xref: /linux/net/ipv4/sysctl_net_ipv4.c (revision 6dfafbd0299a60bfb5d5e277fdf100037c7ded07)
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 static u32 icmp_errors_extension_mask_all =
52 	GENMASK_U8(ICMP_ERR_EXT_COUNT - 1, 0);
53 
54 /* obsolete */
55 static int sysctl_tcp_low_latency __read_mostly;
56 
57 /* Update system visible IP port range */
58 static void set_local_port_range(struct net *net, unsigned int low, unsigned int high)
59 {
60 	bool same_parity = !((low ^ high) & 1);
61 
62 	if (same_parity && !net->ipv4.ip_local_ports.warned) {
63 		net->ipv4.ip_local_ports.warned = true;
64 		pr_err_ratelimited("ip_local_port_range: prefer different parity for start/end values.\n");
65 	}
66 	WRITE_ONCE(net->ipv4.ip_local_ports.range, high << 16 | low);
67 }
68 
69 /* Validate changes from /proc interface. */
70 static int ipv4_local_port_range(const struct ctl_table *table, int write,
71 				 void *buffer, size_t *lenp, loff_t *ppos)
72 {
73 	struct net *net = table->data;
74 	int ret;
75 	int range[2];
76 	struct ctl_table tmp = {
77 		.data = &range,
78 		.maxlen = sizeof(range),
79 		.mode = table->mode,
80 		.extra1 = &ip_local_port_range_min,
81 		.extra2 = &ip_local_port_range_max,
82 	};
83 
84 	inet_get_local_port_range(net, &range[0], &range[1]);
85 
86 	ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
87 
88 	if (write && ret == 0) {
89 		/* Ensure that the upper limit is not smaller than the lower,
90 		 * and that the lower does not encroach upon the privileged
91 		 * port limit.
92 		 */
93 		if ((range[1] < range[0]) ||
94 		    (range[0] < READ_ONCE(net->ipv4.sysctl_ip_prot_sock)))
95 			ret = -EINVAL;
96 		else
97 			set_local_port_range(net, range[0], range[1]);
98 	}
99 
100 	return ret;
101 }
102 
103 /* Validate changes from /proc interface. */
104 static int ipv4_privileged_ports(const struct ctl_table *table, int write,
105 				void *buffer, size_t *lenp, loff_t *ppos)
106 {
107 	struct net *net = container_of(table->data, struct net,
108 	    ipv4.sysctl_ip_prot_sock);
109 	int ret;
110 	int pports;
111 	int range[2];
112 	struct ctl_table tmp = {
113 		.data = &pports,
114 		.maxlen = sizeof(pports),
115 		.mode = table->mode,
116 		.extra1 = &ip_privileged_port_min,
117 		.extra2 = &ip_privileged_port_max,
118 	};
119 
120 	pports = READ_ONCE(net->ipv4.sysctl_ip_prot_sock);
121 
122 	ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
123 
124 	if (write && ret == 0) {
125 		inet_get_local_port_range(net, &range[0], &range[1]);
126 		/* Ensure that the local port range doesn't overlap with the
127 		 * privileged port range.
128 		 */
129 		if (range[0] < pports)
130 			ret = -EINVAL;
131 		else
132 			WRITE_ONCE(net->ipv4.sysctl_ip_prot_sock, pports);
133 	}
134 
135 	return ret;
136 }
137 
138 static void inet_get_ping_group_range_table(const struct ctl_table *table,
139 					    kgid_t *low, kgid_t *high)
140 {
141 	kgid_t *data = table->data;
142 	struct net *net =
143 		container_of(table->data, struct net, ipv4.ping_group_range.range);
144 	unsigned int seq;
145 	do {
146 		seq = read_seqbegin(&net->ipv4.ping_group_range.lock);
147 
148 		*low = data[0];
149 		*high = data[1];
150 	} while (read_seqretry(&net->ipv4.ping_group_range.lock, seq));
151 }
152 
153 /* Update system visible IP port range */
154 static void set_ping_group_range(const struct ctl_table *table,
155 				 kgid_t low, kgid_t high)
156 {
157 	kgid_t *data = table->data;
158 	struct net *net =
159 		container_of(table->data, struct net, ipv4.ping_group_range.range);
160 	write_seqlock(&net->ipv4.ping_group_range.lock);
161 	data[0] = low;
162 	data[1] = high;
163 	write_sequnlock(&net->ipv4.ping_group_range.lock);
164 }
165 
166 /* Validate changes from /proc interface. */
167 static int ipv4_ping_group_range(const struct ctl_table *table, int write,
168 				 void *buffer, size_t *lenp, loff_t *ppos)
169 {
170 	struct user_namespace *user_ns = current_user_ns();
171 	int ret;
172 	unsigned long urange[2];
173 	kgid_t low, high;
174 	struct ctl_table tmp = {
175 		.data = &urange,
176 		.maxlen = sizeof(urange),
177 		.mode = table->mode,
178 		.extra1 = &ip_ping_group_range_min,
179 		.extra2 = &ip_ping_group_range_max,
180 	};
181 
182 	inet_get_ping_group_range_table(table, &low, &high);
183 	urange[0] = from_kgid_munged(user_ns, low);
184 	urange[1] = from_kgid_munged(user_ns, high);
185 	ret = proc_doulongvec_minmax(&tmp, write, buffer, lenp, ppos);
186 
187 	if (write && ret == 0) {
188 		low = make_kgid(user_ns, urange[0]);
189 		high = make_kgid(user_ns, urange[1]);
190 		if (!gid_valid(low) || !gid_valid(high))
191 			return -EINVAL;
192 		if (urange[1] < urange[0] || gid_lt(high, low)) {
193 			low = make_kgid(&init_user_ns, 1);
194 			high = make_kgid(&init_user_ns, 0);
195 		}
196 		set_ping_group_range(table, low, high);
197 	}
198 
199 	return ret;
200 }
201 
202 static int ipv4_fwd_update_priority(const struct ctl_table *table, int write,
203 				    void *buffer, size_t *lenp, loff_t *ppos)
204 {
205 	struct net *net;
206 	int ret;
207 
208 	net = container_of(table->data, struct net,
209 			   ipv4.sysctl_ip_fwd_update_priority);
210 	ret = proc_dou8vec_minmax(table, write, buffer, lenp, ppos);
211 	if (write && ret == 0)
212 		call_netevent_notifiers(NETEVENT_IPV4_FWD_UPDATE_PRIORITY_UPDATE,
213 					net);
214 
215 	return ret;
216 }
217 
218 static int proc_tcp_congestion_control(const struct ctl_table *ctl, int write,
219 				       void *buffer, size_t *lenp, loff_t *ppos)
220 {
221 	struct net *net = container_of(ctl->data, struct net,
222 				       ipv4.tcp_congestion_control);
223 	char val[TCP_CA_NAME_MAX];
224 	struct ctl_table tbl = {
225 		.data = val,
226 		.maxlen = TCP_CA_NAME_MAX,
227 	};
228 	int ret;
229 
230 	tcp_get_default_congestion_control(net, val);
231 
232 	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
233 	if (write && ret == 0)
234 		ret = tcp_set_default_congestion_control(net, val);
235 	return ret;
236 }
237 
238 static int proc_tcp_available_congestion_control(const struct ctl_table *ctl,
239 						 int write, void *buffer,
240 						 size_t *lenp, loff_t *ppos)
241 {
242 	struct ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX, };
243 	int ret;
244 
245 	tbl.data = kmalloc(tbl.maxlen, GFP_USER);
246 	if (!tbl.data)
247 		return -ENOMEM;
248 	tcp_get_available_congestion_control(tbl.data, TCP_CA_BUF_MAX);
249 	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
250 	kfree(tbl.data);
251 	return ret;
252 }
253 
254 static int proc_allowed_congestion_control(const struct ctl_table *ctl,
255 					   int write, void *buffer,
256 					   size_t *lenp, loff_t *ppos)
257 {
258 	struct ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX };
259 	int ret;
260 
261 	tbl.data = kmalloc(tbl.maxlen, GFP_USER);
262 	if (!tbl.data)
263 		return -ENOMEM;
264 
265 	tcp_get_allowed_congestion_control(tbl.data, tbl.maxlen);
266 	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
267 	if (write && ret == 0)
268 		ret = tcp_set_allowed_congestion_control(tbl.data);
269 	kfree(tbl.data);
270 	return ret;
271 }
272 
273 static int sscanf_key(char *buf, __le32 *key)
274 {
275 	u32 user_key[4];
276 	int i, ret = 0;
277 
278 	if (sscanf(buf, "%x-%x-%x-%x", user_key, user_key + 1,
279 		   user_key + 2, user_key + 3) != 4) {
280 		ret = -EINVAL;
281 	} else {
282 		for (i = 0; i < ARRAY_SIZE(user_key); i++)
283 			key[i] = cpu_to_le32(user_key[i]);
284 	}
285 	pr_debug("proc TFO key set 0x%x-%x-%x-%x <- 0x%s: %u\n",
286 		 user_key[0], user_key[1], user_key[2], user_key[3], buf, ret);
287 
288 	return ret;
289 }
290 
291 static int proc_tcp_fastopen_key(const struct ctl_table *table, int write,
292 				 void *buffer, size_t *lenp, loff_t *ppos)
293 {
294 	struct net *net = container_of(table->data, struct net,
295 	    ipv4.sysctl_tcp_fastopen);
296 	/* maxlen to print the list of keys in hex (*2), with dashes
297 	 * separating doublewords and a comma in between keys.
