]> git.proxmox.com Git - mirror_ubuntu-bionic-kernel.git/blob - net/ipv4/sysctl_net_ipv4.c
projects / mirror_ubuntu-bionic-kernel.git / blob
? search:


2fd25d8c44d608d231be5c0688ba8aa7b5b6b250
[mirror_ubuntu-bionic-kernel.git] / net / ipv4 / sysctl_net_ipv4.c
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/route.h>
23 #include <net/tcp.h>
24 #include <net/udp.h>
25 #include <net/cipso_ipv4.h>
26 #include <net/inet_frag.h>
27 #include <net/ping.h>
28 #include <net/protocol.h>
29 #include <net/netevent.h>
30
31 static int zero;
32 static int one = 1;
33 static int four = 4;
34 static int thousand = 1000;
35 static int gso_max_segs = GSO_MAX_SEGS;
36 static int tcp_retr1_max = 255;
37 static int ip_local_port_range_min[] = { 1, 1 };
38 static int ip_local_port_range_max[] = { 65535, 65535 };
39 static int tcp_adv_win_scale_min = -31;
40 static int tcp_adv_win_scale_max = 31;
41 static int ip_privileged_port_min;
42 static int ip_privileged_port_max = 65535;
43 static int ip_ttl_min = 1;
44 static int ip_ttl_max = 255;
45 static int tcp_syn_retries_min = 1;
46 static int tcp_syn_retries_max = MAX_TCP_SYNCNT;
47 static int ip_ping_group_range_min[] = { 0, 0 };
48 static int ip_ping_group_range_max[] = { GID_T_MAX, GID_T_MAX };
49
50 /* obsolete */
51 static int sysctl_tcp_low_latency __read_mostly;
52
53 /* Update system visible IP port range */
54 static void set_local_port_range(struct net *net, int range[2])
55 {
56 bool same_parity = !((range[0] ^ range[1]) & 1);
57
58 write_seqlock_bh(&net->ipv4.ip_local_ports.lock);
59 if (same_parity && !net->ipv4.ip_local_ports.warned) {
60 net->ipv4.ip_local_ports.warned = true;
61 pr_err_ratelimited("ip_local_port_range: prefer different parity for start/end values.\n");
62 }
63 net->ipv4.ip_local_ports.range[0] = range[0];
64 net->ipv4.ip_local_ports.range[1] = range[1];
65 write_sequnlock_bh(&net->ipv4.ip_local_ports.lock);
66 }
67
68 /* Validate changes from /proc interface. */
69 static int ipv4_local_port_range(struct ctl_table *table, int write,
70 void __user *buffer,
71 size_t *lenp, loff_t *ppos)
72 {
73 struct net *net =
74 container_of(table->data, struct net, ipv4.ip_local_ports.range);
75 int ret;
76 int range[2];
77 struct ctl_table tmp = {
78 .data = &range,
79 .maxlen = sizeof(range),
80 .mode = table->mode,
81 .extra1 = &ip_local_port_range_min,
82 .extra2 = &ip_local_port_range_max,
83 };
84
85 inet_get_local_port_range(net, &range[0], &range[1]);
86
87 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
88
89 if (write && ret == 0) {
90 /* Ensure that the upper limit is not smaller than the lower,
91 * and that the lower does not encroach upon the privileged
92 * port limit.
93 */
94 if ((range[1] < range[0]) ||
95 (range[0] < net->ipv4.sysctl_ip_prot_sock))
96 ret = -EINVAL;
97 else
98 set_local_port_range(net, range);
99 }
100
101 return ret;
102 }
103
104 /* Validate changes from /proc interface. */
105 static int ipv4_privileged_ports(struct ctl_table *table, int write,
106 void __user *buffer, size_t *lenp, loff_t *ppos)
107 {
108 struct net *net = container_of(table->data, struct net,
109 ipv4.sysctl_ip_prot_sock);
110 int ret;
111 int pports;
112 int range[2];
113 struct ctl_table tmp = {
114 .data = &pports,
115 .maxlen = sizeof(pports),
116 .mode = table->mode,
117 .extra1 = &ip_privileged_port_min,
118 .extra2 = &ip_privileged_port_max,
119 };
120
121 pports = net->ipv4.sysctl_ip_prot_sock;
122
123 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
124
125 if (write && ret == 0) {
126 inet_get_local_port_range(net, &range[0], &range[1]);
127 /* Ensure that the local port range doesn't overlap with the
128 * privileged port range.
