xref: /linux/net/ipv4/sysctl_net_ipv4.c (revision 07fdad3a93756b872da7b53647715c48d0f4a2d0)
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 = 2;
51 
52 /* obsolete */
53 static int sysctl_tcp_low_latency __read_mostly;
54 
55 /* Update system visible IP port range */
56 static void set_local_port_range(struct net *net, unsigned int low, unsigned int high)
57 {
58 	bool same_parity = !((low ^ high) & 1);
59 
60 	if (same_parity && !net->ipv4.ip_local_ports.warned) {
61 		net->ipv4.ip_local_ports.warned = true;
62 		pr_err_ratelimited("ip_local_port_range: prefer different parity for start/end values.\n");
63 	}
64 	WRITE_ONCE(net->ipv4.ip_local_ports.range, high << 16 | low);
65 }
66 
67 /* Validate changes from /proc interface. */
68 static int ipv4_local_port_range(const struct ctl_table *table, int write,
69 				 void *buffer, size_t *lenp, loff_t *ppos)
70 {
71 	struct net *net = table->data;
72 	int ret;
73 	int range[2];
74 	struct ctl_table tmp = {
75 		.data = &range,
76 		.maxlen = sizeof(range),
77 		.mode = table->mode,
78 		.extra1 = &ip_local_port_range_min,
79 		.extra2 = &ip_local_port_range_max,
80 	};
81 
82 	inet_get_local_port_range(net, &range[0], &range[1]);
83 
84 	ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
85 
86 	if (write && ret == 0) {
87 		/* Ensure that the upper limit is not smaller than the lower,
88 		 * and that the lower does not encroach upon the privileged
89 		 * port limit.
90 		 */
91 		if ((range[1] < range[0]) ||
92 		    (range[0] < READ_ONCE(net->ipv4.sysctl_ip_prot_sock)))
93 			ret = -EINVAL;
94 		else
95 			set_local_port_range(net, range[0], range[1]);
96 	}
97 
98 	return ret;
99 }
100 
101 /* Validate changes from /proc interface. */
102 static int ipv4_privileged_ports(const struct ctl_table *table, int write,
103 				void *buffer, size_t *lenp, loff_t *ppos)
104 {
105 	struct net *net = container_of(table->data, struct net,
106 	    ipv4.sysctl_ip_prot_sock);
107 	int ret;
108 	int pports;
109 	int range[2];
110 	struct ctl_table tmp = {
111 		.data = &pports,
112 		.maxlen = sizeof(pports),
113 		.mode = table->mode,
114 		.extra1 = &ip_privileged_port_min,
115 		.extra2 = &ip_privileged_port_max,
116 	};
117 
118 	pports = READ_ONCE(net->ipv4.sysctl_ip_prot_sock);
119 
120 	ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
121 
122 	if (write && ret == 0) {
123 		inet_get_local_port_range(net, &range[0], &range[1]);
124 		/* Ensure that the local port range doesn't overlap with the
125 		 * privileged port range.
126 		 */
127 		if (range[0] < pports)
128 			ret = -EINVAL;
129 		else
130 			WRITE_ONCE(net->ipv4.sysctl_ip_prot_sock, pports);
131 	}
132 
133 	return ret;
134 }
135 
136 static void inet_get_ping_group_range_table(const struct ctl_table *table,
137 					    kgid_t *low, kgid_t *high)
138 {
139 	kgid_t *data = table->data;
140 	struct net *net =
141 		container_of(table->data, struct net, ipv4.ping_group_range.range);
142 	unsigned int seq;
143 	do {
144 		seq = read_seqbegin(&net->ipv4.ping_group_range.lock);
145 
146 		*low = data[0];
147 		*high = data[1];
148 	} while (read_seqretry(&net->ipv4.ping_group_range.lock, seq));
149 }
150 
151 /* Update system visible IP port range */
152 static void set_ping_group_range(const struct ctl_table *table,
153 				 kgid_t low, kgid_t high)
154 {
155 	kgid_t *data = table->data;
156 	struct net *net =
157 		container_of(table->data, struct net, ipv4.ping_group_range.range);
158 	write_seqlock(&net->ipv4.ping_group_range.lock);
159 	data[0] = low;
160 	data[1] = high;
161 	write_sequnlock(&net->ipv4.ping_group_range.lock);
162 }
163 
164 /* Validate changes from /proc interface. */
165 static int ipv4_ping_group_range(const struct ctl_table *table, int write,
166 				 void *buffer, size_t *lenp, loff_t *ppos)
167 {
168 	struct user_namespace *user_ns = current_user_ns();
169 	int ret;
170 	unsigned long urange[2];
171 	kgid_t low, high;
172 	struct ctl_table tmp = {
173 		.data = &urange,
174 		.maxlen = sizeof(urange),
175 		.mode = table->mode,
176 		.extra1 = &ip_ping_group_range_min,
177 		.extra2 = &ip_ping_group_range_max,
178 	};
179 
180 	inet_get_ping_group_range_table(table, &low, &high);
181 	urange[0] = from_kgid_munged(user_ns, low);
182 	urange[1] = from_kgid_munged(user_ns, high);
183 	ret = proc_doulongvec_minmax(&tmp, write, buffer, lenp, ppos);
184 
185 	if (write && ret == 0) {
186 		low = make_kgid(user_ns, urange[0]);
187 		high = make_kgid(user_ns, urange[1]);
188 		if (!gid_valid(low) || !gid_valid(high))
189 			return -EINVAL;
190 		if (urange[1] < urange[0] || gid_lt(high, low)) {
191 			low = make_kgid(&init_user_ns, 1);
192 			high = make_kgid(&init_user_ns, 0);
193 		}
194 		set_ping_group_range(table, low, high);
195 	}
196 
197 	return ret;
198 }
199 
200 static int ipv4_fwd_update_priority(const struct ctl_table *table, int write,
201 				    void *buffer, size_t *lenp, loff_t *ppos)
202 {
203 	struct net *net;
204 	int ret;
205 
206 	net = container_of(table->data, struct net,
207 			   ipv4.sysctl_ip_fwd_update_priority);
208 	ret = proc_dou8vec_minmax(table, write, buffer, lenp, ppos);
209 	if (write && ret == 0)
210 		call_netevent_notifiers(NETEVENT_IPV4_FWD_UPDATE_PRIORITY_UPDATE,
211 					net);
212 
213 	return ret;
214 }
215 
216 static int proc_tcp_congestion_control(const struct ctl_table *ctl, int write,
217 				       void *buffer, size_t *lenp, loff_t *ppos)
218 {
219 	struct net *net = container_of(ctl->data, struct net,
220 				       ipv4.tcp_congestion_control);
221 	char val[TCP_CA_NAME_MAX];
222 	struct ctl_table tbl = {
223 		.data = val,
224 		.maxlen = TCP_CA_NAME_MAX,
225 	};
226 	int ret;
227 
228 	tcp_get_default_congestion_control(net, val);
229 
230 	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
231 	if (write && ret == 0)
232 		ret = tcp_set_default_congestion_control(net, val);
233 	return ret;
234 }
235 
236 static int proc_tcp_available_congestion_control(const struct ctl_table *ctl,
237 						 int write, void *buffer,
238 						 size_t *lenp, loff_t *ppos)
239 {
240 	struct ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX, };
241 	int ret;
242 
243 	tbl.data = kmalloc(tbl.maxlen, GFP_USER);
244 	if (!tbl.data)
245 		return -ENOMEM;
246 	tcp_get_available_congestion_control(tbl.data, TCP_CA_BUF_MAX);
247 	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
248 	kfree(tbl.data);
249 	return ret;
250 }
251 
252 static int proc_allowed_congestion_control(const struct ctl_table *ctl,
253 					   int write, void *buffer,
254 					   size_t *lenp, loff_t *ppos)
255 {
256 	struct ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX };
257 	int ret;
258 
259 	tbl.data = kmalloc(tbl.maxlen, GFP_USER);
260 	if (!tbl.data)
261 		return -ENOMEM;
262 
263 	tcp_get_allowed_congestion_control(tbl.data, tbl.maxlen);
264 	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
265 	if (write && ret == 0)
266 		ret = tcp_set_allowed_congestion_control(tbl.data);
267 	kfree(tbl.data);
268 	return ret;
269 }
270 
271 static int sscanf_key(char *buf, __le32 *key)
272 {
273 	u32 user_key[4];
274 	int i, ret = 0;
275 
276 	if (sscanf(buf, "%x-%x-%x-%x", user_key, user_key + 1,
277 		   user_key + 2, user_key + 3) != 4) {
278 		ret = -EINVAL;
279 	} else {
280 		for (i = 0; i < ARRAY_SIZE(user_key); i++)
281 			key[i] = cpu_to_le32(user_key[i]);
282 	}
283 	pr_debug("proc TFO key set 0x%x-%x-%x-%x <- 0x%s: %u\n",
284 		 user_key[0], user_key[1], user_key[2], user_key[3], buf, ret);
285 
286 	return ret;
287 }
288 
289 static int proc_tcp_fastopen_key(const struct ctl_table *table, int write,
290 				 void *buffer, size_t *lenp, loff_t *ppos)
291 {
292 	struct net *net = container_of(table->data, struct net,
293 	    ipv4.sysctl_tcp_fastopen);
294 	/* maxlen to print the list of keys in hex (*2), with dashes
295 	 * separating doublewords and a comma in between keys.
