• 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 	},
616 	{
617 		.procname	= "ip_dynaddr",
618 		.data		= &init_net.ipv4.sysctl_ip_dynaddr,
619 		.maxlen		= sizeof(u8),
620 		.mode		= 0644,
621 		.proc_handler	= proc_dou8vec_minmax,
622 	},
623 	{
624 		.procname	= "ip_early_demux",
625 		.data		= &init_net.ipv4.sysctl_ip_early_demux,
626 		.maxlen		= sizeof(u8),
627 		.mode		= 0644,
628 		.proc_handler	= proc_dou8vec_minmax,
629 	},
630 	{
631 		.procname       = "udp_early_demux",
632 		.data           = &init_net.ipv4.sysctl_udp_early_demux,
633 		.maxlen         = sizeof(int),
634 		.mode           = 0644,
635 		.proc_handler   = proc_douintvec_minmax,
636 	},
637 	{
638 		.procname       = "tcp_early_demux",
639 		.data           = &init_net.ipv4.sysctl_tcp_early_demux,
640 		.maxlen         = sizeof(int),
641 		.mode           = 0644,
642 		.proc_handler   = proc_douintvec_minmax,
643 	},
644 	{
645 		.procname       = "nexthop_compat_mode",
646 		.data           = &init_net.ipv4.sysctl_nexthop_compat_mode,
647 		.maxlen         = sizeof(u8),
648 		.mode           = 0644,
649 		.proc_handler   = proc_dou8vec_minmax,
650 		.extra1		= SYSCTL_ZERO,
651 		.extra2		= SYSCTL_ONE,
652 	},
653 	{
654 		.procname	= "ip_default_ttl",
655 		.data		= &init_net.ipv4.sysctl_ip_default_ttl,
656 		.maxlen		= sizeof(u8),
657 		.mode		= 0644,
658 		.proc_handler	= proc_dou8vec_minmax,
659 		.extra1		= &ip_ttl_min,
660 		.extra2		= &ip_ttl_max,
661 	},
662 	{
663 		.procname	= "ip_local_port_range",
664 		.maxlen		= sizeof(init_net.ipv4.ip_local_ports.range),
665 		.data		= &init_net.ipv4.ip_local_ports.range,
666 		.mode		= 0644,
667 		.proc_handler	= ipv4_local_port_range,
668 	},
669 	{
670 		.procname	= "ip_local_reserved_ports",
671 		.data		= &init_net.ipv4.sysctl_local_reserved_ports,
672 		.maxlen		= 65536,
673 		.mode		= 0644,
674 		.proc_handler	= proc_do_large_bitmap,
675 	},
676 	{
677 		.procname	= "ip_no_pmtu_disc",
678 		.data		= &init_net.ipv4.sysctl_ip_no_pmtu_disc,
679 		.maxlen		= sizeof(u8),
680 		.mode		= 0644,
681 		.proc_handler	= proc_dou8vec_minmax,
682 	},
683 	{
684 		.procname	= "ip_forward_use_pmtu",
685 		.data		= &init_net.ipv4.sysctl_ip_fwd_use_pmtu,
686 		.maxlen		= sizeof(u8),
687 		.mode		= 0644,
688 		.proc_handler	= proc_dou8vec_minmax,
689 	},
690 	{
691 		.procname	= "ip_forward_update_priority",
692 		.data		= &init_net.ipv4.sysctl_ip_fwd_update_priority,
693 		.maxlen		= sizeof(int),
694 		.mode		= 0644,
695 		.proc_handler   = ipv4_fwd_update_priority,
696 		.extra1		= SYSCTL_ZERO,
697 		.extra2		= SYSCTL_ONE,
698 	},
699 	{
700 		.procname	= "ip_nonlocal_bind",
701 		.data		= &init_net.ipv4.sysctl_ip_nonlocal_bind,
702 		.maxlen		= sizeof(u8),
703 		.mode		= 0644,
704 		.proc_handler	= proc_dou8vec_minmax,
705 	},
706 	{
707 		.procname	= "ip_autobind_reuse",
708 		.data		= &init_net.ipv4.sysctl_ip_autobind_reuse,
709 		.maxlen		= sizeof(u8),
