xref: /openbmc/linux/net/ipv4/sysctl_net_ipv4.c (revision ccc319dc)
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_min_snd_mss_min = TCP_MIN_SND_MSS;
29 static int tcp_min_snd_mss_max = 65535;
30 static int ip_privileged_port_min;
31 static int ip_privileged_port_max = 65535;
32 static int ip_ttl_min = 1;
33 static int ip_ttl_max = 255;
34 static int tcp_syn_retries_min = 1;
35 static int tcp_syn_retries_max = MAX_TCP_SYNCNT;
36 static int ip_ping_group_range_min[] = { 0, 0 };
37 static int ip_ping_group_range_max[] = { GID_T_MAX, GID_T_MAX };
38 static u32 u32_max_div_HZ = UINT_MAX / HZ;
39 static int one_day_secs = 24 * 3600;
40 static u32 fib_multipath_hash_fields_all_mask __maybe_unused =
41 	FIB_MULTIPATH_HASH_FIELD_ALL_MASK;
42 
43 /* obsolete */
44 static int sysctl_tcp_low_latency __read_mostly;
45 
46 /* Update system visible IP port range */
47 static void set_local_port_range(struct net *net, int range[2])
48 {
49 	bool same_parity = !((range[0] ^ range[1]) & 1);
50 
51 	write_seqlock_bh(&net->ipv4.ip_local_ports.lock);
52 	if (same_parity && !net->ipv4.ip_local_ports.warned) {
53 		net->ipv4.ip_local_ports.warned = true;
54 		pr_err_ratelimited("ip_local_port_range: prefer different parity for start/end values.\n");
55 	}
56 	net->ipv4.ip_local_ports.range[0] = range[0];
57 	net->ipv4.ip_local_ports.range[1] = range[1];
58 	write_sequnlock_bh(&net->ipv4.ip_local_ports.lock);
59 }
60 
61 /* Validate changes from /proc interface. */
62 static int ipv4_local_port_range(struct ctl_table *table, int write,
63 				 void *buffer, size_t *lenp, loff_t *ppos)
64 {
65 	struct net *net =
66 		container_of(table->data, struct net, ipv4.ip_local_ports.range);
67 	int ret;
68 	int range[2];
69 	struct ctl_table tmp = {
70 		.data = &range,
71 		.maxlen = sizeof(range),
72 		.mode = table->mode,
73 		.extra1 = &ip_local_port_range_min,
74 		.extra2 = &ip_local_port_range_max,
75 	};
76 
77 	inet_get_local_port_range(net, &range[0], &range[1]);
78 
79 	ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
80 
81 	if (write && ret == 0) {
82 		/* Ensure that the upper limit is not smaller than the lower,
83 		 * and that the lower does not encroach upon the privileged
84 		 * port limit.
85 		 */
86 		if ((range[1] < range[0]) ||
87 		    (range[0] < net->ipv4.sysctl_ip_prot_sock))
88 			ret = -EINVAL;
89 		else
90 			set_local_port_range(net, range);
91 	}
92 
93 	return ret;
94 }
95 
96 /* Validate changes from /proc interface. */
97 static int ipv4_privileged_ports(struct ctl_table *table, int write,
98 				void *buffer, size_t *lenp, loff_t *ppos)
99 {
100 	struct net *net = container_of(table->data, struct net,
101 	    ipv4.sysctl_ip_prot_sock);
102 	int ret;
103 	int pports;
104 	int range[2];
105 	struct ctl_table tmp = {
106 		.data = &pports,
107 		.maxlen = sizeof(pports),
108 		.mode = table->mode,
109 		.extra1 = &ip_privileged_port_min,
110 		.extra2 = &ip_privileged_port_max,
111 	};
112 
113 	pports = net->ipv4.sysctl_ip_prot_sock;
114 
115 	ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
116 
117 	if (write && ret == 0) {
118 		inet_get_local_port_range(net, &range[0], &range[1]);
119 		/* Ensure that the local port range doesn't overlap with the
120 		 * privileged port range.
121 		 */
122 		if (range[0] < pports)
123 			ret = -EINVAL;
124 		else
125 			net->ipv4.sysctl_ip_prot_sock = pports;
126 	}
127 
128 	return ret;
129 }
130 
131 static void inet_get_ping_group_range_table(struct ctl_table *table, kgid_t *low, kgid_t *high)
132 {
133 	kgid_t *data = table->data;
134 	struct net *net =
135 		container_of(table->data, struct net, ipv4.ping_group_range.range);
136 	unsigned int seq;
137 	do {
138 		seq = read_seqbegin(&net->ipv4.ping_group_range.lock);
139 
140 		*low = data[0];
141 		*high = data[1];
142 	} while (read_seqretry(&net->ipv4.ping_group_range.lock, seq));
143 }
144 
145 /* Update system visible IP port range */
146 static void set_ping_group_range(struct ctl_table *table, kgid_t low, kgid_t high)
147 {
148 	kgid_t *data = table->data;
149 	struct net *net =
150 		container_of(table->data, struct net, ipv4.ping_group_range.range);
151 	write_seqlock(&net->ipv4.ping_group_range.lock);
152 	data[0] = low;
153 	data[1] = high;
154 	write_sequnlock(&net->ipv4.ping_group_range.lock);
155 }
156 
157 /* Validate changes from /proc interface. */
158 static int ipv4_ping_group_range(struct ctl_table *table, int write,
159 				 void *buffer, size_t *lenp, loff_t *ppos)
160 {
161 	struct user_namespace *user_ns = current_user_ns();
162 	int ret;
163 	gid_t urange[2];
164 	kgid_t low, high;
165 	struct ctl_table tmp = {
166 		.data = &urange,
167 		.maxlen = sizeof(urange),
168 		.mode = table->mode,
169 		.extra1 = &ip_ping_group_range_min,
170 		.extra2 = &ip_ping_group_range_max,
171 	};
172 
173 	inet_get_ping_group_range_table(table, &low, &high);
174 	urange[0] = from_kgid_munged(user_ns, low);
175 	urange[1] = from_kgid_munged(user_ns, high);
176 	ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
177 
178 	if (write && ret == 0) {
179 		low = make_kgid(user_ns, urange[0]);
180 		high = make_kgid(user_ns, urange[1]);
181 		if (!gid_valid(low) || !gid_valid(high))
182 			return -EINVAL;
183 		if (urange[1] < urange[0] || gid_lt(high, low)) {
184 			low = make_kgid(&init_user_ns, 1);
185 			high = make_kgid(&init_user_ns, 0);
186 		}
187 		set_ping_group_range(table, low, high);
