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 <net/snmp.h>
18#include <net/icmp.h>
19#include <net/ip.h>
20#include <net/route.h>
21#include <net/tcp.h>
22#include <net/udp.h>
23#include <net/cipso_ipv4.h>
24#include <net/inet_frag.h>
25#include <net/ping.h>
26
27static int zero;
28static int tcp_retr1_max = 255;
29static int ip_local_port_range_min[] = { 1, 1 };
30static int ip_local_port_range_max[] = { 65535, 65535 };
31static int tcp_adv_win_scale_min = -31;
32static int tcp_adv_win_scale_max = 31;
33static int ip_ttl_min = 1;
34static int ip_ttl_max = 255;
35static int ip_ping_group_range_min[] = { 0, 0 };
36static int ip_ping_group_range_max[] = { GID_T_MAX, GID_T_MAX };
37
38/* Update system visible IP port range */
39static void set_local_port_range(int range[2])
40{
41 write_seqlock(&sysctl_local_ports.lock);
42 sysctl_local_ports.range[0] = range[0];
43 sysctl_local_ports.range[1] = range[1];
44 write_sequnlock(&sysctl_local_ports.lock);
45}
46
47/* Validate changes from /proc interface. */
48static int ipv4_local_port_range(ctl_table *table, int write,
49 void __user *buffer,
50 size_t *lenp, loff_t *ppos)
51{
52 int ret;
53 int range[2];
54 ctl_table tmp = {
55 .data = &range,
56 .maxlen = sizeof(range),
57 .mode = table->mode,
58 .extra1 = &ip_local_port_range_min,
59 .extra2 = &ip_local_port_range_max,
60 };
61
62 inet_get_local_port_range(range, range + 1);
63 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
64
65 if (write && ret == 0) {
66 if (range[1] < range[0])
67 ret = -EINVAL;
68 else
69 set_local_port_range(range);
70 }
71
72 return ret;
73}
74
75
76void inet_get_ping_group_range_table(struct ctl_table *table, gid_t *low, gid_t *high)
77{
78 gid_t *data = table->data;
79 unsigned seq;
80 do {
81 seq = read_seqbegin(&sysctl_local_ports.lock);
82
83 *low = data[0];
84 *high = data[1];
85 } while (read_seqretry(&sysctl_local_ports.lock, seq));
86}
87
88/* Update system visible IP port range */
89static void set_ping_group_range(struct ctl_table *table, int range[2])
90{
91 gid_t *data = table->data;
92 write_seqlock(&sysctl_local_ports.lock);
93 data[0] = range[0];
94 data[1] = range[1];
95 write_sequnlock(&sysctl_local_ports.lock);
96}
97
98/* Validate changes from /proc interface. */
99static int ipv4_ping_group_range(ctl_table *table, int write,
100 void __user *buffer,
101 size_t *lenp, loff_t *ppos)
102{
103 int ret;
104 gid_t range[2];
105 ctl_table tmp = {
106 .data = &range,
107 .maxlen = sizeof(range),
108 .mode = table->mode,
109 .extra1 = &ip_ping_group_range_min,
110 .extra2 = &ip_ping_group_range_max,
111 };
112
113 inet_get_ping_group_range_table(table, range, range + 1);
114 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
115
116 if (write && ret == 0)
117 set_ping_group_range(table, range);
118
119 return ret;
120}
121
122static int proc_tcp_congestion_control(ctl_table *ctl, int write,
123 void __user *buffer, size_t *lenp, loff_t *ppos)
124{
125 char val[TCP_CA_NAME_MAX];
126 ctl_table tbl = {
127 .data = val,
128 .maxlen = TCP_CA_NAME_MAX,
129 };
130 int ret;
131
132 tcp_get_default_congestion_control(val);
133
134 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
135 if (write && ret == 0)
136 ret = tcp_set_default_congestion_control(val);
137 return ret;
138}
139
140static int proc_tcp_available_congestion_control(ctl_table *ctl,
141 int write,
142 void __user *buffer, size_t *lenp,
143 loff_t *ppos)
144{
145 ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX, };
146 int ret;
147
148 tbl.data = kmalloc(tbl.maxlen, GFP_USER);
149 if (!tbl.data)
150 return -ENOMEM;
151 tcp_get_available_congestion_control(tbl.data, TCP_CA_BUF_MAX);
152 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
153 kfree(tbl.data);
154 return ret;
155}
156
157static int proc_allowed_congestion_control(ctl_table *ctl,
158 int write,
159 void __user *buffer, size_t *lenp,
160 loff_t *ppos)
161{
162 ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX };
163 int ret;
164
165 tbl.data = kmalloc(tbl.maxlen, GFP_USER);
166 if (!tbl.data)
167 return -ENOMEM;
168
169 tcp_get_allowed_congestion_control(tbl.data, tbl.maxlen);
170 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
171 if (write && ret == 0)
172 ret = tcp_set_allowed_congestion_control(tbl.data);
173 kfree(tbl.data);
174 return ret;
175}
176
177static struct ctl_table ipv4_table[] = {
178 {
179 .procname = "tcp_timestamps",
180 .data = &sysctl_tcp_timestamps,
181 .maxlen = sizeof(int),
182 .mode = 0644,
183 .proc_handler = proc_dointvec
184 },
185 {
186 .procname = "tcp_window_scaling",
187 .data = &sysctl_tcp_window_scaling,
188 .maxlen = sizeof(int),
189 .mode = 0644,
190 .proc_handler = proc_dointvec
191 },
192 {
193 .procname = "tcp_sack",
194 .data = &sysctl_tcp_sack,
195 .maxlen = sizeof(int),
196 .mode = 0644,
197 .proc_handler = proc_dointvec
198 },
199 {
200 .procname = "tcp_retrans_collapse",
201 .data = &sysctl_tcp_retrans_collapse,
202 .maxlen = sizeof(int),
203 .mode = 0644,
204 .proc_handler = proc_dointvec
205 },
206 {
207 .procname = "ip_default_ttl",
208 .data = &sysctl_ip_default_ttl,
209 .maxlen = sizeof(int),
210 .mode = 0644,
211 .proc_handler = proc_dointvec_minmax,
212 .extra1 = &ip_ttl_min,
213 .extra2 = &ip_ttl_max,
214 },
215 {
216 .procname = "ip_no_pmtu_disc",
217 .data = &ipv4_config.no_pmtu_disc,
218 .maxlen = sizeof(int),
219 .mode = 0644,
220 .proc_handler = proc_dointvec
221 },
222 {
223 .procname = "ip_nonlocal_bind",
224 .data = &sysctl_ip_nonlocal_bind,
225 .maxlen = sizeof(int),
226 .mode = 0644,
227 .proc_handler = proc_dointvec
228 },
229 {
230 .procname = "tcp_syn_retries",
231 .data = &sysctl_tcp_syn_retries,
232 .maxlen = sizeof(int),
233 .mode = 0644,
234 .proc_handler = proc_dointvec
235 },
236 {
237 .procname = "tcp_synack_retries",
238 .data = &sysctl_tcp_synack_retries,
239 .maxlen = sizeof(int),
240 .mode = 0644,
241 .proc_handler = proc_dointvec
242 },
243 {
244 .procname = "tcp_max_orphans",
245 .data = &sysctl_tcp_max_orphans,
246 .maxlen = sizeof(int),
247 .mode = 0644,
248 .proc_handler = proc_dointvec
249 },
250 {
251 .procname = "tcp_max_tw_buckets",
252 .data = &tcp_death_row.sysctl_max_tw_buckets,
253 .maxlen = sizeof(int),
254 .mode = 0644,
255 .proc_handler = proc_dointvec
256 },
257 {