129 */
130 if (range[0] < pports)
131 ret = -EINVAL;
132 else
133 net->ipv4.sysctl_ip_prot_sock = pports;
134 }
135
136 return ret;
137 }
138
139 static void inet_get_ping_group_range_table(struct ctl_table *table, kgid_t *low, kgid_t *high)
140 {
141 kgid_t *data = table->data;
142 struct net *net =
143 container_of(table->data, struct net, ipv4.ping_group_range.range);
144 unsigned int seq;
145 do {
146 seq = read_seqbegin(&net->ipv4.ping_group_range.lock);
147
148 *low = data[0];
149 *high = data[1];
150 } while (read_seqretry(&net->ipv4.ping_group_range.lock, seq));
151 }
152
153 /* Update system visible IP port range */
154 static void set_ping_group_range(struct ctl_table *table, kgid_t low, kgid_t high)
155 {
156 kgid_t *data = table->data;
157 struct net *net =
158 container_of(table->data, struct net, ipv4.ping_group_range.range);
159 write_seqlock(&net->ipv4.ping_group_range.lock);
160 data[0] = low;
161 data[1] = high;
162 write_sequnlock(&net->ipv4.ping_group_range.lock);
163 }
164
165 /* Validate changes from /proc interface. */
166 static int ipv4_ping_group_range(struct ctl_table *table, int write,
167 void __user *buffer,
168 size_t *lenp, loff_t *ppos)
169 {
170 struct user_namespace *user_ns = current_user_ns();
171 int ret;
172 gid_t urange[2];
173 kgid_t low, high;
174 struct ctl_table tmp = {
175 .data = &urange,
176 .maxlen = sizeof(urange),
177 .mode = table->mode,
178 .extra1 = &ip_ping_group_range_min,
179 .extra2 = &ip_ping_group_range_max,
180 };
181
182 inet_get_ping_group_range_table(table, &low, &high);
183 urange[0] = from_kgid_munged(user_ns, low);
184 urange[1] = from_kgid_munged(user_ns, high);
185 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
186
187 if (write && ret == 0) {
188 low = make_kgid(user_ns, urange[0]);
189 high = make_kgid(user_ns, urange[1]);
190 if (!gid_valid(low) || !gid_valid(high))
191 return -EINVAL;
192 if (urange[1] < urange[0] || gid_lt(high, low)) {
193 low = make_kgid(&init_user_ns, 1);
194 high = make_kgid(&init_user_ns, 0);
195 }
196 set_ping_group_range(table, low, high);
197 }
198
199 return ret;
200 }
201
202 static int proc_tcp_congestion_control(struct ctl_table *ctl, int write,
203 void __user *buffer, size_t *lenp, loff_t *ppos)
204 {
205 struct net *net = container_of(ctl->data, struct net,
206 ipv4.tcp_congestion_control);
207 char val[TCP_CA_NAME_MAX];
208 struct ctl_table tbl = {
209 .data = val,
210 .maxlen = TCP_CA_NAME_MAX,
211 };
212 int ret;
213
214 tcp_get_default_congestion_control(net, val);
215
216 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
217 if (write && ret == 0)
218 ret = tcp_set_default_congestion_control(net, val);
219 return ret;
220 }
221
222 static int proc_tcp_available_congestion_control(struct ctl_table *ctl,
223 int write,
224 void __user *buffer, size_t *lenp,
225 loff_t *ppos)
226 {
227 struct ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX, };
228 int ret;
229
230 tbl.data = kmalloc(tbl.maxlen, GFP_USER);
231 if (!tbl.data)
232 return -ENOMEM;
233 tcp_get_available_congestion_control(tbl.data, TCP_CA_BUF_MAX);
234 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
235 kfree(tbl.data);
236 return ret;
237 }
238
239 static int proc_allowed_congestion_control(struct ctl_table *ctl,
240 int write,
241 void __user *buffer, size_t *lenp,
242 loff_t *ppos)
243 {
244 struct ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX };
245 int ret;
246
247 tbl.data = kmalloc(tbl.maxlen, GFP_USER);
248 if (!tbl.data)
249 return -ENOMEM;
250
251 tcp_get_allowed_congestion_control(tbl.data, tbl.maxlen);
252 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
253 if (write && ret == 0)
254 ret = tcp_set_allowed_congestion_control(tbl.data);
255 kfree(tbl.data);
256 return ret;
257 }
258
259 static int proc_tcp_fastopen_key(struct ctl_table *table, int write,
260 void __user *buffer, size_t *lenp,
261 loff_t *ppos)
262 {
263 struct net *net = container_of(table->data, struct net,
264 ipv4.sysctl_tcp_fastopen);
265 struct ctl_table tbl = { .maxlen = (TCP_FASTOPEN_KEY_LENGTH * 2 + 10) };
266 struct tcp_fastopen_context *ctxt;
267 u32 user_key[4]; /* 16 bytes, matching TCP_FASTOPEN_KEY_LENGTH */
268 __le32 key[4];
269 int ret, i;
270
271 tbl.data = kmalloc(tbl.maxlen, GFP_KERNEL);
272 if (!tbl.data)
