xref: /linux/net/ipv4/sysctl_net_ipv4.c (revision 9ad28ba1c6f054ef801bcd63c5b73792b4b1d9a4)
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 static unsigned int tcp_child_ehash_entries_max = 16 * 1024 * 1024;
43 static unsigned int udp_child_hash_entries_max = UDP_HTABLE_SIZE_MAX;
44 static int tcp_plb_max_rounds = 31;
45 static int tcp_plb_max_cong_thresh = 256;
46 
47 /* obsolete */
48 static int sysctl_tcp_low_latency __read_mostly;
49 
50 /* Update system visible IP port range */
51 static void set_local_port_range(struct net *net, int range[2])
52 {
53 	bool same_parity = !((range[0] ^ range[1]) & 1);
54 
55 	write_seqlock_bh(&net->ipv4.ip_local_ports.lock);
56 	if (same_parity && !net->ipv4.ip_local_ports.warned) {
57 		net->ipv4.ip_local_ports.warned = true;
58 		pr_err_ratelimited("ip_local_port_range: prefer different parity for start/end values.\n");
59 	}
60 	net->ipv4.ip_local_ports.range[0] = range[0];
61 	net->ipv4.ip_local_ports.range[1] = range[1];
62 	write_sequnlock_bh(&net->ipv4.ip_local_ports.lock);
63 }
64 
65 /* Validate changes from /proc interface. */
66 static int ipv4_local_port_range(struct ctl_table *table, int write,
67 				 void *buffer, size_t *lenp, loff_t *ppos)
68 {
69 	struct net *net =
70 		container_of(table->data, struct net, ipv4.ip_local_ports.range);
71 	int ret;
72 	int range[2];
73 	struct ctl_table tmp = {
74 		.data = &range,
75 		.maxlen = sizeof(range),
76 		.mode = table->mode,
77 		.extra1 = &ip_local_port_range_min,
78 		.extra2 = &ip_local_port_range_max,
79 	};
80 
81 	inet_get_local_port_range(net, &range[0], &range[1]);
82 
83 	ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
84 
85 	if (write && ret == 0) {
86 		/* Ensure that the upper limit is not smaller than the lower,
87 		 * and that the lower does not encroach upon the privileged
88 		 * port limit.
89 		 */
90 		if ((range[1] < range[0]) ||
91 		    (range[0] < READ_ONCE(net->ipv4.sysctl_ip_prot_sock)))
92 			ret = -EINVAL;
93 		else
94 			set_local_port_range(net, range);
95 	}
96 
97 	return ret;
98 }
99 
100 /* Validate changes from /proc interface. */
101 static int ipv4_privileged_ports(struct ctl_table *table, int write,
102 				void *buffer, size_t *lenp, loff_t *ppos)
103 {
104 	struct net *net = container_of(table->data, struct net,
105 	    ipv4.sysctl_ip_prot_sock);
106 	int ret;
107 	int pports;
108 	int range[2];
109 	struct ctl_table tmp = {
110 		.data = &pports,
111 		.maxlen = sizeof(pports),
112 		.mode = table->mode,
113 		.extra1 = &ip_privileged_port_min,
114 		.extra2 = &ip_privileged_port_max,
115 	};
116 
117 	pports = READ_ONCE(net->ipv4.sysctl_ip_prot_sock);
118 
119 	ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
120 
121 	if (write && ret == 0) {
122 		inet_get_local_port_range(net, &range[0], &range[1]);
123 		/* Ensure that the local port range doesn't overlap with the
124 		 * privileged port range.
125 		 */
126 		if (range[0] < pports)
127 			ret = -EINVAL;
128 		else
129 			WRITE_ONCE(net->ipv4.sysctl_ip_prot_sock, pports);
130 	}
131 
132 	return ret;
133 }
134 
135 static void inet_get_ping_group_range_table(struct ctl_table *table, kgid_t *low, kgid_t *high)
136 {
137 	kgid_t *data = table->data;
138 	struct net *net =
139 		container_of(table->data, struct net, ipv4.ping_group_range.range);
140 	unsigned int seq;
141 	do {
142 		seq = read_seqbegin(&net->ipv4.ping_group_range.lock);
143 
144 		*low = data[0];
145 		*high = data[1];
146 	} while (read_seqretry(&net->ipv4.ping_group_range.lock, seq));
147 }
148 
149 /* Update system visible IP port range */
150 static void set_ping_group_range(struct ctl_table *table, kgid_t low, kgid_t high)
151 {
152 	kgid_t *data = table->data;
153 	struct net *net =
154 		container_of(table->data, struct net, ipv4.ping_group_range.range);
155 	write_seqlock(&net->ipv4.ping_group_range.lock);
156 	data[0] = low;
157 	data[1] = high;
158 	write_sequnlock(&net->ipv4.ping_group_range.lock);
159 }
160 
161 /* Validate changes from /proc interface. */
162 static int ipv4_ping_group_range(struct ctl_table *table, int write,
163 				 void *buffer, size_t *lenp, loff_t *ppos)
164 {
165 	struct user_namespace *user_ns = current_user_ns();
166 	int ret;
167 	gid_t urange[2];
168 	kgid_t low, high;
169 	struct ctl_table tmp = {
170 		.data = &urange,
171 		.maxlen = sizeof(urange),
172 		.mode = table->mode,
173 		.extra1 = &ip_ping_group_range_min,
174 		.extra2 = &ip_ping_group_range_max,
175 	};
176 
177 	inet_get_ping_group_range_table(table, &low, &high);
178 	urange[0] = from_kgid_munged(user_ns, low);
179 	urange[1] = from_kgid_munged(user_ns, high);
180 	ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
181 
182 	if (write && ret == 0) {
183 		low = make_kgid(user_ns, urange[0]);
184 		high = make_kgid(user_ns, urange[1]);
185 		if (!gid_valid(low) || !gid_valid(high))
186 			return -EINVAL;
187 		if (urange[1] < urange[0] || gid_lt(high, low)) {
188 			low = make_kgid(&init_user_ns, 1);
189 			high = make_kgid(&init_user_ns, 0);
190 		}
191 		set_ping_group_range(table, low, high);
192 	}
193 
194 	return ret;
195 }
196 
197 static int ipv4_fwd_update_priority(struct ctl_table *table, int write,
198 				    void *buffer, size_t *lenp, loff_t *ppos)
199 {
200 	struct net *net;
201 	int ret;
202 
203 	net = container_of(table->data, struct net,
204 			   ipv4.sysctl_ip_fwd_update_priority);
205 	ret = proc_dou8vec_minmax(table, write, buffer, lenp, ppos);
206 	if (write && ret == 0)
207 		call_netevent_notifiers(NETEVENT_IPV4_FWD_UPDATE_PRIORITY_UPDATE,
208 					net);
209 
210 	return ret;
211 }
212 
213 static int proc_tcp_congestion_control(struct ctl_table *ctl, int write,
214 				       void *buffer, size_t *lenp, loff_t *ppos)
215 {
216 	struct net *net = container_of(ctl->data, struct net,
217 				       ipv4.tcp_congestion_control);
218 	char val[TCP_CA_NAME_MAX];
219 	struct ctl_table tbl = {
220 		.data = val,
221 		.maxlen = TCP_CA_NAME_MAX,
222 	};
223 	int ret;
224 
225 	tcp_get_default_congestion_control(net, val);
226 
227 	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
228 	if (write && ret == 0)
229 		ret = tcp_set_default_congestion_control(net, val);
230 	return ret;
231 }
232 
233 static int proc_tcp_available_congestion_control(struct ctl_table *ctl,
234 						 int write, void *buffer,
235 						 size_t *lenp, loff_t *ppos)
236 {
237 	struct ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX, };
238 	int ret;
239 
240 	tbl.data = kmalloc(tbl.maxlen, GFP_USER);
241 	if (!tbl.data)
242 		return -ENOMEM;
243 	tcp_get_available_congestion_control(tbl.data, TCP_CA_BUF_MAX);
244 	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
245 	kfree(tbl.data);
246 	return ret;
247 }
248 
249 static int proc_allowed_congestion_control(struct ctl_table *ctl,
250 					   int write, void *buffer,
251 					   size_t *lenp, loff_t *ppos)
252 {
253 	struct ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX };
254 	int ret;
255 
256 	tbl.data = kmalloc(tbl.maxlen, GFP_USER);
257 	if (!tbl.data)
258 		return -ENOMEM;
259 
260 	tcp_get_allowed_congestion_control(tbl.data, tbl.maxlen);
261 	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
262 	if (write && ret == 0)
263 		ret = tcp_set_allowed_congestion_control(tbl.data);
264 	kfree(tbl.data);
265 	return ret;
266 }
267 
268 static int sscanf_key(char *buf, __le32 *key)
269 {
270 	u32 user_key[4];
271 	int i, ret = 0;
272 
273 	if (sscanf(buf, "%x-%x-%x-%x", user_key, user_key + 1,
274 		   user_key + 2, user_key + 3) != 4) {
275 		ret = -EINVAL;
276 	} else {
277 		for (i = 0; i < ARRAY_SIZE(user_key); i++)
278 			key[i] = cpu_to_le32(user_key[i]);
279 	}
280 	pr_debug("proc TFO key set 0x%x-%x-%x-%x <- 0x%s: %u\n",
281 		 user_key[0], user_key[1], user_key[2], user_key[3], buf, ret);
282 
283 	return ret;
284 }
285 
286 static int proc_tcp_fastopen_key(struct ctl_table *table, int write,
287 				 void *buffer, size_t *lenp, loff_t *ppos)
288 {
289 	struct net *net = container_of(table->data, struct net,
290 	    ipv4.sysctl_tcp_fastopen);
291 	/* maxlen to print the list of keys in hex (*2), with dashes
292 	 * separating doublewords and a comma in between keys.
