• 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/route.h>
23 #include <net/tcp.h>
24 #include <net/udp.h>
25 #include <net/cipso_ipv4.h>
26 #include <net/inet_frag.h>
27 #include <net/ping.h>
28 #include <net/protocol.h>
29 #include <net/netevent.h>
30 
31 static int two = 2;
32 static int four = 4;
33 static int thousand = 1000;
34 static int tcp_retr1_max = 255;
35 static int ip_local_port_range_min[] = { 1, 1 };
36 static int ip_local_port_range_max[] = { 65535, 65535 };
37 static int tcp_adv_win_scale_min = -31;
38 static int tcp_adv_win_scale_max = 31;
39 static int tcp_app_win_max = 31;
40 static int tcp_min_snd_mss_min = TCP_MIN_SND_MSS;
41 static int tcp_min_snd_mss_max = 65535;
42 static int ip_privileged_port_min;
43 static int ip_privileged_port_max = 65535;
44 static int ip_ttl_min = 1;
45 static int ip_ttl_max = 255;
46 static int tcp_syn_retries_min = 1;
47 static int tcp_syn_retries_max = MAX_TCP_SYNCNT;
48 static int ip_ping_group_range_min[] = { 0, 0 };
49 static int ip_ping_group_range_max[] = { GID_T_MAX, GID_T_MAX };
50 static int comp_sack_nr_max = 255;
51 static u32 u32_max_div_HZ = UINT_MAX / HZ;
52 static int one_day_secs = 24 * 3600;
53 
54 /* obsolete */
55 static int sysctl_tcp_low_latency __read_mostly;
56 
57 /* Update system visible IP port range */
set_local_port_range(struct net * net,int range[2])58 static void set_local_port_range(struct net *net, int range[2])
59 {
60 	bool same_parity = !((range[0] ^ range[1]) & 1);
61 
62 	write_seqlock_bh(&net->ipv4.ip_local_ports.lock);
63 	if (same_parity && !net->ipv4.ip_local_ports.warned) {
64 		net->ipv4.ip_local_ports.warned = true;
65 		pr_err_ratelimited("ip_local_port_range: prefer different parity for start/end values.\n");
66 	}
67 	net->ipv4.ip_local_ports.range[0] = range[0];
68 	net->ipv4.ip_local_ports.range[1] = range[1];
69 	write_sequnlock_bh(&net->ipv4.ip_local_ports.lock);
70 }
71 
72 /* Validate changes from /proc interface. */
ipv4_local_port_range(struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)73 static int ipv4_local_port_range(struct ctl_table *table, int write,
74 				 void *buffer, size_t *lenp, loff_t *ppos)
75 {
76 	struct net *net =
77 		container_of(table->data, struct net, ipv4.ip_local_ports.range);
78 	int ret;
79 	int range[2];
80 	struct ctl_table tmp = {
81 		.data = &range,
82 		.maxlen = sizeof(range),
83 		.mode = table->mode,
84 		.extra1 = &ip_local_port_range_min,
85 		.extra2 = &ip_local_port_range_max,
86 	};
87 
88 	inet_get_local_port_range(net, &range[0], &range[1]);
89 
90 	ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
91 
92 	if (write && ret == 0) {
93 		/* Ensure that the upper limit is not smaller than the lower,
94 		 * and that the lower does not encroach upon the privileged
95 		 * port limit.
96 		 */
97 		if ((range[1] < range[0]) ||
98 		    (range[0] < READ_ONCE(net->ipv4.sysctl_ip_prot_sock)))
99 			ret = -EINVAL;
100 		else
101 			set_local_port_range(net, range);
102 	}
103 
104 	return ret;
105 }
106 
107 /* Validate changes from /proc interface. */
ipv4_privileged_ports(struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)108 static int ipv4_privileged_ports(struct ctl_table *table, int write,
109 				void *buffer, size_t *lenp, loff_t *ppos)
110 {
111 	struct net *net = container_of(table->data, struct net,
112 	    ipv4.sysctl_ip_prot_sock);
113 	int ret;
114 	int pports;
115 	int range[2];
116 	struct ctl_table tmp = {
117 		.data = &pports,
118 		.maxlen = sizeof(pports),
119 		.mode = table->mode,
120 		.extra1 = &ip_privileged_port_min,
121 		.extra2 = &ip_privileged_port_max,
122 	};
123 
124 	pports = READ_ONCE(net->ipv4.sysctl_ip_prot_sock);
125 
126 	ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
127 
128 	if (write && ret == 0) {
129 		inet_get_local_port_range(net, &range[0], &range[1]);
130 		/* Ensure that the local port range doesn't overlap with the
131 		 * privileged port range.
132 		 */
133 		if (range[0] < pports)
134 			ret = -EINVAL;
135 		else
136 			WRITE_ONCE(net->ipv4.sysctl_ip_prot_sock, pports);
137 	}
138 
139 	return ret;
140 }
141 
inet_get_ping_group_range_table(struct ctl_table * table,kgid_t * low,kgid_t * high)142 static void inet_get_ping_group_range_table(struct ctl_table *table, kgid_t *low, kgid_t *high)
143 {
144 	kgid_t *data = table->data;
145 	struct net *net =
146 		container_of(table->data, struct net, ipv4.ping_group_range.range);
147 	unsigned int seq;
148 	do {
149 		seq = read_seqbegin(&net->ipv4.ping_group_range.lock);
150 
151 		*low = data[0];
152 		*high = data[1];
153 	} while (read_seqretry(&net->ipv4.ping_group_range.lock, seq));
154 }
155 
156 /* Update system visible IP port range */
set_ping_group_range(struct ctl_table * table,kgid_t low,kgid_t high)157 static void set_ping_group_range(struct ctl_table *table, kgid_t low, kgid_t high)
158 {
159 	kgid_t *data = table->data;
160 	struct net *net =
161 		container_of(table->data, struct net, ipv4.ping_group_range.range);
162 	write_seqlock(&net->ipv4.ping_group_range.lock);
163 	data[0] = low;
164 	data[1] = high;
165 	write_sequnlock(&net->ipv4.ping_group_range.lock);
166 }
167 
168 /* Validate changes from /proc interface. */
ipv4_ping_group_range(struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)169 static int ipv4_ping_group_range(struct ctl_table *table, int write,
