[linux-block.git] / net / ipv4 / tcp_cong.c

/*
 * Pluggable TCP congestion control support and newReno
 * congestion control.
 * Based on ideas from I/O scheduler support and Web100.
 *
 * Copyright (C) 2005 Stephen Hemminger <shemminger@osdl.org>
 */

#define pr_fmt(fmt) "TCP: " fmt

#include <linux/module.h>
#include <linux/mm.h>
#include <linux/types.h>
#include <linux/list.h>
#include <linux/gfp.h>
#include <linux/jhash.h>
#include <net/tcp.h>

static DEFINE_SPINLOCK(tcp_cong_list_lock);
static LIST_HEAD(tcp_cong_list);

/* Simple linear search, don't expect many entries! */
static struct tcp_congestion_ops *tcp_ca_find(const char *name)
{
	struct tcp_congestion_ops *e;

	list_for_each_entry_rcu(e, &tcp_cong_list, list) {
		if (strcmp(e->name, name) == 0)
			return e;
	}

	return NULL;
}

/* Must be called with rcu lock held */
static const struct tcp_congestion_ops *__tcp_ca_find_autoload(const char *name)
{
	const struct tcp_congestion_ops *ca = tcp_ca_find(name);
#ifdef CONFIG_MODULES
	if (!ca && capable(CAP_NET_ADMIN)) {
		rcu_read_unlock();
		request_module("tcp_%s", name);
		rcu_read_lock();
		ca = tcp_ca_find(name);
	}
#endif
	return ca;
}

/* Simple linear search, not much in here. */
struct tcp_congestion_ops *tcp_ca_find_key(u32 key)
{
	struct tcp_congestion_ops *e;

	list_for_each_entry_rcu(e, &tcp_cong_list, list) {
		if (e->key == key)
			return e;
	}

	return NULL;
}

/*
 * Attach new congestion control algorithm to the list
 * of available options.
 */
int tcp_register_congestion_control(struct tcp_congestion_ops *ca)
{
	int ret = 0;

	/* all algorithms must implement ssthresh and cong_avoid ops */
	if (!ca->ssthresh || !ca->cong_avoid) {
		pr_err("%s does not implement required ops\n", ca->name);
		return -EINVAL;
	}

	ca->key = jhash(ca->name, sizeof(ca->name), strlen(ca->name));

	spin_lock(&tcp_cong_list_lock);
	if (ca->key == TCP_CA_UNSPEC || tcp_ca_find_key(ca->key)) {
		pr_notice("%s already registered or non-unique key\n",
			  ca->name);
		ret = -EEXIST;
	} else {
		list_add_tail_rcu(&ca->list, &tcp_cong_list);
		pr_debug("%s registered\n", ca->name);
	}
	spin_unlock(&tcp_cong_list_lock);

	return ret;
}
EXPORT_SYMBOL_GPL(tcp_register_congestion_control);

/*
 * Remove congestion control algorithm, called from
 * the module's remove function.  Module ref counts are used
 * to ensure that this can't be done till all sockets using
 * that method are closed.
 */
void tcp_unregister_congestion_control(struct tcp_congestion_ops *ca)
{
	spin_lock(&tcp_cong_list_lock);
	list_del_rcu(&ca->list);
	spin_unlock(&tcp_cong_list_lock);

	/* Wait for outstanding readers to complete before the
	 * module gets removed entirely.
	 *
	 * A try_module_get() should fail by now as our module is
	 * in "going" state since no refs are held anymore and
	 * module_exit() handler being called.
	 */
	synchronize_rcu();
}
EXPORT_SYMBOL_GPL(tcp_unregister_congestion_control);

u32 tcp_ca_get_key_by_name(const char *name)
{
	const struct tcp_congestion_ops *ca;
	u32 key;

	might_sleep();

	rcu_read_lock();
	ca = __tcp_ca_find_autoload(name);
	key = ca ? ca->key : TCP_CA_UNSPEC;
	rcu_read_unlock();

	return key;
}
EXPORT_SYMBOL_GPL(tcp_ca_get_key_by_name);

char *tcp_ca_get_name_by_key(u32 key, char *buffer)
{
	const struct tcp_congestion_ops *ca;
	char *ret = NULL;

	rcu_read_lock();
	ca = tcp_ca_find_key(key);
	if (ca)
		ret = strncpy(buffer, ca->name,
			      TCP_CA_NAME_MAX);
	rcu_read_unlock();

	return ret;
}
EXPORT_SYMBOL_GPL(tcp_ca_get_name_by_key);