188 	}
189 
190 	return ret;
191 }
192 
193 static int ipv4_fwd_update_priority(struct ctl_table *table, int write,
194 				    void *buffer, size_t *lenp, loff_t *ppos)
195 {
196 	struct net *net;
197 	int ret;
198 
199 	net = container_of(table->data, struct net,
200 			   ipv4.sysctl_ip_fwd_update_priority);
201 	ret = proc_dou8vec_minmax(table, write, buffer, lenp, ppos);
202 	if (write && ret == 0)
203 		call_netevent_notifiers(NETEVENT_IPV4_FWD_UPDATE_PRIORITY_UPDATE,
204 					net);
205 
206 	return ret;
207 }
208 
209 static int proc_tcp_congestion_control(struct ctl_table *ctl, int write,
210 				       void *buffer, size_t *lenp, loff_t *ppos)
211 {
212 	struct net *net = container_of(ctl->data, struct net,
213 				       ipv4.tcp_congestion_control);
214 	char val[TCP_CA_NAME_MAX];
215 	struct ctl_table tbl = {
216 		.data = val,
217 		.maxlen = TCP_CA_NAME_MAX,
218 	};
219 	int ret;
220 
221 	tcp_get_default_congestion_control(net, val);
222 
223 	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
224 	if (write && ret == 0)
225 		ret = tcp_set_default_congestion_control(net, val);
226 	return ret;
227 }
228 
229 static int proc_tcp_available_congestion_control(struct ctl_table *ctl,
230 						 int write, void *buffer,
231 						 size_t *lenp, loff_t *ppos)
232 {
233 	struct ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX, };
234 	int ret;
235 
236 	tbl.data = kmalloc(tbl.maxlen, GFP_USER);
237 	if (!tbl.data)
238 		return -ENOMEM;
239 	tcp_get_available_congestion_control(tbl.data, TCP_CA_BUF_MAX);
240 	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
241 	kfree(tbl.data);
242 	return ret;
243 }
244 
245 static int proc_allowed_congestion_control(struct ctl_table *ctl,
246 					   int write, void *buffer,
247 					   size_t *lenp, loff_t *ppos)
248 {
249 	struct ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX };
250 	int ret;
251 
252 	tbl.data = kmalloc(tbl.maxlen, GFP_USER);
253 	if (!tbl.data)
254 		return -ENOMEM;
255 
256 	tcp_get_allowed_congestion_control(tbl.data, tbl.maxlen);
257 	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
258 	if (write && ret == 0)
259 		ret = tcp_set_allowed_congestion_control(tbl.data);
260 	kfree(tbl.data);
261 	return ret;
262 }
263 
264 static int sscanf_key(char *buf, __le32 *key)
265 {
266 	u32 user_key[4];
267 	int i, ret = 0;
268 
269 	if (sscanf(buf, "%x-%x-%x-%x", user_key, user_key + 1,
270 		   user_key + 2, user_key + 3) != 4) {
271 		ret = -EINVAL;
272 	} else {
273 		for (i = 0; i < ARRAY_SIZE(user_key); i++)
274 			key[i] = cpu_to_le32(user_key[i]);
275 	}
276 	pr_debug("proc TFO key set 0x%x-%x-%x-%x <- 0x%s: %u\n",
277 		 user_key[0], user_key[1], user_key[2], user_key[3], buf, ret);
278 
279 	return ret;
280 }
281 
282 static int proc_tcp_fastopen_key(struct ctl_table *table, int write,
283 				 void *buffer, size_t *lenp, loff_t *ppos)
284 {
285 	struct net *net = container_of(table->data, struct net,
286 	    ipv4.sysctl_tcp_fastopen);
287 	/* maxlen to print the list of keys in hex (*2), with dashes
288 	 * separating doublewords and a comma in between keys.
289 	 */
290 	struct ctl_table tbl = { .maxlen = ((TCP_FASTOPEN_KEY_LENGTH *
291 					    2 * TCP_FASTOPEN_KEY_MAX) +
292 					    (TCP_FASTOPEN_KEY_MAX * 5)) };
293 	u32 user_key[TCP_FASTOPEN_KEY_BUF_LENGTH / sizeof(u32)];
294 	__le32 key[TCP_FASTOPEN_KEY_BUF_LENGTH / sizeof(__le32)];
295 	char *backup_data;
296 	int ret, i = 0, off = 0, n_keys;
297 
298 	tbl.data = kmalloc(tbl.maxlen, GFP_KERNEL);
299 	if (!tbl.data)
300 		return -ENOMEM;
301 
302 	n_keys = tcp_fastopen_get_cipher(net, NULL, (u64 *)key);
303 	if (!n_keys) {
304 		memset(&key[0], 0, TCP_FASTOPEN_KEY_LENGTH);
305 		n_keys = 1;
306 	}
307 
308 	for (i = 0; i < n_keys * 4; i++)
309 		user_key[i] = le32_to_cpu(key[i]);
310 
311 	for (i = 0; i < n_keys; i++) {
312 		off += snprintf(tbl.data + off, tbl.maxlen - off,
313 				"%08x-%08x-%08x-%08x",
314 				user_key[i * 4],
315 				user_key[i * 4 + 1],
316 				user_key[i * 4 + 2],
317 				user_key[i * 4 + 3]);
318 
319 		if (WARN_ON_ONCE(off >= tbl.maxlen - 1))
320 			break;
321 
322 		if (i + 1 < n_keys)
323 			off += snprintf(tbl.data + off, tbl.maxlen - off, ",");
324 	}
325 
326 	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
327 
328 	if (write && ret == 0) {
329 		backup_data = strchr(tbl.data, ',');
330 		if (backup_data) {
331 			*backup_data = '\0';
332 			backup_data++;
333 		}
334 		if (sscanf_key(tbl.data, key)) {
335 			ret = -EINVAL;
336 			goto bad_key;
337 		}
338 		if (backup_data) {
339 			if (sscanf_key(backup_data, key + 4)) {
340 				ret = -EINVAL;
341 				goto bad_key;
342 			}
343 		}
344 		tcp_fastopen_reset_cipher(net, NULL, key,
345 					  backup_data ? key + 4 : NULL);
346 	}
347 
348 bad_key:
349 	kfree(tbl.data);
350 	return ret;
351 }
352 
353 static void proc_configure_early_demux(int enabled, int protocol)
354 {
355 	struct net_protocol *ipprot;
356 #if IS_ENABLED(CONFIG_IPV6)
357 	struct inet6_protocol *ip6prot;
358 #endif
359 
360 	rcu_read_lock();
361 
362 	ipprot = rcu_dereference(inet_protos[protocol]);
363 	if (ipprot)
364 		ipprot->early_demux = enabled ? ipprot->early_demux_handler :
365 						NULL;
366 
367 #if IS_ENABLED(CONFIG_IPV6)
368 	ip6prot = rcu_dereference(inet6_protos[protocol]);
369 	if (ip6prot)
370 		ip6prot->early_demux = enabled ? ip6prot->early_demux_handler :
371 						 NULL;
372 #endif
373 	rcu_read_unlock();
