1 // SPDX-License-Identifier: GPL-2.0
2 /*
3 * sysctl_net_ipv4.c: sysctl interface to net IPV4 subsystem.
4 *
5 * Begun April 1, 1996, Mike Shaver.
6 * Added /proc/sys/net/ipv4 directory entry (empty =) ). [MS]
7 */
8
9 #include <linux/mm.h>
10 #include <linux/module.h>
11 #include <linux/sysctl.h>
12 #include <linux/igmp.h>
13 #include <linux/inetdevice.h>
14 #include <linux/seqlock.h>
15 #include <linux/init.h>
16 #include <linux/slab.h>
17 #include <linux/nsproxy.h>
18 #include <linux/swap.h>
19 #include <net/snmp.h>
20 #include <net/icmp.h>
21 #include <net/ip.h>
22 #include <net/route.h>
23 #include <net/tcp.h>
24 #include <net/udp.h>
25 #include <net/cipso_ipv4.h>
26 #include <net/inet_frag.h>
27 #include <net/ping.h>
28 #include <net/protocol.h>
29 #include <net/netevent.h>
30
31 static int two = 2;
32 static int four = 4;
33 static int thousand = 1000;
34 static int tcp_retr1_max = 255;
35 static int ip_local_port_range_min[] = { 1, 1 };
36 static int ip_local_port_range_max[] = { 65535, 65535 };
37 static int tcp_adv_win_scale_min = -31;
38 static int tcp_adv_win_scale_max = 31;
39 static int tcp_min_snd_mss_min = TCP_MIN_SND_MSS;
40 static int tcp_min_snd_mss_max = 65535;
41 static int ip_privileged_port_min;
42 static int ip_privileged_port_max = 65535;
43 static int ip_ttl_min = 1;
44 static int ip_ttl_max = 255;
45 static int tcp_syn_retries_min = 1;
46 static int tcp_syn_retries_max = MAX_TCP_SYNCNT;
47 static int ip_ping_group_range_min[] = { 0, 0 };
48 static int ip_ping_group_range_max[] = { GID_T_MAX, GID_T_MAX };
49 static int comp_sack_nr_max = 255;
50 static u32 u32_max_div_HZ = UINT_MAX / HZ;
51 static int one_day_secs = 24 * 3600;
52
53 /* obsolete */
54 static int sysctl_tcp_low_latency __read_mostly;
55
56 /* Update system visible IP port range */
set_local_port_range(struct net * net,int range[2])57 static void set_local_port_range(struct net *net, int range[2])
58 {
59 bool same_parity = !((range[0] ^ range[1]) & 1);
60
61 write_seqlock_bh(&net->ipv4.ip_local_ports.lock);
62 if (same_parity && !net->ipv4.ip_local_ports.warned) {
63 net->ipv4.ip_local_ports.warned = true;
64 pr_err_ratelimited("ip_local_port_range: prefer different parity for start/end values.\n");
65 }
66 net->ipv4.ip_local_ports.range[0] = range[0];
67 net->ipv4.ip_local_ports.range[1] = range[1];
68 write_sequnlock_bh(&net->ipv4.ip_local_ports.lock);
69 }
70
71 /* Validate changes from /proc interface. */
ipv4_local_port_range(struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)72 static int ipv4_local_port_range(struct ctl_table *table, int write,
73 void *buffer, size_t *lenp, loff_t *ppos)
74 {
75 struct net *net =
76 container_of(table->data, struct net, ipv4.ip_local_ports.range);
77 int ret;
78 int range[2];
79 struct ctl_table tmp = {
80 .data = &range,
81 .maxlen = sizeof(range),
82 .mode = table->mode,
83 .extra1 = &ip_local_port_range_min,
84 .extra2 = &ip_local_port_range_max,
85 };
86
87 inet_get_local_port_range(net, &range[0], &range[1]);
88
89 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
90
91 if (write && ret == 0) {
92 /* Ensure that the upper limit is not smaller than the lower,
93 * and that the lower does not encroach upon the privileged
94 * port limit.
95 */
96 if ((range[1] < range[0]) ||
97 (range[0] < READ_ONCE(net->ipv4.sysctl_ip_prot_sock)))
98 ret = -EINVAL;
99 else
100 set_local_port_range(net, range);
101 }
102
103 return ret;
104 }
105
106 /* Validate changes from /proc interface. */
ipv4_privileged_ports(struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)107 static int ipv4_privileged_ports(struct ctl_table *table, int write,
108 void *buffer, size_t *lenp, loff_t *ppos)
109 {
110 struct net *net = container_of(table->data, struct net,
111 ipv4.sysctl_ip_prot_sock);
112 int ret;
113 int pports;
114 int range[2];
115 struct ctl_table tmp = {
116 .data = &pports,
117 .maxlen = sizeof(pports),
118 .mode = table->mode,
119 .extra1 = &ip_privileged_port_min,
120 .extra2 = &ip_privileged_port_max,
121 };
122
123 pports = READ_ONCE(net->ipv4.sysctl_ip_prot_sock);
124
125 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
126
127 if (write && ret == 0) {
128 inet_get_local_port_range(net, &range[0], &range[1]);
129 /* Ensure that the local port range doesn't overlap with the
130 * privileged port range.