293 	 */
294 	struct ctl_table tbl = { .maxlen = ((TCP_FASTOPEN_KEY_LENGTH *
295 					    2 * TCP_FASTOPEN_KEY_MAX) +
296 					    (TCP_FASTOPEN_KEY_MAX * 5)) };
297 	u32 user_key[TCP_FASTOPEN_KEY_BUF_LENGTH / sizeof(u32)];
298 	__le32 key[TCP_FASTOPEN_KEY_BUF_LENGTH / sizeof(__le32)];
299 	char *backup_data;
300 	int ret, i = 0, off = 0, n_keys;
301 
302 	tbl.data = kmalloc(tbl.maxlen, GFP_KERNEL);
303 	if (!tbl.data)
304 		return -ENOMEM;
305 
306 	n_keys = tcp_fastopen_get_cipher(net, NULL, (u64 *)key);
307 	if (!n_keys) {
308 		memset(&key[0], 0, TCP_FASTOPEN_KEY_LENGTH);
309 		n_keys = 1;
310 	}
311 
312 	for (i = 0; i < n_keys * 4; i++)
313 		user_key[i] = le32_to_cpu(key[i]);
314 
315 	for (i = 0; i < n_keys; i++) {
316 		off += snprintf(tbl.data + off, tbl.maxlen - off,
317 				"%08x-%08x-%08x-%08x",
318 				user_key[i * 4],
319 				user_key[i * 4 + 1],
320 				user_key[i * 4 + 2],
321 				user_key[i * 4 + 3]);
322 
323 		if (WARN_ON_ONCE(off >= tbl.maxlen - 1))
324 			break;
325 
326 		if (i + 1 < n_keys)
327 			off += snprintf(tbl.data + off, tbl.maxlen - off, ",");
328 	}
329 
330 	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
331 
332 	if (write && ret == 0) {
333 		backup_data = strchr(tbl.data, ',');
334 		if (backup_data) {
335 			*backup_data = '\0';
336 			backup_data++;
337 		}
338 		if (sscanf_key(tbl.data, key)) {
339 			ret = -EINVAL;
340 			goto bad_key;
341 		}
342 		if (backup_data) {
343 			if (sscanf_key(backup_data, key + 4)) {
344 				ret = -EINVAL;
345 				goto bad_key;
346 			}
347 		}
348 		tcp_fastopen_reset_cipher(net, NULL, key,
349 					  backup_data ? key + 4 : NULL);
350 	}
351 
352 bad_key:
353 	kfree(tbl.data);
354 	return ret;
355 }
356 
357 static int proc_tfo_blackhole_detect_timeout(struct ctl_table *table,
358 					     int write, void *buffer,
359 					     size_t *lenp, loff_t *ppos)
360 {
361 	struct net *net = container_of(table->data, struct net,
362 	    ipv4.sysctl_tcp_fastopen_blackhole_timeout);
363 	int ret;
364 
365 	ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
366 	if (write && ret == 0)
367 		atomic_set(&net->ipv4.tfo_active_disable_times, 0);
368 
369 	return ret;
370 }
371 
372 static int proc_tcp_available_ulp(struct ctl_table *ctl,
373 				  int write, void *buffer, size_t *lenp,
374 				  loff_t *ppos)
375 {
376 	struct ctl_table tbl = { .maxlen = TCP_ULP_BUF_MAX, };
377 	int ret;
378 
379 	tbl.data = kmalloc(tbl.maxlen, GFP_USER);
380 	if (!tbl.data)
381 		return -ENOMEM;
382 	tcp_get_available_ulp(tbl.data, TCP_ULP_BUF_MAX);
383 	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
384 	kfree(tbl.data);
385 
386 	return ret;
387 }
388 
389 static int proc_tcp_ehash_entries(struct ctl_table *table, int write,
390 				  void *buffer, size_t *lenp, loff_t *ppos)
391 {
392 	struct net *net = container_of(table->data, struct net,
393 				       ipv4.sysctl_tcp_child_ehash_entries);
394 	struct inet_hashinfo *hinfo = net->ipv4.tcp_death_row.hashinfo;
395 	int tcp_ehash_entries;
396 	struct ctl_table tbl;
397 
398 	tcp_ehash_entries = hinfo->ehash_mask + 1;
399 
400 	/* A negative number indicates that the child netns
401 	 * shares the global ehash.
402 	 */
403 	if (!net_eq(net, &init_net) && !hinfo->pernet)
404 		tcp_ehash_entries *= -1;
405 
406 	memset(&tbl, 0, sizeof(tbl));
407 	tbl.data = &tcp_ehash_entries;
408 	tbl.maxlen = sizeof(int);
409 
410 	return proc_dointvec(&tbl, write, buffer, lenp, ppos);
411 }
412 
413 static int proc_udp_hash_entries(struct ctl_table *table, int write,
414 				 void *buffer, size_t *lenp, loff_t *ppos)
415 {
416 	struct net *net = container_of(table->data, struct net,
417 				       ipv4.sysctl_udp_child_hash_entries);
418 	int udp_hash_entries;
419 	struct ctl_table tbl;
420 
421 	udp_hash_entries = net->ipv4.udp_table->mask + 1;
422 
423 	/* A negative number indicates that the child netns
424 	 * shares the global udp_table.
