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#include <net/tcp_memcontrol.h>
28
29static int zero;
30static int one = 1;
31static int four = 4;
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 write_seqlock(&net->ipv4.ip_local_ports.lock);
49 net->ipv4.ip_local_ports.range[0] = range[0];
50 net->ipv4.ip_local_ports.range[1] = range[1];
51 write_sequnlock(&net->ipv4.ip_local_ports.lock);
52}
53
54/* Validate changes from /proc interface. */
55static int ipv4_local_port_range(struct ctl_table *table, int write,
56 void __user *buffer,
57 size_t *lenp, loff_t *ppos)
58{
59 struct net *net =
60 container_of(table->data, struct net, ipv4.ip_local_ports.range);
61 int ret;
62 int range[2];
63 struct ctl_table tmp = {
64 .data = &range,
65 .maxlen = sizeof(range),
66 .mode = table->mode,
67 .extra1 = &ip_local_port_range_min,
68 .extra2 = &ip_local_port_range_max,
69 };
70
71 inet_get_local_port_range(net, &range[0], &range[1]);
72
73 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
74
75 if (write && ret == 0) {
76 if (range[1] < range[0])
77 ret = -EINVAL;
78 else
79 set_local_port_range(net, range);
80 }
81
82 return ret;
83}
84
85
86static void inet_get_ping_group_range_table(struct ctl_table *table, kgid_t *low, kgid_t *high)
87{
88 kgid_t *data = table->data;
89 struct net *net =
90 container_of(table->data, struct net, ipv4.ping_group_range.range);
91 unsigned int seq;
92 do {
93 seq = read_seqbegin(&net->ipv4.ip_local_ports.lock);
94
95 *low = data[0];
96 *high = data[1];
97 } while (read_seqretry(&net->ipv4.ip_local_ports.lock, seq));
98}
99
100/* Update system visible IP port range */
101static void set_ping_group_range(struct ctl_table *table, kgid_t low, kgid_t high)
102{
103 kgid_t *data = table->data;
104 struct net *net =
105 container_of(table->data, struct net, ipv4.ping_group_range.range);
106 write_seqlock(&net->ipv4.ip_local_ports.lock);
107 data[0] = low;
108 data[1] = high;
109 write_sequnlock(&net->ipv4.ip_local_ports.lock);
110}
111
112/* Validate changes from /proc interface. */
113static int ipv4_ping_group_range(struct ctl_table *table, int write,
114 void __user *buffer,
115 size_t *lenp, loff_t *ppos)
116{
117 struct user_namespace *user_ns = current_user_ns();
118 int ret;
119 gid_t urange[2];
120 kgid_t low, high;
121 struct ctl_table tmp = {
122 .data = &urange,
123 .maxlen = sizeof(urange),
124 .mode = table->mode,
125 .extra1 = &ip_ping_group_range_min,
126 .extra2 = &ip_ping_group_range_max,
127 };
128
129 inet_get_ping_group_range_table(table, &low, &high);
130 urange[0] = from_kgid_munged(user_ns, low);
131 urange[1] = from_kgid_munged(user_ns, high);
132 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
133
134 if (write && ret == 0) {
135 low = make_kgid(user_ns, urange[0]);
136 high = make_kgid(user_ns, urange[1]);
137 if (!gid_valid(low) || !gid_valid(high) ||
138 (urange[1] < urange[0]) || gid_lt(high, low)) {
139 low = make_kgid(&init_user_ns, 1);
140 high = make_kgid(&init_user_ns, 0);
141 }
142 set_ping_group_range(table, low, high);
143 }
144
145 return ret;
146}
147
148static int proc_tcp_congestion_control(struct ctl_table *ctl, int write,
149 void __user *buffer, size_t *lenp, loff_t *ppos)
150{
151 char val[TCP_CA_NAME_MAX];
152 struct ctl_table tbl = {
153 .data = val,
154 .maxlen = TCP_CA_NAME_MAX,
155 };
156 int ret;
157
158 tcp_get_default_congestion_control(val);
159
160 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
161 if (write && ret == 0)
162 ret = tcp_set_default_congestion_control(val);
163 return ret;
164}
165
166static int proc_tcp_available_congestion_control(struct ctl_table *ctl,
167 int write,
168 void __user *buffer, size_t *lenp,
169 loff_t *ppos)
170{
171 struct ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX, };
172 int ret;
173
174 tbl.data = kmalloc(tbl.maxlen, GFP_USER);
175 if (!tbl.data)
176 return -ENOMEM;
177 tcp_get_available_congestion_control(tbl.data, TCP_CA_BUF_MAX);
178 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
179 kfree(tbl.data);
180 return ret;
181}
182
183static int proc_allowed_congestion_control(struct ctl_table *ctl,
184 int write,
185 void __user *buffer, size_t *lenp,
186 loff_t *ppos)
187{
188 struct ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX };
189 int ret;
190
191 tbl.data = kmalloc(tbl.maxlen, GFP_USER);
192 if (!tbl.data)
193 return -ENOMEM;
194
195 tcp_get_allowed_congestion_control(tbl.data, tbl.maxlen);
196 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
197 if (write && ret == 0)
198 ret = tcp_set_allowed_congestion_control(tbl.data);
199 kfree(tbl.data);
200 return ret;
201}
202
203static int proc_tcp_fastopen_key(struct ctl_table *ctl, int write,
204 void __user *buffer, size_t *lenp,
205 loff_t *ppos)
206{
207 struct ctl_table tbl = { .maxlen = (TCP_FASTOPEN_KEY_LENGTH * 2 + 10) };
208 struct tcp_fastopen_context *ctxt;
209 int ret;
210 u32 user_key[4]; /* 16 bytes, matching TCP_FASTOPEN_KEY_LENGTH */
211
212 tbl.data = kmalloc(tbl.maxlen, GFP_KERNEL);
213 if (!tbl.data)
214 return -ENOMEM;
215
216 rcu_read_lock();
217 ctxt = rcu_dereference(tcp_fastopen_ctx);
218 if (ctxt)
219 memcpy(user_key, ctxt->key, TCP_FASTOPEN_KEY_LENGTH);
220 else
221 memset(user_key, 0, sizeof(user_key));
222 rcu_read_unlock();
223
224 snprintf(tbl.data, tbl.maxlen, "%08x-%08x-%08x-%08x",
225 user_key[0], user_key[1], user_key[2], user_key[3]);
226 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
227
228 if (write && ret == 0) {
229 if (sscanf(tbl.data, "%x-%x-%x-%x", user_key, user_key + 1,
230 user_key + 2, user_key + 3) != 4) {
231 ret = -EINVAL;
232 goto bad_key;
233 }
234 /* Generate a dummy secret but don't publish it. This
235 * is needed so we don't regenerate a new key on the
236 * first invocation of tcp_fastopen_cookie_gen
237 */
238 tcp_fastopen_init_key_once(false);
239 tcp_fastopen_reset_cipher(user_key, TCP_FASTOPEN_KEY_LENGTH);
240 }
241
242bad_key:
243 pr_debug("proc FO key set 0x%x-%x-%x-%x <- 0x%s: %u\n",
244 user_key[0], user_key[1], user_key[2], user_key[3],
245 (char *)tbl.data, ret);
246 kfree(tbl.data);
247 return ret;
248}
249
250static struct ctl_table ipv4_table[] = {
251 {
252 .procname = "tcp_timestamps",
253 .data = &sysctl_tcp_timestamps,
254 .maxlen = sizeof(int),
255 .mode = 0644,
256 .proc_handler = proc_dointvec
257 },
258 {
259 .procname = "tcp_window_scaling",
260 .data = &sysctl_tcp_window_scaling,
261 .maxlen = sizeof(int),
262 .mode = 0644,
263 .proc_handler = proc_dointvec
264 },
265 {
266 .procname = "tcp_sack",
267 .data = &sysctl_tcp_sack,
268 .maxlen = sizeof(int),
269 .mode = 0644,
270 .proc_handler = proc_dointvec
271 },
272 {
273 .procname = "tcp_retrans_collapse",
274 .data = &sysctl_tcp_retrans_collapse,
275 .maxlen = sizeof(int),
276 .mode = 0644,
277 .proc_handler = proc_dointvec
278 },
279 {
280 .procname = "ip_default_ttl",
281 .data = &sysctl_ip_default_ttl,
282 .maxlen = sizeof(int),
283 .mode = 0644,
284 .proc_handler = proc_dointvec_minmax,
285 .extra1 = &ip_ttl_min,
286 .extra2 = &ip_ttl_max,
287 },
288 {
289 .procname = "ip_nonlocal_bind",
290 .data = &sysctl_ip_nonlocal_bind,
291 .maxlen = sizeof(int),
292 .mode = 0644,
293 .proc_handler = proc_dointvec
294 },
295 {
296 .procname = "tcp_syn_retries",
297 .data = &sysctl_tcp_syn_retries,
298 .maxlen = sizeof(int),
299 .mode = 0644,
300 .proc_handler = proc_dointvec_minmax,
301 .extra1 = &tcp_syn_retries_min,
302 .extra2 = &tcp_syn_retries_max
303 },
304 {
305 .procname = "tcp_synack_retries",
306 .data = &sysctl_tcp_synack_retries,
307 .maxlen = sizeof(int),
308 .mode = 0644,
309 .proc_handler = proc_dointvec
310 },
311 {
312 .procname = "tcp_max_orphans",
313 .data = &sysctl_tcp_max_orphans,
314 .maxlen = sizeof(int),
315 .mode = 0644,
316 .proc_handler = proc_dointvec
317 },
318 {
319 .procname = "tcp_max_tw_buckets",
320 .data = &tcp_death_row.sysctl_max_tw_buckets,
321 .maxlen = sizeof(int),
322 .mode = 0644,
323 .proc_handler = proc_dointvec
324 },
325 {
326 .procname = "ip_early_demux",
327 .data = &sysctl_ip_early_demux,
328 .maxlen = sizeof(int),
329 .mode = 0644,
330 .proc_handler = proc_dointvec
331 },
332 {
333 .procname = "ip_dynaddr",
334 .data = &sysctl_ip_dynaddr,
335 .maxlen = sizeof(int),
336 .mode = 0644,
337 .proc_handler = proc_dointvec
338 },
339 {
340 .procname = "tcp_keepalive_time",
341 .data = &sysctl_tcp_keepalive_time,
342 .maxlen = sizeof(int),
343 .mode = 0644,
344 .proc_handler = proc_dointvec_jiffies,
345 },
346 {
347 .procname = "tcp_keepalive_probes",