374 }
375 
376 static int proc_tcp_early_demux(struct ctl_table *table, int write,
377 				void *buffer, size_t *lenp, loff_t *ppos)
378 {
379 	int ret = 0;
380 
381 	ret = proc_dou8vec_minmax(table, write, buffer, lenp, ppos);
382 
383 	if (write && !ret) {
384 		int enabled = init_net.ipv4.sysctl_tcp_early_demux;
385 
386 		proc_configure_early_demux(enabled, IPPROTO_TCP);
387 	}
388 
389 	return ret;
390 }
391 
392 static int proc_udp_early_demux(struct ctl_table *table, int write,
393 				void *buffer, size_t *lenp, loff_t *ppos)
394 {
395 	int ret = 0;
396 
397 	ret = proc_dou8vec_minmax(table, write, buffer, lenp, ppos);
398 
399 	if (write && !ret) {
400 		int enabled = init_net.ipv4.sysctl_udp_early_demux;
401 
402 		proc_configure_early_demux(enabled, IPPROTO_UDP);
403 	}
404 
405 	return ret;
406 }
407 
408 static int proc_tfo_blackhole_detect_timeout(struct ctl_table *table,
409 					     int write, void *buffer,
410 					     size_t *lenp, loff_t *ppos)
411 {
412 	struct net *net = container_of(table->data, struct net,
413 	    ipv4.sysctl_tcp_fastopen_blackhole_timeout);
414 	int ret;
415 
416 	ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
417 	if (write && ret == 0)
418 		atomic_set(&net->ipv4.tfo_active_disable_times, 0);
419 
420 	return ret;
421 }
422 
423 static int proc_tcp_available_ulp(struct ctl_table *ctl,
424 				  int write, void *buffer, size_t *lenp,
425 				  loff_t *ppos)
426 {
427 	struct ctl_table tbl = { .maxlen = TCP_ULP_BUF_MAX, };
428 	int ret;
429 
430 	tbl.data = kmalloc(tbl.maxlen, GFP_USER);
431 	if (!tbl.data)
432 		return -ENOMEM;
433 	tcp_get_available_ulp(tbl.data, TCP_ULP_BUF_MAX);
434 	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
435 	kfree(tbl.data);
436 
437 	return ret;
438 }
439 
440 #ifdef CONFIG_IP_ROUTE_MULTIPATH
441 static int proc_fib_multipath_hash_policy(struct ctl_table *table, int write,
442 					  void *buffer, size_t *lenp,
443 					  loff_t *ppos)
444 {
445 	struct net *net = container_of(table->data, struct net,
446 	    ipv4.sysctl_fib_multipath_hash_policy);
447 	int ret;
448 
449 	ret = proc_dou8vec_minmax(table, write, buffer, lenp, ppos);
450 	if (write && ret == 0)
451 		call_netevent_notifiers(NETEVENT_IPV4_MPATH_HASH_UPDATE, net);
452 
453 	return ret;
454 }
455 
456 static int proc_fib_multipath_hash_fields(struct ctl_table *table, int write,
457 					  void *buffer, size_t *lenp,
458 					  loff_t *ppos)
459 {
460 	struct net *net;
461 	int ret;
462 
463 	net = container_of(table->data, struct net,
464 			   ipv4.sysctl_fib_multipath_hash_fields);
465 	ret = proc_douintvec_minmax(table, write, buffer, lenp, ppos);
466 	if (write && ret == 0)
467 		call_netevent_notifiers(NETEVENT_IPV4_MPATH_HASH_UPDATE, net);
468 
469 	return ret;
470 }
471 #endif
472 
473 static struct ctl_table ipv4_table[] = {
474 	{
475 		.procname	= "tcp_max_orphans",
476 		.data		= &sysctl_tcp_max_orphans,
477 		.maxlen		= sizeof(int),
478 		.mode		= 0644,
479 		.proc_handler	= proc_dointvec
480 	},
481 	{
482 		.procname	= "inet_peer_threshold",
483 		.data		= &inet_peer_threshold,
484 		.maxlen		= sizeof(int),
485 		.mode		= 0644,
486 		.proc_handler	= proc_dointvec
487 	},
488 	{
489 		.procname	= "inet_peer_minttl",
490 		.data		= &inet_peer_minttl,
491 		.maxlen		= sizeof(int),
492 		.mode		= 0644,
493 		.proc_handler	= proc_dointvec_jiffies,
494 	},
495 	{
496 		.procname	= "inet_peer_maxttl",
497 		.data		= &inet_peer_maxttl,
498 		.maxlen		= sizeof(int),
499 		.mode		= 0644,
500 		.proc_handler	= proc_dointvec_jiffies,
501 	},
502 	{
503 		.procname	= "tcp_mem",
504 		.maxlen		= sizeof(sysctl_tcp_mem),
505 		.data		= &sysctl_tcp_mem,
506 		.mode		= 0644,
507 		.proc_handler	= proc_doulongvec_minmax,
508 	},
509 	{
510 		.procname	= "tcp_low_latency",
511 		.data		= &sysctl_tcp_low_latency,
512 		.maxlen		= sizeof(int),
513 		.mode		= 0644,
514 		.proc_handler	= proc_dointvec
515 	},
516 #ifdef CONFIG_NETLABEL
517 	{
518 		.procname	= "cipso_cache_enable",
519 		.data		= &cipso_v4_cache_enabled,
520 		.maxlen		= sizeof(int),
521 		.mode		= 0644,
522 		.proc_handler	= proc_dointvec,
523 	},
524 	{
525 		.procname	= "cipso_cache_bucket_size",
526 		.data		= &cipso_v4_cache_bucketsize,
527 		.maxlen		= sizeof(int),
528 		.mode		= 0644,
529 		.proc_handler	= proc_dointvec,
530 	},
531 	{
532 		.procname	= "cipso_rbm_optfmt",
533 		.data		= &cipso_v4_rbm_optfmt,
534 		.maxlen		= sizeof(int),
535 		.mode		= 0644,
536 		.proc_handler	= proc_dointvec,
537 	},
538 	{
539 		.procname	= "cipso_rbm_strictvalid",
540 		.data		= &cipso_v4_rbm_strictvalid,
541 		.maxlen		= sizeof(int),
542 		.mode		= 0644,
543 		.proc_handler	= proc_dointvec,
544 	},
545 #endif /* CONFIG_NETLABEL */
546 	{
547 		.procname	= "tcp_available_ulp",
548 		.maxlen		= TCP_ULP_BUF_MAX,
549 		.mode		= 0444,
550 		.proc_handler   = proc_tcp_available_ulp,
551 	},
552 	{
553 		.procname	= "icmp_msgs_per_sec",
554 		.data		= &sysctl_icmp_msgs_per_sec,
555 		.maxlen		= sizeof(int),
556 		.mode		= 0644,
557 		.proc_handler	= proc_dointvec_minmax,
558 		.extra1		= SYSCTL_ZERO,
559 	},
560 	{
561 		.procname	= "icmp_msgs_burst",
562 		.data		= &sysctl_icmp_msgs_burst,
563 		.maxlen		= sizeof(int),
564 		.mode		= 0644,
565 		.proc_handler	= proc_dointvec_minmax,
566 		.extra1		= SYSCTL_ZERO,
567 	},
568 	{
569 		.procname	= "udp_mem",
570 		.data		= &sysctl_udp_mem,
571 		.maxlen		= sizeof(sysctl_udp_mem),