298 	 */
299 	struct ctl_table tbl = { .maxlen = ((TCP_FASTOPEN_KEY_LENGTH *
300 					    2 * TCP_FASTOPEN_KEY_MAX) +
301 					    (TCP_FASTOPEN_KEY_MAX * 5)) };
302 	u32 user_key[TCP_FASTOPEN_KEY_BUF_LENGTH / sizeof(u32)];
303 	__le32 key[TCP_FASTOPEN_KEY_BUF_LENGTH / sizeof(__le32)];
304 	char *backup_data;
305 	int ret, i = 0, off = 0, n_keys;
306 
307 	tbl.data = kmalloc(tbl.maxlen, GFP_KERNEL);
308 	if (!tbl.data)
309 		return -ENOMEM;
310 
311 	n_keys = tcp_fastopen_get_cipher(net, NULL, (u64 *)key);
312 	if (!n_keys) {
313 		memset(&key[0], 0, TCP_FASTOPEN_KEY_LENGTH);
314 		n_keys = 1;
315 	}
316 
317 	for (i = 0; i < n_keys * 4; i++)
318 		user_key[i] = le32_to_cpu(key[i]);
319 
320 	for (i = 0; i < n_keys; i++) {
321 		off += snprintf(tbl.data + off, tbl.maxlen - off,
322 				"%08x-%08x-%08x-%08x",
323 				user_key[i * 4],
324 				user_key[i * 4 + 1],
325 				user_key[i * 4 + 2],
326 				user_key[i * 4 + 3]);
327 
328 		if (WARN_ON_ONCE(off >= tbl.maxlen - 1))
329 			break;
330 
331 		if (i + 1 < n_keys)
332 			off += snprintf(tbl.data + off, tbl.maxlen - off, ",");
333 	}
334 
335 	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
336 
337 	if (write && ret == 0) {
338 		backup_data = strchr(tbl.data, ',');
339 		if (backup_data) {
340 			*backup_data = '\0';
341 			backup_data++;
342 		}
343 		if (sscanf_key(tbl.data, key)) {
344 			ret = -EINVAL;
345 			goto bad_key;
346 		}
347 		if (backup_data) {
348 			if (sscanf_key(backup_data, key + 4)) {
349 				ret = -EINVAL;
350 				goto bad_key;
351 			}
352 		}
353 		tcp_fastopen_reset_cipher(net, NULL, key,
354 					  backup_data ? key + 4 : NULL);
355 	}
356 
357 bad_key:
358 	kfree(tbl.data);
359 	return ret;
360 }
361 
362 static int proc_tfo_blackhole_detect_timeout(const struct ctl_table *table,
363 					     int write, void *buffer,
364 					     size_t *lenp, loff_t *ppos)
365 {
366 	struct net *net = container_of(table->data, struct net,
367 	    ipv4.sysctl_tcp_fastopen_blackhole_timeout);
368 	int ret;
369 
370 	ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
371 	if (write && ret == 0)
372 		atomic_set(&net->ipv4.tfo_active_disable_times, 0);
373 
374 	return ret;
375 }
376 
377 static int proc_tcp_available_ulp(const struct ctl_table *ctl,
378 				  int write, void *buffer, size_t *lenp,
379 				  loff_t *ppos)
380 {
381 	struct ctl_table tbl = { .maxlen = TCP_ULP_BUF_MAX, };
382 	int ret;
383 
384 	tbl.data = kmalloc(tbl.maxlen, GFP_USER);
385 	if (!tbl.data)
386 		return -ENOMEM;
387 	tcp_get_available_ulp(tbl.data, TCP_ULP_BUF_MAX);
388 	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
389 	kfree(tbl.data);
390 
391 	return ret;
392 }
393 
394 static int proc_tcp_ehash_entries(const struct ctl_table *table, int write,
395 				  void *buffer, size_t *lenp, loff_t *ppos)
396 {
397 	struct net *net = container_of(table->data, struct net,
398 				       ipv4.sysctl_tcp_child_ehash_entries);
399 	struct inet_hashinfo *hinfo = net->ipv4.tcp_death_row.hashinfo;
400 	int tcp_ehash_entries;
401 	struct ctl_table tbl;
402 
403 	tcp_ehash_entries = hinfo->ehash_mask + 1;
404 
405 	/* A negative number indicates that the child netns
406 	 * shares the global ehash.
407 	 */
408 	if (!net_eq(net, &init_net) && !hinfo->pernet)
409 		tcp_ehash_entries *= -1;
410 
411 	memset(&tbl, 0, sizeof(tbl));
412 	tbl.data = &tcp_ehash_entries;
413 	tbl.maxlen = sizeof(int);
414 
415 	return proc_dointvec(&tbl, write, buffer, lenp, ppos);
416 }
417 
418 static int proc_udp_hash_entries(const struct ctl_table *table, int write,
419 				 void *buffer, size_t *lenp, loff_t *ppos)
420 {
421 	struct net *net = container_of(table->data, struct net,
422 				       ipv4.sysctl_udp_child_hash_entries);
423 	int udp_hash_entries;
424 	struct ctl_table tbl;
425 
426 	udp_hash_entries = net->ipv4.udp_table->mask + 1;
427 
428 	/* A negative number indicates that the child netns
429 	 * shares the global udp_table.
430 	 */
431 	if (!net_eq(net, &init_net) && net->ipv4.udp_table == &udp_table)
432 		udp_hash_entries *= -1;
433 
434 	memset(&tbl, 0, sizeof(tbl));
435 	tbl.data = &udp_hash_entries;
436 	tbl.maxlen = sizeof(int);
437 
438 	return proc_dointvec(&tbl, write, buffer, lenp, ppos);
439 }
440 
441 #ifdef CONFIG_IP_ROUTE_MULTIPATH
442 static int proc_fib_multipath_hash_policy(const struct ctl_table *table, int write,
443 					  void *buffer, size_t *lenp,
444 					  loff_t *ppos)
445 {
446 	struct net *net = container_of(table->data, struct net,
447 	    ipv4.sysctl_fib_multipath_hash_policy);
448 	int ret;
449 
450 	ret = proc_dou8vec_minmax(table, write, buffer, lenp, ppos);
451 	if (write && ret == 0)
452 		call_netevent_notifiers(NETEVENT_IPV4_MPATH_HASH_UPDATE, net);
453 
454 	return ret;
455 }
456 
457 static int proc_fib_multipath_hash_fields(const struct ctl_table *table, int write,
458 					  void *buffer, size_t *lenp,
459 					  loff_t *ppos)
460 {
461 	struct net *net;
462 	int ret;
463 
464 	net = container_of(table->data, struct net,
465 			   ipv4.sysctl_fib_multipath_hash_fields);
466 	ret = proc_douintvec_minmax(table, write, buffer, lenp, ppos);
467 	if (write && ret == 0)
468 		call_netevent_notifiers(NETEVENT_IPV4_MPATH_HASH_UPDATE, net);
469 
470 	return ret;
471 }
472 
473 static u32 proc_fib_multipath_hash_rand_seed __ro_after_init;
474 
475 static void proc_fib_multipath_hash_init_rand_seed(void)
476 {
477 	get_random_bytes(&proc_fib_multipath_hash_rand_seed,
478 			 sizeof(proc_fib_multipath_hash_rand_seed));
479 }
480 
481 static void proc_fib_multipath_hash_set_seed(struct net *net, u32 user_seed)
482 {
483 	struct sysctl_fib_multipath_hash_seed new = {
484 		.user_seed = user_seed,
485 		.mp_seed = (user_seed ? user_seed :
486 			    proc_fib_multipath_hash_rand_seed),
487 	};
488 
489 	WRITE_ONCE(net->ipv4.sysctl_fib_multipath_hash_seed, new);
490 }
491 
492 static int proc_fib_multipath_hash_seed(const struct ctl_table *table, int write,
493 					void *buffer, size_t *lenp,
494 					loff_t *ppos)
495 {
496 	struct sysctl_fib_multipath_hash_seed *mphs;
497 	struct net *net = table->data;
