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/sysctl.h>
10 #include <linux/seqlock.h>
11 #include <linux/init.h>
12 #include <linux/slab.h>
13 #include <net/icmp.h>
14 #include <net/ip.h>
15 #include <net/ip_fib.h>
16 #include <net/tcp.h>
17 #include <net/udp.h>
18 #include <net/cipso_ipv4.h>
19 #include <net/ping.h>
20 #include <net/protocol.h>
21 #include <net/netevent.h>
22
23 static int tcp_retr1_max = 255;
24 static int ip_local_port_range_min[] = { 1, 1 };
25 static int ip_local_port_range_max[] = { 65535, 65535 };
26 static int tcp_adv_win_scale_min = -31;
27 static int tcp_adv_win_scale_max = 31;
28 static int tcp_app_win_max = 31;
29 static int tcp_min_snd_mss_min = TCP_MIN_SND_MSS;
30 static int tcp_min_snd_mss_max = 65535;
31 static int tcp_rto_max_max = TCP_RTO_MAX_SEC * MSEC_PER_SEC;
32 static int ip_privileged_port_min;
33 static int ip_privileged_port_max = 65535;
34 static int ip_ttl_min = 1;
35 static int ip_ttl_max = 255;
36 static int tcp_syn_retries_min = 1;
37 static int tcp_syn_retries_max = MAX_TCP_SYNCNT;
38 static int tcp_syn_linear_timeouts_max = MAX_TCP_SYNCNT;
39 static unsigned long ip_ping_group_range_min[] = { 0, 0 };
40 static unsigned long ip_ping_group_range_max[] = { GID_T_MAX, GID_T_MAX };
41 static u32 u32_max_div_HZ = UINT_MAX / HZ;
42 static int one_day_secs = 24 * 3600;
43 static u32 fib_multipath_hash_fields_all_mask __maybe_unused =
44 FIB_MULTIPATH_HASH_FIELD_ALL_MASK;
45 static unsigned int tcp_child_ehash_entries_max = 16 * 1024 * 1024;
46 static unsigned int udp_child_hash_entries_max = UDP_HTABLE_SIZE_MAX;
47 static int tcp_plb_max_rounds = 31;
48 static int tcp_plb_max_cong_thresh = 256;
49 static unsigned int tcp_tw_reuse_delay_max = TCP_PAWS_MSL * MSEC_PER_SEC;
50 static int tcp_ecn_mode_max = 5;
51 static u32 icmp_errors_extension_mask_all =
52 GENMASK_U8(ICMP_ERR_EXT_COUNT - 1, 0);
53
54 static int tcp_min_rcvbuf = 4096;
55
56 /* obsolete */
57 static int sysctl_tcp_low_latency __read_mostly;
58
59 /* Update system visible IP port range */
set_local_port_range(struct net * net,unsigned int low,unsigned int high)60 static void set_local_port_range(struct net *net, unsigned int low, unsigned int high)
61 {
62 bool same_parity = !((low ^ high) & 1);
63
64 if (same_parity && !net->ipv4.ip_local_ports.warned) {
65 net->ipv4.ip_local_ports.warned = true;
66 pr_err_ratelimited("ip_local_port_range: prefer different parity for start/end values.\n");
67 }
68 WRITE_ONCE(net->ipv4.ip_local_ports.range, high << 16 | low);
69 }
70
71 /* Validate changes from /proc interface. */
ipv4_local_port_range(const struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)72 static int ipv4_local_port_range(const struct ctl_table *table, int write,
73 void *buffer, size_t *lenp, loff_t *ppos)
74 {
75 struct net *net = table->data;
76 int ret;
77 int range[2];
78 struct ctl_table tmp = {
79 .data = &range,
80 .maxlen = sizeof(range),
81 .mode = table->mode,
82 .extra1 = &ip_local_port_range_min,
83 .extra2 = &ip_local_port_range_max,
84 };
85
86 inet_get_local_port_range(net, &range[0], &range[1]);
87
88 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
89
90 if (write && ret == 0) {
91 /* Ensure that the upper limit is not smaller than the lower,
92 * and that the lower does not encroach upon the privileged
93 * port limit.
94 */
95 if ((range[1] < range[0]) ||
96 (range[0] < READ_ONCE(net->ipv4.sysctl_ip_prot_sock)))
97 ret = -EINVAL;
98 else
99 set_local_port_range(net, range[0], range[1]);
100 }
101
102 return ret;
103 }
104
105 /* Validate changes from /proc interface. */
ipv4_privileged_ports(const struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)106 static int ipv4_privileged_ports(const struct ctl_table *table, int write,
107 void *buffer, size_t *lenp, loff_t *ppos)
108 {
109 struct net *net = container_of(table->data, struct net,
110 ipv4.sysctl_ip_prot_sock);
111 int ret;
112 int pports;
113 int range[2];
114 struct ctl_table tmp = {
115 .data = &pports,
116 .maxlen = sizeof(pports),
117 .mode = table->mode,
118 .extra1 = &ip_privileged_port_min,
119 .extra2 = &ip_privileged_port_max,
120 };
121
122 pports = READ_ONCE(net->ipv4.sysctl_ip_prot_sock);
123
124 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
125
126 if (write && ret == 0) {
127 inet_get_local_port_range(net, &range[0], &range[1]);
128 /* Ensure that the local port range doesn't overlap with the
129 * privileged port range.
130 */
131 if (range[0] < pports)
132 ret = -EINVAL;
133 else
134 WRITE_ONCE(net->ipv4.sysctl_ip_prot_sock, pports);
135 }
136
137 return ret;
138 }
139
inet_get_ping_group_range_table(const struct ctl_table * table,kgid_t * low,kgid_t * high)140 static void inet_get_ping_group_range_table(const struct ctl_table *table,
141 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(const struct ctl_table * table,kgid_t low,kgid_t high)156 static void set_ping_group_range(const struct ctl_table *table,
157 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(const struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)169 static int ipv4_ping_group_range(const 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 unsigned long 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_doulongvec_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(const struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)204 static int ipv4_fwd_update_priority(const 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_dou8vec_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(const struct ctl_table * ctl,int write,void * buffer,size_t * lenp,loff_t * ppos)220 static int proc_tcp_congestion_control(const 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(const struct ctl_table * ctl,int write,void * buffer,size_t * lenp,loff_t * ppos)240 static int proc_tcp_available_congestion_control(const 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(const struct ctl_table * ctl,int write,void * buffer,size_t * lenp,loff_t * ppos)256 static int proc_allowed_congestion_control(const 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(const struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)293 static int proc_tcp_fastopen_key(const 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(const struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)364 static int proc_tfo_blackhole_detect_timeout(const 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(const struct ctl_table * ctl,int write,void * buffer,size_t * lenp,loff_t * ppos)379 static int proc_tcp_available_ulp(const 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
proc_tcp_ehash_entries(const struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)396 static int proc_tcp_ehash_entries(const struct ctl_table *table, int write,
397 void *buffer, size_t *lenp, loff_t *ppos)
398 {
399 struct net *net = container_of(table->data, struct net,
400 ipv4.sysctl_tcp_child_ehash_entries);
401 struct inet_hashinfo *hinfo = net->ipv4.tcp_death_row.hashinfo;
402 int tcp_ehash_entries;
403 struct ctl_table tbl;
404
405 tcp_ehash_entries = hinfo->ehash_mask + 1;
406
407 /* A negative number indicates that the child netns
408 * shares the global ehash.