258 .procname = "ip_dynaddr",
259 .data = &sysctl_ip_dynaddr,
260 .maxlen = sizeof(int),
261 .mode = 0644,
262 .proc_handler = proc_dointvec
263 },
264 {
265 .procname = "tcp_keepalive_time",
266 .data = &sysctl_tcp_keepalive_time,
267 .maxlen = sizeof(int),
268 .mode = 0644,
269 .proc_handler = proc_dointvec_jiffies,
270 },
271 {
272 .procname = "tcp_keepalive_probes",
273 .data = &sysctl_tcp_keepalive_probes,
274 .maxlen = sizeof(int),
275 .mode = 0644,
276 .proc_handler = proc_dointvec
277 },
278 {
279 .procname = "tcp_keepalive_intvl",
280 .data = &sysctl_tcp_keepalive_intvl,
281 .maxlen = sizeof(int),
282 .mode = 0644,
283 .proc_handler = proc_dointvec_jiffies,
284 },
285 {
286 .procname = "tcp_retries1",
287 .data = &sysctl_tcp_retries1,
288 .maxlen = sizeof(int),
289 .mode = 0644,
290 .proc_handler = proc_dointvec_minmax,
291 .extra2 = &tcp_retr1_max
292 },
293 {
294 .procname = "tcp_retries2",
295 .data = &sysctl_tcp_retries2,
296 .maxlen = sizeof(int),
297 .mode = 0644,
298 .proc_handler = proc_dointvec
299 },
300 {
301 .procname = "tcp_fin_timeout",
302 .data = &sysctl_tcp_fin_timeout,
303 .maxlen = sizeof(int),
304 .mode = 0644,
305 .proc_handler = proc_dointvec_jiffies,
306 },
307#ifdef CONFIG_SYN_COOKIES
308 {
309 .procname = "tcp_syncookies",
310 .data = &sysctl_tcp_syncookies,
311 .maxlen = sizeof(int),
312 .mode = 0644,
313 .proc_handler = proc_dointvec
314 },
315#endif
316 {
317 .procname = "tcp_tw_recycle",
318 .data = &tcp_death_row.sysctl_tw_recycle,
319 .maxlen = sizeof(int),
320 .mode = 0644,
321 .proc_handler = proc_dointvec
322 },
323 {
324 .procname = "tcp_abort_on_overflow",
325 .data = &sysctl_tcp_abort_on_overflow,
326 .maxlen = sizeof(int),
327 .mode = 0644,
328 .proc_handler = proc_dointvec
329 },
330 {
331 .procname = "tcp_stdurg",
332 .data = &sysctl_tcp_stdurg,
333 .maxlen = sizeof(int),
334 .mode = 0644,
335 .proc_handler = proc_dointvec
336 },
337 {
338 .procname = "tcp_rfc1337",
339 .data = &sysctl_tcp_rfc1337,
340 .maxlen = sizeof(int),
341 .mode = 0644,
342 .proc_handler = proc_dointvec
343 },
344 {
345 .procname = "tcp_max_syn_backlog",
346 .data = &sysctl_max_syn_backlog,
347 .maxlen = sizeof(int),
348 .mode = 0644,
349 .proc_handler = proc_dointvec
350 },
351 {
352 .procname = "ip_local_port_range",
353 .data = &sysctl_local_ports.range,
354 .maxlen = sizeof(sysctl_local_ports.range),
355 .mode = 0644,
356 .proc_handler = ipv4_local_port_range,
357 },
358 {
359 .procname = "ip_local_reserved_ports",
360 .data = NULL, /* initialized in sysctl_ipv4_init */
361 .maxlen = 65536,
362 .mode = 0644,
363 .proc_handler = proc_do_large_bitmap,
364 },
365 {
366 .procname = "igmp_max_memberships",
367 .data = &sysctl_igmp_max_memberships,
368 .maxlen = sizeof(int),
369 .mode = 0644,
370 .proc_handler = proc_dointvec
371 },
372 {
373 .procname = "igmp_max_msf",
374 .data = &sysctl_igmp_max_msf,
375 .maxlen = sizeof(int),
376 .mode = 0644,
377 .proc_handler = proc_dointvec
378 },
379 {
380 .procname = "inet_peer_threshold",
381 .data = &inet_peer_threshold,
382 .maxlen = sizeof(int),
383 .mode = 0644,
384 .proc_handler = proc_dointvec
385 },
386 {
387 .procname = "inet_peer_minttl",
388 .data = &inet_peer_minttl,
389 .maxlen = sizeof(int),
390 .mode = 0644,
391 .proc_handler = proc_dointvec_jiffies,
392 },
393 {
394 .procname = "inet_peer_maxttl",
395 .data = &inet_peer_maxttl,
396 .maxlen = sizeof(int),
397 .mode = 0644,
398 .proc_handler = proc_dointvec_jiffies,
399 },
400 {
401 .procname = "tcp_orphan_retries",
402 .data = &sysctl_tcp_orphan_retries,
403 .maxlen = sizeof(int),
404 .mode = 0644,
405 .proc_handler = proc_dointvec
406 },
407 {
408 .procname = "tcp_fack",
409 .data = &sysctl_tcp_fack,
410 .maxlen = sizeof(int),
411 .mode = 0644,
412 .proc_handler = proc_dointvec
413 },
414 {
415 .procname = "tcp_reordering",
416 .data = &sysctl_tcp_reordering,
417 .maxlen = sizeof(int),
418 .mode = 0644,
419 .proc_handler = proc_dointvec
420 },
421 {
422 .procname = "tcp_ecn",
423 .data = &sysctl_tcp_ecn,
424 .maxlen = sizeof(int),
425 .mode = 0644,
426 .proc_handler = proc_dointvec
427 },
428 {
429 .procname = "tcp_dsack",
430 .data = &sysctl_tcp_dsack,
431 .maxlen = sizeof(int),
432 .mode = 0644,
433 .proc_handler = proc_dointvec
434 },
435 {
436 .procname = "tcp_mem",
437 .data = &sysctl_tcp_mem,
438 .maxlen = sizeof(sysctl_tcp_mem),
439 .mode = 0644,
440 .proc_handler = proc_doulongvec_minmax
441 },
442 {
443 .procname = "tcp_wmem",
444 .data = &sysctl_tcp_wmem,
445 .maxlen = sizeof(sysctl_tcp_wmem),
446 .mode = 0644,
447 .proc_handler = proc_dointvec
448 },
449 {
450 .procname = "tcp_rmem",
451 .data = &sysctl_tcp_rmem,
452 .maxlen = sizeof(sysctl_tcp_rmem),
453 .mode = 0644,
454 .proc_handler = proc_dointvec
455 },
456 {
457 .procname = "tcp_app_win",
458 .data = &sysctl_tcp_app_win,
459 .maxlen = sizeof(int),
460 .mode = 0644,
461 .proc_handler = proc_dointvec
462 },
463 {
464 .procname = "tcp_adv_win_scale",
465 .data = &sysctl_tcp_adv_win_scale,
466 .maxlen = sizeof(int),
467 .mode = 0644,
468 .proc_handler = proc_dointvec_minmax,
469 .extra1 = &tcp_adv_win_scale_min,
470 .extra2 = &tcp_adv_win_scale_max,
471 },
472 {
473 .procname = "tcp_tw_reuse",
474 .data = &sysctl_tcp_tw_reuse,
475 .maxlen = sizeof(int),
476 .mode = 0644,
477 .proc_handler = proc_dointvec
478 },
479 {
480 .procname = "tcp_frto",
481 .data = &sysctl_tcp_frto,
482 .maxlen = sizeof(int),
483 .mode = 0644,
484 .proc_handler = proc_dointvec
485 },
486 {
487 .procname = "tcp_frto_response",
488 .data = &sysctl_tcp_frto_response,
489 .maxlen = sizeof(int),
490 .mode = 0644,
491 .proc_handler = proc_dointvec
492 },
493 {
494 .procname = "tcp_low_latency",
495 .data = &sysctl_tcp_low_latency,
496 .maxlen = sizeof(int),
497 .mode = 0644,
498 .proc_handler = proc_dointvec
499 },
500 {
501 .procname = "tcp_no_metrics_save",
502 .data = &sysctl_tcp_nometrics_save,
503 .maxlen = sizeof(int),
504 .mode = 0644,
505 .proc_handler = proc_dointvec,
506 },
507 {
508 .procname = "tcp_moderate_rcvbuf",
509 .data = &sysctl_tcp_moderate_rcvbuf,
510 .maxlen = sizeof(int),
511 .mode = 0644,
512 .proc_handler = proc_dointvec,
513 },
514 {
515 .procname = "tcp_tso_win_divisor",
516 .data = &sysctl_tcp_tso_win_divisor,
517 .maxlen = sizeof(int),
518 .mode = 0644,