572 		.mode		= 0644,
573 		.proc_handler	= proc_doulongvec_minmax,
574 	},
575 	{
576 		.procname	= "fib_sync_mem",
577 		.data		= &sysctl_fib_sync_mem,
578 		.maxlen		= sizeof(sysctl_fib_sync_mem),
579 		.mode		= 0644,
580 		.proc_handler	= proc_douintvec_minmax,
581 		.extra1		= &sysctl_fib_sync_mem_min,
582 		.extra2		= &sysctl_fib_sync_mem_max,
583 	},
584 	{ }
585 };
586 
587 static struct ctl_table ipv4_net_table[] = {
588 	/* tcp_max_tw_buckets must be first in this table. */
589 	{
590 		.procname	= "tcp_max_tw_buckets",
591 /*		.data		= &init_net.ipv4.tcp_death_row.sysctl_max_tw_buckets, */
592 		.maxlen		= sizeof(int),
593 		.mode		= 0644,
594 		.proc_handler	= proc_dointvec
595 	},
596 	{
597 		.procname	= "icmp_echo_ignore_all",
598 		.data		= &init_net.ipv4.sysctl_icmp_echo_ignore_all,
599 		.maxlen		= sizeof(u8),
600 		.mode		= 0644,
601 		.proc_handler	= proc_dou8vec_minmax,
602 	},
603 	{
604 		.procname	= "icmp_echo_enable_probe",
605 		.data		= &init_net.ipv4.sysctl_icmp_echo_enable_probe,
606 		.maxlen		= sizeof(u8),
607 		.mode		= 0644,
608 		.proc_handler	= proc_dou8vec_minmax,
609 		.extra1		= SYSCTL_ZERO,
610 		.extra2		= SYSCTL_ONE
611 	},
612 	{
613 		.procname	= "icmp_echo_ignore_broadcasts",
614 		.data		= &init_net.ipv4.sysctl_icmp_echo_ignore_broadcasts,
615 		.maxlen		= sizeof(u8),
616 		.mode		= 0644,
617 		.proc_handler	= proc_dou8vec_minmax,
618 	},
619 	{
620 		.procname	= "icmp_ignore_bogus_error_responses",
621 		.data		= &init_net.ipv4.sysctl_icmp_ignore_bogus_error_responses,
622 		.maxlen		= sizeof(u8),
623 		.mode		= 0644,
624 		.proc_handler	= proc_dou8vec_minmax,
625 	},
626 	{
627 		.procname	= "icmp_errors_use_inbound_ifaddr",
628 		.data		= &init_net.ipv4.sysctl_icmp_errors_use_inbound_ifaddr,
629 		.maxlen		= sizeof(u8),
630 		.mode		= 0644,
631 		.proc_handler	= proc_dou8vec_minmax,
632 	},
633 	{
634 		.procname	= "icmp_ratelimit",
635 		.data		= &init_net.ipv4.sysctl_icmp_ratelimit,
636 		.maxlen		= sizeof(int),
637 		.mode		= 0644,
638 		.proc_handler	= proc_dointvec_ms_jiffies,
639 	},
640 	{
641 		.procname	= "icmp_ratemask",
642 		.data		= &init_net.ipv4.sysctl_icmp_ratemask,
643 		.maxlen		= sizeof(int),
644 		.mode		= 0644,
645 		.proc_handler	= proc_dointvec
646 	},
647 	{
648 		.procname	= "ping_group_range",
649 		.data		= &init_net.ipv4.ping_group_range.range,
650 		.maxlen		= sizeof(gid_t)*2,
651 		.mode		= 0644,
652 		.proc_handler	= ipv4_ping_group_range,
653 	},
654 #ifdef CONFIG_NET_L3_MASTER_DEV
655 	{
656 		.procname	= "raw_l3mdev_accept",
657 		.data		= &init_net.ipv4.sysctl_raw_l3mdev_accept,
658 		.maxlen		= sizeof(u8),
659 		.mode		= 0644,
660 		.proc_handler	= proc_dou8vec_minmax,
661 		.extra1		= SYSCTL_ZERO,
662 		.extra2		= SYSCTL_ONE,
663 	},
664 #endif
665 	{
666 		.procname	= "tcp_ecn",
667 		.data		= &init_net.ipv4.sysctl_tcp_ecn,
668 		.maxlen		= sizeof(u8),
669 		.mode		= 0644,
670 		.proc_handler	= proc_dou8vec_minmax,
671 	},
672 	{
673 		.procname	= "tcp_ecn_fallback",
674 		.data		= &init_net.ipv4.sysctl_tcp_ecn_fallback,
675 		.maxlen		= sizeof(u8),
676 		.mode		= 0644,
677 		.proc_handler	= proc_dou8vec_minmax,
678 	},
679 	{
680 		.procname	= "ip_dynaddr",
681 		.data		= &init_net.ipv4.sysctl_ip_dynaddr,
682 		.maxlen		= sizeof(u8),
683 		.mode		= 0644,
684 		.proc_handler	= proc_dou8vec_minmax,
685 	},
686 	{
687 		.procname	= "ip_early_demux",
688 		.data		= &init_net.ipv4.sysctl_ip_early_demux,
689 		.maxlen		= sizeof(u8),
690 		.mode		= 0644,
691 		.proc_handler	= proc_dou8vec_minmax,
692 	},
693 	{
694 		.procname       = "udp_early_demux",
695 		.data           = &init_net.ipv4.sysctl_udp_early_demux,
696 		.maxlen         = sizeof(u8),
697 		.mode           = 0644,
698 		.proc_handler   = proc_udp_early_demux
699 	},
700 	{
701 		.procname       = "tcp_early_demux",
702 		.data           = &init_net.ipv4.sysctl_tcp_early_demux,
703 		.maxlen         = sizeof(u8),
704 		.mode           = 0644,
705 		.proc_handler   = proc_tcp_early_demux
706 	},
707 	{
708 		.procname       = "nexthop_compat_mode",
709 		.data           = &init_net.ipv4.sysctl_nexthop_compat_mode,
710 		.maxlen         = sizeof(u8),
711 		.mode           = 0644,
712 		.proc_handler   = proc_dou8vec_minmax,
713 		.extra1		= SYSCTL_ZERO,
714 		.extra2		= SYSCTL_ONE,
715 	},
716 	{
717 		.procname	= "ip_default_ttl",
718 		.data		= &init_net.ipv4.sysctl_ip_default_ttl,
719 		.maxlen		= sizeof(u8),
720 		.mode		= 0644,
721 		.proc_handler	= proc_dou8vec_minmax,
722 		.extra1		= &ip_ttl_min,
723 		.extra2		= &ip_ttl_max,
724 	},
725 	{
726 		.procname	= "ip_local_port_range",
727 		.maxlen		= sizeof(init_net.ipv4.ip_local_ports.range),
728 		.data		= &init_net.ipv4.ip_local_ports.range,
729 		.mode		= 0644,
730 		.proc_handler	= ipv4_local_port_range,
731 	},
732 	{
733 		.procname	= "ip_local_reserved_ports",
734 		.data		= &init_net.ipv4.sysctl_local_reserved_ports,
735 		.maxlen		= 65536,
736 		.mode		= 0644,
737 		.proc_handler	= proc_do_large_bitmap,
738 	},
739 	{
740 		.procname	= "ip_no_pmtu_disc",
741 		.data		= &init_net.ipv4.sysctl_ip_no_pmtu_disc,
742 		.maxlen		= sizeof(u8),
743 		.mode		= 0644,
744 		.proc_handler	= proc_dou8vec_minmax,
745 	},
746 	{
747 		.procname	= "ip_forward_use_pmtu",
748 		.data		= &init_net.ipv4.sysctl_ip_fwd_use_pmtu,
749 		.maxlen		= sizeof(u8),
750 		.mode		= 0644,
751 		.proc_handler	= proc_dou8vec_minmax,
752 	},
