sysctl_net_ipv4.c

   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 tcp_rto_max_max = TCP_RTO_MAX_SEC * MSEC_PER_SEC;
  32static int ip_privileged_port_min;
  33static int ip_privileged_port_max = 65535;
  34static int ip_ttl_min = 1;
  35static int ip_ttl_max = 255;
  36static int tcp_syn_retries_min = 1;
  37static int tcp_syn_retries_max = MAX_TCP_SYNCNT;
  38static int tcp_syn_linear_timeouts_max = MAX_TCP_SYNCNT;
  39static unsigned long ip_ping_group_range_min[] = { 0, 0 };
  40static unsigned long ip_ping_group_range_max[] = { GID_T_MAX, GID_T_MAX };
  41static u32 u32_max_div_HZ = UINT_MAX / HZ;
  42static int one_day_secs = 24 * 3600;
  43static u32 fib_multipath_hash_fields_all_mask __maybe_unused =
  44	FIB_MULTIPATH_HASH_FIELD_ALL_MASK;
  45static unsigned int tcp_child_ehash_entries_max = 16 * 1024 * 1024;
  46static unsigned int udp_child_hash_entries_max = UDP_HTABLE_SIZE_MAX;
  47static int tcp_plb_max_rounds = 31;
  48static int tcp_plb_max_cong_thresh = 256;
  49static unsigned int tcp_tw_reuse_delay_max = TCP_PAWS_MSL * MSEC_PER_SEC;
  50static int tcp_ecn_mode_max = 2;
  51static u32 icmp_errors_extension_mask_all =
  52	GENMASK_U8(ICMP_ERR_EXT_COUNT - 1, 0);
  53
  54/* obsolete */
  55static int sysctl_tcp_low_latency __read_mostly;
  56
  57/* Update system visible IP port range */
  58static void set_local_port_range(struct net *net, unsigned int low, unsigned int high)
  59{
  60	bool same_parity = !((low ^ high) & 1);
  61
  62	if (same_parity && !net->ipv4.ip_local_ports.warned) {
  63		net->ipv4.ip_local_ports.warned = true;
  64		pr_err_ratelimited("ip_local_port_range: prefer different parity for start/end values.\n");
  65	}
  66	WRITE_ONCE(net->ipv4.ip_local_ports.range, high << 16 | low);
  67}
  68
  69/* Validate changes from /proc interface. */
  70static int ipv4_local_port_range(const struct ctl_table *table, int write,
  71				 void *buffer, size_t *lenp, loff_t *ppos)
  72{
  73	struct net *net = table->data;
  74	int ret;
  75	int range[2];
  76	struct ctl_table tmp = {
  77		.data = &range,
  78		.maxlen = sizeof(range),
  79		.mode = table->mode,
  80		.extra1 = &ip_local_port_range_min,
  81		.extra2 = &ip_local_port_range_max,
  82	};
  83
  84	inet_get_local_port_range(net, &range[0], &range[1]);
  85
  86	ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
  87
  88	if (write && ret == 0) {
  89		/* Ensure that the upper limit is not smaller than the lower,
  90		 * and that the lower does not encroach upon the privileged
  91		 * port limit.
  92		 */
  93		if ((range[1] < range[0]) ||
  94		    (range[0] < READ_ONCE(net->ipv4.sysctl_ip_prot_sock)))
  95			ret = -EINVAL;
  96		else
  97			set_local_port_range(net, range[0], range[1]);
  98	}
  99
 100	return ret;
 101}
 102
 103/* Validate changes from /proc interface. */
 104static int ipv4_privileged_ports(const struct ctl_table *table, int write,
 105				void *buffer, size_t *lenp, loff_t *ppos)
 106{
 107	struct net *net = container_of(table->data, struct net,
 108	    ipv4.sysctl_ip_prot_sock);
 109	int ret;
 110	int pports;
 111	int range[2];
 112	struct ctl_table tmp = {
 113		.data = &pports,
 114		.maxlen = sizeof(pports),
 115		.mode = table->mode,
 116		.extra1 = &ip_privileged_port_min,
 117		.extra2 = &ip_privileged_port_max,
 118	};
 119
 120	pports = READ_ONCE(net->ipv4.sysctl_ip_prot_sock);
 121
 122	ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
 123
 124	if (write && ret == 0) {
 125		inet_get_local_port_range(net, &range[0], &range[1]);
 126		/* Ensure that the local port range doesn't overlap with the
 127		 * privileged port range.
 128		 */
 129		if (range[0] < pports)
 130			ret = -EINVAL;
 131		else
 132			WRITE_ONCE(net->ipv4.sysctl_ip_prot_sock, pports);
 133	}
 134
 135	return ret;
 136}
 137
 138static void inet_get_ping_group_range_table(const struct ctl_table *table,
 139					    kgid_t *low, kgid_t *high)
 140{
 141	kgid_t *data = table->data;
 142	struct net *net =
 143		container_of(table->data, struct net, ipv4.ping_group_range.range);
 144	unsigned int seq;
 145	do {
 146		seq = read_seqbegin(&net->ipv4.ping_group_range.lock);
 147
 148		*low = data[0];
 149		*high = data[1];
 150	} while (read_seqretry(&net->ipv4.ping_group_range.lock, seq));
 151}
 152
 153/* Update system visible IP port range */
 154static void set_ping_group_range(const struct ctl_table *table,
 155				 kgid_t low, kgid_t high)
 156{
 157	kgid_t *data = table->data;
 158	struct net *net =
 159		container_of(table->data, struct net, ipv4.ping_group_range.range);
 160	write_seqlock(&net->ipv4.ping_group_range.lock);
 161	data[0] = low;
 162	data[1] = high;
 163	write_sequnlock(&net->ipv4.ping_group_range.lock);
 164}
 165
 166/* Validate changes from /proc interface. */
 167static int ipv4_ping_group_range(const struct ctl_table *table, int write,
 168				 void *buffer, size_t *lenp, loff_t *ppos)
 169{
 170	struct user_namespace *user_ns = current_user_ns();
 171	int ret;
 172	unsigned long urange[2];
 173	kgid_t low, high;
 174	struct ctl_table tmp = {
 175		.data = &urange,
 176		.maxlen = sizeof(urange),
 177		.mode = table->mode,
 178		.extra1 = &ip_ping_group_range_min,
 179		.extra2 = &ip_ping_group_range_max,
 180	};
 181
 182	inet_get_ping_group_range_table(table, &low, &high);
 183	urange[0] = from_kgid_munged(user_ns, low);
 184	urange[1] = from_kgid_munged(user_ns, high);
 185	ret = proc_doulongvec_minmax(&tmp, write, buffer, lenp, ppos);
 186
 187	if (write && ret == 0) {
 188		low = make_kgid(user_ns, urange[0]);
 189		high = make_kgid(user_ns, urange[1]);
 190		if (!gid_valid(low) || !gid_valid(high))
 191			return -EINVAL;
 192		if (urange[1] < urange[0] || gid_lt(high, low)) {
 193			low = make_kgid(&init_user_ns, 1);
 194			high = make_kgid(&init_user_ns, 0);
 195		}
 196		set_ping_group_range(table, low, high);
 197	}
 198
 199	return ret;
 200}
 201
 202static int ipv4_fwd_update_priority(const struct ctl_table *table, int write,
 203				    void *buffer, size_t *lenp, loff_t *ppos)
 204{
 205	struct net *net;
 206	int ret;
 207
 208	net = container_of(table->data, struct net,
 209			   ipv4.sysctl_ip_fwd_update_priority);
 210	ret = proc_dou8vec_minmax(table, write, buffer, lenp, ppos);
 211	if (write && ret == 0)
 212		call_netevent_notifiers(NETEVENT_IPV4_FWD_UPDATE_PRIORITY_UPDATE,
 213					net);
 214
 215	return ret;
 216}
 217
 218static int proc_tcp_congestion_control(const struct ctl_table *ctl, int write,
 219				       void *buffer, size_t *lenp, loff_t *ppos)
 220{
 221	struct net *net = container_of(ctl->data, struct net,
 222				       ipv4.tcp_congestion_control);
 223	char val[TCP_CA_NAME_MAX];
 224	struct ctl_table tbl = {
 225		.data = val,
 226		.maxlen = TCP_CA_NAME_MAX,
 227	};
 228	int ret;
 229
 230	tcp_get_default_congestion_control(net, val);
 231
 232	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
 233	if (write && ret == 0)
 234		ret = tcp_set_default_congestion_control(net, val);
 235	return ret;
 236}
 237
 238static int proc_tcp_available_congestion_control(const struct ctl_table *ctl,
 239						 int write, void *buffer,
 240						 size_t *lenp, loff_t *ppos)
 241{
 242	struct ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX, };
 243	int ret;
 244
 245	tbl.data = kmalloc(tbl.maxlen, GFP_USER);
 246	if (!tbl.data)
 247		return -ENOMEM;
 248	tcp_get_available_congestion_control(tbl.data, TCP_CA_BUF_MAX);
 249	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
 250	kfree(tbl.data);
 251	return ret;
 252}
 253
 254static int proc_allowed_congestion_control(const struct ctl_table *ctl,
 255					   int write, void *buffer,
 256					   size_t *lenp, loff_t *ppos)
 257{
 258	struct ctl_table tbl = { .maxlen = TCP_CA_BUF_MAX };
 259	int ret;
 260
 261	tbl.data = kmalloc(tbl.maxlen, GFP_USER);
 262	if (!tbl.data)
 263		return -ENOMEM;
 264
 265	tcp_get_allowed_congestion_control(tbl.data, tbl.maxlen);
 266	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
 267	if (write && ret == 0)
 268		ret = tcp_set_allowed_congestion_control(tbl.data);
 269	kfree(tbl.data);
 270	return ret;
 271}
 272
 273static int sscanf_key(char *buf, __le32 *key)
 274{
 275	u32 user_key[4];
 276	int i, ret = 0;
 277
 278	if (sscanf(buf, "%x-%x-%x-%x", user_key, user_key + 1,
 279		   user_key + 2, user_key + 3) != 4) {
 280		ret = -EINVAL;
 281	} else {
 282		for (i = 0; i < ARRAY_SIZE(user_key); i++)
 283			key[i] = cpu_to_le32(user_key[i]);
 284	}
 285	pr_debug("proc TFO key set 0x%x-%x-%x-%x <- 0x%s: %u\n",
 286		 user_key[0], user_key[1], user_key[2], user_key[3], buf, ret);
 287
 288	return ret;
 289}
 290
 291static int proc_tcp_fastopen_key(const struct ctl_table *table, int write,
 292				 void *buffer, size_t *lenp, loff_t *ppos)
 293{
 294	struct net *net = container_of(table->data, struct net,
 295	    ipv4.sysctl_tcp_fastopen);
 296	/* maxlen to print the list of keys in hex (*2), with dashes
 297	 * separating doublewords and a comma in between keys.