498 	struct ctl_table tmp;
499 	u32 user_seed;
500 	int ret;
501 
502 	mphs = &net->ipv4.sysctl_fib_multipath_hash_seed;
503 	user_seed = mphs->user_seed;
504 
505 	tmp = *table;
506 	tmp.data = &user_seed;
507 
508 	ret = proc_douintvec_minmax(&tmp, write, buffer, lenp, ppos);
509 
510 	if (write && ret == 0) {
511 		proc_fib_multipath_hash_set_seed(net, user_seed);
512 		call_netevent_notifiers(NETEVENT_IPV4_MPATH_HASH_UPDATE, net);
513 	}
514 
515 	return ret;
516 }
517 #else
518 
519 static void proc_fib_multipath_hash_init_rand_seed(void)
520 {
521 }
522 
523 static void proc_fib_multipath_hash_set_seed(struct net *net, u32 user_seed)
524 {
525 }
526 
527 #endif
528 
529 static struct ctl_table ipv4_table[] = {
530 	{
531 		.procname	= "tcp_max_orphans",
532 		.data		= &sysctl_tcp_max_orphans,
533 		.maxlen		= sizeof(int),
534 		.mode		= 0644,
535 		.proc_handler	= proc_dointvec
536 	},
537 	{
538 		.procname	= "inet_peer_threshold",
539 		.data		= &inet_peer_threshold,
540 		.maxlen		= sizeof(int),
541 		.mode		= 0644,
542 		.proc_handler	= proc_dointvec
543 	},
544 	{
545 		.procname	= "inet_peer_minttl",
546 		.data		= &inet_peer_minttl,
547 		.maxlen		= sizeof(int),
548 		.mode		= 0644,
549 		.proc_handler	= proc_dointvec_jiffies,
550 	},
551 	{
552 		.procname	= "inet_peer_maxttl",
553 		.data		= &inet_peer_maxttl,
554 		.maxlen		= sizeof(int),
555 		.mode		= 0644,
556 		.proc_handler	= proc_dointvec_jiffies,
557 	},
558 	{
559 		.procname	= "tcp_mem",
560 		.maxlen		= sizeof(sysctl_tcp_mem),
561 		.data		= &sysctl_tcp_mem,
562 		.mode		= 0644,
563 		.proc_handler	= proc_doulongvec_minmax,
564 	},
565 	{
566 		.procname	= "tcp_low_latency",
567 		.data		= &sysctl_tcp_low_latency,
568 		.maxlen		= sizeof(int),
569 		.mode		= 0644,
570 		.proc_handler	= proc_dointvec
571 	},
572 #ifdef CONFIG_NETLABEL
573 	{
574 		.procname	= "cipso_cache_enable",
575 		.data		= &cipso_v4_cache_enabled,
576 		.maxlen		= sizeof(int),
577 		.mode		= 0644,
578 		.proc_handler	= proc_dointvec,
579 	},
580 	{
581 		.procname	= "cipso_cache_bucket_size",
582 		.data		= &cipso_v4_cache_bucketsize,
583 		.maxlen		= sizeof(int),
584 		.mode		= 0644,
585 		.proc_handler	= proc_dointvec,
586 	},
587 	{
588 		.procname	= "cipso_rbm_optfmt",
589 		.data		= &cipso_v4_rbm_optfmt,
590 		.maxlen		= sizeof(int),
591 		.mode		= 0644,
592 		.proc_handler	= proc_dointvec,
593 	},
594 	{
595 		.procname	= "cipso_rbm_strictvalid",
596 		.data		= &cipso_v4_rbm_strictvalid,
597 		.maxlen		= sizeof(int),
598 		.mode		= 0644,
599 		.proc_handler	= proc_dointvec,
600 	},
601 #endif /* CONFIG_NETLABEL */
602 	{
603 		.procname	= "tcp_available_ulp",
604 		.maxlen		= TCP_ULP_BUF_MAX,
605 		.mode		= 0444,
606 		.proc_handler   = proc_tcp_available_ulp,
607 	},
608 	{
609 		.procname	= "udp_mem",
610 		.data		= &sysctl_udp_mem,
611 		.maxlen		= sizeof(sysctl_udp_mem),
612 		.mode		= 0644,
613 		.proc_handler	= proc_doulongvec_minmax,
614 	},
615 	{
616 		.procname	= "fib_sync_mem",
617 		.data		= &sysctl_fib_sync_mem,
618 		.maxlen		= sizeof(sysctl_fib_sync_mem),
619 		.mode		= 0644,
620 		.proc_handler	= proc_douintvec_minmax,
621 		.extra1		= &sysctl_fib_sync_mem_min,
622 		.extra2		= &sysctl_fib_sync_mem_max,
623 	},
624 };
625 
626 static struct ctl_table ipv4_net_table[] = {
627 	{
628 		.procname	= "tcp_max_tw_buckets",
629 		.data		= &init_net.ipv4.tcp_death_row.sysctl_max_tw_buckets,
630 		.maxlen		= sizeof(int),
631 		.mode		= 0644,
632 		.proc_handler	= proc_dointvec
633 	},
634 	{
635 		.procname	= "icmp_echo_ignore_all",
636 		.data		= &init_net.ipv4.sysctl_icmp_echo_ignore_all,
637 		.maxlen		= sizeof(u8),
638 		.mode		= 0644,
639 		.proc_handler	= proc_dou8vec_minmax,
640 		.extra1		= SYSCTL_ZERO,
641 		.extra2		= SYSCTL_ONE
642 	},
643 	{
644 		.procname	= "icmp_echo_enable_probe",
645 		.data		= &init_net.ipv4.sysctl_icmp_echo_enable_probe,
646 		.maxlen		= sizeof(u8),
647 		.mode		= 0644,
648 		.proc_handler	= proc_dou8vec_minmax,
649 		.extra1		= SYSCTL_ZERO,
650 		.extra2		= SYSCTL_ONE
651 	},
652 	{
653 		.procname	= "icmp_echo_ignore_broadcasts",
654 		.data		= &init_net.ipv4.sysctl_icmp_echo_ignore_broadcasts,
655 		.maxlen		= sizeof(u8),
656 		.mode		= 0644,
657 		.proc_handler	= proc_dou8vec_minmax,
658 		.extra1		= SYSCTL_ZERO,
659 		.extra2		= SYSCTL_ONE
660 	},
661 	{
662 		.procname	= "icmp_ignore_bogus_error_responses",
663 		.data		= &init_net.ipv4.sysctl_icmp_ignore_bogus_error_responses,
664 		.maxlen		= sizeof(u8),
665 		.mode		= 0644,
666 		.proc_handler	= proc_dou8vec_minmax,
667 		.extra1		= SYSCTL_ZERO,
668 		.extra2		= SYSCTL_ONE
669 	},
670 	{
671 		.procname	= "icmp_errors_use_inbound_ifaddr",
672 		.data		= &init_net.ipv4.sysctl_icmp_errors_use_inbound_ifaddr,
673 		.maxlen		= sizeof(u8),
674 		.mode		= 0644,
675 		.proc_handler	= proc_dou8vec_minmax,
676 		.extra1		= SYSCTL_ZERO,
677 		.extra2		= SYSCTL_ONE
678 	},
679 	{
680 		.procname	= "icmp_errors_extension_mask",
681 		.data		= &init_net.ipv4.sysctl_icmp_errors_extension_mask,
682 		.maxlen		= sizeof(u8),
683 		.mode		= 0644,
684 		.proc_handler	= proc_dou8vec_minmax,
685 		.extra1		= SYSCTL_ZERO,
686 		.extra2		= &icmp_errors_extension_mask_all,
687 	},
688 	{
689 		.procname	= "icmp_ratelimit",
690 		.data		= &init_net.ipv4.sysctl_icmp_ratelimit,
691 		.maxlen		= sizeof(int),
692 		.mode		= 0644,
693 		.proc_handler	= proc_dointvec_ms_jiffies,
694 	},
695 	{
696 		.procname	= "icmp_ratemask",
697 		.data		= &init_net.ipv4.sysctl_icmp_ratemask,
698 		.maxlen		= sizeof(int),
699 		.mode		= 0644,
700 		.proc_handler	= proc_dointvec
701 	},
702 	{
703 		.procname	= "icmp_msgs_per_sec",
704 		.data		= &init_net.ipv4.sysctl_icmp_msgs_per_sec,
705 		.maxlen		= sizeof(int),
706 		.mode		= 0644,
707 		.proc_handler	= proc_dointvec_minmax,
708 		.extra1		= SYSCTL_ZERO,
709 	},
710 	{
711 		.procname	= "icmp_msgs_burst",
712 		.data		= &init_net.ipv4.sysctl_icmp_msgs_burst,
713 		.maxlen		= sizeof(int),
714 		.mode		= 0644,
715 		.proc_handler	= proc_dointvec_minmax,
716 		.extra1		= SYSCTL_ZERO,
717 	},
718 	{
