1// SPDX-License-Identifier: GPL-2.0
2/*
3 * sysctl_net_ipv4.c: sysctl interface to net IPV4 subsystem.
4 *
5 * Begun April 1, 1996, Mike Shaver.
6 * Added /proc/sys/net/ipv4 directory entry (empty =) ). [MS]
7 */
8
9#include <linux/sysctl.h>
10#include <linux/seqlock.h>
11#include <linux/init.h>
12#include <linux/slab.h>
13#include <net/icmp.h>
14#include <net/ip.h>
15#include <net/ip_fib.h>
16#include <net/tcp.h>
17#include <net/udp.h>
18#include <net/cipso_ipv4.h>
19#include <net/ping.h>
20#include <net/protocol.h>
21#include <net/netevent.h>
22
23static int tcp_retr1_max = 255;
24static int ip_local_port_range_min[] = { 1, 1 };
25static int ip_local_port_range_max[] = { 65535, 65535 };
26static int tcp_adv_win_scale_min = -31;
27static int tcp_adv_win_scale_max = 31;
28static int tcp_app_win_max = 31;
29static int tcp_min_snd_mss_min = TCP_MIN_SND_MSS;
30static int tcp_min_snd_mss_max = 65535;
31static int ip_privileged_port_min;
32static int ip_privileged_port_max = 65535;
33static int ip_ttl_min = 1;
34static int ip_ttl_max = 255;
35static int tcp_syn_retries_min = 1;
36static int tcp_syn_retries_max = MAX_TCP_SYNCNT;
37static int tcp_syn_linear_timeouts_max = MAX_TCP_SYNCNT;
38static unsigned long ip_ping_group_range_min[] = { 0, 0 };
39static unsigned long ip_ping_group_range_max[] = { GID_T_MAX, GID_T_MAX };
40static u32 u32_max_div_HZ = UINT_MAX / HZ;
41static int one_day_secs = 24 * 3600;
42static u32 fib_multipath_hash_fields_all_mask __maybe_unused =
43	FIB_MULTIPATH_HASH_FIELD_ALL_MASK;
44static unsigned int tcp_child_ehash_entries_max = 16 * 1024 * 1024;
45static unsigned int udp_child_hash_entries_max = UDP_HTABLE_SIZE_MAX;
46static int tcp_plb_max_rounds = 31;
47static int tcp_plb_max_cong_thresh = 256;
48
49/* obsolete */
50static int sysctl_tcp_low_latency __read_mostly;
51
52/* Update system visible IP port range */
53static void set_local_port_range(struct net *net, unsigned int low, unsigned int high)
54{
55	bool same_parity = !((low ^ high) & 1);
56
57	if (same_parity && !net->ipv4.ip_local_ports.warned) {
58		net->ipv4.ip_local_ports.warned = true;
59		pr_err_ratelimited("ip_local_port_range: prefer different parity for start/end values.\n");
60	}
61	WRITE_ONCE(net->ipv4.ip_local_ports.range, high << 16 | low);
62}
63
64/* Validate changes from /proc interface. */
65static int ipv4_local_port_range(struct ctl_table *table, int write,
66				 void *buffer, size_t *lenp, loff_t *ppos)
67{
68	struct net *net = table->data;
69	int ret;
70	int range[2];
71	struct ctl_table tmp = {
72		.data = &range,
73		.maxlen = sizeof(range),
74		.mode = table->mode,
75		.extra1 = &ip_local_port_range_min,
76		.extra2 = &ip_local_port_range_max,
77	};
78
79	inet_get_local_port_range(net, &range[0], &range[1]);
80
81	ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
82
83	if (write && ret == 0) {
84		/* Ensure that the upper limit is not smaller than the lower,
85		 * and that the lower does not encroach upon the privileged
86		 * port limit.
87		 */
88		if ((range[1] < range[0]) ||
89		    (range[0] < READ_ONCE(net->ipv4.sysctl_ip_prot_sock)))
90			ret = -EINVAL;
91		else
92			set_local_port_range(net, range[0], range[1]);
93	}
94
95	return ret;
96}
97
98/* Validate changes from /proc interface. */
99static int ipv4_privileged_ports(struct ctl_table *table, int write,
100				void *buffer, size_t *lenp, loff_t *ppos)
101{
102	struct net *net = container_of(table->data, struct net,
103	    ipv4.sysctl_ip_prot_sock);
104	int ret;
105	int pports;
106	int range[2];
107	struct ctl_table tmp = {
108		.data = &pports,
109		.maxlen = sizeof(pports),
110		.mode = table->mode,
111		.extra1 = &ip_privileged_port_min,
112		.extra2 = &ip_privileged_port_max,
113	};
114
115	pports = READ_ONCE(net->ipv4.sysctl_ip_prot_sock);
116
117	ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
118
119	if (write && ret == 0) {
120		inet_get_local_port_range(net, &range[0], &range[1]);
121		/* Ensure that the local port range doesn't overlap with the
122		 * privileged port range.
123		 */
124		if (range[0] < pports)
125			ret = -EINVAL;
126		else
127			WRITE_ONCE(net->ipv4.sysctl_ip_prot_sock, pports);
128	}
129
130	return ret;
131}
132
133static void inet_get_ping_group_range_table(struct ctl_table *table, kgid_t *low, kgid_t *high)
134{
135	kgid_t *data = table->data;
136	struct net *net =
137		container_of(table->data, struct net, ipv4.ping_group_range.range);
138	unsigned int seq;
139	do {
140		seq = read_seqbegin(&net->ipv4.ping_group_range.lock);
141
142		*low = data[0];
143		*high = data[1];
144	} while (read_seqretry(&net->ipv4.ping_group_range.lock, seq));
145}
146
147/* Update system visible IP port range */
148static void set_ping_group_range(struct ctl_table *table, kgid_t low, kgid_t high)
149{
150	kgid_t *data = table->data;
151	struct net *net =
152		container_of(table->data, struct net, ipv4.ping_group_range.range);
153	write_seqlock(&net->ipv4.ping_group_range.lock);
154	data[0] = low;
155	data[1] = high;
156	write_sequnlock(&net->ipv4.ping_group_range.lock);
157}
158
159/* Validate changes from /proc interface. */
160static int ipv4_ping_group_range(struct ctl_table *table, int write,
161				 void *buffer, size_t *lenp, loff_t *ppos)
162{
163	struct user_namespace *user_ns = current_user_ns();
164	int ret;
165	unsigned long urange[2];
166	kgid_t low, high;
167	struct ctl_table tmp = {
168		.data = &urange,
169		.maxlen = sizeof(urange),
170		.mode = table->mode,
171		.extra1 = &ip_ping_group_range_min,
172		.extra2 = &ip_ping_group_range_max,
173	};
174
175	inet_get_ping_group_range_table(table, &low, &high);
176	urange[0] = from_kgid_munged(user_ns, low);
177	urange[1] = from_kgid_munged(user_ns, high);
178	ret = proc_doulongvec_minmax(&tmp, write, buffer, lenp, ppos);
179
180	if (write && ret == 0) {
181		low = make_kgid(user_ns, urange[0]);
182		high = make_kgid(user_ns, urange[1]);
183		if (!gid_valid(low) || !gid_valid(high))
184			return -EINVAL;
185		if (urange[1] < urange[0] || gid_lt(high, low)) {
186			low = make_kgid(&init_user_ns, 1);
187			high = make_kgid(&init_user_ns, 0);
188		}
189		set_ping_group_range(table, low, high);
190	}
191
192	return ret;
193}
194
195static int ipv4_fwd_update_priority(struct ctl_table *table, int write,
196				    void *buffer, size_t *lenp, loff_t *ppos)
197{
198	struct net *net;
199	int ret;
200
201	net = container_of(table->data, struct net,
202			   ipv4.sysctl_ip_fwd_update_priority);