519 .proc_handler = proc_dointvec,
520 },
521 {
522 .procname = "tcp_congestion_control",
523 .mode = 0644,
524 .maxlen = TCP_CA_NAME_MAX,
525 .proc_handler = proc_tcp_congestion_control,
526 },
527 {
528 .procname = "tcp_abc",
529 .data = &sysctl_tcp_abc,
530 .maxlen = sizeof(int),
531 .mode = 0644,
532 .proc_handler = proc_dointvec,
533 },
534 {
535 .procname = "tcp_mtu_probing",
536 .data = &sysctl_tcp_mtu_probing,
537 .maxlen = sizeof(int),
538 .mode = 0644,
539 .proc_handler = proc_dointvec,
540 },
541 {
542 .procname = "tcp_base_mss",
543 .data = &sysctl_tcp_base_mss,
544 .maxlen = sizeof(int),
545 .mode = 0644,
546 .proc_handler = proc_dointvec,
547 },
548 {
549 .procname = "tcp_workaround_signed_windows",
550 .data = &sysctl_tcp_workaround_signed_windows,
551 .maxlen = sizeof(int),
552 .mode = 0644,
553 .proc_handler = proc_dointvec
554 },
555#ifdef CONFIG_NET_DMA
556 {
557 .procname = "tcp_dma_copybreak",
558 .data = &sysctl_tcp_dma_copybreak,
559 .maxlen = sizeof(int),
560 .mode = 0644,
561 .proc_handler = proc_dointvec
562 },
563#endif
564 {
565 .procname = "tcp_slow_start_after_idle",
566 .data = &sysctl_tcp_slow_start_after_idle,
567 .maxlen = sizeof(int),
568 .mode = 0644,
569 .proc_handler = proc_dointvec
570 },
571#ifdef CONFIG_NETLABEL
572 {
573 .procname = "cipso_cache_enable",
574 .data = &cipso_v4_cache_enabled,
575 .maxlen = sizeof(int),
576 .mode = 0644,
577 .proc_handler = proc_dointvec,
578 },
579 {
580 .procname = "cipso_cache_bucket_size",
581 .data = &cipso_v4_cache_bucketsize,
582 .maxlen = sizeof(int),
583 .mode = 0644,
584 .proc_handler = proc_dointvec,
585 },
586 {
587 .procname = "cipso_rbm_optfmt",
588 .data = &cipso_v4_rbm_optfmt,
589 .maxlen = sizeof(int),
590 .mode = 0644,
591 .proc_handler = proc_dointvec,
592 },
593 {
594 .procname = "cipso_rbm_strictvalid",
595 .data = &cipso_v4_rbm_strictvalid,
596 .maxlen = sizeof(int),
597 .mode = 0644,
598 .proc_handler = proc_dointvec,
599 },
600#endif /* CONFIG_NETLABEL */
601 {
602 .procname = "tcp_available_congestion_control",
603 .maxlen = TCP_CA_BUF_MAX,
604 .mode = 0444,
605 .proc_handler = proc_tcp_available_congestion_control,
606 },
607 {
608 .procname = "tcp_allowed_congestion_control",
609 .maxlen = TCP_CA_BUF_MAX,
610 .mode = 0644,
611 .proc_handler = proc_allowed_congestion_control,
612 },
613 {
614 .procname = "tcp_max_ssthresh",
615 .data = &sysctl_tcp_max_ssthresh,
616 .maxlen = sizeof(int),
617 .mode = 0644,
618 .proc_handler = proc_dointvec,
619 },
620 {
621 .procname = "tcp_cookie_size",
622 .data = &sysctl_tcp_cookie_size,
623 .maxlen = sizeof(int),
624 .mode = 0644,
625 .proc_handler = proc_dointvec
626 },
627 {
628 .procname = "tcp_thin_linear_timeouts",
629 .data = &sysctl_tcp_thin_linear_timeouts,
630 .maxlen = sizeof(int),
631 .mode = 0644,
632 .proc_handler = proc_dointvec
633 },
634 {
635 .procname = "tcp_thin_dupack",
636 .data = &sysctl_tcp_thin_dupack,
637 .maxlen = sizeof(int),
638 .mode = 0644,
639 .proc_handler = proc_dointvec
640 },
641 {
642 .procname = "udp_mem",
643 .data = &sysctl_udp_mem,
644 .maxlen = sizeof(sysctl_udp_mem),
645 .mode = 0644,
646 .proc_handler = proc_doulongvec_minmax,
647 },
648 {
649 .procname = "udp_rmem_min",
650 .data = &sysctl_udp_rmem_min,
651 .maxlen = sizeof(sysctl_udp_rmem_min),
652 .mode = 0644,
653 .proc_handler = proc_dointvec_minmax,
654 .extra1 = &zero
655 },
656 {
657 .procname = "udp_wmem_min",
658 .data = &sysctl_udp_wmem_min,
659 .maxlen = sizeof(sysctl_udp_wmem_min),
660 .mode = 0644,
661 .proc_handler = proc_dointvec_minmax,
662 .extra1 = &zero
663 },
664 { }
665};
666
667static struct ctl_table ipv4_net_table[] = {
668 {
669 .procname = "icmp_echo_ignore_all",
670 .data = &init_net.ipv4.sysctl_icmp_echo_ignore_all,
671 .maxlen = sizeof(int),
672 .mode = 0644,
673 .proc_handler = proc_dointvec
674 },
675 {
676 .procname = "icmp_echo_ignore_broadcasts",
677 .data = &init_net.ipv4.sysctl_icmp_echo_ignore_broadcasts,
678 .maxlen = sizeof(int),
679 .mode = 0644,
680 .proc_handler = proc_dointvec
681 },
682 {
683 .procname = "icmp_ignore_bogus_error_responses",
684 .data = &init_net.ipv4.sysctl_icmp_ignore_bogus_error_responses,
685 .maxlen = sizeof(int),
686 .mode = 0644,
687 .proc_handler = proc_dointvec
688 },
689 {
690 .procname = "icmp_errors_use_inbound_ifaddr",
691 .data = &init_net.ipv4.sysctl_icmp_errors_use_inbound_ifaddr,
692 .maxlen = sizeof(int),
693 .mode = 0644,
694 .proc_handler = proc_dointvec
695 },
696 {
697 .procname = "icmp_ratelimit",
698 .data = &init_net.ipv4.sysctl_icmp_ratelimit,
699 .maxlen = sizeof(int),
700 .mode = 0644,
701 .proc_handler = proc_dointvec_ms_jiffies,
702 },
703 {
704 .procname = "icmp_ratemask",
705 .data = &init_net.ipv4.sysctl_icmp_ratemask,
706 .maxlen = sizeof(int),
707 .mode = 0644,
708 .proc_handler = proc_dointvec
709 },
710 {
711 .procname = "rt_cache_rebuild_count",
712 .data = &init_net.ipv4.sysctl_rt_cache_rebuild_count,
713 .maxlen = sizeof(int),
714 .mode = 0644,
715 .proc_handler = proc_dointvec
716 },
717 {
718 .procname = "ping_group_range",
719 .data = &init_net.ipv4.sysctl_ping_group_range,
720 .maxlen = sizeof(init_net.ipv4.sysctl_ping_group_range),
721 .mode = 0644,
722 .proc_handler = ipv4_ping_group_range,
723 },
724 { }
725};
726
727struct ctl_path net_ipv4_ctl_path[] = {
728 { .procname = "net", },
729 { .procname = "ipv4", },
730 { },
731};
732EXPORT_SYMBOL_GPL(net_ipv4_ctl_path);
733
734static __net_init int ipv4_sysctl_init_net(struct net *net)
735{
736 struct ctl_table *table;
737
738 table = ipv4_net_table;
739 if (!net_eq(net, &init_net)) {
740 table = kmemdup(table, sizeof(ipv4_net_table), GFP_KERNEL);
741 if (table == NULL)
742 goto err_alloc;
743
744 table[0].data =
745 &net->ipv4.sysctl_icmp_echo_ignore_all;
746 table[1].data =
747 &net->ipv4.sysctl_icmp_echo_ignore_broadcasts;
748 table[2].data =
749 &net->ipv4.sysctl_icmp_ignore_bogus_error_responses;
750 table[3].data =
751 &net->ipv4.sysctl_icmp_errors_use_inbound_ifaddr;
752 table[4].data =
753 &net->ipv4.sysctl_icmp_ratelimit;
754 table[5].data =
755 &net->ipv4.sysctl_icmp_ratemask;
756 table[6].data =
757 &net->ipv4.sysctl_rt_cache_rebuild_count;
758 table[7].data =
759 &net->ipv4.sysctl_ping_group_range;
760
761 }
762
763 /*
764 * Sane defaults - nobody may create ping sockets.