719 		.procname	= "ping_group_range",
720 		.data		= &init_net.ipv4.ping_group_range.range,
721 		.maxlen		= sizeof(gid_t)*2,
722 		.mode		= 0644,
723 		.proc_handler	= ipv4_ping_group_range,
724 	},
725 #ifdef CONFIG_NET_L3_MASTER_DEV
726 	{
727 		.procname	= "raw_l3mdev_accept",
728 		.data		= &init_net.ipv4.sysctl_raw_l3mdev_accept,
729 		.maxlen		= sizeof(u8),
730 		.mode		= 0644,
731 		.proc_handler	= proc_dou8vec_minmax,
732 		.extra1		= SYSCTL_ZERO,
733 		.extra2		= SYSCTL_ONE,
734 	},
735 #endif
736 	{
737 		.procname	= "tcp_ecn",
738 		.data		= &init_net.ipv4.sysctl_tcp_ecn,
739 		.maxlen		= sizeof(u8),
740 		.mode		= 0644,
741 		.proc_handler	= proc_dou8vec_minmax,
742 		.extra1		= SYSCTL_ZERO,
743 		.extra2		= &tcp_ecn_mode_max,
744 	},
745 	{
746 		.procname	= "tcp_ecn_option",
747 		.data		= &init_net.ipv4.sysctl_tcp_ecn_option,
748 		.maxlen		= sizeof(u8),
749 		.mode		= 0644,
750 		.proc_handler	= proc_dou8vec_minmax,
751 		.extra1		= SYSCTL_ZERO,
752 		.extra2		= SYSCTL_TWO,
753 	},
754 	{
755 		.procname	= "tcp_ecn_option_beacon",
756 		.data		= &init_net.ipv4.sysctl_tcp_ecn_option_beacon,
757 		.maxlen		= sizeof(u8),
758 		.mode		= 0644,
759 		.proc_handler	= proc_dou8vec_minmax,
760 		.extra1		= SYSCTL_ZERO,
761 		.extra2		= SYSCTL_THREE,
762 	},
763 	{
764 		.procname	= "tcp_ecn_fallback",
765 		.data		= &init_net.ipv4.sysctl_tcp_ecn_fallback,
766 		.maxlen		= sizeof(u8),
767 		.mode		= 0644,
768 		.proc_handler	= proc_dou8vec_minmax,
769 		.extra1		= SYSCTL_ZERO,
770 		.extra2		= SYSCTL_ONE,
771 	},
772 	{
773 		.procname	= "ip_dynaddr",
774 		.data		= &init_net.ipv4.sysctl_ip_dynaddr,
775 		.maxlen		= sizeof(u8),
776 		.mode		= 0644,
777 		.proc_handler	= proc_dou8vec_minmax,
778 	},
779 	{
780 		.procname	= "ip_early_demux",
781 		.data		= &init_net.ipv4.sysctl_ip_early_demux,
782 		.maxlen		= sizeof(u8),
783 		.mode		= 0644,
784 		.proc_handler	= proc_dou8vec_minmax,
785 	},
786 	{
787 		.procname       = "udp_early_demux",
788 		.data           = &init_net.ipv4.sysctl_udp_early_demux,
789 		.maxlen         = sizeof(u8),
790 		.mode           = 0644,
791 		.proc_handler   = proc_dou8vec_minmax,
792 	},
793 	{
794 		.procname       = "tcp_early_demux",
795 		.data           = &init_net.ipv4.sysctl_tcp_early_demux,
796 		.maxlen         = sizeof(u8),
797 		.mode           = 0644,
798 		.proc_handler   = proc_dou8vec_minmax,
799 	},
800 	{
801 		.procname       = "nexthop_compat_mode",
802 		.data           = &init_net.ipv4.sysctl_nexthop_compat_mode,
803 		.maxlen         = sizeof(u8),
804 		.mode           = 0644,
805 		.proc_handler   = proc_dou8vec_minmax,
806 		.extra1		= SYSCTL_ZERO,
807 		.extra2		= SYSCTL_ONE,
808 	},
809 	{
810 		.procname	= "ip_default_ttl",
811 		.data		= &init_net.ipv4.sysctl_ip_default_ttl,
812 		.maxlen		= sizeof(u8),
813 		.mode		= 0644,
814 		.proc_handler	= proc_dou8vec_minmax,
815 		.extra1		= &ip_ttl_min,
816 		.extra2		= &ip_ttl_max,
817 	},
818 	{
819 		.procname	= "ip_local_port_range",
820 		.maxlen		= 0,
821 		.data		= &init_net,
822 		.mode		= 0644,
823 		.proc_handler	= ipv4_local_port_range,
824 	},
825 	{
826 		.procname	= "ip_local_reserved_ports",
827 		.data		= &init_net.ipv4.sysctl_local_reserved_ports,
828 		.maxlen		= 65536,
829 		.mode		= 0644,
830 		.proc_handler	= proc_do_large_bitmap,
831 	},
832 	{
833 		.procname	= "ip_no_pmtu_disc",
834 		.data		= &init_net.ipv4.sysctl_ip_no_pmtu_disc,
835 		.maxlen		= sizeof(u8),
836 		.mode		= 0644,
837 		.proc_handler	= proc_dou8vec_minmax,
838 	},
839 	{
840 		.procname	= "ip_forward_use_pmtu",
841 		.data		= &init_net.ipv4.sysctl_ip_fwd_use_pmtu,
842 		.maxlen		= sizeof(u8),
843 		.mode		= 0644,
844 		.proc_handler	= proc_dou8vec_minmax,
845 	},
846 	{
847 		.procname	= "ip_forward_update_priority",
848 		.data		= &init_net.ipv4.sysctl_ip_fwd_update_priority,
849 		.maxlen		= sizeof(u8),
850 		.mode		= 0644,
851 		.proc_handler   = ipv4_fwd_update_priority,
852 		.extra1		= SYSCTL_ZERO,
853 		.extra2		= SYSCTL_ONE,
854 	},
855 	{
856 		.procname	= "ip_nonlocal_bind",
857 		.data		= &init_net.ipv4.sysctl_ip_nonlocal_bind,
858 		.maxlen		= sizeof(u8),
859 		.mode		= 0644,
860 		.proc_handler	= proc_dou8vec_minmax,
861 	},
862 	{
863 		.procname	= "ip_autobind_reuse",
864 		.data		= &init_net.ipv4.sysctl_ip_autobind_reuse,
865 		.maxlen		= sizeof(u8),
866 		.mode		= 0644,
867 		.proc_handler	= proc_dou8vec_minmax,
868 		.extra1         = SYSCTL_ZERO,
869 		.extra2         = SYSCTL_ONE,
870 	},
871 	{
872 		.procname	= "fwmark_reflect",
873 		.data		= &init_net.ipv4.sysctl_fwmark_reflect,
874 		.maxlen		= sizeof(u8),
875 		.mode		= 0644,
876 		.proc_handler	= proc_dou8vec_minmax,
877 	},
878 	{
879 		.procname	= "tcp_fwmark_accept",
880 		.data		= &init_net.ipv4.sysctl_tcp_fwmark_accept,
881 		.maxlen		= sizeof(u8),
882 		.mode		= 0644,
883 		.proc_handler	= proc_dou8vec_minmax,
884 	},
885 #ifdef CONFIG_NET_L3_MASTER_DEV
886 	{
887 		.procname	= "tcp_l3mdev_accept",
888 		.data		= &init_net.ipv4.sysctl_tcp_l3mdev_accept,
889 		.maxlen		= sizeof(u8),
890 		.mode		= 0644,
891 		.proc_handler	= proc_dou8vec_minmax,
892 		.extra1		= SYSCTL_ZERO,
893 		.extra2		= SYSCTL_ONE,
894 	},
895 #endif
896 	{
897 		.procname	= "tcp_mtu_probing",
898 		.data		= &init_net.ipv4.sysctl_tcp_mtu_probing,
899 		.maxlen		= sizeof(u8),
900 		.mode		= 0644,
901 		.proc_handler	= proc_dou8vec_minmax,
902 	},
903 	{
904 		.procname	= "tcp_base_mss",
905 		.data		= &init_net.ipv4.sysctl_tcp_base_mss,
906 		.maxlen		= sizeof(int),
907 		.mode		= 0644,
908 		.proc_handler	= proc_dointvec,
909 	},
910 	{
911 		.procname	= "tcp_min_snd_mss",
912 		.data		= &init_net.ipv4.sysctl_tcp_min_snd_mss,
913 		.maxlen		= sizeof(int),
914 		.mode		= 0644,
915 		.proc_handler	= proc_dointvec_minmax,
916 		.extra1		= &tcp_min_snd_mss_min,
917 		.extra2		= &tcp_min_snd_mss_max,
918 	},
919 	{
920 		.procname	= "tcp_mtu_probe_floor",
921 		.data		= &init_net.ipv4.sysctl_tcp_mtu_probe_floor,
922 		.maxlen		= sizeof(int),
923 		.mode		= 0644,
924 		.proc_handler	= proc_dointvec_minmax,