710 		.mode		= 0644,
711 		.proc_handler	= proc_dou8vec_minmax,
712 		.extra1         = SYSCTL_ZERO,
713 		.extra2         = SYSCTL_ONE,
714 	},
715 	{
716 		.procname	= "fwmark_reflect",
717 		.data		= &init_net.ipv4.sysctl_fwmark_reflect,
718 		.maxlen		= sizeof(u8),
719 		.mode		= 0644,
720 		.proc_handler	= proc_dou8vec_minmax,
721 	},
722 	{
723 		.procname	= "tcp_fwmark_accept",
724 		.data		= &init_net.ipv4.sysctl_tcp_fwmark_accept,
725 		.maxlen		= sizeof(u8),
726 		.mode		= 0644,
727 		.proc_handler	= proc_dou8vec_minmax,
728 	},
729 #ifdef CONFIG_NET_L3_MASTER_DEV
730 	{
731 		.procname	= "tcp_l3mdev_accept",
732 		.data		= &init_net.ipv4.sysctl_tcp_l3mdev_accept,
733 		.maxlen		= sizeof(u8),
734 		.mode		= 0644,
735 		.proc_handler	= proc_dou8vec_minmax,
736 		.extra1		= SYSCTL_ZERO,
737 		.extra2		= SYSCTL_ONE,
738 	},
739 #endif
740 	{
741 		.procname	= "tcp_mtu_probing",
742 		.data		= &init_net.ipv4.sysctl_tcp_mtu_probing,
743 		.maxlen		= sizeof(u8),
744 		.mode		= 0644,
745 		.proc_handler	= proc_dou8vec_minmax,
746 	},
747 	{
748 		.procname	= "tcp_base_mss",
749 		.data		= &init_net.ipv4.sysctl_tcp_base_mss,
750 		.maxlen		= sizeof(int),
751 		.mode		= 0644,
752 		.proc_handler	= proc_dointvec,
753 	},
754 	{
755 		.procname	= "tcp_min_snd_mss",
756 		.data		= &init_net.ipv4.sysctl_tcp_min_snd_mss,
757 		.maxlen		= sizeof(int),
758 		.mode		= 0644,
759 		.proc_handler	= proc_dointvec_minmax,
760 		.extra1		= &tcp_min_snd_mss_min,
761 		.extra2		= &tcp_min_snd_mss_max,
762 	},
763 	{
764 		.procname	= "tcp_mtu_probe_floor",
765 		.data		= &init_net.ipv4.sysctl_tcp_mtu_probe_floor,
766 		.maxlen		= sizeof(int),
767 		.mode		= 0644,
768 		.proc_handler	= proc_dointvec_minmax,
769 		.extra1		= &tcp_min_snd_mss_min,
770 		.extra2		= &tcp_min_snd_mss_max,
771 	},
772 	{
773 		.procname	= "tcp_probe_threshold",
774 		.data		= &init_net.ipv4.sysctl_tcp_probe_threshold,
775 		.maxlen		= sizeof(int),
776 		.mode		= 0644,
777 		.proc_handler	= proc_dointvec,
778 	},
779 	{
780 		.procname	= "tcp_probe_interval",
781 		.data		= &init_net.ipv4.sysctl_tcp_probe_interval,
782 		.maxlen		= sizeof(u32),
783 		.mode		= 0644,
784 		.proc_handler	= proc_douintvec_minmax,
785 		.extra2		= &u32_max_div_HZ,
786 	},
787 	{
788 		.procname	= "igmp_link_local_mcast_reports",
789 		.data		= &init_net.ipv4.sysctl_igmp_llm_reports,
790 		.maxlen		= sizeof(int),
791 		.mode		= 0644,
792 		.proc_handler	= proc_dointvec
793 	},
794 	{
795 		.procname	= "igmp_max_memberships",
796 		.data		= &init_net.ipv4.sysctl_igmp_max_memberships,
797 		.maxlen		= sizeof(int),
798 		.mode		= 0644,
799 		.proc_handler	= proc_dointvec
800 	},
801 	{
802 		.procname	= "igmp_max_msf",
803 		.data		= &init_net.ipv4.sysctl_igmp_max_msf,
804 		.maxlen		= sizeof(int),
805 		.mode		= 0644,
806 		.proc_handler	= proc_dointvec
807 	},
808 #ifdef CONFIG_IP_MULTICAST
809 	{
810 		.procname	= "igmp_qrv",