348 .data = &sysctl_tcp_keepalive_probes,
349 .maxlen = sizeof(int),
350 .mode = 0644,
351 .proc_handler = proc_dointvec
352 },
353 {
354 .procname = "tcp_keepalive_intvl",
355 .data = &sysctl_tcp_keepalive_intvl,
356 .maxlen = sizeof(int),
357 .mode = 0644,
358 .proc_handler = proc_dointvec_jiffies,
359 },
360 {
361 .procname = "tcp_retries1",
362 .data = &sysctl_tcp_retries1,
363 .maxlen = sizeof(int),
364 .mode = 0644,
365 .proc_handler = proc_dointvec_minmax,
366 .extra2 = &tcp_retr1_max
367 },
368 {
369 .procname = "tcp_retries2",
370 .data = &sysctl_tcp_retries2,
371 .maxlen = sizeof(int),
372 .mode = 0644,
373 .proc_handler = proc_dointvec
374 },
375 {
376 .procname = "tcp_fin_timeout",
377 .data = &sysctl_tcp_fin_timeout,
378 .maxlen = sizeof(int),
379 .mode = 0644,
380 .proc_handler = proc_dointvec_jiffies,
381 },
382#ifdef CONFIG_SYN_COOKIES
383 {
384 .procname = "tcp_syncookies",
385 .data = &sysctl_tcp_syncookies,
386 .maxlen = sizeof(int),
387 .mode = 0644,
388 .proc_handler = proc_dointvec
389 },
390#endif
391 {
392 .procname = "tcp_fastopen",
393 .data = &sysctl_tcp_fastopen,
394 .maxlen = sizeof(int),
395 .mode = 0644,
396 .proc_handler = proc_dointvec,
397 },
398 {
399 .procname = "tcp_fastopen_key",
400 .mode = 0600,
401 .maxlen = ((TCP_FASTOPEN_KEY_LENGTH * 2) + 10),
402 .proc_handler = proc_tcp_fastopen_key,
403 },
404 {
405 .procname = "tcp_tw_recycle",
406 .data = &tcp_death_row.sysctl_tw_recycle,
407 .maxlen = sizeof(int),
408 .mode = 0644,
409 .proc_handler = proc_dointvec
410 },
411 {
412 .procname = "tcp_abort_on_overflow",
413 .data = &sysctl_tcp_abort_on_overflow,
414 .maxlen = sizeof(int),
415 .mode = 0644,
416 .proc_handler = proc_dointvec
417 },
418 {
419 .procname = "tcp_stdurg",
420 .data = &sysctl_tcp_stdurg,
421 .maxlen = sizeof(int),
422 .mode = 0644,
423 .proc_handler = proc_dointvec
424 },
425 {
426 .procname = "tcp_rfc1337",
427 .data = &sysctl_tcp_rfc1337,
428 .maxlen = sizeof(int),
429 .mode = 0644,
430 .proc_handler = proc_dointvec
431 },
432 {
433 .procname = "tcp_max_syn_backlog",
434 .data = &sysctl_max_syn_backlog,
435 .maxlen = sizeof(int),
436 .mode = 0644,
437 .proc_handler = proc_dointvec
438 },
439 {
440 .procname = "ip_local_reserved_ports",
441 .data = NULL, /* initialized in sysctl_ipv4_init */
442 .maxlen = 65536,
443 .mode = 0644,
444 .proc_handler = proc_do_large_bitmap,
445 },
446 {
447 .procname = "igmp_max_memberships",
448 .data = &sysctl_igmp_max_memberships,
449 .maxlen = sizeof(int),
450 .mode = 0644,
451 .proc_handler = proc_dointvec
452 },
453 {
454 .procname = "igmp_max_msf",
455 .data = &sysctl_igmp_max_msf,
456 .maxlen = sizeof(int),
457 .mode = 0644,
458 .proc_handler = proc_dointvec
459 },
460 {
461 .procname = "inet_peer_threshold",
462 .data = &inet_peer_threshold,
463 .maxlen = sizeof(int),
464 .mode = 0644,
465 .proc_handler = proc_dointvec
466 },
467 {
468 .procname = "inet_peer_minttl",
469 .data = &inet_peer_minttl,
470 .maxlen = sizeof(int),
471 .mode = 0644,
472 .proc_handler = proc_dointvec_jiffies,
473 },
474 {
475 .procname = "inet_peer_maxttl",
476 .data = &inet_peer_maxttl,
477 .maxlen = sizeof(int),
478 .mode = 0644,
479 .proc_handler = proc_dointvec_jiffies,
480 },
481 {
482 .procname = "tcp_orphan_retries",
483 .data = &sysctl_tcp_orphan_retries,
484 .maxlen = sizeof(int),
485 .mode = 0644,
486 .proc_handler = proc_dointvec
487 },
488 {
489 .procname = "tcp_fack",
490 .data = &sysctl_tcp_fack,
491 .maxlen = sizeof(int),
492 .mode = 0644,
493 .proc_handler = proc_dointvec
494 },
495 {
496 .procname = "tcp_reordering",
497 .data = &sysctl_tcp_reordering,
498 .maxlen = sizeof(int),
499 .mode = 0644,
500 .proc_handler = proc_dointvec
501 },
502 {
503 .procname = "tcp_dsack",
504 .data = &sysctl_tcp_dsack,
505 .maxlen = sizeof(int),
506 .mode = 0644,
507 .proc_handler = proc_dointvec
508 },
509 {
510 .procname = "tcp_mem",
511 .maxlen = sizeof(sysctl_tcp_mem),
512 .data = &sysctl_tcp_mem,
513 .mode = 0644,
514 .proc_handler = proc_doulongvec_minmax,
515 },
516 {
517 .procname = "tcp_wmem",
518 .data = &sysctl_tcp_wmem,
519 .maxlen = sizeof(sysctl_tcp_wmem),
520 .mode = 0644,
521 .proc_handler = proc_dointvec_minmax,
522 .extra1 = &one,
523 },
524 {
525 .procname = "tcp_notsent_lowat",
526 .data = &sysctl_tcp_notsent_lowat,
527 .maxlen = sizeof(sysctl_tcp_notsent_lowat),
528 .mode = 0644,
529 .proc_handler = proc_dointvec,
530 },
531 {
532 .procname = "tcp_rmem",
533 .data = &sysctl_tcp_rmem,
534 .maxlen = sizeof(sysctl_tcp_rmem),
535 .mode = 0644,
536 .proc_handler = proc_dointvec_minmax,
537 .extra1 = &one,
538 },
539 {
540 .procname = "tcp_app_win",
541 .data = &sysctl_tcp_app_win,
542 .maxlen = sizeof(int),
543 .mode = 0644,
544 .proc_handler = proc_dointvec
545 },
546 {
547 .procname = "tcp_adv_win_scale",
548 .data = &sysctl_tcp_adv_win_scale,
549 .maxlen = sizeof(int),
550 .mode = 0644,
551 .proc_handler = proc_dointvec_minmax,
552 .extra1 = &tcp_adv_win_scale_min,
553 .extra2 = &tcp_adv_win_scale_max,
554 },
555 {
556 .procname = "tcp_tw_reuse",
557 .data = &sysctl_tcp_tw_reuse,
558 .maxlen = sizeof(int),
559 .mode = 0644,
560 .proc_handler = proc_dointvec
561 },
562 {
563 .procname = "tcp_frto",
564 .data = &sysctl_tcp_frto,
565 .maxlen = sizeof(int),
566 .mode = 0644,
567 .proc_handler = proc_dointvec
568 },
569 {
570 .procname = "tcp_low_latency",
571 .data = &sysctl_tcp_low_latency,
572 .maxlen = sizeof(int),
573 .mode = 0644,
574 .proc_handler = proc_dointvec
575 },
576 {
577 .procname = "tcp_no_metrics_save",
578 .data = &sysctl_tcp_nometrics_save,
579 .maxlen = sizeof(int),
580 .mode = 0644,
581 .proc_handler = proc_dointvec,
582 },
583 {
584 .procname = "tcp_moderate_rcvbuf",
585 .data = &sysctl_tcp_moderate_rcvbuf,
586 .maxlen = sizeof(int),
587 .mode = 0644,
588 .proc_handler = proc_dointvec,
589 },
590 {
591 .procname = "tcp_tso_win_divisor",
592 .data = &sysctl_tcp_tso_win_divisor,
593 .maxlen = sizeof(int),
594 .mode = 0644,
595 .proc_handler = proc_dointvec,
596 },
597 {
598 .procname = "tcp_congestion_control",
599 .mode = 0644,
600 .maxlen = TCP_CA_NAME_MAX,
601 .proc_handler = proc_tcp_congestion_control,
602 },
603 {
604 .procname = "tcp_mtu_probing",
605 .data = &sysctl_tcp_mtu_probing,
606 .maxlen = sizeof(int),
607 .mode = 0644,
608 .proc_handler = proc_dointvec,
609 },
610 {
611 .procname = "tcp_base_mss",
612 .data = &sysctl_tcp_base_mss,
613 .maxlen = sizeof(int),
614 .mode = 0644,
615 .proc_handler = proc_dointvec,
616 },
617 {
618 .procname = "tcp_workaround_signed_windows",
619 .data = &sysctl_tcp_workaround_signed_windows,
620 .maxlen = sizeof(int),
621 .mode = 0644,
622 .proc_handler = proc_dointvec
623 },
624 {
625 .procname = "tcp_limit_output_bytes",
626 .data = &sysctl_tcp_limit_output_bytes,
627 .maxlen = sizeof(int),
628 .mode = 0644,
629 .proc_handler = proc_dointvec
630 },
631 {
632 .procname = "tcp_challenge_ack_limit",
633 .data = &sysctl_tcp_challenge_ack_limit,
634 .maxlen = sizeof(int),
635 .mode = 0644,
636 .proc_handler = proc_dointvec
637 },
638#ifdef CONFIG_NET_DMA
639 {
640 .procname = "tcp_dma_copybreak",
641 .data = &sysctl_tcp_dma_copybreak,
642 .maxlen = sizeof(int),
643 .mode = 0644,
644 .proc_handler = proc_dointvec
645 },
646#endif
647 {
648 .procname = "tcp_slow_start_after_idle",
649 .data = &sysctl_tcp_slow_start_after_idle,
650 .maxlen = sizeof(int),
651 .mode = 0644,
652 .proc_handler = proc_dointvec
653 },
654#ifdef CONFIG_NETLABEL
655 {
656 .procname = "cipso_cache_enable",
657 .data = &cipso_v4_cache_enabled,
658 .maxlen = sizeof(int),
659 .mode = 0644,
660 .proc_handler = proc_dointvec,
661 },
662 {
663 .procname = "cipso_cache_bucket_size",
664 .data = &cipso_v4_cache_bucketsize,
665 .maxlen = sizeof(int),
666 .mode = 0644,
667 .proc_handler = proc_dointvec,
668 },
669 {
670 .procname = "cipso_rbm_optfmt",
671 .data = &cipso_v4_rbm_optfmt,
672 .maxlen = sizeof(int),
673 .mode = 0644,
674 .proc_handler = proc_dointvec,
675 },
676 {
677 .procname = "cipso_rbm_strictvalid",
678 .data = &cipso_v4_rbm_strictvalid,
679 .maxlen = sizeof(int),
680 .mode = 0644,
681 .proc_handler = proc_dointvec,
682 },
683#endif /* CONFIG_NETLABEL */
684 {
685 .procname = "tcp_available_congestion_control",
686 .maxlen = TCP_CA_BUF_MAX,
687 .mode = 0444,
688 .proc_handler = proc_tcp_available_congestion_control,
689 },
690 {
691 .procname = "tcp_allowed_congestion_control",
692 .maxlen = TCP_CA_BUF_MAX,
693 .mode = 0644,
694 .proc_handler = proc_allowed_congestion_control,
695 },