 298	 */
 299	struct ctl_table tbl = { .maxlen = ((TCP_FASTOPEN_KEY_LENGTH *
 300					    2 * TCP_FASTOPEN_KEY_MAX) +
 301					    (TCP_FASTOPEN_KEY_MAX * 5)) };
 302	u32 user_key[TCP_FASTOPEN_KEY_BUF_LENGTH / sizeof(u32)];
 303	__le32 key[TCP_FASTOPEN_KEY_BUF_LENGTH / sizeof(__le32)];
 304	char *backup_data;
 305	int ret, i = 0, off = 0, n_keys;
 306
 307	tbl.data = kmalloc(tbl.maxlen, GFP_KERNEL);
 308	if (!tbl.data)
 309		return -ENOMEM;
 310
 311	n_keys = tcp_fastopen_get_cipher(net, NULL, (u64 *)key);
 312	if (!n_keys) {
 313		memset(&key[0], 0, TCP_FASTOPEN_KEY_LENGTH);
 314		n_keys = 1;
 315	}
 316
 317	for (i = 0; i < n_keys * 4; i++)
 318		user_key[i] = le32_to_cpu(key[i]);
 319
 320	for (i = 0; i < n_keys; i++) {
 321		off += snprintf(tbl.data + off, tbl.maxlen - off,
 322				"%08x-%08x-%08x-%08x",
 323				user_key[i * 4],
 324				user_key[i * 4 + 1],
 325				user_key[i * 4 + 2],
 326				user_key[i * 4 + 3]);
 327
 328		if (WARN_ON_ONCE(off >= tbl.maxlen - 1))
 329			break;
 330
 331		if (i + 1 < n_keys)
 332			off += snprintf(tbl.data + off, tbl.maxlen - off, ",");
 333	}
 334
 335	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
 336
 337	if (write && ret == 0) {
 338		backup_data = strchr(tbl.data, ',');
 339		if (backup_data) {
 340			*backup_data = '\0';
 341			backup_data++;
 342		}
 343		if (sscanf_key(tbl.data, key)) {
 344			ret = -EINVAL;
 345			goto bad_key;
 346		}
 347		if (backup_data) {
 348			if (sscanf_key(backup_data, key + 4)) {
 349				ret = -EINVAL;
 350				goto bad_key;
 351			}
 352		}
 353		tcp_fastopen_reset_cipher(net, NULL, key,
 354					  backup_data ? key + 4 : NULL);
 355	}
 356
 357bad_key:
 358	kfree(tbl.data);
 359	return ret;
 360}
 361
 362static int proc_tfo_blackhole_detect_timeout(const struct ctl_table *table,
 363					     int write, void *buffer,
 364					     size_t *lenp, loff_t *ppos)
 365{
 366	struct net *net = container_of(table->data, struct net,
 367	    ipv4.sysctl_tcp_fastopen_blackhole_timeout);
 368	int ret;
 369
 370	ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
 371	if (write && ret == 0)
 372		atomic_set(&net->ipv4.tfo_active_disable_times, 0);
 373
 374	return ret;
 375}
 376
 377static int proc_tcp_available_ulp(const struct ctl_table *ctl,
 378				  int write, void *buffer, size_t *lenp,
 379				  loff_t *ppos)
 380{
 381	struct ctl_table tbl = { .maxlen = TCP_ULP_BUF_MAX, };
 382	int ret;
 383
 384	tbl.data = kmalloc(tbl.maxlen, GFP_USER);
 385	if (!tbl.data)
 386		return -ENOMEM;
 387	tcp_get_available_ulp(tbl.data, TCP_ULP_BUF_MAX);
 388	ret = proc_dostring(&tbl, write, buffer, lenp, ppos);
 389	kfree(tbl.data);
 390
 391	return ret;
 392}
 393
 394static int proc_tcp_ehash_entries(const struct ctl_table *table, int write,
 395				  void *buffer, size_t *lenp, loff_t *ppos)
 396{
 397	struct net *net = container_of(table->data, struct net,
 398				       ipv4.sysctl_tcp_child_ehash_entries);
 399	struct inet_hashinfo *hinfo = net->ipv4.tcp_death_row.hashinfo;
 400	int tcp_ehash_entries;
 401	struct ctl_table tbl;
 402
 403	tcp_ehash_entries = hinfo->ehash_mask + 1;
 404
 405	/* A negative number indicates that the child netns
 406	 * shares the global ehash.
 407	 */
 408	if (!net_eq(net, &init_net) && !hinfo->pernet)
 409		tcp_ehash_entries *= -1;
 410
 411	memset(&tbl, 0, sizeof(tbl));
 412	tbl.data = &tcp_ehash_entries;
 413	tbl.maxlen = sizeof(int);
 414
 415	return proc_dointvec(&tbl, write, buffer, lenp, ppos);
 416}
 417
 418static int proc_udp_hash_entries(const struct ctl_table *table, int write,
 419				 void *buffer, size_t *lenp, loff_t *ppos)
 420{
 421	struct net *net = container_of(table->data, struct net,
 422				       ipv4.sysctl_udp_child_hash_entries);
 423	int udp_hash_entries;
 424	struct ctl_table tbl;
 425
 426	udp_hash_entries = net->ipv4.udp_table->mask + 1;
 427
 428	/* A negative number indicates that the child netns
 429	 * shares the global udp_table.
