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