203	ret = proc_dou8vec_minmax(table, write, buffer, lenp, ppos);
204	if (write && ret == 0)
205		call_netevent_notifiers(NETEVENT_IPV4_FWD_UPDATE_PRIORITY_UPDATE,
206					net);
207
208	return ret;
209}
210
211static int proc_tcp_congestion_control(struct ctl_table *ctl, int write,
212				       void *buffer, size_t *lenp, loff_t *ppos)
213{
214	struct net *net = container_of(ctl->data, struct net,
215				       ipv4.tcp_congestion_control);
216	char val[TCP_CA_NAME_MAX];
217	struct ctl_table tbl = {
218		.data = val,
219		.maxlen = TCP_CA_NAME_MAX,
220	};
221	int ret;
222
223	tcp_get_default_congestion_control(net, val);
224
225	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
226	if (write && ret == 0)
227		ret = tcp_set_default_congestion_control(net, val);
228	return ret;
229}
230
231static int proc_tcp_available_congestion_control(struct ctl_table *ctl,
232						 int write, void *buffer,
233						 size_t *lenp, loff_t *ppos)
234{
235	struct ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX, };
236	int ret;
237
238	tbl.data = kmalloc(tbl.maxlen, GFP_USER);
239	if (!tbl.data)
240		return -ENOMEM;
241	tcp_get_available_congestion_control(tbl.data, TCP_CA_BUF_MAX);
242	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
243	kfree(tbl.data);
244	return ret;
245}
246
247static int proc_allowed_congestion_control(struct ctl_table *ctl,
248					   int write, void *buffer,
249					   size_t *lenp, loff_t *ppos)
250{
251	struct ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX };
252	int ret;
253
254	tbl.data = kmalloc(tbl.maxlen, GFP_USER);
255	if (!tbl.data)
256		return -ENOMEM;
257
258	tcp_get_allowed_congestion_control(tbl.data, tbl.maxlen);
259	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
260	if (write && ret == 0)
261		ret = tcp_set_allowed_congestion_control(tbl.data);
262	kfree(tbl.data);
263	return ret;
264}
265
266static int sscanf_key(char *buf, __le32 *key)
267{
268	u32 user_key[4];
269	int i, ret = 0;
270
271	if (sscanf(buf, "%x-%x-%x-%x", user_key, user_key + 1,
272		   user_key + 2, user_key + 3) != 4) {
273		ret = -EINVAL;
274	} else {
275		for (i = 0; i < ARRAY_SIZE(user_key); i++)
276			key[i] = cpu_to_le32(user_key[i]);
277	}
278	pr_debug("proc TFO key set 0x%x-%x-%x-%x <- 0x%s: %u\n",
279		 user_key[0], user_key[1], user_key[2], user_key[3], buf, ret);
280
281	return ret;
282}
283
284static int proc_tcp_fastopen_key(struct ctl_table *table, int write,
285				 void *buffer, size_t *lenp, loff_t *ppos)
286{
287	struct net *net = container_of(table->data, struct net,
288	    ipv4.sysctl_tcp_fastopen);
289	/* maxlen to print the list of keys in hex (*2), with dashes
290	 * separating doublewords and a comma in between keys.
291	 */
292	struct ctl_table tbl = { .maxlen = ((TCP_FASTOPEN_KEY_LENGTH *
293					    2 * TCP_FASTOPEN_KEY_MAX) +
294					    (TCP_FASTOPEN_KEY_MAX * 5)) };
295	u32 user_key[TCP_FASTOPEN_KEY_BUF_LENGTH / sizeof(u32)];
296	__le32 key[TCP_FASTOPEN_KEY_BUF_LENGTH / sizeof(__le32)];
297	char *backup_data;
298	int ret, i = 0, off = 0, n_keys;
299
300	tbl.data = kmalloc(tbl.maxlen, GFP_KERNEL);
301	if (!tbl.data)
302		return -ENOMEM;
303
304	n_keys = tcp_fastopen_get_cipher(net, NULL, (u64 *)key);
305	if (!n_keys) {
306		memset(&key[0], 0, TCP_FASTOPEN_KEY_LENGTH);
307		n_keys = 1;
308	}
309
310	for (i = 0; i < n_keys * 4; i++)
311		user_key[i] = le32_to_cpu(key[i]);
312
313	for (i = 0; i < n_keys; i++) {
314		off += snprintf(tbl.data + off, tbl.maxlen - off,
315				"%08x-%08x-%08x-%08x",
316				user_key[i * 4],
317				user_key[i * 4 + 1],
318				user_key[i * 4 + 2],
319				user_key[i * 4 + 3]);
320
321		if (WARN_ON_ONCE(off >= tbl.maxlen - 1))
322			break;
323
324		if (i + 1 < n_keys)
325			off += snprintf(tbl.data + off, tbl.maxlen - off, ",");
326	}
327
328	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
329
330	if (write && ret == 0) {
331		backup_data = strchr(tbl.data, ',');
332		if (backup_data) {
333			*backup_data = '\0';
334			backup_data++;
335		}
336		if (sscanf_key(tbl.data, key)) {
337			ret = -EINVAL;
338			goto bad_key;
339		}
340		if (backup_data) {
341			if (sscanf_key(backup_data, key + 4)) {
342				ret = -EINVAL;
343				goto bad_key;
344			}
345		}
346		tcp_fastopen_reset_cipher(net, NULL, key,
347					  backup_data ? key + 4 : NULL);
348	}
349
350bad_key:
351	kfree(tbl.data);
352	return ret;
353}
354
355static int proc_tfo_blackhole_detect_timeout(struct ctl_table *table,
356					     int write, void *buffer,
357					     size_t *lenp, loff_t *ppos)
358{
359	struct net *net = container_of(table->data, struct net,
360	    ipv4.sysctl_tcp_fastopen_blackhole_timeout);
361	int ret;
362
363	ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
364	if (write && ret == 0)
365		atomic_set(&net->ipv4.tfo_active_disable_times, 0);
366
367	return ret;
368}
369
370static int proc_tcp_available_ulp(struct ctl_table *ctl,
371				  int write, void *buffer, size_t *lenp,
372				  loff_t *ppos)
373{
374	struct ctl_table tbl = { .maxlen = TCP_ULP_BUF_MAX, };
375	int ret;
376
377	tbl.data = kmalloc(tbl.maxlen, GFP_USER);
378	if (!tbl.data)
379		return -ENOMEM;
380	tcp_get_available_ulp(tbl.data, TCP_ULP_BUF_MAX);
381	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
382	kfree(tbl.data);
383
384	return ret;
385}
386
387static int proc_tcp_ehash_entries(struct ctl_table *table, int write,
388				  void *buffer, size_t *lenp, loff_t *ppos)
389{
390	struct net *net = container_of(table->data, struct net,
391				       ipv4.sysctl_tcp_child_ehash_entries);
392	struct inet_hashinfo *hinfo = net->ipv4.tcp_death_row.hashinfo;
393	int tcp_ehash_entries;
394	struct ctl_table tbl;
395
396	tcp_ehash_entries = hinfo->ehash_mask + 1;
397
398	/* A negative number indicates that the child netns
399	 * shares the global ehash.
400	 */
401	if (!net_eq(net, &init_net) && !hinfo->pernet)
402		tcp_ehash_entries *= -1;
403
404	memset(&tbl, 0, sizeof(tbl));
405	tbl.data = &tcp_ehash_entries;
406	tbl.maxlen = sizeof(int);
407
408	return proc_dointvec(&tbl, write, buffer, lenp, ppos);
409}
410
411static int proc_udp_hash_entries(struct ctl_table *table, int write,
412				 void *buffer, size_t *lenp, loff_t *ppos)
413{
414	struct net *net = container_of(table->data, struct net,
415				       ipv4.sysctl_udp_child_hash_entries);
416	int udp_hash_entries;
417	struct ctl_table tbl;
418
419	udp_hash_entries = net->ipv4.udp_table->mask + 1;
420
421	/* A negative number indicates that the child netns
422	 * shares the global udp_table.