 430	 */
 431	if (!net_eq(net, &init_net) && net->ipv4.udp_table == &udp_table)
 432		udp_hash_entries *= -1;
 433
 434	memset(&tbl, 0, sizeof(tbl));
 435	tbl.data = &udp_hash_entries;
 436	tbl.maxlen = sizeof(int);
 437
 438	return proc_dointvec(&tbl, write, buffer, lenp, ppos);
 439}
 440
 441#ifdef CONFIG_IP_ROUTE_MULTIPATH
 442static int proc_fib_multipath_hash_policy(const struct ctl_table *table, int write,
 443					  void *buffer, size_t *lenp,
 444					  loff_t *ppos)
 445{
 446	struct net *net = container_of(table->data, struct net,
 447	    ipv4.sysctl_fib_multipath_hash_policy);
 448	int ret;
 449
 450	ret = proc_dou8vec_minmax(table, write, buffer, lenp, ppos);
 451	if (write && ret == 0)
 452		call_netevent_notifiers(NETEVENT_IPV4_MPATH_HASH_UPDATE, net);
 453
 454	return ret;
 455}
 456
 457static int proc_fib_multipath_hash_fields(const struct ctl_table *table, int write,
 458					  void *buffer, size_t *lenp,
 459					  loff_t *ppos)
 460{
 461	struct net *net;
 462	int ret;
 463
 464	net = container_of(table->data, struct net,
 465			   ipv4.sysctl_fib_multipath_hash_fields);
 466	ret = proc_douintvec_minmax(table, write, buffer, lenp, ppos);
 467	if (write && ret == 0)
 468		call_netevent_notifiers(NETEVENT_IPV4_MPATH_HASH_UPDATE, net);
 469
 470	return ret;
 471}
 472
 473static u32 proc_fib_multipath_hash_rand_seed __ro_after_init;
 474
 475static void proc_fib_multipath_hash_init_rand_seed(void)
 476{
 477	get_random_bytes(&proc_fib_multipath_hash_rand_seed,
 478			 sizeof(proc_fib_multipath_hash_rand_seed));
 479}
 480
 481static void proc_fib_multipath_hash_set_seed(struct net *net, u32 user_seed)
 482{
 483	struct sysctl_fib_multipath_hash_seed new = {
 484		.user_seed = user_seed,
 485		.mp_seed = (user_seed ? user_seed :
 486			    proc_fib_multipath_hash_rand_seed),
 487	};
 488
 489	WRITE_ONCE(net->ipv4.sysctl_fib_multipath_hash_seed.user_seed, new.user_seed);
 490	WRITE_ONCE(net->ipv4.sysctl_fib_multipath_hash_seed.mp_seed, new.mp_seed);
 491}
 492
 493static int proc_fib_multipath_hash_seed(const struct ctl_table *table, int write,
 494					void *buffer, size_t *lenp,
 495					loff_t *ppos)
 496{
 497	struct sysctl_fib_multipath_hash_seed *mphs;
 498	struct net *net = table->data;
 499	struct ctl_table tmp;
 500	u32 user_seed;
 501	int ret;
 502
 503	mphs = &net->ipv4.sysctl_fib_multipath_hash_seed;
 504	user_seed = READ_ONCE(mphs->user_seed);
 505
 506	tmp = *table;
 507	tmp.data = &user_seed;
 508
 509	ret = proc_douintvec_minmax(&tmp, write, buffer, lenp, ppos);
 510
 511	if (write && ret == 0) {
 512		proc_fib_multipath_hash_set_seed(net, user_seed);
 513		call_netevent_notifiers(NETEVENT_IPV4_MPATH_HASH_UPDATE, net);
 514	}
 515
 516	return ret;
 517}
 518#else
 519
 520static void proc_fib_multipath_hash_init_rand_seed(void)
 521{
 522}
 523
 524static void proc_fib_multipath_hash_set_seed(struct net *net, u32 user_seed)
 525{
 526}
 527
 528#endif
 529
 530static struct ctl_table ipv4_table[] = {
 531	{
 532		.procname	= "tcp_max_orphans",
 533		.data		= &sysctl_tcp_max_orphans,
 534		.maxlen		= sizeof(int),
 535		.mode		= 0644,
 536		.proc_handler	= proc_dointvec
 537	},
 538	{
 539		.procname	= "inet_peer_threshold",
 540		.data		= &inet_peer_threshold,
 541		.maxlen		= sizeof(int),
 542		.mode		= 0644,
 543		.proc_handler	= proc_dointvec
 544	},
 545	{
 546		.procname	= "inet_peer_minttl",
 547		.data		= &inet_peer_minttl,
 548		.maxlen		= sizeof(int),
 549		.mode		= 0644,
 550		.proc_handler	= proc_dointvec_jiffies,
 551	},
 552	{
 553		.procname	= "inet_peer_maxttl",
 554		.data		= &inet_peer_maxttl,
 555		.maxlen		= sizeof(int),
 556		.mode		= 0644,
 557		.proc_handler	= proc_dointvec_jiffies,
 558	},
 559	{
 560		.procname	= "tcp_mem",
 561		.maxlen		= sizeof(sysctl_tcp_mem),
 562		.data		= &sysctl_tcp_mem,
 563		.mode		= 0644,
 564		.proc_handler	= proc_doulongvec_minmax,
 565	},
 566	{
 567		.procname	= "tcp_low_latency",
 568		.data		= &sysctl_tcp_low_latency,
 569		.maxlen		= sizeof(int),
 570		.mode		= 0644,
 571		.proc_handler	= proc_dointvec
 572	},
 573#ifdef CONFIG_NETLABEL
 574	{
 575		.procname	= "cipso_cache_enable",
 576		.data		= &cipso_v4_cache_enabled,
 577		.maxlen		= sizeof(int),
 578		.mode		= 0644,
 579		.proc_handler	= proc_dointvec,
 580	},
 581	{
 582		.procname	= "cipso_cache_bucket_size",
 583		.data		= &cipso_v4_cache_bucketsize,
 584		.maxlen		= sizeof(int),
 585		.mode		= 0644,
 586		.proc_handler	= proc_dointvec,
 587	},
 588	{
 589		.procname	= "cipso_rbm_optfmt",
 590		.data		= &cipso_v4_rbm_optfmt,
 591		.maxlen		= sizeof(int),
 592		.mode		= 0644,
 593		.proc_handler	= proc_dointvec,
 594	},
 595	{
 596		.procname	= "cipso_rbm_strictvalid",
 597		.data		= &cipso_v4_rbm_strictvalid,
 598		.maxlen		= sizeof(int),
 599		.mode		= 0644,
 600		.proc_handler	= proc_dointvec,
 601	},
 602#endif /* CONFIG_NETLABEL */
 603	{
 604		.procname	= "tcp_available_ulp",
 605		.maxlen		= TCP_ULP_BUF_MAX,
 606		.mode		= 0444,
 607		.proc_handler   = proc_tcp_available_ulp,
 608	},
 609	{
 610		.procname	= "udp_mem",
 611		.data		= &sysctl_udp_mem,
 612		.maxlen		= sizeof(sysctl_udp_mem),
 613		.mode		= 0644,
 614		.proc_handler	= proc_doulongvec_minmax,
 615	},
 616	{
 617		.procname	= "fib_sync_mem",
 618		.data		= &sysctl_fib_sync_mem,
 619		.maxlen		= sizeof(sysctl_fib_sync_mem),
 620		.mode		= 0644,
 621		.proc_handler	= proc_douintvec_minmax,
 622		.extra1		= &sysctl_fib_sync_mem_min,
 623		.extra2		= &sysctl_fib_sync_mem_max,
 624	},
 625};
 626
 627static struct ctl_table ipv4_net_table[] = {
 628	{
 629		.procname	= "tcp_max_tw_buckets",
 630		.data		= &init_net.ipv4.tcp_death_row.sysctl_max_tw_buckets,
 631		.maxlen		= sizeof(int),
 632		.mode		= 0644,
 633		.proc_handler	= proc_dointvec
 634	},
 635	{
 636		.procname	= "icmp_echo_ignore_all",
 637		.data		= &init_net.ipv4.sysctl_icmp_echo_ignore_all,
 638		.maxlen		= sizeof(u8),
 639		.mode		= 0644,
 640		.proc_handler	= proc_dou8vec_minmax,
 641		.extra1		= SYSCTL_ZERO,
 642		.extra2		= SYSCTL_ONE
 643	},
 644	{
 645		.procname	= "icmp_echo_enable_probe",
 646		.data		= &init_net.ipv4.sysctl_icmp_echo_enable_probe,
 647		.maxlen		= sizeof(u8),
 648		.mode		= 0644,
 649		.proc_handler	= proc_dou8vec_minmax,
 650		.extra1		= SYSCTL_ZERO,
 651		.extra2		= SYSCTL_ONE
 652	},
 653	{
 654		.procname	= "icmp_echo_ignore_broadcasts",
 655		.data		= &init_net.ipv4.sysctl_icmp_echo_ignore_broadcasts,
 656		.maxlen		= sizeof(u8),
 657		.mode		= 0644,
 658		.proc_handler	= proc_dou8vec_minmax,
 659		.extra1		= SYSCTL_ZERO,
 660		.extra2		= SYSCTL_ONE
 661	},
 662	{
 663		.procname	= "icmp_ignore_bogus_error_responses",
 664		.data		= &init_net.ipv4.sysctl_icmp_ignore_bogus_error_responses,
 665		.maxlen		= sizeof(u8),
 666		.mode		= 0644,