765 * Boot scripts should set this to distro-specific group.
766 */
767 net->ipv4.sysctl_ping_group_range[0] = 1;
768 net->ipv4.sysctl_ping_group_range[1] = 0;
769
770 net->ipv4.sysctl_rt_cache_rebuild_count = 4;
771
772 net->ipv4.ipv4_hdr = register_net_sysctl_table(net,
773 net_ipv4_ctl_path, table);
774 if (net->ipv4.ipv4_hdr == NULL)
775 goto err_reg;
776
777 return 0;
778
779err_reg:
780 if (!net_eq(net, &init_net))
781 kfree(table);
782err_alloc:
783 return -ENOMEM;
784}
785
786static __net_exit void ipv4_sysctl_exit_net(struct net *net)
787{
788 struct ctl_table *table;
789
790 table = net->ipv4.ipv4_hdr->ctl_table_arg;
791 unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
792 kfree(table);
793}
794
795static __net_initdata struct pernet_operations ipv4_sysctl_ops = {
796 .init = ipv4_sysctl_init_net,
797 .exit = ipv4_sysctl_exit_net,
798};
799
800static __init int sysctl_ipv4_init(void)
801{
802 struct ctl_table_header *hdr;
803 struct ctl_table *i;
804
805 for (i = ipv4_table; i->procname; i++) {
806 if (strcmp(i->procname, "ip_local_reserved_ports") == 0) {
807 i->data = sysctl_local_reserved_ports;
808 break;
809 }
810 }
811 if (!i->procname)
812 return -EINVAL;
813
814 hdr = register_sysctl_paths(net_ipv4_ctl_path, ipv4_table);
815 if (hdr == NULL)
816 return -ENOMEM;
817
818 if (register_pernet_subsys(&ipv4_sysctl_ops)) {
819 unregister_sysctl_table(hdr);
820 return -ENOMEM;
821 }
822
823 return 0;
824}
825
826__initcall(sysctl_ipv4_init);
1/*
2 * sysctl_net_ipv4.c: sysctl interface to net IPV4 subsystem.
3 *
4 * Begun April 1, 1996, Mike Shaver.
5 * Added /proc/sys/net/ipv4 directory entry (empty =) ). [MS]
6 */
7
8#include <linux/mm.h>
9#include <linux/module.h>
10#include <linux/sysctl.h>
11#include <linux/igmp.h>
12#include <linux/inetdevice.h>
13#include <linux/seqlock.h>
14#include <linux/init.h>
15#include <linux/slab.h>
16#include <linux/nsproxy.h>
17#include <linux/swap.h>
18#include <net/snmp.h>
19#include <net/icmp.h>
20#include <net/ip.h>
21#include <net/route.h>
22#include <net/tcp.h>
23#include <net/udp.h>
24#include <net/cipso_ipv4.h>
25#include <net/inet_frag.h>
26#include <net/ping.h>
27
28static int zero;
29static int one = 1;
30static int four = 4;
31static int thousand = 1000;
32static int gso_max_segs = GSO_MAX_SEGS;
33static int tcp_retr1_max = 255;
34static int ip_local_port_range_min[] = { 1, 1 };
35static int ip_local_port_range_max[] = { 65535, 65535 };
36static int tcp_adv_win_scale_min = -31;
37static int tcp_adv_win_scale_max = 31;
38static int ip_ttl_min = 1;
39static int ip_ttl_max = 255;
40static int tcp_syn_retries_min = 1;
41static int tcp_syn_retries_max = MAX_TCP_SYNCNT;
42static int ip_ping_group_range_min[] = { 0, 0 };
43static int ip_ping_group_range_max[] = { GID_T_MAX, GID_T_MAX };
44
45/* Update system visible IP port range */
46static void set_local_port_range(struct net *net, int range[2])
47{
48 bool same_parity = !((range[0] ^ range[1]) & 1);
49
50 write_seqlock_bh(&net->ipv4.ip_local_ports.lock);
51 if (same_parity && !net->ipv4.ip_local_ports.warned) {
52 net->ipv4.ip_local_ports.warned = true;
53 pr_err_ratelimited("ip_local_port_range: prefer different parity for start/end values.\n");
54 }
55 net->ipv4.ip_local_ports.range[0] = range[0];
56 net->ipv4.ip_local_ports.range[1] = range[1];
57 write_sequnlock_bh(&net->ipv4.ip_local_ports.lock);
58}
59
60/* Validate changes from /proc interface. */
61static int ipv4_local_port_range(struct ctl_table *table, int write,
62 void __user *buffer,
63 size_t *lenp, loff_t *ppos)
64{
65 struct net *net =
66 container_of(table->data, struct net, ipv4.ip_local_ports.range);
67 int ret;
68 int range[2];
69 struct ctl_table tmp = {
70 .data = &range,
71 .maxlen = sizeof(range),
72 .mode = table->mode,
73 .extra1 = &ip_local_port_range_min,
74 .extra2 = &ip_local_port_range_max,
75 };
76
77 inet_get_local_port_range(net, &range[0], &range[1]);
78
79 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
80
81 if (write && ret == 0) {
82 if (range[1] < range[0])
83 ret = -EINVAL;
84 else
85 set_local_port_range(net, range);
86 }
87
88 return ret;
89}
90
91
92static void inet_get_ping_group_range_table(struct ctl_table *table, kgid_t *low, kgid_t *high)
93{
94 kgid_t *data = table->data;
95 struct net *net =
96 container_of(table->data, struct net, ipv4.ping_group_range.range);
97 unsigned int seq;
98 do {
99 seq = read_seqbegin(&net->ipv4.ip_local_ports.lock);
100
101 *low = data[0];
102 *high = data[1];
103 } while (read_seqretry(&net->ipv4.ip_local_ports.lock, seq));
104}
105
106/* Update system visible IP port range */
107static void set_ping_group_range(struct ctl_table *table, kgid_t low, kgid_t high)
108{
109 kgid_t *data = table->data;
110 struct net *net =
111 container_of(table->data, struct net, ipv4.ping_group_range.range);
112 write_seqlock(&net->ipv4.ip_local_ports.lock);
113 data[0] = low;
114 data[1] = high;
115 write_sequnlock(&net->ipv4.ip_local_ports.lock);
116}
117
118/* Validate changes from /proc interface. */
119static int ipv4_ping_group_range(struct ctl_table *table, int write,
120 void __user *buffer,
121 size_t *lenp, loff_t *ppos)
122{
123 struct user_namespace *user_ns = current_user_ns();
124 int ret;
125 gid_t urange[2];
126 kgid_t low, high;
127 struct ctl_table tmp = {
128 .data = &urange,
129 .maxlen = sizeof(urange),
130 .mode = table->mode,
131 .extra1 = &ip_ping_group_range_min,
132 .extra2 = &ip_ping_group_range_max,
133 };
134
135 inet_get_ping_group_range_table(table, &low, &high);
136 urange[0] = from_kgid_munged(user_ns, low);
137 urange[1] = from_kgid_munged(user_ns, high);
138 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
139
140 if (write && ret == 0) {
141 low = make_kgid(user_ns, urange[0]);