409 */
410 if (!net_eq(net, &init_net) && !hinfo->pernet)
411 tcp_ehash_entries *= -1;
412
413 memset(&tbl, 0, sizeof(tbl));
414 tbl.data = &tcp_ehash_entries;
415 tbl.maxlen = sizeof(int);
416
417 return proc_dointvec(&tbl, write, buffer, lenp, ppos);
418 }
419
proc_udp_hash_entries(const struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)420 static int proc_udp_hash_entries(const struct ctl_table *table, int write,
421 void *buffer, size_t *lenp, loff_t *ppos)
422 {
423 struct net *net = container_of(table->data, struct net,
424 ipv4.sysctl_udp_child_hash_entries);
425 int udp_hash_entries;
426 struct ctl_table tbl;
427
428 udp_hash_entries = net->ipv4.udp_table->mask + 1;
429
430 /* A negative number indicates that the child netns
431 * shares the global udp_table.
432 */
433 if (!net_eq(net, &init_net) && net->ipv4.udp_table == &udp_table)
434 udp_hash_entries *= -1;
435
436 memset(&tbl, 0, sizeof(tbl));
437 tbl.data = &udp_hash_entries;
438 tbl.maxlen = sizeof(int);
439
440 return proc_dointvec(&tbl, write, buffer, lenp, ppos);
441 }
442
443 #ifdef CONFIG_IP_ROUTE_MULTIPATH
proc_fib_multipath_hash_policy(const struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)444 static int proc_fib_multipath_hash_policy(const struct ctl_table *table, int write,
445 void *buffer, size_t *lenp,
446 loff_t *ppos)
447 {
448 struct net *net = container_of(table->data, struct net,
449 ipv4.sysctl_fib_multipath_hash_policy);
450 int ret;
451
452 ret = proc_dou8vec_minmax(table, write, buffer, lenp, ppos);
453 if (write && ret == 0)
454 call_netevent_notifiers(NETEVENT_IPV4_MPATH_HASH_UPDATE, net);
455
456 return ret;
457 }
458
proc_fib_multipath_hash_fields(const struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)459 static int proc_fib_multipath_hash_fields(const struct ctl_table *table, int write,
460 void *buffer, size_t *lenp,
461 loff_t *ppos)
462 {
463 struct net *net;
464 int ret;
465
466 net = container_of(table->data, struct net,
467 ipv4.sysctl_fib_multipath_hash_fields);
468 ret = proc_douintvec_minmax(table, write, buffer, lenp, ppos);
469 if (write && ret == 0)
470 call_netevent_notifiers(NETEVENT_IPV4_MPATH_HASH_UPDATE, net);
471
472 return ret;
473 }
474
475 static u32 proc_fib_multipath_hash_rand_seed __ro_after_init;
476
proc_fib_multipath_hash_init_rand_seed(void)477 static void proc_fib_multipath_hash_init_rand_seed(void)
478 {
479 get_random_bytes(&proc_fib_multipath_hash_rand_seed,
480 sizeof(proc_fib_multipath_hash_rand_seed));
481 }
482
proc_fib_multipath_hash_set_seed(struct net * net,u32 user_seed)483 static void proc_fib_multipath_hash_set_seed(struct net *net, u32 user_seed)
484 {
485 struct sysctl_fib_multipath_hash_seed new = {
486 .user_seed = user_seed,
487 .mp_seed = (user_seed ? user_seed :
488 proc_fib_multipath_hash_rand_seed),
489 };
490
491 WRITE_ONCE(net->ipv4.sysctl_fib_multipath_hash_seed.user_seed, new.user_seed);
492 WRITE_ONCE(net->ipv4.sysctl_fib_multipath_hash_seed.mp_seed, new.mp_seed);
493 }
494
proc_fib_multipath_hash_seed(const struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)495 static int proc_fib_multipath_hash_seed(const struct ctl_table *table, int write,
496 void *buffer, size_t *lenp,
497 loff_t *ppos)
498 {
499 struct sysctl_fib_multipath_hash_seed *mphs;
500 struct net *net = table->data;
501 struct ctl_table tmp;
502 u32 user_seed;
503 int ret;
504
505 mphs = &net->ipv4.sysctl_fib_multipath_hash_seed;
506 user_seed = READ_ONCE(mphs->user_seed);
507
508 tmp = *table;
509 tmp.data = &user_seed;
510
511 ret = proc_douintvec_minmax(&tmp, write, buffer, lenp, ppos);
512
513 if (write && ret == 0) {
514 proc_fib_multipath_hash_set_seed(net, user_seed);
515 call_netevent_notifiers(NETEVENT_IPV4_MPATH_HASH_UPDATE, net);
516 }
517
518 return ret;
519 }
520 #else
521
proc_fib_multipath_hash_init_rand_seed(void)522 static void proc_fib_multipath_hash_init_rand_seed(void)
523 {
524 }
525
proc_fib_multipath_hash_set_seed(struct net * net,u32 user_seed)526 static void proc_fib_multipath_hash_set_seed(struct net *net, u32 user_seed)
527 {
528 }
529
530 #endif
531
532 static struct ctl_table ipv4_table[] = {
533 {
534 .procname = "tcp_max_orphans",
535 .data = &sysctl_tcp_max_orphans,
536 .maxlen = sizeof(int),
537 .mode = 0644,
538 .proc_handler = proc_dointvec
539 },
540 {
541 .procname = "inet_peer_threshold",
542 .data = &inet_peer_threshold,
543 .maxlen = sizeof(int),
544 .mode = 0644,
545 .proc_handler = proc_dointvec
546 },
547 {
548 .procname = "inet_peer_minttl",
549 .data = &inet_peer_minttl,
550 .maxlen = sizeof(int),
551 .mode = 0644,
552 .proc_handler = proc_dointvec_jiffies,
553 },
554 {
555 .procname = "inet_peer_maxttl",
556 .data = &inet_peer_maxttl,
557 .maxlen = sizeof(int),
558 .mode = 0644,
559 .proc_handler = proc_dointvec_jiffies,
560 },
561 {
562 .procname = "tcp_mem",
563 .maxlen = sizeof(sysctl_tcp_mem),
564 .data = &sysctl_tcp_mem,
565 .mode = 0644,
566 .proc_handler = proc_doulongvec_minmax,
567 },
568 {
569 .procname = "tcp_low_latency",
570 .data = &sysctl_tcp_low_latency,
571 .maxlen = sizeof(int),
572 .mode = 0644,
573 .proc_handler = proc_dointvec
574 },
575 #ifdef CONFIG_NETLABEL
576 {
577 .procname = "cipso_cache_enable",
578 .data = &cipso_v4_cache_enabled,
579 .maxlen = sizeof(int),
580 .mode = 0644,
581 .proc_handler = proc_dointvec,
582 },
583 {
584 .procname = "cipso_cache_bucket_size",
585 .data = &cipso_v4_cache_bucketsize,
586 .maxlen = sizeof(int),
587 .mode = 0644,