 667		.proc_handler	= proc_dou8vec_minmax,
 668		.extra1		= SYSCTL_ZERO,
 669		.extra2		= SYSCTL_ONE
 670	},
 671	{
 672		.procname	= "icmp_errors_use_inbound_ifaddr",
 673		.data		= &init_net.ipv4.sysctl_icmp_errors_use_inbound_ifaddr,
 674		.maxlen		= sizeof(u8),
 675		.mode		= 0644,
 676		.proc_handler	= proc_dou8vec_minmax,
 677		.extra1		= SYSCTL_ZERO,
 678		.extra2		= SYSCTL_ONE
 679	},
 680	{
 681		.procname	= "icmp_errors_extension_mask",
 682		.data		= &init_net.ipv4.sysctl_icmp_errors_extension_mask,
 683		.maxlen		= sizeof(u8),
 684		.mode		= 0644,
 685		.proc_handler	= proc_dou8vec_minmax,
 686		.extra1		= SYSCTL_ZERO,
 687		.extra2		= &icmp_errors_extension_mask_all,
 688	},
 689	{
 690		.procname	= "icmp_ratelimit",
 691		.data		= &init_net.ipv4.sysctl_icmp_ratelimit,
 692		.maxlen		= sizeof(int),
 693		.mode		= 0644,
 694		.proc_handler	= proc_dointvec_ms_jiffies,
 695	},
 696	{
 697		.procname	= "icmp_ratemask",
 698		.data		= &init_net.ipv4.sysctl_icmp_ratemask,
 699		.maxlen		= sizeof(int),
 700		.mode		= 0644,
 701		.proc_handler	= proc_dointvec
 702	},
 703	{
 704		.procname	= "icmp_msgs_per_sec",
 705		.data		= &init_net.ipv4.sysctl_icmp_msgs_per_sec,
 706		.maxlen		= sizeof(int),
 707		.mode		= 0644,
 708		.proc_handler	= proc_dointvec_minmax,
 709		.extra1		= SYSCTL_ZERO,
 710	},
 711	{
 712		.procname	= "icmp_msgs_burst",
 713		.data		= &init_net.ipv4.sysctl_icmp_msgs_burst,
 714		.maxlen		= sizeof(int),
 715		.mode		= 0644,
 716		.proc_handler	= proc_dointvec_minmax,
 717		.extra1		= SYSCTL_ZERO,
 718	},
 719	{
 720		.procname	= "ping_group_range",
 721		.data		= &init_net.ipv4.ping_group_range.range,
 722		.maxlen		= sizeof(gid_t)*2,
 723		.mode		= 0644,
 724		.proc_handler	= ipv4_ping_group_range,
 725	},
 726#ifdef CONFIG_NET_L3_MASTER_DEV
 727	{
 728		.procname	= "raw_l3mdev_accept",
 729		.data		= &init_net.ipv4.sysctl_raw_l3mdev_accept,
 730		.maxlen		= sizeof(u8),
 731		.mode		= 0644,
 732		.proc_handler	= proc_dou8vec_minmax,
 733		.extra1		= SYSCTL_ZERO,
 734		.extra2		= SYSCTL_ONE,
 735	},
 736#endif
 737	{
 738		.procname	= "tcp_ecn",
 739		.data		= &init_net.ipv4.sysctl_tcp_ecn,
 740		.maxlen		= sizeof(u8),
 741		.mode		= 0644,
 742		.proc_handler	= proc_dou8vec_minmax,
 743		.extra1		= SYSCTL_ZERO,
 744		.extra2		= &tcp_ecn_mode_max,
 745	},
 746	{
 747		.procname	= "tcp_ecn_option",
 748		.data		= &init_net.ipv4.sysctl_tcp_ecn_option,
 749		.maxlen		= sizeof(u8),
 750		.mode		= 0644,
 751		.proc_handler	= proc_dou8vec_minmax,
 752		.extra1		= SYSCTL_ZERO,
 753		.extra2		= SYSCTL_TWO,
 754	},
 755	{
 756		.procname	= "tcp_ecn_option_beacon",
 757		.data		= &init_net.ipv4.sysctl_tcp_ecn_option_beacon,
 758		.maxlen		= sizeof(u8),
 759		.mode		= 0644,
 760		.proc_handler	= proc_dou8vec_minmax,
 761		.extra1		= SYSCTL_ZERO,
 762		.extra2		= SYSCTL_THREE,
 763	},
 764	{
 765		.procname	= "tcp_ecn_fallback",
 766		.data		= &init_net.ipv4.sysctl_tcp_ecn_fallback,
 767		.maxlen		= sizeof(u8),
 768		.mode		= 0644,
 769		.proc_handler	= proc_dou8vec_minmax,
 770		.extra1		= SYSCTL_ZERO,
 771		.extra2		= SYSCTL_ONE,
 772	},
 773	{
 774		.procname	= "ip_dynaddr",
 775		.data		= &init_net.ipv4.sysctl_ip_dynaddr,
 776		.maxlen		= sizeof(u8),
 777		.mode		= 0644,
 778		.proc_handler	= proc_dou8vec_minmax,
 779	},
 780	{
 781		.procname	= "ip_early_demux",
 782		.data		= &init_net.ipv4.sysctl_ip_early_demux,
 783		.maxlen		= sizeof(u8),
 784		.mode		= 0644,
 785		.proc_handler	= proc_dou8vec_minmax,
 786	},
 787	{
 788		.procname       = "udp_early_demux",
 789		.data           = &init_net.ipv4.sysctl_udp_early_demux,
 790		.maxlen         = sizeof(u8),
 791		.mode           = 0644,
 792		.proc_handler   = proc_dou8vec_minmax,
 793	},
 794	{
 795		.procname       = "tcp_early_demux",
 796		.data           = &init_net.ipv4.sysctl_tcp_early_demux,
 797		.maxlen         = sizeof(u8),
 798		.mode           = 0644,
 799		.proc_handler   = proc_dou8vec_minmax,
 800	},
 801	{
 802		.procname       = "nexthop_compat_mode",
 803		.data           = &init_net.ipv4.sysctl_nexthop_compat_mode,
 804		.maxlen         = sizeof(u8),
 805		.mode           = 0644,
 806		.proc_handler   = proc_dou8vec_minmax,
 807		.extra1		= SYSCTL_ZERO,
 808		.extra2		= SYSCTL_ONE,
 809	},
 810	{
 811		.procname	= "ip_default_ttl",
 812		.data		= &init_net.ipv4.sysctl_ip_default_ttl,
 813		.maxlen		= sizeof(u8),
 814		.mode		= 0644,
 815		.proc_handler	= proc_dou8vec_minmax,
 816		.extra1		= &ip_ttl_min,
 817		.extra2		= &ip_ttl_max,
 818	},
 819	{
 820		.procname	= "ip_local_port_range",
 821		.maxlen		= 0,
 822		.data		= &init_net,
 823		.mode		= 0644,
 824		.proc_handler	= ipv4_local_port_range,
 825	},
 826	{
 827		.procname	= "ip_local_reserved_ports",
 828		.data		= &init_net.ipv4.sysctl_local_reserved_ports,
 829		.maxlen		= 65536,
 830		.mode		= 0644,
 831		.proc_handler	= proc_do_large_bitmap,
 832	},
 833	{
 834		.procname	= "ip_no_pmtu_disc",
 835		.data		= &init_net.ipv4.sysctl_ip_no_pmtu_disc,
 836		.maxlen		= sizeof(u8),
 837		.mode		= 0644,
 838		.proc_handler	= proc_dou8vec_minmax,
 839	},
 840	{
 841		.procname	= "ip_forward_use_pmtu",
 842		.data		= &init_net.ipv4.sysctl_ip_fwd_use_pmtu,
 843		.maxlen		= sizeof(u8),
 844		.mode		= 0644,
 845		.proc_handler	= proc_dou8vec_minmax,
 846	},
 847	{
 848		.procname	= "ip_forward_update_priority",
 849		.data		= &init_net.ipv4.sysctl_ip_fwd_update_priority,
 850		.maxlen		= sizeof(u8),
 851		.mode		= 0644,
 852		.proc_handler   = ipv4_fwd_update_priority,
 853		.extra1		= SYSCTL_ZERO,
 854		.extra2		= SYSCTL_ONE,
 855	},
 856	{
 857		.procname	= "ip_nonlocal_bind",
 858		.data		= &init_net.ipv4.sysctl_ip_nonlocal_bind,
 859		.maxlen		= sizeof(u8),
 860		.mode		= 0644,
 861		.proc_handler	= proc_dou8vec_minmax,
 862	},
 863	{
 864		.procname	= "ip_autobind_reuse",
 865		.data		= &init_net.ipv4.sysctl_ip_autobind_reuse,
 866		.maxlen		= sizeof(u8),
 867		.mode		= 0644,
 868		.proc_handler	= proc_dou8vec_minmax,
 869		.extra1         = SYSCTL_ZERO,
 870		.extra2         = SYSCTL_ONE,
 871	},
 872	{
 873		.procname	= "fwmark_reflect",
 874		.data		= &init_net.ipv4.sysctl_fwmark_reflect,
 875		.maxlen		= sizeof(u8),
 876		.mode		= 0644,
 877		.proc_handler	= proc_dou8vec_minmax,
 878	},
 879	{
 880		.procname	= "tcp_fwmark_accept",
 881		.data		= &init_net.ipv4.sysctl_tcp_fwmark_accept,
 882		.maxlen		= sizeof(u8),
 883		.mode		= 0644,
 884		.proc_handler	= proc_dou8vec_minmax,
 885	},
 886#ifdef CONFIG_NET_L3_MASTER_DEV
 887	{
 888		.procname	= "tcp_l3mdev_accept",
 889		.data		= &init_net.ipv4.sysctl_tcp_l3mdev_accept,
 890		.maxlen		= sizeof(u8),
 891		.mode		= 0644,
 892		.proc_handler	= proc_dou8vec_minmax,
 893		.extra1		= SYSCTL_ZERO,
 894		.extra2		= SYSCTL_ONE,
 895	},
 896#endif
 897	{
 898		.procname	= "tcp_mtu_probing",