696 {
697 .procname = "tcp_thin_linear_timeouts",
698 .data = &sysctl_tcp_thin_linear_timeouts,
699 .maxlen = sizeof(int),
700 .mode = 0644,
701 .proc_handler = proc_dointvec
702 },
703 {
704 .procname = "tcp_thin_dupack",
705 .data = &sysctl_tcp_thin_dupack,
706 .maxlen = sizeof(int),
707 .mode = 0644,
708 .proc_handler = proc_dointvec
709 },
710 {
711 .procname = "tcp_early_retrans",
712 .data = &sysctl_tcp_early_retrans,
713 .maxlen = sizeof(int),
714 .mode = 0644,
715 .proc_handler = proc_dointvec_minmax,
716 .extra1 = &zero,
717 .extra2 = &four,
718 },
719 {
720 .procname = "tcp_min_tso_segs",
721 .data = &sysctl_tcp_min_tso_segs,
722 .maxlen = sizeof(int),
723 .mode = 0644,
724 .proc_handler = proc_dointvec_minmax,
725 .extra1 = &zero,
726 .extra2 = &gso_max_segs,
727 },
728 {
729 .procname = "tcp_autocorking",
730 .data = &sysctl_tcp_autocorking,
731 .maxlen = sizeof(int),
732 .mode = 0644,
733 .proc_handler = proc_dointvec_minmax,
734 .extra1 = &zero,
735 .extra2 = &one,
736 },
737 {
738 .procname = "udp_mem",
739 .data = &sysctl_udp_mem,
740 .maxlen = sizeof(sysctl_udp_mem),
741 .mode = 0644,
742 .proc_handler = proc_doulongvec_minmax,
743 },
744 {
745 .procname = "udp_rmem_min",
746 .data = &sysctl_udp_rmem_min,
747 .maxlen = sizeof(sysctl_udp_rmem_min),
748 .mode = 0644,
749 .proc_handler = proc_dointvec_minmax,
750 .extra1 = &one
751 },
752 {
753 .procname = "udp_wmem_min",
754 .data = &sysctl_udp_wmem_min,
755 .maxlen = sizeof(sysctl_udp_wmem_min),
756 .mode = 0644,
757 .proc_handler = proc_dointvec_minmax,
758 .extra1 = &one
759 },
760 { }
761};
762
763static struct ctl_table ipv4_net_table[] = {
764 {
765 .procname = "icmp_echo_ignore_all",
766 .data = &init_net.ipv4.sysctl_icmp_echo_ignore_all,
767 .maxlen = sizeof(int),
768 .mode = 0644,
769 .proc_handler = proc_dointvec
770 },
771 {
772 .procname = "icmp_echo_ignore_broadcasts",
773 .data = &init_net.ipv4.sysctl_icmp_echo_ignore_broadcasts,
774 .maxlen = sizeof(int),
775 .mode = 0644,
776 .proc_handler = proc_dointvec
777 },
778 {
779 .procname = "icmp_ignore_bogus_error_responses",
780 .data = &init_net.ipv4.sysctl_icmp_ignore_bogus_error_responses,
781 .maxlen = sizeof(int),
782 .mode = 0644,
783 .proc_handler = proc_dointvec
784 },
785 {
786 .procname = "icmp_errors_use_inbound_ifaddr",
787 .data = &init_net.ipv4.sysctl_icmp_errors_use_inbound_ifaddr,
788 .maxlen = sizeof(int),
789 .mode = 0644,
790 .proc_handler = proc_dointvec
791 },
792 {
793 .procname = "icmp_ratelimit",
794 .data = &init_net.ipv4.sysctl_icmp_ratelimit,
795 .maxlen = sizeof(int),
796 .mode = 0644,
797 .proc_handler = proc_dointvec_ms_jiffies,
798 },
799 {
800 .procname = "icmp_ratemask",
801 .data = &init_net.ipv4.sysctl_icmp_ratemask,
802 .maxlen = sizeof(int),
803 .mode = 0644,
804 .proc_handler = proc_dointvec
805 },
806 {
807 .procname = "ping_group_range",
808 .data = &init_net.ipv4.ping_group_range.range,
809 .maxlen = sizeof(gid_t)*2,
810 .mode = 0644,
811 .proc_handler = ipv4_ping_group_range,
812 },
813 {
814 .procname = "tcp_ecn",
815 .data = &init_net.ipv4.sysctl_tcp_ecn,
816 .maxlen = sizeof(int),
817 .mode = 0644,
818 .proc_handler = proc_dointvec
819 },
820 {
821 .procname = "ip_local_port_range",
822 .maxlen = sizeof(init_net.ipv4.ip_local_ports.range),
823 .data = &init_net.ipv4.ip_local_ports.range,
824 .mode = 0644,
825 .proc_handler = ipv4_local_port_range,
826 },
827 {
828 .procname = "ip_no_pmtu_disc",
829 .data = &init_net.ipv4.sysctl_ip_no_pmtu_disc,
830 .maxlen = sizeof(int),
831 .mode = 0644,
832 .proc_handler = proc_dointvec
833 },
834 {
835 .procname = "ip_forward_use_pmtu",
836 .data = &init_net.ipv4.sysctl_ip_fwd_use_pmtu,
837 .maxlen = sizeof(int),
838 .mode = 0644,
839 .proc_handler = proc_dointvec,
840 },
841 { }
842};
843
844static __net_init int ipv4_sysctl_init_net(struct net *net)
845{
846 struct ctl_table *table;
847
848 table = ipv4_net_table;
849 if (!net_eq(net, &init_net)) {
850 int i;
851
852 table = kmemdup(table, sizeof(ipv4_net_table), GFP_KERNEL);
853 if (table == NULL)
854 goto err_alloc;
855
856 /* Update the variables to point into the current struct net */
857 for (i = 0; i < ARRAY_SIZE(ipv4_net_table) - 1; i++)
858 table[i].data += (void *)net - (void *)&init_net;
859 }
860
861 net->ipv4.ipv4_hdr = register_net_sysctl(net, "net/ipv4", table);
862 if (net->ipv4.ipv4_hdr == NULL)
863 goto err_reg;
864
865 return 0;
866
867err_reg:
868 if (!net_eq(net, &init_net))
869 kfree(table);
870err_alloc:
871 return -ENOMEM;
872}
873
874static __net_exit void ipv4_sysctl_exit_net(struct net *net)
875{
876 struct ctl_table *table;
877
878 table = net->ipv4.ipv4_hdr->ctl_table_arg;
879 unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
880 kfree(table);
881}
882
883static __net_initdata struct pernet_operations ipv4_sysctl_ops = {
884 .init = ipv4_sysctl_init_net,
885 .exit = ipv4_sysctl_exit_net,
886};
887
888static __init int sysctl_ipv4_init(void)
889{
890 struct ctl_table_header *hdr;
891 struct ctl_table *i;
892
893 for (i = ipv4_table; i->procname; i++) {
894 if (strcmp(i->procname, "ip_local_reserved_ports") == 0) {
895 i->data = sysctl_local_reserved_ports;
896 break;
897 }
898 }
899 if (!i->procname)
900 return -EINVAL;
901
902 hdr = register_net_sysctl(&init_net, "net/ipv4", ipv4_table);
903 if (hdr == NULL)
904 return -ENOMEM;
905
906 if (register_pernet_subsys(&ipv4_sysctl_ops)) {
907 unregister_net_sysctl_table(hdr);
908 return -ENOMEM;
909 }
910
911 return 0;
912}
913
914__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/mm.h>
10#include <linux/module.h>
11#include <linux/sysctl.h>
12#include <linux/igmp.h>
13#include <linux/inetdevice.h>
14#include <linux/seqlock.h>
15#include <linux/init.h>
16#include <linux/slab.h>
17#include <linux/nsproxy.h>
18#include <linux/swap.h>
19#include <net/snmp.h>
20#include <net/icmp.h>
21#include <net/ip.h>
22#include <net/ip_fib.h>
23#include <net/route.h>
24#include <net/tcp.h>
25#include <net/udp.h>
26#include <net/cipso_ipv4.h>
27#include <net/inet_frag.h>
28#include <net/ping.h>
29#include <net/protocol.h>
30#include <net/netevent.h>
31
32static int two = 2;
33static int three __maybe_unused = 3;
34static int four = 4;
35static int thousand = 1000;
36static int tcp_retr1_max = 255;
37static int ip_local_port_range_min[] = { 1, 1 };
38static int ip_local_port_range_max[] = { 65535, 65535 };
39static int tcp_adv_win_scale_min = -31;
40static int tcp_adv_win_scale_max = 31;
41static int tcp_min_snd_mss_min = TCP_MIN_SND_MSS;
42static int tcp_min_snd_mss_max = 65535;
43static int ip_privileged_port_min;
44static int ip_privileged_port_max = 65535;
45static int ip_ttl_min = 1;
46static int ip_ttl_max = 255;
47static int tcp_syn_retries_min = 1;
48static int tcp_syn_retries_max = MAX_TCP_SYNCNT;
49static int ip_ping_group_range_min[] = { 0, 0 };
50static int ip_ping_group_range_max[] = { GID_T_MAX, GID_T_MAX };
51static u32 u32_max_div_HZ = UINT_MAX / HZ;
52static int one_day_secs = 24 * 3600;
53static u32 fib_multipath_hash_fields_all_mask __maybe_unused =
54 FIB_MULTIPATH_HASH_FIELD_ALL_MASK;
55
56/* obsolete */
57static int sysctl_tcp_low_latency __read_mostly;
58
59/* Update system visible IP port range */
60static void set_local_port_range(struct net *net, int range[2])
61{
62 bool same_parity = !((range[0] ^ range[1]) & 1);
63
64 write_seqlock_bh(&net->ipv4.ip_local_ports.lock);
65 if (same_parity && !net->ipv4.ip_local_ports.warned) {
66 net->ipv4.ip_local_ports.warned = true;
67 pr_err_ratelimited("ip_local_port_range: prefer different parity for start/end values.\n");
68 }
69 net->ipv4.ip_local_ports.range[0] = range[0];
70 net->ipv4.ip_local_ports.range[1] = range[1];
71 write_sequnlock_bh(&net->ipv4.ip_local_ports.lock);
72}
73
74/* Validate changes from /proc interface. */
75static int ipv4_local_port_range(struct ctl_table *table, int write,
76 void *buffer, size_t *lenp, loff_t *ppos)
77{
78 struct net *net =
79 container_of(table->data, struct net, ipv4.ip_local_ports.range);
80 int ret;
81 int range[2];
82 struct ctl_table tmp = {
83 .data = &range,
84 .maxlen = sizeof(range),
85 .mode = table->mode,
86 .extra1 = &ip_local_port_range_min,
87 .extra2 = &ip_local_port_range_max,
88 };
89
90 inet_get_local_port_range(net, &range[0], &range[1]);
91
92 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
93
94 if (write && ret == 0) {
95 /* Ensure that the upper limit is not smaller than the lower,
96 * and that the lower does not encroach upon the privileged
97 * port limit.