588 .proc_handler = proc_dointvec,
589 },
590 {
591 .procname = "cipso_rbm_optfmt",
592 .data = &cipso_v4_rbm_optfmt,
593 .maxlen = sizeof(int),
594 .mode = 0644,
595 .proc_handler = proc_dointvec,
596 },
597 {
598 .procname = "cipso_rbm_strictvalid",
599 .data = &cipso_v4_rbm_strictvalid,
600 .maxlen = sizeof(int),
601 .mode = 0644,
602 .proc_handler = proc_dointvec,
603 },
604 #endif /* CONFIG_NETLABEL */
605 {
606 .procname = "tcp_available_ulp",
607 .maxlen = TCP_ULP_BUF_MAX,
608 .mode = 0444,
609 .proc_handler = proc_tcp_available_ulp,
610 },
611 {
612 .procname = "udp_mem",
613 .data = &sysctl_udp_mem,
614 .maxlen = sizeof(sysctl_udp_mem),
615 .mode = 0644,
616 .proc_handler = proc_doulongvec_minmax,
617 },
618 {
619 .procname = "fib_sync_mem",
620 .data = &sysctl_fib_sync_mem,
621 .maxlen = sizeof(sysctl_fib_sync_mem),
622 .mode = 0644,
623 .proc_handler = proc_douintvec_minmax,
624 .extra1 = &sysctl_fib_sync_mem_min,
625 .extra2 = &sysctl_fib_sync_mem_max,
626 },
627 };
628
629 static const struct ctl_table ipv4_net_table[] = {
630 {
631 .procname = "tcp_max_tw_buckets",
632 .data = &init_net.ipv4.tcp_death_row.sysctl_max_tw_buckets,
633 .maxlen = sizeof(int),
634 .mode = 0644,
635 .proc_handler = proc_dointvec
636 },
637 {
638 .procname = "icmp_echo_ignore_all",
639 .data = &init_net.ipv4.sysctl_icmp_echo_ignore_all,
640 .maxlen = sizeof(u8),
641 .mode = 0644,
642 .proc_handler = proc_dou8vec_minmax,
643 .extra1 = SYSCTL_ZERO,
644 .extra2 = SYSCTL_ONE
645 },
646 {
647 .procname = "icmp_echo_enable_probe",
648 .data = &init_net.ipv4.sysctl_icmp_echo_enable_probe,
649 .maxlen = sizeof(u8),
650 .mode = 0644,
651 .proc_handler = proc_dou8vec_minmax,
652 .extra1 = SYSCTL_ZERO,
653 .extra2 = SYSCTL_ONE
654 },
655 {
656 .procname = "icmp_echo_ignore_broadcasts",
657 .data = &init_net.ipv4.sysctl_icmp_echo_ignore_broadcasts,
658 .maxlen = sizeof(u8),
659 .mode = 0644,
660 .proc_handler = proc_dou8vec_minmax,
661 .extra1 = SYSCTL_ZERO,
662 .extra2 = SYSCTL_ONE
663 },
664 {
665 .procname = "icmp_ignore_bogus_error_responses",
666 .data = &init_net.ipv4.sysctl_icmp_ignore_bogus_error_responses,
667 .maxlen = sizeof(u8),
668 .mode = 0644,
669 .proc_handler = proc_dou8vec_minmax,
670 .extra1 = SYSCTL_ZERO,
671 .extra2 = SYSCTL_ONE
672 },
673 {
674 .procname = "icmp_errors_use_inbound_ifaddr",
675 .data = &init_net.ipv4.sysctl_icmp_errors_use_inbound_ifaddr,
676 .maxlen = sizeof(u8),
677 .mode = 0644,
678 .proc_handler = proc_dou8vec_minmax,
679 .extra1 = SYSCTL_ZERO,
680 .extra2 = SYSCTL_ONE
681 },
682 {
683 .procname = "icmp_errors_extension_mask",
684 .data = &init_net.ipv4.sysctl_icmp_errors_extension_mask,
685 .maxlen = sizeof(u8),
686 .mode = 0644,
687 .proc_handler = proc_dou8vec_minmax,
688 .extra1 = SYSCTL_ZERO,
689 .extra2 = &icmp_errors_extension_mask_all,
690 },
691 {
692 .procname = "icmp_ratelimit",
693 .data = &init_net.ipv4.sysctl_icmp_ratelimit,
694 .maxlen = sizeof(int),
695 .mode = 0644,
696 .proc_handler = proc_dointvec_ms_jiffies,
697 },
698 {
699 .procname = "icmp_ratemask",
700 .data = &init_net.ipv4.sysctl_icmp_ratemask,
701 .maxlen = sizeof(int),
702 .mode = 0644,
703 .proc_handler = proc_dointvec
704 },
705 {
706 .procname = "icmp_msgs_per_sec",
707 .data = &init_net.ipv4.sysctl_icmp_msgs_per_sec,
708 .maxlen = sizeof(int),
709 .mode = 0644,
710 .proc_handler = proc_dointvec_minmax,
711 .extra1 = SYSCTL_ZERO,
712 },
713 {
714 .procname = "icmp_msgs_burst",
715 .data = &init_net.ipv4.sysctl_icmp_msgs_burst,
716 .maxlen = sizeof(int),
717 .mode = 0644,
718 .proc_handler = proc_dointvec_minmax,
719 .extra1 = SYSCTL_ZERO,
720 },
721 {
722 .procname = "ping_group_range",
723 .data = &init_net.ipv4.ping_group_range.range,
724 .maxlen = sizeof(gid_t)*2,
725 .mode = 0644,
726 .proc_handler = ipv4_ping_group_range,
727 },
728 #ifdef CONFIG_NET_L3_MASTER_DEV
729 {
730 .procname = "raw_l3mdev_accept",
731 .data = &init_net.ipv4.sysctl_raw_l3mdev_accept,
732 .maxlen = sizeof(u8),
733 .mode = 0644,
734 .proc_handler = proc_dou8vec_minmax,
735 .extra1 = SYSCTL_ZERO,
736 .extra2 = SYSCTL_ONE,
737 },
738 #endif
739 {
740 .procname = "tcp_ecn",
741 .data = &init_net.ipv4.sysctl_tcp_ecn,
742 .maxlen = sizeof(u8),
743 .mode = 0644,
744 .proc_handler = proc_dou8vec_minmax,
745 .extra1 = SYSCTL_ZERO,
746 .extra2 = &tcp_ecn_mode_max,
747 },
748 {
749 .procname = "tcp_ecn_option",
750 .data = &init_net.ipv4.sysctl_tcp_ecn_option,
751 .maxlen = sizeof(u8),
752 .mode = 0644,
753 .proc_handler = proc_dou8vec_minmax,
754 .extra1 = SYSCTL_ZERO,
755 .extra2 = SYSCTL_THREE,
756 },
757 {
758 .procname = "tcp_ecn_option_beacon",
759 .data = &init_net.ipv4.sysctl_tcp_ecn_option_beacon,
760 .maxlen = sizeof(u8),
761 .mode = 0644,
762 .proc_handler = proc_dou8vec_minmax,
763 .extra1 = SYSCTL_ZERO,
764 .extra2 = SYSCTL_THREE,
765 },
766 {
767 .procname = "tcp_ecn_fallback",
768 .data = &init_net.ipv4.sysctl_tcp_ecn_fallback,
769 .maxlen = sizeof(u8),
770 .mode = 0644,
771 .proc_handler = proc_dou8vec_minmax,
772 .extra1 = SYSCTL_ZERO,
773 .extra2 = SYSCTL_ONE,
774 },
775 {
776 .procname = "ip_dynaddr",
777 .data = &init_net.ipv4.sysctl_ip_dynaddr,
778 .maxlen = sizeof(u8),
779 .mode = 0644,
780 .proc_handler = proc_dou8vec_minmax,
781 },
782 {
783 .procname = "ip_early_demux",
784 .data = &init_net.ipv4.sysctl_ip_early_demux,
785 .maxlen = sizeof(u8),
786 .mode = 0644,
787 .proc_handler = proc_dou8vec_minmax,
788 },
789 {
790 .procname = "udp_early_demux",