296 	 */
297 	struct ctl_table tbl = { .maxlen = ((TCP_FASTOPEN_KEY_LENGTH *
298 					    2 * TCP_FASTOPEN_KEY_MAX) +
299 					    (TCP_FASTOPEN_KEY_MAX * 5)) };
300 	u32 user_key[TCP_FASTOPEN_KEY_BUF_LENGTH / sizeof(u32)];
301 	__le32 key[TCP_FASTOPEN_KEY_BUF_LENGTH / sizeof(__le32)];
302 	char *backup_data;
303 	int ret, i = 0, off = 0, n_keys;
304 
305 	tbl.data = kmalloc(tbl.maxlen, GFP_KERNEL);
306 	if (!tbl.data)
307 		return -ENOMEM;
308 
309 	n_keys = tcp_fastopen_get_cipher(net, NULL, (u64 *)key);
310 	if (!n_keys) {
311 		memset(&key[0], 0, TCP_FASTOPEN_KEY_LENGTH);
312 		n_keys = 1;
313 	}
314 
315 	for (i = 0; i < n_keys * 4; i++)
316 		user_key[i] = le32_to_cpu(key[i]);
317 
318 	for (i = 0; i < n_keys; i++) {
319 		off += snprintf(tbl.data + off, tbl.maxlen - off,
320 				"%08x-%08x-%08x-%08x",
321 				user_key[i * 4],
322 				user_key[i * 4 + 1],
323 				user_key[i * 4 + 2],
324 				user_key[i * 4 + 3]);
325 
326 		if (WARN_ON_ONCE(off >= tbl.maxlen - 1))
327 			break;
328 
329 		if (i + 1 < n_keys)
330 			off += snprintf(tbl.data + off, tbl.maxlen - off, ",");
331 	}
332 
333 	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
334 
335 	if (write && ret == 0) {
336 		backup_data = strchr(tbl.data, ',');
337 		if (backup_data) {
338 			*backup_data = '\0';
339 			backup_data++;
340 		}
341 		if (sscanf_key(tbl.data, key)) {
342 			ret = -EINVAL;
343 			goto bad_key;
344 		}
345 		if (backup_data) {
346 			if (sscanf_key(backup_data, key + 4)) {
347 				ret = -EINVAL;
348 				goto bad_key;
349 			}
350 		}
351 		tcp_fastopen_reset_cipher(net, NULL, key,
352 					  backup_data ? key + 4 : NULL);
353 	}
354 
355 bad_key:
356 	kfree(tbl.data);
357 	return ret;
358 }
359 
360 static int proc_tfo_blackhole_detect_timeout(const struct ctl_table *table,
361 					     int write, void *buffer,
362 					     size_t *lenp, loff_t *ppos)
363 {
364 	struct net *net = container_of(table->data, struct net,
365 	    ipv4.sysctl_tcp_fastopen_blackhole_timeout);
366 	int ret;
367 
368 	ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
369 	if (write && ret == 0)
370 		atomic_set(&net->ipv4.tfo_active_disable_times, 0);
371 
372 	return ret;
373 }
374 
375 static int proc_tcp_available_ulp(const struct ctl_table *ctl,
376 				  int write, void *buffer, size_t *lenp,
377 				  loff_t *ppos)
378 {
379 	struct ctl_table tbl = { .maxlen = TCP_ULP_BUF_MAX, };
380 	int ret;
381 
382 	tbl.data = kmalloc(tbl.maxlen, GFP_USER);
383 	if (!tbl.data)
384 		return -ENOMEM;
385 	tcp_get_available_ulp(tbl.data, TCP_ULP_BUF_MAX);
386 	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
387 	kfree(tbl.data);
388 
389 	return ret;
390 }
391 
392 static int proc_tcp_ehash_entries(const struct ctl_table *table, int write,
393 				  void *buffer, size_t *lenp, loff_t *ppos)
394 {
395 	struct net *net = container_of(table->data, struct net,
396 				       ipv4.sysctl_tcp_child_ehash_entries);
397 	struct inet_hashinfo *hinfo = net->ipv4.tcp_death_row.hashinfo;
398 	int tcp_ehash_entries;
399 	struct ctl_table tbl;
400 
401 	tcp_ehash_entries = hinfo->ehash_mask + 1;
402 
403 	/* A negative number indicates that the child netns
404 	 * shares the global ehash.
405 	 */
406 	if (!net_eq(net, &init_net) && !hinfo->pernet)
407 		tcp_ehash_entries *= -1;
408 
409 	memset(&tbl, 0, sizeof(tbl));
410 	tbl.data = &tcp_ehash_entries;
411 	tbl.maxlen = sizeof(int);
412 
413 	return proc_dointvec(&tbl, write, buffer, lenp, ppos);
414 }
415 
416 static int proc_udp_hash_entries(const struct ctl_table *table, int write,
417 				 void *buffer, size_t *lenp, loff_t *ppos)
418 {
419 	struct net *net = container_of(table->data, struct net,
420 				       ipv4.sysctl_udp_child_hash_entries);
421 	int udp_hash_entries;
422 	struct ctl_table tbl;
423 
424 	udp_hash_entries = net->ipv4.udp_table->mask + 1;
425 
426 	/* A negative number indicates that the child netns
427 	 * shares the global udp_table.
428 	 */
429 	if (!net_eq(net, &init_net) && net->ipv4.udp_table == &udp_table)
430 		udp_hash_entries *= -1;
431 
432 	memset(&tbl, 0, sizeof(tbl));
433 	tbl.data = &udp_hash_entries;
434 	tbl.maxlen = sizeof(int);
435 
436 	return proc_dointvec(&tbl, write, buffer, lenp, ppos);
437 }
438 
439 #ifdef CONFIG_IP_ROUTE_MULTIPATH
440 static int proc_fib_multipath_hash_policy(const struct ctl_table *table, int write,
441 					  void *buffer, size_t *lenp,
442 					  loff_t *ppos)
443 {
444 	struct net *net = container_of(table->data, struct net,
445 	    ipv4.sysctl_fib_multipath_hash_policy);
446 	int ret;
447 
448 	ret = proc_dou8vec_minmax(table, write, buffer, lenp, ppos);
449 	if (write && ret == 0)
450 		call_netevent_notifiers(NETEVENT_IPV4_MPATH_HASH_UPDATE, net);
451 
452 	return ret;
453 }
454 
455 static int proc_fib_multipath_hash_fields(const struct ctl_table *table, int write,
456 					  void *buffer, size_t *lenp,
457 					  loff_t *ppos)
458 {
459 	struct net *net;
460 	int ret;
461 
462 	net = container_of(table->data, struct net,
463 			   ipv4.sysctl_fib_multipath_hash_fields);
464 	ret = proc_douintvec_minmax(table, write, buffer, lenp, ppos);
465 	if (write && ret == 0)
466 		call_netevent_notifiers(NETEVENT_IPV4_MPATH_HASH_UPDATE, net);
467 
468 	return ret;
469 }
470 
471 static u32 proc_fib_multipath_hash_rand_seed __ro_after_init;
472 
473 static void proc_fib_multipath_hash_init_rand_seed(void)
474 {
475 	get_random_bytes(&proc_fib_multipath_hash_rand_seed,
476 			 sizeof(proc_fib_multipath_hash_rand_seed));
477 }
478 
479 static void proc_fib_multipath_hash_set_seed(struct net *net, u32 user_seed)
480 {
481 	struct sysctl_fib_multipath_hash_seed new = {
482 		.user_seed = user_seed,
483 		.mp_seed = (user_seed ? user_seed :
484 			    proc_fib_multipath_hash_rand_seed),
485 	};
486 
487 	WRITE_ONCE(net->ipv4.sysctl_fib_multipath_hash_seed, new);
488 }
489 
490 static int proc_fib_multipath_hash_seed(const struct ctl_table *table, int write,
491 					void *buffer, size_t *lenp,
492 					loff_t *ppos)
493 {