425 	 */
426 	if (!net_eq(net, &init_net) && net->ipv4.udp_table == &udp_table)
427 		udp_hash_entries *= -1;
428 
429 	memset(&tbl, 0, sizeof(tbl));
430 	tbl.data = &udp_hash_entries;
431 	tbl.maxlen = sizeof(int);
432 
433 	return proc_dointvec(&tbl, write, buffer, lenp, ppos);
434 }
435 
436 #ifdef CONFIG_IP_ROUTE_MULTIPATH
437 static int proc_fib_multipath_hash_policy(struct ctl_table *table, int write,
438 					  void *buffer, size_t *lenp,
439 					  loff_t *ppos)
440 {
441 	struct net *net = container_of(table->data, struct net,
442 	    ipv4.sysctl_fib_multipath_hash_policy);
443 	int ret;
444 
445 	ret = proc_dou8vec_minmax(table, write, buffer, lenp, ppos);
446 	if (write && ret == 0)
447 		call_netevent_notifiers(NETEVENT_IPV4_MPATH_HASH_UPDATE, net);
448 
449 	return ret;
450 }
451 
452 static int proc_fib_multipath_hash_fields(struct ctl_table *table, int write,
453 					  void *buffer, size_t *lenp,
454 					  loff_t *ppos)
455 {
456 	struct net *net;
457 	int ret;
458 
459 	net = container_of(table->data, struct net,
460 			   ipv4.sysctl_fib_multipath_hash_fields);
461 	ret = proc_douintvec_minmax(table, write, buffer, lenp, ppos);
462 	if (write && ret == 0)
463 		call_netevent_notifiers(NETEVENT_IPV4_MPATH_HASH_UPDATE, net);
464 
465 	return ret;
466 }
467 #endif
468 
469 static struct ctl_table ipv4_table[] = {
470 	{
471 		.procname	= "tcp_max_orphans",
472 		.data		= &sysctl_tcp_max_orphans,
473 		.maxlen		= sizeof(int),
474 		.mode		= 0644,
475 		.proc_handler	= proc_dointvec
476 	},
477 	{
478 		.procname	= "inet_peer_threshold",
479 		.data		= &inet_peer_threshold,
480 		.maxlen		= sizeof(int),
481 		.mode		= 0644,
482 		.proc_handler	= proc_dointvec
483 	},
484 	{
485 		.procname	= "inet_peer_minttl",
486 		.data		= &inet_peer_minttl,
487 		.maxlen		= sizeof(int),
488 		.mode		= 0644,
489 		.proc_handler	= proc_dointvec_jiffies,
490 	},
491 	{
492 		.procname	= "inet_peer_maxttl",
493 		.data		= &inet_peer_maxttl,
494 		.maxlen		= sizeof(int),
495 		.mode		= 0644,
496 		.proc_handler	= proc_dointvec_jiffies,
497 	},
498 	{
499 		.procname	= "tcp_mem",
500 		.maxlen		= sizeof(sysctl_tcp_mem),
501 		.data		= &sysctl_tcp_mem,
502 		.mode		= 0644,
503 		.proc_handler	= proc_doulongvec_minmax,
504 	},
505 	{
506 		.procname	= "tcp_low_latency",
507 		.data		= &sysctl_tcp_low_latency,
508 		.maxlen		= sizeof(int),
509 		.mode		= 0644,
510 		.proc_handler	= proc_dointvec
511 	},
512 #ifdef CONFIG_NETLABEL
513 	{
514 		.procname	= "cipso_cache_enable",
515 		.data		= &cipso_v4_cache_enabled,
516 		.maxlen		= sizeof(int),
517 		.mode		= 0644,
518 		.proc_handler	= proc_dointvec,
519 	},
520 	{
521 		.procname	= "cipso_cache_bucket_size",
522 		.data		= &cipso_v4_cache_bucketsize,
523 		.maxlen		= sizeof(int),
524 		.mode		= 0644,
525 		.proc_handler	= proc_dointvec,
526 	},
527 	{
528 		.procname	= "cipso_rbm_optfmt",
529 		.data		= &cipso_v4_rbm_optfmt,
530 		.maxlen		= sizeof(int),
531 		.mode		= 0644,
532 		.proc_handler	= proc_dointvec,
533 	},
534 	{
535 		.procname	= "cipso_rbm_strictvalid",
536 		.data		= &cipso_v4_rbm_strictvalid,
537 		.maxlen		= sizeof(int),
538 		.mode		= 0644,
539 		.proc_handler	= proc_dointvec,
540 	},
541 #endif /* CONFIG_NETLABEL */
542 	{
543 		.procname	= "tcp_available_ulp",
544 		.maxlen		= TCP_ULP_BUF_MAX,
545 		.mode		= 0444,
546 		.proc_handler   = proc_tcp_available_ulp,
547 	},
548 	{
549 		.procname	= "icmp_msgs_per_sec",
550 		.data		= &sysctl_icmp_msgs_per_sec,
551 		.maxlen		= sizeof(int),
552 		.mode		= 0644,
553 		.proc_handler	= proc_dointvec_minmax,
554 		.extra1		= SYSCTL_ZERO,
555 	},
556 	{
557 		.procname	= "icmp_msgs_burst",
558 		.data		= &sysctl_icmp_msgs_burst,
559 		.maxlen		= sizeof(int),
560 		.mode		= 0644,
561 		.proc_handler	= proc_dointvec_minmax,
562 		.extra1		= SYSCTL_ZERO,
563 	},
564 	{
565 		.procname	= "udp_mem",
566 		.data		= &sysctl_udp_mem,
567 		.maxlen		= sizeof(sysctl_udp_mem),
568 		.mode		= 0644,
569 		.proc_handler	= proc_doulongvec_minmax,
570 	},
571 	{
572 		.procname	= "fib_sync_mem",
573 		.data		= &sysctl_fib_sync_mem,
574 		.maxlen		= sizeof(sysctl_fib_sync_mem),
575 		.mode		= 0644,
576 		.proc_handler	= proc_douintvec_minmax,
577 		.extra1		= &sysctl_fib_sync_mem_min,
578 		.extra2		= &sysctl_fib_sync_mem_max,
579 	},
580 	{ }
581 };
582 
583 static struct ctl_table ipv4_net_table[] = {
584 	{
585 		.procname	= "tcp_max_tw_buckets",
586 		.data		= &init_net.ipv4.tcp_death_row.sysctl_max_tw_buckets,
587 		.maxlen		= sizeof(int),
588 		.mode		= 0644,
589 		.proc_handler	= proc_dointvec
590 	},
591 	{
592 		.procname	= "icmp_echo_ignore_all",
593 		.data		= &init_net.ipv4.sysctl_icmp_echo_ignore_all,
594 		.maxlen		= sizeof(u8),
595 		.mode		= 0644,
596 		.proc_handler	= proc_dou8vec_minmax,
597 		.extra1		= SYSCTL_ZERO,
598 		.extra2		= SYSCTL_ONE
599 	},
600 	{
601 		.procname	= "icmp_echo_enable_probe",
602 		.data		= &init_net.ipv4.sysctl_icmp_echo_enable_probe,
603 		.maxlen		= sizeof(u8),
604 		.mode		= 0644,
605 		.proc_handler	= proc_dou8vec_minmax,
606 		.extra1		= SYSCTL_ZERO,
607 		.extra2		= SYSCTL_ONE
608 	},
609 	{
610 		.procname	= "icmp_echo_ignore_broadcasts",
611 		.data		= &init_net.ipv4.sysctl_icmp_echo_ignore_broadcasts,
612 		.maxlen		= sizeof(u8),
613 		.mode		= 0644,
614 		.proc_handler	= proc_dou8vec_minmax,
615 		.extra1		= SYSCTL_ZERO,
616 		.extra2		= SYSCTL_ONE
617 	},
618 	{
619 		.procname	= "icmp_ignore_bogus_error_responses",
620 		.data		= &init_net.ipv4.sysctl_icmp_ignore_bogus_error_responses,
621 		.maxlen		= sizeof(u8),
622 		.mode		= 0644,
623 		.proc_handler	= proc_dou8vec_minmax,
624 		.extra1		= SYSCTL_ZERO,
625 		.extra2		= SYSCTL_ONE
626 	},
627 	{
628 		.procname	= "icmp_errors_use_inbound_ifaddr",
629 		.data		= &init_net.ipv4.sysctl_icmp_errors_use_inbound_ifaddr,
630 		.maxlen		= sizeof(u8),
631 		.mode		= 0644,
632 		.proc_handler	= proc_dou8vec_minmax,