/* Assign choice of congestion control. */
void tcp_assign_congestion_control(struct sock *sk)
{
	struct inet_connection_sock *icsk = inet_csk(sk);
	struct tcp_congestion_ops *ca;

	rcu_read_lock();
	list_for_each_entry_rcu(ca, &tcp_cong_list, list) {
		if (likely(try_module_get(ca->owner))) {
			icsk->icsk_ca_ops = ca;
			goto out;
		}
		/* Fallback to next available. The last really
		 * guaranteed fallback is Reno from this list.
		 */
	}
out:
	rcu_read_unlock();

	/* Clear out private data before diag gets it and
	 * the ca has not been initialized.
	 */
	if (ca->get_info)
		memset(icsk->icsk_ca_priv, 0, sizeof(icsk->icsk_ca_priv));
}

void tcp_init_congestion_control(struct sock *sk)
{
	const struct inet_connection_sock *icsk = inet_csk(sk);

	if (icsk->icsk_ca_ops->init)
		icsk->icsk_ca_ops->init(sk);
}

static void tcp_reinit_congestion_control(struct sock *sk,
					  const struct tcp_congestion_ops *ca)
{
	struct inet_connection_sock *icsk = inet_csk(sk);

	tcp_cleanup_congestion_control(sk);
	icsk->icsk_ca_ops = ca;
	icsk->icsk_ca_setsockopt = 1;

	if (sk->sk_state != TCP_CLOSE && icsk->icsk_ca_ops->init)
		icsk->icsk_ca_ops->init(sk);
}

/* Manage refcounts on socket close. */
void tcp_cleanup_congestion_control(struct sock *sk)
{
	struct inet_connection_sock *icsk = inet_csk(sk);

	if (icsk->icsk_ca_ops->release)
		icsk->icsk_ca_ops->release(sk);
	module_put(icsk->icsk_ca_ops->owner);
}

/* Used by sysctl to change default congestion control */
int tcp_set_default_congestion_control(const char *name)
{
	struct tcp_congestion_ops *ca;
	int ret = -ENOENT;

	spin_lock(&tcp_cong_list_lock);
	ca = tcp_ca_find(name);
#ifdef CONFIG_MODULES
	if (!ca && capable(CAP_NET_ADMIN)) {
		spin_unlock(&tcp_cong_list_lock);

		request_module("tcp_%s", name);
		spin_lock(&tcp_cong_list_lock);
		ca = tcp_ca_find(name);
	}
#endif

	if (ca) {
		ca->flags |= TCP_CONG_NON_RESTRICTED;	/* default is always allowed */
		list_move(&ca->list, &tcp_cong_list);
		ret = 0;
	}
	spin_unlock(&tcp_cong_list_lock);

	return ret;
}

/* Set default value from kernel configuration at bootup */
static int __init tcp_congestion_default(void)
{
	return tcp_set_default_congestion_control(CONFIG_DEFAULT_TCP_CONG);
}
late_initcall(tcp_congestion_default);

/* Build string with list of available congestion control values */
void tcp_get_available_congestion_control(char *buf, size_t maxlen)
{
	struct tcp_congestion_ops *ca;
	size_t offs = 0;

	rcu_read_lock();
	list_for_each_entry_rcu(ca, &tcp_cong_list, list) {
		offs += snprintf(buf + offs, maxlen - offs,
				 "%s%s",
				 offs == 0 ? "" : " ", ca->name);
	}
	rcu_read_unlock();
}