 899		.data		= &init_net.ipv4.sysctl_tcp_mtu_probing,
 900		.maxlen		= sizeof(u8),
 901		.mode		= 0644,
 902		.proc_handler	= proc_dou8vec_minmax,
 903	},
 904	{
 905		.procname	= "tcp_base_mss",
 906		.data		= &init_net.ipv4.sysctl_tcp_base_mss,
 907		.maxlen		= sizeof(int),
 908		.mode		= 0644,
 909		.proc_handler	= proc_dointvec,
 910	},
 911	{
 912		.procname	= "tcp_min_snd_mss",
 913		.data		= &init_net.ipv4.sysctl_tcp_min_snd_mss,
 914		.maxlen		= sizeof(int),
 915		.mode		= 0644,
 916		.proc_handler	= proc_dointvec_minmax,
 917		.extra1		= &tcp_min_snd_mss_min,
 918		.extra2		= &tcp_min_snd_mss_max,
 919	},
 920	{
 921		.procname	= "tcp_mtu_probe_floor",
 922		.data		= &init_net.ipv4.sysctl_tcp_mtu_probe_floor,
 923		.maxlen		= sizeof(int),
 924		.mode		= 0644,
 925		.proc_handler	= proc_dointvec_minmax,
 926		.extra1		= &tcp_min_snd_mss_min,
 927		.extra2		= &tcp_min_snd_mss_max,
 928	},
 929	{
 930		.procname	= "tcp_probe_threshold",
 931		.data		= &init_net.ipv4.sysctl_tcp_probe_threshold,
 932		.maxlen		= sizeof(int),
 933		.mode		= 0644,
 934		.proc_handler	= proc_dointvec,
 935	},
 936	{
 937		.procname	= "tcp_probe_interval",
 938		.data		= &init_net.ipv4.sysctl_tcp_probe_interval,
 939		.maxlen		= sizeof(u32),
 940		.mode		= 0644,
 941		.proc_handler	= proc_douintvec_minmax,
 942		.extra2		= &u32_max_div_HZ,
 943	},
 944	{
 945		.procname	= "igmp_link_local_mcast_reports",
 946		.data		= &init_net.ipv4.sysctl_igmp_llm_reports,
 947		.maxlen		= sizeof(u8),
 948		.mode		= 0644,
 949		.proc_handler	= proc_dou8vec_minmax,
 950	},
 951	{
 952		.procname	= "igmp_max_memberships",
 953		.data		= &init_net.ipv4.sysctl_igmp_max_memberships,
 954		.maxlen		= sizeof(int),
 955		.mode		= 0644,
 956		.proc_handler	= proc_dointvec
 957	},
 958	{
 959		.procname	= "igmp_max_msf",
 960		.data		= &init_net.ipv4.sysctl_igmp_max_msf,
 961		.maxlen		= sizeof(int),
 962		.mode		= 0644,
 963		.proc_handler	= proc_dointvec
 964	},
 965#ifdef CONFIG_IP_MULTICAST
 966	{
 967		.procname	= "igmp_qrv",
 968		.data		= &init_net.ipv4.sysctl_igmp_qrv,
 969		.maxlen		= sizeof(int),
 970		.mode		= 0644,
 971		.proc_handler	= proc_dointvec_minmax,
 972		.extra1		= SYSCTL_ONE
 973	},
 974#endif
 975	{
 976		.procname	= "tcp_congestion_control",
 977		.data		= &init_net.ipv4.tcp_congestion_control,
 978		.mode		= 0644,
 979		.maxlen		= TCP_CA_NAME_MAX,
 980		.proc_handler	= proc_tcp_congestion_control,
 981	},
 982	{
 983		.procname	= "tcp_available_congestion_control",
 984		.maxlen		= TCP_CA_BUF_MAX,
 985		.mode		= 0444,
 986		.proc_handler   = proc_tcp_available_congestion_control,
 987	},
 988	{
 989		.procname	= "tcp_allowed_congestion_control",
 990		.maxlen		= TCP_CA_BUF_MAX,
 991		.mode		= 0644,
 992		.proc_handler   = proc_allowed_congestion_control,
 993	},
 994	{
 995		.procname	= "tcp_keepalive_time",
 996		.data		= &init_net.ipv4.sysctl_tcp_keepalive_time,
 997		.maxlen		= sizeof(int),
 998		.mode		= 0644,
 999		.proc_handler	= proc_dointvec_jiffies,
1000	},
1001	{
1002		.procname	= "tcp_keepalive_probes",
1003		.data		= &init_net.ipv4.sysctl_tcp_keepalive_probes,
1004		.maxlen		= sizeof(u8),
1005		.mode		= 0644,
1006		.proc_handler	= proc_dou8vec_minmax,
1007	},
1008	{
1009		.procname	= "tcp_keepalive_intvl",
1010		.data		= &init_net.ipv4.sysctl_tcp_keepalive_intvl,
1011		.maxlen		= sizeof(int),
1012		.mode		= 0644,
1013		.proc_handler	= proc_dointvec_jiffies,
1014	},
1015	{
1016		.procname	= "tcp_syn_retries",
1017		.data		= &init_net.ipv4.sysctl_tcp_syn_retries,
1018		.maxlen		= sizeof(u8),
1019		.mode		= 0644,
1020		.proc_handler	= proc_dou8vec_minmax,
1021		.extra1		= &tcp_syn_retries_min,
1022		.extra2		= &tcp_syn_retries_max
1023	},
1024	{
1025		.procname	= "tcp_synack_retries",
1026		.data		= &init_net.ipv4.sysctl_tcp_synack_retries,
1027		.maxlen		= sizeof(u8),
1028		.mode		= 0644,
1029		.proc_handler	= proc_dou8vec_minmax,
1030	},
1031#ifdef CONFIG_SYN_COOKIES
1032	{
1033		.procname	= "tcp_syncookies",
1034		.data		= &init_net.ipv4.sysctl_tcp_syncookies,
1035		.maxlen		= sizeof(u8),
1036		.mode		= 0644,
1037		.proc_handler	= proc_dou8vec_minmax,
1038	},
1039#endif
1040	{
1041		.procname	= "tcp_migrate_req",
1042		.data		= &init_net.ipv4.sysctl_tcp_migrate_req,
1043		.maxlen		= sizeof(u8),
1044		.mode		= 0644,
1045		.proc_handler	= proc_dou8vec_minmax,
1046		.extra1		= SYSCTL_ZERO,
1047		.extra2		= SYSCTL_ONE
1048	},
1049	{
1050		.procname	= "tcp_reordering",
1051		.data		= &init_net.ipv4.sysctl_tcp_reordering,
1052		.maxlen		= sizeof(int),
1053		.mode		= 0644,
1054		.proc_handler	= proc_dointvec
1055	},
1056	{
1057		.procname	= "tcp_retries1",
1058		.data		= &init_net.ipv4.sysctl_tcp_retries1,
1059		.maxlen		= sizeof(u8),
1060		.mode		= 0644,
1061		.proc_handler	= proc_dou8vec_minmax,
1062		.extra2		= &tcp_retr1_max
1063	},
1064	{
1065		.procname	= "tcp_retries2",
1066		.data		= &init_net.ipv4.sysctl_tcp_retries2,
1067		.maxlen		= sizeof(u8),
1068		.mode		= 0644,
1069		.proc_handler	= proc_dou8vec_minmax,
1070	},
1071	{
1072		.procname	= "tcp_orphan_retries",
1073		.data		= &init_net.ipv4.sysctl_tcp_orphan_retries,
1074		.maxlen		= sizeof(u8),
1075		.mode		= 0644,
1076		.proc_handler	= proc_dou8vec_minmax,
1077	},
1078	{
1079		.procname	= "tcp_fin_timeout",
1080		.data		= &init_net.ipv4.sysctl_tcp_fin_timeout,
1081		.maxlen		= sizeof(int),
1082		.mode		= 0644,
1083		.proc_handler	= proc_dointvec_jiffies,
1084	},
1085	{
1086		.procname	= "tcp_notsent_lowat",
1087		.data		= &init_net.ipv4.sysctl_tcp_notsent_lowat,
1088		.maxlen		= sizeof(unsigned int),
1089		.mode		= 0644,
1090		.proc_handler	= proc_douintvec,
1091	},
1092	{
1093		.procname	= "tcp_tw_reuse",
1094		.data		= &init_net.ipv4.sysctl_tcp_tw_reuse,
1095		.maxlen		= sizeof(u8),
1096		.mode		= 0644,
1097		.proc_handler	= proc_dou8vec_minmax,
1098		.extra1		= SYSCTL_ZERO,
1099		.extra2		= SYSCTL_TWO,
1100	},
1101	{
1102		.procname	= "tcp_tw_reuse_delay",
1103		.data		= &init_net.ipv4.sysctl_tcp_tw_reuse_delay,
1104		.maxlen		= sizeof(unsigned int),
1105		.mode		= 0644,
1106		.proc_handler	= proc_douintvec_minmax,
1107		.extra1		= SYSCTL_ONE,
1108		.extra2		= &tcp_tw_reuse_delay_max,
1109	},
1110	{
1111		.procname	= "tcp_max_syn_backlog",
1112		.data		= &init_net.ipv4.sysctl_max_syn_backlog,
1113		.maxlen		= sizeof(int),
1114		.mode		= 0644,
1115		.proc_handler	= proc_dointvec
1116	},
1117	{
1118		.procname	= "tcp_fastopen",
1119		.data		= &init_net.ipv4.sysctl_tcp_fastopen,
1120		.maxlen		= sizeof(int),
1121		.mode		= 0644,
1122		.proc_handler	= proc_dointvec,
1123	},
1124	{
1125		.procname	= "tcp_fastopen_key",
1126		.mode		= 0600,
1127		.data		= &init_net.ipv4.sysctl_tcp_fastopen,
1128		/* maxlen to print the list of keys in hex (*2), with dashes
1129		 * separating doublewords and a comma in between keys.