170 				 void *buffer, size_t *lenp, loff_t *ppos)
171 {
172 	struct user_namespace *user_ns = current_user_ns();
173 	int ret;
174 	gid_t urange[2];
175 	kgid_t low, high;
176 	struct ctl_table tmp = {
177 		.data = &urange,
178 		.maxlen = sizeof(urange),
179 		.mode = table->mode,
180 		.extra1 = &ip_ping_group_range_min,
181 		.extra2 = &ip_ping_group_range_max,
182 	};
183 
184 	inet_get_ping_group_range_table(table, &low, &high);
185 	urange[0] = from_kgid_munged(user_ns, low);
186 	urange[1] = from_kgid_munged(user_ns, high);
187 	ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
188 
189 	if (write && ret == 0) {
190 		low = make_kgid(user_ns, urange[0]);
191 		high = make_kgid(user_ns, urange[1]);
192 		if (!gid_valid(low) || !gid_valid(high))
193 			return -EINVAL;
194 		if (urange[1] < urange[0] || gid_lt(high, low)) {
195 			low = make_kgid(&init_user_ns, 1);
196 			high = make_kgid(&init_user_ns, 0);
197 		}
198 		set_ping_group_range(table, low, high);
199 	}
200 
201 	return ret;
202 }
203 
ipv4_fwd_update_priority(struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)204 static int ipv4_fwd_update_priority(struct ctl_table *table, int write,
205 				    void *buffer, size_t *lenp, loff_t *ppos)
206 {
207 	struct net *net;
208 	int ret;
209 
210 	net = container_of(table->data, struct net,
211 			   ipv4.sysctl_ip_fwd_update_priority);
212 	ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
213 	if (write && ret == 0)
214 		call_netevent_notifiers(NETEVENT_IPV4_FWD_UPDATE_PRIORITY_UPDATE,
215 					net);
216 
217 	return ret;
218 }
219 
proc_tcp_congestion_control(struct ctl_table * ctl,int write,void * buffer,size_t * lenp,loff_t * ppos)220 static int proc_tcp_congestion_control(struct ctl_table *ctl, int write,
221 				       void *buffer, size_t *lenp, loff_t *ppos)
222 {
223 	struct net *net = container_of(ctl->data, struct net,
224 				       ipv4.tcp_congestion_control);
225 	char val[TCP_CA_NAME_MAX];
226 	struct ctl_table tbl = {
227 		.data = val,
228 		.maxlen = TCP_CA_NAME_MAX,
229 	};
230 	int ret;
231 
232 	tcp_get_default_congestion_control(net, val);
233 
234 	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
235 	if (write && ret == 0)
236 		ret = tcp_set_default_congestion_control(net, val);
237 	return ret;
238 }
239 
proc_tcp_available_congestion_control(struct ctl_table * ctl,int write,void * buffer,size_t * lenp,loff_t * ppos)240 static int proc_tcp_available_congestion_control(struct ctl_table *ctl,
241 						 int write, void *buffer,
242 						 size_t *lenp, loff_t *ppos)
243 {
244 	struct ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX, };
245 	int ret;
246 
247 	tbl.data = kmalloc(tbl.maxlen, GFP_USER);
248 	if (!tbl.data)
249 		return -ENOMEM;
250 	tcp_get_available_congestion_control(tbl.data, TCP_CA_BUF_MAX);
251 	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
252 	kfree(tbl.data);
253 	return ret;
254 }
255 
proc_allowed_congestion_control(struct ctl_table * ctl,int write,void * buffer,size_t * lenp,loff_t * ppos)256 static int proc_allowed_congestion_control(struct ctl_table *ctl,
257 					   int write, void *buffer,
258 					   size_t *lenp, loff_t *ppos)
259 {
260 	struct ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX };
261 	int ret;
262 
263 	tbl.data = kmalloc(tbl.maxlen, GFP_USER);
264 	if (!tbl.data)
265 		return -ENOMEM;
266 
267 	tcp_get_allowed_congestion_control(tbl.data, tbl.maxlen);
268 	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
269 	if (write && ret == 0)
270 		ret = tcp_set_allowed_congestion_control(tbl.data);
271 	kfree(tbl.data);
272 	return ret;
273 }
274 
sscanf_key(char * buf,__le32 * key)275 static int sscanf_key(char *buf, __le32 *key)
276 {
277 	u32 user_key[4];
278 	int i, ret = 0;
279 
280 	if (sscanf(buf, "%x-%x-%x-%x", user_key, user_key + 1,
281 		   user_key + 2, user_key + 3) != 4) {
282 		ret = -EINVAL;
283 	} else {
284 		for (i = 0; i < ARRAY_SIZE(user_key); i++)
285 			key[i] = cpu_to_le32(user_key[i]);
286 	}
287 	pr_debug("proc TFO key set 0x%x-%x-%x-%x <- 0x%s: %u\n",
288 		 user_key[0], user_key[1], user_key[2], user_key[3], buf, ret);
289 
290 	return ret;
291 }
292 
proc_tcp_fastopen_key(struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)293 static int proc_tcp_fastopen_key(struct ctl_table *table, int write,
294 				 void *buffer, size_t *lenp, loff_t *ppos)
295 {
296 	struct net *net = container_of(table->data, struct net,
297 	    ipv4.sysctl_tcp_fastopen);
298 	/* maxlen to print the list of keys in hex (*2), with dashes
299 	 * separating doublewords and a comma in between keys.
300 	 */
301 	struct ctl_table tbl = { .maxlen = ((TCP_FASTOPEN_KEY_LENGTH *
302 					    2 * TCP_FASTOPEN_KEY_MAX) +
303 					    (TCP_FASTOPEN_KEY_MAX * 5)) };
304 	u32 user_key[TCP_FASTOPEN_KEY_BUF_LENGTH / sizeof(u32)];
305 	__le32 key[TCP_FASTOPEN_KEY_BUF_LENGTH / sizeof(__le32)];
306 	char *backup_data;
307 	int ret, i = 0, off = 0, n_keys;
308 
309 	tbl.data = kmalloc(tbl.maxlen, GFP_KERNEL);
310 	if (!tbl.data)
311 		return -ENOMEM;
312 
313 	n_keys = tcp_fastopen_get_cipher(net, NULL, (u64 *)key);
314 	if (!n_keys) {
315 		memset(&key[0], 0, TCP_FASTOPEN_KEY_LENGTH);
316 		n_keys = 1;
317 	}
318 
319 	for (i = 0; i < n_keys * 4; i++)