/* Get current default congestion control */
void tcp_get_default_congestion_control(char *name)
{
	struct tcp_congestion_ops *ca;
	/* We will always have reno... */
	BUG_ON(list_empty(&tcp_cong_list));

	rcu_read_lock();
	ca = list_entry(tcp_cong_list.next, struct tcp_congestion_ops, list);
	strncpy(name, ca->name, TCP_CA_NAME_MAX);
	rcu_read_unlock();
}

/* Built list of non-restricted congestion control values */
void tcp_get_allowed_congestion_control(char *buf, size_t maxlen)
{
	struct tcp_congestion_ops *ca;
	size_t offs = 0;

	*buf = '\0';
	rcu_read_lock();
	list_for_each_entry_rcu(ca, &tcp_cong_list, list) {
		if (!(ca->flags & TCP_CONG_NON_RESTRICTED))
			continue;
		offs += snprintf(buf + offs, maxlen - offs,
				 "%s%s",
				 offs == 0 ? "" : " ", ca->name);
	}
	rcu_read_unlock();
}

/* Change list of non-restricted congestion control */
int tcp_set_allowed_congestion_control(char *val)
{
	struct tcp_congestion_ops *ca;
	char *saved_clone, *clone, *name;
	int ret = 0;

	saved_clone = clone = kstrdup(val, GFP_USER);
	if (!clone)
		return -ENOMEM;

	spin_lock(&tcp_cong_list_lock);
	/* pass 1 check for bad entries */
	while ((name = strsep(&clone, " ")) && *name) {
		ca = tcp_ca_find(name);
		if (!ca) {
			ret = -ENOENT;
			goto out;
		}
	}

	/* pass 2 clear old values */
	list_for_each_entry_rcu(ca, &tcp_cong_list, list)
		ca->flags &= ~TCP_CONG_NON_RESTRICTED;

	/* pass 3 mark as allowed */
	while ((name = strsep(&val, " ")) && *name) {
		ca = tcp_ca_find(name);
		WARN_ON(!ca);
		if (ca)
			ca->flags |= TCP_CONG_NON_RESTRICTED;
	}
out:
	spin_unlock(&tcp_cong_list_lock);
	kfree(saved_clone);

	return ret;
}

/* Change congestion control for socket */
int tcp_set_congestion_control(struct sock *sk, const char *name)
{
	struct inet_connection_sock *icsk = inet_csk(sk);
	const struct tcp_congestion_ops *ca;
	int err = 0;

	if (icsk->icsk_ca_dst_locked)
		return -EPERM;

	rcu_read_lock();
	ca = __tcp_ca_find_autoload(name);
	/* No change asking for existing value */
	if (ca == icsk->icsk_ca_ops) {
		icsk->icsk_ca_setsockopt = 1;
		goto out;
	}
	if (!ca)
		err = -ENOENT;
	else if (!((ca->flags & TCP_CONG_NON_RESTRICTED) ||
		   ns_capable(sock_net(sk)->user_ns, CAP_NET_ADMIN)))
		err = -EPERM;
	else if (!try_module_get(ca->owner))
		err = -EBUSY;
	else
		tcp_reinit_congestion_control(sk, ca);
 out:
	rcu_read_unlock();
	return err;
}

/* Slow start is used when congestion window is no greater than the slow start
 * threshold. We base on RFC2581 and also handle stretch ACKs properly.
 * We do not implement RFC3465 Appropriate Byte Counting (ABC) per se but
 * something better;) a packet is only considered (s)acked in its entirety to
 * defend the ACK attacks described in the RFC. Slow start processes a stretch
 * ACK of degree N as if N acks of degree 1 are received back to back except
 * ABC caps N to 2. Slow start exits when cwnd grows over ssthresh and
 * returns the leftover acks to adjust cwnd in congestion avoidance mode.
 */
u32 tcp_slow_start(struct tcp_sock *tp, u32 acked)
{
	u32 cwnd = tp->snd_cwnd + acked;

	if (cwnd > tp->snd_ssthresh)
		cwnd = tp->snd_ssthresh + 1;
	acked -= cwnd - tp->snd_cwnd;
	tp->snd_cwnd = min(cwnd, tp->snd_cwnd_clamp);

	return acked;
}
EXPORT_SYMBOL_GPL(tcp_slow_start);

