sysctl_net_ipv4.c revision c255a458055e459f65eb7b7f51dc5dbdd0caf1d8 
1/*
2 * sysctl_net_ipv4.c: sysctl interface to net IPV4 subsystem.
3 *
4 * Begun April 1, 1996, Mike Shaver.
5 * Added /proc/sys/net/ipv4 directory entry (empty =) ). [MS]
6 */
7
8#include <linux/mm.h>
9#include <linux/module.h>
10#include <linux/sysctl.h>
11#include <linux/igmp.h>
12#include <linux/inetdevice.h>
13#include <linux/seqlock.h>
14#include <linux/init.h>
15#include <linux/slab.h>
16#include <linux/nsproxy.h>
17#include <linux/swap.h>
18#include <net/snmp.h>
19#include <net/icmp.h>
20#include <net/ip.h>
21#include <net/route.h>
22#include <net/tcp.h>
23#include <net/udp.h>
24#include <net/cipso_ipv4.h>
25#include <net/inet_frag.h>
26#include <net/ping.h>
27#include <net/tcp_memcontrol.h>
28
29static int zero;
30static int two = 2;
31static int tcp_retr1_max = 255;
32static int ip_local_port_range_min[] = { 1, 1 };
33static int ip_local_port_range_max[] = { 65535, 65535 };
34static int tcp_adv_win_scale_min = -31;
35static int tcp_adv_win_scale_max = 31;
36static int ip_ttl_min = 1;
37static int ip_ttl_max = 255;
38static int ip_ping_group_range_min[] = { 0, 0 };
39static int ip_ping_group_range_max[] = { GID_T_MAX, GID_T_MAX };
40
41/* Update system visible IP port range */
42static void set_local_port_range(int range[2])
43{
44	write_seqlock(&sysctl_local_ports.lock);
45	sysctl_local_ports.range[0] = range[0];
46	sysctl_local_ports.range[1] = range[1];
47	write_sequnlock(&sysctl_local_ports.lock);
48}
49
50/* Validate changes from /proc interface. */
51static int ipv4_local_port_range(ctl_table *table, int write,
52				 void __user *buffer,
53				 size_t *lenp, loff_t *ppos)
54{
55	int ret;
56	int range[2];
57	ctl_table tmp = {
58		.data = &range,
59		.maxlen = sizeof(range),
60		.mode = table->mode,
61		.extra1 = &ip_local_port_range_min,
62		.extra2 = &ip_local_port_range_max,
63	};
64
65	inet_get_local_port_range(range, range + 1);
66	ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
67
68	if (write && ret == 0) {
69		if (range[1] < range[0])
70			ret = -EINVAL;
71		else
72			set_local_port_range(range);
73	}
74
75	return ret;
76}
77
78
79static void inet_get_ping_group_range_table(struct ctl_table *table, gid_t *low, gid_t *high)
80{
81	gid_t *data = table->data;
82	unsigned int seq;
83	do {
84		seq = read_seqbegin(&sysctl_local_ports.lock);
85
86		*low = data[0];
87		*high = data[1];
88	} while (read_seqretry(&sysctl_local_ports.lock, seq));
89}
90
91/* Update system visible IP port range */
92static void set_ping_group_range(struct ctl_table *table, gid_t range[2])
93{
94	gid_t *data = table->data;
95	write_seqlock(&sysctl_local_ports.lock);
96	data[0] = range[0];
97	data[1] = range[1];
98	write_sequnlock(&sysctl_local_ports.lock);
99}
100
101/* Validate changes from /proc interface. */
102static int ipv4_ping_group_range(ctl_table *table, int write,
103				 void __user *buffer,