320 		user_key[i] = le32_to_cpu(key[i]);
321 
322 	for (i = 0; i < n_keys; i++) {
323 		off += snprintf(tbl.data + off, tbl.maxlen - off,
324 				"%08x-%08x-%08x-%08x",
325 				user_key[i * 4],
326 				user_key[i * 4 + 1],
327 				user_key[i * 4 + 2],
328 				user_key[i * 4 + 3]);
329 
330 		if (WARN_ON_ONCE(off >= tbl.maxlen - 1))
331 			break;
332 
333 		if (i + 1 < n_keys)
334 			off += snprintf(tbl.data + off, tbl.maxlen - off, ",");
335 	}
336 
337 	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
338 
339 	if (write && ret == 0) {
340 		backup_data = strchr(tbl.data, ',');
341 		if (backup_data) {
342 			*backup_data = '\0';
343 			backup_data++;
344 		}
345 		if (sscanf_key(tbl.data, key)) {
346 			ret = -EINVAL;
347 			goto bad_key;
348 		}
349 		if (backup_data) {
350 			if (sscanf_key(backup_data, key + 4)) {
351 				ret = -EINVAL;
352 				goto bad_key;
353 			}
354 		}
355 		tcp_fastopen_reset_cipher(net, NULL, key,
356 					  backup_data ? key + 4 : NULL);
357 	}
358 
359 bad_key:
360 	kfree(tbl.data);
361 	return ret;
362 }
363 
proc_tfo_blackhole_detect_timeout(struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)364 static int proc_tfo_blackhole_detect_timeout(struct ctl_table *table,
365 					     int write, void *buffer,
366 					     size_t *lenp, loff_t *ppos)
367 {
368 	struct net *net = container_of(table->data, struct net,
369 	    ipv4.sysctl_tcp_fastopen_blackhole_timeout);
370 	int ret;
371 
372 	ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
373 	if (write && ret == 0)
374 		atomic_set(&net->ipv4.tfo_active_disable_times, 0);
375 
376 	return ret;
377 }
378 
proc_tcp_available_ulp(struct ctl_table * ctl,int write,void * buffer,size_t * lenp,loff_t * ppos)379 static int proc_tcp_available_ulp(struct ctl_table *ctl,
380 				  int write, void *buffer, size_t *lenp,
381 				  loff_t *ppos)
382 {
383 	struct ctl_table tbl = { .maxlen = TCP_ULP_BUF_MAX, };
384 	int ret;
385 
386 	tbl.data = kmalloc(tbl.maxlen, GFP_USER);
387 	if (!tbl.data)
388 		return -ENOMEM;
389 	tcp_get_available_ulp(tbl.data, TCP_ULP_BUF_MAX);
390 	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
391 	kfree(tbl.data);
392 
393 	return ret;
394 }
395 
396 #ifdef CONFIG_IP_ROUTE_MULTIPATH
proc_fib_multipath_hash_policy(struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)397 static int proc_fib_multipath_hash_policy(struct ctl_table *table, int write,
398 					  void *buffer, size_t *lenp,
399 					  loff_t *ppos)
400 {
401 	struct net *net = container_of(table->data, struct net,
402 	    ipv4.sysctl_fib_multipath_hash_policy);
403 	int ret;
404 
405 	ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
406 	if (write && ret == 0)
407 		call_netevent_notifiers(NETEVENT_IPV4_MPATH_HASH_UPDATE, net);
408 
409 	return ret;
410 }
411 #endif
412 
413 static struct ctl_table ipv4_table[] = {
414 	{
415 		.procname	= "tcp_max_orphans",
416 		.data		= &sysctl_tcp_max_orphans,
417 		.maxlen		= sizeof(int),
418 		.mode		= 0644,
419 		.proc_handler	= proc_dointvec
420 	},
421 	{
422 		.procname	= "inet_peer_threshold",
423 		.data		= &inet_peer_threshold,
424 		.maxlen		= sizeof(int),
425 		.mode		= 0644,
426 		.proc_handler	= proc_dointvec
427 	},
428 	{
429 		.procname	= "inet_peer_minttl",
430 		.data		= &inet_peer_minttl,
431 		.maxlen		= sizeof(int),
432 		.mode		= 0644,
433 		.proc_handler	= proc_dointvec_jiffies,
434 	},
435 	{
436 		.procname	= "inet_peer_maxttl",
437 		.data		= &inet_peer_maxttl,
438 		.maxlen		= sizeof(int),
439 		.mode		= 0644,
440 		.proc_handler	= proc_dointvec_jiffies,
441 	},
442 	{
443 		.procname	= "tcp_mem",
444 		.maxlen		= sizeof(sysctl_tcp_mem),
445 		.data		= &sysctl_tcp_mem,
446 		.mode		= 0644,
447 		.proc_handler	= proc_doulongvec_minmax,
448 	},
449 	{
450 		.procname	= "tcp_low_latency",
451 		.data		= &sysctl_tcp_low_latency,
452 		.maxlen		= sizeof(int),
453 		.mode		= 0644,
454 		.proc_handler	= proc_dointvec
455 	},
456 #ifdef CONFIG_NETLABEL
457 	{
458 		.procname	= "cipso_cache_enable",
459 		.data		= &cipso_v4_cache_enabled,
460 		.maxlen		= sizeof(int),
461 		.mode		= 0644,
462 		.proc_handler	= proc_dointvec,
463 	},
464 	{
465 		.procname	= "cipso_cache_bucket_size",
466 		.data		= &cipso_v4_cache_bucketsize,
467 		.maxlen		= sizeof(int),
468 		.mode		= 0644,
469 		.proc_handler	= proc_dointvec,
470 	},
471 	{
472 		.procname	= "cipso_rbm_optfmt",
473 		.data		= &cipso_v4_rbm_optfmt,
474 		.maxlen		= sizeof(int),
475 		.mode		= 0644,
476 		.proc_handler	= proc_dointvec,
477 	},
478 	{
479 		.procname	= "cipso_rbm_strictvalid",
480 		.data		= &cipso_v4_rbm_strictvalid,
481 		.maxlen		= sizeof(int),
482 		.mode		= 0644,
483 		.proc_handler	= proc_dointvec,
484 	},
485 #endif /* CONFIG_NETLABEL */
486 	{
487 		.procname	= "tcp_available_ulp",
488 		.maxlen		= TCP_ULP_BUF_MAX,
489 		.mode		= 0444,
490 		.proc_handler   = proc_tcp_available_ulp,
491 	},
492 	{
493 		.procname	= "icmp_msgs_per_sec",
494 		.data		= &sysctl_icmp_msgs_per_sec,
495 		.maxlen		= sizeof(int),
496 		.mode		= 0644,
497 		.proc_handler	= proc_dointvec_minmax,
498 		.extra1		= SYSCTL_ZERO,
499 	},
500 	{
501 		.procname	= "icmp_msgs_burst",
502 		.data		= &sysctl_icmp_msgs_burst,
503 		.maxlen		= sizeof(int),
504 		.mode		= 0644,
505 		.proc_handler	= proc_dointvec_minmax,