273 return -ENOMEM;
274
275 rcu_read_lock();
276 ctxt = rcu_dereference(net->ipv4.tcp_fastopen_ctx);
277 if (ctxt)
278 memcpy(key, ctxt->key, TCP_FASTOPEN_KEY_LENGTH);
279 else
280 memset(key, 0, sizeof(key));
281 rcu_read_unlock();
282
283 for (i = 0; i < ARRAY_SIZE(key); i++)
284 user_key[i] = le32_to_cpu(key[i]);
285
286 snprintf(tbl.data, tbl.maxlen, "%08x-%08x-%08x-%08x",
287 user_key[0], user_key[1], user_key[2], user_key[3]);
288 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
289
290 if (write && ret == 0) {
291 if (sscanf(tbl.data, "%x-%x-%x-%x", user_key, user_key + 1,
292 user_key + 2, user_key + 3) != 4) {
293 ret = -EINVAL;
294 goto bad_key;
295 }
296
297 for (i = 0; i < ARRAY_SIZE(user_key); i++)
298 key[i] = cpu_to_le32(user_key[i]);
299
300 tcp_fastopen_reset_cipher(net, NULL, key,
301 TCP_FASTOPEN_KEY_LENGTH);
302 }
303
304 bad_key:
305 pr_debug("proc FO key set 0x%x-%x-%x-%x <- 0x%s: %u\n",
306 user_key[0], user_key[1], user_key[2], user_key[3],
307 (char *)tbl.data, ret);
308 kfree(tbl.data);
309 return ret;
310 }
311
312 static void proc_configure_early_demux(int enabled, int protocol)
313 {
314 struct net_protocol *ipprot;
315 #if IS_ENABLED(CONFIG_IPV6)
316 struct inet6_protocol *ip6prot;
317 #endif
318
319 rcu_read_lock();
320
321 ipprot = rcu_dereference(inet_protos[protocol]);
322 if (ipprot)
323 ipprot->early_demux = enabled ? ipprot->early_demux_handler :
324 NULL;
325
326 #if IS_ENABLED(CONFIG_IPV6)
327 ip6prot = rcu_dereference(inet6_protos[protocol]);
328 if (ip6prot)
329 ip6prot->early_demux = enabled ? ip6prot->early_demux_handler :
330 NULL;
331 #endif
332 rcu_read_unlock();
333 }
334
335 static int proc_tcp_early_demux(struct ctl_table *table, int write,
336 void __user *buffer, size_t *lenp, loff_t *ppos)
337 {
338 int ret = 0;
339
340 ret = proc_dointvec(table, write, buffer, lenp, ppos);
341
342 if (write && !ret) {
343 int enabled = init_net.ipv4.sysctl_tcp_early_demux;
344
345 proc_configure_early_demux(enabled, IPPROTO_TCP);
346 }
347
348 return ret;
349 }
350
351 static int proc_udp_early_demux(struct ctl_table *table, int write,
352 void __user *buffer, size_t *lenp, loff_t *ppos)
353 {
354 int ret = 0;
355
356 ret = proc_dointvec(table, write, buffer, lenp, ppos);
357
358 if (write && !ret) {
359 int enabled = init_net.ipv4.sysctl_udp_early_demux;
360
361 proc_configure_early_demux(enabled, IPPROTO_UDP);
362 }
363
364 return ret;
365 }
366
367 static int proc_tfo_blackhole_detect_timeout(struct ctl_table *table,
368 int write,
369 void __user *buffer,
370 size_t *lenp, loff_t *ppos)
371 {
372 struct net *net = container_of(table->data, struct net,
373 ipv4.sysctl_tcp_fastopen_blackhole_timeout);
374 int ret;
375
376 ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
377 if (write && ret == 0)
378 atomic_set(&net->ipv4.tfo_active_disable_times, 0);
379
380 return ret;
381 }
382
383 static int proc_tcp_available_ulp(struct ctl_table *ctl,
384 int write,
385 void __user *buffer, size_t *lenp,
386 loff_t *ppos)
387 {
388 struct ctl_table tbl = { .maxlen = TCP_ULP_BUF_MAX, };
389 int ret;
390
391 tbl.data = kmalloc(tbl.maxlen, GFP_USER);
392 if (!tbl.data)
393 return -ENOMEM;
394 tcp_get_available_ulp(tbl.data, TCP_ULP_BUF_MAX);
395 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
396 kfree(tbl.data);
397
398 return ret;
399 }
400
401 #ifdef CONFIG_IP_ROUTE_MULTIPATH
402 static int proc_fib_multipath_hash_policy(struct ctl_table *table, int write,
403 void __user *buffer, size_t *lenp,
404 loff_t *ppos)
405 {
406 struct net *net = container_of(table->data, struct net,
407 ipv4.sysctl_fib_multipath_hash_policy);
408 int ret;
409
410 ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
411 if (write && ret == 0)
412 call_netevent_notifiers(NETEVENT_MULTIPATH_HASH_UPDATE, net);
413
414 return ret;
415 }
416 #endif
417
418 static struct ctl_table ipv4_table[] = {
419 {
420 .procname = "tcp_max_orphans",
421 .data = &sysctl_tcp_max_orphans,
422 .maxlen = sizeof(int),
423 .mode = 0644,
424 .proc_handler = proc_dointvec
425 },
426 {
427 .procname = "inet_peer_threshold",
428 .data = &inet_peer_threshold,
429 .maxlen = sizeof(int),
430 .mode = 0644,
431 .proc_handler = proc_dointvec
432 },
433 {
434 .procname = "inet_peer_minttl",
435 .data = &inet_peer_minttl,
436 .maxlen = sizeof(int),