1130		 */
1131		.maxlen		= ((TCP_FASTOPEN_KEY_LENGTH *
1132				   2 * TCP_FASTOPEN_KEY_MAX) +
1133				   (TCP_FASTOPEN_KEY_MAX * 5)),
1134		.proc_handler	= proc_tcp_fastopen_key,
1135	},
1136	{
1137		.procname	= "tcp_fastopen_blackhole_timeout_sec",
1138		.data		= &init_net.ipv4.sysctl_tcp_fastopen_blackhole_timeout,
1139		.maxlen		= sizeof(int),
1140		.mode		= 0644,
1141		.proc_handler	= proc_tfo_blackhole_detect_timeout,
1142		.extra1		= SYSCTL_ZERO,
1143	},
1144#ifdef CONFIG_IP_ROUTE_MULTIPATH
1145	{
1146		.procname	= "fib_multipath_use_neigh",
1147		.data		= &init_net.ipv4.sysctl_fib_multipath_use_neigh,
1148		.maxlen		= sizeof(u8),
1149		.mode		= 0644,
1150		.proc_handler	= proc_dou8vec_minmax,
1151		.extra1		= SYSCTL_ZERO,
1152		.extra2		= SYSCTL_ONE,
1153	},
1154	{
1155		.procname	= "fib_multipath_hash_policy",
1156		.data		= &init_net.ipv4.sysctl_fib_multipath_hash_policy,
1157		.maxlen		= sizeof(u8),
1158		.mode		= 0644,
1159		.proc_handler	= proc_fib_multipath_hash_policy,
1160		.extra1		= SYSCTL_ZERO,
1161		.extra2		= SYSCTL_THREE,
1162	},
1163	{
1164		.procname	= "fib_multipath_hash_fields",
1165		.data		= &init_net.ipv4.sysctl_fib_multipath_hash_fields,
1166		.maxlen		= sizeof(u32),
1167		.mode		= 0644,
1168		.proc_handler	= proc_fib_multipath_hash_fields,
1169		.extra1		= SYSCTL_ONE,
1170		.extra2		= &fib_multipath_hash_fields_all_mask,
1171	},
1172	{
1173		.procname	= "fib_multipath_hash_seed",
1174		.data		= &init_net,
1175		.maxlen		= sizeof(u32),
1176		.mode		= 0644,
1177		.proc_handler	= proc_fib_multipath_hash_seed,
1178	},
1179#endif
1180	{
1181		.procname	= "ip_unprivileged_port_start",
1182		.maxlen		= sizeof(int),
1183		.data		= &init_net.ipv4.sysctl_ip_prot_sock,
1184		.mode		= 0644,
1185		.proc_handler	= ipv4_privileged_ports,
1186	},
1187#ifdef CONFIG_NET_L3_MASTER_DEV
1188	{
1189		.procname	= "udp_l3mdev_accept",
1190		.data		= &init_net.ipv4.sysctl_udp_l3mdev_accept,
1191		.maxlen		= sizeof(u8),
1192		.mode		= 0644,
1193		.proc_handler	= proc_dou8vec_minmax,
1194		.extra1		= SYSCTL_ZERO,
1195		.extra2		= SYSCTL_ONE,
1196	},
1197#endif
1198	{
1199		.procname	= "tcp_sack",
1200		.data		= &init_net.ipv4.sysctl_tcp_sack,
1201		.maxlen		= sizeof(u8),
1202		.mode		= 0644,
1203		.proc_handler	= proc_dou8vec_minmax,
1204	},
1205	{
1206		.procname	= "tcp_window_scaling",
1207		.data		= &init_net.ipv4.sysctl_tcp_window_scaling,
1208		.maxlen		= sizeof(u8),
1209		.mode		= 0644,
1210		.proc_handler	= proc_dou8vec_minmax,
1211	},
1212	{
1213		.procname	= "tcp_timestamps",
1214		.data		= &init_net.ipv4.sysctl_tcp_timestamps,
1215		.maxlen		= sizeof(u8),
1216		.mode		= 0644,
1217		.proc_handler	= proc_dou8vec_minmax,
1218	},
1219	{
1220		.procname	= "tcp_early_retrans",
1221		.data		= &init_net.ipv4.sysctl_tcp_early_retrans,
1222		.maxlen		= sizeof(u8),
1223		.mode		= 0644,
1224		.proc_handler	= proc_dou8vec_minmax,
1225		.extra1		= SYSCTL_ZERO,
1226		.extra2		= SYSCTL_FOUR,
1227	},
1228	{
1229		.procname	= "tcp_recovery",
1230		.data		= &init_net.ipv4.sysctl_tcp_recovery,
1231		.maxlen		= sizeof(u8),
1232		.mode		= 0644,
1233		.proc_handler	= proc_dou8vec_minmax,
1234	},
1235	{
1236		.procname       = "tcp_thin_linear_timeouts",
1237		.data           = &init_net.ipv4.sysctl_tcp_thin_linear_timeouts,
1238		.maxlen         = sizeof(u8),
1239		.mode           = 0644,
1240		.proc_handler   = proc_dou8vec_minmax,
1241	},
1242	{
1243		.procname	= "tcp_slow_start_after_idle",
1244		.data		= &init_net.ipv4.sysctl_tcp_slow_start_after_idle,
1245		.maxlen		= sizeof(u8),
1246		.mode		= 0644,
1247		.proc_handler	= proc_dou8vec_minmax,
1248	},
1249	{
1250		.procname	= "tcp_retrans_collapse",
1251		.data		= &init_net.ipv4.sysctl_tcp_retrans_collapse,
1252		.maxlen		= sizeof(u8),
1253		.mode		= 0644,
1254		.proc_handler	= proc_dou8vec_minmax,
1255	},
1256	{
1257		.procname	= "tcp_stdurg",
1258		.data		= &init_net.ipv4.sysctl_tcp_stdurg,
1259		.maxlen		= sizeof(u8),
1260		.mode		= 0644,
1261		.proc_handler	= proc_dou8vec_minmax,
1262	},
1263	{
1264		.procname	= "tcp_rfc1337",
1265		.data		= &init_net.ipv4.sysctl_tcp_rfc1337,
1266		.maxlen		= sizeof(u8),
1267		.mode		= 0644,
1268		.proc_handler	= proc_dou8vec_minmax,
1269	},
1270	{
1271		.procname	= "tcp_abort_on_overflow",
1272		.data		= &init_net.ipv4.sysctl_tcp_abort_on_overflow,
1273		.maxlen		= sizeof(u8),
1274		.mode		= 0644,
1275		.proc_handler	= proc_dou8vec_minmax,
1276	},
1277	{
1278		.procname	= "tcp_fack",
1279		.data		= &init_net.ipv4.sysctl_tcp_fack,
1280		.maxlen		= sizeof(u8),
1281		.mode		= 0644,
1282		.proc_handler	= proc_dou8vec_minmax,
1283	},
1284	{
1285		.procname	= "tcp_max_reordering",
1286		.data		= &init_net.ipv4.sysctl_tcp_max_reordering,
1287		.maxlen		= sizeof(int),
1288		.mode		= 0644,
1289		.proc_handler	= proc_dointvec
1290	},
1291	{
1292		.procname	= "tcp_dsack",
1293		.data		= &init_net.ipv4.sysctl_tcp_dsack,
1294		.maxlen		= sizeof(u8),
1295		.mode		= 0644,
1296		.proc_handler	= proc_dou8vec_minmax,
1297	},
1298	{
1299		.procname	= "tcp_app_win",
1300		.data		= &init_net.ipv4.sysctl_tcp_app_win,
1301		.maxlen		= sizeof(u8),
1302		.mode		= 0644,
1303		.proc_handler	= proc_dou8vec_minmax,
1304		.extra1		= SYSCTL_ZERO,
1305		.extra2		= &tcp_app_win_max,
1306	},
1307	{
1308		.procname	= "tcp_adv_win_scale",
1309		.data		= &init_net.ipv4.sysctl_tcp_adv_win_scale,
1310		.maxlen		= sizeof(int),
1311		.mode		= 0644,
1312		.proc_handler	= proc_dointvec_minmax,
1313		.extra1		= &tcp_adv_win_scale_min,
1314		.extra2		= &tcp_adv_win_scale_max,
1315	},
1316	{
1317		.procname	= "tcp_frto",
1318		.data		= &init_net.ipv4.sysctl_tcp_frto,
1319		.maxlen		= sizeof(u8),
1320		.mode		= 0644,
1321		.proc_handler	= proc_dou8vec_minmax,
1322	},
1323	{
1324		.procname	= "tcp_no_metrics_save",
1325		.data		= &init_net.ipv4.sysctl_tcp_nometrics_save,
1326		.maxlen		= sizeof(u8),
1327		.mode		= 0644,
1328		.proc_handler	= proc_dou8vec_minmax,
1329	},
1330	{
1331		.procname	= "tcp_no_ssthresh_metrics_save",