423	 */
424	if (!net_eq(net, &init_net) && net->ipv4.udp_table == &udp_table)
425		udp_hash_entries *= -1;
426
427	memset(&tbl, 0, sizeof(tbl));
428	tbl.data = &udp_hash_entries;
429	tbl.maxlen = sizeof(int);
430
431	return proc_dointvec(&tbl, write, buffer, lenp, ppos);
432}
433
434#ifdef CONFIG_IP_ROUTE_MULTIPATH
435static int proc_fib_multipath_hash_policy(struct ctl_table *table, int write,
436					  void *buffer, size_t *lenp,
437					  loff_t *ppos)
438{
439	struct net *net = container_of(table->data, struct net,
440	    ipv4.sysctl_fib_multipath_hash_policy);
441	int ret;
442
443	ret = proc_dou8vec_minmax(table, write, buffer, lenp, ppos);
444	if (write && ret == 0)
445		call_netevent_notifiers(NETEVENT_IPV4_MPATH_HASH_UPDATE, net);
446
447	return ret;
448}
449
450static int proc_fib_multipath_hash_fields(struct ctl_table *table, int write,
451					  void *buffer, size_t *lenp,
452					  loff_t *ppos)
453{
454	struct net *net;
455	int ret;
456
457	net = container_of(table->data, struct net,
458			   ipv4.sysctl_fib_multipath_hash_fields);
459	ret = proc_douintvec_minmax(table, write, buffer, lenp, ppos);
460	if (write && ret == 0)
461		call_netevent_notifiers(NETEVENT_IPV4_MPATH_HASH_UPDATE, net);
462
463	return ret;
464}
465#endif
466
467static struct ctl_table ipv4_table[] = {
468	{
469		.procname	= "tcp_max_orphans",
470		.data		= &sysctl_tcp_max_orphans,
471		.maxlen		= sizeof(int),
472		.mode		= 0644,
473		.proc_handler	= proc_dointvec
474	},
475	{
476		.procname	= "inet_peer_threshold",
477		.data		= &inet_peer_threshold,
478		.maxlen		= sizeof(int),
479		.mode		= 0644,
480		.proc_handler	= proc_dointvec
481	},
482	{
483		.procname	= "inet_peer_minttl",
484		.data		= &inet_peer_minttl,
485		.maxlen		= sizeof(int),
486		.mode		= 0644,
487		.proc_handler	= proc_dointvec_jiffies,
488	},
489	{
490		.procname	= "inet_peer_maxttl",
491		.data		= &inet_peer_maxttl,
492		.maxlen		= sizeof(int),
493		.mode		= 0644,
494		.proc_handler	= proc_dointvec_jiffies,
495	},
496	{
497		.procname	= "tcp_mem",
498		.maxlen		= sizeof(sysctl_tcp_mem),
499		.data		= &sysctl_tcp_mem,
500		.mode		= 0644,
501		.proc_handler	= proc_doulongvec_minmax,
502	},
503	{
504		.procname	= "tcp_low_latency",
505		.data		= &sysctl_tcp_low_latency,
506		.maxlen		= sizeof(int),
507		.mode		= 0644,
508		.proc_handler	= proc_dointvec
509	},
510#ifdef CONFIG_NETLABEL
511	{
512		.procname	= "cipso_cache_enable",
513		.data		= &cipso_v4_cache_enabled,
514		.maxlen		= sizeof(int),
515		.mode		= 0644,
516		.proc_handler	= proc_dointvec,
517	},
518	{
519		.procname	= "cipso_cache_bucket_size",
520		.data		= &cipso_v4_cache_bucketsize,
521		.maxlen		= sizeof(int),
522		.mode		= 0644,
523		.proc_handler	= proc_dointvec,
524	},
525	{
526		.procname	= "cipso_rbm_optfmt",
527		.data		= &cipso_v4_rbm_optfmt,
528		.maxlen		= sizeof(int),
529		.mode		= 0644,
530		.proc_handler	= proc_dointvec,
531	},
532	{
533		.procname	= "cipso_rbm_strictvalid",
534		.data		= &cipso_v4_rbm_strictvalid,
535		.maxlen		= sizeof(int),
536		.mode		= 0644,
537		.proc_handler	= proc_dointvec,
538	},
539#endif /* CONFIG_NETLABEL */
540	{
541		.procname	= "tcp_available_ulp",
542		.maxlen		= TCP_ULP_BUF_MAX,
543		.mode		= 0444,
544		.proc_handler   = proc_tcp_available_ulp,
545	},
546	{
547		.procname	= "icmp_msgs_per_sec",
548		.data		= &sysctl_icmp_msgs_per_sec,
549		.maxlen		= sizeof(int),
550		.mode		= 0644,
551		.proc_handler	= proc_dointvec_minmax,
552		.extra1		= SYSCTL_ZERO,
553	},
554	{
555		.procname	= "icmp_msgs_burst",
556		.data		= &sysctl_icmp_msgs_burst,
557		.maxlen		= sizeof(int),
558		.mode		= 0644,
559		.proc_handler	= proc_dointvec_minmax,
560		.extra1		= SYSCTL_ZERO,
561	},
562	{
563		.procname	= "udp_mem",
564		.data		= &sysctl_udp_mem,
565		.maxlen		= sizeof(sysctl_udp_mem),
566		.mode		= 0644,
567		.proc_handler	= proc_doulongvec_minmax,
568	},
569	{
570		.procname	= "fib_sync_mem",
571		.data		= &sysctl_fib_sync_mem,
572		.maxlen		= sizeof(sysctl_fib_sync_mem),
573		.mode		= 0644,
574		.proc_handler	= proc_douintvec_minmax,
575		.extra1		= &sysctl_fib_sync_mem_min,
576		.extra2		= &sysctl_fib_sync_mem_max,
577	},
578	{ }
579};
580
581static struct ctl_table ipv4_net_table[] = {
582	{
583		.procname	= "tcp_max_tw_buckets",
584		.data		= &init_net.ipv4.tcp_death_row.sysctl_max_tw_buckets,
585		.maxlen		= sizeof(int),
586		.mode		= 0644,
587		.proc_handler	= proc_dointvec
588	},
589	{
590		.procname	= "icmp_echo_ignore_all",
591		.data		= &init_net.ipv4.sysctl_icmp_echo_ignore_all,
592		.maxlen		= sizeof(u8),
593		.mode		= 0644,
594		.proc_handler	= proc_dou8vec_minmax,
595		.extra1		= SYSCTL_ZERO,
596		.extra2		= SYSCTL_ONE
597	},
598	{
599		.procname	= "icmp_echo_enable_probe",
600		.data		= &init_net.ipv4.sysctl_icmp_echo_enable_probe,
601		.maxlen		= sizeof(u8),
602		.mode		= 0644,
603		.proc_handler	= proc_dou8vec_minmax,
604		.extra1		= SYSCTL_ZERO,
605		.extra2		= SYSCTL_ONE
606	},
607	{
608		.procname	= "icmp_echo_ignore_broadcasts",
609		.data		= &init_net.ipv4.sysctl_icmp_echo_ignore_broadcasts,
610		.maxlen		= sizeof(u8),
611		.mode		= 0644,
612		.proc_handler	= proc_dou8vec_minmax,
613		.extra1		= SYSCTL_ZERO,
614		.extra2		= SYSCTL_ONE
615	},
616	{
617		.procname	= "icmp_ignore_bogus_error_responses",
618		.data		= &init_net.ipv4.sysctl_icmp_ignore_bogus_error_responses,
619		.maxlen		= sizeof(u8),
620		.mode		= 0644,
621		.proc_handler	= proc_dou8vec_minmax,
622		.extra1		= SYSCTL_ZERO,
623		.extra2		= SYSCTL_ONE
624	},
625	{
626		.procname	= "icmp_errors_use_inbound_ifaddr",
627		.data		= &init_net.ipv4.sysctl_icmp_errors_use_inbound_ifaddr,
628		.maxlen		= sizeof(u8),
629		.mode		= 0644,
