?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
x86/speculation/mds: Fix documentation typo
[linux/fpc-iii.git] / net / ipv4 / sysctl_net_ipv4.c
blobe8caab8e2f5cc88f37d2c48b77f08cf7089b5b6f
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
30 static int zero;
31 static int one = 1;
32 static int four = 4;
33 static int thousand = 1000;
34 static int gso_max_segs = GSO_MAX_SEGS;
35 static int tcp_retr1_max = 255;
36 static int ip_local_port_range_min[] = { 1, 1 };
37 static int ip_local_port_range_max[] = { 65535, 65535 };
38 static int tcp_adv_win_scale_min = -31;
39 static int tcp_adv_win_scale_max = 31;
40 static int ip_privileged_port_min;
41 static int ip_privileged_port_max = 65535;
42 static int ip_ttl_min = 1;
43 static int ip_ttl_max = 255;
44 static int tcp_syn_retries_min = 1;
45 static int tcp_syn_retries_max = MAX_TCP_SYNCNT;
46 static int ip_ping_group_range_min[] = { 0, 0 };
47 static int ip_ping_group_range_max[] = { GID_T_MAX, GID_T_MAX };
48 static int one_day_secs = 24 * 3600;
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 char val[TCP_CA_NAME_MAX];
206 struct ctl_table tbl = {
207 .data = val,
208 .maxlen = TCP_CA_NAME_MAX,
209 };
210 int ret;
211
212 tcp_get_default_congestion_control(val);
213
214 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
215 if (write && ret == 0)
216 ret = tcp_set_default_congestion_control(val);
217 return ret;
218 }
219
220 static int proc_tcp_available_congestion_control(struct ctl_table *ctl,
221 int write,
222 void __user *buffer, size_t *lenp,
223 loff_t *ppos)
224 {
225 struct ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX, };
226 int ret;
227
228 tbl.data = kmalloc(tbl.maxlen, GFP_USER);
229 if (!tbl.data)
230 return -ENOMEM;
231 tcp_get_available_congestion_control(tbl.data, TCP_CA_BUF_MAX);
232 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
233 kfree(tbl.data);
234 return ret;
235 }
236
237 static int proc_allowed_congestion_control(struct ctl_table *ctl,
238 int write,
239 void __user *buffer, size_t *lenp,
240 loff_t *ppos)
241 {
242 struct ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX };
243 int ret;
244
245 tbl.data = kmalloc(tbl.maxlen, GFP_USER);
246 if (!tbl.data)
247 return -ENOMEM;
248
249 tcp_get_allowed_congestion_control(tbl.data, tbl.maxlen);
250 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
251 if (write && ret == 0)
252 ret = tcp_set_allowed_congestion_control(tbl.data);
253 kfree(tbl.data);
254 return ret;
255 }
256
257 static int proc_tcp_fastopen_key(struct ctl_table *ctl, int write,
258 void __user *buffer, size_t *lenp,
259 loff_t *ppos)
260 {
261 struct ctl_table tbl = { .maxlen = (TCP_FASTOPEN_KEY_LENGTH * 2 + 10) };
262 struct tcp_fastopen_context *ctxt;
263 u32 user_key[4]; /* 16 bytes, matching TCP_FASTOPEN_KEY_LENGTH */
264 __le32 key[4];
265 int ret, i;
266
267 tbl.data = kmalloc(tbl.maxlen, GFP_KERNEL);
268 if (!tbl.data)
269 return -ENOMEM;
270
271 rcu_read_lock();
272 ctxt = rcu_dereference(tcp_fastopen_ctx);
273 if (ctxt)
274 memcpy(key, ctxt->key, TCP_FASTOPEN_KEY_LENGTH);
275 else
276 memset(key, 0, sizeof(key));
277 rcu_read_unlock();
278
279 for (i = 0; i < ARRAY_SIZE(key); i++)
280 user_key[i] = le32_to_cpu(key[i]);
281
