .maxlen = sizeof(int),
844 .mode = 0644,
845 .proc_handler = proc_dointvec,
846 },
847 {
848 .procname = "tcp_probe_interval",
849 .data = &init_net.ipv4.sysctl_tcp_probe_interval,
850 .maxlen = sizeof(u32),
851 .mode = 0644,
852 .proc_handler = proc_douintvec_minmax,
853 .extra2 = &u32_max_div_HZ,
854 },
855 {
856 .procname = "igmp_link_local_mcast_reports",
857 .data = &init_net.ipv4.sysctl_igmp_llm_reports,
858 .maxlen = sizeof(u8),
859 .mode = 0644,
860 .proc_handler = proc_dou8vec_minmax,
861 },
862 {
863 .procname = "igmp_max_memberships",
864 .data = &init_net.ipv4.sysctl_igmp_max_memberships,
865 .maxlen = sizeof(int),
866 .mode = 0644,
867 .proc_handler = proc_dointvec
868 },
869 {
870 .procname = "igmp_max_msf",
871 .data = &init_net.ipv4.sysctl_igmp_max_msf,
872 .maxlen = sizeof(int),
873 .mode = 0644,
874 .proc_handler = proc_dointvec
875 },
876#ifdef CONFIG_IP_MULTICAST
877 {
878 .procname = "igmp_qrv",
879 .data = &init_net.ipv4.sysctl_igmp_qrv,
880 .maxlen = sizeof(int),
881 .mode = 0644,
882 .proc_handler = proc_dointvec_minmax,
883 .extra1 = SYSCTL_ONE
884 },
885#endif
886 {
887 .procname = "tcp_congestion_control",
888 .data = &init_net.ipv4.tcp_congestion_control,
889 .mode = 0644,
890 .maxlen = TCP_CA_NAME_MAX,
891 .proc_handler = proc_tcp_congestion_control,
892 },
893 {
894 .procname = "tcp_available_congestion_control",
895 .maxlen = TCP_CA_BUF_MAX,
896 .mode = 0444,
897 .proc_handler = proc_tcp_available_congestion_control,
898 },
899 {
900 .procname = "tcp_allowed_congestion_control",
901 .maxlen = TCP_CA_BUF_MAX,
902 .mode = 0644,
903 .proc_handler = proc_allowed_congestion_control,
904 },
905 {
906 .procname = "tcp_keepalive_time",
907 .data = &init_net.ipv4.sysctl_tcp_keepalive_time,
908 .maxlen = sizeof(int),
909 .mode = 0644,
910 .proc_handler = proc_dointvec_jiffies,
911 },
912 {
913 .procname = "tcp_keepalive_probes",
914 .data = &init_net.ipv4.sysctl_tcp_keepalive_probes,
915 .maxlen = sizeof(u8),
916 .mode = 0644,
917 .proc_handler = proc_dou8vec_minmax,
918 },
919 {
920 .procname = "tcp_keepalive_intvl",
921 .data = &init_net.ipv4.sysctl_tcp_keepalive_intvl,
922 .maxlen = sizeof(int),
923 .mode = 0644,
924 .proc_handler = proc_dointvec_jiffies,
925 },
926 {
927 .procname = "tcp_syn_retries",
928 .data = &init_net.ipv4.sysctl_tcp_syn_retries,
929 .maxlen = sizeof(u8),
930 .mode = 0644,
931 .proc_handler = proc_dou8vec_minmax,
932 .extra1 = &tcp_syn_retries_min,
933 .extra2 = &tcp_syn_retries_max
934 },
935 {
936 .procname = "tcp_synack_retries",
937 .data = &init_net.ipv4.sysctl_tcp_synack_retries,