630		.proc_handler	= proc_dou8vec_minmax,
631		.extra1		= SYSCTL_ZERO,
632		.extra2		= SYSCTL_ONE
633	},
634	{
635		.procname	= "icmp_ratelimit",
636		.data		= &init_net.ipv4.sysctl_icmp_ratelimit,
637		.maxlen		= sizeof(int),
638		.mode		= 0644,
639		.proc_handler	= proc_dointvec_ms_jiffies,
640	},
641	{
642		.procname	= "icmp_ratemask",
643		.data		= &init_net.ipv4.sysctl_icmp_ratemask,
644		.maxlen		= sizeof(int),
645		.mode		= 0644,
646		.proc_handler	= proc_dointvec
647	},
648	{
649		.procname	= "ping_group_range",
650		.data		= &init_net.ipv4.ping_group_range.range,
651		.maxlen		= sizeof(gid_t)*2,
652		.mode		= 0644,
653		.proc_handler	= ipv4_ping_group_range,
654	},
655#ifdef CONFIG_NET_L3_MASTER_DEV
656	{
657		.procname	= "raw_l3mdev_accept",
658		.data		= &init_net.ipv4.sysctl_raw_l3mdev_accept,
659		.maxlen		= sizeof(u8),
660		.mode		= 0644,
661		.proc_handler	= proc_dou8vec_minmax,
662		.extra1		= SYSCTL_ZERO,
663		.extra2		= SYSCTL_ONE,
664	},
665#endif
666	{
667		.procname	= "tcp_ecn",
668		.data		= &init_net.ipv4.sysctl_tcp_ecn,
669		.maxlen		= sizeof(u8),
670		.mode		= 0644,
671		.proc_handler	= proc_dou8vec_minmax,
672		.extra1		= SYSCTL_ZERO,
673		.extra2		= SYSCTL_TWO,
674	},
675	{
676		.procname	= "tcp_ecn_fallback",
677		.data		= &init_net.ipv4.sysctl_tcp_ecn_fallback,
678		.maxlen		= sizeof(u8),
679		.mode		= 0644,
680		.proc_handler	= proc_dou8vec_minmax,
681		.extra1		= SYSCTL_ZERO,
682		.extra2		= SYSCTL_ONE,
683	},
684	{
685		.procname	= "ip_dynaddr",
686		.data		= &init_net.ipv4.sysctl_ip_dynaddr,
687		.maxlen		= sizeof(u8),
688		.mode		= 0644,
689		.proc_handler	= proc_dou8vec_minmax,
690	},
691	{
692		.procname	= "ip_early_demux",
693		.data		= &init_net.ipv4.sysctl_ip_early_demux,
694		.maxlen		= sizeof(u8),
695		.mode		= 0644,
696		.proc_handler	= proc_dou8vec_minmax,
697	},
698	{
699		.procname       = "udp_early_demux",
700		.data           = &init_net.ipv4.sysctl_udp_early_demux,
701		.maxlen         = sizeof(u8),
702		.mode           = 0644,
703		.proc_handler   = proc_dou8vec_minmax,
704	},
705	{
706		.procname       = "tcp_early_demux",
707		.data           = &init_net.ipv4.sysctl_tcp_early_demux,
708		.maxlen         = sizeof(u8),
709		.mode           = 0644,
710		.proc_handler   = proc_dou8vec_minmax,
711	},
712	{
713		.procname       = "nexthop_compat_mode",
714		.data           = &init_net.ipv4.sysctl_nexthop_compat_mode,
715		.maxlen         = sizeof(u8),
716		.mode           = 0644,
717		.proc_handler   = proc_dou8vec_minmax,
718		.extra1		= SYSCTL_ZERO,
719		.extra2		= SYSCTL_ONE,
720	},
721	{
722		.procname	= "ip_default_ttl",
723		.data		= &init_net.ipv4.sysctl_ip_default_ttl,
724		.maxlen		= sizeof(u8),
725		.mode		= 0644,
726		.proc_handler	= proc_dou8vec_minmax,
727		.extra1		= &ip_ttl_min,
728		.extra2		= &ip_ttl_max,
729	},
730	{
731		.procname	= "ip_local_port_range",
732		.maxlen		= 0,
733		.data		= &init_net,
734		.mode		= 0644,
735		.proc_handler	= ipv4_local_port_range,
736	},
737	{
738		.procname	= "ip_local_reserved_ports",
739		.data		= &init_net.ipv4.sysctl_local_reserved_ports,
740		.maxlen		= 65536,
741		.mode		= 0644,
742		.proc_handler	= proc_do_large_bitmap,
743	},
744	{
745		.procname	= "ip_no_pmtu_disc",
746		.data		= &init_net.ipv4.sysctl_ip_no_pmtu_disc,
747		.maxlen		= sizeof(u8),
748		.mode		= 0644,
749		.proc_handler	= proc_dou8vec_minmax,
750	},
751	{
752		.procname	= "ip_forward_use_pmtu",
753		.data		= &init_net.ipv4.sysctl_ip_fwd_use_pmtu,
754		.maxlen		= sizeof(u8),
755		.mode		= 0644,
756		.proc_handler	= proc_dou8vec_minmax,
757	},
758	{
759		.procname	= "ip_forward_update_priority",
760		.data		= &init_net.ipv4.sysctl_ip_fwd_update_priority,
761		.maxlen		= sizeof(u8),
762		.mode		= 0644,
763		.proc_handler   = ipv4_fwd_update_priority,
764		.extra1		= SYSCTL_ZERO,
765		.extra2		= SYSCTL_ONE,
766	},
767	{
768		.procname	= "ip_nonlocal_bind",
769		.data		= &init_net.ipv4.sysctl_ip_nonlocal_bind,
770		.maxlen		= sizeof(u8),
771		.mode		= 0644,
772		.proc_handler	= proc_dou8vec_minmax,
773	},
774	{
775		.procname	= "ip_autobind_reuse",
776		.data		= &init_net.ipv4.sysctl_ip_autobind_reuse,
777		.maxlen		= sizeof(u8),
778		.mode		= 0644,
779		.proc_handler	= proc_dou8vec_minmax,
780		.extra1         = SYSCTL_ZERO,
781		.extra2         = SYSCTL_ONE,
782	},
783	{
784		.procname	= "fwmark_reflect",
785		.data		= &init_net.ipv4.sysctl_fwmark_reflect,
786		.maxlen		= sizeof(u8),
787		.mode		= 0644,
788		.proc_handler	= proc_dou8vec_minmax,
789	},
790	{
791		.procname	= "tcp_fwmark_accept",
792		.data		= &init_net.ipv4.sysctl_tcp_fwmark_accept,
793		.maxlen		= sizeof(u8),
794		.mode		= 0644,
795		.proc_handler	= proc_dou8vec_minmax,
796	},
797#ifdef CONFIG_NET_L3_MASTER_DEV
798	{
799		.procname	= "tcp_l3mdev_accept",
800		.data		= &init_net.ipv4.sysctl_tcp_l3mdev_accept,
801		.maxlen		= sizeof(u8),
802		.mode		= 0644,
803		.proc_handler	= proc_dou8vec_minmax,
804		.extra1		= SYSCTL_ZERO,
805		.extra2		= SYSCTL_ONE,
806	},
807#endif
808	{
809		.procname	= "tcp_mtu_probing",
810		.data		= &init_net.ipv4.sysctl_tcp_mtu_probing,
811		.maxlen		= sizeof(u8),
812		.mode		= 0644,
813		.proc_handler	= proc_dou8vec_minmax,
814	},
815	{
816		.procname	= "tcp_base_mss",
817		.data		= &init_net.ipv4.sysctl_tcp_base_mss,
818		.maxlen		= sizeof(int),
819		.mode		= 0644,
820		.proc_handler	= proc_dointvec,
821	},
822	{
823		.procname	= "tcp_min_snd_mss",
824		.data		= &init_net.ipv4.sysctl_tcp_min_snd_mss,
825		.maxlen		= sizeof(int),
826		.mode		= 0644,
827		.proc_handler	= proc_dointvec_minmax,
828		.extra1		= &tcp_min_snd_mss_min,
829		.extra2		= &tcp_min_snd_mss_max,
830	},
831	{
832		.procname	= "tcp_mtu_probe_floor",