494 	struct sysctl_fib_multipath_hash_seed *mphs;
495 	struct net *net = table->data;
496 	struct ctl_table tmp;
497 	u32 user_seed;
498 	int ret;
499 
500 	mphs = &net->ipv4.sysctl_fib_multipath_hash_seed;
501 	user_seed = mphs->user_seed;
502 
503 	tmp = *table;
504 	tmp.data = &user_seed;
505 
506 	ret = proc_douintvec_minmax(&tmp, write, buffer, lenp, ppos);
507 
508 	if (write && ret == 0) {
509 		proc_fib_multipath_hash_set_seed(net, user_seed);
510 		call_netevent_notifiers(NETEVENT_IPV4_MPATH_HASH_UPDATE, net);
511 	}
512 
513 	return ret;
514 }
515 #else
516 
517 static void proc_fib_multipath_hash_init_rand_seed(void)
518 {
519 }
520 
521 static void proc_fib_multipath_hash_set_seed(struct net *net, u32 user_seed)
522 {
523 }
524 
525 #endif
526 
527 static struct ctl_table ipv4_table[] = {
528 	{
529 		.procname	= "tcp_max_orphans",
530 		.data		= &sysctl_tcp_max_orphans,
531 		.maxlen		= sizeof(int),
532 		.mode		= 0644,
533 		.proc_handler	= proc_dointvec
534 	},
535 	{
536 		.procname	= "inet_peer_threshold",
537 		.data		= &inet_peer_threshold,
538 		.maxlen		= sizeof(int),
539 		.mode		= 0644,
540 		.proc_handler	= proc_dointvec
541 	},
542 	{
543 		.procname	= "inet_peer_minttl",
544 		.data		= &inet_peer_minttl,
545 		.maxlen		= sizeof(int),
546 		.mode		= 0644,
547 		.proc_handler	= proc_dointvec_jiffies,
548 	},
549 	{
550 		.procname	= "inet_peer_maxttl",
551 		.data		= &inet_peer_maxttl,
552 		.maxlen		= sizeof(int),
553 		.mode		= 0644,
554 		.proc_handler	= proc_dointvec_jiffies,
555 	},
556 	{
557 		.procname	= "tcp_mem",
558 		.maxlen		= sizeof(sysctl_tcp_mem),
559 		.data		= &sysctl_tcp_mem,
560 		.mode		= 0644,
561 		.proc_handler	= proc_doulongvec_minmax,
562 	},
563 	{
564 		.procname	= "tcp_low_latency",
565 		.data		= &sysctl_tcp_low_latency,
566 		.maxlen		= sizeof(int),
567 		.mode		= 0644,
568 		.proc_handler	= proc_dointvec
569 	},
570 #ifdef CONFIG_NETLABEL
571 	{
572 		.procname	= "cipso_cache_enable",
573 		.data		= &cipso_v4_cache_enabled,
574 		.maxlen		= sizeof(int),
575 		.mode		= 0644,
576 		.proc_handler	= proc_dointvec,
577 	},
578 	{
579 		.procname	= "cipso_cache_bucket_size",
580 		.data		= &cipso_v4_cache_bucketsize,
581 		.maxlen		= sizeof(int),
582 		.mode		= 0644,
583 		.proc_handler	= proc_dointvec,
584 	},
585 	{
586 		.procname	= "cipso_rbm_optfmt",
587 		.data		= &cipso_v4_rbm_optfmt,
588 		.maxlen		= sizeof(int),
589 		.mode		= 0644,
590 		.proc_handler	= proc_dointvec,
591 	},
592 	{
593 		.procname	= "cipso_rbm_strictvalid",
594 		.data		= &cipso_v4_rbm_strictvalid,
595 		.maxlen		= sizeof(int),
596 		.mode		= 0644,
597 		.proc_handler	= proc_dointvec,
598 	},
599 #endif /* CONFIG_NETLABEL */
600 	{
601 		.procname	= "tcp_available_ulp",
602 		.maxlen		= TCP_ULP_BUF_MAX,
603 		.mode		= 0444,
604 		.proc_handler   = proc_tcp_available_ulp,
605 	},
606 	{
607 		.procname	= "udp_mem",
608 		.data		= &sysctl_udp_mem,
609 		.maxlen		= sizeof(sysctl_udp_mem),
610 		.mode		= 0644,
611 		.proc_handler	= proc_doulongvec_minmax,
612 	},
613 	{
614 		.procname	= "fib_sync_mem",
615 		.data		= &sysctl_fib_sync_mem,
616 		.maxlen		= sizeof(sysctl_fib_sync_mem),
617 		.mode		= 0644,
618 		.proc_handler	= proc_douintvec_minmax,
619 		.extra1		= &sysctl_fib_sync_mem_min,
620 		.extra2		= &sysctl_fib_sync_mem_max,
621 	},
622 };
623 
624 static struct ctl_table ipv4_net_table[] = {
625 	{
626 		.procname	= "tcp_max_tw_buckets",
627 		.data		= &init_net.ipv4.tcp_death_row.sysctl_max_tw_buckets,
628 		.maxlen		= sizeof(int),
629 		.mode		= 0644,
630 		.proc_handler	= proc_dointvec
631 	},
632 	{
633 		.procname	= "icmp_echo_ignore_all",
634 		.data		= &init_net.ipv4.sysctl_icmp_echo_ignore_all,
635 		.maxlen		= sizeof(u8),
636 		.mode		= 0644,
637 		.proc_handler	= proc_dou8vec_minmax,
638 		.extra1		= SYSCTL_ZERO,
639 		.extra2		= SYSCTL_ONE
640 	},
641 	{
642 		.procname	= "icmp_echo_enable_probe",
643 		.data		= &init_net.ipv4.sysctl_icmp_echo_enable_probe,
644 		.maxlen		= sizeof(u8),
645 		.mode		= 0644,
646 		.proc_handler	= proc_dou8vec_minmax,
647 		.extra1		= SYSCTL_ZERO,
648 		.extra2		= SYSCTL_ONE
649 	},
650 	{
651 		.procname	= "icmp_echo_ignore_broadcasts",
652 		.data		= &init_net.ipv4.sysctl_icmp_echo_ignore_broadcasts,
653 		.maxlen		= sizeof(u8),
654 		.mode		= 0644,
655 		.proc_handler	= proc_dou8vec_minmax,
656 		.extra1		= SYSCTL_ZERO,
657 		.extra2		= SYSCTL_ONE
658 	},
659 	{
660 		.procname	= "icmp_ignore_bogus_error_responses",
661 		.data		= &init_net.ipv4.sysctl_icmp_ignore_bogus_error_responses,
662 		.maxlen		= sizeof(u8),
663 		.mode		= 0644,
664 		.proc_handler	= proc_dou8vec_minmax,
665 		.extra1		= SYSCTL_ZERO,
666 		.extra2		= SYSCTL_ONE
667 	},
668 	{
669 		.procname	= "icmp_errors_use_inbound_ifaddr",
670 		.data		= &init_net.ipv4.sysctl_icmp_errors_use_inbound_ifaddr,
671 		.maxlen		= sizeof(u8),
672 		.mode		= 0644,
673 		.proc_handler	= proc_dou8vec_minmax,
674 		.extra1		= SYSCTL_ZERO,
675 		.extra2		= SYSCTL_ONE
676 	},
677 	{
678 		.procname	= "icmp_ratelimit",
679 		.data		= &init_net.ipv4.sysctl_icmp_ratelimit,
680 		.maxlen		= sizeof(int),
681 		.mode		= 0644,
682 		.proc_handler	= proc_dointvec_ms_jiffies,
683 	},
684 	{
685 		.procname	= "icmp_ratemask",
686 		.data		= &init_net.ipv4.sysctl_icmp_ratemask,
687 		.maxlen		= sizeof(int),
688 		.mode		= 0644,
689 		.proc_handler	= proc_dointvec
690 	},
691 	{
692 		.procname	= "icmp_msgs_per_sec",
693 		.data		= &init_net.ipv4.sysctl_icmp_msgs_per_sec,
694 		.maxlen		= sizeof(int),
695 		.mode		= 0644,
696 		.proc_handler	= proc_dointvec_minmax,
697 		.extra1		= SYSCTL_ZERO,
698 	},
699 	{
700 		.procname	= "icmp_msgs_burst",
701 		.data		= &init_net.ipv4.sysctl_icmp_msgs_burst,
702 		.maxlen		= sizeof(int),
703 		.mode		= 0644,
704 		.proc_handler	= proc_dointvec_minmax,
705 		.extra1		= SYSCTL_ZERO,
706 	},
707 	{
708 		.procname	= "ping_group_range",
709 		.data		= &init_net.ipv4.ping_group_range.range,
710 		.maxlen		= sizeof(gid_t)*2,
711 		.mode		= 0644,