98 */
99 if ((range[1] < range[0]) ||
100 (range[0] < net->ipv4.sysctl_ip_prot_sock))
101 ret = -EINVAL;
102 else
103 set_local_port_range(net, range);
104 }
105
106 return ret;
107}
108
109/* Validate changes from /proc interface. */
110static int ipv4_privileged_ports(struct ctl_table *table, int write,
111 void *buffer, size_t *lenp, loff_t *ppos)
112{
113 struct net *net = container_of(table->data, struct net,
114 ipv4.sysctl_ip_prot_sock);
115 int ret;
116 int pports;
117 int range[2];
118 struct ctl_table tmp = {
119 .data = &pports,
120 .maxlen = sizeof(pports),
121 .mode = table->mode,
122 .extra1 = &ip_privileged_port_min,
123 .extra2 = &ip_privileged_port_max,
124 };
125
126 pports = net->ipv4.sysctl_ip_prot_sock;
127
128 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
129
130 if (write && ret == 0) {
131 inet_get_local_port_range(net, &range[0], &range[1]);
132 /* Ensure that the local port range doesn't overlap with the
133 * privileged port range.
134 */
135 if (range[0] < pports)
136 ret = -EINVAL;
137 else
138 net->ipv4.sysctl_ip_prot_sock = pports;
139 }
140
141 return ret;
142}
143
144static void inet_get_ping_group_range_table(struct ctl_table *table, kgid_t *low, kgid_t *high)
145{
146 kgid_t *data = table->data;
147 struct net *net =
148 container_of(table->data, struct net, ipv4.ping_group_range.range);
149 unsigned int seq;
150 do {
151 seq = read_seqbegin(&net->ipv4.ping_group_range.lock);
152
153 *low = data[0];
154 *high = data[1];
155 } while (read_seqretry(&net->ipv4.ping_group_range.lock, seq));
156}
157
158/* Update system visible IP port range */
159static void set_ping_group_range(struct ctl_table *table, kgid_t low, kgid_t high)
160{
161 kgid_t *data = table->data;
162 struct net *net =
163 container_of(table->data, struct net, ipv4.ping_group_range.range);
164 write_seqlock(&net->ipv4.ping_group_range.lock);
165 data[0] = low;
166 data[1] = high;
167 write_sequnlock(&net->ipv4.ping_group_range.lock);
168}
169
170/* Validate changes from /proc interface. */
171static int ipv4_ping_group_range(struct ctl_table *table, int write,
172 void *buffer, size_t *lenp, loff_t *ppos)
173{
174 struct user_namespace *user_ns = current_user_ns();
175 int ret;
176 gid_t urange[2];
177 kgid_t low, high;
178 struct ctl_table tmp = {
179 .data = &urange,
180 .maxlen = sizeof(urange),
181 .mode = table->mode,
182 .extra1 = &ip_ping_group_range_min,
183 .extra2 = &ip_ping_group_range_max,
184 };
185
186 inet_get_ping_group_range_table(table, &low, &high);
187 urange[0] = from_kgid_munged(user_ns, low);
188 urange[1] = from_kgid_munged(user_ns, high);
189 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
190
191 if (write && ret == 0) {
192 low = make_kgid(user_ns, urange[0]);
193 high = make_kgid(user_ns, urange[1]);
194 if (!gid_valid(low) || !gid_valid(high))
195 return -EINVAL;
196 if (urange[1] < urange[0] || gid_lt(high, low)) {
197 low = make_kgid(&init_user_ns, 1);
198 high = make_kgid(&init_user_ns, 0);
199 }
200 set_ping_group_range(table, low, high);
201 }
202
203 return ret;
204}
205
206static int ipv4_fwd_update_priority(struct ctl_table *table, int write,
207 void *buffer, size_t *lenp, loff_t *ppos)
208{
209 struct net *net;
210 int ret;
211
212 net = container_of(table->data, struct net,
213 ipv4.sysctl_ip_fwd_update_priority);
214 ret = proc_dou8vec_minmax(table, write, buffer, lenp, ppos);
215 if (write && ret == 0)
216 call_netevent_notifiers(NETEVENT_IPV4_FWD_UPDATE_PRIORITY_UPDATE,
217 net);
218
219 return ret;
220}
221
222static int proc_tcp_congestion_control(struct ctl_table *ctl, int write,
223 void *buffer, size_t *lenp, loff_t *ppos)
224{
225 struct net *net = container_of(ctl->data, struct net,
226 ipv4.tcp_congestion_control);
227 char val[TCP_CA_NAME_MAX];
228 struct ctl_table tbl = {
229 .data = val,
230 .maxlen = TCP_CA_NAME_MAX,
231 };
232 int ret;
233
234 tcp_get_default_congestion_control(net, val);
235
236 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
237 if (write && ret == 0)
238 ret = tcp_set_default_congestion_control(net, val);
239 return ret;
240}
241
242static int proc_tcp_available_congestion_control(struct ctl_table *ctl,
243 int write, void *buffer,
244 size_t *lenp, loff_t *ppos)
245{
246 struct ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX, };
247 int ret;
248
249 tbl.data = kmalloc(tbl.maxlen, GFP_USER);
250 if (!tbl.data)
251 return -ENOMEM;
252 tcp_get_available_congestion_control(tbl.data, TCP_CA_BUF_MAX);
253 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
254 kfree(tbl.data);
255 return ret;
256}
257
258static int proc_allowed_congestion_control(struct ctl_table *ctl,
259 int write, void *buffer,
260 size_t *lenp, loff_t *ppos)
261{
262 struct ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX };
263 int ret;
264
265 tbl.data = kmalloc(tbl.maxlen, GFP_USER);
266 if (!tbl.data)
267 return -ENOMEM;
268
269 tcp_get_allowed_congestion_control(tbl.data, tbl.maxlen);
270 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
271 if (write && ret == 0)
272 ret = tcp_set_allowed_congestion_control(tbl.data);
273 kfree(tbl.data);
274 return ret;
275}
276
277static int sscanf_key(char *buf, __le32 *key)
278{
279 u32 user_key[4];
280 int i, ret = 0;
281
282 if (sscanf(buf, "%x-%x-%x-%x", user_key, user_key + 1,
283 user_key + 2, user_key + 3) != 4) {
284 ret = -EINVAL;
285 } else {
286 for (i = 0; i < ARRAY_SIZE(user_key); i++)
287 key[i] = cpu_to_le32(user_key[i]);
288 }
289 pr_debug("proc TFO key set 0x%x-%x-%x-%x <- 0x%s: %u\n",
290 user_key[0], user_key[1], user_key[2], user_key[3], buf, ret);
291
292 return ret;
293}
294
295static int proc_tcp_fastopen_key(struct ctl_table *table, int write,
296 void *buffer, size_t *lenp, loff_t *ppos)
297{
298 struct net *net = container_of(table->data, struct net,
299 ipv4.sysctl_tcp_fastopen);
300 /* maxlen to print the list of keys in hex (*2), with dashes
301 * separating doublewords and a comma in between keys.