791 .data = &init_net.ipv4.sysctl_udp_early_demux,
792 .maxlen = sizeof(u8),
793 .mode = 0644,
794 .proc_handler = proc_dou8vec_minmax,
795 },
796 {
797 .procname = "tcp_early_demux",
798 .data = &init_net.ipv4.sysctl_tcp_early_demux,
799 .maxlen = sizeof(u8),
800 .mode = 0644,
801 .proc_handler = proc_dou8vec_minmax,
802 },
803 {
804 .procname = "nexthop_compat_mode",
805 .data = &init_net.ipv4.sysctl_nexthop_compat_mode,
806 .maxlen = sizeof(u8),
807 .mode = 0644,
808 .proc_handler = proc_dou8vec_minmax,
809 .extra1 = SYSCTL_ZERO,
810 .extra2 = SYSCTL_ONE,
811 },
812 {
813 .procname = "ip_default_ttl",
814 .data = &init_net.ipv4.sysctl_ip_default_ttl,
815 .maxlen = sizeof(u8),
816 .mode = 0644,
817 .proc_handler = proc_dou8vec_minmax,
818 .extra1 = &ip_ttl_min,
819 .extra2 = &ip_ttl_max,
820 },
821 {
822 .procname = "ip_local_port_range",
823 .maxlen = 0,
824 .data = &init_net,
825 .mode = 0644,
826 .proc_handler = ipv4_local_port_range,
827 },
828 {
829 .procname = "ip_local_port_step_width",
830 .maxlen = sizeof(u32),
831 .data = &init_net.ipv4.sysctl_ip_local_port_step_width,
832 .mode = 0644,
833 .proc_handler = proc_douintvec,
834 },
835 {
836 .procname = "ip_local_reserved_ports",
837 .data = &init_net.ipv4.sysctl_local_reserved_ports,
838 .maxlen = 65536,
839 .mode = 0644,
840 .proc_handler = proc_do_large_bitmap,
841 },
842 {
843 .procname = "ip_no_pmtu_disc",
844 .data = &init_net.ipv4.sysctl_ip_no_pmtu_disc,
845 .maxlen = sizeof(u8),
846 .mode = 0644,
847 .proc_handler = proc_dou8vec_minmax,
848 },
849 {
850 .procname = "ip_forward_use_pmtu",
851 .data = &init_net.ipv4.sysctl_ip_fwd_use_pmtu,
852 .maxlen = sizeof(u8),
853 .mode = 0644,
854 .proc_handler = proc_dou8vec_minmax,
855 },
856 {
857 .procname = "ip_forward_update_priority",
858 .data = &init_net.ipv4.sysctl_ip_fwd_update_priority,
859 .maxlen = sizeof(u8),
860 .mode = 0644,
861 .proc_handler = ipv4_fwd_update_priority,
862 .extra1 = SYSCTL_ZERO,
863 .extra2 = SYSCTL_ONE,
864 },
865 {
866 .procname = "ip_nonlocal_bind",
867 .data = &init_net.ipv4.sysctl_ip_nonlocal_bind,
868 .maxlen = sizeof(u8),
869 .mode = 0644,
870 .proc_handler = proc_dou8vec_minmax,
871 },
872 {
873 .procname = "ip_autobind_reuse",
874 .data = &init_net.ipv4.sysctl_ip_autobind_reuse,
875 .maxlen = sizeof(u8),
876 .mode = 0644,
877 .proc_handler = proc_dou8vec_minmax,
878 .extra1 = SYSCTL_ZERO,
879 .extra2 = SYSCTL_ONE,
880 },
881 {
882 .procname = "fwmark_reflect",
883 .data = &init_net.ipv4.sysctl_fwmark_reflect,
884 .maxlen = sizeof(u8),
885 .mode = 0644,
886 .proc_handler = proc_dou8vec_minmax,
887 },
888 {
889 .procname = "tcp_fwmark_accept",
890 .data = &init_net.ipv4.sysctl_tcp_fwmark_accept,
891 .maxlen = sizeof(u8),
892 .mode = 0644,
893 .proc_handler = proc_dou8vec_minmax,
894 },
895 #ifdef CONFIG_NET_L3_MASTER_DEV
896 {
897 .procname = "tcp_l3mdev_accept",
898 .data = &init_net.ipv4.sysctl_tcp_l3mdev_accept,
899 .maxlen = sizeof(u8),
900 .mode = 0644,
901 .proc_handler = proc_dou8vec_minmax,
902 .extra1 = SYSCTL_ZERO,
903 .extra2 = SYSCTL_ONE,
904 },
905 #endif
906 {
907 .procname = "tcp_mtu_probing",
908 .data = &init_net.ipv4.sysctl_tcp_mtu_probing,
909 .maxlen = sizeof(u8),
910 .mode = 0644,
911 .proc_handler = proc_dou8vec_minmax,
912 },
913 {
914 .procname = "tcp_base_mss",
915 .data = &init_net.ipv4.sysctl_tcp_base_mss,
916 .maxlen = sizeof(int),
917 .mode = 0644,
918 .proc_handler = proc_dointvec,
919 },
920 {
921 .procname = "tcp_min_snd_mss",
922 .data = &init_net.ipv4.sysctl_tcp_min_snd_mss,
923 .maxlen = sizeof(int),
924 .mode = 0644,
925 .proc_handler = proc_dointvec_minmax,
926 .extra1 = &tcp_min_snd_mss_min,
927 .extra2 = &tcp_min_snd_mss_max,
928 },
929 {
930 .procname = "tcp_mtu_probe_floor",
931 .data = &init_net.ipv4.sysctl_tcp_mtu_probe_floor,
932 .maxlen = sizeof(int),
933 .mode = 0644,
934 .proc_handler = proc_dointvec_minmax,
935 .extra1 = &tcp_min_snd_mss_min,
936 .extra2 = &tcp_min_snd_mss_max,
937 },
938 {
939 .procname = "tcp_probe_threshold",
940 .data = &init_net.ipv4.sysctl_tcp_probe_threshold,
941 .maxlen = sizeof(int),
942 .mode = 0644,
943 .proc_handler = proc_dointvec,
944 },
945 {
946 .procname = "tcp_probe_interval",
947 .data = &init_net.ipv4.sysctl_tcp_probe_interval,
948 .maxlen = sizeof(u32),
949 .mode = 0644,
950 .proc_handler = proc_douintvec_minmax,
951 .extra2 = &u32_max_div_HZ,
952 },
953 {
954 .procname = "igmp_link_local_mcast_reports",
955 .data = &init_net.ipv4.sysctl_igmp_llm_reports,
956 .maxlen = sizeof(u8),
957 .mode = 0644,
958 .proc_handler = proc_dou8vec_minmax,
959 },
960 {
961 .procname = "igmp_max_memberships",
962 .data = &init_net.ipv4.sysctl_igmp_max_memberships,
963 .maxlen = sizeof(int),
964 .mode = 0644,
965 .proc_handler = proc_dointvec
966 },
967 {
968 .procname = "igmp_max_msf",
969 .data = &init_net.ipv4.sysctl_igmp_max_msf,
970 .maxlen = sizeof(int),
971 .mode = 0644,
972 .proc_handler = proc_dointvec
973 },
974 #ifdef CONFIG_IP_MULTICAST
975 {
976 .procname = "igmp_qrv",
977 .data = &init_net.ipv4.sysctl_igmp_qrv,
978 .maxlen = sizeof(int),
979 .mode = 0644,
980 .proc_handler = proc_dointvec_minmax,
981 .extra1 = SYSCTL_ONE
982 },
983 #endif
984 {
985 .procname = "tcp_congestion_control",
986 .data = &init_net.ipv4.tcp_congestion_control,
987 .mode = 0644,
988 .maxlen = TCP_CA_NAME_MAX,
989 .proc_handler = proc_tcp_congestion_control,
990 },
991 {
992 .procname = "tcp_available_congestion_control",
993 .maxlen = TCP_CA_BUF_MAX,
994 .mode = 0444,