712 		.proc_handler	= ipv4_ping_group_range,
713 	},
714 #ifdef CONFIG_NET_L3_MASTER_DEV
715 	{
716 		.procname	= "raw_l3mdev_accept",
717 		.data		= &init_net.ipv4.sysctl_raw_l3mdev_accept,
718 		.maxlen		= sizeof(u8),
719 		.mode		= 0644,
720 		.proc_handler	= proc_dou8vec_minmax,
721 		.extra1		= SYSCTL_ZERO,
722 		.extra2		= SYSCTL_ONE,
723 	},
724 #endif
725 	{
726 		.procname	= "tcp_ecn",
727 		.data		= &init_net.ipv4.sysctl_tcp_ecn,
728 		.maxlen		= sizeof(u8),
729 		.mode		= 0644,
730 		.proc_handler	= proc_dou8vec_minmax,
731 		.extra1		= SYSCTL_ZERO,
732 		.extra2		= &tcp_ecn_mode_max,
733 	},
734 	{
735 		.procname	= "tcp_ecn_option",
736 		.data		= &init_net.ipv4.sysctl_tcp_ecn_option,
737 		.maxlen		= sizeof(u8),
738 		.mode		= 0644,
739 		.proc_handler	= proc_dou8vec_minmax,
740 		.extra1		= SYSCTL_ZERO,
741 		.extra2		= SYSCTL_TWO,
742 	},
743 	{
744 		.procname	= "tcp_ecn_option_beacon",
745 		.data		= &init_net.ipv4.sysctl_tcp_ecn_option_beacon,
746 		.maxlen		= sizeof(u8),
747 		.mode		= 0644,
748 		.proc_handler	= proc_dou8vec_minmax,
749 		.extra1		= SYSCTL_ZERO,
750 		.extra2		= SYSCTL_THREE,
751 	},
752 	{
753 		.procname	= "tcp_ecn_fallback",
754 		.data		= &init_net.ipv4.sysctl_tcp_ecn_fallback,
755 		.maxlen		= sizeof(u8),
756 		.mode		= 0644,
757 		.proc_handler	= proc_dou8vec_minmax,
758 		.extra1		= SYSCTL_ZERO,
759 		.extra2		= SYSCTL_ONE,
760 	},
761 	{
762 		.procname	= "ip_dynaddr",
763 		.data		= &init_net.ipv4.sysctl_ip_dynaddr,
764 		.maxlen		= sizeof(u8),
765 		.mode		= 0644,
766 		.proc_handler	= proc_dou8vec_minmax,
767 	},
768 	{
769 		.procname	= "ip_early_demux",
770 		.data		= &init_net.ipv4.sysctl_ip_early_demux,
771 		.maxlen		= sizeof(u8),
772 		.mode		= 0644,
773 		.proc_handler	= proc_dou8vec_minmax,
774 	},
775 	{
776 		.procname       = "udp_early_demux",
777 		.data           = &init_net.ipv4.sysctl_udp_early_demux,
778 		.maxlen         = sizeof(u8),
779 		.mode           = 0644,
780 		.proc_handler   = proc_dou8vec_minmax,
781 	},
782 	{
783 		.procname       = "tcp_early_demux",
784 		.data           = &init_net.ipv4.sysctl_tcp_early_demux,
785 		.maxlen         = sizeof(u8),
786 		.mode           = 0644,
787 		.proc_handler   = proc_dou8vec_minmax,
788 	},
789 	{
790 		.procname       = "nexthop_compat_mode",
791 		.data           = &init_net.ipv4.sysctl_nexthop_compat_mode,
792 		.maxlen         = sizeof(u8),
793 		.mode           = 0644,
794 		.proc_handler   = proc_dou8vec_minmax,
795 		.extra1		= SYSCTL_ZERO,
796 		.extra2		= SYSCTL_ONE,
797 	},
798 	{
799 		.procname	= "ip_default_ttl",
800 		.data		= &init_net.ipv4.sysctl_ip_default_ttl,
801 		.maxlen		= sizeof(u8),
802 		.mode		= 0644,
803 		.proc_handler	= proc_dou8vec_minmax,
804 		.extra1		= &ip_ttl_min,
805 		.extra2		= &ip_ttl_max,
806 	},
807 	{
808 		.procname	= "ip_local_port_range",
809 		.maxlen		= 0,
810 		.data		= &init_net,
811 		.mode		= 0644,
812 		.proc_handler	= ipv4_local_port_range,
813 	},
814 	{
815 		.procname	= "ip_local_reserved_ports",
816 		.data		= &init_net.ipv4.sysctl_local_reserved_ports,
817 		.maxlen		= 65536,
818 		.mode		= 0644,
819 		.proc_handler	= proc_do_large_bitmap,
820 	},
821 	{
822 		.procname	= "ip_no_pmtu_disc",
823 		.data		= &init_net.ipv4.sysctl_ip_no_pmtu_disc,
824 		.maxlen		= sizeof(u8),
825 		.mode		= 0644,
826 		.proc_handler	= proc_dou8vec_minmax,
827 	},
828 	{
829 		.procname	= "ip_forward_use_pmtu",
830 		.data		= &init_net.ipv4.sysctl_ip_fwd_use_pmtu,
831 		.maxlen		= sizeof(u8),
832 		.mode		= 0644,
833 		.proc_handler	= proc_dou8vec_minmax,
834 	},
835 	{
836 		.procname	= "ip_forward_update_priority",
837 		.data		= &init_net.ipv4.sysctl_ip_fwd_update_priority,
838 		.maxlen		= sizeof(u8),
839 		.mode		= 0644,
840 		.proc_handler   = ipv4_fwd_update_priority,
841 		.extra1		= SYSCTL_ZERO,
842 		.extra2		= SYSCTL_ONE,
843 	},
844 	{
845 		.procname	= "ip_nonlocal_bind",
846 		.data		= &init_net.ipv4.sysctl_ip_nonlocal_bind,
847 		.maxlen		= sizeof(u8),
848 		.mode		= 0644,
849 		.proc_handler	= proc_dou8vec_minmax,
850 	},
851 	{
852 		.procname	= "ip_autobind_reuse",
853 		.data		= &init_net.ipv4.sysctl_ip_autobind_reuse,
854 		.maxlen		= sizeof(u8),
855 		.mode		= 0644,
856 		.proc_handler	= proc_dou8vec_minmax,
857 		.extra1         = SYSCTL_ZERO,
858 		.extra2         = SYSCTL_ONE,
859 	},
860 	{
861 		.procname	= "fwmark_reflect",
862 		.data		= &init_net.ipv4.sysctl_fwmark_reflect,
863 		.maxlen		= sizeof(u8),
864 		.mode		= 0644,
865 		.proc_handler	= proc_dou8vec_minmax,
866 	},
867 	{
868 		.procname	= "tcp_fwmark_accept",
869 		.data		= &init_net.ipv4.sysctl_tcp_fwmark_accept,
870 		.maxlen		= sizeof(u8),
871 		.mode		= 0644,
872 		.proc_handler	= proc_dou8vec_minmax,
873 	},
874 #ifdef CONFIG_NET_L3_MASTER_DEV
875 	{
876 		.procname	= "tcp_l3mdev_accept",
877 		.data		= &init_net.ipv4.sysctl_tcp_l3mdev_accept,
878 		.maxlen		= sizeof(u8),
879 		.mode		= 0644,
880 		.proc_handler	= proc_dou8vec_minmax,
881 		.extra1		= SYSCTL_ZERO,
882 		.extra2		= SYSCTL_ONE,
883 	},
884 #endif
885 	{
886 		.procname	= "tcp_mtu_probing",
887 		.data		= &init_net.ipv4.sysctl_tcp_mtu_probing,
888 		.maxlen		= sizeof(u8),
889 		.mode		= 0644,
890 		.proc_handler	= proc_dou8vec_minmax,
891 	},
892 	{
893 		.procname	= "tcp_base_mss",
894 		.data		= &init_net.ipv4.sysctl_tcp_base_mss,
895 		.maxlen		= sizeof(int),
896 		.mode		= 0644,
897 		.proc_handler	= proc_dointvec,
898 	},
899 	{
900 		.procname	= "tcp_min_snd_mss",
901 		.data		= &init_net.ipv4.sysctl_tcp_min_snd_mss,
902 		.maxlen		= sizeof(int),
903 		.mode		= 0644,
904 		.proc_handler	= proc_dointvec_minmax,
905 		.extra1		= &tcp_min_snd_mss_min,
906 		.extra2		= &tcp_min_snd_mss_max,
907 	},
908 	{
909 		.procname	= "tcp_mtu_probe_floor",
910 		.data		= &init_net.ipv4.sysctl_tcp_mtu_probe_floor,
911 		.maxlen		= sizeof(int),
912 		.mode		= 0644,
913 		.proc_handler	= proc_dointvec_minmax,
914 		.extra1		= &tcp_min_snd_mss_min,
915 		.extra2		= &tcp_min_snd_mss_max,
916 	},