302 */
303 struct ctl_table tbl = { .maxlen = ((TCP_FASTOPEN_KEY_LENGTH *
304 2 * TCP_FASTOPEN_KEY_MAX) +
305 (TCP_FASTOPEN_KEY_MAX * 5)) };
306 u32 user_key[TCP_FASTOPEN_KEY_BUF_LENGTH / sizeof(u32)];
307 __le32 key[TCP_FASTOPEN_KEY_BUF_LENGTH / sizeof(__le32)];
308 char *backup_data;
309 int ret, i = 0, off = 0, n_keys;
310
311 tbl.data = kmalloc(tbl.maxlen, GFP_KERNEL);
312 if (!tbl.data)
313 return -ENOMEM;
314
315 n_keys = tcp_fastopen_get_cipher(net, NULL, (u64 *)key);
316 if (!n_keys) {
317 memset(&key[0], 0, TCP_FASTOPEN_KEY_LENGTH);
318 n_keys = 1;
319 }
320
321 for (i = 0; i < n_keys * 4; i++)
322 user_key[i] = le32_to_cpu(key[i]);
323
324 for (i = 0; i < n_keys; i++) {
325 off += snprintf(tbl.data + off, tbl.maxlen - off,
326 "%08x-%08x-%08x-%08x",
327 user_key[i * 4],
328 user_key[i * 4 + 1],
329 user_key[i * 4 + 2],
330 user_key[i * 4 + 3]);
331
332 if (WARN_ON_ONCE(off >= tbl.maxlen - 1))
333 break;
334
335 if (i + 1 < n_keys)
336 off += snprintf(tbl.data + off, tbl.maxlen - off, ",");
337 }
338
339 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
340
341 if (write && ret == 0) {
342 backup_data = strchr(tbl.data, ',');
343 if (backup_data) {
344 *backup_data = '\0';
345 backup_data++;
346 }
347 if (sscanf_key(tbl.data, key)) {
348 ret = -EINVAL;
349 goto bad_key;
350 }
351 if (backup_data) {
352 if (sscanf_key(backup_data, key + 4)) {
353 ret = -EINVAL;
354 goto bad_key;
355 }
356 }
357 tcp_fastopen_reset_cipher(net, NULL, key,
358 backup_data ? key + 4 : NULL);
359 }
360
361bad_key:
362 kfree(tbl.data);
363 return ret;
364}
365
366static void proc_configure_early_demux(int enabled, int protocol)
367{
368 struct net_protocol *ipprot;
369#if IS_ENABLED(CONFIG_IPV6)
370 struct inet6_protocol *ip6prot;
371#endif
372
373 rcu_read_lock();
374
375 ipprot = rcu_dereference(inet_protos[protocol]);
376 if (ipprot)
377 ipprot->early_demux = enabled ? ipprot->early_demux_handler :
378 NULL;
379
380#if IS_ENABLED(CONFIG_IPV6)
381 ip6prot = rcu_dereference(inet6_protos[protocol]);
382 if (ip6prot)
383 ip6prot->early_demux = enabled ? ip6prot->early_demux_handler :
384 NULL;
385#endif
386 rcu_read_unlock();
387}
388
389static int proc_tcp_early_demux(struct ctl_table *table, int write,
390 void *buffer, size_t *lenp, loff_t *ppos)
391{
392 int ret = 0;
393
394 ret = proc_dou8vec_minmax(table, write, buffer, lenp, ppos);
395
396 if (write && !ret) {
397 int enabled = init_net.ipv4.sysctl_tcp_early_demux;
398
399 proc_configure_early_demux(enabled, IPPROTO_TCP);
400 }
401
402 return ret;
403}
404
405static int proc_udp_early_demux(struct ctl_table *table, int write,
406 void *buffer, size_t *lenp, loff_t *ppos)
407{
408 int ret = 0;
409
410 ret = proc_dou8vec_minmax(table, write, buffer, lenp, ppos);
411
412 if (write && !ret) {
413 int enabled = init_net.ipv4.sysctl_udp_early_demux;
414
415 proc_configure_early_demux(enabled, IPPROTO_UDP);
416 }
417
418 return ret;
419}
420
421static int proc_tfo_blackhole_detect_timeout(struct ctl_table *table,
422 int write, void *buffer,
423 size_t *lenp, loff_t *ppos)
424{
425 struct net *net = container_of(table->data, struct net,
426 ipv4.sysctl_tcp_fastopen_blackhole_timeout);
427 int ret;
428
429 ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
430 if (write && ret == 0)
431 atomic_set(&net->ipv4.tfo_active_disable_times, 0);
432
433 return ret;
434}
435
436static int proc_tcp_available_ulp(struct ctl_table *ctl,
437 int write, void *buffer, size_t *lenp,
438 loff_t *ppos)
439{
440 struct ctl_table tbl = { .maxlen = TCP_ULP_BUF_MAX, };
441 int ret;
442
443 tbl.data = kmalloc(tbl.maxlen, GFP_USER);
444 if (!tbl.data)
445 return -ENOMEM;
446 tcp_get_available_ulp(tbl.data, TCP_ULP_BUF_MAX);
447 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
448 kfree(tbl.data);
449
450 return ret;
451}
452
453#ifdef CONFIG_IP_ROUTE_MULTIPATH
454static int proc_fib_multipath_hash_policy(struct ctl_table *table, int write,
455 void *buffer, size_t *lenp,
456 loff_t *ppos)
457{
458 struct net *net = container_of(table->data, struct net,
459 ipv4.sysctl_fib_multipath_hash_policy);
460 int ret;
461
462 ret = proc_dou8vec_minmax(table, write, buffer, lenp, ppos);
463 if (write && ret == 0)
464 call_netevent_notifiers(NETEVENT_IPV4_MPATH_HASH_UPDATE, net);
465
466 return ret;
467}
468
469static int proc_fib_multipath_hash_fields(struct ctl_table *table, int write,
470 void *buffer, size_t *lenp,
471 loff_t *ppos)
472{
473 struct net *net;
474 int ret;
475
476 net = container_of(table->data, struct net,
477 ipv4.sysctl_fib_multipath_hash_fields);
478 ret = proc_douintvec_minmax(table, write, buffer, lenp, ppos);
479 if (write && ret == 0)
480 call_netevent_notifiers(NETEVENT_IPV4_MPATH_HASH_UPDATE, net);
481
482 return ret;
483}
484#endif
485
486static struct ctl_table ipv4_table[] = {
487 {
488 .procname = "tcp_max_orphans",
489 .data = &sysctl_tcp_max_orphans,
490 .maxlen = sizeof(int),
491 .mode = 0644,
492 .proc_handler = proc_dointvec
493 },
494 {
495 .procname = "inet_peer_threshold",
496 .data = &inet_peer_threshold,
497 .maxlen = sizeof(int),
498 .mode = 0644,
499 .proc_handler = proc_dointvec
500 },
501 {
502 .procname = "inet_peer_minttl",
503 .data = &inet_peer_minttl,
504 .maxlen = sizeof(int),
505 .mode = 0644,
506 .proc_handler = proc_dointvec_jiffies,
507 },
508 {
509 .procname = "inet_peer_maxttl",
510 .data = &inet_peer_maxttl,
511 .maxlen = sizeof(int),
512 .mode = 0644,
513 .proc_handler = proc_dointvec_jiffies,
514 },
515 {
516 .procname = "tcp_mem",
517 .maxlen = sizeof(sysctl_tcp_mem),
518 .data = &sysctl_tcp_mem,
519 .mode = 0644,
520 .proc_handler = proc_doulongvec_minmax,
521 },
522 {
523 .procname = "tcp_low_latency",
524 .data = &sysctl_tcp_low_latency,
525 .maxlen = sizeof(int),
526 .mode = 0644,
527 .proc_handler = proc_dointvec
528 },
529#ifdef CONFIG_NETLABEL
530 {
531 .procname = "cipso_cache_enable",
532 .data = &cipso_v4_cache_enabled,
533 .maxlen = sizeof(int),
534 .mode = 0644,
535 .proc_handler = proc_dointvec,
536 },
537 {
538 .procname = "cipso_cache_bucket_size",
539 .data = &cipso_v4_cache_bucketsize,
540 .maxlen = sizeof(int),
541 .mode = 0644,
542 .proc_handler = proc_dointvec,
543 },
544 {
545 .procname = "cipso_rbm_optfmt",
546 .data = &cipso_v4_rbm_optfmt,
547 .maxlen = sizeof(int),
548 .mode = 0644,
549 .proc_handler = proc_dointvec,
550 },
551 {
552 .procname = "cipso_rbm_strictvalid",
553 .data = &cipso_v4_rbm_strictvalid,
554 .maxlen = sizeof(int),
555 .mode = 0644,
556 .proc_handler = proc_dointvec,
557 },
558#endif /* CONFIG_NETLABEL */
559 {
560 .procname = "tcp_available_ulp",
561 .maxlen = TCP_ULP_BUF_MAX,
562 .mode = 0444,
563 .proc_handler = proc_tcp_available_ulp,
564 },
565 {
566 .procname = "icmp_msgs_per_sec",
567 .data = &sysctl_icmp_msgs_per_sec,
568 .maxlen = sizeof(int),
569 .mode = 0644,
570 .proc_handler = proc_dointvec_minmax,
571 .extra1 = SYSCTL_ZERO,
572 },
573 {
574 .procname = "icmp_msgs_burst",
575 .data = &sysctl_icmp_msgs_burst,
576 .maxlen = sizeof(int),
577 .mode = 0644,
578 .proc_handler = proc_dointvec_minmax,
579 .extra1 = SYSCTL_ZERO,
580 },
581 {
582 .procname = "udp_mem",
583 .data = &sysctl_udp_mem,
584 .maxlen = sizeof(sysctl_udp_mem),
585 .mode = 0644,
586 .proc_handler = proc_doulongvec_minmax,
587 },
588 {
589 .procname = "fib_sync_mem",
590 .data = &sysctl_fib_sync_mem,
591 .maxlen = sizeof(sysctl_fib_sync_mem),
592 .mode = 0644,
593 .proc_handler = proc_douintvec_minmax,
594 .extra1 = &sysctl_fib_sync_mem_min,
595 .extra2 = &sysctl_fib_sync_mem_max,
596 },
597 {
598 .procname = "tcp_rx_skb_cache",
599 .data = &tcp_rx_skb_cache_key.key,
600 .mode = 0644,
601 .proc_handler = proc_do_static_key,
602 },
603 {
604 .procname = "tcp_tx_skb_cache",
605 .data = &tcp_tx_skb_cache_key.key,
606 .mode = 0644,
607 .proc_handler = proc_do_static_key,
608 },
609 { }
610};
611
612static struct ctl_table ipv4_net_table[] = {
613 {
614 .procname = "icmp_echo_ignore_all",
615 .data = &init_net.ipv4.sysctl_icmp_echo_ignore_all,
616 .maxlen = sizeof(u8),