131 */
132 if (range[0] < pports)
133 ret = -EINVAL;
134 else
135 WRITE_ONCE(net->ipv4.sysctl_ip_prot_sock, pports);
136 }
137
138 return ret;
139 }
140
inet_get_ping_group_range_table(struct ctl_table * table,kgid_t * low,kgid_t * high)141 static void inet_get_ping_group_range_table(struct ctl_table *table, kgid_t *low, kgid_t *high)
142 {
143 kgid_t *data = table->data;
144 struct net *net =
145 container_of(table->data, struct net, ipv4.ping_group_range.range);
146 unsigned int seq;
147 do {
148 seq = read_seqbegin(&net->ipv4.ping_group_range.lock);
149
150 *low = data[0];
151 *high = data[1];
152 } while (read_seqretry(&net->ipv4.ping_group_range.lock, seq));
153 }
154
155 /* Update system visible IP port range */
set_ping_group_range(struct ctl_table * table,kgid_t low,kgid_t high)156 static void set_ping_group_range(struct ctl_table *table, kgid_t low, kgid_t high)
157 {
158 kgid_t *data = table->data;
159 struct net *net =
160 container_of(table->data, struct net, ipv4.ping_group_range.range);
161 write_seqlock(&net->ipv4.ping_group_range.lock);
162 data[0] = low;
163 data[1] = high;
164 write_sequnlock(&net->ipv4.ping_group_range.lock);
165 }
166
167 /* Validate changes from /proc interface. */
ipv4_ping_group_range(struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)168 static int ipv4_ping_group_range(struct ctl_table *table, int write,
169 void *buffer, size_t *lenp, loff_t *ppos)
170 {
171 struct user_namespace *user_ns = current_user_ns();
172 int ret;
173 gid_t urange[2];
174 kgid_t low, high;
175 struct ctl_table tmp = {
176 .data = &urange,
177 .maxlen = sizeof(urange),
178 .mode = table->mode,
179 .extra1 = &ip_ping_group_range_min,
180 .extra2 = &ip_ping_group_range_max,
181 };
182
183 inet_get_ping_group_range_table(table, &low, &high);
184 urange[0] = from_kgid_munged(user_ns, low);
185 urange[1] = from_kgid_munged(user_ns, high);
186 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
187
188 if (write && ret == 0) {
189 low = make_kgid(user_ns, urange[0]);
190 high = make_kgid(user_ns, urange[1]);
191 if (!gid_valid(low) || !gid_valid(high))
192 return -EINVAL;
193 if (urange[1] < urange[0] || gid_lt(high, low)) {
194 low = make_kgid(&init_user_ns, 1);
195 high = make_kgid(&init_user_ns, 0);
196 }
197 set_ping_group_range(table, low, high);
198 }
199
200 return ret;
201 }
202
ipv4_fwd_update_priority(struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)203 static int ipv4_fwd_update_priority(struct ctl_table *table, int write,
204 void *buffer, size_t *lenp, loff_t *ppos)
205 {
206 struct net *net;
207 int ret;
208
209 net = container_of(table->data, struct net,
210 ipv4.sysctl_ip_fwd_update_priority);
211 ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
212 if (write && ret == 0)
213 call_netevent_notifiers(NETEVENT_IPV4_FWD_UPDATE_PRIORITY_UPDATE,
214 net);
215
216 return ret;
217 }
218
proc_tcp_congestion_control(struct ctl_table * ctl,int write,void * buffer,size_t * lenp,loff_t * ppos)219 static int proc_tcp_congestion_control(struct ctl_table *ctl, int write,
220 void *buffer, size_t *lenp, loff_t *ppos)
221 {
222 struct net *net = container_of(ctl->data, struct net,
223 ipv4.tcp_congestion_control);
224 char val[TCP_CA_NAME_MAX];
225 struct ctl_table tbl = {
226 .data = val,
227 .maxlen = TCP_CA_NAME_MAX,
228 };
229 int ret;
230
231 tcp_get_default_congestion_control(net, val);
232
233 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
234 if (write && ret == 0)
235 ret = tcp_set_default_congestion_control(net, val);
236 return ret;
237 }
238
proc_tcp_available_congestion_control(struct ctl_table * ctl,int write,void * buffer,size_t * lenp,loff_t * ppos)239 static int proc_tcp_available_congestion_control(struct ctl_table *ctl,
240 int write, void *buffer,
241 size_t *lenp, loff_t *ppos)
242 {
243 struct ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX, };
244 int ret;
245
246 tbl.data = kmalloc(tbl.maxlen, GFP_USER);
247 if (!tbl.data)
248 return -ENOMEM;
249 tcp_get_available_congestion_control(tbl.data, TCP_CA_BUF_MAX);
250 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
251 kfree(tbl.data);
252 return ret;
253 }
254
proc_allowed_congestion_control(struct ctl_table * ctl,int write,void * buffer,size_t * lenp,loff_t * ppos)255 static int proc_allowed_congestion_control(struct ctl_table *ctl,
256 int write, void *buffer,
257 size_t *lenp, loff_t *ppos)
258 {
259 struct ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX };
260 int ret;
261
262 tbl.data = kmalloc(tbl.maxlen, GFP_USER);
263 if (!tbl.data)
264 return -ENOMEM;
265
266 tcp_get_allowed_congestion_control(tbl.data, tbl.maxlen);
267 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
268 if (write && ret == 0)
269 ret = tcp_set_allowed_congestion_control(tbl.data);
270 kfree(tbl.data);
271 return ret;
272 }
273
sscanf_key(char * buf,__le32 * key)274 static int sscanf_key(char *buf, __le32 *key)
275 {
276 u32 user_key[4];
277 int i, ret = 0;
278
279 if (sscanf(buf, "%x-%x-%x-%x", user_key, user_key + 1,
280 user_key + 2, user_key + 3) != 4) {
281 ret = -EINVAL;
282 } else {
283 for (i = 0; i < ARRAY_SIZE(user_key); i++)
284 key[i] = cpu_to_le32(user_key[i]);
285 }
286 pr_debug("proc TFO key set 0x%x-%x-%x-%x <- 0x%s: %u\n",
287 user_key[0], user_key[1], user_key[2], user_key[3], buf, ret);
288
289 return ret;
290 }
291
proc_tcp_fastopen_key(struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)292 static int proc_tcp_fastopen_key(struct ctl_table *table, int write,
293 void *buffer, size_t *lenp, loff_t *ppos)
294 {
295 struct net *net = container_of(table->data, struct net,
296 ipv4.sysctl_tcp_fastopen);
297 /* maxlen to print the list of keys in hex (*2), with dashes
298 * separating doublewords and a comma in between keys.