833		.data		= &init_net.ipv4.sysctl_tcp_mtu_probe_floor,
834		.maxlen		= sizeof(int),
835		.mode		= 0644,
836		.proc_handler	= proc_dointvec_minmax,
837		.extra1		= &tcp_min_snd_mss_min,
838		.extra2		= &tcp_min_snd_mss_max,
839	},
840	{
841		.procname	= "tcp_probe_threshold",
842		.data		= &init_net.ipv4.sysctl_tcp_probe_threshold,
843		.maxlen		= sizeof(int),
844		.mode		= 0644,
845		.proc_handler	= proc_dointvec,
846	},
847	{
848		.procname	= "tcp_probe_interval",
849		.data		= &init_net.ipv4.sysctl_tcp_probe_interval,
850		.maxlen		= sizeof(u32),
851		.mode		= 0644,
852		.proc_handler	= proc_douintvec_minmax,
853		.extra2		= &u32_max_div_HZ,
854	},
855	{
856		.procname	= "igmp_link_local_mcast_reports",
857		.data		= &init_net.ipv4.sysctl_igmp_llm_reports,
858		.maxlen		= sizeof(u8),
859		.mode		= 0644,
860		.proc_handler	= proc_dou8vec_minmax,
861	},
862	{
863		.procname	= "igmp_max_memberships",
864		.data		= &init_net.ipv4.sysctl_igmp_max_memberships,
865		.maxlen		= sizeof(int),
866		.mode		= 0644,
867		.proc_handler	= proc_dointvec
868	},
869	{
870		.procname	= "igmp_max_msf",
871		.data		= &init_net.ipv4.sysctl_igmp_max_msf,
872		.maxlen		= sizeof(int),
873		.mode		= 0644,
874		.proc_handler	= proc_dointvec
875	},
876#ifdef CONFIG_IP_MULTICAST
877	{
878		.procname	= "igmp_qrv",
879		.data		= &init_net.ipv4.sysctl_igmp_qrv,
880		.maxlen		= sizeof(int),
881		.mode		= 0644,
882		.proc_handler	= proc_dointvec_minmax,
883		.extra1		= SYSCTL_ONE
884	},
885#endif
886	{
887		.procname	= "tcp_congestion_control",
888		.data		= &init_net.ipv4.tcp_congestion_control,
889		.mode		= 0644,
890		.maxlen		= TCP_CA_NAME_MAX,
891		.proc_handler	= proc_tcp_congestion_control,
892	},
893	{
894		.procname	= "tcp_available_congestion_control",
895		.maxlen		= TCP_CA_BUF_MAX,
896		.mode		= 0444,
897		.proc_handler   = proc_tcp_available_congestion_control,
898	},
899	{
900		.procname	= "tcp_allowed_congestion_control",
901		.maxlen		= TCP_CA_BUF_MAX,
902		.mode		= 0644,
903		.proc_handler   = proc_allowed_congestion_control,
904	},
905	{
906		.procname	= "tcp_keepalive_time",
907		.data		= &init_net.ipv4.sysctl_tcp_keepalive_time,
908		.maxlen		= sizeof(int),
909		.mode		= 0644,
910		.proc_handler	= proc_dointvec_jiffies,
911	},
912	{
913		.procname	= "tcp_keepalive_probes",
914		.data		= &init_net.ipv4.sysctl_tcp_keepalive_probes,
915		.maxlen		= sizeof(u8),
916		.mode		= 0644,
917		.proc_handler	= proc_dou8vec_minmax,
918	},
919	{
920		.procname	= "tcp_keepalive_intvl",
921		.data		= &init_net.ipv4.sysctl_tcp_keepalive_intvl,
922		.maxlen		= sizeof(int),
923		.mode		= 0644,
924		.proc_handler	= proc_dointvec_jiffies,
925	},
926	{
927		.procname	= "tcp_syn_retries",
928		.data		= &init_net.ipv4.sysctl_tcp_syn_retries,
929		.maxlen		= sizeof(u8),
930		.mode		= 0644,
931		.proc_handler	= proc_dou8vec_minmax,
932		.extra1		= &tcp_syn_retries_min,
933		.extra2		= &tcp_syn_retries_max
934	},
935	{
936		.procname	= "tcp_synack_retries",
937		.data		= &init_net.ipv4.sysctl_tcp_synack_retries,
938		.maxlen		= sizeof(u8),
939		.mode		= 0644,
940		.proc_handler	= proc_dou8vec_minmax,
941	},
942#ifdef CONFIG_SYN_COOKIES
943	{
944		.procname	= "tcp_syncookies",
945		.data		= &init_net.ipv4.sysctl_tcp_syncookies,
946		.maxlen		= sizeof(u8),
947		.mode		= 0644,
948		.proc_handler	= proc_dou8vec_minmax,
949	},
950#endif
951	{
952		.procname	= "tcp_migrate_req",
953		.data		= &init_net.ipv4.sysctl_tcp_migrate_req,
954		.maxlen		= sizeof(u8),
955		.mode		= 0644,
956		.proc_handler	= proc_dou8vec_minmax,
957		.extra1		= SYSCTL_ZERO,
958		.extra2		= SYSCTL_ONE
959	},
960	{
961		.procname	= "tcp_reordering",
962		.data		= &init_net.ipv4.sysctl_tcp_reordering,
963		.maxlen		= sizeof(int),
964		.mode		= 0644,
965		.proc_handler	= proc_dointvec
966	},
967	{
968		.procname	= "tcp_retries1",
969		.data		= &init_net.ipv4.sysctl_tcp_retries1,
970		.maxlen		= sizeof(u8),
971		.mode		= 0644,
972		.proc_handler	= proc_dou8vec_minmax,
973		.extra2		= &tcp_retr1_max
974	},
975	{
976		.procname	= "tcp_retries2",
977		.data		= &init_net.ipv4.sysctl_tcp_retries2,
978		.maxlen		= sizeof(u8),
979		.mode		= 0644,
980		.proc_handler	= proc_dou8vec_minmax,
981	},
982	{
983		.procname	= "tcp_orphan_retries",
984		.data		= &init_net.ipv4.sysctl_tcp_orphan_retries,
985		.maxlen		= sizeof(u8),
986		.mode		= 0644,
987		.proc_handler	= proc_dou8vec_minmax,
988	},
989	{
990		.procname	= "tcp_fin_timeout",
991		.data		= &init_net.ipv4.sysctl_tcp_fin_timeout,
992		.maxlen		= sizeof(int),
993		.mode		= 0644,
994		.proc_handler	= proc_dointvec_jiffies,
995	},
996	{
997		.procname	= "tcp_notsent_lowat",
998		.data		= &init_net.ipv4.sysctl_tcp_notsent_lowat,
999		.maxlen		= sizeof(unsigned int),
1000		.mode		= 0644,
1001		.proc_handler	= proc_douintvec,
1002	},
1003	{
1004		.procname	= "tcp_tw_reuse",
1005		.data		= &init_net.ipv4.sysctl_tcp_tw_reuse,
1006		.maxlen		= sizeof(u8),
1007		.mode		= 0644,
1008		.proc_handler	= proc_dou8vec_minmax,
1009		.extra1		= SYSCTL_ZERO,
1010		.extra2		= SYSCTL_TWO,
1011	},
1012	{
1013		.procname	= "tcp_max_syn_backlog",
1014		.data		= &init_net.ipv4.sysctl_max_syn_backlog,
1015		.maxlen		= sizeof(int),
1016		.mode		= 0644,
1017		.proc_handler	= proc_dointvec
1018	},
1019	{
1020		.procname	= "tcp_fastopen",
1021		.data		= &init_net.ipv4.sysctl_tcp_fastopen,
1022		.maxlen		= sizeof(int),
1023		.mode		= 0644,
1024		.proc_handler	= proc_dointvec,
1025	},
1026	{
1027		.procname	= "tcp_fastopen_key",
1028		.mode		= 0600,
1029		.data		= &init_net.ipv4.sysctl_tcp_fastopen,
1030		/* maxlen to print the list of keys in hex (*2), with dashes
1031		 * separating doublewords and a comma in between keys.
