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 gso_max_segs = GSO_MAX_SEGS;
35 static int tcp_retr1_max = 255;
36 static int ip_local_port_range_min[] = { 1, 1 };
37 static int ip_local_port_range_max[] = { 65535, 65535 };
38 static int tcp_adv_win_scale_min = -31;
39 static int tcp_adv_win_scale_max = 31;
40 static int tcp_min_snd_mss_min = TCP_MIN_SND_MSS;
41 static int tcp_min_snd_mss_max = 65535;
42 static int ip_privileged_port_min;
43 static int ip_privileged_port_max = 65535;
44 static int ip_ttl_min = 1;
45 static int ip_ttl_max = 255;
46 static int tcp_syn_retries_min = 1;
47 static int tcp_syn_retries_max = MAX_TCP_SYNCNT;
48 static int ip_ping_group_range_min[] = { 0, 0 };
49 static int ip_ping_group_range_max[] = { GID_T_MAX, GID_T_MAX };
50 static int comp_sack_nr_max = 255;
51 static u32 u32_max_div_HZ = UINT_MAX / HZ;
52 static int one_day_secs = 24 * 3600;
53
54 /* obsolete */
55 static int sysctl_tcp_low_latency __read_mostly;
56
57 /* Update system visible IP port range */
set_local_port_range(struct net * net,int range[2])58 static void set_local_port_range(struct net *net, int range[2])
59 {
60 bool same_parity = !((range[0] ^ range[1]) & 1);
61
62 write_seqlock_bh(&net->ipv4.ip_local_ports.lock);
63 if (same_parity && !net->ipv4.ip_local_ports.warned) {
64 net->ipv4.ip_local_ports.warned = true;
65 pr_err_ratelimited("ip_local_port_range: prefer different parity for start/end values.\n");
66 }
67 net->ipv4.ip_local_ports.range[0] = range[0];
68 net->ipv4.ip_local_ports.range[1] = range[1];
69 write_sequnlock_bh(&net->ipv4.ip_local_ports.lock);
70 }
71
72 /* Validate changes from /proc interface. */
ipv4_local_port_range(struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)73 static int ipv4_local_port_range(struct ctl_table *table, int write,
74 void *buffer, size_t *lenp, loff_t *ppos)
75 {
76 struct net *net =
77 container_of(table->data, struct net, ipv4.ip_local_ports.range);
78 int ret;
79 int range[2];
80 struct ctl_table tmp = {
81 .data = &range,
82 .maxlen = sizeof(range),
83 .mode = table->mode,
84 .extra1 = &ip_local_port_range_min,
85 .extra2 = &ip_local_port_range_max,
86 };
87
88 inet_get_local_port_range(net, &range[0], &range[1]);
89
90 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
91
92 if (write && ret == 0) {
93 /* Ensure that the upper limit is not smaller than the lower,
94 * and that the lower does not encroach upon the privileged
95 * port limit.
96 */
97 if ((range[1] < range[0]) ||
98 (range[0] < READ_ONCE(net->ipv4.sysctl_ip_prot_sock)))
99 ret = -EINVAL;
100 else
101 set_local_port_range(net, range);
102 }
103
104 return ret;
105 }
106
107 /* Validate changes from /proc interface. */
ipv4_privileged_ports(struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)108 static int ipv4_privileged_ports(struct ctl_table *table, int write,
109 void *buffer, size_t *lenp, loff_t *ppos)
110 {
111 struct net *net = container_of(table->data, struct net,
112 ipv4.sysctl_ip_prot_sock);
113 int ret;
114 int pports;
115 int range[2];
116 struct ctl_table tmp = {
117 .data = &pports,
118 .maxlen = sizeof(pports),
119 .mode = table->mode,
120 .extra1 = &ip_privileged_port_min,
121 .extra2 = &ip_privileged_port_max,
122 };
123
124 pports = READ_ONCE(net->ipv4.sysctl_ip_prot_sock);
125
126 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
127
128 if (write && ret == 0) {
129 inet_get_local_port_range(net, &range[0], &range[1]);
130 /* Ensure that the local port range doesn't overlap with the
131 * privileged port range.
132 */
133 if (range[0] < pports)
134 ret = -EINVAL;
135 else
136 WRITE_ONCE(net->ipv4.sysctl_ip_prot_sock, pports);
137 }
138
139 return ret;
140 }
141
inet_get_ping_group_range_table(struct ctl_table * table,kgid_t * low,kgid_t * high)142 static void inet_get_ping_group_range_table(struct ctl_table *table, kgid_t *low, kgid_t *high)
143 {
144 kgid_t *data = table->data;
145 struct net *net =
146 container_of(table->data, struct net, ipv4.ping_group_range.range);
147 unsigned int seq;
148 do {
149 seq = read_seqbegin(&net->ipv4.ping_group_range.lock);
150
151 *low = data[0];
152 *high = data[1];
153 } while (read_seqretry(&net->ipv4.ping_group_range.lock, seq));
154 }
155
156 /* Update system visible IP port range */
set_ping_group_range(struct ctl_table * table,kgid_t low,kgid_t high)157 static void set_ping_group_range(struct ctl_table *table, kgid_t low, kgid_t high)
158 {
159 kgid_t *data = table->data;
160 struct net *net =
161 container_of(table->data, struct net, ipv4.ping_group_range.range);
162 write_seqlock(&net->ipv4.ping_group_range.lock);
163 data[0] = low;
164 data[1] = high;
165 write_sequnlock(&net->ipv4.ping_group_range.lock);
166 }
167
168 /* Validate changes from /proc interface. */
ipv4_ping_group_range(struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)169 static int ipv4_ping_group_range(struct ctl_table *table, int write,