1332		.data		= &init_net.ipv4.sysctl_tcp_no_ssthresh_metrics_save,
1333		.maxlen		= sizeof(u8),
1334		.mode		= 0644,
1335		.proc_handler	= proc_dou8vec_minmax,
1336		.extra1		= SYSCTL_ZERO,
1337		.extra2		= SYSCTL_ONE,
1338	},
1339	{
1340		.procname	= "tcp_moderate_rcvbuf",
1341		.data		= &init_net.ipv4.sysctl_tcp_moderate_rcvbuf,
1342		.maxlen		= sizeof(u8),
1343		.mode		= 0644,
1344		.proc_handler	= proc_dou8vec_minmax,
1345	},
1346	{
1347		.procname	= "tcp_rcvbuf_low_rtt",
1348		.data		= &init_net.ipv4.sysctl_tcp_rcvbuf_low_rtt,
1349		.maxlen		= sizeof(int),
1350		.mode		= 0644,
1351		.proc_handler	= proc_dointvec_minmax,
1352		.extra1		= SYSCTL_ZERO,
1353		.extra2		= SYSCTL_INT_MAX,
1354	},
1355	{
1356		.procname	= "tcp_tso_win_divisor",
1357		.data		= &init_net.ipv4.sysctl_tcp_tso_win_divisor,
1358		.maxlen		= sizeof(u8),
1359		.mode		= 0644,
1360		.proc_handler	= proc_dou8vec_minmax,
1361	},
1362	{
1363		.procname	= "tcp_workaround_signed_windows",
1364		.data		= &init_net.ipv4.sysctl_tcp_workaround_signed_windows,
1365		.maxlen		= sizeof(u8),
1366		.mode		= 0644,
1367		.proc_handler	= proc_dou8vec_minmax,
1368	},
1369	{
1370		.procname	= "tcp_limit_output_bytes",
1371		.data		= &init_net.ipv4.sysctl_tcp_limit_output_bytes,
1372		.maxlen		= sizeof(int),
1373		.mode		= 0644,
1374		.proc_handler	= proc_dointvec
1375	},
1376	{
1377		.procname	= "tcp_challenge_ack_limit",
1378		.data		= &init_net.ipv4.sysctl_tcp_challenge_ack_limit,
1379		.maxlen		= sizeof(int),
1380		.mode		= 0644,
1381		.proc_handler	= proc_dointvec
1382	},
1383	{
1384		.procname	= "tcp_min_tso_segs",
1385		.data		= &init_net.ipv4.sysctl_tcp_min_tso_segs,
1386		.maxlen		= sizeof(u8),
1387		.mode		= 0644,
1388		.proc_handler	= proc_dou8vec_minmax,
1389		.extra1		= SYSCTL_ONE,
1390	},
1391	{
1392		.procname	= "tcp_tso_rtt_log",
1393		.data		= &init_net.ipv4.sysctl_tcp_tso_rtt_log,
1394		.maxlen		= sizeof(u8),
1395		.mode		= 0644,
1396		.proc_handler	= proc_dou8vec_minmax,
1397	},
1398	{
1399		.procname	= "tcp_min_rtt_wlen",
1400		.data		= &init_net.ipv4.sysctl_tcp_min_rtt_wlen,
1401		.maxlen		= sizeof(int),
1402		.mode		= 0644,
1403		.proc_handler	= proc_dointvec_minmax,
1404		.extra1		= SYSCTL_ZERO,
1405		.extra2		= &one_day_secs
1406	},
1407	{
1408		.procname	= "tcp_autocorking",
1409		.data		= &init_net.ipv4.sysctl_tcp_autocorking,
1410		.maxlen		= sizeof(u8),
1411		.mode		= 0644,
1412		.proc_handler	= proc_dou8vec_minmax,
1413		.extra1		= SYSCTL_ZERO,
1414		.extra2		= SYSCTL_ONE,
1415	},
1416	{
1417		.procname	= "tcp_invalid_ratelimit",
1418		.data		= &init_net.ipv4.sysctl_tcp_invalid_ratelimit,
1419		.maxlen		= sizeof(int),
1420		.mode		= 0644,
1421		.proc_handler	= proc_dointvec_ms_jiffies,
1422	},
1423	{
1424		.procname	= "tcp_pacing_ss_ratio",
1425		.data		= &init_net.ipv4.sysctl_tcp_pacing_ss_ratio,
1426		.maxlen		= sizeof(int),
1427		.mode		= 0644,
1428		.proc_handler	= proc_dointvec_minmax,
1429		.extra1		= SYSCTL_ZERO,
1430		.extra2		= SYSCTL_ONE_THOUSAND,
1431	},
1432	{
1433		.procname	= "tcp_pacing_ca_ratio",
1434		.data		= &init_net.ipv4.sysctl_tcp_pacing_ca_ratio,
1435		.maxlen		= sizeof(int),
1436		.mode		= 0644,
1437		.proc_handler	= proc_dointvec_minmax,
1438		.extra1		= SYSCTL_ZERO,
1439		.extra2		= SYSCTL_ONE_THOUSAND,
1440	},
1441	{
1442		.procname	= "tcp_wmem",
1443		.data		= &init_net.ipv4.sysctl_tcp_wmem,
1444		.maxlen		= sizeof(init_net.ipv4.sysctl_tcp_wmem),
1445		.mode		= 0644,
1446		.proc_handler	= proc_dointvec_minmax,
1447		.extra1		= SYSCTL_ONE,
1448	},
1449	{
1450		.procname	= "tcp_rmem",
1451		.data		= &init_net.ipv4.sysctl_tcp_rmem,
1452		.maxlen		= sizeof(init_net.ipv4.sysctl_tcp_rmem),
1453		.mode		= 0644,
1454		.proc_handler	= proc_dointvec_minmax,
1455		.extra1		= SYSCTL_ONE,
1456	},
1457	{
1458		.procname	= "tcp_comp_sack_delay_ns",
1459		.data		= &init_net.ipv4.sysctl_tcp_comp_sack_delay_ns,
1460		.maxlen		= sizeof(unsigned long),
1461		.mode		= 0644,
1462		.proc_handler	= proc_doulongvec_minmax,
1463	},
1464	{
1465		.procname	= "tcp_comp_sack_rtt_percent",
1466		.data		= &init_net.ipv4.sysctl_tcp_comp_sack_rtt_percent,
1467		.maxlen		= sizeof(int),
1468		.mode		= 0644,
1469		.proc_handler	= proc_dointvec_minmax,
1470		.extra1		= SYSCTL_ONE,
1471		.extra2		= SYSCTL_ONE_THOUSAND,
1472	},
1473	{
1474		.procname	= "tcp_comp_sack_slack_ns",
1475		.data		= &init_net.ipv4.sysctl_tcp_comp_sack_slack_ns,
1476		.maxlen		= sizeof(unsigned long),
1477		.mode		= 0644,
1478		.proc_handler	= proc_doulongvec_minmax,
1479	},
1480	{
1481		.procname	= "tcp_comp_sack_nr",
1482		.data		= &init_net.ipv4.sysctl_tcp_comp_sack_nr,
1483		.maxlen		= sizeof(u8),
1484		.mode		= 0644,
1485		.proc_handler	= proc_dou8vec_minmax,
1486		.extra1		= SYSCTL_ZERO,
1487	},
1488	{
1489		.procname	= "tcp_backlog_ack_defer",
1490		.data		= &init_net.ipv4.sysctl_tcp_backlog_ack_defer,
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_reflect_tos",
1499		.data           = &init_net.ipv4.sysctl_tcp_reflect_tos,
1500		.maxlen         = sizeof(u8),
1501		.mode           = 0644,
1502		.proc_handler   = proc_dou8vec_minmax,
1503		.extra1         = SYSCTL_ZERO,
1504		.extra2         = SYSCTL_ONE,
1505	},
1506	{
1507		.procname	= "tcp_ehash_entries",
1508		.data		= &init_net.ipv4.sysctl_tcp_child_ehash_entries,
1509		.mode		= 0444,
1510		.proc_handler	= proc_tcp_ehash_entries,
1511	},
1512	{
1513		.procname	= "tcp_child_ehash_entries",
1514		.data		= &init_net.ipv4.sysctl_tcp_child_ehash_entries,
1515		.maxlen		= sizeof(unsigned int),
1516		.mode		= 0644,
1517		.proc_handler	= proc_douintvec_minmax,
1518		.extra1		= SYSCTL_ZERO,
1519		.extra2		= &tcp_child_ehash_entries_max,
1520	},
1521	{
1522		.procname	= "udp_hash_entries",
1523		.data		= &init_net.ipv4.sysctl_udp_child_hash_entries,
1524		.mode		= 0444,
1525		.proc_handler	= proc_udp_hash_entries,
1526	},
1527	{
1528		.procname	= "udp_child_hash_entries",
1529		.data		= &init_net.ipv4.sysctl_udp_child_hash_entries,