753 	{
754 		.procname	= "ip_forward_update_priority",
755 		.data		= &init_net.ipv4.sysctl_ip_fwd_update_priority,
756 		.maxlen		= sizeof(u8),
757 		.mode		= 0644,
758 		.proc_handler   = ipv4_fwd_update_priority,
759 		.extra1		= SYSCTL_ZERO,
760 		.extra2		= SYSCTL_ONE,
761 	},
762 	{
763 		.procname	= "ip_nonlocal_bind",
764 		.data		= &init_net.ipv4.sysctl_ip_nonlocal_bind,
765 		.maxlen		= sizeof(u8),
766 		.mode		= 0644,
767 		.proc_handler	= proc_dou8vec_minmax,
768 	},
769 	{
770 		.procname	= "ip_autobind_reuse",
771 		.data		= &init_net.ipv4.sysctl_ip_autobind_reuse,
772 		.maxlen		= sizeof(u8),
773 		.mode		= 0644,
774 		.proc_handler	= proc_dou8vec_minmax,
775 		.extra1         = SYSCTL_ZERO,
776 		.extra2         = SYSCTL_ONE,
777 	},
778 	{
779 		.procname	= "fwmark_reflect",
780 		.data		= &init_net.ipv4.sysctl_fwmark_reflect,
781 		.maxlen		= sizeof(u8),
782 		.mode		= 0644,
783 		.proc_handler	= proc_dou8vec_minmax,
784 	},
785 	{
786 		.procname	= "tcp_fwmark_accept",
787 		.data		= &init_net.ipv4.sysctl_tcp_fwmark_accept,
788 		.maxlen		= sizeof(u8),
789 		.mode		= 0644,
790 		.proc_handler	= proc_dou8vec_minmax,
791 	},
792 #ifdef CONFIG_NET_L3_MASTER_DEV
793 	{
794 		.procname	= "tcp_l3mdev_accept",
795 		.data		= &init_net.ipv4.sysctl_tcp_l3mdev_accept,
796 		.maxlen		= sizeof(u8),
797 		.mode		= 0644,
798 		.proc_handler	= proc_dou8vec_minmax,
799 		.extra1		= SYSCTL_ZERO,
800 		.extra2		= SYSCTL_ONE,
801 	},
802 #endif
803 	{
804 		.procname	= "tcp_mtu_probing",
805 		.data		= &init_net.ipv4.sysctl_tcp_mtu_probing,
806 		.maxlen		= sizeof(u8),
807 		.mode		= 0644,
808 		.proc_handler	= proc_dou8vec_minmax,
809 	},
810 	{
811 		.procname	= "tcp_base_mss",
812 		.data		= &init_net.ipv4.sysctl_tcp_base_mss,
813 		.maxlen		= sizeof(int),
814 		.mode		= 0644,
815 		.proc_handler	= proc_dointvec,
816 	},
817 	{
818 		.procname	= "tcp_min_snd_mss",
819 		.data		= &init_net.ipv4.sysctl_tcp_min_snd_mss,
820 		.maxlen		= sizeof(int),
821 		.mode		= 0644,
822 		.proc_handler	= proc_dointvec_minmax,
823 		.extra1		= &tcp_min_snd_mss_min,
824 		.extra2		= &tcp_min_snd_mss_max,
825 	},
826 	{
827 		.procname	= "tcp_mtu_probe_floor",
828 		.data		= &init_net.ipv4.sysctl_tcp_mtu_probe_floor,
829 		.maxlen		= sizeof(int),
830 		.mode		= 0644,
831 		.proc_handler	= proc_dointvec_minmax,
832 		.extra1		= &tcp_min_snd_mss_min,
833 		.extra2		= &tcp_min_snd_mss_max,
834 	},
835 	{
836 		.procname	= "tcp_probe_threshold",
837 		.data		= &init_net.ipv4.sysctl_tcp_probe_threshold,
838 		.maxlen		= sizeof(int),
839 		.mode		= 0644,
840 		.proc_handler	= proc_dointvec,
841 	},
842 	{
843 		.procname	= "tcp_probe_interval",
844 		.data		= &init_net.ipv4.sysctl_tcp_probe_interval,
845 		.maxlen		= sizeof(u32),
846 		.mode		= 0644,
847 		.proc_handler	= proc_douintvec_minmax,
848 		.extra2		= &u32_max_div_HZ,
849 	},
850 	{
851 		.procname	= "igmp_link_local_mcast_reports",
852 		.data		= &init_net.ipv4.sysctl_igmp_llm_reports,
853 		.maxlen		= sizeof(u8),
854 		.mode		= 0644,
855 		.proc_handler	= proc_dou8vec_minmax,
856 	},
857 	{
858 		.procname	= "igmp_max_memberships",
859 		.data		= &init_net.ipv4.sysctl_igmp_max_memberships,
860 		.maxlen		= sizeof(int),
861 		.mode		= 0644,
862 		.proc_handler	= proc_dointvec
863 	},
864 	{
865 		.procname	= "igmp_max_msf",
866 		.data		= &init_net.ipv4.sysctl_igmp_max_msf,
867 		.maxlen		= sizeof(int),
868 		.mode		= 0644,
869 		.proc_handler	= proc_dointvec
870 	},
871 #ifdef CONFIG_IP_MULTICAST
872 	{
873 		.procname	= "igmp_qrv",
874 		.data		= &init_net.ipv4.sysctl_igmp_qrv,
875 		.maxlen		= sizeof(int),
876 		.mode		= 0644,
877 		.proc_handler	= proc_dointvec_minmax,
878 		.extra1		= SYSCTL_ONE
879 	},
880 #endif
881 	{
882 		.procname	= "tcp_congestion_control",
883 		.data		= &init_net.ipv4.tcp_congestion_control,
884 		.mode		= 0644,
885 		.maxlen		= TCP_CA_NAME_MAX,
886 		.proc_handler	= proc_tcp_congestion_control,
887 	},
888 	{
889 		.procname	= "tcp_available_congestion_control",
890 		.maxlen		= TCP_CA_BUF_MAX,
891 		.mode		= 0444,
892 		.proc_handler   = proc_tcp_available_congestion_control,
893 	},
894 	{
895 		.procname	= "tcp_allowed_congestion_control",
896 		.maxlen		= TCP_CA_BUF_MAX,
897 		.mode		= 0644,
898 		.proc_handler   = proc_allowed_congestion_control,
899 	},
900 	{
901 		.procname	= "tcp_keepalive_time",
902 		.data		= &init_net.ipv4.sysctl_tcp_keepalive_time,
903 		.maxlen		= sizeof(int),
904 		.mode		= 0644,
905 		.proc_handler	= proc_dointvec_jiffies,
906 	},
907 	{
908 		.procname	= "tcp_keepalive_probes",
909 		.data		= &init_net.ipv4.sysctl_tcp_keepalive_probes,
910 		.maxlen		= sizeof(u8),
911 		.mode		= 0644,
912 		.proc_handler	= proc_dou8vec_minmax,
913 	},
914 	{
915 		.procname	= "tcp_keepalive_intvl",
916 		.data		= &init_net.ipv4.sysctl_tcp_keepalive_intvl,
917 		.maxlen		= sizeof(int),
918 		.mode		= 0644,
919 		.proc_handler	= proc_dointvec_jiffies,
920 	},
921 	{
922 		.procname	= "tcp_syn_retries",
923 		.data		= &init_net.ipv4.sysctl_tcp_syn_retries,
924 		.maxlen		= sizeof(u8),
925 		.mode		= 0644,
926 		.proc_handler	= proc_dou8vec_minmax,
927 		.extra1		= &tcp_syn_retries_min,
928 		.extra2		= &tcp_syn_retries_max
929 	},
930 	{
931 		.procname	= "tcp_synack_retries",
932 		.data		= &init_net.ipv4.sysctl_tcp_synack_retries,
933 		.maxlen		= sizeof(u8),
934 		.mode		= 0644,
935 		.proc_handler	= proc_dou8vec_minmax,
936 	},