170 void *buffer, size_t *lenp, loff_t *ppos)
171 {
172 struct user_namespace *user_ns = current_user_ns();
173 int ret;
174 gid_t urange[2];
175 kgid_t low, high;
176 struct ctl_table tmp = {
177 .data = &urange,
178 .maxlen = sizeof(urange),
179 .mode = table->mode,
180 .extra1 = &ip_ping_group_range_min,
181 .extra2 = &ip_ping_group_range_max,
182 };
183
184 inet_get_ping_group_range_table(table, &low, &high);
185 urange[0] = from_kgid_munged(user_ns, low);
186 urange[1] = from_kgid_munged(user_ns, high);
187 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
188
189 if (write && ret == 0) {
190 low = make_kgid(user_ns, urange[0]);
191 high = make_kgid(user_ns, urange[1]);
192 if (!gid_valid(low) || !gid_valid(high))
193 return -EINVAL;
194 if (urange[1] < urange[0] || gid_lt(high, low)) {
195 low = make_kgid(&init_user_ns, 1);
196 high = make_kgid(&init_user_ns, 0);
197 }
198 set_ping_group_range(table, low, high);
199 }
200
201 return ret;
202 }
203
ipv4_fwd_update_priority(struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)204 static int ipv4_fwd_update_priority(struct ctl_table *table, int write,
205 void *buffer, size_t *lenp, loff_t *ppos)
206 {
207 struct net *net;
208 int ret;
209
210 net = container_of(table->data, struct net,
211 ipv4.sysctl_ip_fwd_update_priority);
212 ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
213 if (write && ret == 0)
214 call_netevent_notifiers(NETEVENT_IPV4_FWD_UPDATE_PRIORITY_UPDATE,
215 net);
216
217 return ret;
218 }
219
proc_tcp_congestion_control(struct ctl_table * ctl,int write,void * buffer,size_t * lenp,loff_t * ppos)220 static int proc_tcp_congestion_control(struct ctl_table *ctl, int write,
221 void *buffer, size_t *lenp, loff_t *ppos)
222 {
223 struct net *net = container_of(ctl->data, struct net,
224 ipv4.tcp_congestion_control);
225 char val[TCP_CA_NAME_MAX];
226 struct ctl_table tbl = {
227 .data = val,
228 .maxlen = TCP_CA_NAME_MAX,
229 };
230 int ret;
231
232 tcp_get_default_congestion_control(net, val);
233
234 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
235 if (write && ret == 0)
236 ret = tcp_set_default_congestion_control(net, val);
237 return ret;
238 }
239
proc_tcp_available_congestion_control(struct ctl_table * ctl,int write,void * buffer,size_t * lenp,loff_t * ppos)240 static int proc_tcp_available_congestion_control(struct ctl_table *ctl,
241 int write, void *buffer,
242 size_t *lenp, loff_t *ppos)
243 {
244 struct ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX, };
245 int ret;
246
247 tbl.data = kmalloc(tbl.maxlen, GFP_USER);
248 if (!tbl.data)
249 return -ENOMEM;
250 tcp_get_available_congestion_control(tbl.data, TCP_CA_BUF_MAX);
251 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
252 kfree(tbl.data);
253 return ret;
254 }
255
proc_allowed_congestion_control(struct ctl_table * ctl,int write,void * buffer,size_t * lenp,loff_t * ppos)256 static int proc_allowed_congestion_control(struct ctl_table *ctl,
257 int write, void *buffer,
258 size_t *lenp, loff_t *ppos)
259 {
260 struct ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX };
261 int ret;
262
263 tbl.data = kmalloc(tbl.maxlen, GFP_USER);
264 if (!tbl.data)
265 return -ENOMEM;
266
267 tcp_get_allowed_congestion_control(tbl.data, tbl.maxlen);
268 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
269 if (write && ret == 0)
270 ret = tcp_set_allowed_congestion_control(tbl.data);
271 kfree(tbl.data);
272 return ret;
273 }
274
sscanf_key(char * buf,__le32 * key)275 static int sscanf_key(char *buf, __le32 *key)
276 {
277 u32 user_key[4];
278 int i, ret = 0;
279
280 if (sscanf(buf, "%x-%x-%x-%x", user_key, user_key + 1,
281 user_key + 2, user_key + 3) != 4) {
282 ret = -EINVAL;
283 } else {
284 for (i = 0; i < ARRAY_SIZE(user_key); i++)
285 key[i] = cpu_to_le32(user_key[i]);
286 }
287 pr_debug("proc TFO key set 0x%x-%x-%x-%x <- 0x%s: %u\n",
288 user_key[0], user_key[1], user_key[2], user_key[3], buf, ret);
289
290 return ret;
291 }
292
proc_tcp_fastopen_key(struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)293 static int proc_tcp_fastopen_key(struct ctl_table *table, int write,
294 void *buffer, size_t *lenp, loff_t *ppos)
295 {
296 struct net *net = container_of(table->data, struct net,
297 ipv4.sysctl_tcp_fastopen);
298 /* maxlen to print the list of keys in hex (*2), with dashes
299 * separating doublewords and a comma in between keys.
300 */
301 struct ctl_table tbl = { .maxlen = ((TCP_FASTOPEN_KEY_LENGTH *
302 2 * TCP_FASTOPEN_KEY_MAX) +
303 (TCP_FASTOPEN_KEY_MAX * 5)) };
304 u32 user_key[TCP_FASTOPEN_KEY_BUF_LENGTH / sizeof(u32)];
305 __le32 key[TCP_FASTOPEN_KEY_BUF_LENGTH / sizeof(__le32)];
306 char *backup_data;
307 int ret, i = 0, off = 0, n_keys;
308
309 tbl.data = kmalloc(tbl.maxlen, GFP_KERNEL);
310 if (!tbl.data)
311 return -ENOMEM;
312
313 n_keys = tcp_fastopen_get_cipher(net, NULL, (u64 *)key);
314 if (!n_keys) {