1032		 */
1033		.maxlen		= ((TCP_FASTOPEN_KEY_LENGTH *
1034				   2 * TCP_FASTOPEN_KEY_MAX) +
1035				   (TCP_FASTOPEN_KEY_MAX * 5)),
1036		.proc_handler	= proc_tcp_fastopen_key,
1037	},
1038	{
1039		.procname	= "tcp_fastopen_blackhole_timeout_sec",
1040		.data		= &init_net.ipv4.sysctl_tcp_fastopen_blackhole_timeout,
1041		.maxlen		= sizeof(int),
1042		.mode		= 0644,
1043		.proc_handler	= proc_tfo_blackhole_detect_timeout,
1044		.extra1		= SYSCTL_ZERO,
1045	},
1046#ifdef CONFIG_IP_ROUTE_MULTIPATH
1047	{
1048		.procname	= "fib_multipath_use_neigh",
1049		.data		= &init_net.ipv4.sysctl_fib_multipath_use_neigh,
1050		.maxlen		= sizeof(u8),
1051		.mode		= 0644,
1052		.proc_handler	= proc_dou8vec_minmax,
1053		.extra1		= SYSCTL_ZERO,
1054		.extra2		= SYSCTL_ONE,
1055	},
1056	{
1057		.procname	= "fib_multipath_hash_policy",
1058		.data		= &init_net.ipv4.sysctl_fib_multipath_hash_policy,
1059		.maxlen		= sizeof(u8),
1060		.mode		= 0644,
1061		.proc_handler	= proc_fib_multipath_hash_policy,
1062		.extra1		= SYSCTL_ZERO,
1063		.extra2		= SYSCTL_THREE,
1064	},
1065	{
1066		.procname	= "fib_multipath_hash_fields",
1067		.data		= &init_net.ipv4.sysctl_fib_multipath_hash_fields,
1068		.maxlen		= sizeof(u32),
1069		.mode		= 0644,
1070		.proc_handler	= proc_fib_multipath_hash_fields,
1071		.extra1		= SYSCTL_ONE,
1072		.extra2		= &fib_multipath_hash_fields_all_mask,
1073	},
1074#endif
1075	{
1076		.procname	= "ip_unprivileged_port_start",
1077		.maxlen		= sizeof(int),
1078		.data		= &init_net.ipv4.sysctl_ip_prot_sock,
1079		.mode		= 0644,
1080		.proc_handler	= ipv4_privileged_ports,
1081	},
1082#ifdef CONFIG_NET_L3_MASTER_DEV
1083	{
1084		.procname	= "udp_l3mdev_accept",
1085		.data		= &init_net.ipv4.sysctl_udp_l3mdev_accept,
1086		.maxlen		= sizeof(u8),
1087		.mode		= 0644,
1088		.proc_handler	= proc_dou8vec_minmax,
1089		.extra1		= SYSCTL_ZERO,
1090		.extra2		= SYSCTL_ONE,
1091	},
1092#endif
1093	{
1094		.procname	= "tcp_sack",
1095		.data		= &init_net.ipv4.sysctl_tcp_sack,
1096		.maxlen		= sizeof(u8),
1097		.mode		= 0644,
1098		.proc_handler	= proc_dou8vec_minmax,
1099	},
1100	{
1101		.procname	= "tcp_window_scaling",
1102		.data		= &init_net.ipv4.sysctl_tcp_window_scaling,
1103		.maxlen		= sizeof(u8),
1104		.mode		= 0644,
1105		.proc_handler	= proc_dou8vec_minmax,
1106	},
1107	{
1108		.procname	= "tcp_timestamps",
1109		.data		= &init_net.ipv4.sysctl_tcp_timestamps,
1110		.maxlen		= sizeof(u8),
1111		.mode		= 0644,
1112		.proc_handler	= proc_dou8vec_minmax,
1113	},
1114	{
1115		.procname	= "tcp_early_retrans",
1116		.data		= &init_net.ipv4.sysctl_tcp_early_retrans,
1117		.maxlen		= sizeof(u8),
1118		.mode		= 0644,
1119		.proc_handler	= proc_dou8vec_minmax,
1120		.extra1		= SYSCTL_ZERO,
1121		.extra2		= SYSCTL_FOUR,
1122	},
1123	{
1124		.procname	= "tcp_recovery",
1125		.data		= &init_net.ipv4.sysctl_tcp_recovery,
1126		.maxlen		= sizeof(u8),
1127		.mode		= 0644,
1128		.proc_handler	= proc_dou8vec_minmax,
1129	},
1130	{
1131		.procname       = "tcp_thin_linear_timeouts",
1132		.data           = &init_net.ipv4.sysctl_tcp_thin_linear_timeouts,
1133		.maxlen         = sizeof(u8),
1134		.mode           = 0644,
1135		.proc_handler   = proc_dou8vec_minmax,
1136	},
1137	{
1138		.procname	= "tcp_slow_start_after_idle",
1139		.data		= &init_net.ipv4.sysctl_tcp_slow_start_after_idle,
1140		.maxlen		= sizeof(u8),
1141		.mode		= 0644,
1142		.proc_handler	= proc_dou8vec_minmax,
1143	},
1144	{
1145		.procname	= "tcp_retrans_collapse",
1146		.data		= &init_net.ipv4.sysctl_tcp_retrans_collapse,
1147		.maxlen		= sizeof(u8),
1148		.mode		= 0644,
1149		.proc_handler	= proc_dou8vec_minmax,
1150	},
1151	{
1152		.procname	= "tcp_stdurg",
1153		.data		= &init_net.ipv4.sysctl_tcp_stdurg,
1154		.maxlen		= sizeof(u8),
1155		.mode		= 0644,
1156		.proc_handler	= proc_dou8vec_minmax,
1157	},
1158	{
1159		.procname	= "tcp_rfc1337",
1160		.data		= &init_net.ipv4.sysctl_tcp_rfc1337,
1161		.maxlen		= sizeof(u8),
1162		.mode		= 0644,
1163		.proc_handler	= proc_dou8vec_minmax,
1164	},
1165	{
1166		.procname	= "tcp_abort_on_overflow",
1167		.data		= &init_net.ipv4.sysctl_tcp_abort_on_overflow,
1168		.maxlen		= sizeof(u8),
1169		.mode		= 0644,
1170		.proc_handler	= proc_dou8vec_minmax,
1171	},
1172	{
1173		.procname	= "tcp_fack",
1174		.data		= &init_net.ipv4.sysctl_tcp_fack,
1175		.maxlen		= sizeof(u8),
1176		.mode		= 0644,
1177		.proc_handler	= proc_dou8vec_minmax,
1178	},
1179	{
1180		.procname	= "tcp_max_reordering",
1181		.data		= &init_net.ipv4.sysctl_tcp_max_reordering,
1182		.maxlen		= sizeof(int),
1183		.mode		= 0644,
1184		.proc_handler	= proc_dointvec
1185	},
1186	{
1187		.procname	= "tcp_dsack",
1188		.data		= &init_net.ipv4.sysctl_tcp_dsack,
1189		.maxlen		= sizeof(u8),
1190		.mode		= 0644,
1191		.proc_handler	= proc_dou8vec_minmax,
1192	},
1193	{
1194		.procname	= "tcp_app_win",
1195		.data		= &init_net.ipv4.sysctl_tcp_app_win,
1196		.maxlen		= sizeof(u8),
1197		.mode		= 0644,
1198		.proc_handler	= proc_dou8vec_minmax,
1199		.extra1		= SYSCTL_ZERO,
1200		.extra2		= &tcp_app_win_max,
1201	},
1202	{
1203		.procname	= "tcp_adv_win_scale",
1204		.data		= &init_net.ipv4.sysctl_tcp_adv_win_scale,
1205		.maxlen		= sizeof(int),
1206		.mode		= 0644,
1207		.proc_handler	= proc_dointvec_minmax,
1208		.extra1		= &tcp_adv_win_scale_min,
1209		.extra2		= &tcp_adv_win_scale_max,
1210	},
1211	{
1212		.procname	= "tcp_frto",
1213		.data		= &init_net.ipv4.sysctl_tcp_frto,
1214		.maxlen		= sizeof(u8),
1215		.mode		= 0644,
1216		.proc_handler	= proc_dou8vec_minmax,
1217	},
1218	{
1219		.procname	= "tcp_no_metrics_save",