633 		.extra1		= SYSCTL_ZERO,
634 		.extra2		= SYSCTL_ONE
635 	},
636 	{
637 		.procname	= "icmp_ratelimit",
638 		.data		= &init_net.ipv4.sysctl_icmp_ratelimit,
639 		.maxlen		= sizeof(int),
640 		.mode		= 0644,
641 		.proc_handler	= proc_dointvec_ms_jiffies,
642 	},
643 	{
644 		.procname	= "icmp_ratemask",
645 		.data		= &init_net.ipv4.sysctl_icmp_ratemask,
646 		.maxlen		= sizeof(int),
647 		.mode		= 0644,
648 		.proc_handler	= proc_dointvec
649 	},
650 	{
651 		.procname	= "ping_group_range",
652 		.data		= &init_net.ipv4.ping_group_range.range,
653 		.maxlen		= sizeof(gid_t)*2,
654 		.mode		= 0644,
655 		.proc_handler	= ipv4_ping_group_range,
656 	},
657 #ifdef CONFIG_NET_L3_MASTER_DEV
658 	{
659 		.procname	= "raw_l3mdev_accept",
660 		.data		= &init_net.ipv4.sysctl_raw_l3mdev_accept,
661 		.maxlen		= sizeof(u8),
662 		.mode		= 0644,
663 		.proc_handler	= proc_dou8vec_minmax,
664 		.extra1		= SYSCTL_ZERO,
665 		.extra2		= SYSCTL_ONE,
666 	},
667 #endif
668 	{
669 		.procname	= "tcp_ecn",
670 		.data		= &init_net.ipv4.sysctl_tcp_ecn,
671 		.maxlen		= sizeof(u8),
672 		.mode		= 0644,
673 		.proc_handler	= proc_dou8vec_minmax,
674 		.extra1		= SYSCTL_ZERO,
675 		.extra2		= SYSCTL_TWO,
676 	},
677 	{
678 		.procname	= "tcp_ecn_fallback",
679 		.data		= &init_net.ipv4.sysctl_tcp_ecn_fallback,
680 		.maxlen		= sizeof(u8),
681 		.mode		= 0644,
682 		.proc_handler	= proc_dou8vec_minmax,
683 		.extra1		= SYSCTL_ZERO,
684 		.extra2		= SYSCTL_ONE,
685 	},
686 	{
687 		.procname	= "ip_dynaddr",
688 		.data		= &init_net.ipv4.sysctl_ip_dynaddr,
689 		.maxlen		= sizeof(u8),
690 		.mode		= 0644,
691 		.proc_handler	= proc_dou8vec_minmax,
692 	},
693 	{
694 		.procname	= "ip_early_demux",
695 		.data		= &init_net.ipv4.sysctl_ip_early_demux,
696 		.maxlen		= sizeof(u8),
697 		.mode		= 0644,
698 		.proc_handler	= proc_dou8vec_minmax,
699 	},
700 	{
701 		.procname       = "udp_early_demux",
702 		.data           = &init_net.ipv4.sysctl_udp_early_demux,
703 		.maxlen         = sizeof(u8),
704 		.mode           = 0644,
705 		.proc_handler   = proc_dou8vec_minmax,
706 	},
707 	{
708 		.procname       = "tcp_early_demux",
709 		.data           = &init_net.ipv4.sysctl_tcp_early_demux,
710 		.maxlen         = sizeof(u8),
711 		.mode           = 0644,
712 		.proc_handler   = proc_dou8vec_minmax,
713 	},
714 	{
715 		.procname       = "nexthop_compat_mode",
716 		.data           = &init_net.ipv4.sysctl_nexthop_compat_mode,
717 		.maxlen         = sizeof(u8),
718 		.mode           = 0644,
719 		.proc_handler   = proc_dou8vec_minmax,
720 		.extra1		= SYSCTL_ZERO,
721 		.extra2		= SYSCTL_ONE,
722 	},
723 	{
724 		.procname	= "ip_default_ttl",
725 		.data		= &init_net.ipv4.sysctl_ip_default_ttl,
726 		.maxlen		= sizeof(u8),
727 		.mode		= 0644,
728 		.proc_handler	= proc_dou8vec_minmax,
729 		.extra1		= &ip_ttl_min,
730 		.extra2		= &ip_ttl_max,
731 	},
732 	{
733 		.procname	= "ip_local_port_range",
734 		.maxlen		= sizeof(init_net.ipv4.ip_local_ports.range),
735 		.data		= &init_net.ipv4.ip_local_ports.range,
736 		.mode		= 0644,
737 		.proc_handler	= ipv4_local_port_range,
738 	},
739 	{
740 		.procname	= "ip_local_reserved_ports",
741 		.data		= &init_net.ipv4.sysctl_local_reserved_ports,
742 		.maxlen		= 65536,
743 		.mode		= 0644,
744 		.proc_handler	= proc_do_large_bitmap,
745 	},
746 	{
747 		.procname	= "ip_no_pmtu_disc",
748 		.data		= &init_net.ipv4.sysctl_ip_no_pmtu_disc,
749 		.maxlen		= sizeof(u8),
750 		.mode		= 0644,
751 		.proc_handler	= proc_dou8vec_minmax,
752 	},
753 	{
754 		.procname	= "ip_forward_use_pmtu",
755 		.data		= &init_net.ipv4.sysctl_ip_fwd_use_pmtu,
756 		.maxlen		= sizeof(u8),
757 		.mode		= 0644,
758 		.proc_handler	= proc_dou8vec_minmax,
759 	},
760 	{
761 		.procname	= "ip_forward_update_priority",
762 		.data		= &init_net.ipv4.sysctl_ip_fwd_update_priority,
763 		.maxlen		= sizeof(u8),
764 		.mode		= 0644,
765 		.proc_handler   = ipv4_fwd_update_priority,
766 		.extra1		= SYSCTL_ZERO,
767 		.extra2		= SYSCTL_ONE,
768 	},
769 	{
770 		.procname	= "ip_nonlocal_bind",
771 		.data		= &init_net.ipv4.sysctl_ip_nonlocal_bind,
772 		.maxlen		= sizeof(u8),
773 		.mode		= 0644,
774 		.proc_handler	= proc_dou8vec_minmax,
775 	},
776 	{
777 		.procname	= "ip_autobind_reuse",
778 		.data		= &init_net.ipv4.sysctl_ip_autobind_reuse,
779 		.maxlen		= sizeof(u8),
780 		.mode		= 0644,
781 		.proc_handler	= proc_dou8vec_minmax,
782 		.extra1         = SYSCTL_ZERO,
783 		.extra2         = SYSCTL_ONE,
784 	},
785 	{
786 		.procname	= "fwmark_reflect",
787 		.data		= &init_net.ipv4.sysctl_fwmark_reflect,
788 		.maxlen		= sizeof(u8),
789 		.mode		= 0644,
790 		.proc_handler	= proc_dou8vec_minmax,
791 	},
792 	{
793 		.procname	= "tcp_fwmark_accept",
794 		.data		= &init_net.ipv4.sysctl_tcp_fwmark_accept,
795 		.maxlen		= sizeof(u8),
796 		.mode		= 0644,
797 		.proc_handler	= proc_dou8vec_minmax,
798 	},
799 #ifdef CONFIG_NET_L3_MASTER_DEV
800 	{
801 		.procname	= "tcp_l3mdev_accept",
802 		.data		= &init_net.ipv4.sysctl_tcp_l3mdev_accept,
803 		.maxlen		= sizeof(u8),
804 		.mode		= 0644,
805 		.proc_handler	= proc_dou8vec_minmax,
806 		.extra1		= SYSCTL_ZERO,
807 		.extra2		= SYSCTL_ONE,
808 	},
809 #endif
810 	{
811 		.procname	= "tcp_mtu_probing",
812 		.data		= &init_net.ipv4.sysctl_tcp_mtu_probing,
813 		.maxlen		= sizeof(u8),
814 		.mode		= 0644,
815 		.proc_handler	= proc_dou8vec_minmax,
816 	},
817 	{
818 		.procname	= "tcp_base_mss",
819 		.data		= &init_net.ipv4.sysctl_tcp_base_mss,
820 		.maxlen		= sizeof(int),
821 		.mode		= 0644,
822 		.proc_handler	= proc_dointvec,
823 	},
824 	{
825 		.procname	= "tcp_min_snd_mss",
826 		.data		= &init_net.ipv4.sysctl_tcp_min_snd_mss,
827 		.maxlen		= sizeof(int),
828 		.mode		= 0644,
829 		.proc_handler	= proc_dointvec_minmax,
830 		.extra1		= &tcp_min_snd_mss_min,
831 		.extra2		= &tcp_min_snd_mss_max,
832 	},
833 	{