315 memset(&key[0], 0, TCP_FASTOPEN_KEY_LENGTH);
316 n_keys = 1;
317 }
318
319 for (i = 0; i < n_keys * 4; i++)
320 user_key[i] = le32_to_cpu(key[i]);
321
322 for (i = 0; i < n_keys; i++) {
323 off += snprintf(tbl.data + off, tbl.maxlen - off,
324 "%08x-%08x-%08x-%08x",
325 user_key[i * 4],
326 user_key[i * 4 + 1],
327 user_key[i * 4 + 2],
328 user_key[i * 4 + 3]);
329
330 if (WARN_ON_ONCE(off >= tbl.maxlen - 1))
331 break;
332
333 if (i + 1 < n_keys)
334 off += snprintf(tbl.data + off, tbl.maxlen - off, ",");
335 }
336
337 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
338
339 if (write && ret == 0) {
340 backup_data = strchr(tbl.data, ',');
341 if (backup_data) {
342 *backup_data = '\0';
343 backup_data++;
344 }
345 if (sscanf_key(tbl.data, key)) {
346 ret = -EINVAL;
347 goto bad_key;
348 }
349 if (backup_data) {
350 if (sscanf_key(backup_data, key + 4)) {
351 ret = -EINVAL;
352 goto bad_key;
353 }
354 }
355 tcp_fastopen_reset_cipher(net, NULL, key,
356 backup_data ? key + 4 : NULL);
357 }
358
359 bad_key:
360 kfree(tbl.data);
361 return ret;
362 }
363
proc_tfo_blackhole_detect_timeout(struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)364 static int proc_tfo_blackhole_detect_timeout(struct ctl_table *table,
365 int write, void *buffer,
366 size_t *lenp, loff_t *ppos)
367 {
368 struct net *net = container_of(table->data, struct net,
369 ipv4.sysctl_tcp_fastopen_blackhole_timeout);
370 int ret;
371
372 ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
373 if (write && ret == 0)
374 atomic_set(&net->ipv4.tfo_active_disable_times, 0);
375
376 return ret;
377 }
378
proc_tcp_available_ulp(struct ctl_table * ctl,int write,void * buffer,size_t * lenp,loff_t * ppos)379 static int proc_tcp_available_ulp(struct ctl_table *ctl,
380 int write, void *buffer, size_t *lenp,
381 loff_t *ppos)
382 {
383 struct ctl_table tbl = { .maxlen = TCP_ULP_BUF_MAX, };
384 int ret;
385
386 tbl.data = kmalloc(tbl.maxlen, GFP_USER);
387 if (!tbl.data)
388 return -ENOMEM;
389 tcp_get_available_ulp(tbl.data, TCP_ULP_BUF_MAX);
390 ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
391 kfree(tbl.data);
392
393 return ret;
394 }
395
396 #ifdef CONFIG_IP_ROUTE_MULTIPATH
proc_fib_multipath_hash_policy(struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)397 static int proc_fib_multipath_hash_policy(struct ctl_table *table, int write,
398 void *buffer, size_t *lenp,
399 loff_t *ppos)
400 {
401 struct net *net = container_of(table->data, struct net,
402 ipv4.sysctl_fib_multipath_hash_policy);
403 int ret;
404
405 ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
406 if (write && ret == 0)
407 call_netevent_notifiers(NETEVENT_IPV4_MPATH_HASH_UPDATE, net);
408
409 return ret;
410 }
411 #endif
412
413 static struct ctl_table ipv4_table[] = {
414 {
415 .procname = "tcp_max_orphans",
416 .data = &sysctl_tcp_max_orphans,
417 .maxlen = sizeof(int),
418 .mode = 0644,
419 .proc_handler = proc_dointvec
420 },
421 {
422 .procname = "inet_peer_threshold",
423 .data = &inet_peer_threshold,
424 .maxlen = sizeof(int),
425 .mode = 0644,
426 .proc_handler = proc_dointvec
427 },
428 {
429 .procname = "inet_peer_minttl",
430 .data = &inet_peer_minttl,
431 .maxlen = sizeof(int),
432 .mode = 0644,
433 .proc_handler = proc_dointvec_jiffies,
434 },
435 {
436 .procname = "inet_peer_maxttl",
437 .data = &inet_peer_maxttl,
438 .maxlen = sizeof(int),
439 .mode = 0644,
440 .proc_handler = proc_dointvec_jiffies,
441 },
442 {
443 .procname = "tcp_mem",
444 .maxlen = sizeof(sysctl_tcp_mem),
445 .data = &sysctl_tcp_mem,
446 .mode = 0644,
447 .proc_handler = proc_doulongvec_minmax,
448 },
449 {
450 .procname = "tcp_low_latency",
451 .data = &sysctl_tcp_low_latency,
452 .maxlen = sizeof(int),
453 .mode = 0644,
454 .proc_handler = proc_dointvec
455 },
456 #ifdef CONFIG_NETLABEL
457 {
458 .procname = "cipso_cache_enable",
459 .data = &cipso_v4_cache_enabled,
460 .maxlen = sizeof(int),
461 .mode = 0644,
462 .proc_handler = proc_dointvec,
463 },
464 {
465 .procname = "cipso_cache_bucket_size",
466 .data = &cipso_v4_cache_bucketsize,
467 .maxlen = sizeof(int),
468 .mode = 0644,
469 .proc_handler = proc_dointvec,
470 },
471 {
472 .procname = "cipso_rbm_optfmt",
473 .data = &cipso_v4_rbm_optfmt,
474 .maxlen = sizeof(int),
475 .mode = 0644,
476 .proc_handler = proc_dointvec,
477 },
478 {
479 .procname = "cipso_rbm_strictvalid",
480 .data = &cipso_v4_rbm_strictvalid,
481 .maxlen = sizeof(int),
482 .mode = 0644,
483 .proc_handler = proc_dointvec,
484 },
485 #endif /* CONFIG_NETLABEL */
486 {
487 .procname = "tcp_available_ulp",
488 .maxlen = TCP_ULP_BUF_MAX,
489 .mode = 0444,
490 .proc_handler = proc_tcp_available_ulp,
491 },
492 {
493 .procname = "icmp_msgs_per_sec",
494 .data = &sysctl_icmp_msgs_per_sec,
495 .maxlen = sizeof(int),
496 .mode = 0644,
497 .proc_handler = proc_dointvec_minmax,
498 .extra1 = SYSCTL_ZERO,
499 },
500 {
501 .procname = "icmp_msgs_burst",