617 .mode = 0644,
618 .proc_handler = proc_dou8vec_minmax,
619 },
620 {
621 .procname = "icmp_echo_enable_probe",
622 .data = &init_net.ipv4.sysctl_icmp_echo_enable_probe,
623 .maxlen = sizeof(u8),
624 .mode = 0644,
625 .proc_handler = proc_dou8vec_minmax,
626 .extra1 = SYSCTL_ZERO,
627 .extra2 = SYSCTL_ONE
628 },
629 {
630 .procname = "icmp_echo_ignore_broadcasts",
631 .data = &init_net.ipv4.sysctl_icmp_echo_ignore_broadcasts,
632 .maxlen = sizeof(u8),
633 .mode = 0644,
634 .proc_handler = proc_dou8vec_minmax,
635 },
636 {
637 .procname = "icmp_ignore_bogus_error_responses",
638 .data = &init_net.ipv4.sysctl_icmp_ignore_bogus_error_responses,
639 .maxlen = sizeof(u8),
640 .mode = 0644,
641 .proc_handler = proc_dou8vec_minmax,
642 },
643 {
644 .procname = "icmp_errors_use_inbound_ifaddr",
645 .data = &init_net.ipv4.sysctl_icmp_errors_use_inbound_ifaddr,
646 .maxlen = sizeof(u8),
647 .mode = 0644,
648 .proc_handler = proc_dou8vec_minmax,
649 },
650 {
651 .procname = "icmp_ratelimit",
652 .data = &init_net.ipv4.sysctl_icmp_ratelimit,
653 .maxlen = sizeof(int),
654 .mode = 0644,
655 .proc_handler = proc_dointvec_ms_jiffies,
656 },
657 {
658 .procname = "icmp_ratemask",
659 .data = &init_net.ipv4.sysctl_icmp_ratemask,
660 .maxlen = sizeof(int),
661 .mode = 0644,
662 .proc_handler = proc_dointvec
663 },
664 {
665 .procname = "ping_group_range",
666 .data = &init_net.ipv4.ping_group_range.range,
667 .maxlen = sizeof(gid_t)*2,
668 .mode = 0644,
669 .proc_handler = ipv4_ping_group_range,
670 },
671#ifdef CONFIG_NET_L3_MASTER_DEV
672 {
673 .procname = "raw_l3mdev_accept",
674 .data = &init_net.ipv4.sysctl_raw_l3mdev_accept,
675 .maxlen = sizeof(u8),
676 .mode = 0644,
677 .proc_handler = proc_dou8vec_minmax,
678 .extra1 = SYSCTL_ZERO,
679 .extra2 = SYSCTL_ONE,
680 },
681#endif
682 {
683 .procname = "tcp_ecn",
684 .data = &init_net.ipv4.sysctl_tcp_ecn,
685 .maxlen = sizeof(u8),
686 .mode = 0644,
687 .proc_handler = proc_dou8vec_minmax,
688 },
689 {
690 .procname = "tcp_ecn_fallback",
691 .data = &init_net.ipv4.sysctl_tcp_ecn_fallback,
692 .maxlen = sizeof(u8),
693 .mode = 0644,
694 .proc_handler = proc_dou8vec_minmax,
695 },
696 {
697 .procname = "ip_dynaddr",
698 .data = &init_net.ipv4.sysctl_ip_dynaddr,
699 .maxlen = sizeof(u8),
700 .mode = 0644,
701 .proc_handler = proc_dou8vec_minmax,
702 },
703 {
704 .procname = "ip_early_demux",
705 .data = &init_net.ipv4.sysctl_ip_early_demux,
706 .maxlen = sizeof(u8),
707 .mode = 0644,
708 .proc_handler = proc_dou8vec_minmax,
709 },
710 {
711 .procname = "udp_early_demux",
712 .data = &init_net.ipv4.sysctl_udp_early_demux,
713 .maxlen = sizeof(u8),
714 .mode = 0644,
715 .proc_handler = proc_udp_early_demux
716 },
717 {
718 .procname = "tcp_early_demux",
719 .data = &init_net.ipv4.sysctl_tcp_early_demux,
720 .maxlen = sizeof(u8),
721 .mode = 0644,
722 .proc_handler = proc_tcp_early_demux
723 },
724 {
725 .procname = "nexthop_compat_mode",
726 .data = &init_net.ipv4.sysctl_nexthop_compat_mode,
727 .maxlen = sizeof(u8),
728 .mode = 0644,
729 .proc_handler = proc_dou8vec_minmax,
730 .extra1 = SYSCTL_ZERO,
731 .extra2 = SYSCTL_ONE,
732 },
733 {
734 .procname = "ip_default_ttl",
735 .data = &init_net.ipv4.sysctl_ip_default_ttl,
736 .maxlen = sizeof(u8),
737 .mode = 0644,
738 .proc_handler = proc_dou8vec_minmax,
739 .extra1 = &ip_ttl_min,
740 .extra2 = &ip_ttl_max,
741 },
742 {
743 .procname = "ip_local_port_range",
744 .maxlen = sizeof(init_net.ipv4.ip_local_ports.range),
745 .data = &init_net.ipv4.ip_local_ports.range,
746 .mode = 0644,
747 .proc_handler = ipv4_local_port_range,
748 },
749 {
750 .procname = "ip_local_reserved_ports",
751 .data = &init_net.ipv4.sysctl_local_reserved_ports,
752 .maxlen = 65536,
753 .mode = 0644,
754 .proc_handler = proc_do_large_bitmap,
755 },
756 {
757 .procname = "ip_no_pmtu_disc",
758 .data = &init_net.ipv4.sysctl_ip_no_pmtu_disc,
759 .maxlen = sizeof(u8),
760 .mode = 0644,
761 .proc_handler = proc_dou8vec_minmax,
762 },
763 {
764 .procname = "ip_forward_use_pmtu",
765 .data = &init_net.ipv4.sysctl_ip_fwd_use_pmtu,
766 .maxlen = sizeof(u8),
767 .mode = 0644,
768 .proc_handler = proc_dou8vec_minmax,
769 },
770 {
771 .procname = "ip_forward_update_priority",
772 .data = &init_net.ipv4.sysctl_ip_fwd_update_priority,
773 .maxlen = sizeof(u8),
774 .mode = 0644,
775 .proc_handler = ipv4_fwd_update_priority,
776 .extra1 = SYSCTL_ZERO,
777 .extra2 = SYSCTL_ONE,
778 },
779 {
780 .procname = "ip_nonlocal_bind",
781 .data = &init_net.ipv4.sysctl_ip_nonlocal_bind,
782 .maxlen = sizeof(u8),
783 .mode = 0644,
784 .proc_handler = proc_dou8vec_minmax,
785 },
786 {
787 .procname = "ip_autobind_reuse",
788 .data = &init_net.ipv4.sysctl_ip_autobind_reuse,
789 .maxlen = sizeof(u8),
790 .mode = 0644,
791 .proc_handler = proc_dou8vec_minmax,
792 .extra1 = SYSCTL_ZERO,
793 .extra2 = SYSCTL_ONE,
794 },
795 {
796 .procname = "fwmark_reflect",
797 .data = &init_net.ipv4.sysctl_fwmark_reflect,
798 .maxlen = sizeof(u8),
799 .mode = 0644,
800 .proc_handler = proc_dou8vec_minmax,
801 },
802 {
803 .procname = "tcp_fwmark_accept",
804 .data = &init_net.ipv4.sysctl_tcp_fwmark_accept,
805 .maxlen = sizeof(u8),
806 .mode = 0644,
807 .proc_handler = proc_dou8vec_minmax,
808 },
809#ifdef CONFIG_NET_L3_MASTER_DEV
810 {
811 .procname = "tcp_l3mdev_accept",
812 .data = &init_net.ipv4.sysctl_tcp_l3mdev_accept,
813 .maxlen = sizeof(u8),
814 .mode = 0644,
815 .proc_handler = proc_dou8vec_minmax,
816 .extra1 = SYSCTL_ZERO,
817 .extra2 = SYSCTL_ONE,
818 },
819#endif
820 {
821 .procname = "tcp_mtu_probing",
822 .data = &init_net.ipv4.sysctl_tcp_mtu_probing,
823 .maxlen = sizeof(u8),
824 .mode = 0644,
825 .proc_handler = proc_dou8vec_minmax,
826 },
827 {
828 .procname = "tcp_base_mss",
829 .data = &init_net.ipv4.sysctl_tcp_base_mss,
830 .maxlen = sizeof(int),
831 .mode = 0644,
832 .proc_handler = proc_dointvec,
833 },
834 {
835 .procname = "tcp_min_snd_mss",
836 .data = &init_net.ipv4.sysctl_tcp_min_snd_mss,
837 .maxlen = sizeof(int),
838 .mode = 0644,
839 .proc_handler = proc_dointvec_minmax,
840 .extra1 = &tcp_min_snd_mss_min,
841 .extra2 = &tcp_min_snd_mss_max,
842 },
843 {
844 .procname = "tcp_mtu_probe_floor",
845 .data = &init_net.ipv4.sysctl_tcp_mtu_probe_floor,
846 .maxlen = sizeof(int),
847 .mode = 0644,
848 .proc_handler = proc_dointvec_minmax,
849 .extra1 = &tcp_min_snd_mss_min,
850 .extra2 = &tcp_min_snd_mss_max,
851 },
852 {
853 .procname = "tcp_probe_threshold",
854 .data = &init_net.ipv4.sysctl_tcp_probe_threshold,
855 .maxlen = sizeof(int),
856 .mode = 0644,
857 .proc_handler = proc_dointvec,
858 },
859 {
860 .procname = "tcp_probe_interval",
861 .data = &init_net.ipv4.sysctl_tcp_probe_interval,
862 .maxlen = sizeof(u32),
863 .mode = 0644,
864 .proc_handler = proc_douintvec_minmax,
865 .extra2 = &u32_max_div_HZ,
866 },
867 {
868 .procname = "igmp_link_local_mcast_reports",
869 .data = &init_net.ipv4.sysctl_igmp_llm_reports,
870 .maxlen = sizeof(u8),
871 .mode = 0644,
872 .proc_handler = proc_dou8vec_minmax,
873 },
874 {
875 .procname = "igmp_max_memberships",
876 .data = &init_net.ipv4.sysctl_igmp_max_memberships,
877 .maxlen = sizeof(int),
878 .mode = 0644,
879 .proc_handler = proc_dointvec
880 },
881 {
882 .procname = "igmp_max_msf",
883 .data = &init_net.ipv4.sysctl_igmp_max_msf,
884 .maxlen = sizeof(int),
885 .mode = 0644,
886 .proc_handler = proc_dointvec
887 },
888#ifdef CONFIG_IP_MULTICAST
889 {
890 .procname = "igmp_qrv",
891 .data = &init_net.ipv4.sysctl_igmp_qrv,
892 .maxlen = sizeof(int),
893 .mode = 0644,
894 .proc_handler = proc_dointvec_minmax,
895 .extra1 = SYSCTL_ONE
896 },
897#endif
898 {
899 .procname = "tcp_congestion_control",
900 .data = &init_net.ipv4.tcp_congestion_control,
901 .mode = 0644,
902 .maxlen = TCP_CA_NAME_MAX,
903 .proc_handler = proc_tcp_congestion_control,
904 },
905 {
906 .procname = "tcp_available_congestion_control",
907 .maxlen = TCP_CA_BUF_MAX,
908 .mode = 0444,
909 .proc_handler = proc_tcp_available_congestion_control,
910 },
911 {
912 .procname = "tcp_allowed_congestion_control",