282 snprintf(tbl.data, tbl.maxlen, "%08x-%08x-%08x-%08x",
283 user_key[0], user_key[1], user_key[2], user_key[3]);
284 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
285
286 if (write && ret == 0) {
287 if (sscanf(tbl.data, "%x-%x-%x-%x", user_key, user_key + 1,
288 user_key + 2, user_key + 3) != 4) {
289 ret = -EINVAL;
290 goto bad_key;
291 }
292 /* Generate a dummy secret but don't publish it. This
293 * is needed so we don't regenerate a new key on the
294 * first invocation of tcp_fastopen_cookie_gen
295 */
296 tcp_fastopen_init_key_once(false);
297
298 for (i = 0; i < ARRAY_SIZE(user_key); i++)
299 key[i] = cpu_to_le32(user_key[i]);
300
301 tcp_fastopen_reset_cipher(key, 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 int ret;
373
374 ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
375 if (write && ret == 0)
376 tcp_fastopen_active_timeout_reset();
377
378 return ret;
379 }
380
381 static int proc_tcp_available_ulp(struct ctl_table *ctl,
382 int write,
383 void __user *buffer, size_t *lenp,
384 loff_t *ppos)
385 {
386 struct ctl_table tbl = { .maxlen = TCP_ULP_BUF_MAX, };
387 int ret;
388
389 tbl.data = kmalloc(tbl.maxlen, GFP_USER);
390 if (!tbl.data)
391 return -ENOMEM;
392 tcp_get_available_ulp(tbl.data, TCP_ULP_BUF_MAX);
393 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
394 kfree(tbl.data);
395
396 return ret;
397 }
398
399 static struct ctl_table ipv4_table[] = {
400 {
401 .procname = "tcp_retrans_collapse",
402 .data = &sysctl_tcp_retrans_collapse,
403 .maxlen = sizeof(int),
404 .mode = 0644,
405 .proc_handler = proc_dointvec
406 },
407 {
408 .procname = "tcp_max_orphans",
409 .data = &sysctl_tcp_max_orphans,
410 .maxlen = sizeof(int),
411 .mode = 0644,
412 .proc_handler = proc_dointvec
413 },
414 {
415 .procname = "tcp_fastopen",
416 .data = &sysctl_tcp_fastopen,
417 .maxlen = sizeof(int),
418 .mode = 0644,
419 .proc_handler = proc_dointvec,
420 },
421 {
422 .procname = "tcp_fastopen_key",
423 .mode = 0600,
424 .maxlen = ((TCP_FASTOPEN_KEY_LENGTH * 2) + 10),
425 .proc_handler = proc_tcp_fastopen_key,
426 },
427 {
428 .procname = "tcp_fastopen_blackhole_timeout_sec",
429 .data = &sysctl_tcp_fastopen_blackhole_timeout,
430 .maxlen = sizeof(int),
431 .mode = 0644,
432 .proc_handler = proc_tfo_blackhole_detect_timeout,
433 .extra1 = &zero,
434 },
435 {
436 .procname = "tcp_abort_on_overflow",
437 .data = &sysctl_tcp_abort_on_overflow,
438 .maxlen = sizeof(int),
439 .mode = 0644,
440 .proc_handler = proc_dointvec
441 },
442 {
443 .procname = "tcp_stdurg",
444 .data = &sysctl_tcp_stdurg,
445 .maxlen = sizeof(int),
446 .mode = 0644,
447 .proc_handler = proc_dointvec
448 },
449 {
450 .procname = "tcp_rfc1337",
451 .data = &sysctl_tcp_rfc1337,
452 .maxlen = sizeof(int),
453 .mode = 0644,
454 .proc_handler = proc_dointvec
455 },
456 {
457 .procname = "inet_peer_threshold",
458 .data = &inet_peer_threshold,
459 .maxlen = sizeof(int),
460 .mode = 0644,
461 .proc_handler = proc_dointvec
462 },
463 {
464 .procname = "inet_peer_minttl",
465 .data = &inet_peer_minttl,
466 .maxlen = sizeof(int),
467 .mode = 0644,
468 .proc_handler = proc_dointvec_jiffies,
469 },
470 {
471 .procname = "inet_peer_maxttl",
472 .data = &inet_peer_maxttl,
473 .maxlen = sizeof(int),
474 .mode = 0644,
475 .proc_handler = proc_dointvec_jiffies,