502 .data = &sysctl_icmp_msgs_burst,
503 .maxlen = sizeof(int),
504 .mode = 0644,
505 .proc_handler = proc_dointvec_minmax,
506 .extra1 = SYSCTL_ZERO,
507 },
508 {
509 .procname = "udp_mem",
510 .data = &sysctl_udp_mem,
511 .maxlen = sizeof(sysctl_udp_mem),
512 .mode = 0644,
513 .proc_handler = proc_doulongvec_minmax,
514 },
515 {
516 .procname = "fib_sync_mem",
517 .data = &sysctl_fib_sync_mem,
518 .maxlen = sizeof(sysctl_fib_sync_mem),
519 .mode = 0644,
520 .proc_handler = proc_douintvec_minmax,
521 .extra1 = &sysctl_fib_sync_mem_min,
522 .extra2 = &sysctl_fib_sync_mem_max,
523 },
524 {
525 .procname = "tcp_rx_skb_cache",
526 .data = &tcp_rx_skb_cache_key.key,
527 .mode = 0644,
528 .proc_handler = proc_do_static_key,
529 },
530 {
531 .procname = "tcp_tx_skb_cache",
532 .data = &tcp_tx_skb_cache_key.key,
533 .mode = 0644,
534 .proc_handler = proc_do_static_key,
535 },
536 { }
537 };
538
539 static struct ctl_table ipv4_net_table[] = {
540 {
541 .procname = "icmp_echo_ignore_all",
542 .data = &init_net.ipv4.sysctl_icmp_echo_ignore_all,
543 .maxlen = sizeof(int),
544 .mode = 0644,
545 .proc_handler = proc_dointvec
546 },
547 {
548 .procname = "icmp_echo_ignore_broadcasts",
549 .data = &init_net.ipv4.sysctl_icmp_echo_ignore_broadcasts,
550 .maxlen = sizeof(int),
551 .mode = 0644,
552 .proc_handler = proc_dointvec
553 },
554 {
555 .procname = "icmp_ignore_bogus_error_responses",
556 .data = &init_net.ipv4.sysctl_icmp_ignore_bogus_error_responses,
557 .maxlen = sizeof(int),
558 .mode = 0644,
559 .proc_handler = proc_dointvec
560 },
561 {
562 .procname = "icmp_errors_use_inbound_ifaddr",
563 .data = &init_net.ipv4.sysctl_icmp_errors_use_inbound_ifaddr,
564 .maxlen = sizeof(int),
565 .mode = 0644,
566 .proc_handler = proc_dointvec
567 },
568 {
569 .procname = "icmp_ratelimit",
570 .data = &init_net.ipv4.sysctl_icmp_ratelimit,
571 .maxlen = sizeof(int),
572 .mode = 0644,
573 .proc_handler = proc_dointvec_ms_jiffies,
574 },
575 {
576 .procname = "icmp_ratemask",
577 .data = &init_net.ipv4.sysctl_icmp_ratemask,
578 .maxlen = sizeof(int),
579 .mode = 0644,
580 .proc_handler = proc_dointvec
581 },
582 {
583 .procname = "ping_group_range",
584 .data = &init_net.ipv4.ping_group_range.range,
585 .maxlen = sizeof(gid_t)*2,
586 .mode = 0644,
587 .proc_handler = ipv4_ping_group_range,
588 },
589 #ifdef CONFIG_NET_L3_MASTER_DEV
590 {
591 .procname = "raw_l3mdev_accept",
592 .data = &init_net.ipv4.sysctl_raw_l3mdev_accept,
593 .maxlen = sizeof(int),
594 .mode = 0644,
595 .proc_handler = proc_dointvec_minmax,
596 .extra1 = SYSCTL_ZERO,
597 .extra2 = SYSCTL_ONE,
598 },
599 #endif
600 {
601 .procname = "tcp_ecn",
602 .data = &init_net.ipv4.sysctl_tcp_ecn,
603 .maxlen = sizeof(int),
604 .mode = 0644,
605 .proc_handler = proc_dointvec
606 },
607 {
608 .procname = "tcp_ecn_fallback",
609 .data = &init_net.ipv4.sysctl_tcp_ecn_fallback,
610 .maxlen = sizeof(int),
611 .mode = 0644,
612 .proc_handler = proc_dointvec
613 },
614 {
615 .procname = "ip_dynaddr",
616 .data = &init_net.ipv4.sysctl_ip_dynaddr,
617 .maxlen = sizeof(int),
618 .mode = 0644,
619 .proc_handler = proc_dointvec
620 },
621 {
622 .procname = "ip_early_demux",
623 .data = &init_net.ipv4.sysctl_ip_early_demux,
624 .maxlen = sizeof(int),
625 .mode = 0644,
626 .proc_handler = proc_dointvec
627 },
628 {
629 .procname = "udp_early_demux",
630 .data = &init_net.ipv4.sysctl_udp_early_demux,
631 .maxlen = sizeof(int),
632 .mode = 0644,
633 .proc_handler = proc_douintvec_minmax,
634 },
635 {
636 .procname = "tcp_early_demux",
637 .data = &init_net.ipv4.sysctl_tcp_early_demux,
638 .maxlen = sizeof(int),
639 .mode = 0644,
640 .proc_handler = proc_douintvec_minmax,
641 },
642 {
643 .procname = "nexthop_compat_mode",
644 .data = &init_net.ipv4.sysctl_nexthop_compat_mode,
645 .maxlen = sizeof(int),
646 .mode = 0644,
647 .proc_handler = proc_dointvec_minmax,
648 .extra1 = SYSCTL_ZERO,
649 .extra2 = SYSCTL_ONE,
650 },
651 {
652 .procname = "ip_default_ttl",
653 .data = &init_net.ipv4.sysctl_ip_default_ttl,
654 .maxlen = sizeof(int),
655 .mode = 0644,
656 .proc_handler = proc_dointvec_minmax,
657 .extra1 = &ip_ttl_min,
658 .extra2 = &ip_ttl_max,
659 },
660 {
661 .procname = "ip_local_port_range",
662 .maxlen = sizeof(init_net.ipv4.ip_local_ports.range),
663 .data = &init_net.ipv4.ip_local_ports.range,
664 .mode = 0644,
665 .proc_handler = ipv4_local_port_range,
666 },
667 {
668 .procname = "ip_local_reserved_ports",
669 .data = &init_net.ipv4.sysctl_local_reserved_ports,
670 .maxlen = 65536,
671 .mode = 0644,
672 .proc_handler = proc_do_large_bitmap,
673 },
674 {
675 .procname = "ip_local_unbindable_ports",
676 .data = &init_net.ipv4.sysctl_local_unbindable_ports,
677 .maxlen = 65536,
678 .mode = 0644,
679 .proc_handler = proc_do_large_bitmap,
680 },
681 {
682 .procname = "ip_no_pmtu_disc",
683 .data = &init_net.ipv4.sysctl_ip_no_pmtu_disc,
684 .maxlen = sizeof(int),
685 .mode = 0644,