925 		.extra1		= &tcp_min_snd_mss_min,
926 		.extra2		= &tcp_min_snd_mss_max,
927 	},
928 	{
929 		.procname	= "tcp_probe_threshold",
930 		.data		= &init_net.ipv4.sysctl_tcp_probe_threshold,
931 		.maxlen		= sizeof(int),
932 		.mode		= 0644,
933 		.proc_handler	= proc_dointvec,
934 	},
935 	{
936 		.procname	= "tcp_probe_interval",
937 		.data		= &init_net.ipv4.sysctl_tcp_probe_interval,
938 		.maxlen		= sizeof(u32),
939 		.mode		= 0644,
940 		.proc_handler	= proc_douintvec_minmax,
941 		.extra2		= &u32_max_div_HZ,
942 	},
943 	{
944 		.procname	= "igmp_link_local_mcast_reports",
945 		.data		= &init_net.ipv4.sysctl_igmp_llm_reports,
946 		.maxlen		= sizeof(u8),
947 		.mode		= 0644,
948 		.proc_handler	= proc_dou8vec_minmax,
949 	},
950 	{
951 		.procname	= "igmp_max_memberships",
952 		.data		= &init_net.ipv4.sysctl_igmp_max_memberships,
953 		.maxlen		= sizeof(int),
954 		.mode		= 0644,
955 		.proc_handler	= proc_dointvec
956 	},
957 	{
958 		.procname	= "igmp_max_msf",
959 		.data		= &init_net.ipv4.sysctl_igmp_max_msf,
960 		.maxlen		= sizeof(int),
961 		.mode		= 0644,
962 		.proc_handler	= proc_dointvec
963 	},
964 #ifdef CONFIG_IP_MULTICAST
965 	{
966 		.procname	= "igmp_qrv",
967 		.data		= &init_net.ipv4.sysctl_igmp_qrv,
968 		.maxlen		= sizeof(int),
969 		.mode		= 0644,
970 		.proc_handler	= proc_dointvec_minmax,
971 		.extra1		= SYSCTL_ONE
972 	},
973 #endif
974 	{
975 		.procname	= "tcp_congestion_control",
976 		.data		= &init_net.ipv4.tcp_congestion_control,
977 		.mode		= 0644,
978 		.maxlen		= TCP_CA_NAME_MAX,
979 		.proc_handler	= proc_tcp_congestion_control,
980 	},
981 	{
982 		.procname	= "tcp_available_congestion_control",
983 		.maxlen		= TCP_CA_BUF_MAX,
984 		.mode		= 0444,
985 		.proc_handler   = proc_tcp_available_congestion_control,
986 	},
987 	{
988 		.procname	= "tcp_allowed_congestion_control",
989 		.maxlen		= TCP_CA_BUF_MAX,
990 		.mode		= 0644,
991 		.proc_handler   = proc_allowed_congestion_control,
992 	},
993 	{
994 		.procname	= "tcp_keepalive_time",
995 		.data		= &init_net.ipv4.sysctl_tcp_keepalive_time,
996 		.maxlen		= sizeof(int),
997 		.mode		= 0644,
998 		.proc_handler	= proc_dointvec_jiffies,
999 	},
1000 	{
1001 		.procname	= "tcp_keepalive_probes",
1002 		.data		= &init_net.ipv4.sysctl_tcp_keepalive_probes,
1003 		.maxlen		= sizeof(u8),
1004 		.mode		= 0644,
1005 		.proc_handler	= proc_dou8vec_minmax,
1006 	},
1007 	{
1008 		.procname	= "tcp_keepalive_intvl",
1009 		.data		= &init_net.ipv4.sysctl_tcp_keepalive_intvl,
1010 		.maxlen		= sizeof(int),
1011 		.mode		= 0644,
1012 		.proc_handler	= proc_dointvec_jiffies,
1013 	},
1014 	{
1015 		.procname	= "tcp_syn_retries",
1016 		.data		= &init_net.ipv4.sysctl_tcp_syn_retries,
1017 		.maxlen		= sizeof(u8),
1018 		.mode		= 0644,
1019 		.proc_handler	= proc_dou8vec_minmax,
1020 		.extra1		= &tcp_syn_retries_min,
1021 		.extra2		= &tcp_syn_retries_max
1022 	},
1023 	{
1024 		.procname	= "tcp_synack_retries",
1025 		.data		= &init_net.ipv4.sysctl_tcp_synack_retries,
1026 		.maxlen		= sizeof(u8),
1027 		.mode		= 0644,
1028 		.proc_handler	= proc_dou8vec_minmax,
1029 	},
1030 #ifdef CONFIG_SYN_COOKIES
1031 	{
1032 		.procname	= "tcp_syncookies",
1033 		.data		= &init_net.ipv4.sysctl_tcp_syncookies,
1034 		.maxlen		= sizeof(u8),
1035 		.mode		= 0644,
1036 		.proc_handler	= proc_dou8vec_minmax,
1037 	},
1038 #endif
1039 	{
1040 		.procname	= "tcp_migrate_req",
1041 		.data		= &init_net.ipv4.sysctl_tcp_migrate_req,
1042 		.maxlen		= sizeof(u8),
1043 		.mode		= 0644,
1044 		.proc_handler	= proc_dou8vec_minmax,
1045 		.extra1		= SYSCTL_ZERO,
1046 		.extra2		= SYSCTL_ONE
1047 	},
1048 	{
1049 		.procname	= "tcp_reordering",
1050 		.data		= &init_net.ipv4.sysctl_tcp_reordering,
1051 		.maxlen		= sizeof(int),
1052 		.mode		= 0644,
1053 		.proc_handler	= proc_dointvec
1054 	},
1055 	{
1056 		.procname	= "tcp_retries1",
1057 		.data		= &init_net.ipv4.sysctl_tcp_retries1,
1058 		.maxlen		= sizeof(u8),
1059 		.mode		= 0644,
1060 		.proc_handler	= proc_dou8vec_minmax,
1061 		.extra2		= &tcp_retr1_max
1062 	},
1063 	{
1064 		.procname	= "tcp_retries2",
1065 		.data		= &init_net.ipv4.sysctl_tcp_retries2,
1066 		.maxlen		= sizeof(u8),
1067 		.mode		= 0644,
1068 		.proc_handler	= proc_dou8vec_minmax,
1069 	},
1070 	{
1071 		.procname	= "tcp_orphan_retries",
1072 		.data		= &init_net.ipv4.sysctl_tcp_orphan_retries,
1073 		.maxlen		= sizeof(u8),
1074 		.mode		= 0644,
1075 		.proc_handler	= proc_dou8vec_minmax,
1076 	},
1077 	{
1078 		.procname	= "tcp_fin_timeout",
1079 		.data		= &init_net.ipv4.sysctl_tcp_fin_timeout,
1080 		.maxlen		= sizeof(int),
1081 		.mode		= 0644,
1082 		.proc_handler	= proc_dointvec_jiffies,
1083 	},
1084 	{
1085 		.procname	= "tcp_notsent_lowat",
1086 		.data		= &init_net.ipv4.sysctl_tcp_notsent_lowat,
1087 		.maxlen		= sizeof(unsigned int),
1088 		.mode		= 0644,
1089 		.proc_handler	= proc_douintvec,
1090 	},
1091 	{
1092 		.procname	= "tcp_tw_reuse",
1093 		.data		= &init_net.ipv4.sysctl_tcp_tw_reuse,
1094 		.maxlen		= sizeof(u8),
1095 		.mode		= 0644,
1096 		.proc_handler	= proc_dou8vec_minmax,
1097 		.extra1		= SYSCTL_ZERO,
1098 		.extra2		= SYSCTL_TWO,
1099 	},
1100 	{
1101 		.procname	= "tcp_tw_reuse_delay",
1102 		.data		= &init_net.ipv4.sysctl_tcp_tw_reuse_delay,
1103 		.maxlen		= sizeof(unsigned int),
1104 		.mode		= 0644,
1105 		.proc_handler	= proc_douintvec_minmax,
1106 		.extra1		= SYSCTL_ONE,
1107 		.extra2		= &tcp_tw_reuse_delay_max,
1108 	},
1109 	{
1110 		.procname	= "tcp_max_syn_backlog",
1111 		.data		= &init_net.ipv4.sysctl_max_syn_backlog,
1112 		.maxlen		= sizeof(int),
1113 		.mode		= 0644,
1114 		.proc_handler	= proc_dointvec
1115 	},
1116 	{
1117 		.procname	= "tcp_fastopen",
1118 		.data		= &init_net.ipv4.sysctl_tcp_fastopen,
1119 		.maxlen		= sizeof(int),
1120 		.mode		= 0644,
1121 		.proc_handler	= proc_dointvec,
1122 	},
1123 	{
1124 		.procname	= "tcp_fastopen_key",
1125 		.mode		= 0600,
1126 		.data		= &init_net.ipv4.sysctl_tcp_fastopen,
1127 		/* maxlen to print the list of keys in hex (*2), with dashes
1128 		 * separating doublewords and a comma in between keys.