476 },
477 {
478 .procname = "tcp_fack",
479 .data = &sysctl_tcp_fack,
480 .maxlen = sizeof(int),
481 .mode = 0644,
482 .proc_handler = proc_dointvec
483 },
484 {
485 .procname = "tcp_recovery",
486 .data = &sysctl_tcp_recovery,
487 .maxlen = sizeof(int),
488 .mode = 0644,
489 .proc_handler = proc_dointvec,
490 },
491 {
492 .procname = "tcp_max_reordering",
493 .data = &sysctl_tcp_max_reordering,
494 .maxlen = sizeof(int),
495 .mode = 0644,
496 .proc_handler = proc_dointvec
497 },
498 {
499 .procname = "tcp_dsack",
500 .data = &sysctl_tcp_dsack,
501 .maxlen = sizeof(int),
502 .mode = 0644,
503 .proc_handler = proc_dointvec
504 },
505 {
506 .procname = "tcp_mem",
507 .maxlen = sizeof(sysctl_tcp_mem),
508 .data = &sysctl_tcp_mem,
509 .mode = 0644,
510 .proc_handler = proc_doulongvec_minmax,
511 },
512 {
513 .procname = "tcp_wmem",
514 .data = &sysctl_tcp_wmem,
515 .maxlen = sizeof(sysctl_tcp_wmem),
516 .mode = 0644,
517 .proc_handler = proc_dointvec_minmax,
518 .extra1 = &one,
519 },
520 {
521 .procname = "tcp_rmem",
522 .data = &sysctl_tcp_rmem,
523 .maxlen = sizeof(sysctl_tcp_rmem),
524 .mode = 0644,
525 .proc_handler = proc_dointvec_minmax,
526 .extra1 = &one,
527 },
528 {
529 .procname = "tcp_app_win",
530 .data = &sysctl_tcp_app_win,
531 .maxlen = sizeof(int),
532 .mode = 0644,
533 .proc_handler = proc_dointvec
534 },
535 {
536 .procname = "tcp_adv_win_scale",
537 .data = &sysctl_tcp_adv_win_scale,
538 .maxlen = sizeof(int),
539 .mode = 0644,
540 .proc_handler = proc_dointvec_minmax,
541 .extra1 = &tcp_adv_win_scale_min,
542 .extra2 = &tcp_adv_win_scale_max,
543 },
544 {
545 .procname = "tcp_frto",
546 .data = &sysctl_tcp_frto,
547 .maxlen = sizeof(int),
548 .mode = 0644,
549 .proc_handler = proc_dointvec
550 },
551 {
552 .procname = "tcp_min_rtt_wlen",
553 .data = &sysctl_tcp_min_rtt_wlen,
554 .maxlen = sizeof(int),
555 .mode = 0644,
556 .proc_handler = proc_dointvec_minmax,
557 .extra1 = &zero,
558 .extra2 = &one_day_secs
559 },
560 {
561 .procname = "tcp_low_latency",
562 .data = &sysctl_tcp_low_latency,
563 .maxlen = sizeof(int),
564 .mode = 0644,
565 .proc_handler = proc_dointvec
566 },
567 {
568 .procname = "tcp_no_metrics_save",
569 .data = &sysctl_tcp_nometrics_save,
570 .maxlen = sizeof(int),
571 .mode = 0644,
572 .proc_handler = proc_dointvec,
573 },
574 {
575 .procname = "tcp_moderate_rcvbuf",
576 .data = &sysctl_tcp_moderate_rcvbuf,
577 .maxlen = sizeof(int),
578 .mode = 0644,
579 .proc_handler = proc_dointvec,
580 },
581 {
582 .procname = "tcp_tso_win_divisor",
583 .data = &sysctl_tcp_tso_win_divisor,
584 .maxlen = sizeof(int),
585 .mode = 0644,
586 .proc_handler = proc_dointvec,
587 },
588 {
589 .procname = "tcp_congestion_control",
590 .mode = 0644,
591 .maxlen = TCP_CA_NAME_MAX,
592 .proc_handler = proc_tcp_congestion_control,
593 },
594 {
595 .procname = "tcp_workaround_signed_windows",
596 .data = &sysctl_tcp_workaround_signed_windows,
597 .maxlen = sizeof(int),
598 .mode = 0644,
599 .proc_handler = proc_dointvec
600 },
601 {
602 .procname = "tcp_limit_output_bytes",
603 .data = &sysctl_tcp_limit_output_bytes,
604 .maxlen = sizeof(int),
605 .mode = 0644,
606 .proc_handler = proc_dointvec
607 },
608 {
609 .procname = "tcp_challenge_ack_limit",