142 high = make_kgid(user_ns, urange[1]);
143 if (!gid_valid(low) || !gid_valid(high) ||
144 (urange[1] < urange[0]) || gid_lt(high, low)) {
145 low = make_kgid(&init_user_ns, 1);
146 high = make_kgid(&init_user_ns, 0);
147 }
148 set_ping_group_range(table, low, high);
149 }
150
151 return ret;
152}
153
154static int proc_tcp_congestion_control(struct ctl_table *ctl, int write,
155 void __user *buffer, size_t *lenp, loff_t *ppos)
156{
157 char val[TCP_CA_NAME_MAX];
158 struct ctl_table tbl = {
159 .data = val,
160 .maxlen = TCP_CA_NAME_MAX,
161 };
162 int ret;
163
164 tcp_get_default_congestion_control(val);
165
166 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
167 if (write && ret == 0)
168 ret = tcp_set_default_congestion_control(val);
169 return ret;
170}
171
172static int proc_tcp_available_congestion_control(struct ctl_table *ctl,
173 int write,
174 void __user *buffer, size_t *lenp,
175 loff_t *ppos)
176{
177 struct ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX, };
178 int ret;
179
180 tbl.data = kmalloc(tbl.maxlen, GFP_USER);
181 if (!tbl.data)
182 return -ENOMEM;
183 tcp_get_available_congestion_control(tbl.data, TCP_CA_BUF_MAX);
184 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
185 kfree(tbl.data);
186 return ret;
187}
188
189static int proc_allowed_congestion_control(struct ctl_table *ctl,
190 int write,
191 void __user *buffer, size_t *lenp,
192 loff_t *ppos)
193{
194 struct ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX };
195 int ret;
196
197 tbl.data = kmalloc(tbl.maxlen, GFP_USER);
198 if (!tbl.data)
199 return -ENOMEM;
200
201 tcp_get_allowed_congestion_control(tbl.data, tbl.maxlen);
202 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
203 if (write && ret == 0)
204 ret = tcp_set_allowed_congestion_control(tbl.data);
205 kfree(tbl.data);
206 return ret;
207}
208
209static int proc_tcp_fastopen_key(struct ctl_table *ctl, int write,
210 void __user *buffer, size_t *lenp,
211 loff_t *ppos)
212{
213 struct ctl_table tbl = { .maxlen = (TCP_FASTOPEN_KEY_LENGTH * 2 + 10) };
214 struct tcp_fastopen_context *ctxt;
215 int ret;
216 u32 user_key[4]; /* 16 bytes, matching TCP_FASTOPEN_KEY_LENGTH */
217
218 tbl.data = kmalloc(tbl.maxlen, GFP_KERNEL);
219 if (!tbl.data)
220 return -ENOMEM;
221
222 rcu_read_lock();
223 ctxt = rcu_dereference(tcp_fastopen_ctx);
224 if (ctxt)
225 memcpy(user_key, ctxt->key, TCP_FASTOPEN_KEY_LENGTH);
226 else
227 memset(user_key, 0, sizeof(user_key));
228 rcu_read_unlock();
229
230 snprintf(tbl.data, tbl.maxlen, "%08x-%08x-%08x-%08x",
231 user_key[0], user_key[1], user_key[2], user_key[3]);
232 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
233
234 if (write && ret == 0) {
235 if (sscanf(tbl.data, "%x-%x-%x-%x", user_key, user_key + 1,
236 user_key + 2, user_key + 3) != 4) {
237 ret = -EINVAL;
238 goto bad_key;
239 }
240 /* Generate a dummy secret but don't publish it. This
241 * is needed so we don't regenerate a new key on the
242 * first invocation of tcp_fastopen_cookie_gen
243 */
244 tcp_fastopen_init_key_once(false);
245 tcp_fastopen_reset_cipher(user_key, TCP_FASTOPEN_KEY_LENGTH);
246 }
247
248bad_key:
249 pr_debug("proc FO key set 0x%x-%x-%x-%x <- 0x%s: %u\n",
250 user_key[0], user_key[1], user_key[2], user_key[3],
251 (char *)tbl.data, ret);
252 kfree(tbl.data);
253 return ret;
254}
255
256static struct ctl_table ipv4_table[] = {
257 {
258 .procname = "tcp_timestamps",
259 .data = &sysctl_tcp_timestamps,
260 .maxlen = sizeof(int),
261 .mode = 0644,
262 .proc_handler = proc_dointvec
263 },
264 {
265 .procname = "tcp_window_scaling",
266 .data = &sysctl_tcp_window_scaling,
267 .maxlen = sizeof(int),
268 .mode = 0644,
269 .proc_handler = proc_dointvec
270 },
271 {
272 .procname = "tcp_sack",
273 .data = &sysctl_tcp_sack,
274 .maxlen = sizeof(int),
275 .mode = 0644,
276 .proc_handler = proc_dointvec
277 },
278 {
279 .procname = "tcp_retrans_collapse",
280 .data = &sysctl_tcp_retrans_collapse,
281 .maxlen = sizeof(int),
282 .mode = 0644,
283 .proc_handler = proc_dointvec
284 },
285 {
286 .procname = "tcp_max_orphans",
287 .data = &sysctl_tcp_max_orphans,
288 .maxlen = sizeof(int),
289 .mode = 0644,
290 .proc_handler = proc_dointvec
291 },
292 {
293 .procname = "tcp_max_tw_buckets",
294 .data = &tcp_death_row.sysctl_max_tw_buckets,
295 .maxlen = sizeof(int),
296 .mode = 0644,
297 .proc_handler = proc_dointvec
298 },
299 {
300 .procname = "tcp_fastopen",
301 .data = &sysctl_tcp_fastopen,
302 .maxlen = sizeof(int),
303 .mode = 0644,
304 .proc_handler = proc_dointvec,
305 },
306 {
307 .procname = "tcp_fastopen_key",
308 .mode = 0600,
309 .maxlen = ((TCP_FASTOPEN_KEY_LENGTH * 2) + 10),
310 .proc_handler = proc_tcp_fastopen_key,
311 },
312 {
313 .procname = "tcp_tw_recycle",
314 .data = &tcp_death_row.sysctl_tw_recycle,
315 .maxlen = sizeof(int),
316 .mode = 0644,
317 .proc_handler = proc_dointvec
318 },
319 {
320 .procname = "tcp_abort_on_overflow",
321 .data = &sysctl_tcp_abort_on_overflow,
322 .maxlen = sizeof(int),
323 .mode = 0644,
324 .proc_handler = proc_dointvec
325 },
326 {
327 .procname = "tcp_stdurg",
328 .data = &sysctl_tcp_stdurg,
329 .maxlen = sizeof(int),
330 .mode = 0644,
331 .proc_handler = proc_dointvec
332 },
333 {
334 .procname = "tcp_rfc1337",
335 .data = &sysctl_tcp_rfc1337,
336 .maxlen = sizeof(int),
337 .mode = 0644,
338 .proc_handler = proc_dointvec
339 },
340 {
341 .procname = "tcp_max_syn_backlog",
342 .data = &sysctl_max_syn_backlog,
343 .maxlen = sizeof(int),
344 .mode = 0644,
345 .proc_handler = proc_dointvec
346 },
347 {
348 .procname = "inet_peer_threshold",
349 .data = &inet_peer_threshold,
350 .maxlen = sizeof(int),
351 .mode = 0644,
352 .proc_handler = proc_dointvec
353 },
354 {
355 .procname = "inet_peer_minttl",
356 .data = &inet_peer_minttl,
