xref: /openbmc/linux/net/ipv4/sysctl_net_ipv4.c (revision b627b4ed)
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 <net/snmp.h>
16 #include <net/icmp.h>
17 #include <net/ip.h>
18 #include <net/route.h>
19 #include <net/tcp.h>
20 #include <net/udp.h>
21 #include <net/cipso_ipv4.h>
22 #include <net/inet_frag.h>
23 
24 static int zero;
25 static int tcp_retr1_max = 255;
26 static int ip_local_port_range_min[] = { 1, 1 };
27 static int ip_local_port_range_max[] = { 65535, 65535 };
28 
29 /* Update system visible IP port range */
30 static void set_local_port_range(int range[2])
31 {
32 	write_seqlock(&sysctl_local_ports.lock);
33 	sysctl_local_ports.range[0] = range[0];
34 	sysctl_local_ports.range[1] = range[1];
35 	write_sequnlock(&sysctl_local_ports.lock);
36 }
37 
38 /* Validate changes from /proc interface. */
39 static int ipv4_local_port_range(ctl_table *table, int write, struct file *filp,
40 				 void __user *buffer,
41 				 size_t *lenp, loff_t *ppos)
42 {
43 	int ret;
44 	int range[2];
45 	ctl_table tmp = {
46 		.data = &range,
47 		.maxlen = sizeof(range),
48 		.mode = table->mode,
49 		.extra1 = &ip_local_port_range_min,
50 		.extra2 = &ip_local_port_range_max,
51 	};
52 
53 	inet_get_local_port_range(range, range + 1);
54 	ret = proc_dointvec_minmax(&tmp, write, filp, buffer, lenp, ppos);
55 
56 	if (write && ret == 0) {
57 		if (range[1] < range[0])
58 			ret = -EINVAL;
59 		else
60 			set_local_port_range(range);
61 	}
62 
63 	return ret;
64 }
65 
66 /* Validate changes from sysctl interface. */
67 static int ipv4_sysctl_local_port_range(ctl_table *table,
68 					 void __user *oldval,
69 					 size_t __user *oldlenp,
70 					void __user *newval, size_t newlen)
71 {
72 	int ret;
73 	int range[2];
74 	ctl_table tmp = {
75 		.data = &range,
76 		.maxlen = sizeof(range),
77 		.mode = table->mode,
78 		.extra1 = &ip_local_port_range_min,
79 		.extra2 = &ip_local_port_range_max,
80 	};
81 
82 	inet_get_local_port_range(range, range + 1);
83 	ret = sysctl_intvec(&tmp, oldval, oldlenp, newval, newlen);
84 	if (ret == 0 && newval && newlen) {
85 		if (range[1] < range[0])
86 			ret = -EINVAL;
87 		else
88 			set_local_port_range(range);
89 	}
90 	return ret;
91 }
92 
93 
94 static int proc_tcp_congestion_control(ctl_table *ctl, int write, struct file * filp,
95 				       void __user *buffer, size_t *lenp, loff_t *ppos)
96 {
97 	char val[TCP_CA_NAME_MAX];
98 	ctl_table tbl = {
99 		.data = val,
100 		.maxlen = TCP_CA_NAME_MAX,
101 	};
102 	int ret;
103 
104 	tcp_get_default_congestion_control(val);
105 
106 	ret = proc_dostring(&tbl, write, filp, buffer, lenp, ppos);
107 	if (write && ret == 0)
108 		ret = tcp_set_default_congestion_control(val);
109 	return ret;
110 }
111 
112 static int sysctl_tcp_congestion_control(ctl_table *table,
113 					 void __user *oldval,
114 					 size_t __user *oldlenp,
115 					 void __user *newval, size_t newlen)
116 {
117 	char val[TCP_CA_NAME_MAX];
118 	ctl_table tbl = {
119 		.data = val,
120 		.maxlen = TCP_CA_NAME_MAX,
121 	};
122 	int ret;
123 
124 	tcp_get_default_congestion_control(val);
