• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
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/mm.h>
10 #include <linux/module.h>
11 #include <linux/sysctl.h>
12 #include <linux/igmp.h>
13 #include <linux/inetdevice.h>
14 #include <linux/seqlock.h>
15 #include <linux/init.h>
16 #include <linux/slab.h>
17 #include <linux/nsproxy.h>
18 #include <linux/swap.h>
19 #include <net/snmp.h>
20 #include <net/icmp.h>
21 #include <net/ip.h>
22 #include <net/ip_fib.h>
23 #include <net/route.h>
24 #include <net/tcp.h>
25 #include <net/udp.h>
26 #include <net/cipso_ipv4.h>
27 #include <net/inet_frag.h>
28 #include <net/ping.h>
29 #include <net/protocol.h>
30 #include <net/netevent.h>
31 
32 static int two = 2;
33 static int three __maybe_unused = 3;
34 static int four = 4;
35 static int thousand = 1000;
36 static int tcp_retr1_max = 255;
37 static int ip_local_port_range_min[] = { 1, 1 };
38 static int ip_local_port_range_max[] = { 65535, 65535 };
39 static int tcp_adv_win_scale_min = -31;
40 static int tcp_adv_win_scale_max = 31;
41 static int tcp_app_win_max = 31;
42 static int tcp_min_snd_mss_min = TCP_MIN_SND_MSS;
43 static int tcp_min_snd_mss_max = 65535;
44 static int ip_privileged_port_min;
45 static int ip_privileged_port_max = 65535;
46 static int ip_ttl_min = 1;
47 static int ip_ttl_max = 255;
48 static int tcp_syn_retries_min = 1;
49 static int tcp_syn_retries_max = MAX_TCP_SYNCNT;
50 static int ip_ping_group_range_min[] = { 0, 0 };
51 static int ip_ping_group_range_max[] = { GID_T_MAX, GID_T_MAX };
52 static u32 u32_max_div_HZ = UINT_MAX / HZ;
53 static int one_day_secs = 24 * 3600;
54 static u32 fib_multipath_hash_fields_all_mask __maybe_unused =
55 	FIB_MULTIPATH_HASH_FIELD_ALL_MASK;
56 
57 /* obsolete */
58 static int sysctl_tcp_low_latency __read_mostly;
59 
60 /* Update system visible IP port range */
set_local_port_range(struct net * net,int range[2])61 static void set_local_port_range(struct net *net, int range[2])
62 {
63 	bool same_parity = !((range[0] ^ range[1]) & 1);
64 
65 	write_seqlock_bh(&net->ipv4.ip_local_ports.lock);
66 	if (same_parity && !net->ipv4.ip_local_ports.warned) {
67 		net->ipv4.ip_local_ports.warned = true;
68 		pr_err_ratelimited("ip_local_port_range: prefer different parity for start/end values.\n");
69 	}
70 	net->ipv4.ip_local_ports.range[0] = range[0];
71 	net->ipv4.ip_local_ports.range[1] = range[1];
72 	write_sequnlock_bh(&net->ipv4.ip_local_ports.lock);
73 }
74 
75 /* Validate changes from /proc interface. */
ipv4_local_port_range(struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)76 static int ipv4_local_port_range(struct ctl_table *table, int write,
77 				 void *buffer, size_t *lenp, loff_t *ppos)
78 {
79 	struct net *net =
80 		container_of(table->data, struct net, ipv4.ip_local_ports.range);
81 	int ret;
82 	int range[2];
83 	struct ctl_table tmp = {
84 		.data = &range,
85 		.maxlen = sizeof(range),
86 		.mode = table->mode,
87 		.extra1 = &ip_local_port_range_min,
88 		.extra2 = &ip_local_port_range_max,
89 	};
90 
91 	inet_get_local_port_range(net, &range[0], &range[1]);
92 
93 	ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
94 
95 	if (write && ret == 0) {
96 		/* Ensure that the upper limit is not smaller than the lower,
97 		 * and that the lower does not encroach upon the privileged
98 		 * port limit.
99 		 */
100 		if ((range[1] < range[0]) ||
101 		    (range[0] < READ_ONCE(net->ipv4.sysctl_ip_prot_sock)))
102 			ret = -EINVAL;
103 		else
104 			set_local_port_range(net, range);
105 	}
106 
107 	return ret;
108 }
109 
110 /* Validate changes from /proc interface. */
ipv4_privileged_ports(struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)111 static int ipv4_privileged_ports(struct ctl_table *table, int write,
112 				void *buffer, size_t *lenp, loff_t *ppos)
113 {
114 	struct net *net = container_of(table->data, struct net,
115 	    ipv4.sysctl_ip_prot_sock);
116 	int ret;
117 	int pports;
118 	int range[2];
119 	struct ctl_table tmp = {
120 		.data = &pports,
121 		.maxlen = sizeof(pports),
122 		.mode = table->mode,
123 		.extra1 = &ip_privileged_port_min,
124 		.extra2 = &ip_privileged_port_max,
125 	};
126 
127 	pports = READ_ONCE(net->ipv4.sysctl_ip_prot_sock);
128 
129 	ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
130 
131 	if (write && ret == 0) {
132 		inet_get_local_port_range(net, &range[0], &range[1]);
133 		/* Ensure that the local port range doesn't overlap with the
134 		 * privileged port range.
135 		 */
136 		if (range[0] < pports)
137 			ret = -EINVAL;
138 		else
139 			WRITE_ONCE(net->ipv4.sysctl_ip_prot_sock, pports);
140 	}
141 
142 	return ret;
143 }
144 
inet_get_ping_group_range_table(struct ctl_table * table,kgid_t * low,kgid_t * high)145 static void inet_get_ping_group_range_table(struct ctl_table *table, kgid_t *low, kgid_t *high)
146 {
147 	kgid_t *data = table->data;
148 	struct net *net =
149 		container_of(table->data, struct net, ipv4.ping_group_range.range);
150 	unsigned int seq;
151 	do {
152 		seq = read_seqbegin(&net->ipv4.ping_group_range.lock);
153 
154 		*low = data[0];
155 		*high = data[1];
156 	} while (read_seqretry(&net->ipv4.ping_group_range.lock, seq));
157 }
158 
159 /* Update system visible IP port range */
set_ping_group_range(struct ctl_table * table,kgid_t low,kgid_t high)160 static void set_ping_group_range(struct ctl_table *table, kgid_t low, kgid_t high)
161 {
162 	kgid_t *data = table->data;
163 	struct net *net =
164 		container_of(table->data, struct net, ipv4.ping_group_range.range);
165 	write_seqlock(&net->ipv4.ping_group_range.lock);
166 	data[0] = low;