995 .proc_handler = proc_tcp_available_congestion_control,
996 },
997 {
998 .procname = "tcp_allowed_congestion_control",
999 .maxlen = TCP_CA_BUF_MAX,
1000 .mode = 0644,
1001 .proc_handler = proc_allowed_congestion_control,
1002 },
1003 {
1004 .procname = "tcp_keepalive_time",
1005 .data = &init_net.ipv4.sysctl_tcp_keepalive_time,
1006 .maxlen = sizeof(int),
1007 .mode = 0644,
1008 .proc_handler = proc_dointvec_jiffies,
1009 },
1010 {
1011 .procname = "tcp_keepalive_probes",
1012 .data = &init_net.ipv4.sysctl_tcp_keepalive_probes,
1013 .maxlen = sizeof(u8),
1014 .mode = 0644,
1015 .proc_handler = proc_dou8vec_minmax,
1016 },
1017 {
1018 .procname = "tcp_keepalive_intvl",
1019 .data = &init_net.ipv4.sysctl_tcp_keepalive_intvl,
1020 .maxlen = sizeof(int),
1021 .mode = 0644,
1022 .proc_handler = proc_dointvec_jiffies,
1023 },
1024 {
1025 .procname = "tcp_syn_retries",
1026 .data = &init_net.ipv4.sysctl_tcp_syn_retries,
1027 .maxlen = sizeof(u8),
1028 .mode = 0644,
1029 .proc_handler = proc_dou8vec_minmax,
1030 .extra1 = &tcp_syn_retries_min,
1031 .extra2 = &tcp_syn_retries_max
1032 },
1033 {
1034 .procname = "tcp_synack_retries",
1035 .data = &init_net.ipv4.sysctl_tcp_synack_retries,
1036 .maxlen = sizeof(u8),
1037 .mode = 0644,
1038 .proc_handler = proc_dou8vec_minmax,
1039 },
1040 #ifdef CONFIG_SYN_COOKIES
1041 {
1042 .procname = "tcp_syncookies",
1043 .data = &init_net.ipv4.sysctl_tcp_syncookies,
1044 .maxlen = sizeof(u8),
1045 .mode = 0644,
1046 .proc_handler = proc_dou8vec_minmax,
1047 },
1048 #endif
1049 {
1050 .procname = "tcp_migrate_req",
1051 .data = &init_net.ipv4.sysctl_tcp_migrate_req,
1052 .maxlen = sizeof(u8),
1053 .mode = 0644,
1054 .proc_handler = proc_dou8vec_minmax,
1055 .extra1 = SYSCTL_ZERO,
1056 .extra2 = SYSCTL_ONE
1057 },
1058 {
1059 .procname = "tcp_reordering",
1060 .data = &init_net.ipv4.sysctl_tcp_reordering,
1061 .maxlen = sizeof(int),
1062 .mode = 0644,
1063 .proc_handler = proc_dointvec_minmax,
1064 .extra1 = SYSCTL_ONE,
1065 .extra2 = &init_net.ipv4.sysctl_tcp_max_reordering,
1066 },
1067 {
1068 .procname = "tcp_retries1",
1069 .data = &init_net.ipv4.sysctl_tcp_retries1,
1070 .maxlen = sizeof(u8),
1071 .mode = 0644,
1072 .proc_handler = proc_dou8vec_minmax,
1073 .extra2 = &tcp_retr1_max
1074 },
1075 {
1076 .procname = "tcp_retries2",
1077 .data = &init_net.ipv4.sysctl_tcp_retries2,
1078 .maxlen = sizeof(u8),
1079 .mode = 0644,
1080 .proc_handler = proc_dou8vec_minmax,
1081 },
1082 {
1083 .procname = "tcp_orphan_retries",
1084 .data = &init_net.ipv4.sysctl_tcp_orphan_retries,
1085 .maxlen = sizeof(u8),
1086 .mode = 0644,
1087 .proc_handler = proc_dou8vec_minmax,
1088 },
1089 {
1090 .procname = "tcp_fin_timeout",
1091 .data = &init_net.ipv4.sysctl_tcp_fin_timeout,
1092 .maxlen = sizeof(int),
1093 .mode = 0644,
1094 .proc_handler = proc_dointvec_jiffies,
1095 },
1096 {
1097 .procname = "tcp_notsent_lowat",
1098 .data = &init_net.ipv4.sysctl_tcp_notsent_lowat,
1099 .maxlen = sizeof(unsigned int),
1100 .mode = 0644,
1101 .proc_handler = proc_douintvec,
1102 },
1103 {
1104 .procname = "tcp_tw_reuse",
1105 .data = &init_net.ipv4.sysctl_tcp_tw_reuse,
1106 .maxlen = sizeof(u8),
1107 .mode = 0644,
1108 .proc_handler = proc_dou8vec_minmax,
1109 .extra1 = SYSCTL_ZERO,
1110 .extra2 = SYSCTL_TWO,
1111 },
1112 {
1113 .procname = "tcp_tw_reuse_delay",
1114 .data = &init_net.ipv4.sysctl_tcp_tw_reuse_delay,
1115 .maxlen = sizeof(unsigned int),
1116 .mode = 0644,
1117 .proc_handler = proc_douintvec_minmax,
1118 .extra1 = SYSCTL_ONE,
1119 .extra2 = &tcp_tw_reuse_delay_max,
1120 },
1121 {
1122 .procname = "tcp_max_syn_backlog",
1123 .data = &init_net.ipv4.sysctl_max_syn_backlog,
1124 .maxlen = sizeof(int),
1125 .mode = 0644,
1126 .proc_handler = proc_dointvec
1127 },
1128 {
1129 .procname = "tcp_fastopen",
1130 .data = &init_net.ipv4.sysctl_tcp_fastopen,
1131 .maxlen = sizeof(int),
1132 .mode = 0644,
1133 .proc_handler = proc_dointvec,
1134 },
1135 {
1136 .procname = "tcp_fastopen_key",
1137 .mode = 0600,
1138 .data = &init_net.ipv4.sysctl_tcp_fastopen,
1139 /* maxlen to print the list of keys in hex (*2), with dashes
1140 * separating doublewords and a comma in between keys.
1141 */
1142 .maxlen = ((TCP_FASTOPEN_KEY_LENGTH *
1143 2 * TCP_FASTOPEN_KEY_MAX) +
1144 (TCP_FASTOPEN_KEY_MAX * 5)),
1145 .proc_handler = proc_tcp_fastopen_key,
1146 },
1147 {
1148 .procname = "tcp_fastopen_blackhole_timeout_sec",
1149 .data = &init_net.ipv4.sysctl_tcp_fastopen_blackhole_timeout,
1150 .maxlen = sizeof(int),
1151 .mode = 0644,
1152 .proc_handler = proc_tfo_blackhole_detect_timeout,
1153 .extra1 = SYSCTL_ZERO,
1154 },
1155 #ifdef CONFIG_IP_ROUTE_MULTIPATH
1156 {
1157 .procname = "fib_multipath_use_neigh",
1158 .data = &init_net.ipv4.sysctl_fib_multipath_use_neigh,
1159 .maxlen = sizeof(u8),
1160 .mode = 0644,
1161 .proc_handler = proc_dou8vec_minmax,
1162 .extra1 = SYSCTL_ZERO,
1163 .extra2 = SYSCTL_ONE,
1164 },
1165 {
1166 .procname = "fib_multipath_hash_policy",
1167 .data = &init_net.ipv4.sysctl_fib_multipath_hash_policy,
1168 .maxlen = sizeof(u8),
1169 .mode = 0644,
1170 .proc_handler = proc_fib_multipath_hash_policy,
1171 .extra1 = SYSCTL_ZERO,
1172 .extra2 = SYSCTL_THREE,
1173 },
1174 {
1175 .procname = "fib_multipath_hash_fields",
1176 .data = &init_net.ipv4.sysctl_fib_multipath_hash_fields,
1177 .maxlen = sizeof(u32),
1178 .mode = 0644,
1179 .proc_handler = proc_fib_multipath_hash_fields,
1180 .extra1 = SYSCTL_ONE,
1181 .extra2 = &fib_multipath_hash_fields_all_mask,