125 	ret = sysctl_string(&tbl, oldval, oldlenp, newval, newlen);
126 	if (ret == 1 && newval && newlen)
127 		ret = tcp_set_default_congestion_control(val);
128 	return ret;
129 }
130 
131 static int proc_tcp_available_congestion_control(ctl_table *ctl,
132 						 int write, struct file * filp,
133 						 void __user *buffer, size_t *lenp,
134 						 loff_t *ppos)
135 {
136 	ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX, };
137 	int ret;
138 
139 	tbl.data = kmalloc(tbl.maxlen, GFP_USER);
140 	if (!tbl.data)
141 		return -ENOMEM;
142 	tcp_get_available_congestion_control(tbl.data, TCP_CA_BUF_MAX);
143 	ret = proc_dostring(&tbl, write, filp, buffer, lenp, ppos);
144 	kfree(tbl.data);
145 	return ret;
146 }
147 
148 static int proc_allowed_congestion_control(ctl_table *ctl,
149 					   int write, struct file * filp,
150 					   void __user *buffer, size_t *lenp,
151 					   loff_t *ppos)
152 {
153 	ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX };
154 	int ret;
155 
156 	tbl.data = kmalloc(tbl.maxlen, GFP_USER);
157 	if (!tbl.data)
158 		return -ENOMEM;
159 
160 	tcp_get_allowed_congestion_control(tbl.data, tbl.maxlen);
161 	ret = proc_dostring(&tbl, write, filp, buffer, lenp, ppos);
162 	if (write && ret == 0)
163 		ret = tcp_set_allowed_congestion_control(tbl.data);
164 	kfree(tbl.data);
165 	return ret;
166 }
167 
168 static int strategy_allowed_congestion_control(ctl_table *table,
169 					       void __user *oldval,
170 					       size_t __user *oldlenp,
171 					       void __user *newval,
172 					       size_t newlen)
173 {
174 	ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX };
175 	int ret;
176 
177 	tbl.data = kmalloc(tbl.maxlen, GFP_USER);
178 	if (!tbl.data)
179 		return -ENOMEM;
180 
181 	tcp_get_available_congestion_control(tbl.data, tbl.maxlen);
182 	ret = sysctl_string(&tbl, oldval, oldlenp, newval, newlen);
183 	if (ret == 1 && newval && newlen)
184 		ret = tcp_set_allowed_congestion_control(tbl.data);
185 	kfree(tbl.data);
186 
187 	return ret;
188 
189 }
190 
191 static struct ctl_table ipv4_table[] = {
192 	{
193 		.ctl_name	= NET_IPV4_TCP_TIMESTAMPS,
194 		.procname	= "tcp_timestamps",
195 		.data		= &sysctl_tcp_timestamps,
196 		.maxlen		= sizeof(int),
197 		.mode		= 0644,
198 		.proc_handler	= proc_dointvec
199 	},
200 	{
201 		.ctl_name	= NET_IPV4_TCP_WINDOW_SCALING,
202 		.procname	= "tcp_window_scaling",
203 		.data		= &sysctl_tcp_window_scaling,
204 		.maxlen		= sizeof(int),
205 		.mode		= 0644,
206 		.proc_handler	= proc_dointvec
207 	},
208 	{
209 		.ctl_name	= NET_IPV4_TCP_SACK,
210 		.procname	= "tcp_sack",
211 		.data		= &sysctl_tcp_sack,
212 		.maxlen		= sizeof(int),
213 		.mode		= 0644,
214 		.proc_handler	= proc_dointvec
215 	},
216 	{
217 		.ctl_name	= NET_IPV4_TCP_RETRANS_COLLAPSE,
218 		.procname	= "tcp_retrans_collapse",
219 		.data		= &sysctl_tcp_retrans_collapse,
220 		.maxlen		= sizeof(int),
221 		.mode		= 0644,
222 		.proc_handler	= proc_dointvec
223 	},
224 	{
225 		.ctl_name	= NET_IPV4_DEFAULT_TTL,
226 		.procname	= "ip_default_ttl",
227 		.data		= &sysctl_ip_default_ttl,