610 .data = &sysctl_tcp_challenge_ack_limit,
611 .maxlen = sizeof(int),
612 .mode = 0644,
613 .proc_handler = proc_dointvec
614 },
615 {
616 .procname = "tcp_slow_start_after_idle",
617 .data = &sysctl_tcp_slow_start_after_idle,
618 .maxlen = sizeof(int),
619 .mode = 0644,
620 .proc_handler = proc_dointvec
621 },
622 #ifdef CONFIG_NETLABEL
623 {
624 .procname = "cipso_cache_enable",
625 .data = &cipso_v4_cache_enabled,
626 .maxlen = sizeof(int),
627 .mode = 0644,
628 .proc_handler = proc_dointvec,
629 },
630 {
631 .procname = "cipso_cache_bucket_size",
632 .data = &cipso_v4_cache_bucketsize,
633 .maxlen = sizeof(int),
634 .mode = 0644,
635 .proc_handler = proc_dointvec,
636 },
637 {
638 .procname = "cipso_rbm_optfmt",
639 .data = &cipso_v4_rbm_optfmt,
640 .maxlen = sizeof(int),
641 .mode = 0644,
642 .proc_handler = proc_dointvec,
643 },
644 {
645 .procname = "cipso_rbm_strictvalid",
646 .data = &cipso_v4_rbm_strictvalid,
647 .maxlen = sizeof(int),
648 .mode = 0644,
649 .proc_handler = proc_dointvec,
650 },
651 #endif /* CONFIG_NETLABEL */
652 {
653 .procname = "tcp_available_congestion_control",
654 .maxlen = TCP_CA_BUF_MAX,
655 .mode = 0444,
656 .proc_handler = proc_tcp_available_congestion_control,
657 },
658 {
659 .procname = "tcp_allowed_congestion_control",
660 .maxlen = TCP_CA_BUF_MAX,
661 .mode = 0644,
662 .proc_handler = proc_allowed_congestion_control,
663 },
664 {
665 .procname = "tcp_thin_linear_timeouts",
666 .data = &sysctl_tcp_thin_linear_timeouts,
667 .maxlen = sizeof(int),
668 .mode = 0644,
669 .proc_handler = proc_dointvec
670 },
671 {
672 .procname = "tcp_early_retrans",
673 .data = &sysctl_tcp_early_retrans,
674 .maxlen = sizeof(int),
675 .mode = 0644,
676 .proc_handler = proc_dointvec_minmax,
677 .extra1 = &zero,
678 .extra2 = &four,
679 },
680 {
681 .procname = "tcp_min_tso_segs",
682 .data = &sysctl_tcp_min_tso_segs,
683 .maxlen = sizeof(int),
684 .mode = 0644,
685 .proc_handler = proc_dointvec_minmax,
686 .extra1 = &one,
687 .extra2 = &gso_max_segs,
688 },
689 {
690 .procname = "tcp_pacing_ss_ratio",
691 .data = &sysctl_tcp_pacing_ss_ratio,
692 .maxlen = sizeof(int),
693 .mode = 0644,
694 .proc_handler = proc_dointvec_minmax,
695 .extra1 = &zero,
696 .extra2 = &thousand,
697 },
698 {
699 .procname = "tcp_pacing_ca_ratio",
700 .data = &sysctl_tcp_pacing_ca_ratio,
701 .maxlen = sizeof(int),
702 .mode = 0644,
703 .proc_handler = proc_dointvec_minmax,
704 .extra1 = &zero,
705 .extra2 = &thousand,
706 },
707 {
708 .procname = "tcp_autocorking",
709 .data = &sysctl_tcp_autocorking,
710 .maxlen = sizeof(int),
711 .mode = 0644,
712 .proc_handler = proc_dointvec_minmax,
713 .extra1 = &zero,
714 .extra2 = &one,
715 },
716 {
717 .procname = "tcp_invalid_ratelimit",
718 .data = &sysctl_tcp_invalid_ratelimit,
719 .maxlen = sizeof(int),
720 .mode = 0644,
721 .proc_handler = proc_dointvec_ms_jiffies,
722 },
723 {
724 .procname = "tcp_available_ulp",
725 .maxlen = TCP_ULP_BUF_MAX,
726 .mode = 0444,
727 .proc_handler = proc_tcp_available_ulp,
728 },
729 {
730 .procname = "icmp_msgs_per_sec",
731 .data = &sysctl_icmp_msgs_per_sec,
732 .maxlen = sizeof(int),
733 .mode = 0644,
734 .proc_handler = proc_dointvec_minmax,
735 .extra1 = &zero,
736 },
737 {
738 .procname = "icmp_msgs_burst",