1220		.data		= &init_net.ipv4.sysctl_tcp_nometrics_save,
1221		.maxlen		= sizeof(u8),
1222		.mode		= 0644,
1223		.proc_handler	= proc_dou8vec_minmax,
1224	},
1225	{
1226		.procname	= "tcp_no_ssthresh_metrics_save",
1227		.data		= &init_net.ipv4.sysctl_tcp_no_ssthresh_metrics_save,
1228		.maxlen		= sizeof(u8),
1229		.mode		= 0644,
1230		.proc_handler	= proc_dou8vec_minmax,
1231		.extra1		= SYSCTL_ZERO,
1232		.extra2		= SYSCTL_ONE,
1233	},
1234	{
1235		.procname	= "tcp_moderate_rcvbuf",
1236		.data		= &init_net.ipv4.sysctl_tcp_moderate_rcvbuf,
1237		.maxlen		= sizeof(u8),
1238		.mode		= 0644,
1239		.proc_handler	= proc_dou8vec_minmax,
1240	},
1241	{
1242		.procname	= "tcp_tso_win_divisor",
1243		.data		= &init_net.ipv4.sysctl_tcp_tso_win_divisor,
1244		.maxlen		= sizeof(u8),
1245		.mode		= 0644,
1246		.proc_handler	= proc_dou8vec_minmax,
1247	},
1248	{
1249		.procname	= "tcp_workaround_signed_windows",
1250		.data		= &init_net.ipv4.sysctl_tcp_workaround_signed_windows,
1251		.maxlen		= sizeof(u8),
1252		.mode		= 0644,
1253		.proc_handler	= proc_dou8vec_minmax,
1254	},
1255	{
1256		.procname	= "tcp_limit_output_bytes",
1257		.data		= &init_net.ipv4.sysctl_tcp_limit_output_bytes,
1258		.maxlen		= sizeof(int),
1259		.mode		= 0644,
1260		.proc_handler	= proc_dointvec
1261	},
1262	{
1263		.procname	= "tcp_challenge_ack_limit",
1264		.data		= &init_net.ipv4.sysctl_tcp_challenge_ack_limit,
1265		.maxlen		= sizeof(int),
1266		.mode		= 0644,
1267		.proc_handler	= proc_dointvec
1268	},
1269	{
1270		.procname	= "tcp_min_tso_segs",
1271		.data		= &init_net.ipv4.sysctl_tcp_min_tso_segs,
1272		.maxlen		= sizeof(u8),
1273		.mode		= 0644,
1274		.proc_handler	= proc_dou8vec_minmax,
1275		.extra1		= SYSCTL_ONE,
1276	},
1277	{
1278		.procname	= "tcp_tso_rtt_log",
1279		.data		= &init_net.ipv4.sysctl_tcp_tso_rtt_log,
1280		.maxlen		= sizeof(u8),
1281		.mode		= 0644,
1282		.proc_handler	= proc_dou8vec_minmax,
1283	},
1284	{
1285		.procname	= "tcp_min_rtt_wlen",
1286		.data		= &init_net.ipv4.sysctl_tcp_min_rtt_wlen,
1287		.maxlen		= sizeof(int),
1288		.mode		= 0644,
1289		.proc_handler	= proc_dointvec_minmax,
1290		.extra1		= SYSCTL_ZERO,
1291		.extra2		= &one_day_secs
1292	},
1293	{
1294		.procname	= "tcp_autocorking",
1295		.data		= &init_net.ipv4.sysctl_tcp_autocorking,
1296		.maxlen		= sizeof(u8),
1297		.mode		= 0644,
1298		.proc_handler	= proc_dou8vec_minmax,
1299		.extra1		= SYSCTL_ZERO,
1300		.extra2		= SYSCTL_ONE,
1301	},
1302	{
1303		.procname	= "tcp_invalid_ratelimit",
1304		.data		= &init_net.ipv4.sysctl_tcp_invalid_ratelimit,
1305		.maxlen		= sizeof(int),
1306		.mode		= 0644,
1307		.proc_handler	= proc_dointvec_ms_jiffies,
1308	},
1309	{
1310		.procname	= "tcp_pacing_ss_ratio",
1311		.data		= &init_net.ipv4.sysctl_tcp_pacing_ss_ratio,
1312		.maxlen		= sizeof(int),
1313		.mode		= 0644,
1314		.proc_handler	= proc_dointvec_minmax,
1315		.extra1		= SYSCTL_ZERO,
1316		.extra2		= SYSCTL_ONE_THOUSAND,
1317	},
1318	{
1319		.procname	= "tcp_pacing_ca_ratio",
1320		.data		= &init_net.ipv4.sysctl_tcp_pacing_ca_ratio,
1321		.maxlen		= sizeof(int),
1322		.mode		= 0644,
1323		.proc_handler	= proc_dointvec_minmax,
1324		.extra1		= SYSCTL_ZERO,
1325		.extra2		= SYSCTL_ONE_THOUSAND,
1326	},
1327	{
1328		.procname	= "tcp_wmem",
1329		.data		= &init_net.ipv4.sysctl_tcp_wmem,
1330		.maxlen		= sizeof(init_net.ipv4.sysctl_tcp_wmem),
1331		.mode		= 0644,
1332		.proc_handler	= proc_dointvec_minmax,
1333		.extra1		= SYSCTL_ONE,
1334	},
1335	{
1336		.procname	= "tcp_rmem",
1337		.data		= &init_net.ipv4.sysctl_tcp_rmem,
1338		.maxlen		= sizeof(init_net.ipv4.sysctl_tcp_rmem),
1339		.mode		= 0644,
1340		.proc_handler	= proc_dointvec_minmax,
1341		.extra1		= SYSCTL_ONE,
1342	},
1343	{
1344		.procname	= "tcp_comp_sack_delay_ns",
1345		.data		= &init_net.ipv4.sysctl_tcp_comp_sack_delay_ns,
1346		.maxlen		= sizeof(unsigned long),
1347		.mode		= 0644,
1348		.proc_handler	= proc_doulongvec_minmax,
1349	},
1350	{
1351		.procname	= "tcp_comp_sack_slack_ns",
1352		.data		= &init_net.ipv4.sysctl_tcp_comp_sack_slack_ns,
1353		.maxlen		= sizeof(unsigned long),
1354		.mode		= 0644,
1355		.proc_handler	= proc_doulongvec_minmax,
1356	},
1357	{
1358		.procname	= "tcp_comp_sack_nr",
1359		.data		= &init_net.ipv4.sysctl_tcp_comp_sack_nr,
1360		.maxlen		= sizeof(u8),
1361		.mode		= 0644,
1362		.proc_handler	= proc_dou8vec_minmax,
1363		.extra1		= SYSCTL_ZERO,
1364	},
1365	{
1366		.procname	= "tcp_backlog_ack_defer",
1367		.data		= &init_net.ipv4.sysctl_tcp_backlog_ack_defer,
1368		.maxlen		= sizeof(u8),
1369		.mode		= 0644,
1370		.proc_handler	= proc_dou8vec_minmax,
1371		.extra1		= SYSCTL_ZERO,
1372		.extra2		= SYSCTL_ONE,
1373	},
1374	{
1375		.procname       = "tcp_reflect_tos",
1376		.data           = &init_net.ipv4.sysctl_tcp_reflect_tos,
1377		.maxlen         = sizeof(u8),
1378		.mode           = 0644,
1379		.proc_handler   = proc_dou8vec_minmax,
1380		.extra1         = SYSCTL_ZERO,
1381		.extra2         = SYSCTL_ONE,
1382	},
1383	{
1384		.procname	= "tcp_ehash_entries",
1385		.data		= &init_net.ipv4.sysctl_tcp_child_ehash_entries,
1386		.mode		= 0444,
1387		.proc_handler	= proc_tcp_ehash_entries,
1388	},
1389	{
1390		.procname	= "tcp_child_ehash_entries",
1391		.data		= &init_net.ipv4.sysctl_tcp_child_ehash_entries,
1392		.maxlen		= sizeof(unsigned int),
1393		.mode		= 0644,
1394		.proc_handler	= proc_douintvec_minmax,
1395		.extra1		= SYSCTL_ZERO,
1396		.extra2		= &tcp_child_ehash_entries_max,
1397	},
1398	{
1399		.procname	= "udp_hash_entries",
1400		.data		= &init_net.ipv4.sysctl_udp_child_hash_entries,