228 		.maxlen		= sizeof(int),
229 		.mode		= 0644,
230 		.proc_handler	= ipv4_doint_and_flush,
231 		.strategy	= ipv4_doint_and_flush_strategy,
232 		.extra2		= &init_net,
233 	},
234 	{
235 		.ctl_name	= NET_IPV4_NO_PMTU_DISC,
236 		.procname	= "ip_no_pmtu_disc",
237 		.data		= &ipv4_config.no_pmtu_disc,
238 		.maxlen		= sizeof(int),
239 		.mode		= 0644,
240 		.proc_handler	= proc_dointvec
241 	},
242 	{
243 		.ctl_name	= NET_IPV4_NONLOCAL_BIND,
244 		.procname	= "ip_nonlocal_bind",
245 		.data		= &sysctl_ip_nonlocal_bind,
246 		.maxlen		= sizeof(int),
247 		.mode		= 0644,
248 		.proc_handler	= proc_dointvec
249 	},
250 	{
251 		.ctl_name	= NET_IPV4_TCP_SYN_RETRIES,
252 		.procname	= "tcp_syn_retries",
253 		.data		= &sysctl_tcp_syn_retries,
254 		.maxlen		= sizeof(int),
255 		.mode		= 0644,
256 		.proc_handler	= proc_dointvec
257 	},
258 	{
259 		.ctl_name	= NET_TCP_SYNACK_RETRIES,
260 		.procname	= "tcp_synack_retries",
261 		.data		= &sysctl_tcp_synack_retries,
262 		.maxlen		= sizeof(int),
263 		.mode		= 0644,
264 		.proc_handler	= proc_dointvec
265 	},
266 	{
267 		.ctl_name	= NET_TCP_MAX_ORPHANS,
268 		.procname	= "tcp_max_orphans",
269 		.data		= &sysctl_tcp_max_orphans,
270 		.maxlen		= sizeof(int),
271 		.mode		= 0644,
272 		.proc_handler	= proc_dointvec
273 	},
274 	{
275 		.ctl_name	= NET_TCP_MAX_TW_BUCKETS,
276 		.procname	= "tcp_max_tw_buckets",
277 		.data		= &tcp_death_row.sysctl_max_tw_buckets,
278 		.maxlen		= sizeof(int),
279 		.mode		= 0644,
280 		.proc_handler	= proc_dointvec
281 	},
282 	{
283 		.ctl_name	= NET_IPV4_DYNADDR,
284 		.procname	= "ip_dynaddr",
285 		.data		= &sysctl_ip_dynaddr,
286 		.maxlen		= sizeof(int),
287 		.mode		= 0644,
288 		.proc_handler	= proc_dointvec
289 	},
290 	{
291 		.ctl_name	= NET_IPV4_TCP_KEEPALIVE_TIME,
292 		.procname	= "tcp_keepalive_time",
293 		.data		= &sysctl_tcp_keepalive_time,
294 		.maxlen		= sizeof(int),
295 		.mode		= 0644,
296 		.proc_handler	= proc_dointvec_jiffies,
297 		.strategy	= sysctl_jiffies
298 	},
299 	{
300 		.ctl_name	= NET_IPV4_TCP_KEEPALIVE_PROBES,
301 		.procname	= "tcp_keepalive_probes",
302 		.data		= &sysctl_tcp_keepalive_probes,
303 		.maxlen		= sizeof(int),
304 		.mode		= 0644,
305 		.proc_handler	= proc_dointvec
306 	},
307 	{
308 		.ctl_name	= NET_IPV4_TCP_KEEPALIVE_INTVL,
309 		.procname	= "tcp_keepalive_intvl",
310 		.data		= &sysctl_tcp_keepalive_intvl,
311 		.maxlen		= sizeof(int),
312 		.mode		= 0644,
313 		.proc_handler	= proc_dointvec_jiffies,
314 		.strategy	= sysctl_jiffies
315 	},
316 	{
317 		.ctl_name	= NET_IPV4_TCP_RETRIES1,
318 		.procname	= "tcp_retries1",
319 		.data		= &sysctl_tcp_retries1,
320 		.maxlen		= sizeof(int),
321 		.mode		= 0644,
322 		.proc_handler	= proc_dointvec_minmax,
323 		.strategy	= sysctl_intvec,
324 		.extra2		= &tcp_retr1_max
325 	},
326 	{
327 		.ctl_name	= NET_IPV4_TCP_RETRIES2,
328 		.procname	= "tcp_retries2",
329 		.data		= &sysctl_tcp_retries2,
330 		.maxlen		= sizeof(int),
331 		.mode		= 0644,
332 		.proc_handler	= proc_dointvec
333 	},
334 	{
335 		.ctl_name	= NET_IPV4_TCP_FIN_TIMEOUT,
336 		.procname	= "tcp_fin_timeout",