917 	{
918 		.procname	= "tcp_probe_threshold",
919 		.data		= &init_net.ipv4.sysctl_tcp_probe_threshold,
920 		.maxlen		= sizeof(int),
921 		.mode		= 0644,
922 		.proc_handler	= proc_dointvec,
923 	},
924 	{
925 		.procname	= "tcp_probe_interval",
926 		.data		= &init_net.ipv4.sysctl_tcp_probe_interval,
927 		.maxlen		= sizeof(u32),
928 		.mode		= 0644,
929 		.proc_handler	= proc_douintvec_minmax,
930 		.extra2		= &u32_max_div_HZ,
931 	},
932 	{
933 		.procname	= "igmp_link_local_mcast_reports",
934 		.data		= &init_net.ipv4.sysctl_igmp_llm_reports,
935 		.maxlen		= sizeof(u8),
936 		.mode		= 0644,
937 		.proc_handler	= proc_dou8vec_minmax,
938 	},
939 	{
940 		.procname	= "igmp_max_memberships",
941 		.data		= &init_net.ipv4.sysctl_igmp_max_memberships,
942 		.maxlen		= sizeof(int),
943 		.mode		= 0644,
944 		.proc_handler	= proc_dointvec
945 	},
946 	{
947 		.procname	= "igmp_max_msf",
948 		.data		= &init_net.ipv4.sysctl_igmp_max_msf,
949 		.maxlen		= sizeof(int),
950 		.mode		= 0644,
951 		.proc_handler	= proc_dointvec
952 	},
953 #ifdef CONFIG_IP_MULTICAST
954 	{
955 		.procname	= "igmp_qrv",
956 		.data		= &init_net.ipv4.sysctl_igmp_qrv,
957 		.maxlen		= sizeof(int),
958 		.mode		= 0644,
959 		.proc_handler	= proc_dointvec_minmax,
960 		.extra1		= SYSCTL_ONE
961 	},
962 #endif
963 	{
964 		.procname	= "tcp_congestion_control",
965 		.data		= &init_net.ipv4.tcp_congestion_control,
966 		.mode		= 0644,
967 		.maxlen		= TCP_CA_NAME_MAX,
968 		.proc_handler	= proc_tcp_congestion_control,
969 	},
970 	{
971 		.procname	= "tcp_available_congestion_control",
972 		.maxlen		= TCP_CA_BUF_MAX,
973 		.mode		= 0444,
974 		.proc_handler   = proc_tcp_available_congestion_control,
975 	},
976 	{
977 		.procname	= "tcp_allowed_congestion_control",
978 		.maxlen		= TCP_CA_BUF_MAX,
979 		.mode		= 0644,
980 		.proc_handler   = proc_allowed_congestion_control,
981 	},
982 	{
983 		.procname	= "tcp_keepalive_time",
984 		.data		= &init_net.ipv4.sysctl_tcp_keepalive_time,
985 		.maxlen		= sizeof(int),
986 		.mode		= 0644,
987 		.proc_handler	= proc_dointvec_jiffies,
988 	},
989 	{
990 		.procname	= "tcp_keepalive_probes",
991 		.data		= &init_net.ipv4.sysctl_tcp_keepalive_probes,
992 		.maxlen		= sizeof(u8),
993 		.mode		= 0644,
994 		.proc_handler	= proc_dou8vec_minmax,
995 	},
996 	{
997 		.procname	= "tcp_keepalive_intvl",
998 		.data		= &init_net.ipv4.sysctl_tcp_keepalive_intvl,
999 		.maxlen		= sizeof(int),
1000 		.mode		= 0644,
1001 		.proc_handler	= proc_dointvec_jiffies,
1002 	},
1003 	{
1004 		.procname	= "tcp_syn_retries",
1005 		.data		= &init_net.ipv4.sysctl_tcp_syn_retries,
1006 		.maxlen		= sizeof(u8),
1007 		.mode		= 0644,
1008 		.proc_handler	= proc_dou8vec_minmax,
1009 		.extra1		= &tcp_syn_retries_min,
1010 		.extra2		= &tcp_syn_retries_max
1011 	},
1012 	{
1013 		.procname	= "tcp_synack_retries",
1014 		.data		= &init_net.ipv4.sysctl_tcp_synack_retries,
1015 		.maxlen		= sizeof(u8),
1016 		.mode		= 0644,
1017 		.proc_handler	= proc_dou8vec_minmax,
1018 	},
1019 #ifdef CONFIG_SYN_COOKIES
1020 	{
1021 		.procname	= "tcp_syncookies",
1022 		.data		= &init_net.ipv4.sysctl_tcp_syncookies,
1023 		.maxlen		= sizeof(u8),
1024 		.mode		= 0644,
1025 		.proc_handler	= proc_dou8vec_minmax,
1026 	},
1027 #endif
1028 	{
1029 		.procname	= "tcp_migrate_req",
1030 		.data		= &init_net.ipv4.sysctl_tcp_migrate_req,
1031 		.maxlen		= sizeof(u8),
1032 		.mode		= 0644,
1033 		.proc_handler	= proc_dou8vec_minmax,
1034 		.extra1		= SYSCTL_ZERO,
1035 		.extra2		= SYSCTL_ONE
1036 	},
1037 	{
1038 		.procname	= "tcp_reordering",
1039 		.data		= &init_net.ipv4.sysctl_tcp_reordering,
1040 		.maxlen		= sizeof(int),
1041 		.mode		= 0644,
1042 		.proc_handler	= proc_dointvec
1043 	},
1044 	{
1045 		.procname	= "tcp_retries1",
1046 		.data		= &init_net.ipv4.sysctl_tcp_retries1,
1047 		.maxlen		= sizeof(u8),
1048 		.mode		= 0644,
1049 		.proc_handler	= proc_dou8vec_minmax,
1050 		.extra2		= &tcp_retr1_max
1051 	},
1052 	{
1053 		.procname	= "tcp_retries2",
1054 		.data		= &init_net.ipv4.sysctl_tcp_retries2,
1055 		.maxlen		= sizeof(u8),
1056 		.mode		= 0644,
1057 		.proc_handler	= proc_dou8vec_minmax,
1058 	},
1059 	{
1060 		.procname	= "tcp_orphan_retries",
1061 		.data		= &init_net.ipv4.sysctl_tcp_orphan_retries,
1062 		.maxlen		= sizeof(u8),
1063 		.mode		= 0644,
1064 		.proc_handler	= proc_dou8vec_minmax,
1065 	},
1066 	{
1067 		.procname	= "tcp_fin_timeout",
1068 		.data		= &init_net.ipv4.sysctl_tcp_fin_timeout,
1069 		.maxlen		= sizeof(int),
1070 		.mode		= 0644,
1071 		.proc_handler	= proc_dointvec_jiffies,
1072 	},
1073 	{
1074 		.procname	= "tcp_notsent_lowat",
1075 		.data		= &init_net.ipv4.sysctl_tcp_notsent_lowat,
1076 		.maxlen		= sizeof(unsigned int),
1077 		.mode		= 0644,
1078 		.proc_handler	= proc_douintvec,
1079 	},
1080 	{
1081 		.procname	= "tcp_tw_reuse",
1082 		.data		= &init_net.ipv4.sysctl_tcp_tw_reuse,
1083 		.maxlen		= sizeof(u8),
1084 		.mode		= 0644,
1085 		.proc_handler	= proc_dou8vec_minmax,
1086 		.extra1		= SYSCTL_ZERO,
1087 		.extra2		= SYSCTL_TWO,
1088 	},
1089 	{
1090 		.procname	= "tcp_tw_reuse_delay",
1091 		.data		= &init_net.ipv4.sysctl_tcp_tw_reuse_delay,
1092 		.maxlen		= sizeof(unsigned int),
1093 		.mode		= 0644,
1094 		.proc_handler	= proc_douintvec_minmax,
1095 		.extra1		= SYSCTL_ONE,
1096 		.extra2		= &tcp_tw_reuse_delay_max,
1097 	},
1098 	{
1099 		.procname	= "tcp_max_syn_backlog",
1100 		.data		= &init_net.ipv4.sysctl_max_syn_backlog,
1101 		.maxlen		= sizeof(int),
1102 		.mode		= 0644,
1103 		.proc_handler	= proc_dointvec
1104 	},
1105 	{
1106 		.procname	= "tcp_fastopen",
1107 		.data		= &init_net.ipv4.sysctl_tcp_fastopen,
1108 		.maxlen		= sizeof(int),
1109 		.mode		= 0644,
1110 		.proc_handler	= proc_dointvec,
1111 	},
1112 	{
1113 		.procname	= "tcp_fastopen_key",
1114 		.mode		= 0600,
1115 		.data		= &init_net.ipv4.sysctl_tcp_fastopen,
1116 		/* maxlen to print the list of keys in hex (*2), with dashes
1117 		 * separating doublewords and a comma in between keys.