739 .data = &sysctl_icmp_msgs_burst,
740 .maxlen = sizeof(int),
741 .mode = 0644,
742 .proc_handler = proc_dointvec_minmax,
743 .extra1 = &zero,
744 },
745 {
746 .procname = "udp_mem",
747 .data = &sysctl_udp_mem,
748 .maxlen = sizeof(sysctl_udp_mem),
749 .mode = 0644,
750 .proc_handler = proc_doulongvec_minmax,
751 },
752 {
753 .procname = "udp_rmem_min",
754 .data = &sysctl_udp_rmem_min,
755 .maxlen = sizeof(sysctl_udp_rmem_min),
756 .mode = 0644,
757 .proc_handler = proc_dointvec_minmax,
758 .extra1 = &one
759 },
760 {
761 .procname = "udp_wmem_min",
762 .data = &sysctl_udp_wmem_min,
763 .maxlen = sizeof(sysctl_udp_wmem_min),
764 .mode = 0644,
765 .proc_handler = proc_dointvec_minmax,
766 .extra1 = &one
767 },
768 { }
769 };
770
771 static struct ctl_table ipv4_net_table[] = {
772 {
773 .procname = "icmp_echo_ignore_all",
774 .data = &init_net.ipv4.sysctl_icmp_echo_ignore_all,
775 .maxlen = sizeof(int),
776 .mode = 0644,
777 .proc_handler = proc_dointvec
778 },
779 {
780 .procname = "icmp_echo_ignore_broadcasts",
781 .data = &init_net.ipv4.sysctl_icmp_echo_ignore_broadcasts,
782 .maxlen = sizeof(int),
783 .mode = 0644,
784 .proc_handler = proc_dointvec
785 },
786 {
787 .procname = "icmp_ignore_bogus_error_responses",
788 .data = &init_net.ipv4.sysctl_icmp_ignore_bogus_error_responses,
789 .maxlen = sizeof(int),
790 .mode = 0644,
791 .proc_handler = proc_dointvec
792 },
793 {
794 .procname = "icmp_errors_use_inbound_ifaddr",
795 .data = &init_net.ipv4.sysctl_icmp_errors_use_inbound_ifaddr,
796 .maxlen = sizeof(int),
797 .mode = 0644,
798 .proc_handler = proc_dointvec
799 },
800 {
801 .procname = "icmp_ratelimit",
802 .data = &init_net.ipv4.sysctl_icmp_ratelimit,
803 .maxlen = sizeof(int),
804 .mode = 0644,
805 .proc_handler = proc_dointvec_ms_jiffies,
806 },
807 {
808 .procname = "icmp_ratemask",
809 .data = &init_net.ipv4.sysctl_icmp_ratemask,
810 .maxlen = sizeof(int),
811 .mode = 0644,
812 .proc_handler = proc_dointvec
813 },
814 {
815 .procname = "ping_group_range",
816 .data = &init_net.ipv4.ping_group_range.range,
817 .maxlen = sizeof(gid_t)*2,
818 .mode = 0644,
819 .proc_handler = ipv4_ping_group_range,
820 },
821 {
822 .procname = "tcp_ecn",
823 .data = &init_net.ipv4.sysctl_tcp_ecn,
824 .maxlen = sizeof(int),
825 .mode = 0644,
826 .proc_handler = proc_dointvec
827 },
828 {
829 .procname = "tcp_ecn_fallback",
830 .data = &init_net.ipv4.sysctl_tcp_ecn_fallback,
831 .maxlen = sizeof(int),
832 .mode = 0644,
833 .proc_handler = proc_dointvec
834 },
835 {
836 .procname = "ip_dynaddr",
837 .data = &init_net.ipv4.sysctl_ip_dynaddr,
838 .maxlen = sizeof(int),
839 .mode = 0644,
840 .proc_handler = proc_dointvec
841 },
842 {
843 .procname = "ip_early_demux",
844 .data = &init_net.ipv4.sysctl_ip_early_demux,
845 .maxlen = sizeof(int),
846 .mode = 0644,
847 .proc_handler = proc_dointvec
848 },
849 {
850 .procname = "udp_early_demux",
851 .data = &init_net.ipv4.sysctl_udp_early_demux,
852 .maxlen = sizeof(int),
853 .mode = 0644,
854 .proc_handler = proc_udp_early_demux
855 },
856 {
857 .procname = "tcp_early_demux",
858 .data = &init_net.ipv4.sysctl_tcp_early_demux,
859 .maxlen = sizeof(int),
860 .mode = 0644,
861 .proc_handler = proc_tcp_early_demux
862 },
863 {
864 .procname = "ip_default_ttl",