337 		.data		= &sysctl_tcp_fin_timeout,
338 		.maxlen		= sizeof(int),
339 		.mode		= 0644,
340 		.proc_handler	= proc_dointvec_jiffies,
341 		.strategy	= sysctl_jiffies
342 	},
343 #ifdef CONFIG_SYN_COOKIES
344 	{
345 		.ctl_name	= NET_TCP_SYNCOOKIES,
346 		.procname	= "tcp_syncookies",
347 		.data		= &sysctl_tcp_syncookies,
348 		.maxlen		= sizeof(int),
349 		.mode		= 0644,
350 		.proc_handler	= proc_dointvec
351 	},
352 #endif
353 	{
354 		.ctl_name	= NET_TCP_TW_RECYCLE,
355 		.procname	= "tcp_tw_recycle",
356 		.data		= &tcp_death_row.sysctl_tw_recycle,
357 		.maxlen		= sizeof(int),
358 		.mode		= 0644,
359 		.proc_handler	= proc_dointvec
360 	},
361 	{
362 		.ctl_name	= NET_TCP_ABORT_ON_OVERFLOW,
363 		.procname	= "tcp_abort_on_overflow",
364 		.data		= &sysctl_tcp_abort_on_overflow,
365 		.maxlen		= sizeof(int),
366 		.mode		= 0644,
367 		.proc_handler	= proc_dointvec
368 	},
369 	{
370 		.ctl_name	= NET_TCP_STDURG,
371 		.procname	= "tcp_stdurg",
372 		.data		= &sysctl_tcp_stdurg,
373 		.maxlen		= sizeof(int),
374 		.mode		= 0644,
375 		.proc_handler	= proc_dointvec
376 	},
377 	{
378 		.ctl_name	= NET_TCP_RFC1337,
379 		.procname	= "tcp_rfc1337",
380 		.data		= &sysctl_tcp_rfc1337,
381 		.maxlen		= sizeof(int),
382 		.mode		= 0644,
383 		.proc_handler	= proc_dointvec
384 	},
385 	{
386 		.ctl_name	= NET_TCP_MAX_SYN_BACKLOG,
387 		.procname	= "tcp_max_syn_backlog",
388 		.data		= &sysctl_max_syn_backlog,
389 		.maxlen		= sizeof(int),
390 		.mode		= 0644,
391 		.proc_handler	= proc_dointvec
392 	},
393 	{
394 		.ctl_name	= NET_IPV4_LOCAL_PORT_RANGE,
395 		.procname	= "ip_local_port_range",
396 		.data		= &sysctl_local_ports.range,
397 		.maxlen		= sizeof(sysctl_local_ports.range),
398 		.mode		= 0644,
399 		.proc_handler	= ipv4_local_port_range,
400 		.strategy	= ipv4_sysctl_local_port_range,
401 	},
402 #ifdef CONFIG_IP_MULTICAST
403 	{
404 		.ctl_name	= NET_IPV4_IGMP_MAX_MEMBERSHIPS,
405 		.procname	= "igmp_max_memberships",
406 		.data		= &sysctl_igmp_max_memberships,
407 		.maxlen		= sizeof(int),
408 		.mode		= 0644,
409 		.proc_handler	= proc_dointvec
410 	},
411 
412 #endif
413 	{
414 		.ctl_name	= NET_IPV4_IGMP_MAX_MSF,
415 		.procname	= "igmp_max_msf",
416 		.data		= &sysctl_igmp_max_msf,
417 		.maxlen		= sizeof(int),
418 		.mode		= 0644,
419 		.proc_handler	= proc_dointvec
420 	},
421 	{
422 		.ctl_name	= NET_IPV4_INET_PEER_THRESHOLD,
423 		.procname	= "inet_peer_threshold",
424 		.data		= &inet_peer_threshold,
425 		.maxlen		= sizeof(int),
426 		.mode		= 0644,
427 		.proc_handler	= proc_dointvec
428 	},
429 	{
430 		.ctl_name	= NET_IPV4_INET_PEER_MINTTL,
431 		.procname	= "inet_peer_minttl",
432 		.data		= &inet_peer_minttl,
433 		.maxlen		= sizeof(int),
434 		.mode		= 0644,
435 		.proc_handler	= proc_dointvec_jiffies,
436 		.strategy	= sysctl_jiffies
437 	},
438 	{
439 		.ctl_name	= NET_IPV4_INET_PEER_MAXTTL,
440 		.procname	= "inet_peer_maxttl",
441 		.data		= &inet_peer_maxttl,
442 		.maxlen		= sizeof(int),
443 		.mode		= 0644,
444 		.proc_handler	= proc_dointvec_jiffies,
445 		.strategy	= sysctl_jiffies
446 	},
447 	{