104				 size_t *lenp, loff_t *ppos)
105{
106	int ret;
107	gid_t range[2];
108	ctl_table tmp = {
109		.data = &range,
110		.maxlen = sizeof(range),
111		.mode = table->mode,
112		.extra1 = &ip_ping_group_range_min,
113		.extra2 = &ip_ping_group_range_max,
114	};
115
116	inet_get_ping_group_range_table(table, range, range + 1);
117	ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
118
119	if (write && ret == 0)
120		set_ping_group_range(table, range);
121
122	return ret;
123}
124
125static int proc_tcp_congestion_control(ctl_table *ctl, int write,
126				       void __user *buffer, size_t *lenp, loff_t *ppos)
127{
128	char val[TCP_CA_NAME_MAX];
129	ctl_table tbl = {
130		.data = val,
131		.maxlen = TCP_CA_NAME_MAX,
132	};
133	int ret;
134
135	tcp_get_default_congestion_control(val);
136
137	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
138	if (write && ret == 0)
139		ret = tcp_set_default_congestion_control(val);
140	return ret;
141}
142
143static int proc_tcp_available_congestion_control(ctl_table *ctl,
144						 int write,
145						 void __user *buffer, size_t *lenp,
146						 loff_t *ppos)
147{
148	ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX, };
149	int ret;
150
151	tbl.data = kmalloc(tbl.maxlen, GFP_USER);
152	if (!tbl.data)
153		return -ENOMEM;
154	tcp_get_available_congestion_control(tbl.data, TCP_CA_BUF_MAX);
155	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
156	kfree(tbl.data);
157	return ret;
158}
159
160static int proc_allowed_congestion_control(ctl_table *ctl,
161					   int write,
162					   void __user *buffer, size_t *lenp,
163					   loff_t *ppos)
164{
165	ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX };
166	int ret;
167
168	tbl.data = kmalloc(tbl.maxlen, GFP_USER);
169	if (!tbl.data)
170		return -ENOMEM;
171
172	tcp_get_allowed_congestion_control(tbl.data, tbl.maxlen);
173	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
174	if (write && ret == 0)
175		ret = tcp_set_allowed_congestion_control(tbl.data);
176	kfree(tbl.data);
177	return ret;
178}
179
180static int ipv4_tcp_mem(ctl_table *ctl, int write,
181			   void __user *buffer, size_t *lenp,
182			   loff_t *ppos)
183{
184	int ret;
185	unsigned long vec[3];
186	struct net *net = current->nsproxy->net_ns;
187#ifdef CONFIG_MEMCG_KMEM
188	struct mem_cgroup *memcg;
189#endif
190
191	ctl_table tmp = {
192		.data = &vec,
193		.maxlen = sizeof(vec),
194		.mode = ctl->mode,
195	};
196
197	if (!write) {
198		ctl->data = &net->ipv4.sysctl_tcp_mem;
199		return proc_doulongvec_minmax(ctl, write, buffer, lenp, ppos);
200	}
201
202	ret = proc_doulongvec_minmax(&tmp, write, buffer, lenp, ppos);
203	if (ret)
204		return ret;
205
206#ifdef CONFIG_MEMCG_KMEM
207	rcu_read_lock();
208	memcg = mem_cgroup_from_task(current);
209
210	tcp_prot_mem(memcg, vec[0], 0);
211	tcp_prot_mem(memcg, vec[1], 1);
212	tcp_prot_mem(memcg, vec[2], 2);
213	rcu_read_unlock();
214#endif
215
216	net->ipv4.sysctl_tcp_mem[0] = vec[0];