938 .maxlen = sizeof(u8),
939 .mode = 0644,
940 .proc_handler = proc_dou8vec_minmax,
941 },
942#ifdef CONFIG_SYN_COOKIES
943 {
944 .procname = "tcp_syncookies",
945 .data = &init_net.ipv4.sysctl_tcp_syncookies,
946 .maxlen = sizeof(u8),
947 .mode = 0644,
948 .proc_handler = proc_dou8vec_minmax,
949 },
950#endif
951 {
952 .procname = "tcp_migrate_req",
953 .data = &init_net.ipv4.sysctl_tcp_migrate_req,
954 .maxlen = sizeof(u8),
955 .mode = 0644,
956 .proc_handler = proc_dou8vec_minmax,
957 .extra1 = SYSCTL_ZERO,
958 .extra2 = SYSCTL_ONE
959 },
960 {
961 .procname = "tcp_reordering",
962 .data = &init_net.ipv4.sysctl_tcp_reordering,
963 .maxlen = sizeof(int),
964 .mode = 0644,
965 .proc_handler = proc_dointvec
966 },
967 {
968 .procname = "tcp_retries1",
969 .data = &init_net.ipv4.sysctl_tcp_retries1,
970 .maxlen = sizeof(u8),
971 .mode = 0644,
972 .proc_handler = proc_dou8vec_minmax,
973 .extra2 = &tcp_retr1_max
974 },
975 {
976 .procname = "tcp_retries2",
977 .data = &init_net.ipv4.sysctl_tcp_retries2,
978 .maxlen = sizeof(u8),
979 .mode = 0644,
980 .proc_handler = proc_dou8vec_minmax,
981 },
982 {
983 .procname = "tcp_orphan_retries",
984 .data = &init_net.ipv4.sysctl_tcp_orphan_retries,
985 .maxlen = sizeof(u8),
986 .mode = 0644,
987 .proc_handler = proc_dou8vec_minmax,
988 },
989 {
990 .procname = "tcp_fin_timeout",
991 .data = &init_net.ipv4.sysctl_tcp_fin_timeout,
992 .maxlen = sizeof(int),
993 .mode = 0644,
994 .proc_handler = proc_dointvec_jiffies,
995 },
996 {
997 .procname = "tcp_notsent_lowat",
998 .data = &init_net.ipv4.sysctl_tcp_notsent_lowat,
999 .maxlen = sizeof(unsigned int),
1000 .mode = 0644,
1001 .proc_handler = proc_douintvec,
1002 },
1003 {
1004 .procname = "tcp_tw_reuse",
1005 .data = &init_net.ipv4.sysctl_tcp_tw_reuse,
1006 .maxlen = sizeof(u8),
1007 .mode = 0644,
1008 .proc_handler = proc_dou8vec_minmax,
1009 .extra1 = SYSCTL_ZERO,
1010 .extra2 = SYSCTL_TWO,
1011 },
1012 {
1013 .procname = "tcp_max_syn_backlog",
1014 .data = &init_net.ipv4.sysctl_max_syn_backlog,
1015 .maxlen = sizeof(int),
1016 .mode = 0644,
1017 .proc_handler = proc_dointvec
1018 },
1019 {
1020 .procname = "tcp_fastopen",
1021 .data = &init_net.ipv4.sysctl_tcp_fastopen,
1022 .maxlen = sizeof(int),
1023 .mode = 0644,
1024 .proc_handler = proc_dointvec,
1025 },
1026 {
1027 .procname = "tcp_fastopen_key",
1028 .mode = 0600,
1029 .data = &init_net.ipv4.sysctl_tcp_fastopen,
1030 /* maxlen to print the list of keys in hex (*2), with dashes
1031 * separating doublewords and a comma in between keys.