448 		.ctl_name	= NET_IPV4_INET_PEER_GC_MINTIME,
449 		.procname	= "inet_peer_gc_mintime",
450 		.data		= &inet_peer_gc_mintime,
451 		.maxlen		= sizeof(int),
452 		.mode		= 0644,
453 		.proc_handler	= proc_dointvec_jiffies,
454 		.strategy	= sysctl_jiffies
455 	},
456 	{
457 		.ctl_name	= NET_IPV4_INET_PEER_GC_MAXTIME,
458 		.procname	= "inet_peer_gc_maxtime",
459 		.data		= &inet_peer_gc_maxtime,
460 		.maxlen		= sizeof(int),
461 		.mode		= 0644,
462 		.proc_handler	= proc_dointvec_jiffies,
463 		.strategy	= sysctl_jiffies
464 	},
465 	{
466 		.ctl_name	= NET_TCP_ORPHAN_RETRIES,
467 		.procname	= "tcp_orphan_retries",
468 		.data		= &sysctl_tcp_orphan_retries,
469 		.maxlen		= sizeof(int),
470 		.mode		= 0644,
471 		.proc_handler	= proc_dointvec
472 	},
473 	{
474 		.ctl_name	= NET_TCP_FACK,
475 		.procname	= "tcp_fack",
476 		.data		= &sysctl_tcp_fack,
477 		.maxlen		= sizeof(int),
478 		.mode		= 0644,
479 		.proc_handler	= proc_dointvec
480 	},
481 	{
482 		.ctl_name	= NET_TCP_REORDERING,
483 		.procname	= "tcp_reordering",
484 		.data		= &sysctl_tcp_reordering,
485 		.maxlen		= sizeof(int),
486 		.mode		= 0644,
487 		.proc_handler	= proc_dointvec
488 	},
489 	{
490 		.ctl_name	= NET_TCP_ECN,
491 		.procname	= "tcp_ecn",
492 		.data		= &sysctl_tcp_ecn,
493 		.maxlen		= sizeof(int),
494 		.mode		= 0644,
495 		.proc_handler	= proc_dointvec
496 	},
497 	{
498 		.ctl_name	= NET_TCP_DSACK,
499 		.procname	= "tcp_dsack",
500 		.data		= &sysctl_tcp_dsack,
501 		.maxlen		= sizeof(int),
502 		.mode		= 0644,
503 		.proc_handler	= proc_dointvec
504 	},
505 	{
506 		.ctl_name	= NET_TCP_MEM,
507 		.procname	= "tcp_mem",
508 		.data		= &sysctl_tcp_mem,
509 		.maxlen		= sizeof(sysctl_tcp_mem),
510 		.mode		= 0644,
511 		.proc_handler	= proc_dointvec
512 	},
513 	{
514 		.ctl_name	= NET_TCP_WMEM,
515 		.procname	= "tcp_wmem",
516 		.data		= &sysctl_tcp_wmem,
517 		.maxlen		= sizeof(sysctl_tcp_wmem),
518 		.mode		= 0644,
519 		.proc_handler	= proc_dointvec
520 	},
521 	{
522 		.ctl_name	= NET_TCP_RMEM,
523 		.procname	= "tcp_rmem",
524 		.data		= &sysctl_tcp_rmem,
525 		.maxlen		= sizeof(sysctl_tcp_rmem),
526 		.mode		= 0644,
527 		.proc_handler	= proc_dointvec
528 	},
529 	{
530 		.ctl_name	= NET_TCP_APP_WIN,
531 		.procname	= "tcp_app_win",
532 		.data		= &sysctl_tcp_app_win,
533 		.maxlen		= sizeof(int),
534 		.mode		= 0644,
535 		.proc_handler	= proc_dointvec
536 	},
537 	{
538 		.ctl_name	= NET_TCP_ADV_WIN_SCALE,
539 		.procname	= "tcp_adv_win_scale",
540 		.data		= &sysctl_tcp_adv_win_scale,
541 		.maxlen		= sizeof(int),
542 		.mode		= 0644,
543 		.proc_handler	= proc_dointvec
544 	},
545 	{
546 		.ctl_name	= NET_TCP_TW_REUSE,
547 		.procname	= "tcp_tw_reuse",
548 		.data		= &sysctl_tcp_tw_reuse,
549 		.maxlen		= sizeof(int),
550 		.mode		= 0644,
551 		.proc_handler	= proc_dointvec
552 	},
553 	{
554 		.ctl_name	= NET_TCP_FRTO,
555 		.procname	= "tcp_frto",
556 		.data		= &sysctl_tcp_frto,
557 		.maxlen		= sizeof(int),
558 		.mode		= 0644,
559 		.proc_handler	= proc_dointvec
560 	},
561 	{
562 		.ctl_name	= NET_TCP_FRTO_RESPONSE,
563 		.procname	= "tcp_frto_response",