506 		.extra1		= SYSCTL_ZERO,
507 	},
508 	{
509 		.procname	= "udp_mem",
510 		.data		= &sysctl_udp_mem,
511 		.maxlen		= sizeof(sysctl_udp_mem),
512 		.mode		= 0644,
513 		.proc_handler	= proc_doulongvec_minmax,
514 	},
515 	{
516 		.procname	= "fib_sync_mem",
517 		.data		= &sysctl_fib_sync_mem,
518 		.maxlen		= sizeof(sysctl_fib_sync_mem),
519 		.mode		= 0644,
520 		.proc_handler	= proc_douintvec_minmax,
521 		.extra1		= &sysctl_fib_sync_mem_min,
522 		.extra2		= &sysctl_fib_sync_mem_max,
523 	},
524 	{
525 		.procname	= "tcp_rx_skb_cache",
526 		.data		= &tcp_rx_skb_cache_key.key,
527 		.mode		= 0644,
528 		.proc_handler	= proc_do_static_key,
529 	},
530 	{
531 		.procname	= "tcp_tx_skb_cache",
532 		.data		= &tcp_tx_skb_cache_key.key,
533 		.mode		= 0644,
534 		.proc_handler	= proc_do_static_key,
535 	},
536 	{ }
537 };
538 
539 static struct ctl_table ipv4_net_table[] = {
540 	{
541 		.procname	= "icmp_echo_ignore_all",
542 		.data		= &init_net.ipv4.sysctl_icmp_echo_ignore_all,
543 		.maxlen		= sizeof(u8),
544 		.mode		= 0644,
545 		.proc_handler	= proc_dou8vec_minmax,
546 	},
547 	{
548 		.procname	= "icmp_echo_ignore_broadcasts",
549 		.data		= &init_net.ipv4.sysctl_icmp_echo_ignore_broadcasts,
550 		.maxlen		= sizeof(u8),
551 		.mode		= 0644,
552 		.proc_handler	= proc_dou8vec_minmax,
553 	},
554 	{
555 		.procname	= "icmp_ignore_bogus_error_responses",
556 		.data		= &init_net.ipv4.sysctl_icmp_ignore_bogus_error_responses,
557 		.maxlen		= sizeof(u8),
558 		.mode		= 0644,
559 		.proc_handler	= proc_dou8vec_minmax,
560 	},
561 	{
562 		.procname	= "icmp_errors_use_inbound_ifaddr",
563 		.data		= &init_net.ipv4.sysctl_icmp_errors_use_inbound_ifaddr,
564 		.maxlen		= sizeof(u8),
565 		.mode		= 0644,
566 		.proc_handler	= proc_dou8vec_minmax,
567 		.extra1		= SYSCTL_ZERO,
568 		.extra2		= SYSCTL_ONE
569 	},
570 	{
571 		.procname	= "icmp_ratelimit",
572 		.data		= &init_net.ipv4.sysctl_icmp_ratelimit,
573 		.maxlen		= sizeof(int),
574 		.mode		= 0644,
575 		.proc_handler	= proc_dointvec_ms_jiffies,
576 	},
577 	{
578 		.procname	= "icmp_ratemask",
579 		.data		= &init_net.ipv4.sysctl_icmp_ratemask,
580 		.maxlen		= sizeof(int),
581 		.mode		= 0644,
582 		.proc_handler	= proc_dointvec
583 	},
584 	{
585 		.procname	= "ping_group_range",
586 		.data		= &init_net.ipv4.ping_group_range.range,
587 		.maxlen		= sizeof(gid_t)*2,
588 		.mode		= 0644,
589 		.proc_handler	= ipv4_ping_group_range,
590 	},
591 #ifdef CONFIG_NET_L3_MASTER_DEV
592 	{
593 		.procname	= "raw_l3mdev_accept",
594 		.data		= &init_net.ipv4.sysctl_raw_l3mdev_accept,
595 		.maxlen		= sizeof(u8),
596 		.mode		= 0644,
597 		.proc_handler	= proc_dou8vec_minmax,
598 		.extra1		= SYSCTL_ZERO,
599 		.extra2		= SYSCTL_ONE,
600 	},
601 #endif
602 	{
603 		.procname	= "tcp_ecn",
604 		.data		= &init_net.ipv4.sysctl_tcp_ecn,
605 		.maxlen		= sizeof(u8),
606 		.mode		= 0644,
607 		.proc_handler	= proc_dou8vec_minmax,
608 	},
609 	{
610 		.procname	= "tcp_ecn_fallback",
611 		.data		= &init_net.ipv4.sysctl_tcp_ecn_fallback,
612 		.maxlen		= sizeof(u8),
613 		.mode		= 0644,
614 		.proc_handler	= proc_dou8vec_minmax,
615 		.extra1		= SYSCTL_ZERO,
616 		.extra2		= SYSCTL_ONE,
617 	},
618 	{
619 		.procname	= "ip_dynaddr",
620 		.data		= &init_net.ipv4.sysctl_ip_dynaddr,
621 		.maxlen		= sizeof(u8),
622 		.mode		= 0644,
623 		.proc_handler	= proc_dou8vec_minmax,
624 	},
625 	{
626 		.procname	= "ip_early_demux",
627 		.data		= &init_net.ipv4.sysctl_ip_early_demux,
628 		.maxlen		= sizeof(u8),
629 		.mode		= 0644,
630 		.proc_handler	= proc_dou8vec_minmax,
631 	},
632 	{
633 		.procname       = "udp_early_demux",
634 		.data           = &init_net.ipv4.sysctl_udp_early_demux,
635 		.maxlen         = sizeof(int),
636 		.mode           = 0644,
637 		.proc_handler   = proc_douintvec_minmax,
638 	},
639 	{
640 		.procname       = "tcp_early_demux",
641 		.data           = &init_net.ipv4.sysctl_tcp_early_demux,
642 		.maxlen         = sizeof(int),
643 		.mode           = 0644,
644 		.proc_handler   = proc_douintvec_minmax,
645 	},
646 	{
647 		.procname       = "nexthop_compat_mode",
648 		.data           = &init_net.ipv4.sysctl_nexthop_compat_mode,
649 		.maxlen         = sizeof(u8),
650 		.mode           = 0644,
651 		.proc_handler   = proc_dou8vec_minmax,
652 		.extra1		= SYSCTL_ZERO,
653 		.extra2		= SYSCTL_ONE,
654 	},
655 	{
656 		.procname	= "ip_default_ttl",
657 		.data		= &init_net.ipv4.sysctl_ip_default_ttl,
658 		.maxlen		= sizeof(u8),
659 		.mode		= 0644,
660 		.proc_handler	= proc_dou8vec_minmax,
661 		.extra1		= &ip_ttl_min,
662 		.extra2		= &ip_ttl_max,
663 	},
664 	{
665 		.procname	= "ip_local_port_range",
666 		.maxlen		= sizeof(init_net.ipv4.ip_local_ports.range),
667 		.data		= &init_net.ipv4.ip_local_ports.range,
668 		.mode		= 0644,
669 		.proc_handler	= ipv4_local_port_range,
670 	},
671 	{
672 		.procname	= "ip_local_reserved_ports",
673 		.data		= &init_net.ipv4.sysctl_local_reserved_ports,
674 		.maxlen		= 65536,
675 		.mode		= 0644,
676 		.proc_handler	= proc_do_large_bitmap,
677 	},
678 	{
679 		.procname	= "ip_no_pmtu_disc",
680 		.data		= &init_net.ipv4.sysctl_ip_no_pmtu_disc,
681 		.maxlen		= sizeof(u8),
682 		.mode		= 0644,
683 		.proc_handler	= proc_dou8vec_minmax,
684 	},
685 	{
686 		.procname	= "ip_forward_use_pmtu",