437 .mode = 0644,
438 .proc_handler = proc_dointvec_jiffies,
439 },
440 {
441 .procname = "inet_peer_maxttl",
442 .data = &inet_peer_maxttl,
443 .maxlen = sizeof(int),
444 .mode = 0644,
445 .proc_handler = proc_dointvec_jiffies,
446 },
447 {
448 .procname = "tcp_mem",
449 .maxlen = sizeof(sysctl_tcp_mem),
450 .data = &sysctl_tcp_mem,
451 .mode = 0644,
452 .proc_handler = proc_doulongvec_minmax,
453 },
454 {
455 .procname = "tcp_low_latency",
456 .data = &sysctl_tcp_low_latency,
457 .maxlen = sizeof(int),
458 .mode = 0644,
459 .proc_handler = proc_dointvec
460 },
461 #ifdef CONFIG_NETLABEL
462 {
463 .procname = "cipso_cache_enable",
464 .data = &cipso_v4_cache_enabled,
465 .maxlen = sizeof(int),
466 .mode = 0644,
467 .proc_handler = proc_dointvec,
468 },
469 {
470 .procname = "cipso_cache_bucket_size",
471 .data = &cipso_v4_cache_bucketsize,
472 .maxlen = sizeof(int),
473 .mode = 0644,
474 .proc_handler = proc_dointvec,
475 },
476 {
477 .procname = "cipso_rbm_optfmt",
478 .data = &cipso_v4_rbm_optfmt,
479 .maxlen = sizeof(int),
480 .mode = 0644,
481 .proc_handler = proc_dointvec,
482 },
483 {
484 .procname = "cipso_rbm_strictvalid",
485 .data = &cipso_v4_rbm_strictvalid,
486 .maxlen = sizeof(int),
487 .mode = 0644,
488 .proc_handler = proc_dointvec,
489 },
490 #endif /* CONFIG_NETLABEL */
491 {
492 .procname = "tcp_available_congestion_control",
493 .maxlen = TCP_CA_BUF_MAX,
494 .mode = 0444,
495 .proc_handler = proc_tcp_available_congestion_control,
496 },
497 {
498 .procname = "tcp_allowed_congestion_control",
499 .maxlen = TCP_CA_BUF_MAX,
500 .mode = 0644,
501 .proc_handler = proc_allowed_congestion_control,
502 },
503 {
504 .procname = "tcp_available_ulp",
505 .maxlen = TCP_ULP_BUF_MAX,
506 .mode = 0444,
507 .proc_handler = proc_tcp_available_ulp,
508 },
509 {
510 .procname = "icmp_msgs_per_sec",
511 .data = &sysctl_icmp_msgs_per_sec,
512 .maxlen = sizeof(int),
513 .mode = 0644,
514 .proc_handler = proc_dointvec_minmax,
515 .extra1 = &zero,
516 },
517 {
518 .procname = "icmp_msgs_burst",
519 .data = &sysctl_icmp_msgs_burst,
520 .maxlen = sizeof(int),
521 .mode = 0644,
522 .proc_handler = proc_dointvec_minmax,
523 .extra1 = &zero,
524 },
525 {
526 .procname = "udp_mem",
527 .data = &sysctl_udp_mem,
528 .maxlen = sizeof(sysctl_udp_mem),
529 .mode = 0644,
530 .proc_handler = proc_doulongvec_minmax,
531 },
532 {
533 .procname = "udp_rmem_min",
534 .data = &sysctl_udp_rmem_min,
535 .maxlen = sizeof(sysctl_udp_rmem_min),
536 .mode = 0644,
537 .proc_handler = proc_dointvec_minmax,
538 .extra1 = &one
539 },
540 {
541 .procname = "udp_wmem_min",
542 .data = &sysctl_udp_wmem_min,
543 .maxlen = sizeof(sysctl_udp_wmem_min),
544 .mode = 0644,
545 .proc_handler = proc_dointvec_minmax,
546 .extra1 = &one
547 },
548 { }
549 };
550
551 static struct ctl_table ipv4_net_table[] = {
552 {
553 .procname = "icmp_echo_ignore_all",
554 .data = &init_net.ipv4.sysctl_icmp_echo_ignore_all,
555 .maxlen = sizeof(int),
556 .mode = 0644,
557 .proc_handler = proc_dointvec
558 },
559 {
560 .procname = "icmp_echo_ignore_broadcasts",
561 .data = &init_net.ipv4.sysctl_icmp_echo_ignore_broadcasts,
562 .maxlen = sizeof(int),
563 .mode = 0644,
564 .proc_handler = proc_dointvec
565 },
566 {
567 .procname = "icmp_ignore_bogus_error_responses",
568 .data = &init_net.ipv4.sysctl_icmp_ignore_bogus_error_responses,
569 .maxlen = sizeof(int),
570 .mode = 0644,
571 .proc_handler = proc_dointvec
572 },
573 {
574 .procname = "icmp_errors_use_inbound_ifaddr",
575 .data = &init_net.ipv4.sysctl_icmp_errors_use_inbound_ifaddr,
576 .maxlen = sizeof(int),
577 .mode = 0644,
578 .proc_handler = proc_dointvec
579 },
580 {
581 .procname = "icmp_ratelimit",
582 .data = &init_net.ipv4.sysctl_icmp_ratelimit,
583 .maxlen = sizeof(int),
584 .mode = 0644,
585 .proc_handler = proc_dointvec_ms_jiffies,
586 },
587 {
588 .procname = "icmp_ratemask",
589 .data = &init_net.ipv4.sysctl_icmp_ratemask,
590 .maxlen = sizeof(int),
591 .mode = 0644,
592 .proc_handler = proc_dointvec
593 },
594 {
595 .procname = "ping_group_range",
596 .data = &init_net.ipv4.ping_group_range.range,
597 .maxlen = sizeof(gid_t)*2,
598 .mode = 0644,
599 .proc_handler = ipv4_ping_group_range,
600 },
601 {
602 .procname = "tcp_ecn",
603 .data = &init_net.ipv4.sysctl_tcp_ecn,
604 .maxlen = sizeof(int),