1182 },
1183 {
1184 .procname = "fib_multipath_hash_seed",
1185 .data = &init_net,
1186 .maxlen = sizeof(u32),
1187 .mode = 0644,
1188 .proc_handler = proc_fib_multipath_hash_seed,
1189 },
1190 #endif
1191 {
1192 .procname = "ip_unprivileged_port_start",
1193 .maxlen = sizeof(int),
1194 .data = &init_net.ipv4.sysctl_ip_prot_sock,
1195 .mode = 0644,
1196 .proc_handler = ipv4_privileged_ports,
1197 },
1198 #ifdef CONFIG_NET_L3_MASTER_DEV
1199 {
1200 .procname = "udp_l3mdev_accept",
1201 .data = &init_net.ipv4.sysctl_udp_l3mdev_accept,
1202 .maxlen = sizeof(u8),
1203 .mode = 0644,
1204 .proc_handler = proc_dou8vec_minmax,
1205 .extra1 = SYSCTL_ZERO,
1206 .extra2 = SYSCTL_ONE,
1207 },
1208 #endif
1209 {
1210 .procname = "tcp_sack",
1211 .data = &init_net.ipv4.sysctl_tcp_sack,
1212 .maxlen = sizeof(u8),
1213 .mode = 0644,
1214 .proc_handler = proc_dou8vec_minmax,
1215 },
1216 {
1217 .procname = "tcp_window_scaling",
1218 .data = &init_net.ipv4.sysctl_tcp_window_scaling,
1219 .maxlen = sizeof(u8),
1220 .mode = 0644,
1221 .proc_handler = proc_dou8vec_minmax,
1222 },
1223 {
1224 .procname = "tcp_timestamps",
1225 .data = &init_net.ipv4.sysctl_tcp_timestamps,
1226 .maxlen = sizeof(u8),
1227 .mode = 0644,
1228 .proc_handler = proc_dou8vec_minmax,
1229 },
1230 {
1231 .procname = "tcp_early_retrans",
1232 .data = &init_net.ipv4.sysctl_tcp_early_retrans,
1233 .maxlen = sizeof(u8),
1234 .mode = 0644,
1235 .proc_handler = proc_dou8vec_minmax,
1236 .extra1 = SYSCTL_ZERO,
1237 .extra2 = SYSCTL_FOUR,
1238 },
1239 {
1240 .procname = "tcp_recovery",
1241 .data = &init_net.ipv4.sysctl_tcp_recovery,
1242 .maxlen = sizeof(u8),
1243 .mode = 0644,
1244 .proc_handler = proc_dou8vec_minmax,
1245 },
1246 {
1247 .procname = "tcp_thin_linear_timeouts",
1248 .data = &init_net.ipv4.sysctl_tcp_thin_linear_timeouts,
1249 .maxlen = sizeof(u8),
1250 .mode = 0644,
1251 .proc_handler = proc_dou8vec_minmax,
1252 },
1253 {
1254 .procname = "tcp_slow_start_after_idle",
1255 .data = &init_net.ipv4.sysctl_tcp_slow_start_after_idle,
1256 .maxlen = sizeof(u8),
1257 .mode = 0644,
1258 .proc_handler = proc_dou8vec_minmax,
1259 },
1260 {
1261 .procname = "tcp_retrans_collapse",
1262 .data = &init_net.ipv4.sysctl_tcp_retrans_collapse,
1263 .maxlen = sizeof(u8),
1264 .mode = 0644,
1265 .proc_handler = proc_dou8vec_minmax,
1266 },
1267 {
1268 .procname = "tcp_stdurg",
1269 .data = &init_net.ipv4.sysctl_tcp_stdurg,
1270 .maxlen = sizeof(u8),
1271 .mode = 0644,
1272 .proc_handler = proc_dou8vec_minmax,
1273 },
1274 {
1275 .procname = "tcp_rfc1337",
1276 .data = &init_net.ipv4.sysctl_tcp_rfc1337,
1277 .maxlen = sizeof(u8),
1278 .mode = 0644,
1279 .proc_handler = proc_dou8vec_minmax,
1280 },
1281 {
1282 .procname = "tcp_abort_on_overflow",
1283 .data = &init_net.ipv4.sysctl_tcp_abort_on_overflow,
1284 .maxlen = sizeof(u8),
1285 .mode = 0644,
1286 .proc_handler = proc_dou8vec_minmax,
1287 },
1288 {
1289 .procname = "tcp_fack",
1290 .data = &init_net.ipv4.sysctl_tcp_fack,
1291 .maxlen = sizeof(u8),
1292 .mode = 0644,
1293 .proc_handler = proc_dou8vec_minmax,
1294 },
1295 {
1296 .procname = "tcp_max_reordering",
1297 .data = &init_net.ipv4.sysctl_tcp_max_reordering,
1298 .maxlen = sizeof(int),
1299 .mode = 0644,
1300 .proc_handler = proc_dointvec_minmax,
1301 .extra1 = SYSCTL_ONE,
1302 },
1303 {
1304 .procname = "tcp_dsack",
1305 .data = &init_net.ipv4.sysctl_tcp_dsack,
1306 .maxlen = sizeof(u8),
1307 .mode = 0644,
1308 .proc_handler = proc_dou8vec_minmax,
1309 },
1310 {
1311 .procname = "tcp_app_win",
1312 .data = &init_net.ipv4.sysctl_tcp_app_win,
1313 .maxlen = sizeof(u8),
1314 .mode = 0644,
1315 .proc_handler = proc_dou8vec_minmax,
1316 .extra1 = SYSCTL_ZERO,
1317 .extra2 = &tcp_app_win_max,
1318 },
1319 {
1320 .procname = "tcp_adv_win_scale",
1321 .data = &init_net.ipv4.sysctl_tcp_adv_win_scale,
1322 .maxlen = sizeof(int),
1323 .mode = 0644,
1324 .proc_handler = proc_dointvec_minmax,
1325 .extra1 = &tcp_adv_win_scale_min,
1326 .extra2 = &tcp_adv_win_scale_max,
1327 },
1328 {
1329 .procname = "tcp_frto",
1330 .data = &init_net.ipv4.sysctl_tcp_frto,
1331 .maxlen = sizeof(u8),
1332 .mode = 0644,
1333 .proc_handler = proc_dou8vec_minmax,
1334 },
1335 {
1336 .procname = "tcp_no_metrics_save",
1337 .data = &init_net.ipv4.sysctl_tcp_nometrics_save,
1338 .maxlen = sizeof(u8),
1339 .mode = 0644,
1340 .proc_handler = proc_dou8vec_minmax,
1341 },
1342 {
1343 .procname = "tcp_no_ssthresh_metrics_save",
1344 .data = &init_net.ipv4.sysctl_tcp_no_ssthresh_metrics_save,
1345 .maxlen = sizeof(u8),
1346 .mode = 0644,
1347 .proc_handler = proc_dou8vec_minmax,
1348 .extra1 = SYSCTL_ZERO,
1349 .extra2 = SYSCTL_ONE,
1350 },
1351 {
1352 .procname = "tcp_moderate_rcvbuf",
1353 .data = &init_net.ipv4.sysctl_tcp_moderate_rcvbuf,
1354 .maxlen = sizeof(u8),
1355 .mode = 0644,
1356 .proc_handler = proc_dou8vec_minmax,
1357 },
1358 {
1359 .procname = "tcp_rcvbuf_low_rtt",
1360 .data = &init_net.ipv4.sysctl_tcp_rcvbuf_low_rtt,
1361 .maxlen = sizeof(int),
1362 .mode = 0644,
1363 .proc_handler = proc_dointvec_minmax,
1364 .extra1 = SYSCTL_ZERO,
1365 .extra2 = SYSCTL_INT_MAX,
1366 },
1367 {
1368 .procname = "tcp_tso_win_divisor",
1369 .data = &init_net.ipv4.sysctl_tcp_tso_win_divisor,
1370 .maxlen = sizeof(u8),
1371 .mode = 0644,
1372 .proc_handler = proc_dou8vec_minmax,
1373 },
1374 {
1375 .procname = "tcp_workaround_signed_windows",
1376 .data = &init_net.ipv4.sysctl_tcp_workaround_signed_windows,
1377 .maxlen = sizeof(u8),
1378 .mode = 0644,