357 .maxlen = sizeof(int),
358 .mode = 0644,
359 .proc_handler = proc_dointvec_jiffies,
360 },
361 {
362 .procname = "inet_peer_maxttl",
363 .data = &inet_peer_maxttl,
364 .maxlen = sizeof(int),
365 .mode = 0644,
366 .proc_handler = proc_dointvec_jiffies,
367 },
368 {
369 .procname = "tcp_fack",
370 .data = &sysctl_tcp_fack,
371 .maxlen = sizeof(int),
372 .mode = 0644,
373 .proc_handler = proc_dointvec
374 },
375 {
376 .procname = "tcp_recovery",
377 .data = &sysctl_tcp_recovery,
378 .maxlen = sizeof(int),
379 .mode = 0644,
380 .proc_handler = proc_dointvec,
381 },
382 {
383 .procname = "tcp_max_reordering",
384 .data = &sysctl_tcp_max_reordering,
385 .maxlen = sizeof(int),
386 .mode = 0644,
387 .proc_handler = proc_dointvec
388 },
389 {
390 .procname = "tcp_dsack",
391 .data = &sysctl_tcp_dsack,
392 .maxlen = sizeof(int),
393 .mode = 0644,
394 .proc_handler = proc_dointvec
395 },
396 {
397 .procname = "tcp_mem",
398 .maxlen = sizeof(sysctl_tcp_mem),
399 .data = &sysctl_tcp_mem,
400 .mode = 0644,
401 .proc_handler = proc_doulongvec_minmax,
402 },
403 {
404 .procname = "tcp_wmem",
405 .data = &sysctl_tcp_wmem,
406 .maxlen = sizeof(sysctl_tcp_wmem),
407 .mode = 0644,
408 .proc_handler = proc_dointvec_minmax,
409 .extra1 = &one,
410 },
411 {
412 .procname = "tcp_rmem",
413 .data = &sysctl_tcp_rmem,
414 .maxlen = sizeof(sysctl_tcp_rmem),
415 .mode = 0644,
416 .proc_handler = proc_dointvec_minmax,
417 .extra1 = &one,
418 },
419 {
420 .procname = "tcp_app_win",
421 .data = &sysctl_tcp_app_win,
422 .maxlen = sizeof(int),
423 .mode = 0644,
424 .proc_handler = proc_dointvec
425 },
426 {
427 .procname = "tcp_adv_win_scale",
428 .data = &sysctl_tcp_adv_win_scale,
429 .maxlen = sizeof(int),
430 .mode = 0644,
431 .proc_handler = proc_dointvec_minmax,
432 .extra1 = &tcp_adv_win_scale_min,
433 .extra2 = &tcp_adv_win_scale_max,
434 },
435 {
436 .procname = "tcp_tw_reuse",
437 .data = &sysctl_tcp_tw_reuse,
438 .maxlen = sizeof(int),
439 .mode = 0644,
440 .proc_handler = proc_dointvec
441 },
442 {
443 .procname = "tcp_frto",
444 .data = &sysctl_tcp_frto,
445 .maxlen = sizeof(int),
446 .mode = 0644,
447 .proc_handler = proc_dointvec
448 },
449 {
450 .procname = "tcp_min_rtt_wlen",
451 .data = &sysctl_tcp_min_rtt_wlen,
452 .maxlen = sizeof(int),
453 .mode = 0644,
454 .proc_handler = proc_dointvec
455 },
456 {
457 .procname = "tcp_low_latency",
458 .data = &sysctl_tcp_low_latency,
459 .maxlen = sizeof(int),
460 .mode = 0644,
461 .proc_handler = proc_dointvec
462 },
463 {
464 .procname = "tcp_no_metrics_save",
465 .data = &sysctl_tcp_nometrics_save,
466 .maxlen = sizeof(int),
467 .mode = 0644,
468 .proc_handler = proc_dointvec,
469 },
470 {
471 .procname = "tcp_moderate_rcvbuf",
472 .data = &sysctl_tcp_moderate_rcvbuf,
473 .maxlen = sizeof(int),
474 .mode = 0644,
475 .proc_handler = proc_dointvec,
476 },
477 {
478 .procname = "tcp_tso_win_divisor",
479 .data = &sysctl_tcp_tso_win_divisor,
480 .maxlen = sizeof(int),
481 .mode = 0644,
482 .proc_handler = proc_dointvec,
483 },
484 {
485 .procname = "tcp_congestion_control",
486 .mode = 0644,
487 .maxlen = TCP_CA_NAME_MAX,
488 .proc_handler = proc_tcp_congestion_control,
489 },
490 {
491 .procname = "tcp_workaround_signed_windows",
492 .data = &sysctl_tcp_workaround_signed_windows,
493 .maxlen = sizeof(int),
494 .mode = 0644,
495 .proc_handler = proc_dointvec
496 },
497 {
498 .procname = "tcp_limit_output_bytes",
499 .data = &sysctl_tcp_limit_output_bytes,
500 .maxlen = sizeof(int),
501 .mode = 0644,
502 .proc_handler = proc_dointvec
503 },
504 {
505 .procname = "tcp_challenge_ack_limit",
506 .data = &sysctl_tcp_challenge_ack_limit,
507 .maxlen = sizeof(int),
508 .mode = 0644,
509 .proc_handler = proc_dointvec
510 },
511 {
512 .procname = "tcp_slow_start_after_idle",
513 .data = &sysctl_tcp_slow_start_after_idle,
514 .maxlen = sizeof(int),
515 .mode = 0644,
516 .proc_handler = proc_dointvec
517 },
518#ifdef CONFIG_NETLABEL
519 {
520 .procname = "cipso_cache_enable",
521 .data = &cipso_v4_cache_enabled,
522 .maxlen = sizeof(int),
523 .mode = 0644,
524 .proc_handler = proc_dointvec,
525 },
526 {
527 .procname = "cipso_cache_bucket_size",
528 .data = &cipso_v4_cache_bucketsize,
529 .maxlen = sizeof(int),
530 .mode = 0644,
531 .proc_handler = proc_dointvec,
532 },
533 {
534 .procname = "cipso_rbm_optfmt",
535 .data = &cipso_v4_rbm_optfmt,
536 .maxlen = sizeof(int),
537 .mode = 0644,
538 .proc_handler = proc_dointvec,
539 },
540 {
541 .procname = "cipso_rbm_strictvalid",
542 .data = &cipso_v4_rbm_strictvalid,
543 .maxlen = sizeof(int),
544 .mode = 0644,
545 .proc_handler = proc_dointvec,
546 },
547#endif /* CONFIG_NETLABEL */
548 {
549 .procname = "tcp_available_congestion_control",
550 .maxlen = TCP_CA_BUF_MAX,
551 .mode = 0444,
552 .proc_handler = proc_tcp_available_congestion_control,
553 },
554 {
555 .procname = "tcp_allowed_congestion_control",
556 .maxlen = TCP_CA_BUF_MAX,
557 .mode = 0644,
558 .proc_handler = proc_allowed_congestion_control,
559 },
560 {
561 .procname = "tcp_thin_linear_timeouts",
562 .data = &sysctl_tcp_thin_linear_timeouts,
563 .maxlen = sizeof(int),
564 .mode = 0644,
565 .proc_handler = proc_dointvec
566 },
567 {
568 .procname = "tcp_thin_dupack",
569 .data = &sysctl_tcp_thin_dupack,
570 .maxlen = sizeof(int),
571 .mode = 0644,
572 .proc_handler = proc_dointvec
573 },
574 {
575 .procname = "tcp_early_retrans",
576 .data = &sysctl_tcp_early_retrans,
577 .maxlen = sizeof(int),
578 .mode = 0644,
579 .proc_handler = proc_dointvec_minmax,
580 .extra1 = &zero,
581 .extra2 = &four,
582 },
583 {
584 .procname = "tcp_min_tso_segs",
585 .data = &sysctl_tcp_min_tso_segs,
586 .maxlen = sizeof(int),
587 .mode = 0644,
588 .proc_handler = proc_dointvec_minmax,
589 .extra1 = &one,
590 .extra2 = &gso_max_segs,
591 },
592 {