1530		.maxlen		= sizeof(unsigned int),
1531		.mode		= 0644,
1532		.proc_handler	= proc_douintvec_minmax,
1533		.extra1		= SYSCTL_ZERO,
1534		.extra2		= &udp_child_hash_entries_max,
1535	},
1536	{
1537		.procname	= "udp_rmem_min",
1538		.data		= &init_net.ipv4.sysctl_udp_rmem_min,
1539		.maxlen		= sizeof(init_net.ipv4.sysctl_udp_rmem_min),
1540		.mode		= 0644,
1541		.proc_handler	= proc_dointvec_minmax,
1542		.extra1		= SYSCTL_ONE
1543	},
1544	{
1545		.procname	= "udp_wmem_min",
1546		.data		= &init_net.ipv4.sysctl_udp_wmem_min,
1547		.maxlen		= sizeof(init_net.ipv4.sysctl_udp_wmem_min),
1548		.mode		= 0644,
1549		.proc_handler	= proc_dointvec_minmax,
1550		.extra1		= SYSCTL_ONE
1551	},
1552	{
1553		.procname	= "fib_notify_on_flag_change",
1554		.data		= &init_net.ipv4.sysctl_fib_notify_on_flag_change,
1555		.maxlen		= sizeof(u8),
1556		.mode		= 0644,
1557		.proc_handler	= proc_dou8vec_minmax,
1558		.extra1		= SYSCTL_ZERO,
1559		.extra2		= SYSCTL_TWO,
1560	},
1561	{
1562		.procname       = "tcp_plb_enabled",
1563		.data           = &init_net.ipv4.sysctl_tcp_plb_enabled,
1564		.maxlen         = sizeof(u8),
1565		.mode           = 0644,
1566		.proc_handler   = proc_dou8vec_minmax,
1567		.extra1         = SYSCTL_ZERO,
1568		.extra2         = SYSCTL_ONE,
1569	},
1570	{
1571		.procname       = "tcp_plb_idle_rehash_rounds",
1572		.data           = &init_net.ipv4.sysctl_tcp_plb_idle_rehash_rounds,
1573		.maxlen         = sizeof(u8),
1574		.mode           = 0644,
1575		.proc_handler   = proc_dou8vec_minmax,
1576		.extra2		= &tcp_plb_max_rounds,
1577	},
1578	{
1579		.procname       = "tcp_plb_rehash_rounds",
1580		.data           = &init_net.ipv4.sysctl_tcp_plb_rehash_rounds,
1581		.maxlen         = sizeof(u8),
1582		.mode           = 0644,
1583		.proc_handler   = proc_dou8vec_minmax,
1584		.extra2         = &tcp_plb_max_rounds,
1585	},
1586	{
1587		.procname       = "tcp_plb_suspend_rto_sec",
1588		.data           = &init_net.ipv4.sysctl_tcp_plb_suspend_rto_sec,
1589		.maxlen         = sizeof(u8),
1590		.mode           = 0644,
1591		.proc_handler   = proc_dou8vec_minmax,
1592	},
1593	{
1594		.procname       = "tcp_plb_cong_thresh",
1595		.data           = &init_net.ipv4.sysctl_tcp_plb_cong_thresh,
1596		.maxlen         = sizeof(int),
1597		.mode           = 0644,
1598		.proc_handler   = proc_dointvec_minmax,
1599		.extra1         = SYSCTL_ZERO,
1600		.extra2         = &tcp_plb_max_cong_thresh,
1601	},
1602	{
1603		.procname	= "tcp_syn_linear_timeouts",
1604		.data		= &init_net.ipv4.sysctl_tcp_syn_linear_timeouts,
1605		.maxlen		= sizeof(u8),
1606		.mode		= 0644,
1607		.proc_handler	= proc_dou8vec_minmax,
1608		.extra1		= SYSCTL_ZERO,
1609		.extra2		= &tcp_syn_linear_timeouts_max,
1610	},
1611	{
1612		.procname	= "tcp_shrink_window",
1613		.data		= &init_net.ipv4.sysctl_tcp_shrink_window,
1614		.maxlen		= sizeof(u8),
1615		.mode		= 0644,
1616		.proc_handler	= proc_dou8vec_minmax,
1617		.extra1		= SYSCTL_ZERO,
1618		.extra2		= SYSCTL_ONE,
1619	},
1620	{
1621		.procname	= "tcp_pingpong_thresh",
1622		.data		= &init_net.ipv4.sysctl_tcp_pingpong_thresh,
1623		.maxlen		= sizeof(u8),
1624		.mode		= 0644,
1625		.proc_handler	= proc_dou8vec_minmax,
1626		.extra1		= SYSCTL_ONE,
1627	},
1628	{
1629		.procname	= "tcp_rto_min_us",
1630		.data		= &init_net.ipv4.sysctl_tcp_rto_min_us,
1631		.maxlen		= sizeof(int),
1632		.mode		= 0644,
1633		.proc_handler	= proc_dointvec_minmax,
1634		.extra1		= SYSCTL_ONE,
1635	},
1636	{
1637		.procname	= "tcp_rto_max_ms",
1638		.data		= &init_net.ipv4.sysctl_tcp_rto_max_ms,
1639		.maxlen		= sizeof(int),
1640		.mode		= 0644,
1641		.proc_handler	= proc_dointvec_minmax,
1642		.extra1		= SYSCTL_ONE_THOUSAND,
1643		.extra2		= &tcp_rto_max_max,
1644	},
1645};
1646
1647static __net_init int ipv4_sysctl_init_net(struct net *net)
1648{
1649	size_t table_size = ARRAY_SIZE(ipv4_net_table);
1650	struct ctl_table *table;
1651
1652	table = ipv4_net_table;
1653	if (!net_eq(net, &init_net)) {
1654		int i;
1655
1656		table = kmemdup(table, sizeof(ipv4_net_table), GFP_KERNEL);
1657		if (!table)
1658			goto err_alloc;
1659
1660		for (i = 0; i < table_size; i++) {
1661			if (table[i].data) {
1662				/* Update the variables to point into
1663				 * the current struct net
1664				 */
1665				table[i].data += (void *)net - (void *)&init_net;
1666			} else {
1667				/* Entries without data pointer are global;
1668				 * Make them read-only in non-init_net ns
1669				 */
1670				table[i].mode &= ~0222;
1671			}
1672		}
1673	}
1674
1675	net->ipv4.ipv4_hdr = register_net_sysctl_sz(net, "net/ipv4", table,
1676						    table_size);
1677	if (!net->ipv4.ipv4_hdr)
1678		goto err_reg;
1679
1680	net->ipv4.sysctl_local_reserved_ports = kzalloc(65536 / 8, GFP_KERNEL);
1681	if (!net->ipv4.sysctl_local_reserved_ports)
1682		goto err_ports;
1683
1684	proc_fib_multipath_hash_set_seed(net, 0);
1685
1686	return 0;
1687
1688err_ports:
1689	unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
1690err_reg:
1691	if (!net_eq(net, &init_net))
1692		kfree(table);
1693err_alloc:
1694	return -ENOMEM;
1695}
1696
1697static __net_exit void ipv4_sysctl_exit_net(struct net *net)
1698{
1699	const struct ctl_table *table;
1700
1701	kfree(net->ipv4.sysctl_local_reserved_ports);
1702	table = net->ipv4.ipv4_hdr->ctl_table_arg;
1703	unregister_net_sysctl_table(net->ipv4.ipv4_hdr);
1704	kfree(table);
1705}
1706
1707static __net_initdata struct pernet_operations ipv4_sysctl_ops = {
1708	.init = ipv4_sysctl_init_net,
1709	.exit = ipv4_sysctl_exit_net,
1710};
1711
1712static __init int sysctl_ipv4_init(void)
1713{
1714	struct ctl_table_header *hdr;
1715
1716	hdr = register_net_sysctl(&init_net, "net/ipv4", ipv4_table);
1717	if (!hdr)
1718		return -ENOMEM;
1719
1720	proc_fib_multipath_hash_init_rand_seed();
1721
1722	if (register_pernet_subsys(&ipv4_sysctl_ops)) {
1723		unregister_net_sysctl_table(hdr);
1724		return -ENOMEM;
1725	}
1726
1727	return 0;
1728}
1729
1730__initcall(sysctl_ipv4_init);