167 	data[1] = high;
168 	write_sequnlock(&net->ipv4.ping_group_range.lock);
169 }
170 
171 /* Validate changes from /proc interface. */
ipv4_ping_group_range(struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)172 static int ipv4_ping_group_range(struct ctl_table *table, int write,
173 				 void *buffer, size_t *lenp, loff_t *ppos)
174 {
175 	struct user_namespace *user_ns = current_user_ns();
176 	int ret;
177 	gid_t urange[2];
178 	kgid_t low, high;
179 	struct ctl_table tmp = {
180 		.data = &urange,
181 		.maxlen = sizeof(urange),
182 		.mode = table->mode,
183 		.extra1 = &ip_ping_group_range_min,
184 		.extra2 = &ip_ping_group_range_max,
185 	};
186 
187 	inet_get_ping_group_range_table(table, &low, &high);
188 	urange[0] = from_kgid_munged(user_ns, low);
189 	urange[1] = from_kgid_munged(user_ns, high);
190 	ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
191 
192 	if (write && ret == 0) {
193 		low = make_kgid(user_ns, urange[0]);
194 		high = make_kgid(user_ns, urange[1]);
195 		if (!gid_valid(low) || !gid_valid(high))
196 			return -EINVAL;
197 		if (urange[1] < urange[0] || gid_lt(high, low)) {
198 			low = make_kgid(&init_user_ns, 1);
199 			high = make_kgid(&init_user_ns, 0);
200 		}
201 		set_ping_group_range(table, low, high);
202 	}
203 
204 	return ret;
205 }
206 
ipv4_fwd_update_priority(struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)207 static int ipv4_fwd_update_priority(struct ctl_table *table, int write,
208 				    void *buffer, size_t *lenp, loff_t *ppos)
209 {
210 	struct net *net;
211 	int ret;
212 
213 	net = container_of(table->data, struct net,
214 			   ipv4.sysctl_ip_fwd_update_priority);
215 	ret = proc_dou8vec_minmax(table, write, buffer, lenp, ppos);
216 	if (write && ret == 0)
217 		call_netevent_notifiers(NETEVENT_IPV4_FWD_UPDATE_PRIORITY_UPDATE,
218 					net);
219 
220 	return ret;
221 }
222 
proc_tcp_congestion_control(struct ctl_table * ctl,int write,void * buffer,size_t * lenp,loff_t * ppos)223 static int proc_tcp_congestion_control(struct ctl_table *ctl, int write,
224 				       void *buffer, size_t *lenp, loff_t *ppos)
225 {
226 	struct net *net = container_of(ctl->data, struct net,
227 				       ipv4.tcp_congestion_control);
228 	char val[TCP_CA_NAME_MAX];
229 	struct ctl_table tbl = {
230 		.data = val,
231 		.maxlen = TCP_CA_NAME_MAX,
232 	};
233 	int ret;
234 
235 	tcp_get_default_congestion_control(net, val);
236 
237 	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
238 	if (write && ret == 0)
239 		ret = tcp_set_default_congestion_control(net, val);
240 	return ret;
241 }
242 
proc_tcp_available_congestion_control(struct ctl_table * ctl,int write,void * buffer,size_t * lenp,loff_t * ppos)243 static int proc_tcp_available_congestion_control(struct ctl_table *ctl,
244 						 int write, void *buffer,
245 						 size_t *lenp, loff_t *ppos)
246 {
247 	struct ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX, };
248 	int ret;
249 
250 	tbl.data = kmalloc(tbl.maxlen, GFP_USER);
251 	if (!tbl.data)
252 		return -ENOMEM;
253 	tcp_get_available_congestion_control(tbl.data, TCP_CA_BUF_MAX);
254 	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
255 	kfree(tbl.data);
256 	return ret;
257 }
258 
proc_allowed_congestion_control(struct ctl_table * ctl,int write,void * buffer,size_t * lenp,loff_t * ppos)259 static int proc_allowed_congestion_control(struct ctl_table *ctl,
260 					   int write, void *buffer,
261 					   size_t *lenp, loff_t *ppos)
262 {
263 	struct ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX };
264 	int ret;
265 
266 	tbl.data = kmalloc(tbl.maxlen, GFP_USER);
267 	if (!tbl.data)
268 		return -ENOMEM;
269 
270 	tcp_get_allowed_congestion_control(tbl.data, tbl.maxlen);
271 	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
272 	if (write && ret == 0)
273 		ret = tcp_set_allowed_congestion_control(tbl.data);
274 	kfree(tbl.data);
275 	return ret;
276 }
277 
sscanf_key(char * buf,__le32 * key)278 static int sscanf_key(char *buf, __le32 *key)
279 {
280 	u32 user_key[4];
281 	int i, ret = 0;
282 
283 	if (sscanf(buf, "%x-%x-%x-%x", user_key, user_key + 1,
284 		   user_key + 2, user_key + 3) != 4) {
285 		ret = -EINVAL;
286 	} else {
287 		for (i = 0; i < ARRAY_SIZE(user_key); i++)
288 			key[i] = cpu_to_le32(user_key[i]);
289 	}
290 	pr_debug("proc TFO key set 0x%x-%x-%x-%x <- 0x%s: %u\n",
291 		 user_key[0], user_key[1], user_key[2], user_key[3], buf, ret);
292 
293 	return ret;
294 }
295 
proc_tcp_fastopen_key(struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)296 static int proc_tcp_fastopen_key(struct ctl_table *table, int write,
297 				 void *buffer, size_t *lenp, loff_t *ppos)
298 {
299 	struct net *net = container_of(table->data, struct net,
300 	    ipv4.sysctl_tcp_fastopen);
301 	/* maxlen to print the list of keys in hex (*2), with dashes
302 	 * separating doublewords and a comma in between keys.
303 	 */
304 	struct ctl_table tbl = { .maxlen = ((TCP_FASTOPEN_KEY_LENGTH *
305 					    2 * TCP_FASTOPEN_KEY_MAX) +
306 					    (TCP_FASTOPEN_KEY_MAX * 5)) };
307 	u32 user_key[TCP_FASTOPEN_KEY_BUF_LENGTH / sizeof(u32)];
308 	__le32 key[TCP_FASTOPEN_KEY_BUF_LENGTH / sizeof(__le32)];
309 	char *backup_data;
310 	int ret, i = 0, off = 0, n_keys;
311 
312 	tbl.data = kmalloc(tbl.maxlen, GFP_KERNEL);
313 	if (!tbl.data)
314 		return -ENOMEM;
315 
316 	n_keys = tcp_fastopen_get_cipher(net, NULL, (u64 *)key);
317 	if (!n_keys) {
318 		memset(&key[0], 0, TCP_FASTOPEN_KEY_LENGTH);
319 		n_keys = 1;
320 	}
321 
322 	for (i = 0; i < n_keys * 4; i++)
323 		user_key[i] = le32_to_cpu(key[i]);
324 
325 	for (i = 0; i < n_keys; i++) {