/* In theory this is tp->snd_cwnd += 1 / tp->snd_cwnd (or alternative w),
 * for every packet that was ACKed.
 */
void tcp_cong_avoid_ai(struct tcp_sock *tp, u32 w, u32 acked)
{
	/* If credits accumulated at a higher w, apply them gently now. */
	if (tp->snd_cwnd_cnt >= w) {
		tp->snd_cwnd_cnt = 0;
		tp->snd_cwnd++;
	}

	tp->snd_cwnd_cnt += acked;
	if (tp->snd_cwnd_cnt >= w) {
		u32 delta = tp->snd_cwnd_cnt / w;

		tp->snd_cwnd_cnt -= delta * w;
		tp->snd_cwnd += delta;
	}
	tp->snd_cwnd = min(tp->snd_cwnd, tp->snd_cwnd_clamp);
}
EXPORT_SYMBOL_GPL(tcp_cong_avoid_ai);

/*
 * TCP Reno congestion control
 * This is special case used for fallback as well.
 */
/* This is Jacobson's slow start and congestion avoidance.
 * SIGCOMM '88, p. 328.
 */
void tcp_reno_cong_avoid(struct sock *sk, u32 ack, u32 acked)
{
	struct tcp_sock *tp = tcp_sk(sk);

	if (!tcp_is_cwnd_limited(sk))
		return;

	/* In "safe" area, increase. */
	if (tp->snd_cwnd <= tp->snd_ssthresh) {
		acked = tcp_slow_start(tp, acked);
		if (!acked)
			return;
	}
	/* In dangerous area, increase slowly. */
	tcp_cong_avoid_ai(tp, tp->snd_cwnd, acked);
}
EXPORT_SYMBOL_GPL(tcp_reno_cong_avoid);

/* Slow start threshold is half the congestion window (min 2) */
u32 tcp_reno_ssthresh(struct sock *sk)
{
	const struct tcp_sock *tp = tcp_sk(sk);

	return max(tp->snd_cwnd >> 1U, 2U);
}
EXPORT_SYMBOL_GPL(tcp_reno_ssthresh);