1379 .proc_handler = proc_dou8vec_minmax,
1380 },
1381 {
1382 .procname = "tcp_limit_output_bytes",
1383 .data = &init_net.ipv4.sysctl_tcp_limit_output_bytes,
1384 .maxlen = sizeof(int),
1385 .mode = 0644,
1386 .proc_handler = proc_dointvec
1387 },
1388 {
1389 .procname = "tcp_challenge_ack_limit",
1390 .data = &init_net.ipv4.sysctl_tcp_challenge_ack_limit,
1391 .maxlen = sizeof(int),
1392 .mode = 0644,
1393 .proc_handler = proc_dointvec
1394 },
1395 {
1396 .procname = "tcp_min_tso_segs",
1397 .data = &init_net.ipv4.sysctl_tcp_min_tso_segs,
1398 .maxlen = sizeof(u8),
1399 .mode = 0644,
1400 .proc_handler = proc_dou8vec_minmax,
1401 .extra1 = SYSCTL_ONE,
1402 },
1403 {
1404 .procname = "tcp_tso_rtt_log",
1405 .data = &init_net.ipv4.sysctl_tcp_tso_rtt_log,
1406 .maxlen = sizeof(u8),
1407 .mode = 0644,
1408 .proc_handler = proc_dou8vec_minmax,
1409 },
1410 {
1411 .procname = "tcp_min_rtt_wlen",
1412 .data = &init_net.ipv4.sysctl_tcp_min_rtt_wlen,
1413 .maxlen = sizeof(int),
1414 .mode = 0644,
1415 .proc_handler = proc_dointvec_minmax,
1416 .extra1 = SYSCTL_ZERO,
1417 .extra2 = &one_day_secs
1418 },
1419 {
1420 .procname = "tcp_autocorking",
1421 .data = &init_net.ipv4.sysctl_tcp_autocorking,
1422 .maxlen = sizeof(u8),
1423 .mode = 0644,
1424 .proc_handler = proc_dou8vec_minmax,
1425 .extra1 = SYSCTL_ZERO,
1426 .extra2 = SYSCTL_ONE,
1427 },
1428 {
1429 .procname = "tcp_invalid_ratelimit",
1430 .data = &init_net.ipv4.sysctl_tcp_invalid_ratelimit,
1431 .maxlen = sizeof(int),
1432 .mode = 0644,
1433 .proc_handler = proc_dointvec_ms_jiffies,
1434 },
1435 {
1436 .procname = "tcp_pacing_ss_ratio",
1437 .data = &init_net.ipv4.sysctl_tcp_pacing_ss_ratio,
1438 .maxlen = sizeof(int),
1439 .mode = 0644,
1440 .proc_handler = proc_dointvec_minmax,
1441 .extra1 = SYSCTL_ZERO,
1442 .extra2 = SYSCTL_ONE_THOUSAND,
1443 },
1444 {
1445 .procname = "tcp_pacing_ca_ratio",
1446 .data = &init_net.ipv4.sysctl_tcp_pacing_ca_ratio,
1447 .maxlen = sizeof(int),
1448 .mode = 0644,
1449 .proc_handler = proc_dointvec_minmax,
1450 .extra1 = SYSCTL_ZERO,
1451 .extra2 = SYSCTL_ONE_THOUSAND,
1452 },
1453 {
1454 .procname = "tcp_wmem",
1455 .data = &init_net.ipv4.sysctl_tcp_wmem,
1456 .maxlen = sizeof(init_net.ipv4.sysctl_tcp_wmem),
1457 .mode = 0644,
1458 .proc_handler = proc_dointvec_minmax,
1459 .extra1 = SYSCTL_ONE,
1460 },
1461 {
1462 .procname = "tcp_rmem",
1463 .data = &init_net.ipv4.sysctl_tcp_rmem,
1464 .maxlen = sizeof(init_net.ipv4.sysctl_tcp_rmem),
1465 .mode = 0644,
1466 .proc_handler = proc_dointvec_minmax,
1467 .extra1 = &tcp_min_rcvbuf,
1468 },
1469 {
1470 .procname = "tcp_comp_sack_delay_ns",
1471 .data = &init_net.ipv4.sysctl_tcp_comp_sack_delay_ns,
1472 .maxlen = sizeof(unsigned long),
1473 .mode = 0644,
1474 .proc_handler = proc_doulongvec_minmax,
1475 },
1476 {
1477 .procname = "tcp_comp_sack_rtt_percent",
1478 .data = &init_net.ipv4.sysctl_tcp_comp_sack_rtt_percent,
1479 .maxlen = sizeof(int),
1480 .mode = 0644,
1481 .proc_handler = proc_dointvec_minmax,
1482 .extra1 = SYSCTL_ONE,
1483 .extra2 = SYSCTL_ONE_THOUSAND,
1484 },
1485 {
1486 .procname = "tcp_comp_sack_slack_ns",
1487 .data = &init_net.ipv4.sysctl_tcp_comp_sack_slack_ns,
1488 .maxlen = sizeof(unsigned long),
1489 .mode = 0644,
1490 .proc_handler = proc_doulongvec_minmax,
1491 },
1492 {
1493 .procname = "tcp_comp_sack_nr",
1494 .data = &init_net.ipv4.sysctl_tcp_comp_sack_nr,
1495 .maxlen = sizeof(u8),
1496 .mode = 0644,
1497 .proc_handler = proc_dou8vec_minmax,
1498 .extra1 = SYSCTL_ZERO,
1499 },
1500 {
1501 .procname = "tcp_backlog_ack_defer",
1502 .data = &init_net.ipv4.sysctl_tcp_backlog_ack_defer,
1503 .maxlen = sizeof(u8),
1504 .mode = 0644,
1505 .proc_handler = proc_dou8vec_minmax,
1506 .extra1 = SYSCTL_ZERO,
1507 .extra2 = SYSCTL_ONE,
1508 },
1509 {
1510 .procname = "tcp_reflect_tos",
1511 .data = &init_net.ipv4.sysctl_tcp_reflect_tos,
1512 .maxlen = sizeof(u8),
1513 .mode = 0644,
1514 .proc_handler = proc_dou8vec_minmax,
1515 .extra1 = SYSCTL_ZERO,
1516 .extra2 = SYSCTL_ONE,
1517 },
1518 {
1519 .procname = "tcp_ehash_entries",
1520 .data = &init_net.ipv4.sysctl_tcp_child_ehash_entries,
1521 .mode = 0444,
1522 .proc_handler = proc_tcp_ehash_entries,
1523 },
1524 {
1525 .procname = "tcp_child_ehash_entries",
1526 .data = &init_net.ipv4.sysctl_tcp_child_ehash_entries,
1527 .maxlen = sizeof(unsigned int),
1528 .mode = 0644,
1529 .proc_handler = proc_douintvec_minmax,
1530 .extra1 = SYSCTL_ZERO,
1531 .extra2 = &tcp_child_ehash_entries_max,
1532 },
1533 {
1534 .procname = "udp_hash_entries",
1535 .data = &init_net.ipv4.sysctl_udp_child_hash_entries,
1536 .mode = 0444,
1537 .proc_handler = proc_udp_hash_entries,
1538 },
1539 {
1540 .procname = "udp_child_hash_entries",
1541 .data = &init_net.ipv4.sysctl_udp_child_hash_entries,
1542 .maxlen = sizeof(unsigned int),
1543 .mode = 0644,
1544 .proc_handler = proc_douintvec_minmax,
1545 .extra1 = SYSCTL_ZERO,
1546 .extra2 = &udp_child_hash_entries_max,
1547 },
1548 {
1549 .procname = "udp_rmem_min",
1550 .data = &init_net.ipv4.sysctl_udp_rmem_min,
1551 .maxlen = sizeof(init_net.ipv4.sysctl_udp_rmem_min),
1552 .mode = 0644,
1553 .proc_handler = proc_dointvec_minmax,
1554 .extra1 = SYSCTL_ONE
1555 },
1556 {
1557 .procname = "udp_wmem_min",
1558 .data = &init_net.ipv4.sysctl_udp_wmem_min,
1559 .maxlen = sizeof(init_net.ipv4.sysctl_udp_wmem_min),
1560 .mode = 0644,
1561 .proc_handler = proc_dointvec_minmax,
1562 .extra1 = SYSCTL_ONE
1563 },
1564 {
1565 .procname = "fib_notify_on_flag_change",