834 		.procname	= "tcp_mtu_probe_floor",
835 		.data		= &init_net.ipv4.sysctl_tcp_mtu_probe_floor,
836 		.maxlen		= sizeof(int),
837 		.mode		= 0644,
838 		.proc_handler	= proc_dointvec_minmax,
839 		.extra1		= &tcp_min_snd_mss_min,
840 		.extra2		= &tcp_min_snd_mss_max,
841 	},
842 	{
843 		.procname	= "tcp_probe_threshold",
844 		.data		= &init_net.ipv4.sysctl_tcp_probe_threshold,
845 		.maxlen		= sizeof(int),
846 		.mode		= 0644,
847 		.proc_handler	= proc_dointvec,
848 	},
849 	{
850 		.procname	= "tcp_probe_interval",
851 		.data		= &init_net.ipv4.sysctl_tcp_probe_interval,
852 		.maxlen		= sizeof(u32),
853 		.mode		= 0644,
854 		.proc_handler	= proc_douintvec_minmax,
855 		.extra2		= &u32_max_div_HZ,
856 	},
857 	{
858 		.procname	= "igmp_link_local_mcast_reports",
859 		.data		= &init_net.ipv4.sysctl_igmp_llm_reports,
860 		.maxlen		= sizeof(u8),
861 		.mode		= 0644,
862 		.proc_handler	= proc_dou8vec_minmax,
863 	},
864 	{
865 		.procname	= "igmp_max_memberships",
866 		.data		= &init_net.ipv4.sysctl_igmp_max_memberships,
867 		.maxlen		= sizeof(int),
868 		.mode		= 0644,
869 		.proc_handler	= proc_dointvec
870 	},
871 	{
872 		.procname	= "igmp_max_msf",
873 		.data		= &init_net.ipv4.sysctl_igmp_max_msf,
874 		.maxlen		= sizeof(int),
875 		.mode		= 0644,
876 		.proc_handler	= proc_dointvec
877 	},
878 #ifdef CONFIG_IP_MULTICAST
879 	{
880 		.procname	= "igmp_qrv",
881 		.data		= &init_net.ipv4.sysctl_igmp_qrv,
882 		.maxlen		= sizeof(int),
883 		.mode		= 0644,
884 		.proc_handler	= proc_dointvec_minmax,
885 		.extra1		= SYSCTL_ONE
886 	},
887 #endif
888 	{
889 		.procname	= "tcp_congestion_control",
890 		.data		= &init_net.ipv4.tcp_congestion_control,
891 		.mode		= 0644,
892 		.maxlen		= TCP_CA_NAME_MAX,
893 		.proc_handler	= proc_tcp_congestion_control,
894 	},
895 	{
896 		.procname	= "tcp_available_congestion_control",
897 		.maxlen		= TCP_CA_BUF_MAX,
898 		.mode		= 0444,
899 		.proc_handler   = proc_tcp_available_congestion_control,
900 	},
901 	{
902 		.procname	= "tcp_allowed_congestion_control",
903 		.maxlen		= TCP_CA_BUF_MAX,
904 		.mode		= 0644,
905 		.proc_handler   = proc_allowed_congestion_control,
906 	},
907 	{
908 		.procname	= "tcp_keepalive_time",
909 		.data		= &init_net.ipv4.sysctl_tcp_keepalive_time,
910 		.maxlen		= sizeof(int),
911 		.mode		= 0644,
912 		.proc_handler	= proc_dointvec_jiffies,
913 	},
914 	{
915 		.procname	= "tcp_keepalive_probes",
916 		.data		= &init_net.ipv4.sysctl_tcp_keepalive_probes,
917 		.maxlen		= sizeof(u8),
918 		.mode		= 0644,
919 		.proc_handler	= proc_dou8vec_minmax,
920 	},
921 	{
922 		.procname	= "tcp_keepalive_intvl",
923 		.data		= &init_net.ipv4.sysctl_tcp_keepalive_intvl,
924 		.maxlen		= sizeof(int),
925 		.mode		= 0644,
926 		.proc_handler	= proc_dointvec_jiffies,
927 	},
928 	{
929 		.procname	= "tcp_syn_retries",
930 		.data		= &init_net.ipv4.sysctl_tcp_syn_retries,
931 		.maxlen		= sizeof(u8),
932 		.mode		= 0644,
933 		.proc_handler	= proc_dou8vec_minmax,
934 		.extra1		= &tcp_syn_retries_min,
935 		.extra2		= &tcp_syn_retries_max
936 	},
937 	{
938 		.procname	= "tcp_synack_retries",
939 		.data		= &init_net.ipv4.sysctl_tcp_synack_retries,
940 		.maxlen		= sizeof(u8),
941 		.mode		= 0644,
942 		.proc_handler	= proc_dou8vec_minmax,
943 	},
944 #ifdef CONFIG_SYN_COOKIES
945 	{
946 		.procname	= "tcp_syncookies",
947 		.data		= &init_net.ipv4.sysctl_tcp_syncookies,
948 		.maxlen		= sizeof(u8),
949 		.mode		= 0644,
950 		.proc_handler	= proc_dou8vec_minmax,
951 	},
952 #endif
953 	{
954 		.procname	= "tcp_migrate_req",
955 		.data		= &init_net.ipv4.sysctl_tcp_migrate_req,
956 		.maxlen		= sizeof(u8),
957 		.mode		= 0644,
958 		.proc_handler	= proc_dou8vec_minmax,
959 		.extra1		= SYSCTL_ZERO,
960 		.extra2		= SYSCTL_ONE
961 	},
962 	{
963 		.procname	= "tcp_reordering",
964 		.data		= &init_net.ipv4.sysctl_tcp_reordering,
965 		.maxlen		= sizeof(int),
966 		.mode		= 0644,
967 		.proc_handler	= proc_dointvec
968 	},
969 	{
970 		.procname	= "tcp_retries1",
971 		.data		= &init_net.ipv4.sysctl_tcp_retries1,
972 		.maxlen		= sizeof(u8),
973 		.mode		= 0644,
974 		.proc_handler	= proc_dou8vec_minmax,
975 		.extra2		= &tcp_retr1_max
976 	},
977 	{
978 		.procname	= "tcp_retries2",
979 		.data		= &init_net.ipv4.sysctl_tcp_retries2,
980 		.maxlen		= sizeof(u8),
981 		.mode		= 0644,
982 		.proc_handler	= proc_dou8vec_minmax,
983 	},
984 	{
985 		.procname	= "tcp_orphan_retries",
986 		.data		= &init_net.ipv4.sysctl_tcp_orphan_retries,
987 		.maxlen		= sizeof(u8),
988 		.mode		= 0644,
989 		.proc_handler	= proc_dou8vec_minmax,
990 	},
991 	{
992 		.procname	= "tcp_fin_timeout",
993 		.data		= &init_net.ipv4.sysctl_tcp_fin_timeout,
994 		.maxlen		= sizeof(int),
995 		.mode		= 0644,
996 		.proc_handler	= proc_dointvec_jiffies,
997 	},
998 	{
999 		.procname	= "tcp_notsent_lowat",
1000 		.data		= &init_net.ipv4.sysctl_tcp_notsent_lowat,
1001 		.maxlen		= sizeof(unsigned int),
1002 		.mode		= 0644,
1003 		.proc_handler	= proc_douintvec,
1004 	},
1005 	{
1006 		.procname	= "tcp_tw_reuse",
1007 		.data		= &init_net.ipv4.sysctl_tcp_tw_reuse,
1008 		.maxlen		= sizeof(u8),
1009 		.mode		= 0644,
1010 		.proc_handler	= proc_dou8vec_minmax,
1011 		.extra1		= SYSCTL_ZERO,
1012 		.extra2		= SYSCTL_TWO,
1013 	},
1014 	{
1015 		.procname	= "tcp_max_syn_backlog",
1016 		.data		= &init_net.ipv4.sysctl_max_syn_backlog,
1017 		.maxlen		= sizeof(int),
1018 		.mode		= 0644,
1019 		.proc_handler	= proc_dointvec
1020 	},
1021 	{
1022 		.procname	= "tcp_fastopen",
1023 		.data		= &init_net.ipv4.sysctl_tcp_fastopen,
1024 		.maxlen		= sizeof(int),
1025 		.mode		= 0644,
1026 		.proc_handler	= proc_dointvec,
1027 	},
1028 	{
1029 		.procname	= "tcp_fastopen_key",
1030 		.mode		= 0600,
1031 		.data		= &init_net.ipv4.sysctl_tcp_fastopen,
1032 		/* maxlen to print the list of keys in hex (*2), with dashes
1033 		 * separating doublewords and a comma in between keys.