811 		.data		= &init_net.ipv4.sysctl_igmp_qrv,
812 		.maxlen		= sizeof(int),
813 		.mode		= 0644,
814 		.proc_handler	= proc_dointvec_minmax,
815 		.extra1		= SYSCTL_ONE
816 	},
817 #endif
818 	{
819 		.procname	= "tcp_congestion_control",
820 		.data		= &init_net.ipv4.tcp_congestion_control,
821 		.mode		= 0644,
822 		.maxlen		= TCP_CA_NAME_MAX,
823 		.proc_handler	= proc_tcp_congestion_control,
824 	},
825 	{
826 		.procname	= "tcp_available_congestion_control",
827 		.maxlen		= TCP_CA_BUF_MAX,
828 		.mode		= 0444,
829 		.proc_handler   = proc_tcp_available_congestion_control,
830 	},
831 	{
832 		.procname	= "tcp_allowed_congestion_control",
833 		.maxlen		= TCP_CA_BUF_MAX,
834 		.mode		= 0644,
835 		.proc_handler   = proc_allowed_congestion_control,
836 	},
837 	{
838 		.procname	= "tcp_keepalive_time",
839 		.data		= &init_net.ipv4.sysctl_tcp_keepalive_time,
840 		.maxlen		= sizeof(int),
841 		.mode		= 0644,
842 		.proc_handler	= proc_dointvec_jiffies,
843 	},
844 	{
845 		.procname	= "tcp_keepalive_probes",
846 		.data		= &init_net.ipv4.sysctl_tcp_keepalive_probes,
847 		.maxlen		= sizeof(u8),
848 		.mode		= 0644,
849 		.proc_handler	= proc_dou8vec_minmax,
850 	},
851 	{
852 		.procname	= "tcp_keepalive_intvl",
853 		.data		= &init_net.ipv4.sysctl_tcp_keepalive_intvl,
854 		.maxlen		= sizeof(int),
855 		.mode		= 0644,
856 		.proc_handler	= proc_dointvec_jiffies,
857 	},
858 	{
859 		.procname	= "tcp_syn_retries",
860 		.data		= &init_net.ipv4.sysctl_tcp_syn_retries,
861 		.maxlen		= sizeof(u8),
862 		.mode		= 0644,
863 		.proc_handler	= proc_dou8vec_minmax,
864 		.extra1		= &tcp_syn_retries_min,
865 		.extra2		= &tcp_syn_retries_max
866 	},
867 	{
868 		.procname	= "tcp_synack_retries",
869 		.data		= &init_net.ipv4.sysctl_tcp_synack_retries,
870 		.maxlen		= sizeof(u8),
871 		.mode		= 0644,
872 		.proc_handler	= proc_dou8vec_minmax,
873 	},
874 #ifdef CONFIG_SYN_COOKIES
875 	{
876 		.procname	= "tcp_syncookies",
877 		.data		= &init_net.ipv4.sysctl_tcp_syncookies,
878 		.maxlen		= sizeof(u8),
879 		.mode		= 0644,
880 		.proc_handler	= proc_dou8vec_minmax,
881 	},
882 #endif
883 	{
884 		.procname	= "tcp_migrate_req",
885 		.data		= &init_net.ipv4.sysctl_tcp_migrate_req,
886 		.maxlen		= sizeof(u8),
887 		.mode		= 0644,
888 		.proc_handler	= proc_dou8vec_minmax,
889 		.extra1		= SYSCTL_ZERO,
890 		.extra2		= SYSCTL_ONE
891 	},
892 	{
893 		.procname	= "tcp_reordering",
894 		.data		= &init_net.ipv4.sysctl_tcp_reordering,
895 		.maxlen		= sizeof(int),
896 		.mode		= 0644,
897 		.proc_handler	= proc_dointvec
898 	},
899 	{
900 		.procname	= "tcp_retries1",
901 		.data		= &init_net.ipv4.sysctl_tcp_retries1,
902 		.maxlen		= sizeof(u8),
903 		.mode		= 0644,
904 		.proc_handler	= proc_dou8vec_minmax,
905 		.extra2		= &tcp_retr1_max
906 	},
907 	{
908 		.procname	= "tcp_retries2",
909 		.data		= &init_net.ipv4.sysctl_tcp_retries2,
910 		.maxlen		= sizeof(u8),