1129 		 */
1130 		.maxlen		= ((TCP_FASTOPEN_KEY_LENGTH *
1131 				   2 * TCP_FASTOPEN_KEY_MAX) +
1132 				   (TCP_FASTOPEN_KEY_MAX * 5)),
1133 		.proc_handler	= proc_tcp_fastopen_key,
1134 	},
1135 	{
1136 		.procname	= "tcp_fastopen_blackhole_timeout_sec",
1137 		.data		= &init_net.ipv4.sysctl_tcp_fastopen_blackhole_timeout,
1138 		.maxlen		= sizeof(int),
1139 		.mode		= 0644,
1140 		.proc_handler	= proc_tfo_blackhole_detect_timeout,
1141 		.extra1		= SYSCTL_ZERO,
1142 	},
1143 #ifdef CONFIG_IP_ROUTE_MULTIPATH
1144 	{
1145 		.procname	= "fib_multipath_use_neigh",
1146 		.data		= &init_net.ipv4.sysctl_fib_multipath_use_neigh,
1147 		.maxlen		= sizeof(u8),
1148 		.mode		= 0644,
1149 		.proc_handler	= proc_dou8vec_minmax,
1150 		.extra1		= SYSCTL_ZERO,
1151 		.extra2		= SYSCTL_ONE,
1152 	},
1153 	{
1154 		.procname	= "fib_multipath_hash_policy",
1155 		.data		= &init_net.ipv4.sysctl_fib_multipath_hash_policy,
1156 		.maxlen		= sizeof(u8),
1157 		.mode		= 0644,
1158 		.proc_handler	= proc_fib_multipath_hash_policy,
1159 		.extra1		= SYSCTL_ZERO,
1160 		.extra2		= SYSCTL_THREE,
1161 	},
1162 	{
1163 		.procname	= "fib_multipath_hash_fields",
1164 		.data		= &init_net.ipv4.sysctl_fib_multipath_hash_fields,
1165 		.maxlen		= sizeof(u32),
1166 		.mode		= 0644,
1167 		.proc_handler	= proc_fib_multipath_hash_fields,
1168 		.extra1		= SYSCTL_ONE,
1169 		.extra2		= &fib_multipath_hash_fields_all_mask,
1170 	},
1171 	{
1172 		.procname	= "fib_multipath_hash_seed",
1173 		.data		= &init_net,
1174 		.maxlen		= sizeof(u32),
1175 		.mode		= 0644,
1176 		.proc_handler	= proc_fib_multipath_hash_seed,
1177 	},
1178 #endif
1179 	{
1180 		.procname	= "ip_unprivileged_port_start",
1181 		.maxlen		= sizeof(int),
1182 		.data		= &init_net.ipv4.sysctl_ip_prot_sock,
1183 		.mode		= 0644,
1184 		.proc_handler	= ipv4_privileged_ports,
1185 	},
1186 #ifdef CONFIG_NET_L3_MASTER_DEV
1187 	{
1188 		.procname	= "udp_l3mdev_accept",
1189 		.data		= &init_net.ipv4.sysctl_udp_l3mdev_accept,
1190 		.maxlen		= sizeof(u8),
1191 		.mode		= 0644,
1192 		.proc_handler	= proc_dou8vec_minmax,
1193 		.extra1		= SYSCTL_ZERO,
1194 		.extra2		= SYSCTL_ONE,
1195 	},
1196 #endif
1197 	{
1198 		.procname	= "tcp_sack",
1199 		.data		= &init_net.ipv4.sysctl_tcp_sack,
1200 		.maxlen		= sizeof(u8),
1201 		.mode		= 0644,
1202 		.proc_handler	= proc_dou8vec_minmax,
1203 	},
1204 	{
1205 		.procname	= "tcp_window_scaling",
1206 		.data		= &init_net.ipv4.sysctl_tcp_window_scaling,
1207 		.maxlen		= sizeof(u8),
1208 		.mode		= 0644,
1209 		.proc_handler	= proc_dou8vec_minmax,
1210 	},
1211 	{
1212 		.procname	= "tcp_timestamps",
1213 		.data		= &init_net.ipv4.sysctl_tcp_timestamps,
1214 		.maxlen		= sizeof(u8),
1215 		.mode		= 0644,
1216 		.proc_handler	= proc_dou8vec_minmax,
1217 	},
1218 	{
1219 		.procname	= "tcp_early_retrans",
1220 		.data		= &init_net.ipv4.sysctl_tcp_early_retrans,
1221 		.maxlen		= sizeof(u8),
1222 		.mode		= 0644,
1223 		.proc_handler	= proc_dou8vec_minmax,
1224 		.extra1		= SYSCTL_ZERO,
1225 		.extra2		= SYSCTL_FOUR,
1226 	},
1227 	{
1228 		.procname	= "tcp_recovery",
1229 		.data		= &init_net.ipv4.sysctl_tcp_recovery,
1230 		.maxlen		= sizeof(u8),
1231 		.mode		= 0644,
1232 		.proc_handler	= proc_dou8vec_minmax,
1233 	},
1234 	{
1235 		.procname       = "tcp_thin_linear_timeouts",
1236 		.data           = &init_net.ipv4.sysctl_tcp_thin_linear_timeouts,
1237 		.maxlen         = sizeof(u8),
1238 		.mode           = 0644,
1239 		.proc_handler   = proc_dou8vec_minmax,
1240 	},
1241 	{
1242 		.procname	= "tcp_slow_start_after_idle",
1243 		.data		= &init_net.ipv4.sysctl_tcp_slow_start_after_idle,
1244 		.maxlen		= sizeof(u8),
1245 		.mode		= 0644,
1246 		.proc_handler	= proc_dou8vec_minmax,
1247 	},
1248 	{
1249 		.procname	= "tcp_retrans_collapse",
1250 		.data		= &init_net.ipv4.sysctl_tcp_retrans_collapse,
1251 		.maxlen		= sizeof(u8),
1252 		.mode		= 0644,
1253 		.proc_handler	= proc_dou8vec_minmax,
1254 	},
1255 	{
1256 		.procname	= "tcp_stdurg",
1257 		.data		= &init_net.ipv4.sysctl_tcp_stdurg,
1258 		.maxlen		= sizeof(u8),
1259 		.mode		= 0644,
1260 		.proc_handler	= proc_dou8vec_minmax,
1261 	},
1262 	{
1263 		.procname	= "tcp_rfc1337",
1264 		.data		= &init_net.ipv4.sysctl_tcp_rfc1337,
1265 		.maxlen		= sizeof(u8),
1266 		.mode		= 0644,
1267 		.proc_handler	= proc_dou8vec_minmax,
1268 	},
1269 	{
1270 		.procname	= "tcp_abort_on_overflow",
1271 		.data		= &init_net.ipv4.sysctl_tcp_abort_on_overflow,
1272 		.maxlen		= sizeof(u8),
1273 		.mode		= 0644,
1274 		.proc_handler	= proc_dou8vec_minmax,
1275 	},
1276 	{
1277 		.procname	= "tcp_fack",
1278 		.data		= &init_net.ipv4.sysctl_tcp_fack,
1279 		.maxlen		= sizeof(u8),
1280 		.mode		= 0644,
1281 		.proc_handler	= proc_dou8vec_minmax,
1282 	},
1283 	{
1284 		.procname	= "tcp_max_reordering",
1285 		.data		= &init_net.ipv4.sysctl_tcp_max_reordering,
1286 		.maxlen		= sizeof(int),
1287 		.mode		= 0644,
1288 		.proc_handler	= proc_dointvec
1289 	},
1290 	{
1291 		.procname	= "tcp_dsack",
1292 		.data		= &init_net.ipv4.sysctl_tcp_dsack,
1293 		.maxlen		= sizeof(u8),
1294 		.mode		= 0644,
1295 		.proc_handler	= proc_dou8vec_minmax,
1296 	},
1297 	{
1298 		.procname	= "tcp_app_win",
1299 		.data		= &init_net.ipv4.sysctl_tcp_app_win,
1300 		.maxlen		= sizeof(u8),
1301 		.mode		= 0644,
1302 		.proc_handler	= proc_dou8vec_minmax,
1303 		.extra1		= SYSCTL_ZERO,
1304 		.extra2		= &tcp_app_win_max,
1305 	},
1306 	{
1307 		.procname	= "tcp_adv_win_scale",
1308 		.data		= &init_net.ipv4.sysctl_tcp_adv_win_scale,
1309 		.maxlen		= sizeof(int),
1310 		.mode		= 0644,
1311 		.proc_handler	= proc_dointvec_minmax,
1312 		.extra1		= &tcp_adv_win_scale_min,
1313 		.extra2		= &tcp_adv_win_scale_max,
1314 	},
1315 	{
1316 		.procname	= "tcp_frto",
1317 		.data		= &init_net.ipv4.sysctl_tcp_frto,
1318 		.maxlen		= sizeof(u8),
1319 		.mode		= 0644,
1320 		.proc_handler	= proc_dou8vec_minmax,
1321 	},
1322 	{
1323 		.procname	= "tcp_no_metrics_save",
1324 		.data		= &init_net.ipv4.sysctl_tcp_nometrics_save,
1325 		.maxlen		= sizeof(u8),
1326 		.mode		= 0644,
1327 		.proc_handler	= proc_dou8vec_minmax,
1328 	},
1329 	{
1330 		.procname	= "tcp_no_ssthresh_metrics_save",
1331 		.data		= &init_net.ipv4.sysctl_tcp_no_ssthresh_metrics_save,