1034 		 */
1035 		.maxlen		= ((TCP_FASTOPEN_KEY_LENGTH *
1036 				   2 * TCP_FASTOPEN_KEY_MAX) +
1037 				   (TCP_FASTOPEN_KEY_MAX * 5)),
1038 		.proc_handler	= proc_tcp_fastopen_key,
1039 	},
1040 	{
1041 		.procname	= "tcp_fastopen_blackhole_timeout_sec",
1042 		.data		= &init_net.ipv4.sysctl_tcp_fastopen_blackhole_timeout,
1043 		.maxlen		= sizeof(int),
1044 		.mode		= 0644,
1045 		.proc_handler	= proc_tfo_blackhole_detect_timeout,
1046 		.extra1		= SYSCTL_ZERO,
1047 	},
1048 #ifdef CONFIG_IP_ROUTE_MULTIPATH
1049 	{
1050 		.procname	= "fib_multipath_use_neigh",
1051 		.data		= &init_net.ipv4.sysctl_fib_multipath_use_neigh,
1052 		.maxlen		= sizeof(u8),
1053 		.mode		= 0644,
1054 		.proc_handler	= proc_dou8vec_minmax,
1055 		.extra1		= SYSCTL_ZERO,
1056 		.extra2		= SYSCTL_ONE,
1057 	},
1058 	{
1059 		.procname	= "fib_multipath_hash_policy",
1060 		.data		= &init_net.ipv4.sysctl_fib_multipath_hash_policy,
1061 		.maxlen		= sizeof(u8),
1062 		.mode		= 0644,
1063 		.proc_handler	= proc_fib_multipath_hash_policy,
1064 		.extra1		= SYSCTL_ZERO,
1065 		.extra2		= SYSCTL_THREE,
1066 	},
1067 	{
1068 		.procname	= "fib_multipath_hash_fields",
1069 		.data		= &init_net.ipv4.sysctl_fib_multipath_hash_fields,
1070 		.maxlen		= sizeof(u32),
1071 		.mode		= 0644,
1072 		.proc_handler	= proc_fib_multipath_hash_fields,
1073 		.extra1		= SYSCTL_ONE,
1074 		.extra2		= &fib_multipath_hash_fields_all_mask,
1075 	},
1076 #endif
1077 	{
1078 		.procname	= "ip_unprivileged_port_start",
1079 		.maxlen		= sizeof(int),
1080 		.data		= &init_net.ipv4.sysctl_ip_prot_sock,
1081 		.mode		= 0644,
1082 		.proc_handler	= ipv4_privileged_ports,
1083 	},
1084 #ifdef CONFIG_NET_L3_MASTER_DEV
1085 	{
1086 		.procname	= "udp_l3mdev_accept",
1087 		.data		= &init_net.ipv4.sysctl_udp_l3mdev_accept,
1088 		.maxlen		= sizeof(u8),
1089 		.mode		= 0644,
1090 		.proc_handler	= proc_dou8vec_minmax,
1091 		.extra1		= SYSCTL_ZERO,
1092 		.extra2		= SYSCTL_ONE,
1093 	},
1094 #endif
1095 	{
1096 		.procname	= "tcp_sack",
1097 		.data		= &init_net.ipv4.sysctl_tcp_sack,
1098 		.maxlen		= sizeof(u8),
1099 		.mode		= 0644,
1100 		.proc_handler	= proc_dou8vec_minmax,
1101 	},
1102 	{
1103 		.procname	= "tcp_window_scaling",
1104 		.data		= &init_net.ipv4.sysctl_tcp_window_scaling,
1105 		.maxlen		= sizeof(u8),
1106 		.mode		= 0644,
1107 		.proc_handler	= proc_dou8vec_minmax,
1108 	},
1109 	{
1110 		.procname	= "tcp_timestamps",
1111 		.data		= &init_net.ipv4.sysctl_tcp_timestamps,
1112 		.maxlen		= sizeof(u8),
1113 		.mode		= 0644,
1114 		.proc_handler	= proc_dou8vec_minmax,
1115 	},
1116 	{
1117 		.procname	= "tcp_early_retrans",
1118 		.data		= &init_net.ipv4.sysctl_tcp_early_retrans,
1119 		.maxlen		= sizeof(u8),
1120 		.mode		= 0644,
1121 		.proc_handler	= proc_dou8vec_minmax,
1122 		.extra1		= SYSCTL_ZERO,
1123 		.extra2		= SYSCTL_FOUR,
1124 	},
1125 	{
1126 		.procname	= "tcp_recovery",
1127 		.data		= &init_net.ipv4.sysctl_tcp_recovery,
1128 		.maxlen		= sizeof(u8),
1129 		.mode		= 0644,
1130 		.proc_handler	= proc_dou8vec_minmax,
1131 	},
1132 	{
1133 		.procname       = "tcp_thin_linear_timeouts",
1134 		.data           = &init_net.ipv4.sysctl_tcp_thin_linear_timeouts,
1135 		.maxlen         = sizeof(u8),
1136 		.mode           = 0644,
1137 		.proc_handler   = proc_dou8vec_minmax,
1138 	},
1139 	{
1140 		.procname	= "tcp_slow_start_after_idle",
1141 		.data		= &init_net.ipv4.sysctl_tcp_slow_start_after_idle,
1142 		.maxlen		= sizeof(u8),
1143 		.mode		= 0644,
1144 		.proc_handler	= proc_dou8vec_minmax,
1145 	},
1146 	{
1147 		.procname	= "tcp_retrans_collapse",
1148 		.data		= &init_net.ipv4.sysctl_tcp_retrans_collapse,
1149 		.maxlen		= sizeof(u8),
1150 		.mode		= 0644,
1151 		.proc_handler	= proc_dou8vec_minmax,
1152 	},
1153 	{
1154 		.procname	= "tcp_stdurg",
1155 		.data		= &init_net.ipv4.sysctl_tcp_stdurg,
1156 		.maxlen		= sizeof(u8),
1157 		.mode		= 0644,
1158 		.proc_handler	= proc_dou8vec_minmax,
1159 	},
1160 	{
1161 		.procname	= "tcp_rfc1337",
1162 		.data		= &init_net.ipv4.sysctl_tcp_rfc1337,
1163 		.maxlen		= sizeof(u8),
1164 		.mode		= 0644,
1165 		.proc_handler	= proc_dou8vec_minmax,
1166 	},
1167 	{
1168 		.procname	= "tcp_abort_on_overflow",
1169 		.data		= &init_net.ipv4.sysctl_tcp_abort_on_overflow,
1170 		.maxlen		= sizeof(u8),
1171 		.mode		= 0644,
1172 		.proc_handler	= proc_dou8vec_minmax,
1173 	},
1174 	{
1175 		.procname	= "tcp_fack",
1176 		.data		= &init_net.ipv4.sysctl_tcp_fack,
1177 		.maxlen		= sizeof(u8),
1178 		.mode		= 0644,
1179 		.proc_handler	= proc_dou8vec_minmax,
1180 	},
1181 	{
1182 		.procname	= "tcp_max_reordering",
1183 		.data		= &init_net.ipv4.sysctl_tcp_max_reordering,
1184 		.maxlen		= sizeof(int),
1185 		.mode		= 0644,
1186 		.proc_handler	= proc_dointvec
1187 	},
1188 	{
1189 		.procname	= "tcp_dsack",
1190 		.data		= &init_net.ipv4.sysctl_tcp_dsack,
1191 		.maxlen		= sizeof(u8),
1192 		.mode		= 0644,
1193 		.proc_handler	= proc_dou8vec_minmax,
1194 	},
1195 	{
1196 		.procname	= "tcp_app_win",
1197 		.data		= &init_net.ipv4.sysctl_tcp_app_win,
1198 		.maxlen		= sizeof(u8),
1199 		.mode		= 0644,
1200 		.proc_handler	= proc_dou8vec_minmax,
1201 	},
1202 	{
1203 		.procname	= "tcp_adv_win_scale",
1204 		.data		= &init_net.ipv4.sysctl_tcp_adv_win_scale,
1205 		.maxlen		= sizeof(int),
1206 		.mode		= 0644,
1207 		.proc_handler	= proc_dointvec_minmax,