605 .mode = 0644,
606 .proc_handler = proc_dointvec
607 },
608 {
609 .procname = "tcp_ecn_fallback",
610 .data = &init_net.ipv4.sysctl_tcp_ecn_fallback,
611 .maxlen = sizeof(int),
612 .mode = 0644,
613 .proc_handler = proc_dointvec
614 },
615 {
616 .procname = "ip_dynaddr",
617 .data = &init_net.ipv4.sysctl_ip_dynaddr,
618 .maxlen = sizeof(int),
619 .mode = 0644,
620 .proc_handler = proc_dointvec
621 },
622 {
623 .procname = "ip_early_demux",
624 .data = &init_net.ipv4.sysctl_ip_early_demux,
625 .maxlen = sizeof(int),
626 .mode = 0644,
627 .proc_handler = proc_dointvec
628 },
629 {
630 .procname = "udp_early_demux",
631 .data = &init_net.ipv4.sysctl_udp_early_demux,
632 .maxlen = sizeof(int),
633 .mode = 0644,
634 .proc_handler = proc_udp_early_demux
635 },
636 {
637 .procname = "tcp_early_demux",
638 .data = &init_net.ipv4.sysctl_tcp_early_demux,
639 .maxlen = sizeof(int),
640 .mode = 0644,
641 .proc_handler = proc_tcp_early_demux
642 },
643 {
644 .procname = "ip_default_ttl",
645 .data = &init_net.ipv4.sysctl_ip_default_ttl,
646 .maxlen = sizeof(int),
647 .mode = 0644,
648 .proc_handler = proc_dointvec_minmax,
649 .extra1 = &ip_ttl_min,
650 .extra2 = &ip_ttl_max,
651 },
652 {
653 .procname = "ip_local_port_range",
654 .maxlen = sizeof(init_net.ipv4.ip_local_ports.range),
655 .data = &init_net.ipv4.ip_local_ports.range,
656 .mode = 0644,
657 .proc_handler = ipv4_local_port_range,
658 },
659 {
660 .procname = "ip_local_reserved_ports",
661 .data = &init_net.ipv4.sysctl_local_reserved_ports,
662 .maxlen = 65536,
663 .mode = 0644,
664 .proc_handler = proc_do_large_bitmap,
665 },
666 {
667 .procname = "ip_no_pmtu_disc",
668 .data = &init_net.ipv4.sysctl_ip_no_pmtu_disc,
669 .maxlen = sizeof(int),
670 .mode = 0644,
671 .proc_handler = proc_dointvec
672 },
673 {
674 .procname = "ip_forward_use_pmtu",
675 .data = &init_net.ipv4.sysctl_ip_fwd_use_pmtu,
676 .maxlen = sizeof(int),
677 .mode = 0644,
678 .proc_handler = proc_dointvec,
679 },
680 {
681 .procname = "ip_nonlocal_bind",
682 .data = &init_net.ipv4.sysctl_ip_nonlocal_bind,
683 .maxlen = sizeof(int),
684 .mode = 0644,
685 .proc_handler = proc_dointvec
686 },
687 {
688 .procname = "fwmark_reflect",
689 .data = &init_net.ipv4.sysctl_fwmark_reflect,
690 .maxlen = sizeof(int),
691 .mode = 0644,
692 .proc_handler = proc_dointvec,
693 },
694 {
695 .procname = "tcp_fwmark_accept",
696 .data = &init_net.ipv4.sysctl_tcp_fwmark_accept,
697 .maxlen = sizeof(int),
698 .mode = 0644,
699 .proc_handler = proc_dointvec,
700 },
701 #ifdef CONFIG_NET_L3_MASTER_DEV
702 {
703 .procname = "tcp_l3mdev_accept",
704 .data = &init_net.ipv4.sysctl_tcp_l3mdev_accept,
705 .maxlen = sizeof(int),
706 .mode = 0644,
707 .proc_handler = proc_dointvec_minmax,
708 .extra1 = &zero,
709 .extra2 = &one,
710 },
711 #endif
712 {
713 .procname = "tcp_mtu_probing",
714 .data = &init_net.ipv4.sysctl_tcp_mtu_probing,
715 .maxlen = sizeof(int),
716 .mode = 0644,
717 .proc_handler = proc_dointvec,
718 },
719 {
720 .procname = "tcp_base_mss",
721 .data = &init_net.ipv4.sysctl_tcp_base_mss,
722 .maxlen = sizeof(int),
723 .mode = 0644,
724 .proc_handler = proc_dointvec,
725 },
726 {
727 .procname = "tcp_probe_threshold",
728 .data = &init_net.ipv4.sysctl_tcp_probe_threshold,
729 .maxlen = sizeof(int),
730 .mode = 0644,
731 .proc_handler = proc_dointvec,
732 },
733 {
734 .procname = "tcp_probe_interval",
735 .data = &init_net.ipv4.sysctl_tcp_probe_interval,
736 .maxlen = sizeof(int),
737 .mode = 0644,
738 .proc_handler = proc_dointvec,
739 },
740 {
741 .procname = "igmp_link_local_mcast_reports",
742 .data = &init_net.ipv4.sysctl_igmp_llm_reports,
743 .maxlen = sizeof(int),
744 .mode = 0644,
745 .proc_handler = proc_dointvec
746 },
747 {
748 .procname = "igmp_max_memberships",
749 .data = &init_net.ipv4.sysctl_igmp_max_memberships,
750 .maxlen = sizeof(int),
751 .mode = 0644,
752 .proc_handler = proc_dointvec
753 },
754 {
755 .procname = "igmp_max_msf",
756 .data = &init_net.ipv4.sysctl_igmp_max_msf,
757 .maxlen = sizeof(int),
758 .mode = 0644,
759 .proc_handler = proc_dointvec
760 },
761 #ifdef CONFIG_IP_MULTICAST
762 {
763 .procname = "igmp_qrv",
764 .data = &init_net.ipv4.sysctl_igmp_qrv,
765 .maxlen = sizeof(int),
766 .mode = 0644,
767 .proc_handler = proc_dointvec_minmax,
768 .extra1 = &one
769 },
770 #endif