564 		.data		= &sysctl_tcp_frto_response,
565 		.maxlen		= sizeof(int),
566 		.mode		= 0644,
567 		.proc_handler	= proc_dointvec
568 	},
569 	{
570 		.ctl_name	= NET_TCP_LOW_LATENCY,
571 		.procname	= "tcp_low_latency",
572 		.data		= &sysctl_tcp_low_latency,
573 		.maxlen		= sizeof(int),
574 		.mode		= 0644,
575 		.proc_handler	= proc_dointvec
576 	},
577 	{
578 		.ctl_name	= NET_TCP_NO_METRICS_SAVE,
579 		.procname	= "tcp_no_metrics_save",
580 		.data		= &sysctl_tcp_nometrics_save,
581 		.maxlen		= sizeof(int),
582 		.mode		= 0644,
583 		.proc_handler	= proc_dointvec,
584 	},
585 	{
586 		.ctl_name	= NET_TCP_MODERATE_RCVBUF,
587 		.procname	= "tcp_moderate_rcvbuf",
588 		.data		= &sysctl_tcp_moderate_rcvbuf,
589 		.maxlen		= sizeof(int),
590 		.mode		= 0644,
591 		.proc_handler	= proc_dointvec,
592 	},
593 	{
594 		.ctl_name	= NET_TCP_TSO_WIN_DIVISOR,
595 		.procname	= "tcp_tso_win_divisor",
596 		.data		= &sysctl_tcp_tso_win_divisor,
597 		.maxlen		= sizeof(int),
598 		.mode		= 0644,
599 		.proc_handler	= proc_dointvec,
600 	},
601 	{
602 		.ctl_name	= NET_TCP_CONG_CONTROL,
603 		.procname	= "tcp_congestion_control",
604 		.mode		= 0644,
605 		.maxlen		= TCP_CA_NAME_MAX,
606 		.proc_handler	= proc_tcp_congestion_control,
607 		.strategy	= sysctl_tcp_congestion_control,
608 	},
609 	{
610 		.ctl_name	= NET_TCP_ABC,
611 		.procname	= "tcp_abc",
612 		.data		= &sysctl_tcp_abc,
613 		.maxlen		= sizeof(int),
614 		.mode		= 0644,
615 		.proc_handler	= proc_dointvec,
616 	},
617 	{
618 		.ctl_name	= NET_TCP_MTU_PROBING,
619 		.procname	= "tcp_mtu_probing",
620 		.data		= &sysctl_tcp_mtu_probing,
621 		.maxlen		= sizeof(int),
622 		.mode		= 0644,
623 		.proc_handler	= proc_dointvec,
624 	},
625 	{
626 		.ctl_name	= NET_TCP_BASE_MSS,
627 		.procname	= "tcp_base_mss",
628 		.data		= &sysctl_tcp_base_mss,
629 		.maxlen		= sizeof(int),
630 		.mode		= 0644,
631 		.proc_handler	= proc_dointvec,
632 	},
633 	{
634 		.ctl_name	= NET_IPV4_TCP_WORKAROUND_SIGNED_WINDOWS,
635 		.procname	= "tcp_workaround_signed_windows",
636 		.data		= &sysctl_tcp_workaround_signed_windows,
637 		.maxlen		= sizeof(int),
638 		.mode		= 0644,
639 		.proc_handler	= proc_dointvec
640 	},
641 #ifdef CONFIG_NET_DMA
642 	{
643 		.ctl_name	= NET_TCP_DMA_COPYBREAK,
644 		.procname	= "tcp_dma_copybreak",
645 		.data		= &sysctl_tcp_dma_copybreak,
646 		.maxlen		= sizeof(int),
647 		.mode		= 0644,
648 		.proc_handler	= proc_dointvec
649 	},
650 #endif
651 	{
652 		.ctl_name	= NET_TCP_SLOW_START_AFTER_IDLE,
653 		.procname	= "tcp_slow_start_after_idle",
654 		.data		= &sysctl_tcp_slow_start_after_idle,
655 		.maxlen		= sizeof(int),
656 		.mode		= 0644,
657 		.proc_handler	= proc_dointvec
658 	},
659 #ifdef CONFIG_NETLABEL
660 	{
661 		.ctl_name	= NET_CIPSOV4_CACHE_ENABLE,
662 		.procname	= "cipso_cache_enable",
663 		.data		= &cipso_v4_cache_enabled,
664 		.maxlen		= sizeof(int),
665 		.mode		= 0644,
666 		.proc_handler	= proc_dointvec,
667 	},
668 	{
669 		.ctl_name	= NET_CIPSOV4_CACHE_BUCKET_SIZE,
670 		.procname	= "cipso_cache_bucket_size",
671 		.data		= &cipso_v4_cache_bucketsize,