911 		.mode		= 0644,
912 		.proc_handler	= proc_dou8vec_minmax,
913 	},
914 	{
915 		.procname	= "tcp_orphan_retries",
916 		.data		= &init_net.ipv4.sysctl_tcp_orphan_retries,
917 		.maxlen		= sizeof(u8),
918 		.mode		= 0644,
919 		.proc_handler	= proc_dou8vec_minmax,
920 	},
921 	{
922 		.procname	= "tcp_fin_timeout",
923 		.data		= &init_net.ipv4.sysctl_tcp_fin_timeout,
924 		.maxlen		= sizeof(int),
925 		.mode		= 0644,
926 		.proc_handler	= proc_dointvec_jiffies,
927 	},
928 	{
929 		.procname	= "tcp_notsent_lowat",
930 		.data		= &init_net.ipv4.sysctl_tcp_notsent_lowat,
931 		.maxlen		= sizeof(unsigned int),
932 		.mode		= 0644,
933 		.proc_handler	= proc_douintvec,
934 	},
935 	{
936 		.procname	= "tcp_tw_reuse",
937 		.data		= &init_net.ipv4.sysctl_tcp_tw_reuse,
938 		.maxlen		= sizeof(u8),
939 		.mode		= 0644,
940 		.proc_handler	= proc_dou8vec_minmax,
941 		.extra1		= SYSCTL_ZERO,
942 		.extra2		= &two,
943 	},
944 	{
945 		.procname	= "tcp_max_tw_buckets",
946 		.data		= &init_net.ipv4.tcp_death_row.sysctl_max_tw_buckets,
947 		.maxlen		= sizeof(int),
948 		.mode		= 0644,
949 		.proc_handler	= proc_dointvec
950 	},
951 	{
952 		.procname	= "tcp_max_syn_backlog",
953 		.data		= &init_net.ipv4.sysctl_max_syn_backlog,
954 		.maxlen		= sizeof(int),
955 		.mode		= 0644,
956 		.proc_handler	= proc_dointvec
957 	},
958 	{
959 		.procname	= "tcp_fastopen",
960 		.data		= &init_net.ipv4.sysctl_tcp_fastopen,
961 		.maxlen		= sizeof(int),
962 		.mode		= 0644,
963 		.proc_handler	= proc_dointvec,
964 	},
965 	{
966 		.procname	= "tcp_fastopen_key",
967 		.mode		= 0600,
968 		.data		= &init_net.ipv4.sysctl_tcp_fastopen,
969 		/* maxlen to print the list of keys in hex (*2), with dashes
970 		 * separating doublewords and a comma in between keys.
971 		 */
972 		.maxlen		= ((TCP_FASTOPEN_KEY_LENGTH *
973 				   2 * TCP_FASTOPEN_KEY_MAX) +
974 				   (TCP_FASTOPEN_KEY_MAX * 5)),
975 		.proc_handler	= proc_tcp_fastopen_key,
976 	},
977 	{
978 		.procname	= "tcp_fastopen_blackhole_timeout_sec",
979 		.data		= &init_net.ipv4.sysctl_tcp_fastopen_blackhole_timeout,
980 		.maxlen		= sizeof(int),
981 		.mode		= 0644,
982 		.proc_handler	= proc_tfo_blackhole_detect_timeout,
983 		.extra1		= SYSCTL_ZERO,
984 	},
985 #ifdef CONFIG_IP_ROUTE_MULTIPATH
986 	{
987 		.procname	= "fib_multipath_use_neigh",
988 		.data		= &init_net.ipv4.sysctl_fib_multipath_use_neigh,
989 		.maxlen		= sizeof(int),
990 		.mode		= 0644,
991 		.proc_handler	= proc_dointvec_minmax,
992 		.extra1		= SYSCTL_ZERO,
993 		.extra2		= SYSCTL_ONE,
994 	},
995 	{
996 		.procname	= "fib_multipath_hash_policy",
997 		.data		= &init_net.ipv4.sysctl_fib_multipath_hash_policy,
998 		.maxlen		= sizeof(int),
999 		.mode		= 0644,
1000 		.proc_handler	= proc_fib_multipath_hash_policy,
1001 		.extra1		= SYSCTL_ZERO,
1002 		.extra2		= &two,
1003 	},
1004 #endif
1005 	{
1006 		.procname	= "ip_unprivileged_port_start",
1007 		.maxlen		= sizeof(int),