686 .proc_handler = proc_dointvec
687 },
688 {
689 .procname = "ip_forward_use_pmtu",
690 .data = &init_net.ipv4.sysctl_ip_fwd_use_pmtu,
691 .maxlen = sizeof(int),
692 .mode = 0644,
693 .proc_handler = proc_dointvec,
694 },
695 {
696 .procname = "ip_forward_update_priority",
697 .data = &init_net.ipv4.sysctl_ip_fwd_update_priority,
698 .maxlen = sizeof(int),
699 .mode = 0644,
700 .proc_handler = ipv4_fwd_update_priority,
701 .extra1 = SYSCTL_ZERO,
702 .extra2 = SYSCTL_ONE,
703 },
704 {
705 .procname = "ip_nonlocal_bind",
706 .data = &init_net.ipv4.sysctl_ip_nonlocal_bind,
707 .maxlen = sizeof(int),
708 .mode = 0644,
709 .proc_handler = proc_dointvec
710 },
711 {
712 .procname = "ip_autobind_reuse",
713 .data = &init_net.ipv4.sysctl_ip_autobind_reuse,
714 .maxlen = sizeof(int),
715 .mode = 0644,
716 .proc_handler = proc_dointvec_minmax,
717 .extra1 = SYSCTL_ZERO,
718 .extra2 = SYSCTL_ONE,
719 },
720 {
721 .procname = "fwmark_reflect",
722 .data = &init_net.ipv4.sysctl_fwmark_reflect,
723 .maxlen = sizeof(int),
724 .mode = 0644,
725 .proc_handler = proc_dointvec,
726 },
727 {
728 .procname = "tcp_fwmark_accept",
729 .data = &init_net.ipv4.sysctl_tcp_fwmark_accept,
730 .maxlen = sizeof(int),
731 .mode = 0644,
732 .proc_handler = proc_dointvec,
733 },
734 #ifdef CONFIG_NET_L3_MASTER_DEV
735 {
736 .procname = "tcp_l3mdev_accept",
737 .data = &init_net.ipv4.sysctl_tcp_l3mdev_accept,
738 .maxlen = sizeof(int),
739 .mode = 0644,
740 .proc_handler = proc_dointvec_minmax,
741 .extra1 = SYSCTL_ZERO,
742 .extra2 = SYSCTL_ONE,
743 },
744 #endif
745 {
746 .procname = "tcp_mtu_probing",
747 .data = &init_net.ipv4.sysctl_tcp_mtu_probing,
748 .maxlen = sizeof(int),
749 .mode = 0644,
750 .proc_handler = proc_dointvec,
751 },
752 {
753 .procname = "tcp_base_mss",
754 .data = &init_net.ipv4.sysctl_tcp_base_mss,
755 .maxlen = sizeof(int),
756 .mode = 0644,
757 .proc_handler = proc_dointvec,
758 },
759 {
760 .procname = "tcp_min_snd_mss",
761 .data = &init_net.ipv4.sysctl_tcp_min_snd_mss,
762 .maxlen = sizeof(int),
763 .mode = 0644,
764 .proc_handler = proc_dointvec_minmax,
765 .extra1 = &tcp_min_snd_mss_min,
766 .extra2 = &tcp_min_snd_mss_max,
767 },
768 {
769 .procname = "tcp_mtu_probe_floor",
770 .data = &init_net.ipv4.sysctl_tcp_mtu_probe_floor,
771 .maxlen = sizeof(int),
772 .mode = 0644,
773 .proc_handler = proc_dointvec_minmax,
774 .extra1 = &tcp_min_snd_mss_min,
775 .extra2 = &tcp_min_snd_mss_max,
776 },
777 {
778 .procname = "tcp_probe_threshold",
779 .data = &init_net.ipv4.sysctl_tcp_probe_threshold,
780 .maxlen = sizeof(int),
781 .mode = 0644,
782 .proc_handler = proc_dointvec,
783 },
784 {
785 .procname = "tcp_probe_interval",
786 .data = &init_net.ipv4.sysctl_tcp_probe_interval,
787 .maxlen = sizeof(u32),
788 .mode = 0644,
789 .proc_handler = proc_douintvec_minmax,
790 .extra2 = &u32_max_div_HZ,
791 },
792 {
793 .procname = "igmp_link_local_mcast_reports",
794 .data = &init_net.ipv4.sysctl_igmp_llm_reports,
795 .maxlen = sizeof(int),
796 .mode = 0644,
797 .proc_handler = proc_dointvec
798 },
799 {
800 .procname = "igmp_max_memberships",
801 .data = &init_net.ipv4.sysctl_igmp_max_memberships,
802 .maxlen = sizeof(int),
803 .mode = 0644,
804 .proc_handler = proc_dointvec
805 },
806 {
807 .procname = "igmp_max_msf",
808 .data = &init_net.ipv4.sysctl_igmp_max_msf,
809 .maxlen = sizeof(int),
810 .mode = 0644,
811 .proc_handler = proc_dointvec
812 },
813 #ifdef CONFIG_IP_MULTICAST
814 {
815 .procname = "igmp_qrv",
816 .data = &init_net.ipv4.sysctl_igmp_qrv,
817 .maxlen = sizeof(int),
818 .mode = 0644,
819 .proc_handler = proc_dointvec_minmax,
820 .extra1 = SYSCTL_ONE
821 },
822 #endif
823 {
824 .procname = "tcp_congestion_control",
825 .data = &init_net.ipv4.tcp_congestion_control,
826 .mode = 0644,
827 .maxlen = TCP_CA_NAME_MAX,
828 .proc_handler = proc_tcp_congestion_control,
829 },
830 {
831 .procname = "tcp_available_congestion_control",
832 .maxlen = TCP_CA_BUF_MAX,
833 .mode = 0444,
834 .proc_handler = proc_tcp_available_congestion_control,
835 },
836 {
837 .procname = "tcp_allowed_congestion_control",
838 .maxlen = TCP_CA_BUF_MAX,
839 .mode = 0644,
840 .proc_handler = proc_allowed_congestion_control,
841 },
842 {
843 .procname = "tcp_keepalive_time",
844 .data = &init_net.ipv4.sysctl_tcp_keepalive_time,
845 .maxlen = sizeof(int),
846 .mode = 0644,
847 .proc_handler = proc_dointvec_jiffies,
848 },
849 {
850 .procname = "tcp_keepalive_probes",
851 .data = &init_net.ipv4.sysctl_tcp_keepalive_probes,
852 .maxlen = sizeof(int),
853 .mode = 0644,
854 .proc_handler = proc_dointvec
855 },
856 {
857 .procname = "tcp_keepalive_intvl",
858 .data = &init_net.ipv4.sysctl_tcp_keepalive_intvl,
859 .maxlen = sizeof(int),
860 .mode = 0644,
861 .proc_handler = proc_dointvec_jiffies,
862 },
863 {
864 .procname = "tcp_syn_retries",
865 .data = &init_net.ipv4.sysctl_tcp_syn_retries,
866 .maxlen = sizeof(int),
867 .mode = 0644,