1032 */
1033 .maxlen = ((TCP_FASTOPEN_KEY_LENGTH *
1034 2 * TCP_FASTOPEN_KEY_MAX) +
1035 (TCP_FASTOPEN_KEY_MAX * 5)),
1036 .proc_handler = proc_tcp_fastopen_key,
1037 },
1038 {
1039 .procname = "tcp_fastopen_blackhole_timeout_sec",
1040 .data = &init_net.ipv4.sysctl_tcp_fastopen_blackhole_timeout,
1041 .maxlen = sizeof(int),
1042 .mode = 0644,
1043 .proc_handler = proc_tfo_blackhole_detect_timeout,
1044 .extra1 = SYSCTL_ZERO,
1045 },
1046#ifdef CONFIG_IP_ROUTE_MULTIPATH
1047 {
1048 .procname = "fib_multipath_use_neigh",
1049 .data = &init_net.ipv4.sysctl_fib_multipath_use_neigh,
1050 .maxlen = sizeof(u8),
1051 .mode = 0644,
1052 .proc_handler = proc_dou8vec_minmax,
1053 .extra1 = SYSCTL_ZERO,
1054 .extra2 = SYSCTL_ONE,
1055 },
1056 {
1057 .procname = "fib_multipath_hash_policy",
1058 .data = &init_net.ipv4.sysctl_fib_multipath_hash_policy,
1059 .maxlen = sizeof(u8),
1060 .mode = 0644,
1061 .proc_handler = proc_fib_multipath_hash_policy,
1062 .extra1 = SYSCTL_ZERO,
1063 .extra2 = SYSCTL_THREE,
1064 },
1065 {
1066 .procname = "fib_multipath_hash_fields",
1067 .data = &init_net.ipv4.sysctl_fib_multipath_hash_fields,
1068 .maxlen = sizeof(u32),
1069 .mode = 0644,
1070 .proc_handler = proc_fib_multipath_hash_fields,
1071 .extra1 = SYSCTL_ONE,
1072 .extra2 = &fib_multipath_hash_fields_all_mask,
1073 },
1074#endif
1075 {
1076 .procname = "ip_unprivileged_port_start",
1077 .maxlen = sizeof(int),
1078 .data = &init_net.ipv4.sysctl_ip_prot_sock,
1079 .mode = 0644,
1080 .proc_handler = ipv4_privileged_ports,
1081 },
1082#ifdef CONFIG_NET_L3_MASTER_DEV
1083 {
1084 .procname = "udp_l3mdev_accept",
1085 .data = &init_net.ipv4.sysctl_udp_l3mdev_accept,
1086 .maxlen = sizeof(u8),
1087 .mode = 0644,
1088 .proc_handler = proc_dou8vec_minmax,
1089 .extra1 = SYSCTL_ZERO,
1090 .extra2 = SYSCTL_ONE,
1091 },
1092#endif
1093 {
1094 .procname = "tcp_sack",
1095 .data = &init_net.ipv4.sysctl_tcp_sack,
1096 .maxlen = sizeof(u8),
1097 .mode = 0644,
1098 .proc_handler = proc_dou8vec_minmax,
1099 },
1100 {
1101 .procname = "tcp_window_scaling",
1102 .data = &init_net.ipv4.sysctl_tcp_window_scaling,
1103 .maxlen = sizeof(u8),
1104 .mode = 0644,
1105 .proc_handler = proc_dou8vec_minmax,
1106 },
1107 {
1108 .procname = "tcp_timestamps",
1109 .data = &init_net.ipv4.sysctl_tcp_timestamps,
1110 .maxlen = sizeof(u8),
1111 .mode = 0644,
1112 .proc_handler = proc_dou8vec_minmax,
1113 },
1114 {
1115 .procname = "tcp_early_retrans",
1116 .data = &init_net.ipv4.sysctl_tcp_early_retrans,
1117 .maxlen = sizeof(u8),
1118 .mode = 0644,
1119 .proc_handler = proc_dou8vec_minmax,
1120 .extra1 = SYSCTL_ZERO,
1121 .extra2 = SYSCTL_FOUR,
1122 },
1123 {
1124 .procname = "tcp_recovery",
1125 .data = &init_net.ipv4.sysctl_tcp_recovery,