913 .maxlen = TCP_CA_BUF_MAX,
914 .mode = 0644,
915 .proc_handler = proc_allowed_congestion_control,
916 },
917 {
918 .procname = "tcp_keepalive_time",
919 .data = &init_net.ipv4.sysctl_tcp_keepalive_time,
920 .maxlen = sizeof(int),
921 .mode = 0644,
922 .proc_handler = proc_dointvec_jiffies,
923 },
924 {
925 .procname = "tcp_keepalive_probes",
926 .data = &init_net.ipv4.sysctl_tcp_keepalive_probes,
927 .maxlen = sizeof(u8),
928 .mode = 0644,
929 .proc_handler = proc_dou8vec_minmax,
930 },
931 {
932 .procname = "tcp_keepalive_intvl",
933 .data = &init_net.ipv4.sysctl_tcp_keepalive_intvl,
934 .maxlen = sizeof(int),
935 .mode = 0644,
936 .proc_handler = proc_dointvec_jiffies,
937 },
938 {
939 .procname = "tcp_syn_retries",
940 .data = &init_net.ipv4.sysctl_tcp_syn_retries,
941 .maxlen = sizeof(u8),
942 .mode = 0644,
943 .proc_handler = proc_dou8vec_minmax,
944 .extra1 = &tcp_syn_retries_min,
945 .extra2 = &tcp_syn_retries_max
946 },
947 {
948 .procname = "tcp_synack_retries",
949 .data = &init_net.ipv4.sysctl_tcp_synack_retries,
950 .maxlen = sizeof(u8),
951 .mode = 0644,
952 .proc_handler = proc_dou8vec_minmax,
953 },
954#ifdef CONFIG_SYN_COOKIES
955 {
956 .procname = "tcp_syncookies",
957 .data = &init_net.ipv4.sysctl_tcp_syncookies,
958 .maxlen = sizeof(u8),
959 .mode = 0644,
960 .proc_handler = proc_dou8vec_minmax,
961 },
962#endif
963 {
964 .procname = "tcp_migrate_req",
965 .data = &init_net.ipv4.sysctl_tcp_migrate_req,
966 .maxlen = sizeof(u8),
967 .mode = 0644,
968 .proc_handler = proc_dou8vec_minmax,
969 .extra1 = SYSCTL_ZERO,
970 .extra2 = SYSCTL_ONE
971 },
972 {
973 .procname = "tcp_reordering",
974 .data = &init_net.ipv4.sysctl_tcp_reordering,
975 .maxlen = sizeof(int),
976 .mode = 0644,
977 .proc_handler = proc_dointvec
978 },
979 {
980 .procname = "tcp_retries1",
981 .data = &init_net.ipv4.sysctl_tcp_retries1,
982 .maxlen = sizeof(u8),
983 .mode = 0644,
984 .proc_handler = proc_dou8vec_minmax,
985 .extra2 = &tcp_retr1_max
986 },
987 {
988 .procname = "tcp_retries2",
989 .data = &init_net.ipv4.sysctl_tcp_retries2,
990 .maxlen = sizeof(u8),
991 .mode = 0644,
992 .proc_handler = proc_dou8vec_minmax,
993 },
994 {
995 .procname = "tcp_orphan_retries",
996 .data = &init_net.ipv4.sysctl_tcp_orphan_retries,
997 .maxlen = sizeof(u8),
998 .mode = 0644,
999 .proc_handler = proc_dou8vec_minmax,
1000 },
1001 {
1002 .procname = "tcp_fin_timeout",
1003 .data = &init_net.ipv4.sysctl_tcp_fin_timeout,
1004 .maxlen = sizeof(int),
1005 .mode = 0644,
1006 .proc_handler = proc_dointvec_jiffies,
1007 },
1008 {
1009 .procname = "tcp_notsent_lowat",
1010 .data = &init_net.ipv4.sysctl_tcp_notsent_lowat,
1011 .maxlen = sizeof(unsigned int),
1012 .mode = 0644,
1013 .proc_handler = proc_douintvec,
1014 },
1015 {
1016 .procname = "tcp_tw_reuse",
1017 .data = &init_net.ipv4.sysctl_tcp_tw_reuse,
1018 .maxlen = sizeof(u8),
1019 .mode = 0644,
1020 .proc_handler = proc_dou8vec_minmax,
1021 .extra1 = SYSCTL_ZERO,
1022 .extra2 = &two,
1023 },
1024 {
1025 .procname = "tcp_max_tw_buckets",
1026 .data = &init_net.ipv4.tcp_death_row.sysctl_max_tw_buckets,
1027 .maxlen = sizeof(int),
1028 .mode = 0644,
1029 .proc_handler = proc_dointvec
1030 },
1031 {
1032 .procname = "tcp_max_syn_backlog",
1033 .data = &init_net.ipv4.sysctl_max_syn_backlog,
1034 .maxlen = sizeof(int),
1035 .mode = 0644,
1036 .proc_handler = proc_dointvec
1037 },
1038 {
1039 .procname = "tcp_fastopen",
1040 .data = &init_net.ipv4.sysctl_tcp_fastopen,
1041 .maxlen = sizeof(int),
1042 .mode = 0644,
1043 .proc_handler = proc_dointvec,
1044 },
1045 {
1046 .procname = "tcp_fastopen_key",
1047 .mode = 0600,
1048 .data = &init_net.ipv4.sysctl_tcp_fastopen,
1049 /* maxlen to print the list of keys in hex (*2), with dashes
1050 * separating doublewords and a comma in between keys.
1051 */
1052 .maxlen = ((TCP_FASTOPEN_KEY_LENGTH *
1053 2 * TCP_FASTOPEN_KEY_MAX) +
1054 (TCP_FASTOPEN_KEY_MAX * 5)),
1055 .proc_handler = proc_tcp_fastopen_key,
1056 },
1057 {
1058 .procname = "tcp_fastopen_blackhole_timeout_sec",
1059 .data = &init_net.ipv4.sysctl_tcp_fastopen_blackhole_timeout,
1060 .maxlen = sizeof(int),
1061 .mode = 0644,
1062 .proc_handler = proc_tfo_blackhole_detect_timeout,
1063 .extra1 = SYSCTL_ZERO,
1064 },
1065#ifdef CONFIG_IP_ROUTE_MULTIPATH
1066 {
1067 .procname = "fib_multipath_use_neigh",
1068 .data = &init_net.ipv4.sysctl_fib_multipath_use_neigh,
1069 .maxlen = sizeof(u8),
1070 .mode = 0644,
1071 .proc_handler = proc_dou8vec_minmax,
1072 .extra1 = SYSCTL_ZERO,
1073 .extra2 = SYSCTL_ONE,
1074 },
1075 {
1076 .procname = "fib_multipath_hash_policy",
1077 .data = &init_net.ipv4.sysctl_fib_multipath_hash_policy,
1078 .maxlen = sizeof(u8),
1079 .mode = 0644,
1080 .proc_handler = proc_fib_multipath_hash_policy,
1081 .extra1 = SYSCTL_ZERO,
1082 .extra2 = &three,
1083 },
1084 {
1085 .procname = "fib_multipath_hash_fields",
1086 .data = &init_net.ipv4.sysctl_fib_multipath_hash_fields,
1087 .maxlen = sizeof(u32),
1088 .mode = 0644,
1089 .proc_handler = proc_fib_multipath_hash_fields,
1090 .extra1 = SYSCTL_ONE,
1091 .extra2 = &fib_multipath_hash_fields_all_mask,
1092 },
1093#endif
1094 {
1095 .procname = "ip_unprivileged_port_start",
1096 .maxlen = sizeof(int),
1097 .data = &init_net.ipv4.sysctl_ip_prot_sock,
1098 .mode = 0644,
1099 .proc_handler = ipv4_privileged_ports,
1100 },
1101#ifdef CONFIG_NET_L3_MASTER_DEV
1102 {
1103 .procname = "udp_l3mdev_accept",
1104 .data = &init_net.ipv4.sysctl_udp_l3mdev_accept,
1105 .maxlen = sizeof(u8),
1106 .mode = 0644,
1107 .proc_handler = proc_dou8vec_minmax,
1108 .extra1 = SYSCTL_ZERO,
1109 .extra2 = SYSCTL_ONE,
1110 },
1111#endif
1112 {
1113 .procname = "tcp_sack",
1114 .data = &init_net.ipv4.sysctl_tcp_sack,
1115 .maxlen = sizeof(u8),
1116 .mode = 0644,
1117 .proc_handler = proc_dou8vec_minmax,
1118 },
1119 {
1120 .procname = "tcp_window_scaling",
1121 .data = &init_net.ipv4.sysctl_tcp_window_scaling,
1122 .maxlen = sizeof(u8),
1123 .mode = 0644,
1124 .proc_handler = proc_dou8vec_minmax,
1125 },
1126 {
1127 .procname = "tcp_timestamps",
1128 .data = &init_net.ipv4.sysctl_tcp_timestamps,
1129 .maxlen = sizeof(u8),
1130 .mode = 0644,
1131 .proc_handler = proc_dou8vec_minmax,
1132 },
1133 {
1134 .procname = "tcp_early_retrans",
1135 .data = &init_net.ipv4.sysctl_tcp_early_retrans,
1136 .maxlen = sizeof(u8),
1137 .mode = 0644,
1138 .proc_handler = proc_dou8vec_minmax,
1139 .extra1 = SYSCTL_ZERO,
1140 .extra2 = &four,
1141 },
1142 {
1143 .procname = "tcp_recovery",
1144 .data = &init_net.ipv4.sysctl_tcp_recovery,
1145 .maxlen = sizeof(u8),
1146 .mode = 0644,
1147 .proc_handler = proc_dou8vec_minmax,
1148 },
1149 {
1150 .procname = "tcp_thin_linear_timeouts",
1151 .data = &init_net.ipv4.sysctl_tcp_thin_linear_timeouts,
1152 .maxlen = sizeof(u8),
1153 .mode = 0644,
1154 .proc_handler = proc_dou8vec_minmax,
1155 },
1156 {
1157 .procname = "tcp_slow_start_after_idle",
1158 .data = &init_net.ipv4.sysctl_tcp_slow_start_after_idle,
1159 .maxlen = sizeof(u8),
1160 .mode = 0644,
1161 .proc_handler = proc_dou8vec_minmax,
1162 },
1163 {
1164 .procname = "tcp_retrans_collapse",
1165 .data = &init_net.ipv4.sysctl_tcp_retrans_collapse,
1166 .maxlen = sizeof(u8),
1167 .mode = 0644,
1168 .proc_handler = proc_dou8vec_minmax,
1169 },
1170 {
1171 .procname = "tcp_stdurg",
1172 .data = &init_net.ipv4.sysctl_tcp_stdurg,
1173 .maxlen = sizeof(u8),
1174 .mode = 0644,
1175 .proc_handler = proc_dou8vec_minmax,
1176 },
1177 {
1178 .procname = "tcp_rfc1337",
1179 .data = &init_net.ipv4.sysctl_tcp_rfc1337,
1180 .maxlen = sizeof(u8),
1181 .mode = 0644,
1182 .proc_handler = proc_dou8vec_minmax,
1183 },
1184 {
1185 .procname = "tcp_abort_on_overflow",
1186 .data = &init_net.ipv4.sysctl_tcp_abort_on_overflow,
1187 .maxlen = sizeof(u8),
1188 .mode = 0644,
1189 .proc_handler = proc_dou8vec_minmax,
1190 },
1191 {
1192 .procname = "tcp_fack",
1193 .data = &init_net.ipv4.sysctl_tcp_fack,
1194 .maxlen = sizeof(u8),
1195 .mode = 0644,
1196 .proc_handler = proc_dou8vec_minmax,
1197 },
1198 {
1199 .procname = "tcp_max_reordering",