593 .procname = "tcp_pacing_ss_ratio",
594 .data = &sysctl_tcp_pacing_ss_ratio,
595 .maxlen = sizeof(int),
596 .mode = 0644,
597 .proc_handler = proc_dointvec_minmax,
598 .extra1 = &zero,
599 .extra2 = &thousand,
600 },
601 {
602 .procname = "tcp_pacing_ca_ratio",
603 .data = &sysctl_tcp_pacing_ca_ratio,
604 .maxlen = sizeof(int),
605 .mode = 0644,
606 .proc_handler = proc_dointvec_minmax,
607 .extra1 = &zero,
608 .extra2 = &thousand,
609 },
610 {
611 .procname = "tcp_autocorking",
612 .data = &sysctl_tcp_autocorking,
613 .maxlen = sizeof(int),
614 .mode = 0644,
615 .proc_handler = proc_dointvec_minmax,
616 .extra1 = &zero,
617 .extra2 = &one,
618 },
619 {
620 .procname = "tcp_invalid_ratelimit",
621 .data = &sysctl_tcp_invalid_ratelimit,
622 .maxlen = sizeof(int),
623 .mode = 0644,
624 .proc_handler = proc_dointvec_ms_jiffies,
625 },
626 {
627 .procname = "icmp_msgs_per_sec",
628 .data = &sysctl_icmp_msgs_per_sec,
629 .maxlen = sizeof(int),
630 .mode = 0644,
631 .proc_handler = proc_dointvec_minmax,
632 .extra1 = &zero,
633 },
634 {
635 .procname = "icmp_msgs_burst",
636 .data = &sysctl_icmp_msgs_burst,
637 .maxlen = sizeof(int),
638 .mode = 0644,
639 .proc_handler = proc_dointvec_minmax,
640 .extra1 = &zero,
641 },
642 {
643 .procname = "udp_mem",
644 .data = &sysctl_udp_mem,
645 .maxlen = sizeof(sysctl_udp_mem),
646 .mode = 0644,
647 .proc_handler = proc_doulongvec_minmax,
648 },
649 {
650 .procname = "udp_rmem_min",
651 .data = &sysctl_udp_rmem_min,
652 .maxlen = sizeof(sysctl_udp_rmem_min),
653 .mode = 0644,
654 .proc_handler = proc_dointvec_minmax,
655 .extra1 = &one
656 },
657 {
658 .procname = "udp_wmem_min",
659 .data = &sysctl_udp_wmem_min,
660 .maxlen = sizeof(sysctl_udp_wmem_min),
661 .mode = 0644,
662 .proc_handler = proc_dointvec_minmax,
663 .extra1 = &one
664 },
665 { }
666};
667
668static struct ctl_table ipv4_net_table[] = {
669 {
670 .procname = "icmp_echo_ignore_all",
671 .data = &init_net.ipv4.sysctl_icmp_echo_ignore_all,
672 .maxlen = sizeof(int),
673 .mode = 0644,
674 .proc_handler = proc_dointvec
675 },
676 {
677 .procname = "icmp_echo_ignore_broadcasts",
678 .data = &init_net.ipv4.sysctl_icmp_echo_ignore_broadcasts,
679 .maxlen = sizeof(int),
680 .mode = 0644,
681 .proc_handler = proc_dointvec
682 },
683 {
684 .procname = "icmp_ignore_bogus_error_responses",
685 .data = &init_net.ipv4.sysctl_icmp_ignore_bogus_error_responses,
686 .maxlen = sizeof(int),
687 .mode = 0644,
688 .proc_handler = proc_dointvec
689 },
690 {
691 .procname = "icmp_errors_use_inbound_ifaddr",
692 .data = &init_net.ipv4.sysctl_icmp_errors_use_inbound_ifaddr,
693 .maxlen = sizeof(int),
694 .mode = 0644,
695 .proc_handler = proc_dointvec
696 },
697 {
698 .procname = "icmp_ratelimit",
699 .data = &init_net.ipv4.sysctl_icmp_ratelimit,
700 .maxlen = sizeof(int),
701 .mode = 0644,
702 .proc_handler = proc_dointvec_ms_jiffies,
703 },
704 {
705 .procname = "icmp_ratemask",
706 .data = &init_net.ipv4.sysctl_icmp_ratemask,
707 .maxlen = sizeof(int),
708 .mode = 0644,
709 .proc_handler = proc_dointvec
710 },
711 {
712 .procname = "ping_group_range",
713 .data = &init_net.ipv4.ping_group_range.range,
714 .maxlen = sizeof(gid_t)*2,
715 .mode = 0644,
716 .proc_handler = ipv4_ping_group_range,
717 },
718 {
719 .procname = "tcp_ecn",
720 .data = &init_net.ipv4.sysctl_tcp_ecn,
721 .maxlen = sizeof(int),
722 .mode = 0644,
723 .proc_handler = proc_dointvec
724 },
725 {
726 .procname = "tcp_ecn_fallback",
727 .data = &init_net.ipv4.sysctl_tcp_ecn_fallback,
728 .maxlen = sizeof(int),
729 .mode = 0644,
730 .proc_handler = proc_dointvec
731 },
732 {
733 .procname = "ip_dynaddr",
734 .data = &init_net.ipv4.sysctl_ip_dynaddr,
735 .maxlen = sizeof(int),
736 .mode = 0644,
737 .proc_handler = proc_dointvec
738 },
739 {
740 .procname = "ip_early_demux",
741 .data = &init_net.ipv4.sysctl_ip_early_demux,
742 .maxlen = sizeof(int),
743 .mode = 0644,
744 .proc_handler = proc_dointvec
745 },
746 {
747 .procname = "ip_default_ttl",
748 .data = &init_net.ipv4.sysctl_ip_default_ttl,
749 .maxlen = sizeof(int),
750 .mode = 0644,
751 .proc_handler = proc_dointvec_minmax,
752 .extra1 = &ip_ttl_min,
753 .extra2 = &ip_ttl_max,
754 },
755 {
756 .procname = "ip_local_port_range",
757 .maxlen = sizeof(init_net.ipv4.ip_local_ports.range),
758 .data = &init_net.ipv4.ip_local_ports.range,
759 .mode = 0644,
760 .proc_handler = ipv4_local_port_range,
761 },
762 {
763 .procname = "ip_local_reserved_ports",
764 .data = &init_net.ipv4.sysctl_local_reserved_ports,
765 .maxlen = 65536,
766 .mode = 0644,
767 .proc_handler = proc_do_large_bitmap,
768 },
769 {
770 .procname = "ip_no_pmtu_disc",
771 .data = &init_net.ipv4.sysctl_ip_no_pmtu_disc,
772 .maxlen = sizeof(int),
773 .mode = 0644,
774 .proc_handler = proc_dointvec
775 },
776 {
777 .procname = "ip_forward_use_pmtu",
778 .data = &init_net.ipv4.sysctl_ip_fwd_use_pmtu,
779 .maxlen = sizeof(int),
780 .mode = 0644,
781 .proc_handler = proc_dointvec,
782 },
783 {
784 .procname = "ip_nonlocal_bind",
785 .data = &init_net.ipv4.sysctl_ip_nonlocal_bind,
786 .maxlen = sizeof(int),
787 .mode = 0644,
788 .proc_handler = proc_dointvec
789 },
790 {
791 .procname = "fwmark_reflect",
792 .data = &init_net.ipv4.sysctl_fwmark_reflect,
793 .maxlen = sizeof(int),
794 .mode = 0644,
795 .proc_handler = proc_dointvec,
796 },
797 {
798 .procname = "tcp_fwmark_accept",
799 .data = &init_net.ipv4.sysctl_tcp_fwmark_accept,
800 .maxlen = sizeof(int),
801 .mode = 0644,
802 .proc_handler = proc_dointvec,
803 },
804#ifdef CONFIG_NET_L3_MASTER_DEV
805 {
806 .procname = "tcp_l3mdev_accept",
807 .data = &init_net.ipv4.sysctl_tcp_l3mdev_accept,
808 .maxlen = sizeof(int),
809 .mode = 0644,
810 .proc_handler = proc_dointvec_minmax,
811 .extra1 = &zero,
812 .extra2 = &one,
813 },
814#endif
815 {