299 */
300 struct ctl_table tbl = { .maxlen = ((TCP_FASTOPEN_KEY_LENGTH *
301 2 * TCP_FASTOPEN_KEY_MAX) +
302 (TCP_FASTOPEN_KEY_MAX * 5)) };
303 u32 user_key[TCP_FASTOPEN_KEY_BUF_LENGTH / sizeof(u32)];
304 __le32 key[TCP_FASTOPEN_KEY_BUF_LENGTH / sizeof(__le32)];
305 char *backup_data;
306 int ret, i = 0, off = 0, n_keys;
307
308 tbl.data = kmalloc(tbl.maxlen, GFP_KERNEL);
309 if (!tbl.data)
310 return -ENOMEM;
311
312 n_keys = tcp_fastopen_get_cipher(net, NULL, (u64 *)key);
313 if (!n_keys) {
314 memset(&key[0], 0, TCP_FASTOPEN_KEY_LENGTH);
315 n_keys = 1;
316 }
317
318 for (i = 0; i < n_keys * 4; i++)
319 user_key[i] = le32_to_cpu(key[i]);
320
321 for (i = 0; i < n_keys; i++) {
322 off += snprintf(tbl.data + off, tbl.maxlen - off,
323 "%08x-%08x-%08x-%08x",
324 user_key[i * 4],
325 user_key[i * 4 + 1],
326 user_key[i * 4 + 2],
327 user_key[i * 4 + 3]);
328
329 if (WARN_ON_ONCE(off >= tbl.maxlen - 1))
330 break;
331
332 if (i + 1 < n_keys)
333 off += snprintf(tbl.data + off, tbl.maxlen - off, ",");
334 }
335
336 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
337
338 if (write && ret == 0) {
339 backup_data = strchr(tbl.data, ',');
340 if (backup_data) {
341 *backup_data = '\0';
342 backup_data++;
343 }
344 if (sscanf_key(tbl.data, key)) {
345 ret = -EINVAL;
346 goto bad_key;
347 }
348 if (backup_data) {
349 if (sscanf_key(backup_data, key + 4)) {
350 ret = -EINVAL;
351 goto bad_key;
352 }
353 }
354 tcp_fastopen_reset_cipher(net, NULL, key,
355 backup_data ? key + 4 : NULL);
356 }
357
358 bad_key:
359 kfree(tbl.data);
360 return ret;
361 }
362
proc_tfo_blackhole_detect_timeout(struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)363 static int proc_tfo_blackhole_detect_timeout(struct ctl_table *table,
364 int write, void *buffer,
365 size_t *lenp, loff_t *ppos)
366 {
367 struct net *net = container_of(table->data, struct net,
368 ipv4.sysctl_tcp_fastopen_blackhole_timeout);
369 int ret;
370
371 ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
372 if (write && ret == 0)
373 atomic_set(&net->ipv4.tfo_active_disable_times, 0);
374
375 return ret;
376 }
377
proc_tcp_available_ulp(struct ctl_table * ctl,int write,void * buffer,size_t * lenp,loff_t * ppos)378 static int proc_tcp_available_ulp(struct ctl_table *ctl,
379 int write, void *buffer, size_t *lenp,
380 loff_t *ppos)
381 {
382 struct ctl_table tbl = { .maxlen = TCP_ULP_BUF_MAX, };
383 int ret;
384
385 tbl.data = kmalloc(tbl.maxlen, GFP_USER);
386 if (!tbl.data)
387 return -ENOMEM;
388 tcp_get_available_ulp(tbl.data, TCP_ULP_BUF_MAX);
389 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
390 kfree(tbl.data);
391
392 return ret;
393 }
394
395 #ifdef CONFIG_IP_ROUTE_MULTIPATH
proc_fib_multipath_hash_policy(struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)396 static int proc_fib_multipath_hash_policy(struct ctl_table *table, int write,
397 void *buffer, size_t *lenp,
398 loff_t *ppos)
399 {
400 struct net *net = container_of(table->data, struct net,
401 ipv4.sysctl_fib_multipath_hash_policy);
402 int ret;
403
404 ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
405 if (write && ret == 0)
406 call_netevent_notifiers(NETEVENT_IPV4_MPATH_HASH_UPDATE, net);
407
408 return ret;
409 }
410 #endif
411
412 static struct ctl_table ipv4_table[] = {
413 {
414 .procname = "tcp_max_orphans",
415 .data = &sysctl_tcp_max_orphans,
416 .maxlen = sizeof(int),
417 .mode = 0644,
418 .proc_handler = proc_dointvec
419 },
420 {
421 .procname = "inet_peer_threshold",
422 .data = &inet_peer_threshold,
423 .maxlen = sizeof(int),
424 .mode = 0644,
425 .proc_handler = proc_dointvec
426 },
427 {
428 .procname = "inet_peer_minttl",
429 .data = &inet_peer_minttl,
430 .maxlen = sizeof(int),
431 .mode = 0644,
432 .proc_handler = proc_dointvec_jiffies,
433 },
434 {
435 .procname = "inet_peer_maxttl",
436 .data = &inet_peer_maxttl,
437 .maxlen = sizeof(int),
438 .mode = 0644,
439 .proc_handler = proc_dointvec_jiffies,
440 },
441 {
442 .procname = "tcp_mem",
443 .maxlen = sizeof(sysctl_tcp_mem),
444 .data = &sysctl_tcp_mem,
445 .mode = 0644,
446 .proc_handler = proc_doulongvec_minmax,
447 },
448 {
449 .procname = "tcp_low_latency",
450 .data = &sysctl_tcp_low_latency,
451 .maxlen = sizeof(int),
452 .mode = 0644,
453 .proc_handler = proc_dointvec
454 },
455 #ifdef CONFIG_NETLABEL
456 {
457 .procname = "cipso_cache_enable",
458 .data = &cipso_v4_cache_enabled,
459 .maxlen = sizeof(int),
460 .mode = 0644,
461 .proc_handler = proc_dointvec,
462 },
463 {
464 .procname = "cipso_cache_bucket_size",
465 .data = &cipso_v4_cache_bucketsize,
466 .maxlen = sizeof(int),
467 .mode = 0644,
468 .proc_handler = proc_dointvec,
469 },
470 {
471 .procname = "cipso_rbm_optfmt",
472 .data = &cipso_v4_rbm_optfmt,
473 .maxlen = sizeof(int),
474 .mode = 0644,
475 .proc_handler = proc_dointvec,
476 },
477 {
478 .procname = "cipso_rbm_strictvalid",
479 .data = &cipso_v4_rbm_strictvalid,
480 .maxlen = sizeof(int),
481 .mode = 0644,
482 .proc_handler = proc_dointvec,