937 #ifdef CONFIG_SYN_COOKIES
938 	{
939 		.procname	= "tcp_syncookies",
940 		.data		= &init_net.ipv4.sysctl_tcp_syncookies,
941 		.maxlen		= sizeof(u8),
942 		.mode		= 0644,
943 		.proc_handler	= proc_dou8vec_minmax,
944 	},
945 #endif
946 	{
947 		.procname	= "tcp_migrate_req",
948 		.data		= &init_net.ipv4.sysctl_tcp_migrate_req,
949 		.maxlen		= sizeof(u8),
950 		.mode		= 0644,
951 		.proc_handler	= proc_dou8vec_minmax,
952 		.extra1		= SYSCTL_ZERO,
953 		.extra2		= SYSCTL_ONE
954 	},
955 	{
956 		.procname	= "tcp_reordering",
957 		.data		= &init_net.ipv4.sysctl_tcp_reordering,
958 		.maxlen		= sizeof(int),
959 		.mode		= 0644,
960 		.proc_handler	= proc_dointvec
961 	},
962 	{
963 		.procname	= "tcp_retries1",
964 		.data		= &init_net.ipv4.sysctl_tcp_retries1,
965 		.maxlen		= sizeof(u8),
966 		.mode		= 0644,
967 		.proc_handler	= proc_dou8vec_minmax,
968 		.extra2		= &tcp_retr1_max
969 	},
970 	{
971 		.procname	= "tcp_retries2",
972 		.data		= &init_net.ipv4.sysctl_tcp_retries2,
973 		.maxlen		= sizeof(u8),
974 		.mode		= 0644,
975 		.proc_handler	= proc_dou8vec_minmax,
976 	},
977 	{
978 		.procname	= "tcp_orphan_retries",
979 		.data		= &init_net.ipv4.sysctl_tcp_orphan_retries,
980 		.maxlen		= sizeof(u8),
981 		.mode		= 0644,
982 		.proc_handler	= proc_dou8vec_minmax,
983 	},
984 	{
985 		.procname	= "tcp_fin_timeout",
986 		.data		= &init_net.ipv4.sysctl_tcp_fin_timeout,
987 		.maxlen		= sizeof(int),
988 		.mode		= 0644,
989 		.proc_handler	= proc_dointvec_jiffies,
990 	},
991 	{
992 		.procname	= "tcp_notsent_lowat",
993 		.data		= &init_net.ipv4.sysctl_tcp_notsent_lowat,
994 		.maxlen		= sizeof(unsigned int),
995 		.mode		= 0644,
996 		.proc_handler	= proc_douintvec,
997 	},
998 	{
999 		.procname	= "tcp_tw_reuse",
1000 		.data		= &init_net.ipv4.sysctl_tcp_tw_reuse,
1001 		.maxlen		= sizeof(u8),
1002 		.mode		= 0644,
1003 		.proc_handler	= proc_dou8vec_minmax,
1004 		.extra1		= SYSCTL_ZERO,
1005 		.extra2		= SYSCTL_TWO,
1006 	},
1007 	{
1008 		.procname	= "tcp_max_syn_backlog",
1009 		.data		= &init_net.ipv4.sysctl_max_syn_backlog,
1010 		.maxlen		= sizeof(int),
1011 		.mode		= 0644,
1012 		.proc_handler	= proc_dointvec
1013 	},
1014 	{
1015 		.procname	= "tcp_fastopen",
1016 		.data		= &init_net.ipv4.sysctl_tcp_fastopen,
1017 		.maxlen		= sizeof(int),
1018 		.mode		= 0644,
1019 		.proc_handler	= proc_dointvec,
1020 	},
1021 	{
1022 		.procname	= "tcp_fastopen_key",
1023 		.mode		= 0600,
1024 		.data		= &init_net.ipv4.sysctl_tcp_fastopen,
1025 		/* maxlen to print the list of keys in hex (*2), with dashes
1026 		 * separating doublewords and a comma in between keys.
1027 		 */
1028 		.maxlen		= ((TCP_FASTOPEN_KEY_LENGTH *
1029 				   2 * TCP_FASTOPEN_KEY_MAX) +
1030 				   (TCP_FASTOPEN_KEY_MAX * 5)),
1031 		.proc_handler	= proc_tcp_fastopen_key,
1032 	},
1033 	{
1034 		.procname	= "tcp_fastopen_blackhole_timeout_sec",
1035 		.data		= &init_net.ipv4.sysctl_tcp_fastopen_blackhole_timeout,
1036 		.maxlen		= sizeof(int),
1037 		.mode		= 0644,
1038 		.proc_handler	= proc_tfo_blackhole_detect_timeout,
1039 		.extra1		= SYSCTL_ZERO,
1040 	},
1041 #ifdef CONFIG_IP_ROUTE_MULTIPATH
1042 	{
1043 		.procname	= "fib_multipath_use_neigh",
1044 		.data		= &init_net.ipv4.sysctl_fib_multipath_use_neigh,
1045 		.maxlen		= sizeof(u8),
1046 		.mode		= 0644,
1047 		.proc_handler	= proc_dou8vec_minmax,
1048 		.extra1		= SYSCTL_ZERO,
1049 		.extra2		= SYSCTL_ONE,
1050 	},
1051 	{
1052 		.procname	= "fib_multipath_hash_policy",
1053 		.data		= &init_net.ipv4.sysctl_fib_multipath_hash_policy,
1054 		.maxlen		= sizeof(u8),
1055 		.mode		= 0644,
1056 		.proc_handler	= proc_fib_multipath_hash_policy,
1057 		.extra1		= SYSCTL_ZERO,
1058 		.extra2		= SYSCTL_THREE,
1059 	},
1060 	{
1061 		.procname	= "fib_multipath_hash_fields",
1062 		.data		= &init_net.ipv4.sysctl_fib_multipath_hash_fields,
1063 		.maxlen		= sizeof(u32),
1064 		.mode		= 0644,
1065 		.proc_handler	= proc_fib_multipath_hash_fields,
1066 		.extra1		= SYSCTL_ONE,
1067 		.extra2		= &fib_multipath_hash_fields_all_mask,
1068 	},
1069 #endif
1070 	{
1071 		.procname	= "ip_unprivileged_port_start",
1072 		.maxlen		= sizeof(int),
1073 		.data		= &init_net.ipv4.sysctl_ip_prot_sock,
1074 		.mode		= 0644,
1075 		.proc_handler	= ipv4_privileged_ports,
1076 	},
1077 #ifdef CONFIG_NET_L3_MASTER_DEV
1078 	{
1079 		.procname	= "udp_l3mdev_accept",
1080 		.data		= &init_net.ipv4.sysctl_udp_l3mdev_accept,
1081 		.maxlen		= sizeof(u8),
1082 		.mode		= 0644,
1083 		.proc_handler	= proc_dou8vec_minmax,
1084 		.extra1		= SYSCTL_ZERO,
1085 		.extra2		= SYSCTL_ONE,
1086 	},
1087 #endif
1088 	{
1089 		.procname	= "tcp_sack",
1090 		.data		= &init_net.ipv4.sysctl_tcp_sack,
1091 		.maxlen		= sizeof(u8),
1092 		.mode		= 0644,
1093 		.proc_handler	= proc_dou8vec_minmax,
1094 	},
1095 	{
1096 		.procname	= "tcp_window_scaling",
1097 		.data		= &init_net.ipv4.sysctl_tcp_window_scaling,
1098 		.maxlen		= sizeof(u8),
1099 		.mode		= 0644,
1100 		.proc_handler	= proc_dou8vec_minmax,
1101 	},
1102 	{
1103 		.procname	= "tcp_timestamps",
1104 		.data		= &init_net.ipv4.sysctl_tcp_timestamps,
1105 		.maxlen		= sizeof(u8),
1106 		.mode		= 0644,
1107 		.proc_handler	= proc_dou8vec_minmax,
1108 	},
1109 	{
1110 		.procname	= "tcp_early_retrans",
1111 		.data		= &init_net.ipv4.sysctl_tcp_early_retrans,
1112 		.maxlen		= sizeof(u8),
1113 		.mode		= 0644,
1114 		.proc_handler	= proc_dou8vec_minmax,