771 {
772 .procname = "tcp_congestion_control",
773 .data = &init_net.ipv4.tcp_congestion_control,
774 .mode = 0644,
775 .maxlen = TCP_CA_NAME_MAX,
776 .proc_handler = proc_tcp_congestion_control,
777 },
778 {
779 .procname = "tcp_keepalive_time",
780 .data = &init_net.ipv4.sysctl_tcp_keepalive_time,
781 .maxlen = sizeof(int),
782 .mode = 0644,
783 .proc_handler = proc_dointvec_jiffies,
784 },
785 {
786 .procname = "tcp_keepalive_probes",
787 .data = &init_net.ipv4.sysctl_tcp_keepalive_probes,
788 .maxlen = sizeof(int),
789 .mode = 0644,
790 .proc_handler = proc_dointvec
791 },
792 {
793 .procname = "tcp_keepalive_intvl",
794 .data = &init_net.ipv4.sysctl_tcp_keepalive_intvl,
795 .maxlen = sizeof(int),
796 .mode = 0644,
797 .proc_handler = proc_dointvec_jiffies,
798 },
799 {
800 .procname = "tcp_syn_retries",
801 .data = &init_net.ipv4.sysctl_tcp_syn_retries,
802 .maxlen = sizeof(int),
803 .mode = 0644,
804 .proc_handler = proc_dointvec_minmax,
805 .extra1 = &tcp_syn_retries_min,
806 .extra2 = &tcp_syn_retries_max
807 },
808 {
809 .procname = "tcp_synack_retries",
810 .data = &init_net.ipv4.sysctl_tcp_synack_retries,
811 .maxlen = sizeof(int),
812 .mode = 0644,
813 .proc_handler = proc_dointvec
814 },
815 #ifdef CONFIG_SYN_COOKIES
816 {
817 .procname = "tcp_syncookies",
818 .data = &init_net.ipv4.sysctl_tcp_syncookies,
819 .maxlen = sizeof(int),
820 .mode = 0644,
821 .proc_handler = proc_dointvec
822 },
823 #endif
824 {
825 .procname = "tcp_reordering",
826 .data = &init_net.ipv4.sysctl_tcp_reordering,
827 .maxlen = sizeof(int),
828 .mode = 0644,
829 .proc_handler = proc_dointvec
830 },
831 {
832 .procname = "tcp_retries1",
833 .data = &init_net.ipv4.sysctl_tcp_retries1,
834 .maxlen = sizeof(int),
835 .mode = 0644,
836 .proc_handler = proc_dointvec_minmax,
837 .extra2 = &tcp_retr1_max
838 },
839 {
840 .procname = "tcp_retries2",
841 .data = &init_net.ipv4.sysctl_tcp_retries2,
842 .maxlen = sizeof(int),
843 .mode = 0644,
844 .proc_handler = proc_dointvec
845 },
846 {
847 .procname = "tcp_orphan_retries",
848 .data = &init_net.ipv4.sysctl_tcp_orphan_retries,
849 .maxlen = sizeof(int),
850 .mode = 0644,
851 .proc_handler = proc_dointvec
852 },
853 {
854 .procname = "tcp_fin_timeout",
855 .data = &init_net.ipv4.sysctl_tcp_fin_timeout,
856 .maxlen = sizeof(int),
857 .mode = 0644,
858 .proc_handler = proc_dointvec_jiffies,
859 },
860 {
861 .procname = "tcp_notsent_lowat",
862 .data = &init_net.ipv4.sysctl_tcp_notsent_lowat,
863 .maxlen = sizeof(unsigned int),
864 .mode = 0644,
865 .proc_handler = proc_douintvec,
866 },
867 {
868 .procname = "tcp_tw_reuse",
869 .data = &init_net.ipv4.sysctl_tcp_tw_reuse,
870 .maxlen = sizeof(int),
871 .mode = 0644,
872 .proc_handler = proc_dointvec
873 },
874 {
875 .procname = "tcp_max_tw_buckets",
876 .data = &init_net.ipv4.tcp_death_row.sysctl_max_tw_buckets,
877 .maxlen = sizeof(int),
878 .mode = 0644,
879 .proc_handler = proc_dointvec
880 },
881 {
882 .procname = "tcp_max_syn_backlog",
883 .data = &init_net.ipv4.sysctl_max_syn_backlog,
884 .maxlen = sizeof(int),
885 .mode = 0644,
886 .proc_handler = proc_dointvec
887 },
888 {
889 .procname = "tcp_fastopen",
890 .data = &init_net.ipv4.sysctl_tcp_fastopen,
891 .maxlen = sizeof(int),
892 .mode = 0644,
893 .proc_handler = proc_dointvec,
894 },
895 {
896 .procname = "tcp_fastopen_key",
897 .mode = 0600,
898 .data = &init_net.ipv4.sysctl_tcp_fastopen,
899 .maxlen = ((TCP_FASTOPEN_KEY_LENGTH * 2) + 10),
900 .proc_handler = proc_tcp_fastopen_key,
901 },
902 {
903 .procname = "tcp_fastopen_blackhole_timeout_sec",
904 .data = &init_net.ipv4.sysctl_tcp_fastopen_blackhole_timeout,
905 .maxlen = sizeof(int),
906 .mode = 0644,
907 .proc_handler = proc_tfo_blackhole_detect_timeout,
908 .extra1 = &zero,
909 },
910 #ifdef CONFIG_IP_ROUTE_MULTIPATH
911 {
912 .procname = "fib_multipath_use_neigh",
913 .data = &init_net.ipv4.sysctl_fib_multipath_use_neigh,
914 .maxlen = sizeof(int),
915 .mode = 0644,
916 .proc_handler = proc_dointvec_minmax,
917 .extra1 = &zero,
918 .extra2 = &one,
919 },
920 {
921 .procname = "fib_multipath_hash_policy",
922 .data = &init_net.ipv4.sysctl_fib_multipath_hash_policy,
923 .maxlen = sizeof(int),
924 .mode = 0644,
925 .proc_handler = proc_fib_multipath_hash_policy,
926 .extra1 = &zero,
927 .extra2 = &one,
928 },
929 #endif