483 },
484 #endif /* CONFIG_NETLABEL */
485 {
486 .procname = "tcp_available_ulp",
487 .maxlen = TCP_ULP_BUF_MAX,
488 .mode = 0444,
489 .proc_handler = proc_tcp_available_ulp,
490 },
491 {
492 .procname = "icmp_msgs_per_sec",
493 .data = &sysctl_icmp_msgs_per_sec,
494 .maxlen = sizeof(int),
495 .mode = 0644,
496 .proc_handler = proc_dointvec_minmax,
497 .extra1 = SYSCTL_ZERO,
498 },
499 {
500 .procname = "icmp_msgs_burst",
501 .data = &sysctl_icmp_msgs_burst,
502 .maxlen = sizeof(int),
503 .mode = 0644,
504 .proc_handler = proc_dointvec_minmax,
505 .extra1 = SYSCTL_ZERO,
506 },
507 {
508 .procname = "udp_mem",
509 .data = &sysctl_udp_mem,
510 .maxlen = sizeof(sysctl_udp_mem),
511 .mode = 0644,
512 .proc_handler = proc_doulongvec_minmax,
513 },
514 {
515 .procname = "fib_sync_mem",
516 .data = &sysctl_fib_sync_mem,
517 .maxlen = sizeof(sysctl_fib_sync_mem),
518 .mode = 0644,
519 .proc_handler = proc_douintvec_minmax,
520 .extra1 = &sysctl_fib_sync_mem_min,
521 .extra2 = &sysctl_fib_sync_mem_max,
522 },
523 {
524 .procname = "tcp_rx_skb_cache",
525 .data = &tcp_rx_skb_cache_key.key,
526 .mode = 0644,
527 .proc_handler = proc_do_static_key,
528 },
529 {
530 .procname = "tcp_tx_skb_cache",
531 .data = &tcp_tx_skb_cache_key.key,
532 .mode = 0644,
533 .proc_handler = proc_do_static_key,
534 },
535 { }
536 };
537
538 static struct ctl_table ipv4_net_table[] = {
539 {
540 .procname = "icmp_echo_ignore_all",
541 .data = &init_net.ipv4.sysctl_icmp_echo_ignore_all,
542 .maxlen = sizeof(int),
543 .mode = 0644,
544 .proc_handler = proc_dointvec
545 },
546 {
547 .procname = "icmp_echo_ignore_broadcasts",
548 .data = &init_net.ipv4.sysctl_icmp_echo_ignore_broadcasts,
549 .maxlen = sizeof(int),
550 .mode = 0644,
551 .proc_handler = proc_dointvec
552 },
553 {
554 .procname = "icmp_ignore_bogus_error_responses",
555 .data = &init_net.ipv4.sysctl_icmp_ignore_bogus_error_responses,
556 .maxlen = sizeof(int),
557 .mode = 0644,
558 .proc_handler = proc_dointvec
559 },
560 {
561 .procname = "icmp_errors_use_inbound_ifaddr",
562 .data = &init_net.ipv4.sysctl_icmp_errors_use_inbound_ifaddr,
563 .maxlen = sizeof(int),
564 .mode = 0644,
565 .proc_handler = proc_dointvec
566 },
567 {
568 .procname = "icmp_ratelimit",
569 .data = &init_net.ipv4.sysctl_icmp_ratelimit,
570 .maxlen = sizeof(int),
571 .mode = 0644,
572 .proc_handler = proc_dointvec_ms_jiffies,
573 },
574 {
575 .procname = "icmp_ratemask",
576 .data = &init_net.ipv4.sysctl_icmp_ratemask,
577 .maxlen = sizeof(int),
578 .mode = 0644,
579 .proc_handler = proc_dointvec
580 },
581 {
582 .procname = "ping_group_range",
583 .data = &init_net.ipv4.ping_group_range.range,
584 .maxlen = sizeof(gid_t)*2,
585 .mode = 0644,
586 .proc_handler = ipv4_ping_group_range,
587 },
588 #ifdef CONFIG_NET_L3_MASTER_DEV
589 {
590 .procname = "raw_l3mdev_accept",
591 .data = &init_net.ipv4.sysctl_raw_l3mdev_accept,
592 .maxlen = sizeof(int),
593 .mode = 0644,
594 .proc_handler = proc_dointvec_minmax,
595 .extra1 = SYSCTL_ZERO,
596 .extra2 = SYSCTL_ONE,
597 },
598 #endif
599 {
600 .procname = "tcp_ecn",
601 .data = &init_net.ipv4.sysctl_tcp_ecn,
602 .maxlen = sizeof(int),
603 .mode = 0644,
604 .proc_handler = proc_dointvec
605 },
606 {
607 .procname = "tcp_ecn_fallback",
608 .data = &init_net.ipv4.sysctl_tcp_ecn_fallback,
609 .maxlen = sizeof(int),
610 .mode = 0644,
611 .proc_handler = proc_dointvec
612 },
613 {
614 .procname = "ip_dynaddr",
615 .data = &init_net.ipv4.sysctl_ip_dynaddr,
616 .maxlen = sizeof(int),
617 .mode = 0644,
618 .proc_handler = proc_dointvec
619 },
620 {
621 .procname = "ip_early_demux",
622 .data = &init_net.ipv4.sysctl_ip_early_demux,
623 .maxlen = sizeof(int),
624 .mode = 0644,
625 .proc_handler = proc_dointvec
626 },
627 {
628 .procname = "udp_early_demux",
629 .data = &init_net.ipv4.sysctl_udp_early_demux,
630 .maxlen = sizeof(int),
631 .mode = 0644,
632 .proc_handler = proc_douintvec_minmax,
633 },
634 {
635 .procname = "tcp_early_demux",
636 .data = &init_net.ipv4.sysctl_tcp_early_demux,
637 .maxlen = sizeof(int),
638 .mode = 0644,
639 .proc_handler = proc_douintvec_minmax,
640 },
641 {
642 .procname = "nexthop_compat_mode",
643 .data = &init_net.ipv4.sysctl_nexthop_compat_mode,
644 .maxlen = sizeof(int),
645 .mode = 0644,
646 .proc_handler = proc_dointvec_minmax,
647 .extra1 = SYSCTL_ZERO,
648 .extra2 = SYSCTL_ONE,
649 },
650 {
651 .procname = "ip_default_ttl",
652 .data = &init_net.ipv4.sysctl_ip_default_ttl,
653 .maxlen = sizeof(int),
654 .mode = 0644,
655 .proc_handler = proc_dointvec_minmax,
656 .extra1 = &ip_ttl_min,
657 .extra2 = &ip_ttl_max,
658 },
659 {
660 .procname = "ip_local_port_range",
661 .maxlen = sizeof(init_net.ipv4.ip_local_ports.range),
662 .data = &init_net.ipv4.ip_local_ports.range,
663 .mode = 0644,
664 .proc_handler = ipv4_local_port_range,
665 },
666 {
667 .procname = "ip_local_reserved_ports",
668 .data = &init_net.ipv4.sysctl_local_reserved_ports,
669 .maxlen = 65536,
670 .mode = 0644,
671 .proc_handler = proc_do_large_bitmap,