868 .proc_handler = proc_dointvec_minmax,
869 .extra1 = &tcp_syn_retries_min,
870 .extra2 = &tcp_syn_retries_max
871 },
872 {
873 .procname = "tcp_synack_retries",
874 .data = &init_net.ipv4.sysctl_tcp_synack_retries,
875 .maxlen = sizeof(int),
876 .mode = 0644,
877 .proc_handler = proc_dointvec
878 },
879 #ifdef CONFIG_SYN_COOKIES
880 {
881 .procname = "tcp_syncookies",
882 .data = &init_net.ipv4.sysctl_tcp_syncookies,
883 .maxlen = sizeof(int),
884 .mode = 0644,
885 .proc_handler = proc_dointvec
886 },
887 #endif
888 {
889 .procname = "tcp_reordering",
890 .data = &init_net.ipv4.sysctl_tcp_reordering,
891 .maxlen = sizeof(int),
892 .mode = 0644,
893 .proc_handler = proc_dointvec
894 },
895 {
896 .procname = "tcp_retries1",
897 .data = &init_net.ipv4.sysctl_tcp_retries1,
898 .maxlen = sizeof(int),
899 .mode = 0644,
900 .proc_handler = proc_dointvec_minmax,
901 .extra2 = &tcp_retr1_max
902 },
903 {
904 .procname = "tcp_retries2",
905 .data = &init_net.ipv4.sysctl_tcp_retries2,
906 .maxlen = sizeof(int),
907 .mode = 0644,
908 .proc_handler = proc_dointvec
909 },
910 {
911 .procname = "tcp_orphan_retries",
912 .data = &init_net.ipv4.sysctl_tcp_orphan_retries,
913 .maxlen = sizeof(int),
914 .mode = 0644,
915 .proc_handler = proc_dointvec
916 },
917 {
918 .procname = "tcp_fin_timeout",
919 .data = &init_net.ipv4.sysctl_tcp_fin_timeout,
920 .maxlen = sizeof(int),
921 .mode = 0644,
922 .proc_handler = proc_dointvec_jiffies,
923 },
924 {
925 .procname = "tcp_notsent_lowat",
926 .data = &init_net.ipv4.sysctl_tcp_notsent_lowat,
927 .maxlen = sizeof(unsigned int),
928 .mode = 0644,
929 .proc_handler = proc_douintvec,
930 },
931 {
932 .procname = "tcp_tw_reuse",
933 .data = &init_net.ipv4.sysctl_tcp_tw_reuse,
934 .maxlen = sizeof(int),
935 .mode = 0644,
936 .proc_handler = proc_dointvec_minmax,
937 .extra1 = SYSCTL_ZERO,
938 .extra2 = &two,
939 },
940 {
941 .procname = "tcp_max_tw_buckets",
942 .data = &init_net.ipv4.tcp_death_row.sysctl_max_tw_buckets,
943 .maxlen = sizeof(int),
944 .mode = 0644,
945 .proc_handler = proc_dointvec
946 },
947 {
948 .procname = "tcp_max_syn_backlog",
949 .data = &init_net.ipv4.sysctl_max_syn_backlog,
950 .maxlen = sizeof(int),
951 .mode = 0644,
952 .proc_handler = proc_dointvec
953 },
954 {
955 .procname = "tcp_fastopen",
956 .data = &init_net.ipv4.sysctl_tcp_fastopen,
957 .maxlen = sizeof(int),
958 .mode = 0644,
959 .proc_handler = proc_dointvec,
960 },
961 {
962 .procname = "tcp_fastopen_key",
963 .mode = 0600,
964 .data = &init_net.ipv4.sysctl_tcp_fastopen,
965 /* maxlen to print the list of keys in hex (*2), with dashes
966 * separating doublewords and a comma in between keys.
967 */
968 .maxlen = ((TCP_FASTOPEN_KEY_LENGTH *
969 2 * TCP_FASTOPEN_KEY_MAX) +
970 (TCP_FASTOPEN_KEY_MAX * 5)),
971 .proc_handler = proc_tcp_fastopen_key,
972 },
973 {
974 .procname = "tcp_fastopen_blackhole_timeout_sec",
975 .data = &init_net.ipv4.sysctl_tcp_fastopen_blackhole_timeout,
976 .maxlen = sizeof(int),
977 .mode = 0644,
978 .proc_handler = proc_tfo_blackhole_detect_timeout,
979 .extra1 = SYSCTL_ZERO,
980 },
981 #ifdef CONFIG_IP_ROUTE_MULTIPATH
982 {
983 .procname = "fib_multipath_use_neigh",
984 .data = &init_net.ipv4.sysctl_fib_multipath_use_neigh,
985 .maxlen = sizeof(int),
986 .mode = 0644,
987 .proc_handler = proc_dointvec_minmax,
988 .extra1 = SYSCTL_ZERO,
989 .extra2 = SYSCTL_ONE,
990 },
991 {
992 .procname = "fib_multipath_hash_policy",
993 .data = &init_net.ipv4.sysctl_fib_multipath_hash_policy,
994 .maxlen = sizeof(int),
995 .mode = 0644,
996 .proc_handler = proc_fib_multipath_hash_policy,
997 .extra1 = SYSCTL_ZERO,
998 .extra2 = &two,
999 },
1000 #endif
1001 {
1002 .procname = "ip_unprivileged_port_start",
1003 .maxlen = sizeof(int),
1004 .data = &init_net.ipv4.sysctl_ip_prot_sock,
1005 .mode = 0644,
1006 .proc_handler = ipv4_privileged_ports,
1007 },
1008 #ifdef CONFIG_NET_L3_MASTER_DEV
1009 {
1010 .procname = "udp_l3mdev_accept",
1011 .data = &init_net.ipv4.sysctl_udp_l3mdev_accept,
1012 .maxlen = sizeof(int),
1013 .mode = 0644,
1014 .proc_handler = proc_dointvec_minmax,
1015 .extra1 = SYSCTL_ZERO,
1016 .extra2 = SYSCTL_ONE,
1017 },
1018 #endif
1019 {
1020 .procname = "tcp_sack",
1021 .data = &init_net.ipv4.sysctl_tcp_sack,
1022 .maxlen = sizeof(int),
1023 .mode = 0644,
1024 .proc_handler = proc_dointvec
1025 },
1026 {
1027 .procname = "tcp_window_scaling",
1028 .data = &init_net.ipv4.sysctl_tcp_window_scaling,
1029 .maxlen = sizeof(int),
1030 .mode = 0644,
1031 .proc_handler = proc_dointvec
1032 },
1033 {
1034 .procname = "tcp_timestamps",
1035 .data = &init_net.ipv4.sysctl_tcp_timestamps,
1036 .maxlen = sizeof(int),
1037 .mode = 0644,
1038 .proc_handler = proc_dointvec
1039 },
1040 {
1041 .procname = "tcp_early_retrans",
1042 .data = &init_net.ipv4.sysctl_tcp_early_retrans,
1043 .maxlen = sizeof(int),
1044 .mode = 0644,
1045 .proc_handler = proc_dointvec_minmax,
1046 .extra1 = SYSCTL_ZERO,
1047 .extra2 = &four,