687 		.data		= &init_net.ipv4.sysctl_ip_fwd_use_pmtu,
688 		.maxlen		= sizeof(u8),
689 		.mode		= 0644,
690 		.proc_handler	= proc_dou8vec_minmax,
691 	},
692 	{
693 		.procname	= "ip_forward_update_priority",
694 		.data		= &init_net.ipv4.sysctl_ip_fwd_update_priority,
695 		.maxlen		= sizeof(int),
696 		.mode		= 0644,
697 		.proc_handler   = ipv4_fwd_update_priority,
698 		.extra1		= SYSCTL_ZERO,
699 		.extra2		= SYSCTL_ONE,
700 	},
701 	{
702 		.procname	= "ip_nonlocal_bind",
703 		.data		= &init_net.ipv4.sysctl_ip_nonlocal_bind,
704 		.maxlen		= sizeof(u8),
705 		.mode		= 0644,
706 		.proc_handler	= proc_dou8vec_minmax,
707 	},
708 	{
709 		.procname	= "ip_autobind_reuse",
710 		.data		= &init_net.ipv4.sysctl_ip_autobind_reuse,
711 		.maxlen		= sizeof(u8),
712 		.mode		= 0644,
713 		.proc_handler	= proc_dou8vec_minmax,
714 		.extra1         = SYSCTL_ZERO,
715 		.extra2         = SYSCTL_ONE,
716 	},
717 	{
718 		.procname	= "fwmark_reflect",
719 		.data		= &init_net.ipv4.sysctl_fwmark_reflect,
720 		.maxlen		= sizeof(u8),
721 		.mode		= 0644,
722 		.proc_handler	= proc_dou8vec_minmax,
723 	},
724 	{
725 		.procname	= "tcp_fwmark_accept",
726 		.data		= &init_net.ipv4.sysctl_tcp_fwmark_accept,
727 		.maxlen		= sizeof(u8),
728 		.mode		= 0644,
729 		.proc_handler	= proc_dou8vec_minmax,
730 	},
731 #ifdef CONFIG_NET_L3_MASTER_DEV
732 	{
733 		.procname	= "tcp_l3mdev_accept",
734 		.data		= &init_net.ipv4.sysctl_tcp_l3mdev_accept,
735 		.maxlen		= sizeof(u8),
736 		.mode		= 0644,
737 		.proc_handler	= proc_dou8vec_minmax,
738 		.extra1		= SYSCTL_ZERO,
739 		.extra2		= SYSCTL_ONE,
740 	},
741 #endif
742 	{
743 		.procname	= "tcp_mtu_probing",
744 		.data		= &init_net.ipv4.sysctl_tcp_mtu_probing,
745 		.maxlen		= sizeof(u8),
746 		.mode		= 0644,
747 		.proc_handler	= proc_dou8vec_minmax,
748 	},
749 	{
750 		.procname	= "tcp_base_mss",
751 		.data		= &init_net.ipv4.sysctl_tcp_base_mss,
752 		.maxlen		= sizeof(int),
753 		.mode		= 0644,
754 		.proc_handler	= proc_dointvec,
755 	},
756 	{
757 		.procname	= "tcp_min_snd_mss",
758 		.data		= &init_net.ipv4.sysctl_tcp_min_snd_mss,
759 		.maxlen		= sizeof(int),
760 		.mode		= 0644,
761 		.proc_handler	= proc_dointvec_minmax,
762 		.extra1		= &tcp_min_snd_mss_min,
763 		.extra2		= &tcp_min_snd_mss_max,
764 	},
765 	{
766 		.procname	= "tcp_mtu_probe_floor",
767 		.data		= &init_net.ipv4.sysctl_tcp_mtu_probe_floor,
768 		.maxlen		= sizeof(int),
769 		.mode		= 0644,
770 		.proc_handler	= proc_dointvec_minmax,
771 		.extra1		= &tcp_min_snd_mss_min,
772 		.extra2		= &tcp_min_snd_mss_max,
773 	},
774 	{
775 		.procname	= "tcp_probe_threshold",
776 		.data		= &init_net.ipv4.sysctl_tcp_probe_threshold,
777 		.maxlen		= sizeof(int),
778 		.mode		= 0644,
779 		.proc_handler	= proc_dointvec,
780 	},
781 	{
782 		.procname	= "tcp_probe_interval",
783 		.data		= &init_net.ipv4.sysctl_tcp_probe_interval,
784 		.maxlen		= sizeof(u32),
785 		.mode		= 0644,
786 		.proc_handler	= proc_douintvec_minmax,
787 		.extra2		= &u32_max_div_HZ,
788 	},
789 	{
790 		.procname	= "igmp_link_local_mcast_reports",
791 		.data		= &init_net.ipv4.sysctl_igmp_llm_reports,
792 		.maxlen		= sizeof(int),
793 		.mode		= 0644,
794 		.proc_handler	= proc_dointvec
795 	},
796 	{
797 		.procname	= "igmp_max_memberships",
798 		.data		= &init_net.ipv4.sysctl_igmp_max_memberships,
799 		.maxlen		= sizeof(int),
800 		.mode		= 0644,
801 		.proc_handler	= proc_dointvec
802 	},
803 	{
804 		.procname	= "igmp_max_msf",
805 		.data		= &init_net.ipv4.sysctl_igmp_max_msf,
806 		.maxlen		= sizeof(int),
807 		.mode		= 0644,
808 		.proc_handler	= proc_dointvec
809 	},
810 #ifdef CONFIG_IP_MULTICAST
811 	{
812 		.procname	= "igmp_qrv",
813 		.data		= &init_net.ipv4.sysctl_igmp_qrv,
814 		.maxlen		= sizeof(int),
815 		.mode		= 0644,
816 		.proc_handler	= proc_dointvec_minmax,
817 		.extra1		= SYSCTL_ONE
818 	},
819 #endif
820 	{
821 		.procname	= "tcp_congestion_control",
822 		.data		= &init_net.ipv4.tcp_congestion_control,
823 		.mode		= 0644,
824 		.maxlen		= TCP_CA_NAME_MAX,
825 		.proc_handler	= proc_tcp_congestion_control,
826 	},
827 	{
828 		.procname	= "tcp_available_congestion_control",
829 		.maxlen		= TCP_CA_BUF_MAX,
830 		.mode		= 0444,
831 		.proc_handler   = proc_tcp_available_congestion_control,
832 	},
833 	{
834 		.procname	= "tcp_allowed_congestion_control",
835 		.maxlen		= TCP_CA_BUF_MAX,
836 		.mode		= 0644,
837 		.proc_handler   = proc_allowed_congestion_control,
838 	},
839 	{
840 		.procname	= "tcp_keepalive_time",
841 		.data		= &init_net.ipv4.sysctl_tcp_keepalive_time,
842 		.maxlen		= sizeof(int),
843 		.mode		= 0644,
844 		.proc_handler	= proc_dointvec_jiffies,
845 	},
846 	{
847 		.procname	= "tcp_keepalive_probes",
848 		.data		= &init_net.ipv4.sysctl_tcp_keepalive_probes,
849 		.maxlen		= sizeof(u8),
850 		.mode		= 0644,
851 		.proc_handler	= proc_dou8vec_minmax,
852 	},
853 	{
854 		.procname	= "tcp_keepalive_intvl",
855 		.data		= &init_net.ipv4.sysctl_tcp_keepalive_intvl,
856 		.maxlen		= sizeof(int),
857 		.mode		= 0644,
858 		.proc_handler	= proc_dointvec_jiffies,
859 	},
860 	{
861 		.procname	= "tcp_syn_retries",
862 		.data		= &init_net.ipv4.sysctl_tcp_syn_retries,
863 		.maxlen		= sizeof(u8),
864 		.mode		= 0644,
865 		.proc_handler	= proc_dou8vec_minmax,
866 		.extra1		= &tcp_syn_retries_min,