217	net->ipv4.sysctl_tcp_mem[1] = vec[1];
218	net->ipv4.sysctl_tcp_mem[2] = vec[2];
219
220	return 0;
221}
222
223static struct ctl_table ipv4_table[] = {
224	{
225		.procname	= "tcp_timestamps",
226		.data		= &sysctl_tcp_timestamps,
227		.maxlen		= sizeof(int),
228		.mode		= 0644,
229		.proc_handler	= proc_dointvec
230	},
231	{
232		.procname	= "tcp_window_scaling",
233		.data		= &sysctl_tcp_window_scaling,
234		.maxlen		= sizeof(int),
235		.mode		= 0644,
236		.proc_handler	= proc_dointvec
237	},
238	{
239		.procname	= "tcp_sack",
240		.data		= &sysctl_tcp_sack,
241		.maxlen		= sizeof(int),
242		.mode		= 0644,
243		.proc_handler	= proc_dointvec
244	},
245	{
246		.procname	= "tcp_retrans_collapse",
247		.data		= &sysctl_tcp_retrans_collapse,
248		.maxlen		= sizeof(int),
249		.mode		= 0644,
250		.proc_handler	= proc_dointvec
251	},
252	{
253		.procname	= "ip_default_ttl",
254		.data		= &sysctl_ip_default_ttl,
255		.maxlen		= sizeof(int),
256		.mode		= 0644,
257		.proc_handler	= proc_dointvec_minmax,
258		.extra1		= &ip_ttl_min,
259		.extra2		= &ip_ttl_max,
260	},
261	{
262		.procname	= "ip_no_pmtu_disc",
263		.data		= &ipv4_config.no_pmtu_disc,
264		.maxlen		= sizeof(int),
265		.mode		= 0644,
266		.proc_handler	= proc_dointvec
267	},
268	{
269		.procname	= "ip_nonlocal_bind",
270		.data		= &sysctl_ip_nonlocal_bind,
271		.maxlen		= sizeof(int),
272		.mode		= 0644,
273		.proc_handler	= proc_dointvec
274	},
275	{
276		.procname	= "tcp_syn_retries",
277		.data		= &sysctl_tcp_syn_retries,
278		.maxlen		= sizeof(int),
279		.mode		= 0644,
280		.proc_handler	= proc_dointvec
281	},
282	{
283		.procname	= "tcp_synack_retries",
284		.data		= &sysctl_tcp_synack_retries,
285		.maxlen		= sizeof(int),
286		.mode		= 0644,
287		.proc_handler	= proc_dointvec
288	},
289	{
290		.procname	= "tcp_max_orphans",
291		.data		= &sysctl_tcp_max_orphans,
292		.maxlen		= sizeof(int),
293		.mode		= 0644,
294		.proc_handler	= proc_dointvec
295	},
296	{
297		.procname	= "tcp_max_tw_buckets",
298		.data		= &tcp_death_row.sysctl_max_tw_buckets,
299		.maxlen		= sizeof(int),
300		.mode		= 0644,
301		.proc_handler	= proc_dointvec
302	},
303	{
304		.procname	= "ip_early_demux",
305		.data		= &sysctl_ip_early_demux,
306		.maxlen		= sizeof(int),
307		.mode		= 0644,
308		.proc_handler	= proc_dointvec
309	},
310	{
311		.procname	= "ip_dynaddr",
312		.data		= &sysctl_ip_dynaddr,
313		.maxlen		= sizeof(int),
314		.mode		= 0644,
315		.proc_handler	= proc_dointvec
316	},
317	{
318		.procname	= "tcp_keepalive_time",
319		.data		= &sysctl_tcp_keepalive_time,
320		.maxlen		= sizeof(int),
321		.mode		= 0644,
322		.proc_handler	= proc_dointvec_jiffies,
323	},
324	{
325		.procname	= "tcp_keepalive_probes",
326		.data		= &sysctl_tcp_keepalive_probes,
327		.maxlen		= sizeof(int),
328		.mode		= 0644,
329		.proc_handler	= proc_dointvec
330	},
331	{
332		.procname	= "tcp_keepalive_intvl",
333		.data		= &sysctl_tcp_keepalive_intvl,