1208 		.extra1		= &tcp_adv_win_scale_min,
1209 		.extra2		= &tcp_adv_win_scale_max,
1210 	},
1211 	{
1212 		.procname	= "tcp_frto",
1213 		.data		= &init_net.ipv4.sysctl_tcp_frto,
1214 		.maxlen		= sizeof(u8),
1215 		.mode		= 0644,
1216 		.proc_handler	= proc_dou8vec_minmax,
1217 	},
1218 	{
1219 		.procname	= "tcp_no_metrics_save",
1220 		.data		= &init_net.ipv4.sysctl_tcp_nometrics_save,
1221 		.maxlen		= sizeof(u8),
1222 		.mode		= 0644,
1223 		.proc_handler	= proc_dou8vec_minmax,
1224 	},
1225 	{
1226 		.procname	= "tcp_no_ssthresh_metrics_save",
1227 		.data		= &init_net.ipv4.sysctl_tcp_no_ssthresh_metrics_save,
1228 		.maxlen		= sizeof(u8),
1229 		.mode		= 0644,
1230 		.proc_handler	= proc_dou8vec_minmax,
1231 		.extra1		= SYSCTL_ZERO,
1232 		.extra2		= SYSCTL_ONE,
1233 	},
1234 	{
1235 		.procname	= "tcp_moderate_rcvbuf",
1236 		.data		= &init_net.ipv4.sysctl_tcp_moderate_rcvbuf,
1237 		.maxlen		= sizeof(u8),
1238 		.mode		= 0644,
1239 		.proc_handler	= proc_dou8vec_minmax,
1240 	},
1241 	{
1242 		.procname	= "tcp_tso_win_divisor",
1243 		.data		= &init_net.ipv4.sysctl_tcp_tso_win_divisor,
1244 		.maxlen		= sizeof(u8),
1245 		.mode		= 0644,
1246 		.proc_handler	= proc_dou8vec_minmax,
1247 	},
1248 	{
1249 		.procname	= "tcp_workaround_signed_windows",
1250 		.data		= &init_net.ipv4.sysctl_tcp_workaround_signed_windows,
1251 		.maxlen		= sizeof(u8),
1252 		.mode		= 0644,
1253 		.proc_handler	= proc_dou8vec_minmax,
1254 	},
1255 	{
1256 		.procname	= "tcp_limit_output_bytes",
1257 		.data		= &init_net.ipv4.sysctl_tcp_limit_output_bytes,
1258 		.maxlen		= sizeof(int),
1259 		.mode		= 0644,
1260 		.proc_handler	= proc_dointvec
1261 	},
1262 	{
1263 		.procname	= "tcp_challenge_ack_limit",
1264 		.data		= &init_net.ipv4.sysctl_tcp_challenge_ack_limit,
1265 		.maxlen		= sizeof(int),
1266 		.mode		= 0644,
1267 		.proc_handler	= proc_dointvec
1268 	},
1269 	{
1270 		.procname	= "tcp_min_tso_segs",
1271 		.data		= &init_net.ipv4.sysctl_tcp_min_tso_segs,
1272 		.maxlen		= sizeof(u8),
1273 		.mode		= 0644,
1274 		.proc_handler	= proc_dou8vec_minmax,
1275 		.extra1		= SYSCTL_ONE,
1276 	},
1277 	{
1278 		.procname	= "tcp_tso_rtt_log",
1279 		.data		= &init_net.ipv4.sysctl_tcp_tso_rtt_log,
1280 		.maxlen		= sizeof(u8),
1281 		.mode		= 0644,
1282 		.proc_handler	= proc_dou8vec_minmax,
1283 	},
1284 	{
1285 		.procname	= "tcp_min_rtt_wlen",
1286 		.data		= &init_net.ipv4.sysctl_tcp_min_rtt_wlen,
1287 		.maxlen		= sizeof(int),
1288 		.mode		= 0644,
1289 		.proc_handler	= proc_dointvec_minmax,
1290 		.extra1		= SYSCTL_ZERO,
1291 		.extra2		= &one_day_secs
1292 	},
1293 	{
1294 		.procname	= "tcp_autocorking",
1295 		.data		= &init_net.ipv4.sysctl_tcp_autocorking,
1296 		.maxlen		= sizeof(u8),
1297 		.mode		= 0644,
1298 		.proc_handler	= proc_dou8vec_minmax,
1299 		.extra1		= SYSCTL_ZERO,
1300 		.extra2		= SYSCTL_ONE,
1301 	},
1302 	{
1303 		.procname	= "tcp_invalid_ratelimit",
1304 		.data		= &init_net.ipv4.sysctl_tcp_invalid_ratelimit,
1305 		.maxlen		= sizeof(int),
1306 		.mode		= 0644,
1307 		.proc_handler	= proc_dointvec_ms_jiffies,
1308 	},
1309 	{
1310 		.procname	= "tcp_pacing_ss_ratio",
1311 		.data		= &init_net.ipv4.sysctl_tcp_pacing_ss_ratio,
1312 		.maxlen		= sizeof(int),
1313 		.mode		= 0644,
1314 		.proc_handler	= proc_dointvec_minmax,
1315 		.extra1		= SYSCTL_ZERO,
1316 		.extra2		= SYSCTL_ONE_THOUSAND,
1317 	},
1318 	{
1319 		.procname	= "tcp_pacing_ca_ratio",
1320 		.data		= &init_net.ipv4.sysctl_tcp_pacing_ca_ratio,
1321 		.maxlen		= sizeof(int),
1322 		.mode		= 0644,
1323 		.proc_handler	= proc_dointvec_minmax,
1324 		.extra1		= SYSCTL_ZERO,
1325 		.extra2		= SYSCTL_ONE_THOUSAND,
1326 	},
1327 	{
1328 		.procname	= "tcp_wmem",
1329 		.data		= &init_net.ipv4.sysctl_tcp_wmem,
1330 		.maxlen		= sizeof(init_net.ipv4.sysctl_tcp_wmem),
1331 		.mode		= 0644,
1332 		.proc_handler	= proc_dointvec_minmax,
1333 		.extra1		= SYSCTL_ONE,
1334 	},
1335 	{
1336 		.procname	= "tcp_rmem",
1337 		.data		= &init_net.ipv4.sysctl_tcp_rmem,
1338 		.maxlen		= sizeof(init_net.ipv4.sysctl_tcp_rmem),
1339 		.mode		= 0644,
1340 		.proc_handler	= proc_dointvec_minmax,
1341 		.extra1		= SYSCTL_ONE,
1342 	},
1343 	{
1344 		.procname	= "tcp_comp_sack_delay_ns",
1345 		.data		= &init_net.ipv4.sysctl_tcp_comp_sack_delay_ns,
1346 		.maxlen		= sizeof(unsigned long),
1347 		.mode		= 0644,
1348 		.proc_handler	= proc_doulongvec_minmax,
1349 	},
1350 	{
1351 		.procname	= "tcp_comp_sack_slack_ns",
1352 		.data		= &init_net.ipv4.sysctl_tcp_comp_sack_slack_ns,
1353 		.maxlen		= sizeof(unsigned long),
1354 		.mode		= 0644,
1355 		.proc_handler	= proc_doulongvec_minmax,
1356 	},
1357 	{
1358 		.procname	= "tcp_comp_sack_nr",
1359 		.data		= &init_net.ipv4.sysctl_tcp_comp_sack_nr,
1360 		.maxlen		= sizeof(u8),
1361 		.mode		= 0644,
1362 		.proc_handler	= proc_dou8vec_minmax,
1363 		.extra1		= SYSCTL_ZERO,
1364 	},
1365 	{
1366 		.procname       = "tcp_reflect_tos",
1367 		.data           = &init_net.ipv4.sysctl_tcp_reflect_tos,
1368 		.maxlen         = sizeof(u8),
1369 		.mode           = 0644,
1370 		.proc_handler   = proc_dou8vec_minmax,
1371 		.extra1         = SYSCTL_ZERO,
1372 		.extra2         = SYSCTL_ONE,
1373 	},
1374 	{
1375 		.procname	= "tcp_ehash_entries",
1376 		.data		= &init_net.ipv4.sysctl_tcp_child_ehash_entries,
1377 		.mode		= 0444,