867 		.extra2		= &tcp_syn_retries_max
868 	},
869 	{
870 		.procname	= "tcp_synack_retries",
871 		.data		= &init_net.ipv4.sysctl_tcp_synack_retries,
872 		.maxlen		= sizeof(u8),
873 		.mode		= 0644,
874 		.proc_handler	= proc_dou8vec_minmax,
875 	},
876 #ifdef CONFIG_SYN_COOKIES
877 	{
878 		.procname	= "tcp_syncookies",
879 		.data		= &init_net.ipv4.sysctl_tcp_syncookies,
880 		.maxlen		= sizeof(u8),
881 		.mode		= 0644,
882 		.proc_handler	= proc_dou8vec_minmax,
883 	},
884 #endif
885 	{
886 		.procname	= "tcp_migrate_req",
887 		.data		= &init_net.ipv4.sysctl_tcp_migrate_req,
888 		.maxlen		= sizeof(u8),
889 		.mode		= 0644,
890 		.proc_handler	= proc_dou8vec_minmax,
891 		.extra1		= SYSCTL_ZERO,
892 		.extra2		= SYSCTL_ONE
893 	},
894 	{
895 		.procname	= "tcp_reordering",
896 		.data		= &init_net.ipv4.sysctl_tcp_reordering,
897 		.maxlen		= sizeof(int),
898 		.mode		= 0644,
899 		.proc_handler	= proc_dointvec
900 	},
901 	{
902 		.procname	= "tcp_retries1",
903 		.data		= &init_net.ipv4.sysctl_tcp_retries1,
904 		.maxlen		= sizeof(u8),
905 		.mode		= 0644,
906 		.proc_handler	= proc_dou8vec_minmax,
907 		.extra2		= &tcp_retr1_max
908 	},
909 	{
910 		.procname	= "tcp_retries2",
911 		.data		= &init_net.ipv4.sysctl_tcp_retries2,
912 		.maxlen		= sizeof(u8),
913 		.mode		= 0644,
914 		.proc_handler	= proc_dou8vec_minmax,
915 	},
916 	{
917 		.procname	= "tcp_orphan_retries",
918 		.data		= &init_net.ipv4.sysctl_tcp_orphan_retries,
919 		.maxlen		= sizeof(u8),
920 		.mode		= 0644,
921 		.proc_handler	= proc_dou8vec_minmax,
922 	},
923 	{
924 		.procname	= "tcp_fin_timeout",
925 		.data		= &init_net.ipv4.sysctl_tcp_fin_timeout,
926 		.maxlen		= sizeof(int),
927 		.mode		= 0644,
928 		.proc_handler	= proc_dointvec_jiffies,
929 	},
930 	{
931 		.procname	= "tcp_notsent_lowat",
932 		.data		= &init_net.ipv4.sysctl_tcp_notsent_lowat,
933 		.maxlen		= sizeof(unsigned int),
934 		.mode		= 0644,
935 		.proc_handler	= proc_douintvec,
936 	},
937 	{
938 		.procname	= "tcp_tw_reuse",
939 		.data		= &init_net.ipv4.sysctl_tcp_tw_reuse,
940 		.maxlen		= sizeof(u8),
941 		.mode		= 0644,
942 		.proc_handler	= proc_dou8vec_minmax,
943 		.extra1		= SYSCTL_ZERO,
944 		.extra2		= &two,
945 	},
946 	{
947 		.procname	= "tcp_max_tw_buckets",
948 		.data		= &init_net.ipv4.tcp_death_row.sysctl_max_tw_buckets,
949 		.maxlen		= sizeof(int),
950 		.mode		= 0644,
951 		.proc_handler	= proc_dointvec
952 	},
953 	{
954 		.procname	= "tcp_max_syn_backlog",
955 		.data		= &init_net.ipv4.sysctl_max_syn_backlog,
956 		.maxlen		= sizeof(int),
957 		.mode		= 0644,
958 		.proc_handler	= proc_dointvec
959 	},
960 	{
961 		.procname	= "tcp_fastopen",
962 		.data		= &init_net.ipv4.sysctl_tcp_fastopen,
963 		.maxlen		= sizeof(int),
964 		.mode		= 0644,
965 		.proc_handler	= proc_dointvec,
966 	},
967 	{
968 		.procname	= "tcp_fastopen_key",
969 		.mode		= 0600,
970 		.data		= &init_net.ipv4.sysctl_tcp_fastopen,
971 		/* maxlen to print the list of keys in hex (*2), with dashes
972 		 * separating doublewords and a comma in between keys.
973 		 */
974 		.maxlen		= ((TCP_FASTOPEN_KEY_LENGTH *
975 				   2 * TCP_FASTOPEN_KEY_MAX) +
976 				   (TCP_FASTOPEN_KEY_MAX * 5)),
977 		.proc_handler	= proc_tcp_fastopen_key,
978 	},
979 	{
980 		.procname	= "tcp_fastopen_blackhole_timeout_sec",
981 		.data		= &init_net.ipv4.sysctl_tcp_fastopen_blackhole_timeout,
982 		.maxlen		= sizeof(int),
983 		.mode		= 0644,
984 		.proc_handler	= proc_tfo_blackhole_detect_timeout,
985 		.extra1		= SYSCTL_ZERO,
986 	},
987 #ifdef CONFIG_IP_ROUTE_MULTIPATH
988 	{
989 		.procname	= "fib_multipath_use_neigh",
990 		.data		= &init_net.ipv4.sysctl_fib_multipath_use_neigh,
991 		.maxlen		= sizeof(int),
992 		.mode		= 0644,
993 		.proc_handler	= proc_dointvec_minmax,
994 		.extra1		= SYSCTL_ZERO,
995 		.extra2		= SYSCTL_ONE,
996 	},
997 	{
998 		.procname	= "fib_multipath_hash_policy",
999 		.data		= &init_net.ipv4.sysctl_fib_multipath_hash_policy,
1000 		.maxlen		= sizeof(int),
1001 		.mode		= 0644,
1002 		.proc_handler	= proc_fib_multipath_hash_policy,
1003 		.extra1		= SYSCTL_ZERO,
1004 		.extra2		= &two,
1005 	},
1006 #endif
1007 	{
1008 		.procname	= "ip_unprivileged_port_start",
1009 		.maxlen		= sizeof(int),
1010 		.data		= &init_net.ipv4.sysctl_ip_prot_sock,
1011 		.mode		= 0644,
1012 		.proc_handler	= ipv4_privileged_ports,
1013 	},
1014 #ifdef CONFIG_NET_L3_MASTER_DEV
1015 	{
1016 		.procname	= "udp_l3mdev_accept",
1017 		.data		= &init_net.ipv4.sysctl_udp_l3mdev_accept,
1018 		.maxlen		= sizeof(int),
1019 		.mode		= 0644,
1020 		.proc_handler	= proc_dointvec_minmax,
1021 		.extra1		= SYSCTL_ZERO,
1022 		.extra2		= SYSCTL_ONE,
1023 	},
1024 #endif
1025 	{
1026 		.procname	= "tcp_sack",
1027 		.data		= &init_net.ipv4.sysctl_tcp_sack,
1028 		.maxlen		= sizeof(u8),
1029 		.mode		= 0644,
1030 		.proc_handler	= proc_dou8vec_minmax,
1031 	},
1032 	{
1033 		.procname	= "tcp_window_scaling",
1034 		.data		= &init_net.ipv4.sysctl_tcp_window_scaling,
1035 		.maxlen		= sizeof(u8),
1036 		.mode		= 0644,
1037 		.proc_handler	= proc_dou8vec_minmax,
1038 	},
1039 	{
1040 		.procname	= "tcp_timestamps",
1041 		.data		= &init_net.ipv4.sysctl_tcp_timestamps,
1042 		.maxlen		= sizeof(u8),
1043 		.mode		= 0644,
1044 		.proc_handler	= proc_dou8vec_minmax,
1045 	},
1046 	{