334		.maxlen		= sizeof(int),
335		.mode		= 0644,
336		.proc_handler	= proc_dointvec_jiffies,
337	},
338	{
339		.procname	= "tcp_retries1",
340		.data		= &sysctl_tcp_retries1,
341		.maxlen		= sizeof(int),
342		.mode		= 0644,
343		.proc_handler	= proc_dointvec_minmax,
344		.extra2		= &tcp_retr1_max
345	},
346	{
347		.procname	= "tcp_retries2",
348		.data		= &sysctl_tcp_retries2,
349		.maxlen		= sizeof(int),
350		.mode		= 0644,
351		.proc_handler	= proc_dointvec
352	},
353	{
354		.procname	= "tcp_fin_timeout",
355		.data		= &sysctl_tcp_fin_timeout,
356		.maxlen		= sizeof(int),
357		.mode		= 0644,
358		.proc_handler	= proc_dointvec_jiffies,
359	},
360#ifdef CONFIG_SYN_COOKIES
361	{
362		.procname	= "tcp_syncookies",
363		.data		= &sysctl_tcp_syncookies,
364		.maxlen		= sizeof(int),
365		.mode		= 0644,
366		.proc_handler	= proc_dointvec
367	},
368#endif
369	{
370		.procname	= "tcp_fastopen",
371		.data		= &sysctl_tcp_fastopen,
372		.maxlen		= sizeof(int),
373		.mode		= 0644,
374		.proc_handler	= proc_dointvec,
375	},
376	{
377		.procname	= "tcp_tw_recycle",
378		.data		= &tcp_death_row.sysctl_tw_recycle,
379		.maxlen		= sizeof(int),
380		.mode		= 0644,
381		.proc_handler	= proc_dointvec
382	},
383	{
384		.procname	= "tcp_abort_on_overflow",
385		.data		= &sysctl_tcp_abort_on_overflow,
386		.maxlen		= sizeof(int),
387		.mode		= 0644,
388		.proc_handler	= proc_dointvec
389	},
390	{
391		.procname	= "tcp_stdurg",
392		.data		= &sysctl_tcp_stdurg,
393		.maxlen		= sizeof(int),
394		.mode		= 0644,
395		.proc_handler	= proc_dointvec
396	},
397	{
398		.procname	= "tcp_rfc1337",
399		.data		= &sysctl_tcp_rfc1337,
400		.maxlen		= sizeof(int),
401		.mode		= 0644,
402		.proc_handler	= proc_dointvec
403	},
404	{
405		.procname	= "tcp_max_syn_backlog",
406		.data		= &sysctl_max_syn_backlog,
407		.maxlen		= sizeof(int),
408		.mode		= 0644,
409		.proc_handler	= proc_dointvec
410	},
411	{
412		.procname	= "ip_local_port_range",
413		.data		= &sysctl_local_ports.range,
414		.maxlen		= sizeof(sysctl_local_ports.range),
415		.mode		= 0644,
416		.proc_handler	= ipv4_local_port_range,
417	},
418	{
419		.procname	= "ip_local_reserved_ports",
420		.data		= NULL, /* initialized in sysctl_ipv4_init */
421		.maxlen		= 65536,
422		.mode		= 0644,
423		.proc_handler	= proc_do_large_bitmap,
424	},
425	{
426		.procname	= "igmp_max_memberships",
427		.data		= &sysctl_igmp_max_memberships,
428		.maxlen		= sizeof(int),
429		.mode		= 0644,
430		.proc_handler	= proc_dointvec
431	},
432	{
433		.procname	= "igmp_max_msf",
434		.data		= &sysctl_igmp_max_msf,
435		.maxlen		= sizeof(int),
436		.mode		= 0644,
437		.proc_handler	= proc_dointvec
438	},
439	{
440		.procname	= "inet_peer_threshold",
441		.data		= &inet_peer_threshold,
442		.maxlen		= sizeof(int),
443		.mode		= 0644,
444		.proc_handler	= proc_dointvec
445	},
446	{
447		.procname	= "inet_peer_minttl",
448		.data		= &inet_peer_minttl,
449		.maxlen		= sizeof(int),