1008 		.data		= &init_net.ipv4.sysctl_ip_prot_sock,
1009 		.mode		= 0644,
1010 		.proc_handler	= ipv4_privileged_ports,
1011 	},
1012 #ifdef CONFIG_NET_L3_MASTER_DEV
1013 	{
1014 		.procname	= "udp_l3mdev_accept",
1015 		.data		= &init_net.ipv4.sysctl_udp_l3mdev_accept,
1016 		.maxlen		= sizeof(int),
1017 		.mode		= 0644,
1018 		.proc_handler	= proc_dointvec_minmax,
1019 		.extra1		= SYSCTL_ZERO,
1020 		.extra2		= SYSCTL_ONE,
1021 	},
1022 #endif
1023 	{
1024 		.procname	= "tcp_sack",
1025 		.data		= &init_net.ipv4.sysctl_tcp_sack,
1026 		.maxlen		= sizeof(u8),
1027 		.mode		= 0644,
1028 		.proc_handler	= proc_dou8vec_minmax,
1029 	},
1030 	{
1031 		.procname	= "tcp_window_scaling",
1032 		.data		= &init_net.ipv4.sysctl_tcp_window_scaling,
1033 		.maxlen		= sizeof(u8),
1034 		.mode		= 0644,
1035 		.proc_handler	= proc_dou8vec_minmax,
1036 	},
1037 	{
1038 		.procname	= "tcp_timestamps",
1039 		.data		= &init_net.ipv4.sysctl_tcp_timestamps,
1040 		.maxlen		= sizeof(u8),
1041 		.mode		= 0644,
1042 		.proc_handler	= proc_dou8vec_minmax,
1043 	},
1044 	{
1045 		.procname	= "tcp_early_retrans",
1046 		.data		= &init_net.ipv4.sysctl_tcp_early_retrans,
1047 		.maxlen		= sizeof(u8),
1048 		.mode		= 0644,
1049 		.proc_handler	= proc_dou8vec_minmax,
1050 		.extra1		= SYSCTL_ZERO,
1051 		.extra2		= &four,
1052 	},
1053 	{
1054 		.procname	= "tcp_recovery",
1055 		.data		= &init_net.ipv4.sysctl_tcp_recovery,
1056 		.maxlen		= sizeof(u8),
1057 		.mode		= 0644,
1058 		.proc_handler	= proc_dou8vec_minmax,
1059 	},
1060 	{
1061 		.procname       = "tcp_thin_linear_timeouts",
1062 		.data           = &init_net.ipv4.sysctl_tcp_thin_linear_timeouts,
1063 		.maxlen         = sizeof(u8),
1064 		.mode           = 0644,
1065 		.proc_handler   = proc_dou8vec_minmax,
1066 	},
1067 	{
1068 		.procname	= "tcp_slow_start_after_idle",
1069 		.data		= &init_net.ipv4.sysctl_tcp_slow_start_after_idle,
1070 		.maxlen		= sizeof(u8),
1071 		.mode		= 0644,
1072 		.proc_handler	= proc_dou8vec_minmax,
1073 	},
1074 	{
1075 		.procname	= "tcp_retrans_collapse",
1076 		.data		= &init_net.ipv4.sysctl_tcp_retrans_collapse,
1077 		.maxlen		= sizeof(u8),
1078 		.mode		= 0644,
1079 		.proc_handler	= proc_dou8vec_minmax,
1080 	},
1081 	{
1082 		.procname	= "tcp_stdurg",
1083 		.data		= &init_net.ipv4.sysctl_tcp_stdurg,
1084 		.maxlen		= sizeof(u8),
1085 		.mode		= 0644,
1086 		.proc_handler	= proc_dou8vec_minmax,
1087 	},
1088 	{
1089 		.procname	= "tcp_rfc1337",
1090 		.data		= &init_net.ipv4.sysctl_tcp_rfc1337,
1091 		.maxlen		= sizeof(u8),
1092 		.mode		= 0644,
1093 		.proc_handler	= proc_dou8vec_minmax,
1094 	},
1095 	{
1096 		.procname	= "tcp_abort_on_overflow",
1097 		.data		= &init_net.ipv4.sysctl_tcp_abort_on_overflow,
1098 		.maxlen		= sizeof(u8),
1099 		.mode		= 0644,
1100 		.proc_handler	= proc_dou8vec_minmax,
1101 	},
1102 	{
1103 		.procname	= "tcp_fack",
1104 		.data		= &init_net.ipv4.sysctl_tcp_fack,