1115 		.extra1		= SYSCTL_ZERO,
1116 		.extra2		= SYSCTL_FOUR,
1117 	},
1118 	{
1119 		.procname	= "tcp_recovery",
1120 		.data		= &init_net.ipv4.sysctl_tcp_recovery,
1121 		.maxlen		= sizeof(u8),
1122 		.mode		= 0644,
1123 		.proc_handler	= proc_dou8vec_minmax,
1124 	},
1125 	{
1126 		.procname       = "tcp_thin_linear_timeouts",
1127 		.data           = &init_net.ipv4.sysctl_tcp_thin_linear_timeouts,
1128 		.maxlen         = sizeof(u8),
1129 		.mode           = 0644,
1130 		.proc_handler   = proc_dou8vec_minmax,
1131 	},
1132 	{
1133 		.procname	= "tcp_slow_start_after_idle",
1134 		.data		= &init_net.ipv4.sysctl_tcp_slow_start_after_idle,
1135 		.maxlen		= sizeof(u8),
1136 		.mode		= 0644,
1137 		.proc_handler	= proc_dou8vec_minmax,
1138 	},
1139 	{
1140 		.procname	= "tcp_retrans_collapse",
1141 		.data		= &init_net.ipv4.sysctl_tcp_retrans_collapse,
1142 		.maxlen		= sizeof(u8),
1143 		.mode		= 0644,
1144 		.proc_handler	= proc_dou8vec_minmax,
1145 	},
1146 	{
1147 		.procname	= "tcp_stdurg",
1148 		.data		= &init_net.ipv4.sysctl_tcp_stdurg,
1149 		.maxlen		= sizeof(u8),
1150 		.mode		= 0644,
1151 		.proc_handler	= proc_dou8vec_minmax,
1152 	},
1153 	{
1154 		.procname	= "tcp_rfc1337",
1155 		.data		= &init_net.ipv4.sysctl_tcp_rfc1337,
1156 		.maxlen		= sizeof(u8),
1157 		.mode		= 0644,
1158 		.proc_handler	= proc_dou8vec_minmax,
1159 	},
1160 	{
1161 		.procname	= "tcp_abort_on_overflow",
1162 		.data		= &init_net.ipv4.sysctl_tcp_abort_on_overflow,
1163 		.maxlen		= sizeof(u8),
1164 		.mode		= 0644,
1165 		.proc_handler	= proc_dou8vec_minmax,
1166 	},
1167 	{
1168 		.procname	= "tcp_fack",
1169 		.data		= &init_net.ipv4.sysctl_tcp_fack,
1170 		.maxlen		= sizeof(u8),
1171 		.mode		= 0644,
1172 		.proc_handler	= proc_dou8vec_minmax,
1173 	},
1174 	{
1175 		.procname	= "tcp_max_reordering",
1176 		.data		= &init_net.ipv4.sysctl_tcp_max_reordering,
1177 		.maxlen		= sizeof(int),
1178 		.mode		= 0644,
1179 		.proc_handler	= proc_dointvec
1180 	},
1181 	{
1182 		.procname	= "tcp_dsack",
1183 		.data		= &init_net.ipv4.sysctl_tcp_dsack,
1184 		.maxlen		= sizeof(u8),
1185 		.mode		= 0644,
1186 		.proc_handler	= proc_dou8vec_minmax,
1187 	},
1188 	{
1189 		.procname	= "tcp_app_win",
1190 		.data		= &init_net.ipv4.sysctl_tcp_app_win,
1191 		.maxlen		= sizeof(u8),
1192 		.mode		= 0644,
1193 		.proc_handler	= proc_dou8vec_minmax,
1194 	},
1195 	{
1196 		.procname	= "tcp_adv_win_scale",
1197 		.data		= &init_net.ipv4.sysctl_tcp_adv_win_scale,
1198 		.maxlen		= sizeof(int),
1199 		.mode		= 0644,
1200 		.proc_handler	= proc_dointvec_minmax,
1201 		.extra1		= &tcp_adv_win_scale_min,
1202 		.extra2		= &tcp_adv_win_scale_max,
1203 	},
1204 	{
1205 		.procname	= "tcp_frto",
1206 		.data		= &init_net.ipv4.sysctl_tcp_frto,
1207 		.maxlen		= sizeof(u8),
1208 		.mode		= 0644,
1209 		.proc_handler	= proc_dou8vec_minmax,
1210 	},
1211 	{
1212 		.procname	= "tcp_no_metrics_save",
1213 		.data		= &init_net.ipv4.sysctl_tcp_nometrics_save,
1214 		.maxlen		= sizeof(u8),
1215 		.mode		= 0644,
1216 		.proc_handler	= proc_dou8vec_minmax,
1217 	},
1218 	{
1219 		.procname	= "tcp_no_ssthresh_metrics_save",
1220 		.data		= &init_net.ipv4.sysctl_tcp_no_ssthresh_metrics_save,
1221 		.maxlen		= sizeof(u8),
1222 		.mode		= 0644,
1223 		.proc_handler	= proc_dou8vec_minmax,
1224 		.extra1		= SYSCTL_ZERO,
1225 		.extra2		= SYSCTL_ONE,
1226 	},
1227 	{
1228 		.procname	= "tcp_moderate_rcvbuf",
1229 		.data		= &init_net.ipv4.sysctl_tcp_moderate_rcvbuf,
1230 		.maxlen		= sizeof(u8),
1231 		.mode		= 0644,
1232 		.proc_handler	= proc_dou8vec_minmax,
1233 	},
1234 	{
1235 		.procname	= "tcp_tso_win_divisor",
1236 		.data		= &init_net.ipv4.sysctl_tcp_tso_win_divisor,
1237 		.maxlen		= sizeof(u8),
1238 		.mode		= 0644,
1239 		.proc_handler	= proc_dou8vec_minmax,
1240 	},
1241 	{
1242 		.procname	= "tcp_workaround_signed_windows",
1243 		.data		= &init_net.ipv4.sysctl_tcp_workaround_signed_windows,
1244 		.maxlen		= sizeof(u8),
1245 		.mode		= 0644,
1246 		.proc_handler	= proc_dou8vec_minmax,
1247 	},
1248 	{
1249 		.procname	= "tcp_limit_output_bytes",
1250 		.data		= &init_net.ipv4.sysctl_tcp_limit_output_bytes,
1251 		.maxlen		= sizeof(int),
1252 		.mode		= 0644,
1253 		.proc_handler	= proc_dointvec
1254 	},
1255 	{
1256 		.procname	= "tcp_challenge_ack_limit",
1257 		.data		= &init_net.ipv4.sysctl_tcp_challenge_ack_limit,
1258 		.maxlen		= sizeof(int),
1259 		.mode		= 0644,
1260 		.proc_handler	= proc_dointvec
1261 	},
1262 	{
1263 		.procname	= "tcp_min_tso_segs",
1264 		.data		= &init_net.ipv4.sysctl_tcp_min_tso_segs,
1265 		.maxlen		= sizeof(u8),
1266 		.mode		= 0644,
1267 		.proc_handler	= proc_dou8vec_minmax,
1268 		.extra1		= SYSCTL_ONE,
1269 	},
1270 	{
1271 		.procname	= "tcp_tso_rtt_log",
1272 		.data		= &init_net.ipv4.sysctl_tcp_tso_rtt_log,
1273 		.maxlen		= sizeof(u8),
1274 		.mode		= 0644,
1275 		.proc_handler	= proc_dou8vec_minmax,
1276 	},
1277 	{
1278 		.procname	= "tcp_min_rtt_wlen",
1279 		.data		= &init_net.ipv4.sysctl_tcp_min_rtt_wlen,
1280 		.maxlen		= sizeof(int),
1281 		.mode		= 0644,
1282 		.proc_handler	= proc_dointvec_minmax,
1283 		.extra1		= SYSCTL_ZERO,
1284 		.extra2		= &one_day_secs
1285 	},
1286 	{
1287 		.procname	= "tcp_autocorking",
1288 		.data		= &init_net.ipv4.sysctl_tcp_autocorking,
1289 		.maxlen		= sizeof(u8),
1290 		.mode		= 0644,
1291 		.proc_handler	= proc_dou8vec_minmax,
1292 		.extra1		= SYSCTL_ZERO,
1293 		.extra2		= SYSCTL_ONE,
1294 	},