1126 .maxlen = sizeof(u8),
1127 .mode = 0644,
1128 .proc_handler = proc_dou8vec_minmax,
1129 },
1130 {
1131 .procname = "tcp_thin_linear_timeouts",
1132 .data = &init_net.ipv4.sysctl_tcp_thin_linear_timeouts,
1133 .maxlen = sizeof(u8),
1134 .mode = 0644,
1135 .proc_handler = proc_dou8vec_minmax,
1136 },
1137 {
1138 .procname = "tcp_slow_start_after_idle",
1139 .data = &init_net.ipv4.sysctl_tcp_slow_start_after_idle,
1140 .maxlen = sizeof(u8),
1141 .mode = 0644,
1142 .proc_handler = proc_dou8vec_minmax,
1143 },
1144 {
1145 .procname = "tcp_retrans_collapse",
1146 .data = &init_net.ipv4.sysctl_tcp_retrans_collapse,
1147 .maxlen = sizeof(u8),
1148 .mode = 0644,
1149 .proc_handler = proc_dou8vec_minmax,
1150 },
1151 {
1152 .procname = "tcp_stdurg",
1153 .data = &init_net.ipv4.sysctl_tcp_stdurg,
1154 .maxlen = sizeof(u8),
1155 .mode = 0644,
1156 .proc_handler = proc_dou8vec_minmax,
1157 },
1158 {
1159 .procname = "tcp_rfc1337",
1160 .data = &init_net.ipv4.sysctl_tcp_rfc1337,
1161 .maxlen = sizeof(u8),
1162 .mode = 0644,
1163 .proc_handler = proc_dou8vec_minmax,
1164 },
1165 {
1166 .procname = "tcp_abort_on_overflow",
1167 .data = &init_net.ipv4.sysctl_tcp_abort_on_overflow,
1168 .maxlen = sizeof(u8),
1169 .mode = 0644,
1170 .proc_handler = proc_dou8vec_minmax,
1171 },
1172 {
1173 .procname = "tcp_fack",
1174 .data = &init_net.ipv4.sysctl_tcp_fack,
1175 .maxlen = sizeof(u8),
1176 .mode = 0644,
1177 .proc_handler = proc_dou8vec_minmax,
1178 },
1179 {
1180 .procname = "tcp_max_reordering",
1181 .data = &init_net.ipv4.sysctl_tcp_max_reordering,
1182 .maxlen = sizeof(int),
1183 .mode = 0644,
1184 .proc_handler = proc_dointvec
1185 },
1186 {
1187 .procname = "tcp_dsack",
1188 .data = &init_net.ipv4.sysctl_tcp_dsack,
1189 .maxlen = sizeof(u8),
1190 .mode = 0644,
1191 .proc_handler = proc_dou8vec_minmax,
1192 },
1193 {
1194 .procname = "tcp_app_win",
1195 .data = &init_net.ipv4.sysctl_tcp_app_win,
1196 .maxlen = sizeof(u8),
1197 .mode = 0644,
1198 .proc_handler = proc_dou8vec_minmax,
1199 .extra1 = SYSCTL_ZERO,
1200 .extra2 = &tcp_app_win_max,
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_backlog_ack_defer",
1367 .data = &init_net.ipv4.sysctl_tcp_backlog_ack_defer,
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_reflect_tos",
1376 .data = &init_net.ipv4.sysctl_tcp_reflect_tos,
1377 .maxlen = sizeof(u8),
1378 .mode = 0644,
1379 .proc_handler = proc_dou8vec_minmax,
1380 .extra1 = SYSCTL_ZERO,
1381 .extra2 = SYSCTL_ONE,
1382 },
1383 {
1384 .procname = "tcp_ehash_entries",
1385 .data = &init_net.ipv4.sysctl_tcp_child_ehash_entries,
1386 .mode = 0444,
1387 .proc_handler = proc_tcp_ehash_entries,
1388 },
1389 {
1390 .procname = "tcp_child_ehash_entries",