1566 .data = &init_net.ipv4.sysctl_fib_notify_on_flag_change,
1567 .maxlen = sizeof(u8),
1568 .mode = 0644,
1569 .proc_handler = proc_dou8vec_minmax,
1570 .extra1 = SYSCTL_ZERO,
1571 .extra2 = SYSCTL_TWO,
1572 },
1573 {
1574 .procname = "tcp_plb_enabled",
1575 .data = &init_net.ipv4.sysctl_tcp_plb_enabled,
1576 .maxlen = sizeof(u8),
1577 .mode = 0644,
1578 .proc_handler = proc_dou8vec_minmax,
1579 .extra1 = SYSCTL_ZERO,
1580 .extra2 = SYSCTL_ONE,
1581 },
1582 {
1583 .procname = "tcp_plb_idle_rehash_rounds",
1584 .data = &init_net.ipv4.sysctl_tcp_plb_idle_rehash_rounds,
1585 .maxlen = sizeof(u8),
1586 .mode = 0644,
1587 .proc_handler = proc_dou8vec_minmax,
1588 .extra2 = &tcp_plb_max_rounds,
1589 },
1590 {
1591 .procname = "tcp_plb_rehash_rounds",
1592 .data = &init_net.ipv4.sysctl_tcp_plb_rehash_rounds,
1593 .maxlen = sizeof(u8),
1594 .mode = 0644,
1595 .proc_handler = proc_dou8vec_minmax,
1596 .extra2 = &tcp_plb_max_rounds,
1597 },
1598 {
1599 .procname = "tcp_plb_suspend_rto_sec",
1600 .data = &init_net.ipv4.sysctl_tcp_plb_suspend_rto_sec,
1601 .maxlen = sizeof(u8),
1602 .mode = 0644,
1603 .proc_handler = proc_dou8vec_minmax,
1604 },
1605 {
1606 .procname = "tcp_plb_cong_thresh",
1607 .data = &init_net.ipv4.sysctl_tcp_plb_cong_thresh,
1608 .maxlen = sizeof(int),
1609 .mode = 0644,
1610 .proc_handler = proc_dointvec_minmax,
1611 .extra1 = SYSCTL_ZERO,
1612 .extra2 = &tcp_plb_max_cong_thresh,
1613 },
1614 {
1615 .procname = "tcp_syn_linear_timeouts",
1616 .data = &init_net.ipv4.sysctl_tcp_syn_linear_timeouts,
1617 .maxlen = sizeof(u8),
1618 .mode = 0644,
1619 .proc_handler = proc_dou8vec_minmax,
1620 .extra1 = SYSCTL_ZERO,
1621 .extra2 = &tcp_syn_linear_timeouts_max,
1622 },
1623 {
1624 .procname = "tcp_shrink_window",
1625 .data = &init_net.ipv4.sysctl_tcp_shrink_window,
1626 .maxlen = sizeof(u8),
1627 .mode = 0644,
1628 .proc_handler = proc_dou8vec_minmax,
1629 .extra1 = SYSCTL_ZERO,
1630 .extra2 = SYSCTL_ONE,
1631 },
1632 {
1633 .procname = "tcp_pingpong_thresh",
1634 .data = &init_net.ipv4.sysctl_tcp_pingpong_thresh,
1635 .maxlen = sizeof(u8),
1636 .mode = 0644,
1637 .proc_handler = proc_dou8vec_minmax,
1638 .extra1 = SYSCTL_ONE,
1639 },
1640 {
1641 .procname = "tcp_rto_min_us",
1642 .data = &init_net.ipv4.sysctl_tcp_rto_min_us,
1643 .maxlen = sizeof(int),
1644 .mode = 0644,
1645 .proc_handler = proc_dointvec_minmax,
1646 .extra1 = SYSCTL_ONE,
1647 },
1648 {
1649 .procname = "tcp_rto_max_ms",
1650 .data = &init_net.ipv4.sysctl_tcp_rto_max_ms,
1651 .maxlen = sizeof(int),
1652 .mode = 0644,
1653 .proc_handler = proc_dointvec_minmax,
1654 .extra1 = SYSCTL_ONE_THOUSAND,
1655 .extra2 = &tcp_rto_max_max,
1656 },
1657 };
1658
ipv4_net_table_dup(struct net * net)1659 static const struct ctl_table *ipv4_net_table_dup(struct net *net)
1660 {
1661 size_t table_size = ARRAY_SIZE(ipv4_net_table);
1662 struct ctl_table *table;
1663 int i;
1664
1665 table = kmemdup(ipv4_net_table, sizeof(ipv4_net_table), GFP_KERNEL);
1666 if (!table)
1667 return NULL;
1668
1669 for (i = 0; i < table_size; i++) {
1670 if (table[i].data) {
1671 /* Update the variables to point into
1672 * the current struct net
1673 */
1674 table[i].data += (void *)net - (void *)&init_net;
1675 } else {
1676 /* Entries without data pointer are global;
1677 * Make them read-only in non-init_net ns
1678 */
1679 table[i].mode &= ~0222;
1680 }
1681 if (table[i].extra2 >= (void *)&init_net.ipv4 &&
1682 table[i].extra2 < (void *)(&init_net.ipv4 + 1))
1683 table[i].extra2 += (void *)net - (void *)&init_net;
1684 }
1685 return table;
1686 }
1687
ipv4_sysctl_init_net(struct net * net)1688 static __net_init int ipv4_sysctl_init_net(struct net *net)
1689 {
1690 size_t table_size = ARRAY_SIZE(ipv4_net_table);
1691 const struct ctl_table *table;
1692
1693 table = ipv4_net_table;
1694 if (!net_eq(net, &init_net)) {
1695 table = ipv4_net_table_dup(net);
1696 if (!table)
1697 goto err_alloc;
1698 }
1699
1700 net->ipv4.ipv4_hdr = register_net_sysctl_sz(net, "net/ipv4", table,
1701 table_size);
1702 if (!net->ipv4.ipv4_hdr)
1703 goto err_reg;
1704
1705 net->ipv4.sysctl_local_reserved_ports = kzalloc(65536 / 8, GFP_KERNEL);
1706 if (!net->ipv4.sysctl_local_reserved_ports)
1707 goto err_ports;
1708
1709 proc_fib_multipath_hash_set_seed(net, 0);
1710
1711 return 0;
1712
1713 err_ports:
1714 unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
1715 err_reg:
1716 if (!net_eq(net, &init_net))
1717 kfree(table);
1718 err_alloc:
1719 return -ENOMEM;
1720 }
1721
ipv4_sysctl_exit_net(struct net * net)1722 static __net_exit void ipv4_sysctl_exit_net(struct net *net)
1723 {
1724 const struct ctl_table *table;
1725
1726 table = net->ipv4.ipv4_hdr->ctl_table_arg;
1727 unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
1728 kfree(table);
1729 kfree(net->ipv4.sysctl_local_reserved_ports);
1730 }
1731
1732 static __net_initdata struct pernet_operations ipv4_sysctl_ops = {
1733 .init = ipv4_sysctl_init_net,
1734 .exit = ipv4_sysctl_exit_net,
1735 };
1736
sysctl_ipv4_init(void)1737 static __init int sysctl_ipv4_init(void)
1738 {
1739 struct ctl_table_header *hdr;
1740
1741 hdr = register_net_sysctl(&init_net, "net/ipv4", ipv4_table);
1742 if (!hdr)
1743 return -ENOMEM;
1744
1745 proc_fib_multipath_hash_init_rand_seed();
1746
1747 if (register_pernet_subsys(&ipv4_sysctl_ops)) {
1748 unregister_net_sysctl_table(hdr);
1749 return -ENOMEM;
1750 }
1751
1752 return 0;
1753 }
1754
1755 __initcall(sysctl_ipv4_init);
1756