450		.mode		= 0644,
451		.proc_handler	= proc_dointvec_jiffies,
452	},
453	{
454		.procname	= "inet_peer_maxttl",
455		.data		= &inet_peer_maxttl,
456		.maxlen		= sizeof(int),
457		.mode		= 0644,
458		.proc_handler	= proc_dointvec_jiffies,
459	},
460	{
461		.procname	= "tcp_orphan_retries",
462		.data		= &sysctl_tcp_orphan_retries,
463		.maxlen		= sizeof(int),
464		.mode		= 0644,
465		.proc_handler	= proc_dointvec
466	},
467	{
468		.procname	= "tcp_fack",
469		.data		= &sysctl_tcp_fack,
470		.maxlen		= sizeof(int),
471		.mode		= 0644,
472		.proc_handler	= proc_dointvec
473	},
474	{
475		.procname	= "tcp_reordering",
476		.data		= &sysctl_tcp_reordering,
477		.maxlen		= sizeof(int),
478		.mode		= 0644,
479		.proc_handler	= proc_dointvec
480	},
481	{
482		.procname	= "tcp_ecn",
483		.data		= &sysctl_tcp_ecn,
484		.maxlen		= sizeof(int),
485		.mode		= 0644,
486		.proc_handler	= proc_dointvec
487	},
488	{
489		.procname	= "tcp_dsack",
490		.data		= &sysctl_tcp_dsack,
491		.maxlen		= sizeof(int),
492		.mode		= 0644,
493		.proc_handler	= proc_dointvec
494	},
495	{
496		.procname	= "tcp_wmem",
497		.data		= &sysctl_tcp_wmem,
498		.maxlen		= sizeof(sysctl_tcp_wmem),
499		.mode		= 0644,
500		.proc_handler	= proc_dointvec
501	},
502	{
503		.procname	= "tcp_rmem",
504		.data		= &sysctl_tcp_rmem,
505		.maxlen		= sizeof(sysctl_tcp_rmem),
506		.mode		= 0644,
507		.proc_handler	= proc_dointvec
508	},
509	{
510		.procname	= "tcp_app_win",
511		.data		= &sysctl_tcp_app_win,
512		.maxlen		= sizeof(int),
513		.mode		= 0644,
514		.proc_handler	= proc_dointvec
515	},
516	{
517		.procname	= "tcp_adv_win_scale",
518		.data		= &sysctl_tcp_adv_win_scale,
519		.maxlen		= sizeof(int),
520		.mode		= 0644,
521		.proc_handler	= proc_dointvec_minmax,
522		.extra1		= &tcp_adv_win_scale_min,
523		.extra2		= &tcp_adv_win_scale_max,
524	},
525	{
526		.procname	= "tcp_tw_reuse",
527		.data		= &sysctl_tcp_tw_reuse,
528		.maxlen		= sizeof(int),
529		.mode		= 0644,
530		.proc_handler	= proc_dointvec
531	},
532	{
533		.procname	= "tcp_frto",
534		.data		= &sysctl_tcp_frto,
535		.maxlen		= sizeof(int),
536		.mode		= 0644,
537		.proc_handler	= proc_dointvec
538	},
539	{
540		.procname	= "tcp_frto_response",
541		.data		= &sysctl_tcp_frto_response,
542		.maxlen		= sizeof(int),
543		.mode		= 0644,
544		.proc_handler	= proc_dointvec
545	},
546	{
547		.procname	= "tcp_low_latency",
548		.data		= &sysctl_tcp_low_latency,
549		.maxlen		= sizeof(int),
550		.mode		= 0644,
551		.proc_handler	= proc_dointvec
552	},
553	{
554		.procname	= "tcp_no_metrics_save",
555		.data		= &sysctl_tcp_nometrics_save,
556		.maxlen		= sizeof(int),
557		.mode		= 0644,
558		.proc_handler	= proc_dointvec,
559	},
560	{
561		.procname	= "tcp_moderate_rcvbuf",
562		.data		= &sysctl_tcp_moderate_rcvbuf,
563		.maxlen		= sizeof(int),
564		.mode		= 0644,
565		.proc_handler	= proc_dointvec,
566	},
567	{
568		.procname	= "tcp_tso_win_divisor",