1391 .data = &init_net.ipv4.sysctl_tcp_child_ehash_entries,
1392 .maxlen = sizeof(unsigned int),
1393 .mode = 0644,
1394 .proc_handler = proc_douintvec_minmax,
1395 .extra1 = SYSCTL_ZERO,
1396 .extra2 = &tcp_child_ehash_entries_max,
1397 },
1398 {
1399 .procname = "udp_hash_entries",
1400 .data = &init_net.ipv4.sysctl_udp_child_hash_entries,
1401 .mode = 0444,
1402 .proc_handler = proc_udp_hash_entries,
1403 },
1404 {
1405 .procname = "udp_child_hash_entries",
1406 .data = &init_net.ipv4.sysctl_udp_child_hash_entries,
1407 .maxlen = sizeof(unsigned int),
1408 .mode = 0644,
1409 .proc_handler = proc_douintvec_minmax,
1410 .extra1 = SYSCTL_ZERO,
1411 .extra2 = &udp_child_hash_entries_max,
1412 },
1413 {
1414 .procname = "udp_rmem_min",
1415 .data = &init_net.ipv4.sysctl_udp_rmem_min,
1416 .maxlen = sizeof(init_net.ipv4.sysctl_udp_rmem_min),
1417 .mode = 0644,
1418 .proc_handler = proc_dointvec_minmax,
1419 .extra1 = SYSCTL_ONE
1420 },
1421 {
1422 .procname = "udp_wmem_min",
1423 .data = &init_net.ipv4.sysctl_udp_wmem_min,
1424 .maxlen = sizeof(init_net.ipv4.sysctl_udp_wmem_min),
1425 .mode = 0644,
1426 .proc_handler = proc_dointvec_minmax,
1427 .extra1 = SYSCTL_ONE
1428 },
1429 {
1430 .procname = "fib_notify_on_flag_change",
1431 .data = &init_net.ipv4.sysctl_fib_notify_on_flag_change,
1432 .maxlen = sizeof(u8),
1433 .mode = 0644,
1434 .proc_handler = proc_dou8vec_minmax,
1435 .extra1 = SYSCTL_ZERO,
1436 .extra2 = SYSCTL_TWO,
1437 },
1438 {
1439 .procname = "tcp_plb_enabled",
1440 .data = &init_net.ipv4.sysctl_tcp_plb_enabled,
1441 .maxlen = sizeof(u8),
1442 .mode = 0644,
1443 .proc_handler = proc_dou8vec_minmax,
1444 .extra1 = SYSCTL_ZERO,
1445 .extra2 = SYSCTL_ONE,
1446 },
1447 {
1448 .procname = "tcp_plb_idle_rehash_rounds",
1449 .data = &init_net.ipv4.sysctl_tcp_plb_idle_rehash_rounds,
1450 .maxlen = sizeof(u8),
1451 .mode = 0644,
1452 .proc_handler = proc_dou8vec_minmax,
1453 .extra2 = &tcp_plb_max_rounds,
1454 },
1455 {
1456 .procname = "tcp_plb_rehash_rounds",
1457 .data = &init_net.ipv4.sysctl_tcp_plb_rehash_rounds,
1458 .maxlen = sizeof(u8),
1459 .mode = 0644,
1460 .proc_handler = proc_dou8vec_minmax,
1461 .extra2 = &tcp_plb_max_rounds,
1462 },
1463 {
1464 .procname = "tcp_plb_suspend_rto_sec",
1465 .data = &init_net.ipv4.sysctl_tcp_plb_suspend_rto_sec,
1466 .maxlen = sizeof(u8),
1467 .mode = 0644,
1468 .proc_handler = proc_dou8vec_minmax,
1469 },
1470 {
1471 .procname = "tcp_plb_cong_thresh",
1472 .data = &init_net.ipv4.sysctl_tcp_plb_cong_thresh,
1473 .maxlen = sizeof(int),
1474 .mode = 0644,
1475 .proc_handler = proc_dointvec_minmax,
1476 .extra1 = SYSCTL_ZERO,
1477 .extra2 = &tcp_plb_max_cong_thresh,
1478 },
1479 {
1480 .procname = "tcp_syn_linear_timeouts",
1481 .data = &init_net.ipv4.sysctl_tcp_syn_linear_timeouts,