struct tcp_congestion_ops tcp_reno = {
	.flags		= TCP_CONG_NON_RESTRICTED,
	.name		= "reno",
	.owner		= THIS_MODULE,
	.ssthresh	= tcp_reno_ssthresh,
	.cong_avoid	= tcp_reno_cong_avoid,
};
Commit	Line	Data
317a76f9	1	/*
b92022f3	2	* Pluggable TCP congestion control support and newReno
317a76f9	3	* congestion control.
02582e9b	4	* Based on ideas from I/O scheduler support and Web100.
317a76f9 SH	5	*
	6	* Copyright (C) 2005 Stephen Hemminger <shemminger@osdl.org>
	7	*/
	8
afd46503 JP	9	#define pr_fmt(fmt) "TCP: " fmt
afd46503 JP	10
317a76f9 SH	11	#include <linux/module.h>
	12	#include <linux/mm.h>
	13	#include <linux/types.h>
	14	#include <linux/list.h>
5a0e3ad6	15	#include <linux/gfp.h>
c5c6a8ab	16	#include <linux/jhash.h>
317a76f9 SH	17	#include <net/tcp.h>
	18
	19	static DEFINE_SPINLOCK(tcp_cong_list_lock);
	20	static LIST_HEAD(tcp_cong_list);
	21
	22	/* Simple linear search, don't expect many entries! */
	23	static struct tcp_congestion_ops tcp_ca_find(const char name)
	24	{
	25	struct tcp_congestion_ops *e;
	26
5f8ef48d	27	list_for_each_entry_rcu(e, &tcp_cong_list, list) {
317a76f9 SH	28	if (strcmp(e->name, name) == 0)
	29	return e;
	30	}
	31
	32	return NULL;
	33	}
	34
c5c6a8ab DB	35	/* Must be called with rcu lock held */
	36	static const struct tcp_congestion_ops __tcp_ca_find_autoload(const char name)
	37	{
	38	const struct tcp_congestion_ops *ca = tcp_ca_find(name);
	39	#ifdef CONFIG_MODULES
	40	if (!ca && capable(CAP_NET_ADMIN)) {
	41	rcu_read_unlock();
	42	request_module("tcp_%s", name);
	43	rcu_read_lock();
	44	ca = tcp_ca_find(name);
	45	}
	46	#endif
	47	return ca;
	48	}
	49
	50	/* Simple linear search, not much in here. */
	51	struct tcp_congestion_ops *tcp_ca_find_key(u32 key)
	52	{
	53	struct tcp_congestion_ops *e;
	54
	55	list_for_each_entry_rcu(e, &tcp_cong_list, list) {
	56	if (e->key == key)
	57	return e;
	58	}
	59
	60	return NULL;
	61	}
	62
317a76f9	63	/*
d08df601	64	* Attach new congestion control algorithm to the list
317a76f9 SH	65	* of available options.
	66	*/
	67	int tcp_register_congestion_control(struct tcp_congestion_ops *ca)
	68	{
	69	int ret = 0;
	70
	71	/* all algorithms must implement ssthresh and cong_avoid ops */
72dc5b92	72	if (!ca->ssthresh \|\| !ca->cong_avoid) {
afd46503	73	pr_err("%s does not implement required ops\n", ca->name);
317a76f9 SH	74	return -EINVAL;
	75	}
	76
c5c6a8ab DB	77	ca->key = jhash(ca->name, sizeof(ca->name), strlen(ca->name));
c5c6a8ab DB	78
317a76f9	79	spin_lock(&tcp_cong_list_lock);
c5c6a8ab DB	80	if (ca->key == TCP_CA_UNSPEC \|\| tcp_ca_find_key(ca->key)) {
	81	pr_notice("%s already registered or non-unique key\n",
	82	ca->name);
317a76f9 SH	83	ret = -EEXIST;
317a76f9 SH	84	} else {
3d2573f7	85	list_add_tail_rcu(&ca->list, &tcp_cong_list);
db2855ae	86	pr_debug("%s registered\n", ca->name);
317a76f9 SH	87	}
	88	spin_unlock(&tcp_cong_list_lock);
	89
	90	return ret;
	91	}
	92	EXPORT_SYMBOL_GPL(tcp_register_congestion_control);
	93
	94	/*
	95	* Remove congestion control algorithm, called from
	96	* the module's remove function. Module ref counts are used
	97	* to ensure that this can't be done till all sockets using
	98	* that method are closed.
	99	*/
	100	void tcp_unregister_congestion_control(struct tcp_congestion_ops *ca)
	101	{
	102	spin_lock(&tcp_cong_list_lock);
	103	list_del_rcu(&ca->list);
	104	spin_unlock(&tcp_cong_list_lock);
c5c6a8ab DB	105
	106	/* Wait for outstanding readers to complete before the
	107	* module gets removed entirely.
	108	*