816 .procname = "tcp_mtu_probing",
817 .data = &init_net.ipv4.sysctl_tcp_mtu_probing,
818 .maxlen = sizeof(int),
819 .mode = 0644,
820 .proc_handler = proc_dointvec,
821 },
822 {
823 .procname = "tcp_base_mss",
824 .data = &init_net.ipv4.sysctl_tcp_base_mss,
825 .maxlen = sizeof(int),
826 .mode = 0644,
827 .proc_handler = proc_dointvec,
828 },
829 {
830 .procname = "tcp_probe_threshold",
831 .data = &init_net.ipv4.sysctl_tcp_probe_threshold,
832 .maxlen = sizeof(int),
833 .mode = 0644,
834 .proc_handler = proc_dointvec,
835 },
836 {
837 .procname = "tcp_probe_interval",
838 .data = &init_net.ipv4.sysctl_tcp_probe_interval,
839 .maxlen = sizeof(int),
840 .mode = 0644,
841 .proc_handler = proc_dointvec,
842 },
843 {
844 .procname = "igmp_link_local_mcast_reports",
845 .data = &init_net.ipv4.sysctl_igmp_llm_reports,
846 .maxlen = sizeof(int),
847 .mode = 0644,
848 .proc_handler = proc_dointvec
849 },
850 {
851 .procname = "igmp_max_memberships",
852 .data = &init_net.ipv4.sysctl_igmp_max_memberships,
853 .maxlen = sizeof(int),
854 .mode = 0644,
855 .proc_handler = proc_dointvec
856 },
857 {
858 .procname = "igmp_max_msf",
859 .data = &init_net.ipv4.sysctl_igmp_max_msf,
860 .maxlen = sizeof(int),
861 .mode = 0644,
862 .proc_handler = proc_dointvec
863 },
864#ifdef CONFIG_IP_MULTICAST
865 {
866 .procname = "igmp_qrv",
867 .data = &init_net.ipv4.sysctl_igmp_qrv,
868 .maxlen = sizeof(int),
869 .mode = 0644,
870 .proc_handler = proc_dointvec_minmax,
871 .extra1 = &one
872 },
873#endif
874 {
875 .procname = "tcp_keepalive_time",
876 .data = &init_net.ipv4.sysctl_tcp_keepalive_time,
877 .maxlen = sizeof(int),
878 .mode = 0644,
879 .proc_handler = proc_dointvec_jiffies,
880 },
881 {
882 .procname = "tcp_keepalive_probes",
883 .data = &init_net.ipv4.sysctl_tcp_keepalive_probes,
884 .maxlen = sizeof(int),
885 .mode = 0644,
886 .proc_handler = proc_dointvec
887 },
888 {
889 .procname = "tcp_keepalive_intvl",
890 .data = &init_net.ipv4.sysctl_tcp_keepalive_intvl,
891 .maxlen = sizeof(int),
892 .mode = 0644,
893 .proc_handler = proc_dointvec_jiffies,
894 },
895 {
896 .procname = "tcp_syn_retries",
897 .data = &init_net.ipv4.sysctl_tcp_syn_retries,
898 .maxlen = sizeof(int),
899 .mode = 0644,
900 .proc_handler = proc_dointvec_minmax,
901 .extra1 = &tcp_syn_retries_min,
902 .extra2 = &tcp_syn_retries_max
903 },
904 {
905 .procname = "tcp_synack_retries",
906 .data = &init_net.ipv4.sysctl_tcp_synack_retries,
907 .maxlen = sizeof(int),
908 .mode = 0644,
909 .proc_handler = proc_dointvec
910 },
911#ifdef CONFIG_SYN_COOKIES
912 {
913 .procname = "tcp_syncookies",
914 .data = &init_net.ipv4.sysctl_tcp_syncookies,
915 .maxlen = sizeof(int),
916 .mode = 0644,
917 .proc_handler = proc_dointvec
918 },
919#endif
920 {
921 .procname = "tcp_reordering",
922 .data = &init_net.ipv4.sysctl_tcp_reordering,
923 .maxlen = sizeof(int),
924 .mode = 0644,
925 .proc_handler = proc_dointvec
926 },
927 {
928 .procname = "tcp_retries1",
929 .data = &init_net.ipv4.sysctl_tcp_retries1,
930 .maxlen = sizeof(int),
931 .mode = 0644,
932 .proc_handler = proc_dointvec_minmax,
933 .extra2 = &tcp_retr1_max
934 },
935 {
936 .procname = "tcp_retries2",
937 .data = &init_net.ipv4.sysctl_tcp_retries2,
938 .maxlen = sizeof(int),
939 .mode = 0644,
940 .proc_handler = proc_dointvec
941 },
942 {
943 .procname = "tcp_orphan_retries",
944 .data = &init_net.ipv4.sysctl_tcp_orphan_retries,
945 .maxlen = sizeof(int),
946 .mode = 0644,
947 .proc_handler = proc_dointvec
948 },
949 {
950 .procname = "tcp_fin_timeout",
951 .data = &init_net.ipv4.sysctl_tcp_fin_timeout,
952 .maxlen = sizeof(int),
953 .mode = 0644,
954 .proc_handler = proc_dointvec_jiffies,
955 },
956 {
957 .procname = "tcp_notsent_lowat",
958 .data = &init_net.ipv4.sysctl_tcp_notsent_lowat,
959 .maxlen = sizeof(unsigned int),
960 .mode = 0644,
961 .proc_handler = proc_dointvec,
962 },
963 { }
964};
965
966static __net_init int ipv4_sysctl_init_net(struct net *net)
967{
968 struct ctl_table *table;
969
970 table = ipv4_net_table;
971 if (!net_eq(net, &init_net)) {
972 int i;
973
974 table = kmemdup(table, sizeof(ipv4_net_table), GFP_KERNEL);
975 if (!table)
976 goto err_alloc;
977
978 /* Update the variables to point into the current struct net */
979 for (i = 0; i < ARRAY_SIZE(ipv4_net_table) - 1; i++)
980 table[i].data += (void *)net - (void *)&init_net;
981 }
982
983 net->ipv4.ipv4_hdr = register_net_sysctl(net, "net/ipv4", table);
984 if (!net->ipv4.ipv4_hdr)
985 goto err_reg;
986
987 net->ipv4.sysctl_local_reserved_ports = kzalloc(65536 / 8, GFP_KERNEL);
988 if (!net->ipv4.sysctl_local_reserved_ports)
989 goto err_ports;
990
991 net->ipv4.sysctl_ip_default_ttl = IPDEFTTL;
992 net->ipv4.sysctl_ip_dynaddr = 0;
993 net->ipv4.sysctl_ip_early_demux = 1;
994
995 return 0;
996
997err_ports:
998 unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
999err_reg:
1000 if (!net_eq(net, &init_net))
1001 kfree(table);
1002err_alloc:
1003 return -ENOMEM;
1004}
1005
1006static __net_exit void ipv4_sysctl_exit_net(struct net *net)
1007{
1008 struct ctl_table *table;
1009
1010 kfree(net->ipv4.sysctl_local_reserved_ports);
1011 table = net->ipv4.ipv4_hdr->ctl_table_arg;
1012 unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
1013 kfree(table);
1014}
1015
1016static __net_initdata struct pernet_operations ipv4_sysctl_ops = {
1017 .init = ipv4_sysctl_init_net,
1018 .exit = ipv4_sysctl_exit_net,
1019};
1020
1021static __init int sysctl_ipv4_init(void)
1022{
1023 struct ctl_table_header *hdr;
1024
1025 hdr = register_net_sysctl(&init_net, "net/ipv4", ipv4_table);
1026 if (!hdr)
1027 return -ENOMEM;
1028
1029 if (register_pernet_subsys(&ipv4_sysctl_ops)) {
1030 unregister_net_sysctl_table(hdr);
1031 return -ENOMEM;
1032 }
1033
1034 return 0;
1035}
1036
1037__initcall(sysctl_ipv4_init);