569		.data		= &sysctl_tcp_tso_win_divisor,
570		.maxlen		= sizeof(int),
571		.mode		= 0644,
572		.proc_handler	= proc_dointvec,
573	},
574	{
575		.procname	= "tcp_congestion_control",
576		.mode		= 0644,
577		.maxlen		= TCP_CA_NAME_MAX,
578		.proc_handler	= proc_tcp_congestion_control,
579	},
580	{
581		.procname	= "tcp_abc",
582		.data		= &sysctl_tcp_abc,
583		.maxlen		= sizeof(int),
584		.mode		= 0644,
585		.proc_handler	= proc_dointvec,
586	},
587	{
588		.procname	= "tcp_mtu_probing",
589		.data		= &sysctl_tcp_mtu_probing,
590		.maxlen		= sizeof(int),
591		.mode		= 0644,
592		.proc_handler	= proc_dointvec,
593	},
594	{
595		.procname	= "tcp_base_mss",
596		.data		= &sysctl_tcp_base_mss,
597		.maxlen		= sizeof(int),
598		.mode		= 0644,
599		.proc_handler	= proc_dointvec,
600	},
601	{
602		.procname	= "tcp_workaround_signed_windows",
603		.data		= &sysctl_tcp_workaround_signed_windows,
604		.maxlen		= sizeof(int),
605		.mode		= 0644,
606		.proc_handler	= proc_dointvec
607	},
608	{
609		.procname	= "tcp_limit_output_bytes",
610		.data		= &sysctl_tcp_limit_output_bytes,
611		.maxlen		= sizeof(int),
612		.mode		= 0644,
613		.proc_handler	= proc_dointvec
614	},
615	{
616		.procname	= "tcp_challenge_ack_limit",
617		.data		= &sysctl_tcp_challenge_ack_limit,
618		.maxlen		= sizeof(int),
619		.mode		= 0644,
620		.proc_handler	= proc_dointvec
621	},
622#ifdef CONFIG_NET_DMA
623	{
624		.procname	= "tcp_dma_copybreak",
625		.data		= &sysctl_tcp_dma_copybreak,
626		.maxlen		= sizeof(int),
627		.mode		= 0644,
628		.proc_handler	= proc_dointvec
629	},
630#endif
631	{
632		.procname	= "tcp_slow_start_after_idle",
633		.data		= &sysctl_tcp_slow_start_after_idle,
634		.maxlen		= sizeof(int),
635		.mode		= 0644,
636		.proc_handler	= proc_dointvec
637	},
638#ifdef CONFIG_NETLABEL
639	{
640		.procname	= "cipso_cache_enable",
641		.data		= &cipso_v4_cache_enabled,
642		.maxlen		= sizeof(int),
643		.mode		= 0644,
644		.proc_handler	= proc_dointvec,
645	},
646	{
647		.procname	= "cipso_cache_bucket_size",
648		.data		= &cipso_v4_cache_bucketsize,
649		.maxlen		= sizeof(int),
650		.mode		= 0644,
651		.proc_handler	= proc_dointvec,
652	},
653	{
654		.procname	= "cipso_rbm_optfmt",
655		.data		= &cipso_v4_rbm_optfmt,
656		.maxlen		= sizeof(int),
657		.mode		= 0644,
658		.proc_handler	= proc_dointvec,
659	},
660	{
661		.procname	= "cipso_rbm_strictvalid",
662		.data		= &cipso_v4_rbm_strictvalid,
663		.maxlen		= sizeof(int),
664		.mode		= 0644,
665		.proc_handler	= proc_dointvec,
666	},
667#endif /* CONFIG_NETLABEL */
668	{
669		.procname	= "tcp_available_congestion_control",
670		.maxlen		= TCP_CA_BUF_MAX,
671		.mode		= 0444,
672		.proc_handler   = proc_tcp_available_congestion_control,
673	},
674	{
675		.procname	= "tcp_allowed_congestion_control",
676		.maxlen		= TCP_CA_BUF_MAX,
677		.mode		= 0644,
678		.proc_handler   = proc_allowed_congestion_control,
679	},
680	{
681		.procname	= "tcp_max_ssthresh",