1295 	{
1296 		.procname	= "tcp_invalid_ratelimit",
1297 		.data		= &init_net.ipv4.sysctl_tcp_invalid_ratelimit,
1298 		.maxlen		= sizeof(int),
1299 		.mode		= 0644,
1300 		.proc_handler	= proc_dointvec_ms_jiffies,
1301 	},
1302 	{
1303 		.procname	= "tcp_pacing_ss_ratio",
1304 		.data		= &init_net.ipv4.sysctl_tcp_pacing_ss_ratio,
1305 		.maxlen		= sizeof(int),
1306 		.mode		= 0644,
1307 		.proc_handler	= proc_dointvec_minmax,
1308 		.extra1		= SYSCTL_ZERO,
1309 		.extra2		= SYSCTL_ONE_THOUSAND,
1310 	},
1311 	{
1312 		.procname	= "tcp_pacing_ca_ratio",
1313 		.data		= &init_net.ipv4.sysctl_tcp_pacing_ca_ratio,
1314 		.maxlen		= sizeof(int),
1315 		.mode		= 0644,
1316 		.proc_handler	= proc_dointvec_minmax,
1317 		.extra1		= SYSCTL_ZERO,
1318 		.extra2		= SYSCTL_ONE_THOUSAND,
1319 	},
1320 	{
1321 		.procname	= "tcp_wmem",
1322 		.data		= &init_net.ipv4.sysctl_tcp_wmem,
1323 		.maxlen		= sizeof(init_net.ipv4.sysctl_tcp_wmem),
1324 		.mode		= 0644,
1325 		.proc_handler	= proc_dointvec_minmax,
1326 		.extra1		= SYSCTL_ONE,
1327 	},
1328 	{
1329 		.procname	= "tcp_rmem",
1330 		.data		= &init_net.ipv4.sysctl_tcp_rmem,
1331 		.maxlen		= sizeof(init_net.ipv4.sysctl_tcp_rmem),
1332 		.mode		= 0644,
1333 		.proc_handler	= proc_dointvec_minmax,
1334 		.extra1		= SYSCTL_ONE,
1335 	},
1336 	{
1337 		.procname	= "tcp_comp_sack_delay_ns",
1338 		.data		= &init_net.ipv4.sysctl_tcp_comp_sack_delay_ns,
1339 		.maxlen		= sizeof(unsigned long),
1340 		.mode		= 0644,
1341 		.proc_handler	= proc_doulongvec_minmax,
1342 	},
1343 	{
1344 		.procname	= "tcp_comp_sack_slack_ns",
1345 		.data		= &init_net.ipv4.sysctl_tcp_comp_sack_slack_ns,
1346 		.maxlen		= sizeof(unsigned long),
1347 		.mode		= 0644,
1348 		.proc_handler	= proc_doulongvec_minmax,
1349 	},
1350 	{
1351 		.procname	= "tcp_comp_sack_nr",
1352 		.data		= &init_net.ipv4.sysctl_tcp_comp_sack_nr,
1353 		.maxlen		= sizeof(u8),
1354 		.mode		= 0644,
1355 		.proc_handler	= proc_dou8vec_minmax,
1356 		.extra1		= SYSCTL_ZERO,
1357 	},
1358 	{
1359 		.procname       = "tcp_reflect_tos",
1360 		.data           = &init_net.ipv4.sysctl_tcp_reflect_tos,
1361 		.maxlen         = sizeof(u8),
1362 		.mode           = 0644,
1363 		.proc_handler   = proc_dou8vec_minmax,
1364 		.extra1         = SYSCTL_ZERO,
1365 		.extra2         = SYSCTL_ONE,
1366 	},
1367 	{
1368 		.procname	= "udp_rmem_min",
1369 		.data		= &init_net.ipv4.sysctl_udp_rmem_min,
1370 		.maxlen		= sizeof(init_net.ipv4.sysctl_udp_rmem_min),
1371 		.mode		= 0644,
1372 		.proc_handler	= proc_dointvec_minmax,
1373 		.extra1		= SYSCTL_ONE
1374 	},
1375 	{
1376 		.procname	= "udp_wmem_min",
1377 		.data		= &init_net.ipv4.sysctl_udp_wmem_min,
1378 		.maxlen		= sizeof(init_net.ipv4.sysctl_udp_wmem_min),
1379 		.mode		= 0644,
1380 		.proc_handler	= proc_dointvec_minmax,
1381 		.extra1		= SYSCTL_ONE
1382 	},
1383 	{
1384 		.procname	= "fib_notify_on_flag_change",
1385 		.data		= &init_net.ipv4.sysctl_fib_notify_on_flag_change,
1386 		.maxlen		= sizeof(u8),
1387 		.mode		= 0644,
1388 		.proc_handler	= proc_dou8vec_minmax,
1389 		.extra1		= SYSCTL_ZERO,
1390 		.extra2		= SYSCTL_TWO,
1391 	},
1392 	{ }
1393 };
1394 
1395 static __net_init int ipv4_sysctl_init_net(struct net *net)
1396 {
1397 	struct ctl_table *table;
1398 
1399 	table = ipv4_net_table;
1400 	if (!net_eq(net, &init_net)) {
1401 		int i;
1402 
1403 		table = kmemdup(table, sizeof(ipv4_net_table), GFP_KERNEL);
1404 		if (!table)
1405 			goto err_alloc;
1406 
1407 		/* skip first entry (sysctl_max_tw_buckets) */
1408 		for (i = 1; i < ARRAY_SIZE(ipv4_net_table) - 1; i++) {
1409 			if (table[i].data) {
1410 				/* Update the variables to point into
1411 				 * the current struct net
1412 				 */
1413 				table[i].data += (void *)net - (void *)&init_net;
1414 			} else {
1415 				/* Entries without data pointer are global;
1416 				 * Make them read-only in non-init_net ns
1417 				 */
1418 				table[i].mode &= ~0222;
1419 			}
1420 		}
1421 	}
1422 
1423 	table[0].data = &net->ipv4.tcp_death_row->sysctl_max_tw_buckets;
1424 
1425 	net->ipv4.ipv4_hdr = register_net_sysctl(net, "net/ipv4", table);
1426 	if (!net->ipv4.ipv4_hdr)
1427 		goto err_reg;
1428 
1429 	net->ipv4.sysctl_local_reserved_ports = kzalloc(65536 / 8, GFP_KERNEL);
1430 	if (!net->ipv4.sysctl_local_reserved_ports)
1431 		goto err_ports;
1432 
1433 	return 0;
1434 
1435 err_ports:
1436 	unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
1437 err_reg:
1438 	if (!net_eq(net, &init_net))
1439 		kfree(table);
1440 err_alloc:
1441 	return -ENOMEM;
1442 }
1443 
1444 static __net_exit void ipv4_sysctl_exit_net(struct net *net)
1445 {
1446 	struct ctl_table *table;
1447 
1448 	kfree(net->ipv4.sysctl_local_reserved_ports);
1449 	table = net->ipv4.ipv4_hdr->ctl_table_arg;
1450 	unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
1451 	kfree(table);
1452 }
1453 
1454 static __net_initdata struct pernet_operations ipv4_sysctl_ops = {
1455 	.init = ipv4_sysctl_init_net,
1456 	.exit = ipv4_sysctl_exit_net,
1457 };
1458 
1459 static __init int sysctl_ipv4_init(void)
1460 {
1461 	struct ctl_table_header *hdr;
1462 
1463 	hdr = register_net_sysctl(&init_net, "net/ipv4", ipv4_table);
1464 	if (!hdr)
1465 		return -ENOMEM;
1466 
1467 	if (register_pernet_subsys(&ipv4_sysctl_ops)) {
1468 		unregister_net_sysctl_table(hdr);
1469 		return -ENOMEM;
1470 	}
1471 
1472 	return 0;
1473 }
1474 
1475 __initcall(sysctl_ipv4_init);
1476