1105 		.maxlen		= sizeof(u8),
1106 		.mode		= 0644,
1107 		.proc_handler	= proc_dou8vec_minmax,
1108 	},
1109 	{
1110 		.procname	= "tcp_max_reordering",
1111 		.data		= &init_net.ipv4.sysctl_tcp_max_reordering,
1112 		.maxlen		= sizeof(int),
1113 		.mode		= 0644,
1114 		.proc_handler	= proc_dointvec
1115 	},
1116 	{
1117 		.procname	= "tcp_dsack",
1118 		.data		= &init_net.ipv4.sysctl_tcp_dsack,
1119 		.maxlen		= sizeof(u8),
1120 		.mode		= 0644,
1121 		.proc_handler	= proc_dou8vec_minmax,
1122 	},
1123 	{
1124 		.procname	= "tcp_app_win",
1125 		.data		= &init_net.ipv4.sysctl_tcp_app_win,
1126 		.maxlen		= sizeof(u8),
1127 		.mode		= 0644,
1128 		.proc_handler	= proc_dou8vec_minmax,
1129 		.extra1		= SYSCTL_ZERO,
1130 		.extra2		= &tcp_app_win_max,
1131 	},
1132 	{
1133 		.procname	= "tcp_adv_win_scale",
1134 		.data		= &init_net.ipv4.sysctl_tcp_adv_win_scale,
1135 		.maxlen		= sizeof(int),
1136 		.mode		= 0644,
1137 		.proc_handler	= proc_dointvec_minmax,
1138 		.extra1		= &tcp_adv_win_scale_min,
1139 		.extra2		= &tcp_adv_win_scale_max,
1140 	},
1141 	{
1142 		.procname	= "tcp_frto",
1143 		.data		= &init_net.ipv4.sysctl_tcp_frto,
1144 		.maxlen		= sizeof(u8),
1145 		.mode		= 0644,
1146 		.proc_handler	= proc_dou8vec_minmax,
1147 	},
1148 	{
1149 		.procname	= "tcp_no_metrics_save",
1150 		.data		= &init_net.ipv4.sysctl_tcp_nometrics_save,
1151 		.maxlen		= sizeof(u8),
1152 		.mode		= 0644,
1153 		.proc_handler	= proc_dou8vec_minmax,
1154 	},
1155 	{
1156 		.procname	= "tcp_no_ssthresh_metrics_save",
1157 		.data		= &init_net.ipv4.sysctl_tcp_no_ssthresh_metrics_save,
1158 		.maxlen		= sizeof(u8),
1159 		.mode		= 0644,
1160 		.proc_handler	= proc_dou8vec_minmax,
1161 		.extra1		= SYSCTL_ZERO,
1162 		.extra2		= SYSCTL_ONE,
1163 	},
1164 	{
1165 		.procname	= "tcp_moderate_rcvbuf",
1166 		.data		= &init_net.ipv4.sysctl_tcp_moderate_rcvbuf,
1167 		.maxlen		= sizeof(u8),
1168 		.mode		= 0644,
1169 		.proc_handler	= proc_dou8vec_minmax,
1170 	},
1171 	{
1172 		.procname	= "tcp_tso_win_divisor",
1173 		.data		= &init_net.ipv4.sysctl_tcp_tso_win_divisor,
1174 		.maxlen		= sizeof(u8),
1175 		.mode		= 0644,
1176 		.proc_handler	= proc_dou8vec_minmax,
1177 	},
1178 	{
1179 		.procname	= "tcp_workaround_signed_windows",
1180 		.data		= &init_net.ipv4.sysctl_tcp_workaround_signed_windows,
1181 		.maxlen		= sizeof(u8),
1182 		.mode		= 0644,
1183 		.proc_handler	= proc_dou8vec_minmax,
1184 	},
1185 	{
1186 		.procname	= "tcp_limit_output_bytes",
1187 		.data		= &init_net.ipv4.sysctl_tcp_limit_output_bytes,
1188 		.maxlen		= sizeof(int),
1189 		.mode		= 0644,
1190 		.proc_handler	= proc_dointvec
1191 	},
1192 	{
1193 		.procname	= "tcp_challenge_ack_limit",
1194 		.data		= &init_net.ipv4.sysctl_tcp_challenge_ack_limit,
1195 		.maxlen		= sizeof(int),
1196 		.mode		= 0644,
1197 		.proc_handler	= proc_dointvec
1198 	},
1199 	{
1200 		.procname	= "tcp_min_tso_segs",