672 },
673 {
674 .procname = "ip_local_unbindable_ports",
675 .data = &init_net.ipv4.sysctl_local_unbindable_ports,
676 .maxlen = 65536,
677 .mode = 0644,
678 .proc_handler = proc_do_large_bitmap,
679 },
680 {
681 .procname = "ip_no_pmtu_disc",
682 .data = &init_net.ipv4.sysctl_ip_no_pmtu_disc,
683 .maxlen = sizeof(int),
684 .mode = 0644,
685 .proc_handler = proc_dointvec
686 },
687 {
688 .procname = "ip_forward_use_pmtu",
689 .data = &init_net.ipv4.sysctl_ip_fwd_use_pmtu,
690 .maxlen = sizeof(int),
691 .mode = 0644,
692 .proc_handler = proc_dointvec,
693 },
694 {
695 .procname = "ip_forward_update_priority",
696 .data = &init_net.ipv4.sysctl_ip_fwd_update_priority,
697 .maxlen = sizeof(int),
698 .mode = 0644,
699 .proc_handler = ipv4_fwd_update_priority,
700 .extra1 = SYSCTL_ZERO,
701 .extra2 = SYSCTL_ONE,
702 },
703 {
704 .procname = "ip_nonlocal_bind",
705 .data = &init_net.ipv4.sysctl_ip_nonlocal_bind,
706 .maxlen = sizeof(int),
707 .mode = 0644,
708 .proc_handler = proc_dointvec
709 },
710 {
711 .procname = "ip_autobind_reuse",
712 .data = &init_net.ipv4.sysctl_ip_autobind_reuse,
713 .maxlen = sizeof(int),
714 .mode = 0644,
715 .proc_handler = proc_dointvec_minmax,
716 .extra1 = SYSCTL_ZERO,
717 .extra2 = SYSCTL_ONE,
718 },
719 {
720 .procname = "fwmark_reflect",
721 .data = &init_net.ipv4.sysctl_fwmark_reflect,
722 .maxlen = sizeof(int),
723 .mode = 0644,
724 .proc_handler = proc_dointvec,
725 },
726 {
727 .procname = "tcp_fwmark_accept",
728 .data = &init_net.ipv4.sysctl_tcp_fwmark_accept,
729 .maxlen = sizeof(int),
730 .mode = 0644,
731 .proc_handler = proc_dointvec,
732 },
733 #ifdef CONFIG_NET_L3_MASTER_DEV
734 {
735 .procname = "tcp_l3mdev_accept",
736 .data = &init_net.ipv4.sysctl_tcp_l3mdev_accept,
737 .maxlen = sizeof(int),
738 .mode = 0644,
739 .proc_handler = proc_dointvec_minmax,
740 .extra1 = SYSCTL_ZERO,
741 .extra2 = SYSCTL_ONE,
742 },
743 #endif
744 {
745 .procname = "tcp_mtu_probing",
746 .data = &init_net.ipv4.sysctl_tcp_mtu_probing,
747 .maxlen = sizeof(int),
748 .mode = 0644,
749 .proc_handler = proc_dointvec,
750 },
751 {
752 .procname = "tcp_base_mss",
753 .data = &init_net.ipv4.sysctl_tcp_base_mss,
754 .maxlen = sizeof(int),
755 .mode = 0644,
756 .proc_handler = proc_dointvec,
757 },
758 {
759 .procname = "tcp_min_snd_mss",
760 .data = &init_net.ipv4.sysctl_tcp_min_snd_mss,
761 .maxlen = sizeof(int),
762 .mode = 0644,
763 .proc_handler = proc_dointvec_minmax,
764 .extra1 = &tcp_min_snd_mss_min,
765 .extra2 = &tcp_min_snd_mss_max,
766 },
767 {
768 .procname = "tcp_mtu_probe_floor",
769 .data = &init_net.ipv4.sysctl_tcp_mtu_probe_floor,
770 .maxlen = sizeof(int),
771 .mode = 0644,
772 .proc_handler = proc_dointvec_minmax,
773 .extra1 = &tcp_min_snd_mss_min,
774 .extra2 = &tcp_min_snd_mss_max,
775 },
776 {
777 .procname = "tcp_probe_threshold",
778 .data = &init_net.ipv4.sysctl_tcp_probe_threshold,
779 .maxlen = sizeof(int),
780 .mode = 0644,
781 .proc_handler = proc_dointvec,
782 },
783 {
784 .procname = "tcp_probe_interval",
785 .data = &init_net.ipv4.sysctl_tcp_probe_interval,
786 .maxlen = sizeof(u32),
787 .mode = 0644,
788 .proc_handler = proc_douintvec_minmax,
789 .extra2 = &u32_max_div_HZ,
790 },
791 {
792 .procname = "igmp_link_local_mcast_reports",
793 .data = &init_net.ipv4.sysctl_igmp_llm_reports,
794 .maxlen = sizeof(int),
795 .mode = 0644,
796 .proc_handler = proc_dointvec
797 },
798 {
799 .procname = "igmp_max_memberships",
800 .data = &init_net.ipv4.sysctl_igmp_max_memberships,
801 .maxlen = sizeof(int),
802 .mode = 0644,
803 .proc_handler = proc_dointvec
804 },
805 {
806 .procname = "igmp_max_msf",
807 .data = &init_net.ipv4.sysctl_igmp_max_msf,
808 .maxlen = sizeof(int),
809 .mode = 0644,
810 .proc_handler = proc_dointvec
811 },
812 #ifdef CONFIG_IP_MULTICAST
813 {
814 .procname = "igmp_qrv",
815 .data = &init_net.ipv4.sysctl_igmp_qrv,
816 .maxlen = sizeof(int),
817 .mode = 0644,
818 .proc_handler = proc_dointvec_minmax,
819 .extra1 = SYSCTL_ONE
820 },
821 #endif
822 {
823 .procname = "tcp_congestion_control",
824 .data = &init_net.ipv4.tcp_congestion_control,
825 .mode = 0644,
826 .maxlen = TCP_CA_NAME_MAX,
827 .proc_handler = proc_tcp_congestion_control,
828 },
829 {
830 .procname = "tcp_available_congestion_control",
831 .maxlen = TCP_CA_BUF_MAX,
832 .mode = 0444,
833 .proc_handler = proc_tcp_available_congestion_control,
834 },
835 {
836 .procname = "tcp_allowed_congestion_control",
837 .maxlen = TCP_CA_BUF_MAX,
838 .mode = 0644,
839 .proc_handler = proc_allowed_congestion_control,
840 },
841 {
842 .procname = "tcp_keepalive_time",
843 .data = &init_net.ipv4.sysctl_tcp_keepalive_time,
844 .maxlen = sizeof(int),
845 .mode = 0644,
846 .proc_handler = proc_dointvec_jiffies,
847 },
848 {
849 .procname = "tcp_keepalive_probes",
850 .data = &init_net.ipv4.sysctl_tcp_keepalive_probes,
851 .maxlen = sizeof(int),
852 .mode = 0644,
853 .proc_handler = proc_dointvec
854 },
855 {
856 .procname = "tcp_keepalive_intvl",
857 .data = &init_net.ipv4.sysctl_tcp_keepalive_intvl,