326 		off += snprintf(tbl.data + off, tbl.maxlen - off,
327 				"%08x-%08x-%08x-%08x",
328 				user_key[i * 4],
329 				user_key[i * 4 + 1],
330 				user_key[i * 4 + 2],
331 				user_key[i * 4 + 3]);
332 
333 		if (WARN_ON_ONCE(off >= tbl.maxlen - 1))
334 			break;
335 
336 		if (i + 1 < n_keys)
337 			off += snprintf(tbl.data + off, tbl.maxlen - off, ",");
338 	}
339 
340 	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
341 
342 	if (write && ret == 0) {
343 		backup_data = strchr(tbl.data, ',');
344 		if (backup_data) {
345 			*backup_data = '\0';
346 			backup_data++;
347 		}
348 		if (sscanf_key(tbl.data, key)) {
349 			ret = -EINVAL;
350 			goto bad_key;
351 		}
352 		if (backup_data) {
353 			if (sscanf_key(backup_data, key + 4)) {
354 				ret = -EINVAL;
355 				goto bad_key;
356 			}
357 		}
358 		tcp_fastopen_reset_cipher(net, NULL, key,
359 					  backup_data ? key + 4 : NULL);
360 	}
361 
362 bad_key:
363 	kfree(tbl.data);
364 	return ret;
365 }
366 
proc_tfo_blackhole_detect_timeout(struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)367 static int proc_tfo_blackhole_detect_timeout(struct ctl_table *table,
368 					     int write, void *buffer,
369 					     size_t *lenp, loff_t *ppos)
370 {
371 	struct net *net = container_of(table->data, struct net,
372 	    ipv4.sysctl_tcp_fastopen_blackhole_timeout);
373 	int ret;
374 
375 	ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
376 	if (write && ret == 0)
377 		atomic_set(&net->ipv4.tfo_active_disable_times, 0);
378 
379 	return ret;
380 }
381 
proc_tcp_available_ulp(struct ctl_table * ctl,int write,void * buffer,size_t * lenp,loff_t * ppos)382 static int proc_tcp_available_ulp(struct ctl_table *ctl,
383 				  int write, void *buffer, size_t *lenp,
384 				  loff_t *ppos)
385 {
386 	struct ctl_table tbl = { .maxlen = TCP_ULP_BUF_MAX, };
387 	int ret;
388 
389 	tbl.data = kmalloc(tbl.maxlen, GFP_USER);
390 	if (!tbl.data)
391 		return -ENOMEM;
392 	tcp_get_available_ulp(tbl.data, TCP_ULP_BUF_MAX);
393 	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
394 	kfree(tbl.data);
395 
396 	return ret;
397 }
398 
399 #ifdef CONFIG_IP_ROUTE_MULTIPATH
proc_fib_multipath_hash_policy(struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)400 static int proc_fib_multipath_hash_policy(struct ctl_table *table, int write,
401 					  void *buffer, size_t *lenp,
402 					  loff_t *ppos)
403 {
404 	struct net *net = container_of(table->data, struct net,
405 	    ipv4.sysctl_fib_multipath_hash_policy);
406 	int ret;
407 
408 	ret = proc_dou8vec_minmax(table, write, buffer, lenp, ppos);
409 	if (write && ret == 0)
410 		call_netevent_notifiers(NETEVENT_IPV4_MPATH_HASH_UPDATE, net);
411 
412 	return ret;
413 }
414 
proc_fib_multipath_hash_fields(struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)415 static int proc_fib_multipath_hash_fields(struct ctl_table *table, int write,
416 					  void *buffer, size_t *lenp,
417 					  loff_t *ppos)
418 {
419 	struct net *net;
420 	int ret;
421 
422 	net = container_of(table->data, struct net,
423 			   ipv4.sysctl_fib_multipath_hash_fields);
424 	ret = proc_douintvec_minmax(table, write, buffer, lenp, ppos);
425 	if (write && ret == 0)
426 		call_netevent_notifiers(NETEVENT_IPV4_MPATH_HASH_UPDATE, net);
427 
428 	return ret;
429 }
430 #endif
431 
432 static struct ctl_table ipv4_table[] = {
433 	{
434 		.procname	= "tcp_max_orphans",
435 		.data		= &sysctl_tcp_max_orphans,
436 		.maxlen		= sizeof(int),
437 		.mode		= 0644,
438 		.proc_handler	= proc_dointvec
439 	},
440 	{
441 		.procname	= "inet_peer_threshold",
442 		.data		= &inet_peer_threshold,
443 		.maxlen		= sizeof(int),
444 		.mode		= 0644,
445 		.proc_handler	= proc_dointvec
446 	},
447 	{
448 		.procname	= "inet_peer_minttl",
449 		.data		= &inet_peer_minttl,
450 		.maxlen		= sizeof(int),
451 		.mode		= 0644,
452 		.proc_handler	= proc_dointvec_jiffies,
453 	},
454 	{
455 		.procname	= "inet_peer_maxttl",
456 		.data		= &inet_peer_maxttl,
457 		.maxlen		= sizeof(int),
458 		.mode		= 0644,
459 		.proc_handler	= proc_dointvec_jiffies,
460 	},
461 	{
462 		.procname	= "tcp_mem",
463 		.maxlen		= sizeof(sysctl_tcp_mem),
464 		.data		= &sysctl_tcp_mem,
465 		.mode		= 0644,
466 		.proc_handler	= proc_doulongvec_minmax,
467 	},
468 	{
469 		.procname	= "tcp_low_latency",
470 		.data		= &sysctl_tcp_low_latency,
471 		.maxlen		= sizeof(int),
472 		.mode		= 0644,
473 		.proc_handler	= proc_dointvec
474 	},
475 #ifdef CONFIG_NETLABEL
476 	{
477 		.procname	= "cipso_cache_enable",
478 		.data		= &cipso_v4_cache_enabled,
479 		.maxlen		= sizeof(int),
480 		.mode		= 0644,
481 		.proc_handler	= proc_dointvec,
482 	},
483 	{
484 		.procname	= "cipso_cache_bucket_size",
485 		.data		= &cipso_v4_cache_bucketsize,
486 		.maxlen		= sizeof(int),
487 		.mode		= 0644,
488 		.proc_handler	= proc_dointvec,
489 	},
490 	{
491 		.procname	= "cipso_rbm_optfmt",
492 		.data		= &cipso_v4_rbm_optfmt,
493 		.maxlen		= sizeof(int),
494 		.mode		= 0644,
495 		.proc_handler	= proc_dointvec,
496 	},
497 	{
498 		.procname	= "cipso_rbm_strictvalid",
499 		.data		= &cipso_v4_rbm_strictvalid,
500 		.maxlen		= sizeof(int),
501 		.mode		= 0644,
502 		.proc_handler	= proc_dointvec,
503 	},
504 #endif /* CONFIG_NETLABEL */
505 	{
506 		.procname	= "tcp_available_ulp",
507 		.maxlen		= TCP_ULP_BUF_MAX,
508 		.mode		= 0444,
509 		.proc_handler   = proc_tcp_available_ulp,
510 	},
511 	{
512 		.procname	= "icmp_msgs_per_sec",
513 		.data		= &sysctl_icmp_msgs_per_sec,
514 		.maxlen		= sizeof(int),
515 		.mode		= 0644,
516 		.proc_handler	= proc_dointvec_minmax,
517 		.extra1		= SYSCTL_ZERO,