1401		.mode		= 0444,
1402		.proc_handler	= proc_udp_hash_entries,
1403	},
1404	{
1405		.procname	= "udp_child_hash_entries",
1406		.data		= &init_net.ipv4.sysctl_udp_child_hash_entries,
1407		.maxlen		= sizeof(unsigned int),
1408		.mode		= 0644,
1409		.proc_handler	= proc_douintvec_minmax,
1410		.extra1		= SYSCTL_ZERO,
1411		.extra2		= &udp_child_hash_entries_max,
1412	},
1413	{
1414		.procname	= "udp_rmem_min",
1415		.data		= &init_net.ipv4.sysctl_udp_rmem_min,
1416		.maxlen		= sizeof(init_net.ipv4.sysctl_udp_rmem_min),
1417		.mode		= 0644,
1418		.proc_handler	= proc_dointvec_minmax,
1419		.extra1		= SYSCTL_ONE
1420	},
1421	{
1422		.procname	= "udp_wmem_min",
1423		.data		= &init_net.ipv4.sysctl_udp_wmem_min,
1424		.maxlen		= sizeof(init_net.ipv4.sysctl_udp_wmem_min),
1425		.mode		= 0644,
1426		.proc_handler	= proc_dointvec_minmax,
1427		.extra1		= SYSCTL_ONE
1428	},
1429	{
1430		.procname	= "fib_notify_on_flag_change",
1431		.data		= &init_net.ipv4.sysctl_fib_notify_on_flag_change,
1432		.maxlen		= sizeof(u8),
1433		.mode		= 0644,
1434		.proc_handler	= proc_dou8vec_minmax,
1435		.extra1		= SYSCTL_ZERO,
1436		.extra2		= SYSCTL_TWO,
1437	},
1438	{
1439		.procname       = "tcp_plb_enabled",
1440		.data           = &init_net.ipv4.sysctl_tcp_plb_enabled,
1441		.maxlen         = sizeof(u8),
1442		.mode           = 0644,
1443		.proc_handler   = proc_dou8vec_minmax,
1444		.extra1         = SYSCTL_ZERO,
1445		.extra2         = SYSCTL_ONE,
1446	},
1447	{
1448		.procname       = "tcp_plb_idle_rehash_rounds",
1449		.data           = &init_net.ipv4.sysctl_tcp_plb_idle_rehash_rounds,
1450		.maxlen         = sizeof(u8),
1451		.mode           = 0644,
1452		.proc_handler   = proc_dou8vec_minmax,
1453		.extra2		= &tcp_plb_max_rounds,
1454	},
1455	{
1456		.procname       = "tcp_plb_rehash_rounds",
1457		.data           = &init_net.ipv4.sysctl_tcp_plb_rehash_rounds,
1458		.maxlen         = sizeof(u8),
1459		.mode           = 0644,
1460		.proc_handler   = proc_dou8vec_minmax,
1461		.extra2         = &tcp_plb_max_rounds,
1462	},
1463	{
1464		.procname       = "tcp_plb_suspend_rto_sec",
1465		.data           = &init_net.ipv4.sysctl_tcp_plb_suspend_rto_sec,
1466		.maxlen         = sizeof(u8),
1467		.mode           = 0644,
1468		.proc_handler   = proc_dou8vec_minmax,
1469	},
1470	{
1471		.procname       = "tcp_plb_cong_thresh",
1472		.data           = &init_net.ipv4.sysctl_tcp_plb_cong_thresh,
1473		.maxlen         = sizeof(int),
1474		.mode           = 0644,
1475		.proc_handler   = proc_dointvec_minmax,
1476		.extra1         = SYSCTL_ZERO,
1477		.extra2         = &tcp_plb_max_cong_thresh,
1478	},
1479	{
1480		.procname	= "tcp_syn_linear_timeouts",
1481		.data		= &init_net.ipv4.sysctl_tcp_syn_linear_timeouts,
1482		.maxlen		= sizeof(u8),
1483		.mode		= 0644,
1484		.proc_handler	= proc_dou8vec_minmax,
1485		.extra1		= SYSCTL_ZERO,
1486		.extra2		= &tcp_syn_linear_timeouts_max,
1487	},
1488	{
1489		.procname	= "tcp_shrink_window",
1490		.data		= &init_net.ipv4.sysctl_tcp_shrink_window,
1491		.maxlen		= sizeof(u8),
1492		.mode		= 0644,
1493		.proc_handler	= proc_dou8vec_minmax,
1494		.extra1		= SYSCTL_ZERO,
1495		.extra2		= SYSCTL_ONE,
1496	},
1497	{
1498		.procname	= "tcp_pingpong_thresh",
1499		.data		= &init_net.ipv4.sysctl_tcp_pingpong_thresh,
1500		.maxlen		= sizeof(u8),
1501		.mode		= 0644,
1502		.proc_handler	= proc_dou8vec_minmax,
1503		.extra1		= SYSCTL_ONE,
1504	},
1505	{ }
1506};
1507
1508static __net_init int ipv4_sysctl_init_net(struct net *net)
1509{
1510	struct ctl_table *table;
1511
1512	table = ipv4_net_table;
1513	if (!net_eq(net, &init_net)) {
1514		int i;
1515
1516		table = kmemdup(table, sizeof(ipv4_net_table), GFP_KERNEL);
1517		if (!table)
1518			goto err_alloc;
1519
1520		for (i = 0; i < ARRAY_SIZE(ipv4_net_table) - 1; i++) {
1521			if (table[i].data) {
1522				/* Update the variables to point into
1523				 * the current struct net
1524				 */
1525				table[i].data += (void *)net - (void *)&init_net;
1526			} else {
1527				/* Entries without data pointer are global;
1528				 * Make them read-only in non-init_net ns
1529				 */
1530				table[i].mode &= ~0222;
1531			}
1532		}
1533	}
1534
1535	net->ipv4.ipv4_hdr = register_net_sysctl_sz(net, "net/ipv4", table,
1536						    ARRAY_SIZE(ipv4_net_table));
1537	if (!net->ipv4.ipv4_hdr)
1538		goto err_reg;
1539
1540	net->ipv4.sysctl_local_reserved_ports = kzalloc(65536 / 8, GFP_KERNEL);
1541	if (!net->ipv4.sysctl_local_reserved_ports)
1542		goto err_ports;
1543
1544	return 0;
1545
1546err_ports:
1547	unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
1548err_reg:
1549	if (!net_eq(net, &init_net))
1550		kfree(table);
1551err_alloc:
1552	return -ENOMEM;
1553}
1554
1555static __net_exit void ipv4_sysctl_exit_net(struct net *net)
1556{
1557	struct ctl_table *table;
1558
1559	kfree(net->ipv4.sysctl_local_reserved_ports);
1560	table = net->ipv4.ipv4_hdr->ctl_table_arg;
1561	unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
1562	kfree(table);
1563}
1564
1565static __net_initdata struct pernet_operations ipv4_sysctl_ops = {
1566	.init = ipv4_sysctl_init_net,
1567	.exit = ipv4_sysctl_exit_net,
1568};
1569
1570static __init int sysctl_ipv4_init(void)
1571{
1572	struct ctl_table_header *hdr;
1573
1574	hdr = register_net_sysctl(&init_net, "net/ipv4", ipv4_table);
1575	if (!hdr)
1576		return -ENOMEM;
1577
1578	if (register_pernet_subsys(&ipv4_sysctl_ops)) {
1579		unregister_net_sysctl_table(hdr);
1580		return -ENOMEM;
1581	}
1582
1583	return 0;
1584}
1585
1586__initcall(sysctl_ipv4_init);
1587