865 .data = &init_net.ipv4.sysctl_ip_default_ttl,
866 .maxlen = sizeof(int),
867 .mode = 0644,
868 .proc_handler = proc_dointvec_minmax,
869 .extra1 = &ip_ttl_min,
870 .extra2 = &ip_ttl_max,
871 },
872 {
873 .procname = "ip_local_port_range",
874 .maxlen = sizeof(init_net.ipv4.ip_local_ports.range),
875 .data = &init_net.ipv4.ip_local_ports.range,
876 .mode = 0644,
877 .proc_handler = ipv4_local_port_range,
878 },
879 {
880 .procname = "ip_local_reserved_ports",
881 .data = &init_net.ipv4.sysctl_local_reserved_ports,
882 .maxlen = 65536,
883 .mode = 0644,
884 .proc_handler = proc_do_large_bitmap,
885 },
886 {
887 .procname = "ip_no_pmtu_disc",
888 .data = &init_net.ipv4.sysctl_ip_no_pmtu_disc,
889 .maxlen = sizeof(int),
890 .mode = 0644,
891 .proc_handler = proc_dointvec
892 },
893 {
894 .procname = "ip_forward_use_pmtu",
895 .data = &init_net.ipv4.sysctl_ip_fwd_use_pmtu,
896 .maxlen = sizeof(int),
897 .mode = 0644,
898 .proc_handler = proc_dointvec,
899 },
900 {
901 .procname = "ip_nonlocal_bind",
902 .data = &init_net.ipv4.sysctl_ip_nonlocal_bind,
903 .maxlen = sizeof(int),
904 .mode = 0644,
905 .proc_handler = proc_dointvec
906 },
907 {
908 .procname = "fwmark_reflect",
909 .data = &init_net.ipv4.sysctl_fwmark_reflect,
910 .maxlen = sizeof(int),
911 .mode = 0644,
912 .proc_handler = proc_dointvec,
913 },
914 {
915 .procname = "tcp_fwmark_accept",
916 .data = &init_net.ipv4.sysctl_tcp_fwmark_accept,
917 .maxlen = sizeof(int),
918 .mode = 0644,
919 .proc_handler = proc_dointvec,
920 },
921 #ifdef CONFIG_NET_L3_MASTER_DEV
922 {
923 .procname = "tcp_l3mdev_accept",
924 .data = &init_net.ipv4.sysctl_tcp_l3mdev_accept,
925 .maxlen = sizeof(int),
926 .mode = 0644,
927 .proc_handler = proc_dointvec_minmax,
928 .extra1 = &zero,
929 .extra2 = &one,
930 },
931 #endif
932 {
933 .procname = "tcp_mtu_probing",
934 .data = &init_net.ipv4.sysctl_tcp_mtu_probing,
935 .maxlen = sizeof(int),
936 .mode = 0644,
937 .proc_handler = proc_dointvec,
938 },
939 {
940 .procname = "tcp_base_mss",
941 .data = &init_net.ipv4.sysctl_tcp_base_mss,
942 .maxlen = sizeof(int),
943 .mode = 0644,
944 .proc_handler = proc_dointvec,
945 },
946 {
947 .procname = "tcp_probe_threshold",
948 .data = &init_net.ipv4.sysctl_tcp_probe_threshold,
949 .maxlen = sizeof(int),
950 .mode = 0644,
951 .proc_handler = proc_dointvec,
952 },
953 {
954 .procname = "tcp_probe_interval",
955 .data = &init_net.ipv4.sysctl_tcp_probe_interval,
956 .maxlen = sizeof(int),
957 .mode = 0644,
958 .proc_handler = proc_dointvec,
959 },
960 {
961 .procname = "igmp_link_local_mcast_reports",
962 .data = &init_net.ipv4.sysctl_igmp_llm_reports,
963 .maxlen = sizeof(int),
964 .mode = 0644,
965 .proc_handler = proc_dointvec
966 },
967 {
968 .procname = "igmp_max_memberships",
969 .data = &init_net.ipv4.sysctl_igmp_max_memberships,
970 .maxlen = sizeof(int),
971 .mode = 0644,
972 .proc_handler = proc_dointvec
973 },
974 {
975 .procname = "igmp_max_msf",
976 .data = &init_net.ipv4.sysctl_igmp_max_msf,
977 .maxlen = sizeof(int),
978 .mode = 0644,
979 .proc_handler = proc_dointvec
980 },
981 #ifdef CONFIG_IP_MULTICAST
982 {
983 .procname = "igmp_qrv",
984 .data = &init_net.ipv4.sysctl_igmp_qrv,
985 .maxlen = sizeof(int),
986 .mode = 0644,
987 .proc_handler = proc_dointvec_minmax,