518 	},
519 	{
520 		.procname	= "icmp_msgs_burst",
521 		.data		= &sysctl_icmp_msgs_burst,
522 		.maxlen		= sizeof(int),
523 		.mode		= 0644,
524 		.proc_handler	= proc_dointvec_minmax,
525 		.extra1		= SYSCTL_ZERO,
526 	},
527 	{
528 		.procname	= "udp_mem",
529 		.data		= &sysctl_udp_mem,
530 		.maxlen		= sizeof(sysctl_udp_mem),
531 		.mode		= 0644,
532 		.proc_handler	= proc_doulongvec_minmax,
533 	},
534 	{
535 		.procname	= "fib_sync_mem",
536 		.data		= &sysctl_fib_sync_mem,
537 		.maxlen		= sizeof(sysctl_fib_sync_mem),
538 		.mode		= 0644,
539 		.proc_handler	= proc_douintvec_minmax,
540 		.extra1		= &sysctl_fib_sync_mem_min,
541 		.extra2		= &sysctl_fib_sync_mem_max,
542 	},
543 	{
544 		.procname	= "tcp_rx_skb_cache",
545 		.data		= &tcp_rx_skb_cache_key.key,
546 		.mode		= 0644,
547 		.proc_handler	= proc_do_static_key,
548 	},
549 	{
550 		.procname	= "tcp_tx_skb_cache",
551 		.data		= &tcp_tx_skb_cache_key.key,
552 		.mode		= 0644,
553 		.proc_handler	= proc_do_static_key,
554 	},
555 	{ }
556 };
557 
558 static struct ctl_table ipv4_net_table[] = {
559 	{
560 		.procname	= "icmp_echo_ignore_all",
561 		.data		= &init_net.ipv4.sysctl_icmp_echo_ignore_all,
562 		.maxlen		= sizeof(u8),
563 		.mode		= 0644,
564 		.proc_handler	= proc_dou8vec_minmax,
565 	},
566 	{
567 		.procname	= "icmp_echo_enable_probe",
568 		.data		= &init_net.ipv4.sysctl_icmp_echo_enable_probe,
569 		.maxlen		= sizeof(u8),
570 		.mode		= 0644,
571 		.proc_handler	= proc_dou8vec_minmax,
572 		.extra1		= SYSCTL_ZERO,
573 		.extra2		= SYSCTL_ONE
574 	},
575 	{
576 		.procname	= "icmp_echo_ignore_broadcasts",
577 		.data		= &init_net.ipv4.sysctl_icmp_echo_ignore_broadcasts,
578 		.maxlen		= sizeof(u8),
579 		.mode		= 0644,
580 		.proc_handler	= proc_dou8vec_minmax,
581 	},
582 	{
583 		.procname	= "icmp_ignore_bogus_error_responses",
584 		.data		= &init_net.ipv4.sysctl_icmp_ignore_bogus_error_responses,
585 		.maxlen		= sizeof(u8),
586 		.mode		= 0644,
587 		.proc_handler	= proc_dou8vec_minmax,
588 		.extra1		= SYSCTL_ZERO,
589 		.extra2		= SYSCTL_ONE
590 	},
591 	{
592 		.procname	= "icmp_errors_use_inbound_ifaddr",
593 		.data		= &init_net.ipv4.sysctl_icmp_errors_use_inbound_ifaddr,
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_ratelimit",
602 		.data		= &init_net.ipv4.sysctl_icmp_ratelimit,
603 		.maxlen		= sizeof(int),
604 		.mode		= 0644,
605 		.proc_handler	= proc_dointvec_ms_jiffies,
606 	},
607 	{
608 		.procname	= "icmp_ratemask",
609 		.data		= &init_net.ipv4.sysctl_icmp_ratemask,
610 		.maxlen		= sizeof(int),
611 		.mode		= 0644,
612 		.proc_handler	= proc_dointvec
613 	},
614 	{
615 		.procname	= "ping_group_range",
616 		.data		= &init_net.ipv4.ping_group_range.range,
617 		.maxlen		= sizeof(gid_t)*2,
618 		.mode		= 0644,
619 		.proc_handler	= ipv4_ping_group_range,
620 	},
621 #ifdef CONFIG_NET_L3_MASTER_DEV
622 	{
623 		.procname	= "raw_l3mdev_accept",
624 		.data		= &init_net.ipv4.sysctl_raw_l3mdev_accept,
625 		.maxlen		= sizeof(u8),
626 		.mode		= 0644,
627 		.proc_handler	= proc_dou8vec_minmax,
628 		.extra1		= SYSCTL_ZERO,
629 		.extra2		= SYSCTL_ONE,
630 	},
631 #endif
632 	{
633 		.procname	= "tcp_ecn",
634 		.data		= &init_net.ipv4.sysctl_tcp_ecn,
635 		.maxlen		= sizeof(u8),
636 		.mode		= 0644,
637 		.proc_handler	= proc_dou8vec_minmax,
638 		.extra1		= SYSCTL_ZERO,
639 		.extra2		= SYSCTL_TWO,
640 	},
641 	{
642 		.procname	= "tcp_ecn_fallback",
643 		.data		= &init_net.ipv4.sysctl_tcp_ecn_fallback,
644 		.maxlen		= sizeof(u8),
645 		.mode		= 0644,
646 		.proc_handler	= proc_dou8vec_minmax,
647 		.extra1		= SYSCTL_ZERO,
648 		.extra2		= SYSCTL_ONE,
649 	},
650 	{
651 		.procname	= "ip_dynaddr",
652 		.data		= &init_net.ipv4.sysctl_ip_dynaddr,
653 		.maxlen		= sizeof(u8),
654 		.mode		= 0644,
655 		.proc_handler	= proc_dou8vec_minmax,
656 	},
657 	{
658 		.procname	= "ip_early_demux",
659 		.data		= &init_net.ipv4.sysctl_ip_early_demux,
660 		.maxlen		= sizeof(u8),
661 		.mode		= 0644,
662 		.proc_handler	= proc_dou8vec_minmax,
663 	},
664 	{
665 		.procname       = "udp_early_demux",
666 		.data           = &init_net.ipv4.sysctl_udp_early_demux,
667 		.maxlen         = sizeof(u8),
668 		.mode           = 0644,
669 		.proc_handler   = proc_dou8vec_minmax,
670 	},
671 	{
672 		.procname       = "tcp_early_demux",
673 		.data           = &init_net.ipv4.sysctl_tcp_early_demux,
674 		.maxlen         = sizeof(u8),
675 		.mode           = 0644,
676 		.proc_handler   = proc_dou8vec_minmax,
677 	},
678 	{
679 		.procname       = "nexthop_compat_mode",
680 		.data           = &init_net.ipv4.sysctl_nexthop_compat_mode,
681 		.maxlen         = sizeof(u8),
682 		.mode           = 0644,
683 		.proc_handler   = proc_dou8vec_minmax,
684 		.extra1		= SYSCTL_ZERO,
685 		.extra2		= SYSCTL_ONE,
686 	},
687 	{
688 		.procname	= "ip_default_ttl",
689 		.data		= &init_net.ipv4.sysctl_ip_default_ttl,
690 		.maxlen		= sizeof(u8),
691 		.mode		= 0644,
692 		.proc_handler	= proc_dou8vec_minmax,
693 		.extra1		= &ip_ttl_min,
694 		.extra2		= &ip_ttl_max,
695 	},
696 	{
697 		.procname	= "ip_local_port_range",
698 		.maxlen		= sizeof(init_net.ipv4.ip_local_ports.range),
699 		.data		= &init_net.ipv4.ip_local_ports.range,
700 		.mode		= 0644,
701 		.proc_handler	= ipv4_local_port_range,
702 	},
703 	{
704 		.procname	= "ip_local_reserved_ports",
705 		.data		= &init_net.ipv4.sysctl_local_reserved_ports,
706 		.maxlen		= 65536,
707 		.mode		= 0644,
708 		.proc_handler	= proc_do_large_bitmap,
709 	},
710 	{
711 		.procname	= "ip_local_unbindable_ports",
712 		.data		= &init_net.ipv4.sysctl_local_unbindable_ports,