672 		.maxlen		= sizeof(int),
673 		.mode		= 0644,
674 		.proc_handler	= proc_dointvec,
675 	},
676 	{
677 		.ctl_name	= NET_CIPSOV4_RBM_OPTFMT,
678 		.procname	= "cipso_rbm_optfmt",
679 		.data		= &cipso_v4_rbm_optfmt,
680 		.maxlen		= sizeof(int),
681 		.mode		= 0644,
682 		.proc_handler	= proc_dointvec,
683 	},
684 	{
685 		.ctl_name	= NET_CIPSOV4_RBM_STRICTVALID,
686 		.procname	= "cipso_rbm_strictvalid",
687 		.data		= &cipso_v4_rbm_strictvalid,
688 		.maxlen		= sizeof(int),
689 		.mode		= 0644,
690 		.proc_handler	= proc_dointvec,
691 	},
692 #endif /* CONFIG_NETLABEL */
693 	{
694 		.procname	= "tcp_available_congestion_control",
695 		.maxlen		= TCP_CA_BUF_MAX,
696 		.mode		= 0444,
697 		.proc_handler   = proc_tcp_available_congestion_control,
698 	},
699 	{
700 		.ctl_name	= NET_TCP_ALLOWED_CONG_CONTROL,
701 		.procname	= "tcp_allowed_congestion_control",
702 		.maxlen		= TCP_CA_BUF_MAX,
703 		.mode		= 0644,
704 		.proc_handler   = proc_allowed_congestion_control,
705 		.strategy	= strategy_allowed_congestion_control,
706 	},
707 	{
708 		.ctl_name	= NET_TCP_MAX_SSTHRESH,
709 		.procname	= "tcp_max_ssthresh",
710 		.data		= &sysctl_tcp_max_ssthresh,
711 		.maxlen		= sizeof(int),
712 		.mode		= 0644,
713 		.proc_handler	= proc_dointvec,
714 	},
715 	{
716 		.ctl_name	= CTL_UNNUMBERED,
717 		.procname	= "udp_mem",
718 		.data		= &sysctl_udp_mem,
719 		.maxlen		= sizeof(sysctl_udp_mem),
720 		.mode		= 0644,
721 		.proc_handler	= proc_dointvec_minmax,
722 		.strategy	= sysctl_intvec,
723 		.extra1		= &zero
724 	},
725 	{
726 		.ctl_name	= CTL_UNNUMBERED,
727 		.procname	= "udp_rmem_min",
728 		.data		= &sysctl_udp_rmem_min,
729 		.maxlen		= sizeof(sysctl_udp_rmem_min),
730 		.mode		= 0644,
731 		.proc_handler	= proc_dointvec_minmax,
732 		.strategy	= sysctl_intvec,
733 		.extra1		= &zero
734 	},
735 	{
736 		.ctl_name	= CTL_UNNUMBERED,
737 		.procname	= "udp_wmem_min",
738 		.data		= &sysctl_udp_wmem_min,
739 		.maxlen		= sizeof(sysctl_udp_wmem_min),
740 		.mode		= 0644,
741 		.proc_handler	= proc_dointvec_minmax,
742 		.strategy	= sysctl_intvec,
743 		.extra1		= &zero
744 	},
745 	{ .ctl_name = 0 }
746 };
747 
748 static struct ctl_table ipv4_net_table[] = {
749 	{
750 		.ctl_name	= NET_IPV4_ICMP_ECHO_IGNORE_ALL,
751 		.procname	= "icmp_echo_ignore_all",
752 		.data		= &init_net.ipv4.sysctl_icmp_echo_ignore_all,
753 		.maxlen		= sizeof(int),
754 		.mode		= 0644,
755 		.proc_handler	= proc_dointvec
756 	},
757 	{
758 		.ctl_name	= NET_IPV4_ICMP_ECHO_IGNORE_BROADCASTS,
759 		.procname	= "icmp_echo_ignore_broadcasts",
760 		.data		= &init_net.ipv4.sysctl_icmp_echo_ignore_broadcasts,
761 		.maxlen		= sizeof(int),
762 		.mode		= 0644,
763 		.proc_handler	= proc_dointvec
764 	},
765 	{
766 		.ctl_name	= NET_IPV4_ICMP_IGNORE_BOGUS_ERROR_RESPONSES,
767 		.procname	= "icmp_ignore_bogus_error_responses",
768 		.data		= &init_net.ipv4.sysctl_icmp_ignore_bogus_error_responses,
769 		.maxlen		= sizeof(int),
770 		.mode		= 0644,
771 		.proc_handler	= proc_dointvec
772 	},
773 	{
774 		.ctl_name	= NET_IPV4_ICMP_ERRORS_USE_INBOUND_IFADDR,