1118 		 */
1119 		.maxlen		= ((TCP_FASTOPEN_KEY_LENGTH *
1120 				   2 * TCP_FASTOPEN_KEY_MAX) +
1121 				   (TCP_FASTOPEN_KEY_MAX * 5)),
1122 		.proc_handler	= proc_tcp_fastopen_key,
1123 	},
1124 	{
1125 		.procname	= "tcp_fastopen_blackhole_timeout_sec",
1126 		.data		= &init_net.ipv4.sysctl_tcp_fastopen_blackhole_timeout,
1127 		.maxlen		= sizeof(int),
1128 		.mode		= 0644,
1129 		.proc_handler	= proc_tfo_blackhole_detect_timeout,
1130 		.extra1		= SYSCTL_ZERO,
1131 	},
1132 #ifdef CONFIG_IP_ROUTE_MULTIPATH
1133 	{
1134 		.procname	= "fib_multipath_use_neigh",
1135 		.data		= &init_net.ipv4.sysctl_fib_multipath_use_neigh,
1136 		.maxlen		= sizeof(u8),
1137 		.mode		= 0644,
1138 		.proc_handler	= proc_dou8vec_minmax,
1139 		.extra1		= SYSCTL_ZERO,
1140 		.extra2		= SYSCTL_ONE,
1141 	},
1142 	{
1143 		.procname	= "fib_multipath_hash_policy",
1144 		.data		= &init_net.ipv4.sysctl_fib_multipath_hash_policy,
1145 		.maxlen		= sizeof(u8),
1146 		.mode		= 0644,
1147 		.proc_handler	= proc_fib_multipath_hash_policy,
1148 		.extra1		= SYSCTL_ZERO,
1149 		.extra2		= SYSCTL_THREE,
1150 	},
1151 	{
1152 		.procname	= "fib_multipath_hash_fields",
1153 		.data		= &init_net.ipv4.sysctl_fib_multipath_hash_fields,
1154 		.maxlen		= sizeof(u32),
1155 		.mode		= 0644,
1156 		.proc_handler	= proc_fib_multipath_hash_fields,
1157 		.extra1		= SYSCTL_ONE,
1158 		.extra2		= &fib_multipath_hash_fields_all_mask,
1159 	},
1160 	{
1161 		.procname	= "fib_multipath_hash_seed",
1162 		.data		= &init_net,
1163 		.maxlen		= sizeof(u32),
1164 		.mode		= 0644,
1165 		.proc_handler	= proc_fib_multipath_hash_seed,
1166 	},
1167 #endif
1168 	{
1169 		.procname	= "ip_unprivileged_port_start",
1170 		.maxlen		= sizeof(int),
1171 		.data		= &init_net.ipv4.sysctl_ip_prot_sock,
1172 		.mode		= 0644,
1173 		.proc_handler	= ipv4_privileged_ports,
1174 	},
1175 #ifdef CONFIG_NET_L3_MASTER_DEV
1176 	{
1177 		.procname	= "udp_l3mdev_accept",
1178 		.data		= &init_net.ipv4.sysctl_udp_l3mdev_accept,
1179 		.maxlen		= sizeof(u8),
1180 		.mode		= 0644,
1181 		.proc_handler	= proc_dou8vec_minmax,
1182 		.extra1		= SYSCTL_ZERO,
1183 		.extra2		= SYSCTL_ONE,
1184 	},
1185 #endif
1186 	{
1187 		.procname	= "tcp_sack",
1188 		.data		= &init_net.ipv4.sysctl_tcp_sack,
1189 		.maxlen		= sizeof(u8),
1190 		.mode		= 0644,
1191 		.proc_handler	= proc_dou8vec_minmax,
1192 	},
1193 	{
1194 		.procname	= "tcp_window_scaling",
1195 		.data		= &init_net.ipv4.sysctl_tcp_window_scaling,
1196 		.maxlen		= sizeof(u8),
1197 		.mode		= 0644,
1198 		.proc_handler	= proc_dou8vec_minmax,
1199 	},
1200 	{
1201 		.procname	= "tcp_timestamps",
1202 		.data		= &init_net.ipv4.sysctl_tcp_timestamps,
1203 		.maxlen		= sizeof(u8),
1204 		.mode		= 0644,
1205 		.proc_handler	= proc_dou8vec_minmax,
1206 	},
1207 	{
1208 		.procname	= "tcp_early_retrans",
1209 		.data		= &init_net.ipv4.sysctl_tcp_early_retrans,
1210 		.maxlen		= sizeof(u8),
1211 		.mode		= 0644,
1212 		.proc_handler	= proc_dou8vec_minmax,
1213 		.extra1		= SYSCTL_ZERO,
1214 		.extra2		= SYSCTL_FOUR,
1215 	},
1216 	{
1217 		.procname	= "tcp_recovery",
1218 		.data		= &init_net.ipv4.sysctl_tcp_recovery,
1219 		.maxlen		= sizeof(u8),
1220 		.mode		= 0644,
1221 		.proc_handler	= proc_dou8vec_minmax,
1222 	},
1223 	{
1224 		.procname       = "tcp_thin_linear_timeouts",
1225 		.data           = &init_net.ipv4.sysctl_tcp_thin_linear_timeouts,
1226 		.maxlen         = sizeof(u8),
1227 		.mode           = 0644,
1228 		.proc_handler   = proc_dou8vec_minmax,
1229 	},
1230 	{
1231 		.procname	= "tcp_slow_start_after_idle",
1232 		.data		= &init_net.ipv4.sysctl_tcp_slow_start_after_idle,
1233 		.maxlen		= sizeof(u8),
1234 		.mode		= 0644,
1235 		.proc_handler	= proc_dou8vec_minmax,
1236 	},
1237 	{
1238 		.procname	= "tcp_retrans_collapse",
1239 		.data		= &init_net.ipv4.sysctl_tcp_retrans_collapse,
1240 		.maxlen		= sizeof(u8),
1241 		.mode		= 0644,
1242 		.proc_handler	= proc_dou8vec_minmax,
1243 	},
1244 	{
1245 		.procname	= "tcp_stdurg",
1246 		.data		= &init_net.ipv4.sysctl_tcp_stdurg,
1247 		.maxlen		= sizeof(u8),
1248 		.mode		= 0644,
1249 		.proc_handler	= proc_dou8vec_minmax,
1250 	},
1251 	{
1252 		.procname	= "tcp_rfc1337",
1253 		.data		= &init_net.ipv4.sysctl_tcp_rfc1337,
1254 		.maxlen		= sizeof(u8),
1255 		.mode		= 0644,
1256 		.proc_handler	= proc_dou8vec_minmax,
1257 	},
1258 	{
1259 		.procname	= "tcp_abort_on_overflow",
1260 		.data		= &init_net.ipv4.sysctl_tcp_abort_on_overflow,
1261 		.maxlen		= sizeof(u8),
1262 		.mode		= 0644,
1263 		.proc_handler	= proc_dou8vec_minmax,
1264 	},
1265 	{
1266 		.procname	= "tcp_fack",
1267 		.data		= &init_net.ipv4.sysctl_tcp_fack,
1268 		.maxlen		= sizeof(u8),
1269 		.mode		= 0644,
1270 		.proc_handler	= proc_dou8vec_minmax,
1271 	},
1272 	{
1273 		.procname	= "tcp_max_reordering",
1274 		.data		= &init_net.ipv4.sysctl_tcp_max_reordering,
1275 		.maxlen		= sizeof(int),
1276 		.mode		= 0644,
1277 		.proc_handler	= proc_dointvec
1278 	},
1279 	{
1280 		.procname	= "tcp_dsack",
1281 		.data		= &init_net.ipv4.sysctl_tcp_dsack,
1282 		.maxlen		= sizeof(u8),
1283 		.mode		= 0644,
1284 		.proc_handler	= proc_dou8vec_minmax,
1285 	},
1286 	{
1287 		.procname	= "tcp_app_win",
1288 		.data		= &init_net.ipv4.sysctl_tcp_app_win,
1289 		.maxlen		= sizeof(u8),
1290 		.mode		= 0644,
1291 		.proc_handler	= proc_dou8vec_minmax,
1292 		.extra1		= SYSCTL_ZERO,
1293 		.extra2		= &tcp_app_win_max,
1294 	},
1295 	{
1296 		.procname	= "tcp_adv_win_scale",
1297 		.data		= &init_net.ipv4.sysctl_tcp_adv_win_scale,
1298 		.maxlen		= sizeof(int),
1299 		.mode		= 0644,
1300 		.proc_handler	= proc_dointvec_minmax,
1301 		.extra1		= &tcp_adv_win_scale_min,
1302 		.extra2		= &tcp_adv_win_scale_max,
1303 	},
1304 	{
1305 		.procname	= "tcp_frto",
1306 		.data		= &init_net.ipv4.sysctl_tcp_frto,
1307 		.maxlen		= sizeof(u8),
1308 		.mode		= 0644,
1309 		.proc_handler	= proc_dou8vec_minmax,
1310 	},
1311 	{
1312 		.procname	= "tcp_no_metrics_save",
1313 		.data		= &init_net.ipv4.sysctl_tcp_nometrics_save,