682		.data		= &sysctl_tcp_max_ssthresh,
683		.maxlen		= sizeof(int),
684		.mode		= 0644,
685		.proc_handler	= proc_dointvec,
686	},
687	{
688		.procname	= "tcp_cookie_size",
689		.data		= &sysctl_tcp_cookie_size,
690		.maxlen		= sizeof(int),
691		.mode		= 0644,
692		.proc_handler	= proc_dointvec
693	},
694	{
695		.procname       = "tcp_thin_linear_timeouts",
696		.data           = &sysctl_tcp_thin_linear_timeouts,
697		.maxlen         = sizeof(int),
698		.mode           = 0644,
699		.proc_handler   = proc_dointvec
700	},
701        {
702		.procname       = "tcp_thin_dupack",
703		.data           = &sysctl_tcp_thin_dupack,
704		.maxlen         = sizeof(int),
705		.mode           = 0644,
706		.proc_handler   = proc_dointvec
707	},
708	{
709		.procname	= "tcp_early_retrans",
710		.data		= &sysctl_tcp_early_retrans,
711		.maxlen		= sizeof(int),
712		.mode		= 0644,
713		.proc_handler	= proc_dointvec_minmax,
714		.extra1		= &zero,
715		.extra2		= &two,
716	},
717	{
718		.procname	= "udp_mem",
719		.data		= &sysctl_udp_mem,
720		.maxlen		= sizeof(sysctl_udp_mem),
721		.mode		= 0644,
722		.proc_handler	= proc_doulongvec_minmax,
723	},
724	{
725		.procname	= "udp_rmem_min",
726		.data		= &sysctl_udp_rmem_min,
727		.maxlen		= sizeof(sysctl_udp_rmem_min),
728		.mode		= 0644,
729		.proc_handler	= proc_dointvec_minmax,
730		.extra1		= &zero
731	},
732	{
733		.procname	= "udp_wmem_min",
734		.data		= &sysctl_udp_wmem_min,
735		.maxlen		= sizeof(sysctl_udp_wmem_min),
736		.mode		= 0644,
737		.proc_handler	= proc_dointvec_minmax,
738		.extra1		= &zero
739	},
740	{ }
741};
742
743static struct ctl_table ipv4_net_table[] = {
744	{
745		.procname	= "icmp_echo_ignore_all",
746		.data		= &init_net.ipv4.sysctl_icmp_echo_ignore_all,
747		.maxlen		= sizeof(int),
748		.mode		= 0644,
749		.proc_handler	= proc_dointvec
750	},
751	{
752		.procname	= "icmp_echo_ignore_broadcasts",
753		.data		= &init_net.ipv4.sysctl_icmp_echo_ignore_broadcasts,
754		.maxlen		= sizeof(int),
755		.mode		= 0644,
756		.proc_handler	= proc_dointvec
757	},
758	{
759		.procname	= "icmp_ignore_bogus_error_responses",
760		.data		= &init_net.ipv4.sysctl_icmp_ignore_bogus_error_responses,
761		.maxlen		= sizeof(int),
762		.mode		= 0644,
763		.proc_handler	= proc_dointvec
764	},
765	{
766		.procname	= "icmp_errors_use_inbound_ifaddr",
767		.data		= &init_net.ipv4.sysctl_icmp_errors_use_inbound_ifaddr,
768		.maxlen		= sizeof(int),
769		.mode		= 0644,
770		.proc_handler	= proc_dointvec
771	},
772	{
773		.procname	= "icmp_ratelimit",
774		.data		= &init_net.ipv4.sysctl_icmp_ratelimit,
775		.maxlen		= sizeof(int),
776		.mode		= 0644,
777		.proc_handler	= proc_dointvec_ms_jiffies,
778	},
779	{
780		.procname	= "icmp_ratemask",
781		.data		= &init_net.ipv4.sysctl_icmp_ratemask,
782		.maxlen		= sizeof(int),
783		.mode		= 0644,
784		.proc_handler	= proc_dointvec
785	},
786	{
787		.procname	= "rt_cache_rebuild_count",
788		.data		= &init_net.ipv4.sysctl_rt_cache_rebuild_count,