775 		.procname	= "icmp_errors_use_inbound_ifaddr",
776 		.data		= &init_net.ipv4.sysctl_icmp_errors_use_inbound_ifaddr,
777 		.maxlen		= sizeof(int),
778 		.mode		= 0644,
779 		.proc_handler	= proc_dointvec
780 	},
781 	{
782 		.ctl_name	= NET_IPV4_ICMP_RATELIMIT,
783 		.procname	= "icmp_ratelimit",
784 		.data		= &init_net.ipv4.sysctl_icmp_ratelimit,
785 		.maxlen		= sizeof(int),
786 		.mode		= 0644,
787 		.proc_handler	= proc_dointvec_ms_jiffies,
788 		.strategy	= sysctl_ms_jiffies
789 	},
790 	{
791 		.ctl_name	= NET_IPV4_ICMP_RATEMASK,
792 		.procname	= "icmp_ratemask",
793 		.data		= &init_net.ipv4.sysctl_icmp_ratemask,
794 		.maxlen		= sizeof(int),
795 		.mode		= 0644,
796 		.proc_handler	= proc_dointvec
797 	},
798 	{
799 		.ctl_name	= CTL_UNNUMBERED,
800 		.procname	= "rt_cache_rebuild_count",
801 		.data		= &init_net.ipv4.sysctl_rt_cache_rebuild_count,
802 		.maxlen		= sizeof(int),
803 		.mode		= 0644,
804 		.proc_handler	= proc_dointvec
805 	},
806 	{ }
807 };
808 
809 struct ctl_path net_ipv4_ctl_path[] = {
810 	{ .procname = "net", .ctl_name = CTL_NET, },
811 	{ .procname = "ipv4", .ctl_name = NET_IPV4, },
812 	{ },
813 };
814 EXPORT_SYMBOL_GPL(net_ipv4_ctl_path);
815 
816 static __net_init int ipv4_sysctl_init_net(struct net *net)
817 {
818 	struct ctl_table *table;
819 
820 	table = ipv4_net_table;
821 	if (net != &init_net) {
822 		table = kmemdup(table, sizeof(ipv4_net_table), GFP_KERNEL);
823 		if (table == NULL)
824 			goto err_alloc;
825 
826 		table[0].data =
827 			&net->ipv4.sysctl_icmp_echo_ignore_all;
828 		table[1].data =
829 			&net->ipv4.sysctl_icmp_echo_ignore_broadcasts;
830 		table[2].data =
831 			&net->ipv4.sysctl_icmp_ignore_bogus_error_responses;
832 		table[3].data =
833 			&net->ipv4.sysctl_icmp_errors_use_inbound_ifaddr;
834 		table[4].data =
835 			&net->ipv4.sysctl_icmp_ratelimit;
836 		table[5].data =
837 			&net->ipv4.sysctl_icmp_ratemask;
838 		table[6].data =
839 			&net->ipv4.sysctl_rt_cache_rebuild_count;
840 	}
841 
842 	net->ipv4.sysctl_rt_cache_rebuild_count = 4;
843 
844 	net->ipv4.ipv4_hdr = register_net_sysctl_table(net,
845 			net_ipv4_ctl_path, table);
846 	if (net->ipv4.ipv4_hdr == NULL)
847 		goto err_reg;
848 
849 	return 0;
850 
851 err_reg:
852 	if (net != &init_net)
853 		kfree(table);
854 err_alloc:
855 	return -ENOMEM;
856 }
857 
858 static __net_exit void ipv4_sysctl_exit_net(struct net *net)
859 {
860 	struct ctl_table *table;
861 
862 	table = net->ipv4.ipv4_hdr->ctl_table_arg;
863 	unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
864 	kfree(table);
865 }
866 
867 static __net_initdata struct pernet_operations ipv4_sysctl_ops = {
868 	.init = ipv4_sysctl_init_net,
869 	.exit = ipv4_sysctl_exit_net,
870 };
871 
872 static __init int sysctl_ipv4_init(void)
873 {
874 	struct ctl_table_header *hdr;
875 
876 	hdr = register_sysctl_paths(net_ipv4_ctl_path, ipv4_table);
877 	if (hdr == NULL)
878 		return -ENOMEM;
879 
880 	if (register_pernet_subsys(&ipv4_sysctl_ops)) {
881 		unregister_sysctl_table(hdr);
882 		return -ENOMEM;
883 	}
884 
885 	return 0;
886 }
887 
888 __initcall(sysctl_ipv4_init);
889