858 .maxlen = sizeof(int),
859 .mode = 0644,
860 .proc_handler = proc_dointvec_jiffies,
861 },
862 {
863 .procname = "tcp_syn_retries",
864 .data = &init_net.ipv4.sysctl_tcp_syn_retries,
865 .maxlen = sizeof(int),
866 .mode = 0644,
867 .proc_handler = proc_dointvec_minmax,
868 .extra1 = &tcp_syn_retries_min,
869 .extra2 = &tcp_syn_retries_max
870 },
871 {
872 .procname = "tcp_synack_retries",
873 .data = &init_net.ipv4.sysctl_tcp_synack_retries,
874 .maxlen = sizeof(int),
875 .mode = 0644,
876 .proc_handler = proc_dointvec
877 },
878 #ifdef CONFIG_SYN_COOKIES
879 {
880 .procname = "tcp_syncookies",
881 .data = &init_net.ipv4.sysctl_tcp_syncookies,
882 .maxlen = sizeof(int),
883 .mode = 0644,
884 .proc_handler = proc_dointvec
885 },
886 #endif
887 {
888 .procname = "tcp_reordering",
889 .data = &init_net.ipv4.sysctl_tcp_reordering,
890 .maxlen = sizeof(int),
891 .mode = 0644,
892 .proc_handler = proc_dointvec
893 },
894 {
895 .procname = "tcp_retries1",
896 .data = &init_net.ipv4.sysctl_tcp_retries1,
897 .maxlen = sizeof(int),
898 .mode = 0644,
899 .proc_handler = proc_dointvec_minmax,
900 .extra2 = &tcp_retr1_max
901 },
902 {
903 .procname = "tcp_retries2",
904 .data = &init_net.ipv4.sysctl_tcp_retries2,
905 .maxlen = sizeof(int),
906 .mode = 0644,
907 .proc_handler = proc_dointvec
908 },
909 {
910 .procname = "tcp_orphan_retries",
911 .data = &init_net.ipv4.sysctl_tcp_orphan_retries,
912 .maxlen = sizeof(int),
913 .mode = 0644,
914 .proc_handler = proc_dointvec
915 },
916 {
917 .procname = "tcp_fin_timeout",
918 .data = &init_net.ipv4.sysctl_tcp_fin_timeout,
919 .maxlen = sizeof(int),
920 .mode = 0644,
921 .proc_handler = proc_dointvec_jiffies,
922 },
923 {
924 .procname = "tcp_notsent_lowat",
925 .data = &init_net.ipv4.sysctl_tcp_notsent_lowat,
926 .maxlen = sizeof(unsigned int),
927 .mode = 0644,
928 .proc_handler = proc_douintvec,
929 },
930 {
931 .procname = "tcp_tw_reuse",
932 .data = &init_net.ipv4.sysctl_tcp_tw_reuse,
933 .maxlen = sizeof(int),
934 .mode = 0644,
935 .proc_handler = proc_dointvec_minmax,
936 .extra1 = SYSCTL_ZERO,
937 .extra2 = &two,
938 },
939 {
940 .procname = "tcp_max_tw_buckets",
941 .data = &init_net.ipv4.tcp_death_row.sysctl_max_tw_buckets,
942 .maxlen = sizeof(int),
943 .mode = 0644,
944 .proc_handler = proc_dointvec
945 },
946 {
947 .procname = "tcp_max_syn_backlog",
948 .data = &init_net.ipv4.sysctl_max_syn_backlog,
949 .maxlen = sizeof(int),
950 .mode = 0644,
951 .proc_handler = proc_dointvec
952 },
953 {
954 .procname = "tcp_fastopen",
955 .data = &init_net.ipv4.sysctl_tcp_fastopen,
956 .maxlen = sizeof(int),
957 .mode = 0644,
958 .proc_handler = proc_dointvec,
959 },
960 {
961 .procname = "tcp_fastopen_key",
962 .mode = 0600,
963 .data = &init_net.ipv4.sysctl_tcp_fastopen,
964 /* maxlen to print the list of keys in hex (*2), with dashes
965 * separating doublewords and a comma in between keys.
966 */
967 .maxlen = ((TCP_FASTOPEN_KEY_LENGTH *
968 2 * TCP_FASTOPEN_KEY_MAX) +
969 (TCP_FASTOPEN_KEY_MAX * 5)),
970 .proc_handler = proc_tcp_fastopen_key,
971 },
972 {
973 .procname = "tcp_fastopen_blackhole_timeout_sec",
974 .data = &init_net.ipv4.sysctl_tcp_fastopen_blackhole_timeout,
975 .maxlen = sizeof(int),
976 .mode = 0644,
977 .proc_handler = proc_tfo_blackhole_detect_timeout,
978 .extra1 = SYSCTL_ZERO,
979 },
980 #ifdef CONFIG_IP_ROUTE_MULTIPATH
981 {
982 .procname = "fib_multipath_use_neigh",
983 .data = &init_net.ipv4.sysctl_fib_multipath_use_neigh,
984 .maxlen = sizeof(int),
985 .mode = 0644,
986 .proc_handler = proc_dointvec_minmax,
987 .extra1 = SYSCTL_ZERO,
988 .extra2 = SYSCTL_ONE,
989 },
990 {
991 .procname = "fib_multipath_hash_policy",
992 .data = &init_net.ipv4.sysctl_fib_multipath_hash_policy,
993 .maxlen = sizeof(int),
994 .mode = 0644,
995 .proc_handler = proc_fib_multipath_hash_policy,
996 .extra1 = SYSCTL_ZERO,
997 .extra2 = &two,
998 },
999 #endif
1000 {
1001 .procname = "ip_unprivileged_port_start",
1002 .maxlen = sizeof(int),
1003 .data = &init_net.ipv4.sysctl_ip_prot_sock,
1004 .mode = 0644,
1005 .proc_handler = ipv4_privileged_ports,
1006 },
1007 #ifdef CONFIG_NET_L3_MASTER_DEV
1008 {
1009 .procname = "udp_l3mdev_accept",
1010 .data = &init_net.ipv4.sysctl_udp_l3mdev_accept,
1011 .maxlen = sizeof(int),
1012 .mode = 0644,
1013 .proc_handler = proc_dointvec_minmax,
1014 .extra1 = SYSCTL_ZERO,
1015 .extra2 = SYSCTL_ONE,
1016 },
1017 #endif
1018 {
1019 .procname = "tcp_sack",
1020 .data = &init_net.ipv4.sysctl_tcp_sack,
1021 .maxlen = sizeof(int),
1022 .mode = 0644,
1023 .proc_handler = proc_dointvec
1024 },
1025 {
1026 .procname = "tcp_window_scaling",
1027 .data = &init_net.ipv4.sysctl_tcp_window_scaling,
1028 .maxlen = sizeof(int),
1029 .mode = 0644,
1030 .proc_handler = proc_dointvec
1031 },
1032 {
1033 .procname = "tcp_timestamps",
1034 .data = &init_net.ipv4.sysctl_tcp_timestamps,
1035 .maxlen = sizeof(int),
1036 .mode = 0644,
1037 .proc_handler = proc_dointvec
1038 },
1039 {
1040 .procname = "tcp_early_retrans",