789		.maxlen		= sizeof(int),
790		.mode		= 0644,
791		.proc_handler	= proc_dointvec
792	},
793	{
794		.procname	= "ping_group_range",
795		.data		= &init_net.ipv4.sysctl_ping_group_range,
796		.maxlen		= sizeof(init_net.ipv4.sysctl_ping_group_range),
797		.mode		= 0644,
798		.proc_handler	= ipv4_ping_group_range,
799	},
800	{
801		.procname	= "tcp_mem",
802		.maxlen		= sizeof(init_net.ipv4.sysctl_tcp_mem),
803		.mode		= 0644,
804		.proc_handler	= ipv4_tcp_mem,
805	},
806	{ }
807};
808
809static __net_init int ipv4_sysctl_init_net(struct net *net)
810{
811	struct ctl_table *table;
812
813	table = ipv4_net_table;
814	if (!net_eq(net, &init_net)) {
815		table = kmemdup(table, sizeof(ipv4_net_table), GFP_KERNEL);
816		if (table == NULL)
817			goto err_alloc;
818
819		table[0].data =
820			&net->ipv4.sysctl_icmp_echo_ignore_all;
821		table[1].data =
822			&net->ipv4.sysctl_icmp_echo_ignore_broadcasts;
823		table[2].data =
824			&net->ipv4.sysctl_icmp_ignore_bogus_error_responses;
825		table[3].data =
826			&net->ipv4.sysctl_icmp_errors_use_inbound_ifaddr;
827		table[4].data =
828			&net->ipv4.sysctl_icmp_ratelimit;
829		table[5].data =
830			&net->ipv4.sysctl_icmp_ratemask;
831		table[6].data =
832			&net->ipv4.sysctl_rt_cache_rebuild_count;
833		table[7].data =
834			&net->ipv4.sysctl_ping_group_range;
835
836	}
837
838	/*
839	 * Sane defaults - nobody may create ping sockets.
840	 * Boot scripts should set this to distro-specific group.
841	 */
842	net->ipv4.sysctl_ping_group_range[0] = 1;
843	net->ipv4.sysctl_ping_group_range[1] = 0;
844
845	net->ipv4.sysctl_rt_cache_rebuild_count = 4;
846
847	tcp_init_mem(net);
848
849	net->ipv4.ipv4_hdr = register_net_sysctl(net, "net/ipv4", table);
850	if (net->ipv4.ipv4_hdr == NULL)
851		goto err_reg;
852
853	return 0;
854
855err_reg:
856	if (!net_eq(net, &init_net))
857		kfree(table);
858err_alloc:
859	return -ENOMEM;
860}
861
862static __net_exit void ipv4_sysctl_exit_net(struct net *net)
863{
864	struct ctl_table *table;
865
866	table = net->ipv4.ipv4_hdr->ctl_table_arg;
867	unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
868	kfree(table);
869}
870
871static __net_initdata struct pernet_operations ipv4_sysctl_ops = {
872	.init = ipv4_sysctl_init_net,
873	.exit = ipv4_sysctl_exit_net,
874};
875
876static __init int sysctl_ipv4_init(void)
877{
878	struct ctl_table_header *hdr;
879	struct ctl_table *i;
880
881	for (i = ipv4_table; i->procname; i++) {
882		if (strcmp(i->procname, "ip_local_reserved_ports") == 0) {
883			i->data = sysctl_local_reserved_ports;
884			break;
885		}
886	}
887	if (!i->procname)
888		return -EINVAL;
889
890	hdr = register_net_sysctl(&init_net, "net/ipv4", ipv4_table);
891	if (hdr == NULL)
892		return -ENOMEM;
893
894	if (register_pernet_subsys(&ipv4_sysctl_ops)) {
895		unregister_net_sysctl_table(hdr);
896		return -ENOMEM;
897	}
898
899	return 0;
900}
901
902__initcall(sysctl_ipv4_init);
903