713 		.maxlen		= 65536,
714 		.mode		= 0644,
715 		.proc_handler	= proc_do_large_bitmap,
716 	},
717 	{
718 		.procname	= "ip_no_pmtu_disc",
719 		.data		= &init_net.ipv4.sysctl_ip_no_pmtu_disc,
720 		.maxlen		= sizeof(u8),
721 		.mode		= 0644,
722 		.proc_handler	= proc_dou8vec_minmax,
723 	},
724 	{
725 		.procname	= "ip_forward_use_pmtu",
726 		.data		= &init_net.ipv4.sysctl_ip_fwd_use_pmtu,
727 		.maxlen		= sizeof(u8),
728 		.mode		= 0644,
729 		.proc_handler	= proc_dou8vec_minmax,
730 	},
731 	{
732 		.procname	= "ip_forward_update_priority",
733 		.data		= &init_net.ipv4.sysctl_ip_fwd_update_priority,
734 		.maxlen		= sizeof(u8),
735 		.mode		= 0644,
736 		.proc_handler   = ipv4_fwd_update_priority,
737 		.extra1		= SYSCTL_ZERO,
738 		.extra2		= SYSCTL_ONE,
739 	},
740 	{
741 		.procname	= "ip_nonlocal_bind",
742 		.data		= &init_net.ipv4.sysctl_ip_nonlocal_bind,
743 		.maxlen		= sizeof(u8),
744 		.mode		= 0644,
745 		.proc_handler	= proc_dou8vec_minmax,
746 	},
747 	{
748 		.procname	= "ip_autobind_reuse",
749 		.data		= &init_net.ipv4.sysctl_ip_autobind_reuse,
750 		.maxlen		= sizeof(u8),
751 		.mode		= 0644,
752 		.proc_handler	= proc_dou8vec_minmax,
753 		.extra1         = SYSCTL_ZERO,
754 		.extra2         = SYSCTL_ONE,
755 	},
756 	{
757 		.procname	= "fwmark_reflect",
758 		.data		= &init_net.ipv4.sysctl_fwmark_reflect,
759 		.maxlen		= sizeof(u8),
760 		.mode		= 0644,
761 		.proc_handler	= proc_dou8vec_minmax,
762 	},
763 	{
764 		.procname	= "tcp_fwmark_accept",
765 		.data		= &init_net.ipv4.sysctl_tcp_fwmark_accept,
766 		.maxlen		= sizeof(u8),
767 		.mode		= 0644,
768 		.proc_handler	= proc_dou8vec_minmax,
769 	},
770 #ifdef CONFIG_NET_L3_MASTER_DEV
771 	{
772 		.procname	= "tcp_l3mdev_accept",
773 		.data		= &init_net.ipv4.sysctl_tcp_l3mdev_accept,
774 		.maxlen		= sizeof(u8),
775 		.mode		= 0644,
776 		.proc_handler	= proc_dou8vec_minmax,
777 		.extra1		= SYSCTL_ZERO,
778 		.extra2		= SYSCTL_ONE,
779 	},
780 #endif
781 	{
782 		.procname	= "tcp_mtu_probing",
783 		.data		= &init_net.ipv4.sysctl_tcp_mtu_probing,
784 		.maxlen		= sizeof(u8),
785 		.mode		= 0644,
786 		.proc_handler	= proc_dou8vec_minmax,
787 	},
788 	{
789 		.procname	= "tcp_base_mss",
790 		.data		= &init_net.ipv4.sysctl_tcp_base_mss,
791 		.maxlen		= sizeof(int),
792 		.mode		= 0644,
793 		.proc_handler	= proc_dointvec,
794 	},
795 	{
796 		.procname	= "tcp_min_snd_mss",
797 		.data		= &init_net.ipv4.sysctl_tcp_min_snd_mss,
798 		.maxlen		= sizeof(int),
799 		.mode		= 0644,
800 		.proc_handler	= proc_dointvec_minmax,
801 		.extra1		= &tcp_min_snd_mss_min,
802 		.extra2		= &tcp_min_snd_mss_max,
803 	},
804 	{
805 		.procname	= "tcp_mtu_probe_floor",
806 		.data		= &init_net.ipv4.sysctl_tcp_mtu_probe_floor,
807 		.maxlen		= sizeof(int),
808 		.mode		= 0644,
809 		.proc_handler	= proc_dointvec_minmax,
810 		.extra1		= &tcp_min_snd_mss_min,
811 		.extra2		= &tcp_min_snd_mss_max,
812 	},
813 	{
814 		.procname	= "tcp_probe_threshold",
815 		.data		= &init_net.ipv4.sysctl_tcp_probe_threshold,
816 		.maxlen		= sizeof(int),
817 		.mode		= 0644,
818 		.proc_handler	= proc_dointvec,
819 	},
820 	{
821 		.procname	= "tcp_probe_interval",
822 		.data		= &init_net.ipv4.sysctl_tcp_probe_interval,
823 		.maxlen		= sizeof(u32),
824 		.mode		= 0644,
825 		.proc_handler	= proc_douintvec_minmax,
826 		.extra2		= &u32_max_div_HZ,
827 	},
828 	{
829 		.procname	= "igmp_link_local_mcast_reports",
830 		.data		= &init_net.ipv4.sysctl_igmp_llm_reports,
831 		.maxlen		= sizeof(u8),
832 		.mode		= 0644,
833 		.proc_handler	= proc_dou8vec_minmax,
834 	},
835 	{
836 		.procname	= "igmp_max_memberships",
837 		.data		= &init_net.ipv4.sysctl_igmp_max_memberships,
838 		.maxlen		= sizeof(int),
839 		.mode		= 0644,
840 		.proc_handler	= proc_dointvec
841 	},
842 	{
843 		.procname	= "igmp_max_msf",
844 		.data		= &init_net.ipv4.sysctl_igmp_max_msf,
845 		.maxlen		= sizeof(int),
846 		.mode		= 0644,
847 		.proc_handler	= proc_dointvec
848 	},
849 #ifdef CONFIG_IP_MULTICAST
850 	{
851 		.procname	= "igmp_qrv",
852 		.data		= &init_net.ipv4.sysctl_igmp_qrv,
853 		.maxlen		= sizeof(int),
854 		.mode		= 0644,
855 		.proc_handler	= proc_dointvec_minmax,
856 		.extra1		= SYSCTL_ONE
857 	},
858 #endif
859 	{
860 		.procname	= "tcp_congestion_control",
861 		.data		= &init_net.ipv4.tcp_congestion_control,
862 		.mode		= 0644,
863 		.maxlen		= TCP_CA_NAME_MAX,
864 		.proc_handler	= proc_tcp_congestion_control,
865 	},
866 	{
867 		.procname	= "tcp_available_congestion_control",
868 		.maxlen		= TCP_CA_BUF_MAX,
869 		.mode		= 0444,
870 		.proc_handler   = proc_tcp_available_congestion_control,
871 	},
872 	{
873 		.procname	= "tcp_allowed_congestion_control",
874 		.maxlen		= TCP_CA_BUF_MAX,
875 		.mode		= 0644,
876 		.proc_handler   = proc_allowed_congestion_control,
877 	},
878 	{
879 		.procname	= "tcp_keepalive_time",
880 		.data		= &init_net.ipv4.sysctl_tcp_keepalive_time,
881 		.maxlen		= sizeof(int),
882 		.mode		= 0644,
883 		.proc_handler	= proc_dointvec_jiffies,
884 	},
885 	{
886 		.procname	= "tcp_keepalive_probes",
887 		.data		= &init_net.ipv4.sysctl_tcp_keepalive_probes,
888 		.maxlen		= sizeof(u8),
889 		.mode		= 0644,
890 		.proc_handler	= proc_dou8vec_minmax,
891 	},
892 	{
893 		.procname	= "tcp_keepalive_intvl",
894 		.data		= &init_net.ipv4.sysctl_tcp_keepalive_intvl,
895 		.maxlen		= sizeof(int),
896 		.mode		= 0644,
897 		.proc_handler	= proc_dointvec_jiffies,
898 	},
899 	{
900 		.procname	= "tcp_syn_retries",
901 		.data		= &init_net.ipv4.sysctl_tcp_syn_retries,
902 		.maxlen		= sizeof(u8),
903 		.mode		= 0644,
904 		.proc_handler	= proc_dou8vec_minmax,