1048 },
1049 {
1050 .procname = "tcp_recovery",
1051 .data = &init_net.ipv4.sysctl_tcp_recovery,
1052 .maxlen = sizeof(int),
1053 .mode = 0644,
1054 .proc_handler = proc_dointvec,
1055 },
1056 {
1057 .procname = "tcp_thin_linear_timeouts",
1058 .data = &init_net.ipv4.sysctl_tcp_thin_linear_timeouts,
1059 .maxlen = sizeof(int),
1060 .mode = 0644,
1061 .proc_handler = proc_dointvec
1062 },
1063 {
1064 .procname = "tcp_slow_start_after_idle",
1065 .data = &init_net.ipv4.sysctl_tcp_slow_start_after_idle,
1066 .maxlen = sizeof(int),
1067 .mode = 0644,
1068 .proc_handler = proc_dointvec
1069 },
1070 {
1071 .procname = "tcp_retrans_collapse",
1072 .data = &init_net.ipv4.sysctl_tcp_retrans_collapse,
1073 .maxlen = sizeof(int),
1074 .mode = 0644,
1075 .proc_handler = proc_dointvec
1076 },
1077 {
1078 .procname = "tcp_stdurg",
1079 .data = &init_net.ipv4.sysctl_tcp_stdurg,
1080 .maxlen = sizeof(int),
1081 .mode = 0644,
1082 .proc_handler = proc_dointvec
1083 },
1084 {
1085 .procname = "tcp_rfc1337",
1086 .data = &init_net.ipv4.sysctl_tcp_rfc1337,
1087 .maxlen = sizeof(int),
1088 .mode = 0644,
1089 .proc_handler = proc_dointvec
1090 },
1091 {
1092 .procname = "tcp_abort_on_overflow",
1093 .data = &init_net.ipv4.sysctl_tcp_abort_on_overflow,
1094 .maxlen = sizeof(int),
1095 .mode = 0644,
1096 .proc_handler = proc_dointvec
1097 },
1098 {
1099 .procname = "tcp_fack",
1100 .data = &init_net.ipv4.sysctl_tcp_fack,
1101 .maxlen = sizeof(int),
1102 .mode = 0644,
1103 .proc_handler = proc_dointvec
1104 },
1105 {
1106 .procname = "tcp_max_reordering",
1107 .data = &init_net.ipv4.sysctl_tcp_max_reordering,
1108 .maxlen = sizeof(int),
1109 .mode = 0644,
1110 .proc_handler = proc_dointvec
1111 },
1112 {
1113 .procname = "tcp_dsack",
1114 .data = &init_net.ipv4.sysctl_tcp_dsack,
1115 .maxlen = sizeof(int),
1116 .mode = 0644,
1117 .proc_handler = proc_dointvec
1118 },
1119 {
1120 .procname = "tcp_app_win",
1121 .data = &init_net.ipv4.sysctl_tcp_app_win,
1122 .maxlen = sizeof(int),
1123 .mode = 0644,
1124 .proc_handler = proc_dointvec
1125 },
1126 {
1127 .procname = "tcp_adv_win_scale",
1128 .data = &init_net.ipv4.sysctl_tcp_adv_win_scale,
1129 .maxlen = sizeof(int),
1130 .mode = 0644,
1131 .proc_handler = proc_dointvec_minmax,
1132 .extra1 = &tcp_adv_win_scale_min,
1133 .extra2 = &tcp_adv_win_scale_max,
1134 },
1135 {
1136 .procname = "tcp_frto",
1137 .data = &init_net.ipv4.sysctl_tcp_frto,
1138 .maxlen = sizeof(int),
1139 .mode = 0644,
1140 .proc_handler = proc_dointvec
1141 },
1142 {
1143 .procname = "tcp_no_metrics_save",
1144 .data = &init_net.ipv4.sysctl_tcp_nometrics_save,
1145 .maxlen = sizeof(int),
1146 .mode = 0644,
1147 .proc_handler = proc_dointvec,
1148 },
1149 {
1150 .procname = "tcp_no_ssthresh_metrics_save",
1151 .data = &init_net.ipv4.sysctl_tcp_no_ssthresh_metrics_save,
1152 .maxlen = sizeof(int),
1153 .mode = 0644,
1154 .proc_handler = proc_dointvec_minmax,
1155 .extra1 = SYSCTL_ZERO,
1156 .extra2 = SYSCTL_ONE,
1157 },
1158 {
1159 .procname = "tcp_moderate_rcvbuf",
1160 .data = &init_net.ipv4.sysctl_tcp_moderate_rcvbuf,
1161 .maxlen = sizeof(int),
1162 .mode = 0644,
1163 .proc_handler = proc_dointvec,
1164 },
1165 {
1166 .procname = "tcp_tso_win_divisor",
1167 .data = &init_net.ipv4.sysctl_tcp_tso_win_divisor,
1168 .maxlen = sizeof(int),
1169 .mode = 0644,
1170 .proc_handler = proc_dointvec,
1171 },
1172 {
1173 .procname = "tcp_workaround_signed_windows",
1174 .data = &init_net.ipv4.sysctl_tcp_workaround_signed_windows,
1175 .maxlen = sizeof(int),
1176 .mode = 0644,
1177 .proc_handler = proc_dointvec
1178 },
1179 {
1180 .procname = "tcp_limit_output_bytes",
1181 .data = &init_net.ipv4.sysctl_tcp_limit_output_bytes,
1182 .maxlen = sizeof(int),
1183 .mode = 0644,
1184 .proc_handler = proc_dointvec
1185 },
1186 {
1187 .procname = "tcp_challenge_ack_limit",
1188 .data = &init_net.ipv4.sysctl_tcp_challenge_ack_limit,
1189 .maxlen = sizeof(int),
1190 .mode = 0644,
1191 .proc_handler = proc_dointvec
1192 },
1193 {
1194 .procname = "tcp_min_tso_segs",
1195 .data = &init_net.ipv4.sysctl_tcp_min_tso_segs,
1196 .maxlen = sizeof(int),
1197 .mode = 0644,
1198 .proc_handler = proc_dointvec_minmax,
1199 .extra1 = SYSCTL_ONE,
1200 .extra2 = &gso_max_segs,
1201 },
1202 {
1203 .procname = "tcp_min_rtt_wlen",
1204 .data = &init_net.ipv4.sysctl_tcp_min_rtt_wlen,
1205 .maxlen = sizeof(int),
1206 .mode = 0644,
1207 .proc_handler = proc_dointvec_minmax,
1208 .extra1 = SYSCTL_ZERO,
1209 .extra2 = &one_day_secs
1210 },
1211 {
1212 .procname = "tcp_autocorking",
1213 .data = &init_net.ipv4.sysctl_tcp_autocorking,
1214 .maxlen = sizeof(int),
1215 .mode = 0644,
1216 .proc_handler = proc_dointvec_minmax,
1217 .extra1 = SYSCTL_ZERO,
1218 .extra2 = SYSCTL_ONE,
1219 },
1220 {
1221 .procname = "tcp_invalid_ratelimit",
1222 .data = &init_net.ipv4.sysctl_tcp_invalid_ratelimit,
1223 .maxlen = sizeof(int),
1224 .mode = 0644,