1201 		.data		= &init_net.ipv4.sysctl_tcp_min_tso_segs,
1202 		.maxlen		= sizeof(u8),
1203 		.mode		= 0644,
1204 		.proc_handler	= proc_dou8vec_minmax,
1205 		.extra1		= SYSCTL_ONE,
1206 	},
1207 	{
1208 		.procname	= "tcp_min_rtt_wlen",
1209 		.data		= &init_net.ipv4.sysctl_tcp_min_rtt_wlen,
1210 		.maxlen		= sizeof(int),
1211 		.mode		= 0644,
1212 		.proc_handler	= proc_dointvec_minmax,
1213 		.extra1		= SYSCTL_ZERO,
1214 		.extra2		= &one_day_secs
1215 	},
1216 	{
1217 		.procname	= "tcp_autocorking",
1218 		.data		= &init_net.ipv4.sysctl_tcp_autocorking,
1219 		.maxlen		= sizeof(u8),
1220 		.mode		= 0644,
1221 		.proc_handler	= proc_dou8vec_minmax,
1222 		.extra1		= SYSCTL_ZERO,
1223 		.extra2		= SYSCTL_ONE,
1224 	},
1225 	{
1226 		.procname	= "tcp_invalid_ratelimit",
1227 		.data		= &init_net.ipv4.sysctl_tcp_invalid_ratelimit,
1228 		.maxlen		= sizeof(int),
1229 		.mode		= 0644,
1230 		.proc_handler	= proc_dointvec_ms_jiffies,
1231 	},
1232 	{
1233 		.procname	= "tcp_pacing_ss_ratio",
1234 		.data		= &init_net.ipv4.sysctl_tcp_pacing_ss_ratio,
1235 		.maxlen		= sizeof(int),
1236 		.mode		= 0644,
1237 		.proc_handler	= proc_dointvec_minmax,
1238 		.extra1		= SYSCTL_ZERO,
1239 		.extra2		= &thousand,
1240 	},
1241 	{
1242 		.procname	= "tcp_pacing_ca_ratio",
1243 		.data		= &init_net.ipv4.sysctl_tcp_pacing_ca_ratio,
1244 		.maxlen		= sizeof(int),
1245 		.mode		= 0644,
1246 		.proc_handler	= proc_dointvec_minmax,
1247 		.extra1		= SYSCTL_ZERO,
1248 		.extra2		= &thousand,
1249 	},
1250 	{
1251 		.procname	= "tcp_wmem",
1252 		.data		= &init_net.ipv4.sysctl_tcp_wmem,
1253 		.maxlen		= sizeof(init_net.ipv4.sysctl_tcp_wmem),
1254 		.mode		= 0644,
1255 		.proc_handler	= proc_dointvec_minmax,
1256 		.extra1		= SYSCTL_ONE,
1257 	},
1258 	{
1259 		.procname	= "tcp_rmem",
1260 		.data		= &init_net.ipv4.sysctl_tcp_rmem,
1261 		.maxlen		= sizeof(init_net.ipv4.sysctl_tcp_rmem),
1262 		.mode		= 0644,
1263 		.proc_handler	= proc_dointvec_minmax,
1264 		.extra1		= SYSCTL_ONE,
1265 	},
1266 	{
1267 		.procname	= "tcp_comp_sack_delay_ns",
1268 		.data		= &init_net.ipv4.sysctl_tcp_comp_sack_delay_ns,
1269 		.maxlen		= sizeof(unsigned long),
1270 		.mode		= 0644,
1271 		.proc_handler	= proc_doulongvec_minmax,
1272 	},
1273 	{
1274 		.procname	= "tcp_comp_sack_slack_ns",
1275 		.data		= &init_net.ipv4.sysctl_tcp_comp_sack_slack_ns,
1276 		.maxlen		= sizeof(unsigned long),
1277 		.mode		= 0644,
1278 		.proc_handler	= proc_doulongvec_minmax,
1279 	},
1280 	{
1281 		.procname	= "tcp_comp_sack_nr",
1282 		.data		= &init_net.ipv4.sysctl_tcp_comp_sack_nr,
1283 		.maxlen		= sizeof(int),
1284 		.mode		= 0644,
1285 		.proc_handler	= proc_dointvec_minmax,
1286 		.extra1		= SYSCTL_ZERO,
1287 		.extra2		= &comp_sack_nr_max,
1288 	},
1289 	{
1290 		.procname       = "tcp_reflect_tos",
1291 		.data           = &init_net.ipv4.sysctl_tcp_reflect_tos,
1292 		.maxlen         = sizeof(u8),
1293 		.mode           = 0644,