905 		.extra1		= &tcp_syn_retries_min,
906 		.extra2		= &tcp_syn_retries_max
907 	},
908 	{
909 		.procname	= "tcp_synack_retries",
910 		.data		= &init_net.ipv4.sysctl_tcp_synack_retries,
911 		.maxlen		= sizeof(u8),
912 		.mode		= 0644,
913 		.proc_handler	= proc_dou8vec_minmax,
914 	},
915 #ifdef CONFIG_SYN_COOKIES
916 	{
917 		.procname	= "tcp_syncookies",
918 		.data		= &init_net.ipv4.sysctl_tcp_syncookies,
919 		.maxlen		= sizeof(u8),
920 		.mode		= 0644,
921 		.proc_handler	= proc_dou8vec_minmax,
922 	},
923 #endif
924 	{
925 		.procname	= "tcp_migrate_req",
926 		.data		= &init_net.ipv4.sysctl_tcp_migrate_req,
927 		.maxlen		= sizeof(u8),
928 		.mode		= 0644,
929 		.proc_handler	= proc_dou8vec_minmax,
930 		.extra1		= SYSCTL_ZERO,
931 		.extra2		= SYSCTL_ONE
932 	},
933 	{
934 		.procname	= "tcp_reordering",
935 		.data		= &init_net.ipv4.sysctl_tcp_reordering,
936 		.maxlen		= sizeof(int),
937 		.mode		= 0644,
938 		.proc_handler	= proc_dointvec
939 	},
940 	{
941 		.procname	= "tcp_retries1",
942 		.data		= &init_net.ipv4.sysctl_tcp_retries1,
943 		.maxlen		= sizeof(u8),
944 		.mode		= 0644,
945 		.proc_handler	= proc_dou8vec_minmax,
946 		.extra2		= &tcp_retr1_max
947 	},
948 	{
949 		.procname	= "tcp_retries2",
950 		.data		= &init_net.ipv4.sysctl_tcp_retries2,
951 		.maxlen		= sizeof(u8),
952 		.mode		= 0644,
953 		.proc_handler	= proc_dou8vec_minmax,
954 	},
955 	{
956 		.procname	= "tcp_orphan_retries",
957 		.data		= &init_net.ipv4.sysctl_tcp_orphan_retries,
958 		.maxlen		= sizeof(u8),
959 		.mode		= 0644,
960 		.proc_handler	= proc_dou8vec_minmax,
961 	},
962 	{
963 		.procname	= "tcp_fin_timeout",
964 		.data		= &init_net.ipv4.sysctl_tcp_fin_timeout,
965 		.maxlen		= sizeof(int),
966 		.mode		= 0644,
967 		.proc_handler	= proc_dointvec_jiffies,
968 	},
969 	{
970 		.procname	= "tcp_notsent_lowat",
971 		.data		= &init_net.ipv4.sysctl_tcp_notsent_lowat,
972 		.maxlen		= sizeof(unsigned int),
973 		.mode		= 0644,
974 		.proc_handler	= proc_douintvec,
975 	},
976 	{
977 		.procname	= "tcp_tw_reuse",
978 		.data		= &init_net.ipv4.sysctl_tcp_tw_reuse,
979 		.maxlen		= sizeof(u8),
980 		.mode		= 0644,
981 		.proc_handler	= proc_dou8vec_minmax,
982 		.extra1		= SYSCTL_ZERO,
983 		.extra2		= &two,
984 	},
985 	{
986 		.procname	= "tcp_max_tw_buckets",
987 		.data		= &init_net.ipv4.tcp_death_row.sysctl_max_tw_buckets,
988 		.maxlen		= sizeof(int),
989 		.mode		= 0644,
990 		.proc_handler	= proc_dointvec
991 	},
992 	{
993 		.procname	= "tcp_max_syn_backlog",
994 		.data		= &init_net.ipv4.sysctl_max_syn_backlog,
995 		.maxlen		= sizeof(int),
996 		.mode		= 0644,
997 		.proc_handler	= proc_dointvec
998 	},
999 	{
1000 		.procname	= "tcp_fastopen",
1001 		.data		= &init_net.ipv4.sysctl_tcp_fastopen,
1002 		.maxlen		= sizeof(int),
1003 		.mode		= 0644,
1004 		.proc_handler	= proc_dointvec,
1005 	},
1006 	{
1007 		.procname	= "tcp_fastopen_key",
1008 		.mode		= 0600,
1009 		.data		= &init_net.ipv4.sysctl_tcp_fastopen,
1010 		/* maxlen to print the list of keys in hex (*2), with dashes
1011 		 * separating doublewords and a comma in between keys.
1012 		 */
1013 		.maxlen		= ((TCP_FASTOPEN_KEY_LENGTH *
1014 				   2 * TCP_FASTOPEN_KEY_MAX) +
1015 				   (TCP_FASTOPEN_KEY_MAX * 5)),
1016 		.proc_handler	= proc_tcp_fastopen_key,
1017 	},
1018 	{
1019 		.procname	= "tcp_fastopen_blackhole_timeout_sec",
1020 		.data		= &init_net.ipv4.sysctl_tcp_fastopen_blackhole_timeout,
1021 		.maxlen		= sizeof(int),
1022 		.mode		= 0644,
1023 		.proc_handler	= proc_tfo_blackhole_detect_timeout,
1024 		.extra1		= SYSCTL_ZERO,
1025 	},
1026 #ifdef CONFIG_IP_ROUTE_MULTIPATH
1027 	{
1028 		.procname	= "fib_multipath_use_neigh",
1029 		.data		= &init_net.ipv4.sysctl_fib_multipath_use_neigh,
1030 		.maxlen		= sizeof(u8),
1031 		.mode		= 0644,
1032 		.proc_handler	= proc_dou8vec_minmax,
1033 		.extra1		= SYSCTL_ZERO,
1034 		.extra2		= SYSCTL_ONE,
1035 	},
1036 	{
1037 		.procname	= "fib_multipath_hash_policy",
1038 		.data		= &init_net.ipv4.sysctl_fib_multipath_hash_policy,
1039 		.maxlen		= sizeof(u8),
1040 		.mode		= 0644,
1041 		.proc_handler	= proc_fib_multipath_hash_policy,
1042 		.extra1		= SYSCTL_ZERO,
1043 		.extra2		= &three,
1044 	},
1045 	{
1046 		.procname	= "fib_multipath_hash_fields",
1047 		.data		= &init_net.ipv4.sysctl_fib_multipath_hash_fields,
1048 		.maxlen		= sizeof(u32),
1049 		.mode		= 0644,
1050 		.proc_handler	= proc_fib_multipath_hash_fields,
1051 		.extra1		= SYSCTL_ONE,
1052 		.extra2		= &fib_multipath_hash_fields_all_mask,
1053 	},
1054 #endif
1055 	{
1056 		.procname	= "ip_unprivileged_port_start",
1057 		.maxlen		= sizeof(int),
1058 		.data		= &init_net.ipv4.sysctl_ip_prot_sock,
1059 		.mode		= 0644,
1060 		.proc_handler	= ipv4_privileged_ports,
1061 	},
1062 #ifdef CONFIG_NET_L3_MASTER_DEV
1063 	{
1064 		.procname	= "udp_l3mdev_accept",
1065 		.data		= &init_net.ipv4.sysctl_udp_l3mdev_accept,
1066 		.maxlen		= sizeof(u8),
1067 		.mode		= 0644,
1068 		.proc_handler	= proc_dou8vec_minmax,
1069 		.extra1		= SYSCTL_ZERO,
1070 		.extra2		= SYSCTL_ONE,
1071 	},
1072 #endif
1073 	{
1074 		.procname	= "tcp_sack",
1075 		.data		= &init_net.ipv4.sysctl_tcp_sack,
1076 		.maxlen		= sizeof(u8),
1077 		.mode		= 0644,
1078 		.proc_handler	= proc_dou8vec_minmax,
1079 	},
1080 	{
1081 		.procname	= "tcp_window_scaling",
1082 		.data		= &init_net.ipv4.sysctl_tcp_window_scaling,
1083 		.maxlen		= sizeof(u8),
1084 		.mode		= 0644,
1085 		.proc_handler	= proc_dou8vec_minmax,
1086 	},
1087 	{
1088 		.procname	= "tcp_timestamps",
1089 		.data		= &init_net.ipv4.sysctl_tcp_timestamps,
1090 		.maxlen		= sizeof(u8),
1091 		.mode		= 0644,