1332 		.maxlen		= sizeof(u8),
1333 		.mode		= 0644,
1334 		.proc_handler	= proc_dou8vec_minmax,
1335 		.extra1		= SYSCTL_ZERO,
1336 		.extra2		= SYSCTL_ONE,
1337 	},
1338 	{
1339 		.procname	= "tcp_moderate_rcvbuf",
1340 		.data		= &init_net.ipv4.sysctl_tcp_moderate_rcvbuf,
1341 		.maxlen		= sizeof(u8),
1342 		.mode		= 0644,
1343 		.proc_handler	= proc_dou8vec_minmax,
1344 	},
1345 	{
1346 		.procname	= "tcp_rcvbuf_low_rtt",
1347 		.data		= &init_net.ipv4.sysctl_tcp_rcvbuf_low_rtt,
1348 		.maxlen		= sizeof(int),
1349 		.mode		= 0644,
1350 		.proc_handler	= proc_dointvec_minmax,
1351 		.extra1		= SYSCTL_ZERO,
1352 		.extra2		= SYSCTL_INT_MAX,
1353 	},
1354 	{
1355 		.procname	= "tcp_tso_win_divisor",
1356 		.data		= &init_net.ipv4.sysctl_tcp_tso_win_divisor,
1357 		.maxlen		= sizeof(u8),
1358 		.mode		= 0644,
1359 		.proc_handler	= proc_dou8vec_minmax,
1360 	},
1361 	{
1362 		.procname	= "tcp_workaround_signed_windows",
1363 		.data		= &init_net.ipv4.sysctl_tcp_workaround_signed_windows,
1364 		.maxlen		= sizeof(u8),
1365 		.mode		= 0644,
1366 		.proc_handler	= proc_dou8vec_minmax,
1367 	},
1368 	{
1369 		.procname	= "tcp_limit_output_bytes",
1370 		.data		= &init_net.ipv4.sysctl_tcp_limit_output_bytes,
1371 		.maxlen		= sizeof(int),
1372 		.mode		= 0644,
1373 		.proc_handler	= proc_dointvec
1374 	},
1375 	{
1376 		.procname	= "tcp_challenge_ack_limit",
1377 		.data		= &init_net.ipv4.sysctl_tcp_challenge_ack_limit,
1378 		.maxlen		= sizeof(int),
1379 		.mode		= 0644,
1380 		.proc_handler	= proc_dointvec
1381 	},
1382 	{
1383 		.procname	= "tcp_min_tso_segs",
1384 		.data		= &init_net.ipv4.sysctl_tcp_min_tso_segs,
1385 		.maxlen		= sizeof(u8),
1386 		.mode		= 0644,
1387 		.proc_handler	= proc_dou8vec_minmax,
1388 		.extra1		= SYSCTL_ONE,
1389 	},
1390 	{
1391 		.procname	= "tcp_tso_rtt_log",
1392 		.data		= &init_net.ipv4.sysctl_tcp_tso_rtt_log,
1393 		.maxlen		= sizeof(u8),
1394 		.mode		= 0644,
1395 		.proc_handler	= proc_dou8vec_minmax,
1396 	},
1397 	{
1398 		.procname	= "tcp_min_rtt_wlen",
1399 		.data		= &init_net.ipv4.sysctl_tcp_min_rtt_wlen,
1400 		.maxlen		= sizeof(int),
1401 		.mode		= 0644,
1402 		.proc_handler	= proc_dointvec_minmax,
1403 		.extra1		= SYSCTL_ZERO,
1404 		.extra2		= &one_day_secs
1405 	},
1406 	{
1407 		.procname	= "tcp_autocorking",
1408 		.data		= &init_net.ipv4.sysctl_tcp_autocorking,
1409 		.maxlen		= sizeof(u8),
1410 		.mode		= 0644,
1411 		.proc_handler	= proc_dou8vec_minmax,
1412 		.extra1		= SYSCTL_ZERO,
1413 		.extra2		= SYSCTL_ONE,
1414 	},
1415 	{
1416 		.procname	= "tcp_invalid_ratelimit",
1417 		.data		= &init_net.ipv4.sysctl_tcp_invalid_ratelimit,
1418 		.maxlen		= sizeof(int),
1419 		.mode		= 0644,
1420 		.proc_handler	= proc_dointvec_ms_jiffies,
1421 	},
1422 	{
1423 		.procname	= "tcp_pacing_ss_ratio",
1424 		.data		= &init_net.ipv4.sysctl_tcp_pacing_ss_ratio,
1425 		.maxlen		= sizeof(int),
1426 		.mode		= 0644,
1427 		.proc_handler	= proc_dointvec_minmax,
1428 		.extra1		= SYSCTL_ZERO,
1429 		.extra2		= SYSCTL_ONE_THOUSAND,
1430 	},
1431 	{
1432 		.procname	= "tcp_pacing_ca_ratio",
1433 		.data		= &init_net.ipv4.sysctl_tcp_pacing_ca_ratio,
1434 		.maxlen		= sizeof(int),
1435 		.mode		= 0644,
1436 		.proc_handler	= proc_dointvec_minmax,
1437 		.extra1		= SYSCTL_ZERO,
1438 		.extra2		= SYSCTL_ONE_THOUSAND,
1439 	},
1440 	{
1441 		.procname	= "tcp_wmem",
1442 		.data		= &init_net.ipv4.sysctl_tcp_wmem,
1443 		.maxlen		= sizeof(init_net.ipv4.sysctl_tcp_wmem),
1444 		.mode		= 0644,
1445 		.proc_handler	= proc_dointvec_minmax,
1446 		.extra1		= SYSCTL_ONE,
1447 	},
1448 	{
1449 		.procname	= "tcp_rmem",
1450 		.data		= &init_net.ipv4.sysctl_tcp_rmem,
1451 		.maxlen		= sizeof(init_net.ipv4.sysctl_tcp_rmem),
1452 		.mode		= 0644,
1453 		.proc_handler	= proc_dointvec_minmax,
1454 		.extra1		= SYSCTL_ONE,
1455 	},
1456 	{
1457 		.procname	= "tcp_comp_sack_delay_ns",
1458 		.data		= &init_net.ipv4.sysctl_tcp_comp_sack_delay_ns,
1459 		.maxlen		= sizeof(unsigned long),
1460 		.mode		= 0644,
1461 		.proc_handler	= proc_doulongvec_minmax,
1462 	},
1463 	{
1464 		.procname	= "tcp_comp_sack_rtt_percent",
1465 		.data		= &init_net.ipv4.sysctl_tcp_comp_sack_rtt_percent,
1466 		.maxlen		= sizeof(int),
1467 		.mode		= 0644,
1468 		.proc_handler	= proc_dointvec_minmax,
1469 		.extra1		= SYSCTL_ONE,
1470 		.extra2		= SYSCTL_ONE_THOUSAND,
1471 	},
1472 	{
1473 		.procname	= "tcp_comp_sack_slack_ns",
1474 		.data		= &init_net.ipv4.sysctl_tcp_comp_sack_slack_ns,
1475 		.maxlen		= sizeof(unsigned long),
1476 		.mode		= 0644,
1477 		.proc_handler	= proc_doulongvec_minmax,
1478 	},
1479 	{
1480 		.procname	= "tcp_comp_sack_nr",
1481 		.data		= &init_net.ipv4.sysctl_tcp_comp_sack_nr,
1482 		.maxlen		= sizeof(u8),
1483 		.mode		= 0644,
1484 		.proc_handler	= proc_dou8vec_minmax,
1485 		.extra1		= SYSCTL_ZERO,
1486 	},
1487 	{
1488 		.procname	= "tcp_backlog_ack_defer",
1489 		.data		= &init_net.ipv4.sysctl_tcp_backlog_ack_defer,
1490 		.maxlen		= sizeof(u8),
1491 		.mode		= 0644,
1492 		.proc_handler	= proc_dou8vec_minmax,
1493 		.extra1		= SYSCTL_ZERO,
1494 		.extra2		= SYSCTL_ONE,
1495 	},
1496 	{
1497 		.procname       = "tcp_reflect_tos",
1498 		.data           = &init_net.ipv4.sysctl_tcp_reflect_tos,
1499 		.maxlen         = sizeof(u8),
1500 		.mode           = 0644,
1501 		.proc_handler   = proc_dou8vec_minmax,
1502 		.extra1         = SYSCTL_ZERO,
1503 		.extra2         = SYSCTL_ONE,
1504 	},
1505 	{
1506 		.procname	= "tcp_ehash_entries",
1507 		.data		= &init_net.ipv4.sysctl_tcp_child_ehash_entries,
1508 		.mode		= 0444,
1509 		.proc_handler	= proc_tcp_ehash_entries,
1510 	},
1511 	{
1512 		.procname	= "tcp_child_ehash_entries",
1513 		.data		= &init_net.ipv4.sysctl_tcp_child_ehash_entries,
1514 		.maxlen		= sizeof(unsigned int),
1515 		.mode		= 0644,
1516 		.proc_handler	= proc_douintvec_minmax,
1517 		.extra1		= SYSCTL_ZERO,
1518 		.extra2		= &tcp_child_ehash_entries_max,
1519 	},
1520 	{
1521 		.procname	= "udp_hash_entries",
1522 		.data		= &init_net.ipv4.sysctl_udp_child_hash_entries,
1523 		.mode		= 0444,
1524 		.proc_handler	= proc_udp_hash_entries,
1525 	},
1526 	{
1527 		.procname	= "udp_child_hash_entries",