1047 		.procname	= "tcp_early_retrans",
1048 		.data		= &init_net.ipv4.sysctl_tcp_early_retrans,
1049 		.maxlen		= sizeof(u8),
1050 		.mode		= 0644,
1051 		.proc_handler	= proc_dou8vec_minmax,
1052 		.extra1		= SYSCTL_ZERO,
1053 		.extra2		= &four,
1054 	},
1055 	{
1056 		.procname	= "tcp_recovery",
1057 		.data		= &init_net.ipv4.sysctl_tcp_recovery,
1058 		.maxlen		= sizeof(u8),
1059 		.mode		= 0644,
1060 		.proc_handler	= proc_dou8vec_minmax,
1061 	},
1062 	{
1063 		.procname       = "tcp_thin_linear_timeouts",
1064 		.data           = &init_net.ipv4.sysctl_tcp_thin_linear_timeouts,
1065 		.maxlen         = sizeof(u8),
1066 		.mode           = 0644,
1067 		.proc_handler   = proc_dou8vec_minmax,
1068 	},
1069 	{
1070 		.procname	= "tcp_slow_start_after_idle",
1071 		.data		= &init_net.ipv4.sysctl_tcp_slow_start_after_idle,
1072 		.maxlen		= sizeof(u8),
1073 		.mode		= 0644,
1074 		.proc_handler	= proc_dou8vec_minmax,
1075 	},
1076 	{
1077 		.procname	= "tcp_retrans_collapse",
1078 		.data		= &init_net.ipv4.sysctl_tcp_retrans_collapse,
1079 		.maxlen		= sizeof(u8),
1080 		.mode		= 0644,
1081 		.proc_handler	= proc_dou8vec_minmax,
1082 	},
1083 	{
1084 		.procname	= "tcp_stdurg",
1085 		.data		= &init_net.ipv4.sysctl_tcp_stdurg,
1086 		.maxlen		= sizeof(u8),
1087 		.mode		= 0644,
1088 		.proc_handler	= proc_dou8vec_minmax,
1089 	},
1090 	{
1091 		.procname	= "tcp_rfc1337",
1092 		.data		= &init_net.ipv4.sysctl_tcp_rfc1337,
1093 		.maxlen		= sizeof(u8),
1094 		.mode		= 0644,
1095 		.proc_handler	= proc_dou8vec_minmax,
1096 	},
1097 	{
1098 		.procname	= "tcp_abort_on_overflow",
1099 		.data		= &init_net.ipv4.sysctl_tcp_abort_on_overflow,
1100 		.maxlen		= sizeof(u8),
1101 		.mode		= 0644,
1102 		.proc_handler	= proc_dou8vec_minmax,
1103 	},
1104 	{
1105 		.procname	= "tcp_fack",
1106 		.data		= &init_net.ipv4.sysctl_tcp_fack,
1107 		.maxlen		= sizeof(u8),
1108 		.mode		= 0644,
1109 		.proc_handler	= proc_dou8vec_minmax,
1110 	},
1111 	{
1112 		.procname	= "tcp_max_reordering",
1113 		.data		= &init_net.ipv4.sysctl_tcp_max_reordering,
1114 		.maxlen		= sizeof(int),
1115 		.mode		= 0644,
1116 		.proc_handler	= proc_dointvec
1117 	},
1118 	{
1119 		.procname	= "tcp_dsack",
1120 		.data		= &init_net.ipv4.sysctl_tcp_dsack,
1121 		.maxlen		= sizeof(u8),
1122 		.mode		= 0644,
1123 		.proc_handler	= proc_dou8vec_minmax,
1124 	},
1125 	{
1126 		.procname	= "tcp_app_win",
1127 		.data		= &init_net.ipv4.sysctl_tcp_app_win,
1128 		.maxlen		= sizeof(u8),
1129 		.mode		= 0644,
1130 		.proc_handler	= proc_dou8vec_minmax,
1131 		.extra1		= SYSCTL_ZERO,
1132 		.extra2		= &tcp_app_win_max,
1133 	},
1134 	{
1135 		.procname	= "tcp_adv_win_scale",
1136 		.data		= &init_net.ipv4.sysctl_tcp_adv_win_scale,
1137 		.maxlen		= sizeof(int),
1138 		.mode		= 0644,
1139 		.proc_handler	= proc_dointvec_minmax,
1140 		.extra1		= &tcp_adv_win_scale_min,
1141 		.extra2		= &tcp_adv_win_scale_max,
1142 	},
1143 	{
1144 		.procname	= "tcp_frto",
1145 		.data		= &init_net.ipv4.sysctl_tcp_frto,
1146 		.maxlen		= sizeof(u8),
1147 		.mode		= 0644,
1148 		.proc_handler	= proc_dou8vec_minmax,
1149 	},
1150 	{
1151 		.procname	= "tcp_no_metrics_save",
1152 		.data		= &init_net.ipv4.sysctl_tcp_nometrics_save,
1153 		.maxlen		= sizeof(u8),
1154 		.mode		= 0644,
1155 		.proc_handler	= proc_dou8vec_minmax,
1156 	},
1157 	{
1158 		.procname	= "tcp_no_ssthresh_metrics_save",
1159 		.data		= &init_net.ipv4.sysctl_tcp_no_ssthresh_metrics_save,
1160 		.maxlen		= sizeof(u8),
1161 		.mode		= 0644,
1162 		.proc_handler	= proc_dou8vec_minmax,
1163 		.extra1		= SYSCTL_ZERO,
1164 		.extra2		= SYSCTL_ONE,
1165 	},
1166 	{
1167 		.procname	= "tcp_moderate_rcvbuf",
1168 		.data		= &init_net.ipv4.sysctl_tcp_moderate_rcvbuf,
1169 		.maxlen		= sizeof(u8),
1170 		.mode		= 0644,
1171 		.proc_handler	= proc_dou8vec_minmax,
1172 	},
1173 	{
1174 		.procname	= "tcp_tso_win_divisor",
1175 		.data		= &init_net.ipv4.sysctl_tcp_tso_win_divisor,
1176 		.maxlen		= sizeof(u8),
1177 		.mode		= 0644,
1178 		.proc_handler	= proc_dou8vec_minmax,
1179 	},
1180 	{
1181 		.procname	= "tcp_workaround_signed_windows",
1182 		.data		= &init_net.ipv4.sysctl_tcp_workaround_signed_windows,
1183 		.maxlen		= sizeof(u8),
1184 		.mode		= 0644,
1185 		.proc_handler	= proc_dou8vec_minmax,
1186 	},
1187 	{
1188 		.procname	= "tcp_limit_output_bytes",
1189 		.data		= &init_net.ipv4.sysctl_tcp_limit_output_bytes,
1190 		.maxlen		= sizeof(int),
1191 		.mode		= 0644,
1192 		.proc_handler	= proc_dointvec
1193 	},
1194 	{
1195 		.procname	= "tcp_challenge_ack_limit",
1196 		.data		= &init_net.ipv4.sysctl_tcp_challenge_ack_limit,
1197 		.maxlen		= sizeof(int),
1198 		.mode		= 0644,
1199 		.proc_handler	= proc_dointvec
1200 	},
1201 	{
1202 		.procname	= "tcp_min_tso_segs",
1203 		.data		= &init_net.ipv4.sysctl_tcp_min_tso_segs,
1204 		.maxlen		= sizeof(u8),
1205 		.mode		= 0644,
1206 		.proc_handler	= proc_dou8vec_minmax,
1207 		.extra1		= SYSCTL_ONE,
1208 	},
1209 	{
1210 		.procname	= "tcp_min_rtt_wlen",
1211 		.data		= &init_net.ipv4.sysctl_tcp_min_rtt_wlen,
1212 		.maxlen		= sizeof(int),
1213 		.mode		= 0644,
1214 		.proc_handler	= proc_dointvec_minmax,
1215 		.extra1		= SYSCTL_ZERO,
1216 		.extra2		= &one_day_secs
1217 	},
1218 	{
1219 		.procname	= "tcp_autocorking",
1220 		.data		= &init_net.ipv4.sysctl_tcp_autocorking,