988 .extra1 = &one
989 },
990 #endif
991 {
992 .procname = "tcp_keepalive_time",
993 .data = &init_net.ipv4.sysctl_tcp_keepalive_time,
994 .maxlen = sizeof(int),
995 .mode = 0644,
996 .proc_handler = proc_dointvec_jiffies,
997 },
998 {
999 .procname = "tcp_keepalive_probes",
1000 .data = &init_net.ipv4.sysctl_tcp_keepalive_probes,
1001 .maxlen = sizeof(int),
1002 .mode = 0644,
1003 .proc_handler = proc_dointvec
1004 },
1005 {
1006 .procname = "tcp_keepalive_intvl",
1007 .data = &init_net.ipv4.sysctl_tcp_keepalive_intvl,
1008 .maxlen = sizeof(int),
1009 .mode = 0644,
1010 .proc_handler = proc_dointvec_jiffies,
1011 },
1012 {
1013 .procname = "tcp_syn_retries",
1014 .data = &init_net.ipv4.sysctl_tcp_syn_retries,
1015 .maxlen = sizeof(int),
1016 .mode = 0644,
1017 .proc_handler = proc_dointvec_minmax,
1018 .extra1 = &tcp_syn_retries_min,
1019 .extra2 = &tcp_syn_retries_max
1020 },
1021 {
1022 .procname = "tcp_synack_retries",
1023 .data = &init_net.ipv4.sysctl_tcp_synack_retries,
1024 .maxlen = sizeof(int),
1025 .mode = 0644,
1026 .proc_handler = proc_dointvec
1027 },
1028 #ifdef CONFIG_SYN_COOKIES
1029 {
1030 .procname = "tcp_syncookies",
1031 .data = &init_net.ipv4.sysctl_tcp_syncookies,
1032 .maxlen = sizeof(int),
1033 .mode = 0644,
1034 .proc_handler = proc_dointvec
1035 },
1036 #endif
1037 {
1038 .procname = "tcp_reordering",
1039 .data = &init_net.ipv4.sysctl_tcp_reordering,
1040 .maxlen = sizeof(int),
1041 .mode = 0644,
1042 .proc_handler = proc_dointvec
1043 },
1044 {
1045 .procname = "tcp_retries1",
1046 .data = &init_net.ipv4.sysctl_tcp_retries1,
1047 .maxlen = sizeof(int),
1048 .mode = 0644,
1049 .proc_handler = proc_dointvec_minmax,
1050 .extra2 = &tcp_retr1_max
1051 },
1052 {
1053 .procname = "tcp_retries2",
1054 .data = &init_net.ipv4.sysctl_tcp_retries2,
1055 .maxlen = sizeof(int),
1056 .mode = 0644,
1057 .proc_handler = proc_dointvec
1058 },
1059 {
1060 .procname = "tcp_orphan_retries",
1061 .data = &init_net.ipv4.sysctl_tcp_orphan_retries,
1062 .maxlen = sizeof(int),
1063 .mode = 0644,
1064 .proc_handler = proc_dointvec
1065 },
1066 {
1067 .procname = "tcp_fin_timeout",
1068 .data = &init_net.ipv4.sysctl_tcp_fin_timeout,
1069 .maxlen = sizeof(int),
1070 .mode = 0644,
1071 .proc_handler = proc_dointvec_jiffies,
1072 },
1073 {
1074 .procname = "tcp_notsent_lowat",
1075 .data = &init_net.ipv4.sysctl_tcp_notsent_lowat,
1076 .maxlen = sizeof(unsigned int),
1077 .mode = 0644,
1078 .proc_handler = proc_douintvec,
1079 },
1080 {
1081 .procname = "tcp_tw_reuse",
1082 .data = &init_net.ipv4.sysctl_tcp_tw_reuse,
1083 .maxlen = sizeof(int),
1084 .mode = 0644,
1085 .proc_handler = proc_dointvec
1086 },
1087 {
1088 .procname = "tcp_max_tw_buckets",
1089 .data = &init_net.ipv4.tcp_death_row.sysctl_max_tw_buckets,
1090 .maxlen = sizeof(int),
1091 .mode = 0644,
1092 .proc_handler = proc_dointvec
1093 },
1094 {
1095 .procname = "tcp_max_syn_backlog",
1096 .data = &init_net.ipv4.sysctl_max_syn_backlog,
1097 .maxlen = sizeof(int),
1098 .mode = 0644,
1099 .proc_handler = proc_dointvec
1100 },
1101 #ifdef CONFIG_IP_ROUTE_MULTIPATH
1102 {
1103 .procname = "fib_multipath_use_neigh",
1104 .data = &init_net.ipv4.sysctl_fib_multipath_use_neigh,
1105 .maxlen = sizeof(int),
1106 .mode = 0644,
1107 .proc_handler = proc_dointvec_minmax,