1041 .data = &init_net.ipv4.sysctl_tcp_early_retrans,
1042 .maxlen = sizeof(int),
1043 .mode = 0644,
1044 .proc_handler = proc_dointvec_minmax,
1045 .extra1 = SYSCTL_ZERO,
1046 .extra2 = &four,
1047 },
1048 {
1049 .procname = "tcp_recovery",
1050 .data = &init_net.ipv4.sysctl_tcp_recovery,
1051 .maxlen = sizeof(int),
1052 .mode = 0644,
1053 .proc_handler = proc_dointvec,
1054 },
1055 {
1056 .procname = "tcp_thin_linear_timeouts",
1057 .data = &init_net.ipv4.sysctl_tcp_thin_linear_timeouts,
1058 .maxlen = sizeof(int),
1059 .mode = 0644,
1060 .proc_handler = proc_dointvec
1061 },
1062 {
1063 .procname = "tcp_slow_start_after_idle",
1064 .data = &init_net.ipv4.sysctl_tcp_slow_start_after_idle,
1065 .maxlen = sizeof(int),
1066 .mode = 0644,
1067 .proc_handler = proc_dointvec
1068 },
1069 {
1070 .procname = "tcp_retrans_collapse",
1071 .data = &init_net.ipv4.sysctl_tcp_retrans_collapse,
1072 .maxlen = sizeof(int),
1073 .mode = 0644,
1074 .proc_handler = proc_dointvec
1075 },
1076 {
1077 .procname = "tcp_stdurg",
1078 .data = &init_net.ipv4.sysctl_tcp_stdurg,
1079 .maxlen = sizeof(int),
1080 .mode = 0644,
1081 .proc_handler = proc_dointvec
1082 },
1083 {
1084 .procname = "tcp_rfc1337",
1085 .data = &init_net.ipv4.sysctl_tcp_rfc1337,
1086 .maxlen = sizeof(int),
1087 .mode = 0644,
1088 .proc_handler = proc_dointvec
1089 },
1090 {
1091 .procname = "tcp_abort_on_overflow",
1092 .data = &init_net.ipv4.sysctl_tcp_abort_on_overflow,
1093 .maxlen = sizeof(int),
1094 .mode = 0644,
1095 .proc_handler = proc_dointvec
1096 },
1097 {
1098 .procname = "tcp_fack",
1099 .data = &init_net.ipv4.sysctl_tcp_fack,
1100 .maxlen = sizeof(int),
1101 .mode = 0644,
1102 .proc_handler = proc_dointvec
1103 },
1104 {
1105 .procname = "tcp_max_reordering",
1106 .data = &init_net.ipv4.sysctl_tcp_max_reordering,
1107 .maxlen = sizeof(int),
1108 .mode = 0644,
1109 .proc_handler = proc_dointvec
1110 },
1111 {
1112 .procname = "tcp_dsack",
1113 .data = &init_net.ipv4.sysctl_tcp_dsack,
1114 .maxlen = sizeof(int),
1115 .mode = 0644,
1116 .proc_handler = proc_dointvec
1117 },
1118 {
1119 .procname = "tcp_app_win",
1120 .data = &init_net.ipv4.sysctl_tcp_app_win,
1121 .maxlen = sizeof(int),
1122 .mode = 0644,
1123 .proc_handler = proc_dointvec
1124 },
1125 {
1126 .procname = "tcp_adv_win_scale",
1127 .data = &init_net.ipv4.sysctl_tcp_adv_win_scale,
1128 .maxlen = sizeof(int),
1129 .mode = 0644,
1130 .proc_handler = proc_dointvec_minmax,
1131 .extra1 = &tcp_adv_win_scale_min,
1132 .extra2 = &tcp_adv_win_scale_max,
1133 },
1134 {
1135 .procname = "tcp_frto",
1136 .data = &init_net.ipv4.sysctl_tcp_frto,
1137 .maxlen = sizeof(int),
1138 .mode = 0644,
1139 .proc_handler = proc_dointvec
1140 },
1141 {
1142 .procname = "tcp_no_metrics_save",
1143 .data = &init_net.ipv4.sysctl_tcp_nometrics_save,
1144 .maxlen = sizeof(int),
1145 .mode = 0644,
1146 .proc_handler = proc_dointvec,
1147 },
1148 {
1149 .procname = "tcp_no_ssthresh_metrics_save",
1150 .data = &init_net.ipv4.sysctl_tcp_no_ssthresh_metrics_save,
1151 .maxlen = sizeof(int),
1152 .mode = 0644,
1153 .proc_handler = proc_dointvec_minmax,
1154 .extra1 = SYSCTL_ZERO,
1155 .extra2 = SYSCTL_ONE,
1156 },
1157 {
1158 .procname = "tcp_moderate_rcvbuf",
1159 .data = &init_net.ipv4.sysctl_tcp_moderate_rcvbuf,
1160 .maxlen = sizeof(int),
1161 .mode = 0644,
1162 .proc_handler = proc_dointvec,
1163 },
1164 {
1165 .procname = "tcp_tso_win_divisor",
1166 .data = &init_net.ipv4.sysctl_tcp_tso_win_divisor,
1167 .maxlen = sizeof(int),
1168 .mode = 0644,
1169 .proc_handler = proc_dointvec,
1170 },
1171 {
1172 .procname = "tcp_workaround_signed_windows",
1173 .data = &init_net.ipv4.sysctl_tcp_workaround_signed_windows,
1174 .maxlen = sizeof(int),
1175 .mode = 0644,
1176 .proc_handler = proc_dointvec
1177 },
1178 {
1179 .procname = "tcp_limit_output_bytes",
1180 .data = &init_net.ipv4.sysctl_tcp_limit_output_bytes,
1181 .maxlen = sizeof(int),
1182 .mode = 0644,
1183 .proc_handler = proc_dointvec
1184 },
1185 {
1186 .procname = "tcp_challenge_ack_limit",
1187 .data = &init_net.ipv4.sysctl_tcp_challenge_ack_limit,
1188 .maxlen = sizeof(int),
1189 .mode = 0644,
1190 .proc_handler = proc_dointvec
1191 },
1192 {
1193 .procname = "tcp_min_tso_segs",
1194 .data = &init_net.ipv4.sysctl_tcp_min_tso_segs,
1195 .maxlen = sizeof(int),
1196 .mode = 0644,
1197 .proc_handler = proc_dointvec_minmax,
1198 .extra1 = SYSCTL_ONE,
1199 },
1200 {
1201 .procname = "tcp_min_rtt_wlen",
1202 .data = &init_net.ipv4.sysctl_tcp_min_rtt_wlen,
1203 .maxlen = sizeof(int),
1204 .mode = 0644,
1205 .proc_handler = proc_dointvec_minmax,
1206 .extra1 = SYSCTL_ZERO,
1207 .extra2 = &one_day_secs
1208 },
1209 {
1210 .procname = "tcp_autocorking",
1211 .data = &init_net.ipv4.sysctl_tcp_autocorking,
1212 .maxlen = sizeof(int),
1213 .mode = 0644,
1214 .proc_handler = proc_dointvec_minmax,
1215 .extra1 = SYSCTL_ZERO,
1216 .extra2 = SYSCTL_ONE,
1217 },
1218 {
1219 .procname = "tcp_invalid_ratelimit",
1220 .data = &init_net.ipv4.sysctl_tcp_invalid_ratelimit,