1200 .data = &init_net.ipv4.sysctl_tcp_max_reordering,
1201 .maxlen = sizeof(int),
1202 .mode = 0644,
1203 .proc_handler = proc_dointvec
1204 },
1205 {
1206 .procname = "tcp_dsack",
1207 .data = &init_net.ipv4.sysctl_tcp_dsack,
1208 .maxlen = sizeof(u8),
1209 .mode = 0644,
1210 .proc_handler = proc_dou8vec_minmax,
1211 },
1212 {
1213 .procname = "tcp_app_win",
1214 .data = &init_net.ipv4.sysctl_tcp_app_win,
1215 .maxlen = sizeof(u8),
1216 .mode = 0644,
1217 .proc_handler = proc_dou8vec_minmax,
1218 },
1219 {
1220 .procname = "tcp_adv_win_scale",
1221 .data = &init_net.ipv4.sysctl_tcp_adv_win_scale,
1222 .maxlen = sizeof(int),
1223 .mode = 0644,
1224 .proc_handler = proc_dointvec_minmax,
1225 .extra1 = &tcp_adv_win_scale_min,
1226 .extra2 = &tcp_adv_win_scale_max,
1227 },
1228 {
1229 .procname = "tcp_frto",
1230 .data = &init_net.ipv4.sysctl_tcp_frto,
1231 .maxlen = sizeof(u8),
1232 .mode = 0644,
1233 .proc_handler = proc_dou8vec_minmax,
1234 },
1235 {
1236 .procname = "tcp_no_metrics_save",
1237 .data = &init_net.ipv4.sysctl_tcp_nometrics_save,
1238 .maxlen = sizeof(u8),
1239 .mode = 0644,
1240 .proc_handler = proc_dou8vec_minmax,
1241 },
1242 {
1243 .procname = "tcp_no_ssthresh_metrics_save",
1244 .data = &init_net.ipv4.sysctl_tcp_no_ssthresh_metrics_save,
1245 .maxlen = sizeof(u8),
1246 .mode = 0644,
1247 .proc_handler = proc_dou8vec_minmax,
1248 .extra1 = SYSCTL_ZERO,
1249 .extra2 = SYSCTL_ONE,
1250 },
1251 {
1252 .procname = "tcp_moderate_rcvbuf",
1253 .data = &init_net.ipv4.sysctl_tcp_moderate_rcvbuf,
1254 .maxlen = sizeof(u8),
1255 .mode = 0644,
1256 .proc_handler = proc_dou8vec_minmax,
1257 },
1258 {
1259 .procname = "tcp_tso_win_divisor",
1260 .data = &init_net.ipv4.sysctl_tcp_tso_win_divisor,
1261 .maxlen = sizeof(u8),
1262 .mode = 0644,
1263 .proc_handler = proc_dou8vec_minmax,
1264 },
1265 {
1266 .procname = "tcp_workaround_signed_windows",
1267 .data = &init_net.ipv4.sysctl_tcp_workaround_signed_windows,
1268 .maxlen = sizeof(u8),
1269 .mode = 0644,
1270 .proc_handler = proc_dou8vec_minmax,
1271 },
1272 {
1273 .procname = "tcp_limit_output_bytes",
1274 .data = &init_net.ipv4.sysctl_tcp_limit_output_bytes,
1275 .maxlen = sizeof(int),
1276 .mode = 0644,
1277 .proc_handler = proc_dointvec
1278 },
1279 {
1280 .procname = "tcp_challenge_ack_limit",
1281 .data = &init_net.ipv4.sysctl_tcp_challenge_ack_limit,
1282 .maxlen = sizeof(int),
1283 .mode = 0644,
1284 .proc_handler = proc_dointvec
1285 },
1286 {
1287 .procname = "tcp_min_tso_segs",
1288 .data = &init_net.ipv4.sysctl_tcp_min_tso_segs,
1289 .maxlen = sizeof(u8),
1290 .mode = 0644,
1291 .proc_handler = proc_dou8vec_minmax,
1292 .extra1 = SYSCTL_ONE,
1293 },
1294 {
1295 .procname = "tcp_min_rtt_wlen",
1296 .data = &init_net.ipv4.sysctl_tcp_min_rtt_wlen,
1297 .maxlen = sizeof(int),
1298 .mode = 0644,
1299 .proc_handler = proc_dointvec_minmax,
1300 .extra1 = SYSCTL_ZERO,
1301 .extra2 = &one_day_secs
1302 },
1303 {
1304 .procname = "tcp_autocorking",
1305 .data = &init_net.ipv4.sysctl_tcp_autocorking,
1306 .maxlen = sizeof(u8),
1307 .mode = 0644,
1308 .proc_handler = proc_dou8vec_minmax,
1309 .extra1 = SYSCTL_ZERO,
1310 .extra2 = SYSCTL_ONE,
1311 },
1312 {
1313 .procname = "tcp_invalid_ratelimit",
1314 .data = &init_net.ipv4.sysctl_tcp_invalid_ratelimit,
1315 .maxlen = sizeof(int),
1316 .mode = 0644,
1317 .proc_handler = proc_dointvec_ms_jiffies,
1318 },
1319 {
1320 .procname = "tcp_pacing_ss_ratio",
1321 .data = &init_net.ipv4.sysctl_tcp_pacing_ss_ratio,
1322 .maxlen = sizeof(int),
1323 .mode = 0644,
1324 .proc_handler = proc_dointvec_minmax,
1325 .extra1 = SYSCTL_ZERO,
1326 .extra2 = &thousand,
1327 },
1328 {
1329 .procname = "tcp_pacing_ca_ratio",
1330 .data = &init_net.ipv4.sysctl_tcp_pacing_ca_ratio,
1331 .maxlen = sizeof(int),
1332 .mode = 0644,
1333 .proc_handler = proc_dointvec_minmax,
1334 .extra1 = SYSCTL_ZERO,
1335 .extra2 = &thousand,
1336 },
1337 {
1338 .procname = "tcp_wmem",
1339 .data = &init_net.ipv4.sysctl_tcp_wmem,
1340 .maxlen = sizeof(init_net.ipv4.sysctl_tcp_wmem),
1341 .mode = 0644,
1342 .proc_handler = proc_dointvec_minmax,
1343 .extra1 = SYSCTL_ONE,
1344 },
1345 {
1346 .procname = "tcp_rmem",
1347 .data = &init_net.ipv4.sysctl_tcp_rmem,
1348 .maxlen = sizeof(init_net.ipv4.sysctl_tcp_rmem),
1349 .mode = 0644,
1350 .proc_handler = proc_dointvec_minmax,
1351 .extra1 = SYSCTL_ONE,
1352 },
1353 {
1354 .procname = "tcp_comp_sack_delay_ns",
1355 .data = &init_net.ipv4.sysctl_tcp_comp_sack_delay_ns,
1356 .maxlen = sizeof(unsigned long),
1357 .mode = 0644,
1358 .proc_handler = proc_doulongvec_minmax,
1359 },
1360 {
1361 .procname = "tcp_comp_sack_slack_ns",
1362 .data = &init_net.ipv4.sysctl_tcp_comp_sack_slack_ns,
1363 .maxlen = sizeof(unsigned long),
1364 .mode = 0644,
1365 .proc_handler = proc_doulongvec_minmax,
1366 },
1367 {
1368 .procname = "tcp_comp_sack_nr",
1369 .data = &init_net.ipv4.sysctl_tcp_comp_sack_nr,
1370 .maxlen = sizeof(u8),
1371 .mode = 0644,
1372 .proc_handler = proc_dou8vec_minmax,
1373 .extra1 = SYSCTL_ZERO,
1374 },
1375 {
1376 .procname = "tcp_reflect_tos",
1377 .data = &init_net.ipv4.sysctl_tcp_reflect_tos,
1378 .maxlen = sizeof(u8),
1379 .mode = 0644,
1380 .proc_handler = proc_dou8vec_minmax,
1381 .extra1 = SYSCTL_ZERO,
1382 .extra2 = SYSCTL_ONE,
1383 },
1384 {
1385 .procname = "udp_rmem_min",
1386 .data = &init_net.ipv4.sysctl_udp_rmem_min,
1387 .maxlen = sizeof(init_net.ipv4.sysctl_udp_rmem_min),
1388 .mode = 0644,
1389 .proc_handler = proc_dointvec_minmax,
1390 .extra1 = SYSCTL_ONE
1391 },
1392 {
1393 .procname = "udp_wmem_min",
1394 .data = &init_net.ipv4.sysctl_udp_wmem_min,
1395 .maxlen = sizeof(init_net.ipv4.sysctl_udp_wmem_min),
1396 .mode = 0644,
1397 .proc_handler = proc_dointvec_minmax,
1398 .extra1 = SYSCTL_ONE
1399 },
1400 {
1401 .procname = "fib_notify_on_flag_change",
1402 .data = &init_net.ipv4.sysctl_fib_notify_on_flag_change,
1403 .maxlen = sizeof(u8),
1404 .mode = 0644,
1405 .proc_handler = proc_dou8vec_minmax,
1406 .extra1 = SYSCTL_ZERO,
1407 .extra2 = &two,
1408 },
1409 { }
1410};
1411
1412static __net_init int ipv4_sysctl_init_net(struct net *net)
1413{
1414 struct ctl_table *table;
1415
1416 table = ipv4_net_table;
1417 if (!net_eq(net, &init_net)) {
1418 int i;
1419
1420 table = kmemdup(table, sizeof(ipv4_net_table), GFP_KERNEL);
1421 if (!table)
1422 goto err_alloc;
1423
1424 for (i = 0; i < ARRAY_SIZE(ipv4_net_table) - 1; i++) {
1425 if (table[i].data) {
1426 /* Update the variables to point into
1427 * the current struct net
1428 */
1429 table[i].data += (void *)net - (void *)&init_net;
1430 } else {
1431 /* Entries without data pointer are global;
1432 * Make them read-only in non-init_net ns
1433 */
1434 table[i].mode &= ~0222;
1435 }
1436 }
1437 }
1438
1439 net->ipv4.ipv4_hdr = register_net_sysctl(net, "net/ipv4", table);
1440 if (!net->ipv4.ipv4_hdr)
1441 goto err_reg;
1442
1443 net->ipv4.sysctl_local_reserved_ports = kzalloc(65536 / 8, GFP_KERNEL);
1444 if (!net->ipv4.sysctl_local_reserved_ports)
1445 goto err_ports;
1446
1447 return 0;
1448
1449err_ports:
1450 unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
1451err_reg:
1452 if (!net_eq(net, &init_net))
1453 kfree(table);
1454err_alloc:
1455 return -ENOMEM;
1456}
1457
1458static __net_exit void ipv4_sysctl_exit_net(struct net *net)
1459{
1460 struct ctl_table *table;
1461
1462 kfree(net->ipv4.sysctl_local_reserved_ports);
1463 table = net->ipv4.ipv4_hdr->ctl_table_arg;
1464 unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
1465 kfree(table);
1466}
1467
1468static __net_initdata struct pernet_operations ipv4_sysctl_ops = {
1469 .init = ipv4_sysctl_init_net,
1470 .exit = ipv4_sysctl_exit_net,
1471};
1472
1473static __init int sysctl_ipv4_init(void)
1474{
1475 struct ctl_table_header *hdr;
1476
1477 hdr = register_net_sysctl(&init_net, "net/ipv4", ipv4_table);
1478 if (!hdr)
1479 return -ENOMEM;
1480
1481 if (register_pernet_subsys(&ipv4_sysctl_ops)) {
1482 unregister_net_sysctl_table(hdr);
1483 return -ENOMEM;
1484 }
1485
1486 return 0;
1487}
1488
1489__initcall(sysctl_ipv4_init);