1294 		.proc_handler   = proc_dou8vec_minmax,
1295 		.extra1         = SYSCTL_ZERO,
1296 		.extra2         = SYSCTL_ONE,
1297 	},
1298 	{
1299 		.procname	= "udp_rmem_min",
1300 		.data		= &init_net.ipv4.sysctl_udp_rmem_min,
1301 		.maxlen		= sizeof(init_net.ipv4.sysctl_udp_rmem_min),
1302 		.mode		= 0644,
1303 		.proc_handler	= proc_dointvec_minmax,
1304 		.extra1		= SYSCTL_ONE
1305 	},
1306 	{
1307 		.procname	= "udp_wmem_min",
1308 		.data		= &init_net.ipv4.sysctl_udp_wmem_min,
1309 		.maxlen		= sizeof(init_net.ipv4.sysctl_udp_wmem_min),
1310 		.mode		= 0644,
1311 		.proc_handler	= proc_dointvec_minmax,
1312 		.extra1		= SYSCTL_ONE
1313 	},
1314 	{ }
1315 };
1316 
ipv4_sysctl_init_net(struct net * net)1317 static __net_init int ipv4_sysctl_init_net(struct net *net)
1318 {
1319 	struct ctl_table *table;
1320 
1321 	table = ipv4_net_table;
1322 	if (!net_eq(net, &init_net)) {
1323 		int i;
1324 
1325 		table = kmemdup(table, sizeof(ipv4_net_table), GFP_KERNEL);
1326 		if (!table)
1327 			goto err_alloc;
1328 
1329 		for (i = 0; i < ARRAY_SIZE(ipv4_net_table) - 1; i++) {
1330 			if (table[i].data) {
1331 				/* Update the variables to point into
1332 				 * the current struct net
1333 				 */
1334 				table[i].data += (void *)net - (void *)&init_net;
1335 			} else {
1336 				/* Entries without data pointer are global;
1337 				 * Make them read-only in non-init_net ns
1338 				 */
1339 				table[i].mode &= ~0222;
1340 			}
1341 		}
1342 	}
1343 
1344 	net->ipv4.ipv4_hdr = register_net_sysctl(net, "net/ipv4", table);
1345 	if (!net->ipv4.ipv4_hdr)
1346 		goto err_reg;
1347 
1348 	net->ipv4.sysctl_local_reserved_ports = kzalloc(65536 / 8, GFP_KERNEL);
1349 	if (!net->ipv4.sysctl_local_reserved_ports)
1350 		goto err_ports;
1351 
1352 	return 0;
1353 
1354 err_ports:
1355 	unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
1356 err_reg:
1357 	if (!net_eq(net, &init_net))
1358 		kfree(table);
1359 err_alloc:
1360 	return -ENOMEM;
1361 }
1362 
ipv4_sysctl_exit_net(struct net * net)1363 static __net_exit void ipv4_sysctl_exit_net(struct net *net)
1364 {
1365 	struct ctl_table *table;
1366 
1367 	kfree(net->ipv4.sysctl_local_reserved_ports);
1368 	table = net->ipv4.ipv4_hdr->ctl_table_arg;
1369 	unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
1370 	kfree(table);
1371 }
1372 
1373 static __net_initdata struct pernet_operations ipv4_sysctl_ops = {
1374 	.init = ipv4_sysctl_init_net,
1375 	.exit = ipv4_sysctl_exit_net,
1376 };
1377 
sysctl_ipv4_init(void)1378 static __init int sysctl_ipv4_init(void)
1379 {
1380 	struct ctl_table_header *hdr;
1381 
1382 	hdr = register_net_sysctl(&init_net, "net/ipv4", ipv4_table);
1383 	if (!hdr)
1384 		return -ENOMEM;
1385 
1386 	if (register_pernet_subsys(&ipv4_sysctl_ops)) {
1387 		unregister_net_sysctl_table(hdr);
1388 		return -ENOMEM;
1389 	}
1390 
1391 	return 0;
1392 }
1393 
1394 __initcall(sysctl_ipv4_init);
1395