1221 .maxlen = sizeof(int),
1222 .mode = 0644,
1223 .proc_handler = proc_dointvec_ms_jiffies,
1224 },
1225 {
1226 .procname = "tcp_pacing_ss_ratio",
1227 .data = &init_net.ipv4.sysctl_tcp_pacing_ss_ratio,
1228 .maxlen = sizeof(int),
1229 .mode = 0644,
1230 .proc_handler = proc_dointvec_minmax,
1231 .extra1 = SYSCTL_ZERO,
1232 .extra2 = &thousand,
1233 },
1234 {
1235 .procname = "tcp_pacing_ca_ratio",
1236 .data = &init_net.ipv4.sysctl_tcp_pacing_ca_ratio,
1237 .maxlen = sizeof(int),
1238 .mode = 0644,
1239 .proc_handler = proc_dointvec_minmax,
1240 .extra1 = SYSCTL_ZERO,
1241 .extra2 = &thousand,
1242 },
1243 {
1244 .procname = "tcp_wmem",
1245 .data = &init_net.ipv4.sysctl_tcp_wmem,
1246 .maxlen = sizeof(init_net.ipv4.sysctl_tcp_wmem),
1247 .mode = 0644,
1248 .proc_handler = proc_dointvec_minmax,
1249 .extra1 = SYSCTL_ONE,
1250 },
1251 {
1252 .procname = "tcp_rmem",
1253 .data = &init_net.ipv4.sysctl_tcp_rmem,
1254 .maxlen = sizeof(init_net.ipv4.sysctl_tcp_rmem),
1255 .mode = 0644,
1256 .proc_handler = proc_dointvec_minmax,
1257 .extra1 = SYSCTL_ONE,
1258 },
1259 {
1260 .procname = "tcp_comp_sack_delay_ns",
1261 .data = &init_net.ipv4.sysctl_tcp_comp_sack_delay_ns,
1262 .maxlen = sizeof(unsigned long),
1263 .mode = 0644,
1264 .proc_handler = proc_doulongvec_minmax,
1265 },
1266 {
1267 .procname = "tcp_comp_sack_slack_ns",
1268 .data = &init_net.ipv4.sysctl_tcp_comp_sack_slack_ns,
1269 .maxlen = sizeof(unsigned long),
1270 .mode = 0644,
1271 .proc_handler = proc_doulongvec_minmax,
1272 },
1273 {
1274 .procname = "tcp_comp_sack_nr",
1275 .data = &init_net.ipv4.sysctl_tcp_comp_sack_nr,
1276 .maxlen = sizeof(int),
1277 .mode = 0644,
1278 .proc_handler = proc_dointvec_minmax,
1279 .extra1 = SYSCTL_ZERO,
1280 .extra2 = &comp_sack_nr_max,
1281 },
1282 {
1283 .procname = "tcp_reflect_tos",
1284 .data = &init_net.ipv4.sysctl_tcp_reflect_tos,
1285 .maxlen = sizeof(int),
1286 .mode = 0644,
1287 .proc_handler = proc_dointvec_minmax,
1288 .extra1 = SYSCTL_ZERO,
1289 .extra2 = SYSCTL_ONE,
1290 },
1291 {
1292 .procname = "udp_rmem_min",
1293 .data = &init_net.ipv4.sysctl_udp_rmem_min,
1294 .maxlen = sizeof(init_net.ipv4.sysctl_udp_rmem_min),
1295 .mode = 0644,
1296 .proc_handler = proc_dointvec_minmax,
1297 .extra1 = SYSCTL_ONE
1298 },
1299 {
1300 .procname = "udp_wmem_min",
1301 .data = &init_net.ipv4.sysctl_udp_wmem_min,
1302 .maxlen = sizeof(init_net.ipv4.sysctl_udp_wmem_min),
1303 .mode = 0644,
1304 .proc_handler = proc_dointvec_minmax,
1305 .extra1 = SYSCTL_ONE
1306 },
1307 { }
1308 };
1309
ipv4_sysctl_init_net(struct net * net)1310 static __net_init int ipv4_sysctl_init_net(struct net *net)
1311 {
1312 struct ctl_table *table;
1313
1314 table = ipv4_net_table;
1315 if (!net_eq(net, &init_net)) {
1316 int i;
1317
1318 table = kmemdup(table, sizeof(ipv4_net_table), GFP_KERNEL);
1319 if (!table)
1320 goto err_alloc;
1321
1322 for (i = 0; i < ARRAY_SIZE(ipv4_net_table) - 1; i++) {
1323 if (table[i].data) {
1324 /* Update the variables to point into
1325 * the current struct net
1326 */
1327 table[i].data += (void *)net - (void *)&init_net;
1328 } else {
1329 /* Entries without data pointer are global;
1330 * Make them read-only in non-init_net ns
1331 */
1332 table[i].mode &= ~0222;
1333 }
1334 }
1335 }
1336
1337 net->ipv4.ipv4_hdr = register_net_sysctl(net, "net/ipv4", table);
1338 if (!net->ipv4.ipv4_hdr)
1339 goto err_reg;
1340
1341 net->ipv4.sysctl_local_reserved_ports = kzalloc(65536 / 8, GFP_KERNEL);
1342 if (!net->ipv4.sysctl_local_reserved_ports)
1343 goto err_reserved_ports;
1344
1345 net->ipv4.sysctl_local_unbindable_ports = kzalloc(65536 / 8, GFP_KERNEL);
1346 if (!net->ipv4.sysctl_local_unbindable_ports)
1347 goto err_unbindable_ports;
1348
1349 return 0;
1350
1351 err_unbindable_ports:
1352 kfree(net->ipv4.sysctl_local_reserved_ports);
1353 err_reserved_ports:
1354 unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
1355 err_reg:
1356 if (!net_eq(net, &init_net))
1357 kfree(table);
1358 err_alloc:
1359 return -ENOMEM;
1360 }
1361
ipv4_sysctl_exit_net(struct net * net)1362 static __net_exit void ipv4_sysctl_exit_net(struct net *net)
1363 {
1364 struct ctl_table *table;
1365
1366 kfree(net->ipv4.sysctl_local_unbindable_ports);
1367 kfree(net->ipv4.sysctl_local_reserved_ports);
1368 table = net->ipv4.ipv4_hdr->ctl_table_arg;
1369 unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
1370 kfree(table);
1371 }
1372
1373 static __net_initdata struct pernet_operations ipv4_sysctl_ops = {
1374 .init = ipv4_sysctl_init_net,
1375 .exit = ipv4_sysctl_exit_net,
1376 };
1377
sysctl_ipv4_init(void)1378 static __init int sysctl_ipv4_init(void)
1379 {
1380 struct ctl_table_header *hdr;
1381
1382 hdr = register_net_sysctl(&init_net, "net/ipv4", ipv4_table);
1383 if (!hdr)
1384 return -ENOMEM;
1385
1386 if (register_pernet_subsys(&ipv4_sysctl_ops)) {
1387 unregister_net_sysctl_table(hdr);
1388 return -ENOMEM;
1389 }
1390
1391 return 0;
1392 }
1393
1394 __initcall(sysctl_ipv4_init);
1395