1108 .extra1 = &zero,
1109 .extra2 = &one,
1110 },
1111 {
1112 .procname = "fib_multipath_hash_policy",
1113 .data = &init_net.ipv4.sysctl_fib_multipath_hash_policy,
1114 .maxlen = sizeof(int),
1115 .mode = 0644,
1116 .proc_handler = proc_dointvec_minmax,
1117 .extra1 = &zero,
1118 .extra2 = &one,
1119 },
1120 #endif
1121 {
1122 .procname = "ip_unprivileged_port_start",
1123 .maxlen = sizeof(int),
1124 .data = &init_net.ipv4.sysctl_ip_prot_sock,
1125 .mode = 0644,
1126 .proc_handler = ipv4_privileged_ports,
1127 },
1128 #ifdef CONFIG_NET_L3_MASTER_DEV
1129 {
1130 .procname = "udp_l3mdev_accept",
1131 .data = &init_net.ipv4.sysctl_udp_l3mdev_accept,
1132 .maxlen = sizeof(int),
1133 .mode = 0644,
1134 .proc_handler = proc_dointvec_minmax,
1135 .extra1 = &zero,
1136 .extra2 = &one,
1137 },
1138 #endif
1139 {
1140 .procname = "tcp_sack",
1141 .data = &init_net.ipv4.sysctl_tcp_sack,
1142 .maxlen = sizeof(int),
1143 .mode = 0644,
1144 .proc_handler = proc_dointvec
1145 },
1146 {
1147 .procname = "tcp_window_scaling",
1148 .data = &init_net.ipv4.sysctl_tcp_window_scaling,
1149 .maxlen = sizeof(int),
1150 .mode = 0644,
1151 .proc_handler = proc_dointvec
1152 },
1153 {
1154 .procname = "tcp_timestamps",
1155 .data = &init_net.ipv4.sysctl_tcp_timestamps,
1156 .maxlen = sizeof(int),
1157 .mode = 0644,
1158 .proc_handler = proc_dointvec
1159 },
1160 { }
1161 };
1162
1163 static __net_init int ipv4_sysctl_init_net(struct net *net)
1164 {
1165 struct ctl_table *table;
1166
1167 table = ipv4_net_table;
1168 if (!net_eq(net, &init_net)) {
1169 int i;
1170
1171 table = kmemdup(table, sizeof(ipv4_net_table), GFP_KERNEL);
1172 if (!table)
1173 goto err_alloc;
1174
1175 /* Update the variables to point into the current struct net */
1176 for (i = 0; i < ARRAY_SIZE(ipv4_net_table) - 1; i++)
1177 table[i].data += (void *)net - (void *)&init_net;
1178 }
1179
1180 net->ipv4.ipv4_hdr = register_net_sysctl(net, "net/ipv4", table);
1181 if (!net->ipv4.ipv4_hdr)
1182 goto err_reg;
1183
1184 net->ipv4.sysctl_local_reserved_ports = kzalloc(65536 / 8, GFP_KERNEL);
1185 if (!net->ipv4.sysctl_local_reserved_ports)
1186 goto err_ports;
1187
1188 return 0;
1189
1190 err_ports:
1191 unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
1192 err_reg:
1193 if (!net_eq(net, &init_net))
1194 kfree(table);
1195 err_alloc:
1196 return -ENOMEM;
1197 }
1198
1199 static __net_exit void ipv4_sysctl_exit_net(struct net *net)
1200 {
1201 struct ctl_table *table;
1202
1203 kfree(net->ipv4.sysctl_local_reserved_ports);
1204 table = net->ipv4.ipv4_hdr->ctl_table_arg;
1205 unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
1206 kfree(table);
1207 }
1208
1209 static __net_initdata struct pernet_operations ipv4_sysctl_ops = {
1210 .init = ipv4_sysctl_init_net,
1211 .exit = ipv4_sysctl_exit_net,
1212 };
1213
1214 static __init int sysctl_ipv4_init(void)
1215 {
1216 struct ctl_table_header *hdr;
1217
1218 hdr = register_net_sysctl(&init_net, "net/ipv4", ipv4_table);
1219 if (!hdr)
1220 return -ENOMEM;
1221
1222 if (register_pernet_subsys(&ipv4_sysctl_ops)) {
1223 unregister_net_sysctl_table(hdr);
1224 return -ENOMEM;
1225 }
1226
1227 return 0;
1228 }
1229
1230 __initcall(sysctl_ipv4_init);
Linux with added FPC-III support