930 {
931 .procname = "ip_unprivileged_port_start",
932 .maxlen = sizeof(int),
933 .data = &init_net.ipv4.sysctl_ip_prot_sock,
934 .mode = 0644,
935 .proc_handler = ipv4_privileged_ports,
936 },
937 #ifdef CONFIG_NET_L3_MASTER_DEV
938 {
939 .procname = "udp_l3mdev_accept",
940 .data = &init_net.ipv4.sysctl_udp_l3mdev_accept,
941 .maxlen = sizeof(int),
942 .mode = 0644,
943 .proc_handler = proc_dointvec_minmax,
944 .extra1 = &zero,
945 .extra2 = &one,
946 },
947 #endif
948 {
949 .procname = "tcp_sack",
950 .data = &init_net.ipv4.sysctl_tcp_sack,
951 .maxlen = sizeof(int),
952 .mode = 0644,
953 .proc_handler = proc_dointvec
954 },
955 {
956 .procname = "tcp_window_scaling",
957 .data = &init_net.ipv4.sysctl_tcp_window_scaling,
958 .maxlen = sizeof(int),
959 .mode = 0644,
960 .proc_handler = proc_dointvec
961 },
962 {
963 .procname = "tcp_timestamps",
964 .data = &init_net.ipv4.sysctl_tcp_timestamps,
965 .maxlen = sizeof(int),
966 .mode = 0644,
967 .proc_handler = proc_dointvec
968 },
969 {
970 .procname = "tcp_early_retrans",
971 .data = &init_net.ipv4.sysctl_tcp_early_retrans,
972 .maxlen = sizeof(int),
973 .mode = 0644,
974 .proc_handler = proc_dointvec_minmax,
975 .extra1 = &zero,
976 .extra2 = &four,
977 },
978 {
979 .procname = "tcp_recovery",
980 .data = &init_net.ipv4.sysctl_tcp_recovery,
981 .maxlen = sizeof(int),
982 .mode = 0644,
983 .proc_handler = proc_dointvec,
984 },
985 {
986 .procname = "tcp_thin_linear_timeouts",
987 .data = &init_net.ipv4.sysctl_tcp_thin_linear_timeouts,
988 .maxlen = sizeof(int),
989 .mode = 0644,
990 .proc_handler = proc_dointvec
991 },
992 {
993 .procname = "tcp_slow_start_after_idle",
994 .data = &init_net.ipv4.sysctl_tcp_slow_start_after_idle,
995 .maxlen = sizeof(int),
996 .mode = 0644,
997 .proc_handler = proc_dointvec
998 },
999 {
1000 .procname = "tcp_retrans_collapse",
1001 .data = &init_net.ipv4.sysctl_tcp_retrans_collapse,
1002 .maxlen = sizeof(int),
1003 .mode = 0644,
1004 .proc_handler = proc_dointvec
1005 },
1006 {
1007 .procname = "tcp_stdurg",
1008 .data = &init_net.ipv4.sysctl_tcp_stdurg,
1009 .maxlen = sizeof(int),
1010 .mode = 0644,
1011 .proc_handler = proc_dointvec
1012 },
1013 {
1014 .procname = "tcp_rfc1337",
1015 .data = &init_net.ipv4.sysctl_tcp_rfc1337,
1016 .maxlen = sizeof(int),
1017 .mode = 0644,
1018 .proc_handler = proc_dointvec
1019 },
1020 {
1021 .procname = "tcp_abort_on_overflow",
1022 .data = &init_net.ipv4.sysctl_tcp_abort_on_overflow,
1023 .maxlen = sizeof(int),
1024 .mode = 0644,
1025 .proc_handler = proc_dointvec
1026 },
1027 {
1028 .procname = "tcp_fack",
1029 .data = &init_net.ipv4.sysctl_tcp_fack,
1030 .maxlen = sizeof(int),
1031 .mode = 0644,
1032 .proc_handler = proc_dointvec
1033 },
1034 {
1035 .procname = "tcp_max_reordering",
1036 .data = &init_net.ipv4.sysctl_tcp_max_reordering,
1037 .maxlen = sizeof(int),
1038 .mode = 0644,
1039 .proc_handler = proc_dointvec
1040 },
1041 {
1042 .procname = "tcp_dsack",
1043 .data = &init_net.ipv4.sysctl_tcp_dsack,
1044 .maxlen = sizeof(int),
1045 .mode = 0644,
1046 .proc_handler = proc_dointvec
1047 },
1048 {
1049 .procname = "tcp_app_win",
1050 .data = &init_net.ipv4.sysctl_tcp_app_win,
1051 .maxlen = sizeof(int),
1052 .mode = 0644,
1053 .proc_handler = proc_dointvec
1054 },
1055 {
1056 .procname = "tcp_adv_win_scale",
1057 .data = &init_net.ipv4.sysctl_tcp_adv_win_scale,
1058 .maxlen = sizeof(int),
1059 .mode = 0644,
1060 .proc_handler = proc_dointvec_minmax,
1061 .extra1 = &tcp_adv_win_scale_min,
1062 .extra2 = &tcp_adv_win_scale_max,
1063 },
1064 {
1065 .procname = "tcp_frto",
1066 .data = &init_net.ipv4.sysctl_tcp_frto,
1067 .maxlen = sizeof(int),
1068 .mode = 0644,
1069 .proc_handler = proc_dointvec
1070 },
1071 {
1072 .procname = "tcp_no_metrics_save",
1073 .data = &init_net.ipv4.sysctl_tcp_nometrics_save,
1074 .maxlen = sizeof(int),
1075 .mode = 0644,
1076 .proc_handler = proc_dointvec,
1077 },
1078 {
1079 .procname = "tcp_moderate_rcvbuf",
1080 .data = &init_net.ipv4.sysctl_tcp_moderate_rcvbuf,
1081 .maxlen = sizeof(int),
1082 .mode = 0644,
1083 .proc_handler = proc_dointvec,
1084 },
1085 {
1086 .procname = "tcp_tso_win_divisor",
1087 .data = &init_net.ipv4.sysctl_tcp_tso_win_divisor,
1088 .maxlen = sizeof(int),
1089 .mode = 0644,
1090 .proc_handler = proc_dointvec,
1091 },
1092 {