1314 		.maxlen		= sizeof(u8),
1315 		.mode		= 0644,
1316 		.proc_handler	= proc_dou8vec_minmax,
1317 	},
1318 	{
1319 		.procname	= "tcp_no_ssthresh_metrics_save",
1320 		.data		= &init_net.ipv4.sysctl_tcp_no_ssthresh_metrics_save,
1321 		.maxlen		= sizeof(u8),
1322 		.mode		= 0644,
1323 		.proc_handler	= proc_dou8vec_minmax,
1324 		.extra1		= SYSCTL_ZERO,
1325 		.extra2		= SYSCTL_ONE,
1326 	},
1327 	{
1328 		.procname	= "tcp_moderate_rcvbuf",
1329 		.data		= &init_net.ipv4.sysctl_tcp_moderate_rcvbuf,
1330 		.maxlen		= sizeof(u8),
1331 		.mode		= 0644,
1332 		.proc_handler	= proc_dou8vec_minmax,
1333 	},
1334 	{
1335 		.procname	= "tcp_tso_win_divisor",
1336 		.data		= &init_net.ipv4.sysctl_tcp_tso_win_divisor,
1337 		.maxlen		= sizeof(u8),
1338 		.mode		= 0644,
1339 		.proc_handler	= proc_dou8vec_minmax,
1340 	},
1341 	{
1342 		.procname	= "tcp_workaround_signed_windows",
1343 		.data		= &init_net.ipv4.sysctl_tcp_workaround_signed_windows,
1344 		.maxlen		= sizeof(u8),
1345 		.mode		= 0644,
1346 		.proc_handler	= proc_dou8vec_minmax,
1347 	},
1348 	{
1349 		.procname	= "tcp_limit_output_bytes",
1350 		.data		= &init_net.ipv4.sysctl_tcp_limit_output_bytes,
1351 		.maxlen		= sizeof(int),
1352 		.mode		= 0644,
1353 		.proc_handler	= proc_dointvec
1354 	},
1355 	{
1356 		.procname	= "tcp_challenge_ack_limit",
1357 		.data		= &init_net.ipv4.sysctl_tcp_challenge_ack_limit,
1358 		.maxlen		= sizeof(int),
1359 		.mode		= 0644,
1360 		.proc_handler	= proc_dointvec
1361 	},
1362 	{
1363 		.procname	= "tcp_min_tso_segs",
1364 		.data		= &init_net.ipv4.sysctl_tcp_min_tso_segs,
1365 		.maxlen		= sizeof(u8),
1366 		.mode		= 0644,
1367 		.proc_handler	= proc_dou8vec_minmax,
1368 		.extra1		= SYSCTL_ONE,
1369 	},
1370 	{
1371 		.procname	= "tcp_tso_rtt_log",
1372 		.data		= &init_net.ipv4.sysctl_tcp_tso_rtt_log,
1373 		.maxlen		= sizeof(u8),
1374 		.mode		= 0644,
1375 		.proc_handler	= proc_dou8vec_minmax,
1376 	},
1377 	{
1378 		.procname	= "tcp_min_rtt_wlen",
1379 		.data		= &init_net.ipv4.sysctl_tcp_min_rtt_wlen,
1380 		.maxlen		= sizeof(int),
1381 		.mode		= 0644,
1382 		.proc_handler	= proc_dointvec_minmax,
1383 		.extra1		= SYSCTL_ZERO,
1384 		.extra2		= &one_day_secs
1385 	},
1386 	{
1387 		.procname	= "tcp_autocorking",
1388 		.data		= &init_net.ipv4.sysctl_tcp_autocorking,
1389 		.maxlen		= sizeof(u8),
1390 		.mode		= 0644,
1391 		.proc_handler	= proc_dou8vec_minmax,
1392 		.extra1		= SYSCTL_ZERO,
1393 		.extra2		= SYSCTL_ONE,
1394 	},
1395 	{
1396 		.procname	= "tcp_invalid_ratelimit",
1397 		.data		= &init_net.ipv4.sysctl_tcp_invalid_ratelimit,
1398 		.maxlen		= sizeof(int),
1399 		.mode		= 0644,
1400 		.proc_handler	= proc_dointvec_ms_jiffies,
1401 	},
1402 	{
1403 		.procname	= "tcp_pacing_ss_ratio",
1404 		.data		= &init_net.ipv4.sysctl_tcp_pacing_ss_ratio,
1405 		.maxlen		= sizeof(int),
1406 		.mode		= 0644,
1407 		.proc_handler	= proc_dointvec_minmax,
1408 		.extra1		= SYSCTL_ZERO,
1409 		.extra2		= SYSCTL_ONE_THOUSAND,
1410 	},
1411 	{
1412 		.procname	= "tcp_pacing_ca_ratio",
1413 		.data		= &init_net.ipv4.sysctl_tcp_pacing_ca_ratio,
1414 		.maxlen		= sizeof(int),
1415 		.mode		= 0644,
1416 		.proc_handler	= proc_dointvec_minmax,
1417 		.extra1		= SYSCTL_ZERO,
1418 		.extra2		= SYSCTL_ONE_THOUSAND,
1419 	},
1420 	{
1421 		.procname	= "tcp_wmem",
1422 		.data		= &init_net.ipv4.sysctl_tcp_wmem,
1423 		.maxlen		= sizeof(init_net.ipv4.sysctl_tcp_wmem),
1424 		.mode		= 0644,
1425 		.proc_handler	= proc_dointvec_minmax,
1426 		.extra1		= SYSCTL_ONE,
1427 	},
1428 	{
1429 		.procname	= "tcp_rmem",
1430 		.data		= &init_net.ipv4.sysctl_tcp_rmem,
1431 		.maxlen		= sizeof(init_net.ipv4.sysctl_tcp_rmem),
1432 		.mode		= 0644,
1433 		.proc_handler	= proc_dointvec_minmax,
1434 		.extra1		= SYSCTL_ONE,
1435 	},
1436 	{
1437 		.procname	= "tcp_comp_sack_delay_ns",
1438 		.data		= &init_net.ipv4.sysctl_tcp_comp_sack_delay_ns,
1439 		.maxlen		= sizeof(unsigned long),
1440 		.mode		= 0644,
1441 		.proc_handler	= proc_doulongvec_minmax,
1442 	},
1443 	{
1444 		.procname	= "tcp_comp_sack_slack_ns",
1445 		.data		= &init_net.ipv4.sysctl_tcp_comp_sack_slack_ns,
1446 		.maxlen		= sizeof(unsigned long),
1447 		.mode		= 0644,
1448 		.proc_handler	= proc_doulongvec_minmax,
1449 	},
1450 	{
1451 		.procname	= "tcp_comp_sack_nr",
1452 		.data		= &init_net.ipv4.sysctl_tcp_comp_sack_nr,
1453 		.maxlen		= sizeof(u8),
1454 		.mode		= 0644,
1455 		.proc_handler	= proc_dou8vec_minmax,
1456 		.extra1		= SYSCTL_ZERO,
1457 	},
1458 	{
1459 		.procname	= "tcp_backlog_ack_defer",
1460 		.data		= &init_net.ipv4.sysctl_tcp_backlog_ack_defer,
1461 		.maxlen		= sizeof(u8),
1462 		.mode		= 0644,
1463 		.proc_handler	= proc_dou8vec_minmax,
1464 		.extra1		= SYSCTL_ZERO,
1465 		.extra2		= SYSCTL_ONE,
1466 	},
1467 	{
1468 		.procname       = "tcp_reflect_tos",
1469 		.data           = &init_net.ipv4.sysctl_tcp_reflect_tos,
1470 		.maxlen         = sizeof(u8),
1471 		.mode           = 0644,
1472 		.proc_handler   = proc_dou8vec_minmax,
1473 		.extra1         = SYSCTL_ZERO,
1474 		.extra2         = SYSCTL_ONE,
1475 	},
1476 	{
1477 		.procname	= "tcp_ehash_entries",
1478 		.data		= &init_net.ipv4.sysctl_tcp_child_ehash_entries,
1479 		.mode		= 0444,
1480 		.proc_handler	= proc_tcp_ehash_entries,
1481 	},
1482 	{
1483 		.procname	= "tcp_child_ehash_entries",
1484 		.data		= &init_net.ipv4.sysctl_tcp_child_ehash_entries,
1485 		.maxlen		= sizeof(unsigned int),
1486 		.mode		= 0644,
1487 		.proc_handler	= proc_douintvec_minmax,
1488 		.extra1		= SYSCTL_ZERO,
1489 		.extra2		= &tcp_child_ehash_entries_max,
1490 	},
1491 	{
1492 		.procname	= "udp_hash_entries",
1493 		.data		= &init_net.ipv4.sysctl_udp_child_hash_entries,
1494 		.mode		= 0444,
1495 		.proc_handler	= proc_udp_hash_entries,
1496 	},
1497 	{
1498 		.procname	= "udp_child_hash_entries",
1499 		.data		= &init_net.ipv4.sysctl_udp_child_hash_entries,
1500 		.maxlen		= sizeof(unsigned int),
1501 		.mode		= 0644,
1502 		.proc_handler	= proc_douintvec_minmax,
1503 		.extra1		= SYSCTL_ZERO,