1482 .maxlen = sizeof(u8),
1483 .mode = 0644,
1484 .proc_handler = proc_dou8vec_minmax,
1485 .extra1 = SYSCTL_ZERO,
1486 .extra2 = &tcp_syn_linear_timeouts_max,
1487 },
1488 {
1489 .procname = "tcp_shrink_window",
1490 .data = &init_net.ipv4.sysctl_tcp_shrink_window,
1491 .maxlen = sizeof(u8),
1492 .mode = 0644,
1493 .proc_handler = proc_dou8vec_minmax,
1494 .extra1 = SYSCTL_ZERO,
1495 .extra2 = SYSCTL_ONE,
1496 },
1497 {
1498 .procname = "tcp_pingpong_thresh",
1499 .data = &init_net.ipv4.sysctl_tcp_pingpong_thresh,
1500 .maxlen = sizeof(u8),
1501 .mode = 0644,
1502 .proc_handler = proc_dou8vec_minmax,
1503 .extra1 = SYSCTL_ONE,
1504 },
1505 { }
1506};
1507
1508static __net_init int ipv4_sysctl_init_net(struct net *net)
1509{
1510 struct ctl_table *table;
1511
1512 table = ipv4_net_table;
1513 if (!net_eq(net, &init_net)) {
1514 int i;
1515
1516 table = kmemdup(table, sizeof(ipv4_net_table), GFP_KERNEL);
1517 if (!table)
1518 goto err_alloc;
1519
1520 for (i = 0; i < ARRAY_SIZE(ipv4_net_table) - 1; i++) {
1521 if (table[i].data) {
1522 /* Update the variables to point into
1523 * the current struct net
1524 */
1525 table[i].data += (void *)net - (void *)&init_net;
1526 } else {
1527 /* Entries without data pointer are global;
1528 * Make them read-only in non-init_net ns
1529 */
1530 table[i].mode &= ~0222;
1531 }
1532 }
1533 }
1534
1535 net->ipv4.ipv4_hdr = register_net_sysctl_sz(net, "net/ipv4", table,
1536 ARRAY_SIZE(ipv4_net_table));
1537 if (!net->ipv4.ipv4_hdr)
1538 goto err_reg;
1539
1540 net->ipv4.sysctl_local_reserved_ports = kzalloc(65536 / 8, GFP_KERNEL);
1541 if (!net->ipv4.sysctl_local_reserved_ports)
1542 goto err_ports;
1543
1544 return 0;
1545
1546err_ports:
1547 unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
1548err_reg:
1549 if (!net_eq(net, &init_net))
1550 kfree(table);
1551err_alloc:
1552 return -ENOMEM;
1553}
1554
1555static __net_exit void ipv4_sysctl_exit_net(struct net *net)
1556{
1557 struct ctl_table *table;
1558
1559 kfree(net->ipv4.sysctl_local_reserved_ports);
1560 table = net->ipv4.ipv4_hdr->ctl_table_arg;
1561 unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
1562 kfree(table);
1563}
1564
1565static __net_initdata struct pernet_operations ipv4_sysctl_ops = {
1566 .init = ipv4_sysctl_init_net,
1567 .exit = ipv4_sysctl_exit_net,
1568};
1569
1570static __init int sysctl_ipv4_init(void)
1571{
1572 struct ctl_table_header *hdr;
1573
1574 hdr = register_net_sysctl(&init_net, "net/ipv4", ipv4_table);
1575 if (!hdr)
1576 return -ENOMEM;
1577
1578 if (register_pernet_subsys(&ipv4_sysctl_ops)) {
1579 unregister_net_sysctl_table(hdr);
1580 return -ENOMEM;
1581 }
1582
1583 return 0;
1584}
1585
1586__initcall(sysctl_ipv4_init);