1378 		.proc_handler	= proc_tcp_ehash_entries,
1379 	},
1380 	{
1381 		.procname	= "tcp_child_ehash_entries",
1382 		.data		= &init_net.ipv4.sysctl_tcp_child_ehash_entries,
1383 		.maxlen		= sizeof(unsigned int),
1384 		.mode		= 0644,
1385 		.proc_handler	= proc_douintvec_minmax,
1386 		.extra1		= SYSCTL_ZERO,
1387 		.extra2		= &tcp_child_ehash_entries_max,
1388 	},
1389 	{
1390 		.procname	= "udp_hash_entries",
1391 		.data		= &init_net.ipv4.sysctl_udp_child_hash_entries,
1392 		.mode		= 0444,
1393 		.proc_handler	= proc_udp_hash_entries,
1394 	},
1395 	{
1396 		.procname	= "udp_child_hash_entries",
1397 		.data		= &init_net.ipv4.sysctl_udp_child_hash_entries,
1398 		.maxlen		= sizeof(unsigned int),
1399 		.mode		= 0644,
1400 		.proc_handler	= proc_douintvec_minmax,
1401 		.extra1		= SYSCTL_ZERO,
1402 		.extra2		= &udp_child_hash_entries_max,
1403 	},
1404 	{
1405 		.procname	= "udp_rmem_min",
1406 		.data		= &init_net.ipv4.sysctl_udp_rmem_min,
1407 		.maxlen		= sizeof(init_net.ipv4.sysctl_udp_rmem_min),
1408 		.mode		= 0644,
1409 		.proc_handler	= proc_dointvec_minmax,
1410 		.extra1		= SYSCTL_ONE
1411 	},
1412 	{
1413 		.procname	= "udp_wmem_min",
1414 		.data		= &init_net.ipv4.sysctl_udp_wmem_min,
1415 		.maxlen		= sizeof(init_net.ipv4.sysctl_udp_wmem_min),
1416 		.mode		= 0644,
1417 		.proc_handler	= proc_dointvec_minmax,
1418 		.extra1		= SYSCTL_ONE
1419 	},
1420 	{
1421 		.procname	= "fib_notify_on_flag_change",
1422 		.data		= &init_net.ipv4.sysctl_fib_notify_on_flag_change,
1423 		.maxlen		= sizeof(u8),
1424 		.mode		= 0644,
1425 		.proc_handler	= proc_dou8vec_minmax,
1426 		.extra1		= SYSCTL_ZERO,
1427 		.extra2		= SYSCTL_TWO,
1428 	},
1429 	{
1430 		.procname       = "tcp_plb_enabled",
1431 		.data           = &init_net.ipv4.sysctl_tcp_plb_enabled,
1432 		.maxlen         = sizeof(u8),
1433 		.mode           = 0644,
1434 		.proc_handler   = proc_dou8vec_minmax,
1435 		.extra1         = SYSCTL_ZERO,
1436 		.extra2         = SYSCTL_ONE,
1437 	},
1438 	{
1439 		.procname       = "tcp_plb_idle_rehash_rounds",
1440 		.data           = &init_net.ipv4.sysctl_tcp_plb_idle_rehash_rounds,
1441 		.maxlen         = sizeof(u8),
1442 		.mode           = 0644,
1443 		.proc_handler   = proc_dou8vec_minmax,
1444 		.extra2		= &tcp_plb_max_rounds,
1445 	},
1446 	{
1447 		.procname       = "tcp_plb_rehash_rounds",
1448 		.data           = &init_net.ipv4.sysctl_tcp_plb_rehash_rounds,
1449 		.maxlen         = sizeof(u8),
1450 		.mode           = 0644,
1451 		.proc_handler   = proc_dou8vec_minmax,
1452 		.extra2         = &tcp_plb_max_rounds,
1453 	},
1454 	{
1455 		.procname       = "tcp_plb_suspend_rto_sec",
1456 		.data           = &init_net.ipv4.sysctl_tcp_plb_suspend_rto_sec,
1457 		.maxlen         = sizeof(u8),
1458 		.mode           = 0644,
1459 		.proc_handler   = proc_dou8vec_minmax,
1460 	},
1461 	{
1462 		.procname       = "tcp_plb_cong_thresh",
1463 		.data           = &init_net.ipv4.sysctl_tcp_plb_cong_thresh,
1464 		.maxlen         = sizeof(int),
1465 		.mode           = 0644,
1466 		.proc_handler   = proc_dointvec_minmax,
1467 		.extra1         = SYSCTL_ZERO,
1468 		.extra2         = &tcp_plb_max_cong_thresh,
1469 	},
1470 	{ }
1471 };
1472 
1473 static __net_init int ipv4_sysctl_init_net(struct net *net)
1474 {
1475 	struct ctl_table *table;
1476 
1477 	table = ipv4_net_table;
1478 	if (!net_eq(net, &init_net)) {
1479 		int i;
1480 
1481 		table = kmemdup(table, sizeof(ipv4_net_table), GFP_KERNEL);
1482 		if (!table)
1483 			goto err_alloc;
1484 
1485 		for (i = 0; i < ARRAY_SIZE(ipv4_net_table) - 1; i++) {
1486 			if (table[i].data) {
1487 				/* Update the variables to point into
1488 				 * the current struct net
1489 				 */
1490 				table[i].data += (void *)net - (void *)&init_net;
1491 			} else {
1492 				/* Entries without data pointer are global;
1493 				 * Make them read-only in non-init_net ns
1494 				 */
1495 				table[i].mode &= ~0222;
1496 			}
1497 		}
1498 	}
1499 
1500 	net->ipv4.ipv4_hdr = register_net_sysctl(net, "net/ipv4", table);
1501 	if (!net->ipv4.ipv4_hdr)
1502 		goto err_reg;
1503 
1504 	net->ipv4.sysctl_local_reserved_ports = kzalloc(65536 / 8, GFP_KERNEL);
1505 	if (!net->ipv4.sysctl_local_reserved_ports)
1506 		goto err_ports;
1507 
1508 	return 0;
1509 
1510 err_ports:
1511 	unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
1512 err_reg:
1513 	if (!net_eq(net, &init_net))
1514 		kfree(table);
1515 err_alloc:
1516 	return -ENOMEM;
1517 }
1518 
1519 static __net_exit void ipv4_sysctl_exit_net(struct net *net)
1520 {
1521 	struct ctl_table *table;
1522 
1523 	kfree(net->ipv4.sysctl_local_reserved_ports);
1524 	table = net->ipv4.ipv4_hdr->ctl_table_arg;
1525 	unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
1526 	kfree(table);
1527 }
1528 
1529 static __net_initdata struct pernet_operations ipv4_sysctl_ops = {
1530 	.init = ipv4_sysctl_init_net,
1531 	.exit = ipv4_sysctl_exit_net,
1532 };
1533 
1534 static __init int sysctl_ipv4_init(void)
1535 {
1536 	struct ctl_table_header *hdr;
1537 
1538 	hdr = register_net_sysctl(&init_net, "net/ipv4", ipv4_table);
1539 	if (!hdr)
1540 		return -ENOMEM;
1541 
1542 	if (register_pernet_subsys(&ipv4_sysctl_ops)) {
1543 		unregister_net_sysctl_table(hdr);
1544 		return -ENOMEM;
1545 	}
1546 
1547 	return 0;
1548 }
1549 
1550 __initcall(sysctl_ipv4_init);
1551