1504 		.extra2		= &udp_child_hash_entries_max,
1505 	},
1506 	{
1507 		.procname	= "udp_rmem_min",
1508 		.data		= &init_net.ipv4.sysctl_udp_rmem_min,
1509 		.maxlen		= sizeof(init_net.ipv4.sysctl_udp_rmem_min),
1510 		.mode		= 0644,
1511 		.proc_handler	= proc_dointvec_minmax,
1512 		.extra1		= SYSCTL_ONE
1513 	},
1514 	{
1515 		.procname	= "udp_wmem_min",
1516 		.data		= &init_net.ipv4.sysctl_udp_wmem_min,
1517 		.maxlen		= sizeof(init_net.ipv4.sysctl_udp_wmem_min),
1518 		.mode		= 0644,
1519 		.proc_handler	= proc_dointvec_minmax,
1520 		.extra1		= SYSCTL_ONE
1521 	},
1522 	{
1523 		.procname	= "fib_notify_on_flag_change",
1524 		.data		= &init_net.ipv4.sysctl_fib_notify_on_flag_change,
1525 		.maxlen		= sizeof(u8),
1526 		.mode		= 0644,
1527 		.proc_handler	= proc_dou8vec_minmax,
1528 		.extra1		= SYSCTL_ZERO,
1529 		.extra2		= SYSCTL_TWO,
1530 	},
1531 	{
1532 		.procname       = "tcp_plb_enabled",
1533 		.data           = &init_net.ipv4.sysctl_tcp_plb_enabled,
1534 		.maxlen         = sizeof(u8),
1535 		.mode           = 0644,
1536 		.proc_handler   = proc_dou8vec_minmax,
1537 		.extra1         = SYSCTL_ZERO,
1538 		.extra2         = SYSCTL_ONE,
1539 	},
1540 	{
1541 		.procname       = "tcp_plb_idle_rehash_rounds",
1542 		.data           = &init_net.ipv4.sysctl_tcp_plb_idle_rehash_rounds,
1543 		.maxlen         = sizeof(u8),
1544 		.mode           = 0644,
1545 		.proc_handler   = proc_dou8vec_minmax,
1546 		.extra2		= &tcp_plb_max_rounds,
1547 	},
1548 	{
1549 		.procname       = "tcp_plb_rehash_rounds",
1550 		.data           = &init_net.ipv4.sysctl_tcp_plb_rehash_rounds,
1551 		.maxlen         = sizeof(u8),
1552 		.mode           = 0644,
1553 		.proc_handler   = proc_dou8vec_minmax,
1554 		.extra2         = &tcp_plb_max_rounds,
1555 	},
1556 	{
1557 		.procname       = "tcp_plb_suspend_rto_sec",
1558 		.data           = &init_net.ipv4.sysctl_tcp_plb_suspend_rto_sec,
1559 		.maxlen         = sizeof(u8),
1560 		.mode           = 0644,
1561 		.proc_handler   = proc_dou8vec_minmax,
1562 	},
1563 	{
1564 		.procname       = "tcp_plb_cong_thresh",
1565 		.data           = &init_net.ipv4.sysctl_tcp_plb_cong_thresh,
1566 		.maxlen         = sizeof(int),
1567 		.mode           = 0644,
1568 		.proc_handler   = proc_dointvec_minmax,
1569 		.extra1         = SYSCTL_ZERO,
1570 		.extra2         = &tcp_plb_max_cong_thresh,
1571 	},
1572 	{
1573 		.procname	= "tcp_syn_linear_timeouts",
1574 		.data		= &init_net.ipv4.sysctl_tcp_syn_linear_timeouts,
1575 		.maxlen		= sizeof(u8),
1576 		.mode		= 0644,
1577 		.proc_handler	= proc_dou8vec_minmax,
1578 		.extra1		= SYSCTL_ZERO,
1579 		.extra2		= &tcp_syn_linear_timeouts_max,
1580 	},
1581 	{
1582 		.procname	= "tcp_shrink_window",
1583 		.data		= &init_net.ipv4.sysctl_tcp_shrink_window,
1584 		.maxlen		= sizeof(u8),
1585 		.mode		= 0644,
1586 		.proc_handler	= proc_dou8vec_minmax,
1587 		.extra1		= SYSCTL_ZERO,
1588 		.extra2		= SYSCTL_ONE,
1589 	},
1590 	{
1591 		.procname	= "tcp_pingpong_thresh",
1592 		.data		= &init_net.ipv4.sysctl_tcp_pingpong_thresh,
1593 		.maxlen		= sizeof(u8),
1594 		.mode		= 0644,
1595 		.proc_handler	= proc_dou8vec_minmax,
1596 		.extra1		= SYSCTL_ONE,
1597 	},
1598 	{
1599 		.procname	= "tcp_rto_min_us",
1600 		.data		= &init_net.ipv4.sysctl_tcp_rto_min_us,
1601 		.maxlen		= sizeof(int),
1602 		.mode		= 0644,
1603 		.proc_handler	= proc_dointvec_minmax,
1604 		.extra1		= SYSCTL_ONE,
1605 	},
1606 	{
1607 		.procname	= "tcp_rto_max_ms",
1608 		.data		= &init_net.ipv4.sysctl_tcp_rto_max_ms,
1609 		.maxlen		= sizeof(int),
1610 		.mode		= 0644,
1611 		.proc_handler	= proc_dointvec_minmax,
1612 		.extra1		= SYSCTL_ONE_THOUSAND,
1613 		.extra2		= &tcp_rto_max_max,
1614 	},
1615 };
1616 
1617 static __net_init int ipv4_sysctl_init_net(struct net *net)
1618 {
1619 	size_t table_size = ARRAY_SIZE(ipv4_net_table);
1620 	struct ctl_table *table;
1621 
1622 	table = ipv4_net_table;
1623 	if (!net_eq(net, &init_net)) {
1624 		int i;
1625 
1626 		table = kmemdup(table, sizeof(ipv4_net_table), GFP_KERNEL);
1627 		if (!table)
1628 			goto err_alloc;
1629 
1630 		for (i = 0; i < table_size; i++) {
1631 			if (table[i].data) {
1632 				/* Update the variables to point into
1633 				 * the current struct net
1634 				 */
1635 				table[i].data += (void *)net - (void *)&init_net;
1636 			} else {
1637 				/* Entries without data pointer are global;
1638 				 * Make them read-only in non-init_net ns
1639 				 */
1640 				table[i].mode &= ~0222;
1641 			}
1642 		}
1643 	}
1644 
1645 	net->ipv4.ipv4_hdr = register_net_sysctl_sz(net, "net/ipv4", table,
1646 						    table_size);
1647 	if (!net->ipv4.ipv4_hdr)
1648 		goto err_reg;
1649 
1650 	net->ipv4.sysctl_local_reserved_ports = kzalloc(65536 / 8, GFP_KERNEL);
1651 	if (!net->ipv4.sysctl_local_reserved_ports)
1652 		goto err_ports;
1653 
1654 	proc_fib_multipath_hash_set_seed(net, 0);
1655 
1656 	return 0;
1657 
1658 err_ports:
1659 	unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
1660 err_reg:
1661 	if (!net_eq(net, &init_net))
1662 		kfree(table);
1663 err_alloc:
1664 	return -ENOMEM;
1665 }
1666 
1667 static __net_exit void ipv4_sysctl_exit_net(struct net *net)
1668 {
1669 	const struct ctl_table *table;
1670 
1671 	kfree(net->ipv4.sysctl_local_reserved_ports);
1672 	table = net->ipv4.ipv4_hdr->ctl_table_arg;
1673 	unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
1674 	kfree(table);
1675 }
1676 
1677 static __net_initdata struct pernet_operations ipv4_sysctl_ops = {
1678 	.init = ipv4_sysctl_init_net,
1679 	.exit = ipv4_sysctl_exit_net,
1680 };
1681 
1682 static __init int sysctl_ipv4_init(void)
1683 {
1684 	struct ctl_table_header *hdr;
1685 
1686 	hdr = register_net_sysctl(&init_net, "net/ipv4", ipv4_table);
1687 	if (!hdr)
1688 		return -ENOMEM;
1689 
1690 	proc_fib_multipath_hash_init_rand_seed();
1691 
1692 	if (register_pernet_subsys(&ipv4_sysctl_ops)) {
1693 		unregister_net_sysctl_table(hdr);
1694 		return -ENOMEM;
1695 	}
1696 
1697 	return 0;
1698 }
1699 
1700 __initcall(sysctl_ipv4_init);
1701