1221 		.maxlen		= sizeof(u8),
1222 		.mode		= 0644,
1223 		.proc_handler	= proc_dou8vec_minmax,
1224 		.extra1		= SYSCTL_ZERO,
1225 		.extra2		= SYSCTL_ONE,
1226 	},
1227 	{
1228 		.procname	= "tcp_invalid_ratelimit",
1229 		.data		= &init_net.ipv4.sysctl_tcp_invalid_ratelimit,
1230 		.maxlen		= sizeof(int),
1231 		.mode		= 0644,
1232 		.proc_handler	= proc_dointvec_ms_jiffies,
1233 	},
1234 	{
1235 		.procname	= "tcp_pacing_ss_ratio",
1236 		.data		= &init_net.ipv4.sysctl_tcp_pacing_ss_ratio,
1237 		.maxlen		= sizeof(int),
1238 		.mode		= 0644,
1239 		.proc_handler	= proc_dointvec_minmax,
1240 		.extra1		= SYSCTL_ZERO,
1241 		.extra2		= &thousand,
1242 	},
1243 	{
1244 		.procname	= "tcp_pacing_ca_ratio",
1245 		.data		= &init_net.ipv4.sysctl_tcp_pacing_ca_ratio,
1246 		.maxlen		= sizeof(int),
1247 		.mode		= 0644,
1248 		.proc_handler	= proc_dointvec_minmax,
1249 		.extra1		= SYSCTL_ZERO,
1250 		.extra2		= &thousand,
1251 	},
1252 	{
1253 		.procname	= "tcp_wmem",
1254 		.data		= &init_net.ipv4.sysctl_tcp_wmem,
1255 		.maxlen		= sizeof(init_net.ipv4.sysctl_tcp_wmem),
1256 		.mode		= 0644,
1257 		.proc_handler	= proc_dointvec_minmax,
1258 		.extra1		= SYSCTL_ONE,
1259 	},
1260 	{
1261 		.procname	= "tcp_rmem",
1262 		.data		= &init_net.ipv4.sysctl_tcp_rmem,
1263 		.maxlen		= sizeof(init_net.ipv4.sysctl_tcp_rmem),
1264 		.mode		= 0644,
1265 		.proc_handler	= proc_dointvec_minmax,
1266 		.extra1		= SYSCTL_ONE,
1267 	},
1268 	{
1269 		.procname	= "tcp_comp_sack_delay_ns",
1270 		.data		= &init_net.ipv4.sysctl_tcp_comp_sack_delay_ns,
1271 		.maxlen		= sizeof(unsigned long),
1272 		.mode		= 0644,
1273 		.proc_handler	= proc_doulongvec_minmax,
1274 	},
1275 	{
1276 		.procname	= "tcp_comp_sack_slack_ns",
1277 		.data		= &init_net.ipv4.sysctl_tcp_comp_sack_slack_ns,
1278 		.maxlen		= sizeof(unsigned long),
1279 		.mode		= 0644,
1280 		.proc_handler	= proc_doulongvec_minmax,
1281 	},
1282 	{
1283 		.procname	= "tcp_comp_sack_nr",
1284 		.data		= &init_net.ipv4.sysctl_tcp_comp_sack_nr,
1285 		.maxlen		= sizeof(int),
1286 		.mode		= 0644,
1287 		.proc_handler	= proc_dointvec_minmax,
1288 		.extra1		= SYSCTL_ZERO,
1289 		.extra2		= &comp_sack_nr_max,
1290 	},
1291 	{
1292 		.procname       = "tcp_reflect_tos",
1293 		.data           = &init_net.ipv4.sysctl_tcp_reflect_tos,
1294 		.maxlen         = sizeof(u8),
1295 		.mode           = 0644,
1296 		.proc_handler   = proc_dou8vec_minmax,
1297 		.extra1         = SYSCTL_ZERO,
1298 		.extra2         = SYSCTL_ONE,
1299 	},
1300 	{
1301 		.procname	= "udp_rmem_min",
1302 		.data		= &init_net.ipv4.sysctl_udp_rmem_min,
1303 		.maxlen		= sizeof(init_net.ipv4.sysctl_udp_rmem_min),
1304 		.mode		= 0644,
1305 		.proc_handler	= proc_dointvec_minmax,
1306 		.extra1		= SYSCTL_ONE
1307 	},
1308 	{
1309 		.procname	= "udp_wmem_min",
1310 		.data		= &init_net.ipv4.sysctl_udp_wmem_min,
1311 		.maxlen		= sizeof(init_net.ipv4.sysctl_udp_wmem_min),
1312 		.mode		= 0644,
1313 		.proc_handler	= proc_dointvec_minmax,
1314 		.extra1		= SYSCTL_ONE
1315 	},
1316 	{ }
1317 };
1318 
ipv4_sysctl_init_net(struct net * net)1319 static __net_init int ipv4_sysctl_init_net(struct net *net)
1320 {
1321 	struct ctl_table *table;
1322 
1323 	table = ipv4_net_table;
1324 	if (!net_eq(net, &init_net)) {
1325 		int i;
1326 
1327 		table = kmemdup(table, sizeof(ipv4_net_table), GFP_KERNEL);
1328 		if (!table)
1329 			goto err_alloc;
1330 
1331 		for (i = 0; i < ARRAY_SIZE(ipv4_net_table) - 1; i++) {
1332 			if (table[i].data) {
1333 				/* Update the variables to point into
1334 				 * the current struct net
1335 				 */
1336 				table[i].data += (void *)net - (void *)&init_net;
1337 			} else {
1338 				/* Entries without data pointer are global;
1339 				 * Make them read-only in non-init_net ns
1340 				 */
1341 				table[i].mode &= ~0222;
1342 			}
1343 		}
1344 	}
1345 
1346 	net->ipv4.ipv4_hdr = register_net_sysctl(net, "net/ipv4", table);
1347 	if (!net->ipv4.ipv4_hdr)
1348 		goto err_reg;
1349 
1350 	net->ipv4.sysctl_local_reserved_ports = kzalloc(65536 / 8, GFP_KERNEL);
1351 	if (!net->ipv4.sysctl_local_reserved_ports)
1352 		goto err_ports;
1353 
1354 	return 0;
1355 
1356 err_ports:
1357 	unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
1358 err_reg:
1359 	if (!net_eq(net, &init_net))
1360 		kfree(table);
1361 err_alloc:
1362 	return -ENOMEM;
1363 }
1364 
ipv4_sysctl_exit_net(struct net * net)1365 static __net_exit void ipv4_sysctl_exit_net(struct net *net)
1366 {
1367 	struct ctl_table *table;
1368 
1369 	kfree(net->ipv4.sysctl_local_reserved_ports);
1370 	table = net->ipv4.ipv4_hdr->ctl_table_arg;
1371 	unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
1372 	kfree(table);
1373 }
1374 
1375 static __net_initdata struct pernet_operations ipv4_sysctl_ops = {
1376 	.init = ipv4_sysctl_init_net,
1377 	.exit = ipv4_sysctl_exit_net,
1378 };
1379 
sysctl_ipv4_init(void)1380 static __init int sysctl_ipv4_init(void)
1381 {
1382 	struct ctl_table_header *hdr;
1383 
1384 	hdr = register_net_sysctl(&init_net, "net/ipv4", ipv4_table);
1385 	if (!hdr)
1386 		return -ENOMEM;
1387 
1388 	if (register_pernet_subsys(&ipv4_sysctl_ops)) {
1389 		unregister_net_sysctl_table(hdr);
1390 		return -ENOMEM;
1391 	}
1392 
1393 	return 0;
1394 }
1395 
1396 __initcall(sysctl_ipv4_init);
1397