1092 		.proc_handler	= proc_dou8vec_minmax,
1093 	},
1094 	{
1095 		.procname	= "tcp_early_retrans",
1096 		.data		= &init_net.ipv4.sysctl_tcp_early_retrans,
1097 		.maxlen		= sizeof(u8),
1098 		.mode		= 0644,
1099 		.proc_handler	= proc_dou8vec_minmax,
1100 		.extra1		= SYSCTL_ZERO,
1101 		.extra2		= &four,
1102 	},
1103 	{
1104 		.procname	= "tcp_recovery",
1105 		.data		= &init_net.ipv4.sysctl_tcp_recovery,
1106 		.maxlen		= sizeof(u8),
1107 		.mode		= 0644,
1108 		.proc_handler	= proc_dou8vec_minmax,
1109 	},
1110 	{
1111 		.procname       = "tcp_thin_linear_timeouts",
1112 		.data           = &init_net.ipv4.sysctl_tcp_thin_linear_timeouts,
1113 		.maxlen         = sizeof(u8),
1114 		.mode           = 0644,
1115 		.proc_handler   = proc_dou8vec_minmax,
1116 	},
1117 	{
1118 		.procname	= "tcp_slow_start_after_idle",
1119 		.data		= &init_net.ipv4.sysctl_tcp_slow_start_after_idle,
1120 		.maxlen		= sizeof(u8),
1121 		.mode		= 0644,
1122 		.proc_handler	= proc_dou8vec_minmax,
1123 	},
1124 	{
1125 		.procname	= "tcp_retrans_collapse",
1126 		.data		= &init_net.ipv4.sysctl_tcp_retrans_collapse,
1127 		.maxlen		= sizeof(u8),
1128 		.mode		= 0644,
1129 		.proc_handler	= proc_dou8vec_minmax,
1130 	},
1131 	{
1132 		.procname	= "tcp_stdurg",
1133 		.data		= &init_net.ipv4.sysctl_tcp_stdurg,
1134 		.maxlen		= sizeof(u8),
1135 		.mode		= 0644,
1136 		.proc_handler	= proc_dou8vec_minmax,
1137 	},
1138 	{
1139 		.procname	= "tcp_rfc1337",
1140 		.data		= &init_net.ipv4.sysctl_tcp_rfc1337,
1141 		.maxlen		= sizeof(u8),
1142 		.mode		= 0644,
1143 		.proc_handler	= proc_dou8vec_minmax,
1144 	},
1145 	{
1146 		.procname	= "tcp_abort_on_overflow",
1147 		.data		= &init_net.ipv4.sysctl_tcp_abort_on_overflow,
1148 		.maxlen		= sizeof(u8),
1149 		.mode		= 0644,
1150 		.proc_handler	= proc_dou8vec_minmax,
1151 	},
1152 	{
1153 		.procname	= "tcp_fack",
1154 		.data		= &init_net.ipv4.sysctl_tcp_fack,
1155 		.maxlen		= sizeof(u8),
1156 		.mode		= 0644,
1157 		.proc_handler	= proc_dou8vec_minmax,
1158 	},
1159 	{
1160 		.procname	= "tcp_max_reordering",
1161 		.data		= &init_net.ipv4.sysctl_tcp_max_reordering,
1162 		.maxlen		= sizeof(int),
1163 		.mode		= 0644,
1164 		.proc_handler	= proc_dointvec
1165 	},
1166 	{
1167 		.procname	= "tcp_dsack",
1168 		.data		= &init_net.ipv4.sysctl_tcp_dsack,
1169 		.maxlen		= sizeof(u8),
1170 		.mode		= 0644,
1171 		.proc_handler	= proc_dou8vec_minmax,
1172 	},
1173 	{
1174 		.procname	= "tcp_app_win",
1175 		.data		= &init_net.ipv4.sysctl_tcp_app_win,
1176 		.maxlen		= sizeof(u8),
1177 		.mode		= 0644,
1178 		.proc_handler	= proc_dou8vec_minmax,
1179 		.extra1		= SYSCTL_ZERO,
1180 		.extra2		= &tcp_app_win_max,
1181 	},
1182 	{
1183 		.procname	= "tcp_adv_win_scale",
1184 		.data		= &init_net.ipv4.sysctl_tcp_adv_win_scale,
1185 		.maxlen		= sizeof(int),
1186 		.mode		= 0644,
1187 		.proc_handler	= proc_dointvec_minmax,
1188 		.extra1		= &tcp_adv_win_scale_min,
1189 		.extra2		= &tcp_adv_win_scale_max,
1190 	},
1191 	{
1192 		.procname	= "tcp_frto",
1193 		.data		= &init_net.ipv4.sysctl_tcp_frto,
1194 		.maxlen		= sizeof(u8),
1195 		.mode		= 0644,
1196 		.proc_handler	= proc_dou8vec_minmax,
1197 	},
1198 	{
1199 		.procname	= "tcp_no_metrics_save",
1200 		.data		= &init_net.ipv4.sysctl_tcp_nometrics_save,
1201 		.maxlen		= sizeof(u8),
1202 		.mode		= 0644,
1203 		.proc_handler	= proc_dou8vec_minmax,
1204 	},
1205 	{
1206 		.procname	= "tcp_no_ssthresh_metrics_save",
1207 		.data		= &init_net.ipv4.sysctl_tcp_no_ssthresh_metrics_save,
1208 		.maxlen		= sizeof(u8),
1209 		.mode		= 0644,
1210 		.proc_handler	= proc_dou8vec_minmax,
1211 		.extra1		= SYSCTL_ZERO,
1212 		.extra2		= SYSCTL_ONE,
1213 	},
1214 	{
1215 		.procname	= "tcp_moderate_rcvbuf",
1216 		.data		= &init_net.ipv4.sysctl_tcp_moderate_rcvbuf,
1217 		.maxlen		= sizeof(u8),
1218 		.mode		= 0644,
1219 		.proc_handler	= proc_dou8vec_minmax,
1220 	},
1221 	{
1222 		.procname	= "tcp_tso_win_divisor",
1223 		.data		= &init_net.ipv4.sysctl_tcp_tso_win_divisor,
1224 		.maxlen		= sizeof(u8),
1225 		.mode		= 0644,
1226 		.proc_handler	= proc_dou8vec_minmax,
1227 	},
1228 	{
1229 		.procname	= "tcp_workaround_signed_windows",
1230 		.data		= &init_net.ipv4.sysctl_tcp_workaround_signed_windows,
1231 		.maxlen		= sizeof(u8),
1232 		.mode		= 0644,
1233 		.proc_handler	= proc_dou8vec_minmax,
1234 	},
1235 	{
1236 		.procname	= "tcp_limit_output_bytes",
1237 		.data		= &init_net.ipv4.sysctl_tcp_limit_output_bytes,
1238 		.maxlen		= sizeof(int),
1239 		.mode		= 0644,
1240 		.proc_handler	= proc_dointvec
1241 	},
1242 	{
1243 		.procname	= "tcp_challenge_ack_limit",
1244 		.data		= &init_net.ipv4.sysctl_tcp_challenge_ack_limit,
1245 		.maxlen		= sizeof(int),
1246 		.mode		= 0644,
1247 		.proc_handler	= proc_dointvec
1248 	},
1249 	{
1250 		.procname	= "tcp_min_tso_segs",
1251 		.data		= &init_net.ipv4.sysctl_tcp_min_tso_segs,
1252 		.maxlen		= sizeof(u8),
1253 		.mode		= 0644,
1254 		.proc_handler	= proc_dou8vec_minmax,
1255 		.extra1		= SYSCTL_ONE,
1256 	},
1257 	{
1258 		.procname	= "tcp_min_rtt_wlen",
1259 		.data		= &init_net.ipv4.sysctl_tcp_min_rtt_wlen,
1260 		.maxlen		= sizeof(int),
1261 		.mode		= 0644,
1262 		.proc_handler	= proc_dointvec_minmax,
1263 		.extra1		= SYSCTL_ZERO,
1264 		.extra2		= &one_day_secs
1265 	},
1266 	{
1267 		.procname	= "tcp_autocorking",
1268 		.data		= &init_net.ipv4.sysctl_tcp_autocorking,
1269 		.maxlen		= sizeof(u8),
1270 		.mode		= 0644,
1271 		.proc_handler	= proc_dou8vec_minmax,
1272 		.extra1		= SYSCTL_ZERO,
1273 		.extra2		= SYSCTL_ONE,
1274 	},
1275 	{
1276 		.procname	= "tcp_invalid_ratelimit",