1528 		.data		= &init_net.ipv4.sysctl_udp_child_hash_entries,
1529 		.maxlen		= sizeof(unsigned int),
1530 		.mode		= 0644,
1531 		.proc_handler	= proc_douintvec_minmax,
1532 		.extra1		= SYSCTL_ZERO,
1533 		.extra2		= &udp_child_hash_entries_max,
1534 	},
1535 	{
1536 		.procname	= "udp_rmem_min",
1537 		.data		= &init_net.ipv4.sysctl_udp_rmem_min,
1538 		.maxlen		= sizeof(init_net.ipv4.sysctl_udp_rmem_min),
1539 		.mode		= 0644,
1540 		.proc_handler	= proc_dointvec_minmax,
1541 		.extra1		= SYSCTL_ONE
1542 	},
1543 	{
1544 		.procname	= "udp_wmem_min",
1545 		.data		= &init_net.ipv4.sysctl_udp_wmem_min,
1546 		.maxlen		= sizeof(init_net.ipv4.sysctl_udp_wmem_min),
1547 		.mode		= 0644,
1548 		.proc_handler	= proc_dointvec_minmax,
1549 		.extra1		= SYSCTL_ONE
1550 	},
1551 	{
1552 		.procname	= "fib_notify_on_flag_change",
1553 		.data		= &init_net.ipv4.sysctl_fib_notify_on_flag_change,
1554 		.maxlen		= sizeof(u8),
1555 		.mode		= 0644,
1556 		.proc_handler	= proc_dou8vec_minmax,
1557 		.extra1		= SYSCTL_ZERO,
1558 		.extra2		= SYSCTL_TWO,
1559 	},
1560 	{
1561 		.procname       = "tcp_plb_enabled",
1562 		.data           = &init_net.ipv4.sysctl_tcp_plb_enabled,
1563 		.maxlen         = sizeof(u8),
1564 		.mode           = 0644,
1565 		.proc_handler   = proc_dou8vec_minmax,
1566 		.extra1         = SYSCTL_ZERO,
1567 		.extra2         = SYSCTL_ONE,
1568 	},
1569 	{
1570 		.procname       = "tcp_plb_idle_rehash_rounds",
1571 		.data           = &init_net.ipv4.sysctl_tcp_plb_idle_rehash_rounds,
1572 		.maxlen         = sizeof(u8),
1573 		.mode           = 0644,
1574 		.proc_handler   = proc_dou8vec_minmax,
1575 		.extra2		= &tcp_plb_max_rounds,
1576 	},
1577 	{
1578 		.procname       = "tcp_plb_rehash_rounds",
1579 		.data           = &init_net.ipv4.sysctl_tcp_plb_rehash_rounds,
1580 		.maxlen         = sizeof(u8),
1581 		.mode           = 0644,
1582 		.proc_handler   = proc_dou8vec_minmax,
1583 		.extra2         = &tcp_plb_max_rounds,
1584 	},
1585 	{
1586 		.procname       = "tcp_plb_suspend_rto_sec",
1587 		.data           = &init_net.ipv4.sysctl_tcp_plb_suspend_rto_sec,
1588 		.maxlen         = sizeof(u8),
1589 		.mode           = 0644,
1590 		.proc_handler   = proc_dou8vec_minmax,
1591 	},
1592 	{
1593 		.procname       = "tcp_plb_cong_thresh",
1594 		.data           = &init_net.ipv4.sysctl_tcp_plb_cong_thresh,
1595 		.maxlen         = sizeof(int),
1596 		.mode           = 0644,
1597 		.proc_handler   = proc_dointvec_minmax,
1598 		.extra1         = SYSCTL_ZERO,
1599 		.extra2         = &tcp_plb_max_cong_thresh,
1600 	},
1601 	{
1602 		.procname	= "tcp_syn_linear_timeouts",
1603 		.data		= &init_net.ipv4.sysctl_tcp_syn_linear_timeouts,
1604 		.maxlen		= sizeof(u8),
1605 		.mode		= 0644,
1606 		.proc_handler	= proc_dou8vec_minmax,
1607 		.extra1		= SYSCTL_ZERO,
1608 		.extra2		= &tcp_syn_linear_timeouts_max,
1609 	},
1610 	{
1611 		.procname	= "tcp_shrink_window",
1612 		.data		= &init_net.ipv4.sysctl_tcp_shrink_window,
1613 		.maxlen		= sizeof(u8),
1614 		.mode		= 0644,
1615 		.proc_handler	= proc_dou8vec_minmax,
1616 		.extra1		= SYSCTL_ZERO,
1617 		.extra2		= SYSCTL_ONE,
1618 	},
1619 	{
1620 		.procname	= "tcp_pingpong_thresh",
1621 		.data		= &init_net.ipv4.sysctl_tcp_pingpong_thresh,
1622 		.maxlen		= sizeof(u8),
1623 		.mode		= 0644,
1624 		.proc_handler	= proc_dou8vec_minmax,
1625 		.extra1		= SYSCTL_ONE,
1626 	},
1627 	{
1628 		.procname	= "tcp_rto_min_us",
1629 		.data		= &init_net.ipv4.sysctl_tcp_rto_min_us,
1630 		.maxlen		= sizeof(int),
1631 		.mode		= 0644,
1632 		.proc_handler	= proc_dointvec_minmax,
1633 		.extra1		= SYSCTL_ONE,
1634 	},
1635 	{
1636 		.procname	= "tcp_rto_max_ms",
1637 		.data		= &init_net.ipv4.sysctl_tcp_rto_max_ms,
1638 		.maxlen		= sizeof(int),
1639 		.mode		= 0644,
1640 		.proc_handler	= proc_dointvec_minmax,
1641 		.extra1		= SYSCTL_ONE_THOUSAND,
1642 		.extra2		= &tcp_rto_max_max,
1643 	},
1644 };
1645 
1646 static __net_init int ipv4_sysctl_init_net(struct net *net)
1647 {
1648 	size_t table_size = ARRAY_SIZE(ipv4_net_table);
1649 	struct ctl_table *table;
1650 
1651 	table = ipv4_net_table;
1652 	if (!net_eq(net, &init_net)) {
1653 		int i;
1654 
1655 		table = kmemdup(table, sizeof(ipv4_net_table), GFP_KERNEL);
1656 		if (!table)
1657 			goto err_alloc;
1658 
1659 		for (i = 0; i < table_size; i++) {
1660 			if (table[i].data) {
1661 				/* Update the variables to point into
1662 				 * the current struct net
1663 				 */
1664 				table[i].data += (void *)net - (void *)&init_net;
1665 			} else {
1666 				/* Entries without data pointer are global;
1667 				 * Make them read-only in non-init_net ns
1668 				 */
1669 				table[i].mode &= ~0222;
1670 			}
1671 		}
1672 	}
1673 
1674 	net->ipv4.ipv4_hdr = register_net_sysctl_sz(net, "net/ipv4", table,
1675 						    table_size);
1676 	if (!net->ipv4.ipv4_hdr)
1677 		goto err_reg;
1678 
1679 	net->ipv4.sysctl_local_reserved_ports = kzalloc(65536 / 8, GFP_KERNEL);
1680 	if (!net->ipv4.sysctl_local_reserved_ports)
1681 		goto err_ports;
1682 
1683 	proc_fib_multipath_hash_set_seed(net, 0);
1684 
1685 	return 0;
1686 
1687 err_ports:
1688 	unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
1689 err_reg:
1690 	if (!net_eq(net, &init_net))
1691 		kfree(table);
1692 err_alloc:
1693 	return -ENOMEM;
1694 }
1695 
1696 static __net_exit void ipv4_sysctl_exit_net(struct net *net)
1697 {
1698 	const struct ctl_table *table;
1699 
1700 	kfree(net->ipv4.sysctl_local_reserved_ports);
1701 	table = net->ipv4.ipv4_hdr->ctl_table_arg;
1702 	unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
1703 	kfree(table);
1704 }
1705 
1706 static __net_initdata struct pernet_operations ipv4_sysctl_ops = {
1707 	.init = ipv4_sysctl_init_net,
1708 	.exit = ipv4_sysctl_exit_net,
1709 };
1710 
1711 static __init int sysctl_ipv4_init(void)
1712 {
1713 	struct ctl_table_header *hdr;
1714 
1715 	hdr = register_net_sysctl(&init_net, "net/ipv4", ipv4_table);
1716 	if (!hdr)
1717 		return -ENOMEM;
1718 
1719 	proc_fib_multipath_hash_init_rand_seed();
1720 
1721 	if (register_pernet_subsys(&ipv4_sysctl_ops)) {
1722 		unregister_net_sysctl_table(hdr);
1723 		return -ENOMEM;
1724 	}
1725 
1726 	return 0;
1727 }
1728 
1729 __initcall(sysctl_ipv4_init);
1730