1093 .procname = "tcp_workaround_signed_windows",
1094 .data = &init_net.ipv4.sysctl_tcp_workaround_signed_windows,
1095 .maxlen = sizeof(int),
1096 .mode = 0644,
1097 .proc_handler = proc_dointvec
1098 },
1099 {
1100 .procname = "tcp_limit_output_bytes",
1101 .data = &init_net.ipv4.sysctl_tcp_limit_output_bytes,
1102 .maxlen = sizeof(int),
1103 .mode = 0644,
1104 .proc_handler = proc_dointvec
1105 },
1106 {
1107 .procname = "tcp_challenge_ack_limit",
1108 .data = &init_net.ipv4.sysctl_tcp_challenge_ack_limit,
1109 .maxlen = sizeof(int),
1110 .mode = 0644,
1111 .proc_handler = proc_dointvec
1112 },
1113 {
1114 .procname = "tcp_min_tso_segs",
1115 .data = &init_net.ipv4.sysctl_tcp_min_tso_segs,
1116 .maxlen = sizeof(int),
1117 .mode = 0644,
1118 .proc_handler = proc_dointvec_minmax,
1119 .extra1 = &one,
1120 .extra2 = &gso_max_segs,
1121 },
1122 {
1123 .procname = "tcp_min_rtt_wlen",
1124 .data = &init_net.ipv4.sysctl_tcp_min_rtt_wlen,
1125 .maxlen = sizeof(int),
1126 .mode = 0644,
1127 .proc_handler = proc_dointvec
1128 },
1129 {
1130 .procname = "tcp_autocorking",
1131 .data = &init_net.ipv4.sysctl_tcp_autocorking,
1132 .maxlen = sizeof(int),
1133 .mode = 0644,
1134 .proc_handler = proc_dointvec_minmax,
1135 .extra1 = &zero,
1136 .extra2 = &one,
1137 },
1138 {
1139 .procname = "tcp_invalid_ratelimit",
1140 .data = &init_net.ipv4.sysctl_tcp_invalid_ratelimit,
1141 .maxlen = sizeof(int),
1142 .mode = 0644,
1143 .proc_handler = proc_dointvec_ms_jiffies,
1144 },
1145 {
1146 .procname = "tcp_pacing_ss_ratio",
1147 .data = &init_net.ipv4.sysctl_tcp_pacing_ss_ratio,
1148 .maxlen = sizeof(int),
1149 .mode = 0644,
1150 .proc_handler = proc_dointvec_minmax,
1151 .extra1 = &zero,
1152 .extra2 = &thousand,
1153 },
1154 {
1155 .procname = "tcp_pacing_ca_ratio",
1156 .data = &init_net.ipv4.sysctl_tcp_pacing_ca_ratio,
1157 .maxlen = sizeof(int),
1158 .mode = 0644,
1159 .proc_handler = proc_dointvec_minmax,
1160 .extra1 = &zero,
1161 .extra2 = &thousand,
1162 },
1163 {
1164 .procname = "tcp_wmem",
1165 .data = &init_net.ipv4.sysctl_tcp_wmem,
1166 .maxlen = sizeof(init_net.ipv4.sysctl_tcp_wmem),
1167 .mode = 0644,
1168 .proc_handler = proc_dointvec_minmax,
1169 .extra1 = &one,
1170 },
1171 {
1172 .procname = "tcp_rmem",
1173 .data = &init_net.ipv4.sysctl_tcp_rmem,
1174 .maxlen = sizeof(init_net.ipv4.sysctl_tcp_rmem),
1175 .mode = 0644,
1176 .proc_handler = proc_dointvec_minmax,
1177 .extra1 = &one,
1178 },
1179 { }
1180 };
1181
1182 static __net_init int ipv4_sysctl_init_net(struct net *net)
1183 {
1184 struct ctl_table *table;
1185
1186 table = ipv4_net_table;
1187 if (!net_eq(net, &init_net)) {
1188 int i;
1189
1190 table = kmemdup(table, sizeof(ipv4_net_table), GFP_KERNEL);
1191 if (!table)
1192 goto err_alloc;
1193
1194 /* Update the variables to point into the current struct net */
1195 for (i = 0; i < ARRAY_SIZE(ipv4_net_table) - 1; i++)
1196 table[i].data += (void *)net - (void *)&init_net;
1197 }
1198
1199 net->ipv4.ipv4_hdr = register_net_sysctl(net, "net/ipv4", table);
1200 if (!net->ipv4.ipv4_hdr)
1201 goto err_reg;
1202
1203 net->ipv4.sysctl_local_reserved_ports = kzalloc(65536 / 8, GFP_KERNEL);
1204 if (!net->ipv4.sysctl_local_reserved_ports)
1205 goto err_ports;
1206
1207 return 0;
1208
1209 err_ports:
1210 unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
1211 err_reg:
1212 if (!net_eq(net, &init_net))
1213 kfree(table);
1214 err_alloc:
1215 return -ENOMEM;
1216 }
1217
1218 static __net_exit void ipv4_sysctl_exit_net(struct net *net)
1219 {
1220 struct ctl_table *table;
1221
1222 kfree(net->ipv4.sysctl_local_reserved_ports);
1223 table = net->ipv4.ipv4_hdr->ctl_table_arg;
1224 unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
1225 kfree(table);
1226 }
1227
1228 static __net_initdata struct pernet_operations ipv4_sysctl_ops = {
1229 .init = ipv4_sysctl_init_net,
1230 .exit = ipv4_sysctl_exit_net,
1231 };
1232
1233 static __init int sysctl_ipv4_init(void)
1234 {
1235 struct ctl_table_header *hdr;
1236
1237 hdr = register_net_sysctl(&init_net, "net/ipv4", ipv4_table);
1238 if (!hdr)
1239 return -ENOMEM;
1240
1241 if (register_pernet_subsys(&ipv4_sysctl_ops)) {
1242 unregister_net_sysctl_table(hdr);
1243 return -ENOMEM;
1244 }
1245
1246 return 0;
1247 }
1248
1249 __initcall(sysctl_ipv4_init);
ubuntu-bionic-kernel mirror