1277 		.data		= &init_net.ipv4.sysctl_tcp_invalid_ratelimit,
1278 		.maxlen		= sizeof(int),
1279 		.mode		= 0644,
1280 		.proc_handler	= proc_dointvec_ms_jiffies,
1281 	},
1282 	{
1283 		.procname	= "tcp_pacing_ss_ratio",
1284 		.data		= &init_net.ipv4.sysctl_tcp_pacing_ss_ratio,
1285 		.maxlen		= sizeof(int),
1286 		.mode		= 0644,
1287 		.proc_handler	= proc_dointvec_minmax,
1288 		.extra1		= SYSCTL_ZERO,
1289 		.extra2		= &thousand,
1290 	},
1291 	{
1292 		.procname	= "tcp_pacing_ca_ratio",
1293 		.data		= &init_net.ipv4.sysctl_tcp_pacing_ca_ratio,
1294 		.maxlen		= sizeof(int),
1295 		.mode		= 0644,
1296 		.proc_handler	= proc_dointvec_minmax,
1297 		.extra1		= SYSCTL_ZERO,
1298 		.extra2		= &thousand,
1299 	},
1300 	{
1301 		.procname	= "tcp_wmem",
1302 		.data		= &init_net.ipv4.sysctl_tcp_wmem,
1303 		.maxlen		= sizeof(init_net.ipv4.sysctl_tcp_wmem),
1304 		.mode		= 0644,
1305 		.proc_handler	= proc_dointvec_minmax,
1306 		.extra1		= SYSCTL_ONE,
1307 	},
1308 	{
1309 		.procname	= "tcp_rmem",
1310 		.data		= &init_net.ipv4.sysctl_tcp_rmem,
1311 		.maxlen		= sizeof(init_net.ipv4.sysctl_tcp_rmem),
1312 		.mode		= 0644,
1313 		.proc_handler	= proc_dointvec_minmax,
1314 		.extra1		= SYSCTL_ONE,
1315 	},
1316 	{
1317 		.procname	= "tcp_comp_sack_delay_ns",
1318 		.data		= &init_net.ipv4.sysctl_tcp_comp_sack_delay_ns,
1319 		.maxlen		= sizeof(unsigned long),
1320 		.mode		= 0644,
1321 		.proc_handler	= proc_doulongvec_minmax,
1322 	},
1323 	{
1324 		.procname	= "tcp_comp_sack_slack_ns",
1325 		.data		= &init_net.ipv4.sysctl_tcp_comp_sack_slack_ns,
1326 		.maxlen		= sizeof(unsigned long),
1327 		.mode		= 0644,
1328 		.proc_handler	= proc_doulongvec_minmax,
1329 	},
1330 	{
1331 		.procname	= "tcp_comp_sack_nr",
1332 		.data		= &init_net.ipv4.sysctl_tcp_comp_sack_nr,
1333 		.maxlen		= sizeof(u8),
1334 		.mode		= 0644,
1335 		.proc_handler	= proc_dou8vec_minmax,
1336 		.extra1		= SYSCTL_ZERO,
1337 	},
1338 	{
1339 		.procname       = "tcp_reflect_tos",
1340 		.data           = &init_net.ipv4.sysctl_tcp_reflect_tos,
1341 		.maxlen         = sizeof(u8),
1342 		.mode           = 0644,
1343 		.proc_handler   = proc_dou8vec_minmax,
1344 		.extra1         = SYSCTL_ZERO,
1345 		.extra2         = SYSCTL_ONE,
1346 	},
1347 	{
1348 		.procname	= "udp_rmem_min",
1349 		.data		= &init_net.ipv4.sysctl_udp_rmem_min,
1350 		.maxlen		= sizeof(init_net.ipv4.sysctl_udp_rmem_min),
1351 		.mode		= 0644,
1352 		.proc_handler	= proc_dointvec_minmax,
1353 		.extra1		= SYSCTL_ONE
1354 	},
1355 	{
1356 		.procname	= "udp_wmem_min",
1357 		.data		= &init_net.ipv4.sysctl_udp_wmem_min,
1358 		.maxlen		= sizeof(init_net.ipv4.sysctl_udp_wmem_min),
1359 		.mode		= 0644,
1360 		.proc_handler	= proc_dointvec_minmax,
1361 		.extra1		= SYSCTL_ONE
1362 	},
1363 	{
1364 		.procname	= "fib_notify_on_flag_change",
1365 		.data		= &init_net.ipv4.sysctl_fib_notify_on_flag_change,
1366 		.maxlen		= sizeof(u8),
1367 		.mode		= 0644,
1368 		.proc_handler	= proc_dou8vec_minmax,
1369 		.extra1		= SYSCTL_ZERO,
1370 		.extra2		= &two,
1371 	},
1372 	{ }
1373 };
1374 
ipv4_sysctl_init_net(struct net * net)1375 static __net_init int ipv4_sysctl_init_net(struct net *net)
1376 {
1377 	struct ctl_table *table;
1378 
1379 	table = ipv4_net_table;
1380 	if (!net_eq(net, &init_net)) {
1381 		int i;
1382 
1383 		table = kmemdup(table, sizeof(ipv4_net_table), GFP_KERNEL);
1384 		if (!table)
1385 			goto err_alloc;
1386 
1387 		for (i = 0; i < ARRAY_SIZE(ipv4_net_table) - 1; i++) {
1388 			if (table[i].data) {
1389 				/* Update the variables to point into
1390 				 * the current struct net
1391 				 */
1392 				table[i].data += (void *)net - (void *)&init_net;
1393 			} else {
1394 				/* Entries without data pointer are global;
1395 				 * Make them read-only in non-init_net ns
1396 				 */
1397 				table[i].mode &= ~0222;
1398 			}
1399 		}
1400 	}
1401 
1402 	net->ipv4.ipv4_hdr = register_net_sysctl(net, "net/ipv4", table);
1403 	if (!net->ipv4.ipv4_hdr)
1404 		goto err_reg;
1405 
1406 	net->ipv4.sysctl_local_reserved_ports = kzalloc(65536 / 8, GFP_KERNEL);
1407 	if (!net->ipv4.sysctl_local_reserved_ports)
1408 		goto err_reserved_ports;
1409 
1410 	net->ipv4.sysctl_local_unbindable_ports = kzalloc(65536 / 8, GFP_KERNEL);
1411 	if (!net->ipv4.sysctl_local_unbindable_ports)
1412 		goto err_unbindable_ports;
1413 
1414 	return 0;
1415 
1416 err_unbindable_ports:
1417 	kfree(net->ipv4.sysctl_local_reserved_ports);
1418 err_reserved_ports:
1419 	unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
1420 err_reg:
1421 	if (!net_eq(net, &init_net))
1422 		kfree(table);
1423 err_alloc:
1424 	return -ENOMEM;
1425 }
1426 
ipv4_sysctl_exit_net(struct net * net)1427 static __net_exit void ipv4_sysctl_exit_net(struct net *net)
1428 {
1429 	struct ctl_table *table;
1430 
1431 	kfree(net->ipv4.sysctl_local_unbindable_ports);
1432 	kfree(net->ipv4.sysctl_local_reserved_ports);
1433 	table = net->ipv4.ipv4_hdr->ctl_table_arg;
1434 	unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
1435 	kfree(table);
1436 }
1437 
1438 static __net_initdata struct pernet_operations ipv4_sysctl_ops = {
1439 	.init = ipv4_sysctl_init_net,
1440 	.exit = ipv4_sysctl_exit_net,
1441 };
1442 
sysctl_ipv4_init(void)1443 static __init int sysctl_ipv4_init(void)
1444 {
1445 	struct ctl_table_header *hdr;
1446 
1447 	hdr = register_net_sysctl(&init_net, "net/ipv4", ipv4_table);
1448 	if (!hdr)
1449 		return -ENOMEM;
1450 
1451 	if (register_pernet_subsys(&ipv4_sysctl_ops)) {
1452 		unregister_net_sysctl_table(hdr);
1453 		return -ENOMEM;
1454 	}
1455 
1456 	return 0;
1457 }
1458 
1459 __initcall(sysctl_ipv4_init);
1460