1225 .proc_handler = proc_dointvec_ms_jiffies,
1226 },
1227 {
1228 .procname = "tcp_pacing_ss_ratio",
1229 .data = &init_net.ipv4.sysctl_tcp_pacing_ss_ratio,
1230 .maxlen = sizeof(int),
1231 .mode = 0644,
1232 .proc_handler = proc_dointvec_minmax,
1233 .extra1 = SYSCTL_ZERO,
1234 .extra2 = &thousand,
1235 },
1236 {
1237 .procname = "tcp_pacing_ca_ratio",
1238 .data = &init_net.ipv4.sysctl_tcp_pacing_ca_ratio,
1239 .maxlen = sizeof(int),
1240 .mode = 0644,
1241 .proc_handler = proc_dointvec_minmax,
1242 .extra1 = SYSCTL_ZERO,
1243 .extra2 = &thousand,
1244 },
1245 {
1246 .procname = "tcp_wmem",
1247 .data = &init_net.ipv4.sysctl_tcp_wmem,
1248 .maxlen = sizeof(init_net.ipv4.sysctl_tcp_wmem),
1249 .mode = 0644,
1250 .proc_handler = proc_dointvec_minmax,
1251 .extra1 = SYSCTL_ONE,
1252 },
1253 {
1254 .procname = "tcp_rmem",
1255 .data = &init_net.ipv4.sysctl_tcp_rmem,
1256 .maxlen = sizeof(init_net.ipv4.sysctl_tcp_rmem),
1257 .mode = 0644,
1258 .proc_handler = proc_dointvec_minmax,
1259 .extra1 = SYSCTL_ONE,
1260 },
1261 {
1262 .procname = "tcp_comp_sack_delay_ns",
1263 .data = &init_net.ipv4.sysctl_tcp_comp_sack_delay_ns,
1264 .maxlen = sizeof(unsigned long),
1265 .mode = 0644,
1266 .proc_handler = proc_doulongvec_minmax,
1267 },
1268 {
1269 .procname = "tcp_comp_sack_slack_ns",
1270 .data = &init_net.ipv4.sysctl_tcp_comp_sack_slack_ns,
1271 .maxlen = sizeof(unsigned long),
1272 .mode = 0644,
1273 .proc_handler = proc_doulongvec_minmax,
1274 },
1275 {
1276 .procname = "tcp_comp_sack_nr",
1277 .data = &init_net.ipv4.sysctl_tcp_comp_sack_nr,
1278 .maxlen = sizeof(int),
1279 .mode = 0644,
1280 .proc_handler = proc_dointvec_minmax,
1281 .extra1 = SYSCTL_ZERO,
1282 .extra2 = &comp_sack_nr_max,
1283 },
1284 {
1285 .procname = "tcp_reflect_tos",
1286 .data = &init_net.ipv4.sysctl_tcp_reflect_tos,
1287 .maxlen = sizeof(int),
1288 .mode = 0644,
1289 .proc_handler = proc_dointvec_minmax,
1290 .extra1 = SYSCTL_ZERO,
1291 .extra2 = SYSCTL_ONE,
1292 },
1293 {
1294 .procname = "udp_rmem_min",
1295 .data = &init_net.ipv4.sysctl_udp_rmem_min,
1296 .maxlen = sizeof(init_net.ipv4.sysctl_udp_rmem_min),
1297 .mode = 0644,
1298 .proc_handler = proc_dointvec_minmax,
1299 .extra1 = SYSCTL_ONE
1300 },
1301 {
1302 .procname = "udp_wmem_min",
1303 .data = &init_net.ipv4.sysctl_udp_wmem_min,
1304 .maxlen = sizeof(init_net.ipv4.sysctl_udp_wmem_min),
1305 .mode = 0644,
1306 .proc_handler = proc_dointvec_minmax,
1307 .extra1 = SYSCTL_ONE
1308 },
1309 { }
1310 };
1311
ipv4_sysctl_init_net(struct net * net)1312 static __net_init int ipv4_sysctl_init_net(struct net *net)
1313 {
1314 struct ctl_table *table;
1315
1316 table = ipv4_net_table;
1317 if (!net_eq(net, &init_net)) {
1318 int i;
1319
1320 table = kmemdup(table, sizeof(ipv4_net_table), GFP_KERNEL);
1321 if (!table)
1322 goto err_alloc;
1323
1324 for (i = 0; i < ARRAY_SIZE(ipv4_net_table) - 1; i++) {
1325 if (table[i].data) {
1326 /* Update the variables to point into
1327 * the current struct net
1328 */
1329 table[i].data += (void *)net - (void *)&init_net;
1330 } else {
1331 /* Entries without data pointer are global;
1332 * Make them read-only in non-init_net ns
1333 */
1334 table[i].mode &= ~0222;
1335 }
1336 }
1337 }
1338
1339 net->ipv4.ipv4_hdr = register_net_sysctl(net, "net/ipv4", table);
1340 if (!net->ipv4.ipv4_hdr)
1341 goto err_reg;
1342
1343 net->ipv4.sysctl_local_reserved_ports = kzalloc(65536 / 8, GFP_KERNEL);
1344 if (!net->ipv4.sysctl_local_reserved_ports)
1345 goto err_reserved_ports;
1346
1347 net->ipv4.sysctl_local_unbindable_ports = kzalloc(65536 / 8, GFP_KERNEL);
1348 if (!net->ipv4.sysctl_local_unbindable_ports)
1349 goto err_unbindable_ports;
1350
1351 return 0;
1352
1353 err_unbindable_ports:
1354 kfree(net->ipv4.sysctl_local_reserved_ports);
1355 err_reserved_ports:
1356 unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
1357 err_reg:
1358 if (!net_eq(net, &init_net))
1359 kfree(table);
1360 err_alloc:
1361 return -ENOMEM;
1362 }
1363
ipv4_sysctl_exit_net(struct net * net)1364 static __net_exit void ipv4_sysctl_exit_net(struct net *net)
1365 {
1366 struct ctl_table *table;
1367
1368 kfree(net->ipv4.sysctl_local_unbindable_ports);
1369 kfree(net->ipv4.sysctl_local_reserved_ports);
1370 table = net->ipv4.ipv4_hdr->ctl_table_arg;
1371 unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
1372 kfree(table);
1373 }
1374
1375 static __net_initdata struct pernet_operations ipv4_sysctl_ops = {
1376 .init = ipv4_sysctl_init_net,
1377 .exit = ipv4_sysctl_exit_net,
1378 };
1379
sysctl_ipv4_init(void)1380 static __init int sysctl_ipv4_init(void)
1381 {
1382 struct ctl_table_header *hdr;
1383
1384 hdr = register_net_sysctl(&init_net, "net/ipv4", ipv4_table);
1385 if (!hdr)
1386 return -ENOMEM;
1387
1388 if (register_pernet_subsys(&ipv4_sysctl_ops)) {
1389 unregister_net_sysctl_table(hdr);
1390 return -ENOMEM;
1391 }
1392
1393 return 0;
1394 }
1395
1396 __initcall(sysctl_ipv4_init);
1397