	109	* A try_module_get() should fail by now as our module is
	110	* in "going" state since no refs are held anymore and
	111	* module_exit() handler being called.
	112	*/
	113	synchronize_rcu();
317a76f9 SH	114	}
	115	EXPORT_SYMBOL_GPL(tcp_unregister_congestion_control);
	116
c5c6a8ab DB	117	u32 tcp_ca_get_key_by_name(const char *name)
	118	{
	119	const struct tcp_congestion_ops *ca;
	120	u32 key;
	121
	122	might_sleep();
	123
	124	rcu_read_lock();
	125	ca = __tcp_ca_find_autoload(name);
	126	key = ca ? ca->key : TCP_CA_UNSPEC;
	127	rcu_read_unlock();
	128
	129	return key;
	130	}
	131	EXPORT_SYMBOL_GPL(tcp_ca_get_key_by_name);
	132
	133	char tcp_ca_get_name_by_key(u32 key, char buffer)
	134	{
	135	const struct tcp_congestion_ops *ca;
	136	char *ret = NULL;
	137
	138	rcu_read_lock();
	139	ca = tcp_ca_find_key(key);
	140	if (ca)
	141	ret = strncpy(buffer, ca->name,
	142	TCP_CA_NAME_MAX);
	143	rcu_read_unlock();
	144
	145	return ret;
	146	}
	147	EXPORT_SYMBOL_GPL(tcp_ca_get_name_by_key);
	148
317a76f9	149	/* Assign choice of congestion control. */
55d8694f	150	void tcp_assign_congestion_control(struct sock *sk)
317a76f9	151	{
6687e988	152	struct inet_connection_sock *icsk = inet_csk(sk);
317a76f9 SH	153	struct tcp_congestion_ops *ca;
317a76f9 SH	154
55d8694f FW	155	rcu_read_lock();
	156	list_for_each_entry_rcu(ca, &tcp_cong_list, list) {
	157	if (likely(try_module_get(ca->owner))) {
	158	icsk->icsk_ca_ops = ca;
	159	goto out;
317a76f9	160	}
55d8694f FW	161	/* Fallback to next available. The last really
	162	* guaranteed fallback is Reno from this list.
	163	*/
317a76f9	164	}
55d8694f FW	165	out:
	166	rcu_read_unlock();
	167
	168	/* Clear out private data before diag gets it and
	169	* the ca has not been initialized.
	170	*/
	171	if (ca->get_info)
	172	memset(icsk->icsk_ca_priv, 0, sizeof(icsk->icsk_ca_priv));
	173	}
	174
	175	void tcp_init_congestion_control(struct sock *sk)
	176	{
	177	const struct inet_connection_sock *icsk = inet_csk(sk);
317a76f9	178
6687e988 ACM	179	if (icsk->icsk_ca_ops->init)
6687e988 ACM	180	icsk->icsk_ca_ops->init(sk);
317a76f9 SH	181	}
317a76f9 SH	182
29ba4fff DB	183	static void tcp_reinit_congestion_control(struct sock *sk,
	184	const struct tcp_congestion_ops *ca)
	185	{
	186	struct inet_connection_sock *icsk = inet_csk(sk);
	187
	188	tcp_cleanup_congestion_control(sk);
	189	icsk->icsk_ca_ops = ca;
9f950415	190	icsk->icsk_ca_setsockopt = 1;
29ba4fff DB	191
	192	if (sk->sk_state != TCP_CLOSE && icsk->icsk_ca_ops->init)
	193	icsk->icsk_ca_ops->init(sk);
	194	}
	195
317a76f9	196	/* Manage refcounts on socket close. */
6687e988	197	void tcp_cleanup_congestion_control(struct sock *sk)
317a76f9	198	{
6687e988 ACM	199	struct inet_connection_sock *icsk = inet_csk(sk);
	200
	201	if (icsk->icsk_ca_ops->release)
	202	icsk->icsk_ca_ops->release(sk);
	203	module_put(icsk->icsk_ca_ops->owner);
317a76f9 SH	204	}
	205
	206	/* Used by sysctl to change default congestion control */
	207	int tcp_set_default_congestion_control(const char *name)
	208	{
	209	struct tcp_congestion_ops *ca;
	210	int ret = -ENOENT;
	211
	212	spin_lock(&tcp_cong_list_lock);
	213	ca = tcp_ca_find(name);
95a5afca	214	#ifdef CONFIG_MODULES
a8f80e8f	215	if (!ca && capable(CAP_NET_ADMIN)) {
317a76f9 SH	216	spin_unlock(&tcp_cong_list_lock);
	217
	218	request_module("tcp_%s", name);
	219	spin_lock(&tcp_cong_list_lock);
	220	ca = tcp_ca_find(name);
	221	}
	222	#endif
	223
	224	if (ca) {
164891aa	225	ca->flags \|= TCP_CONG_NON_RESTRICTED; /* default is always allowed */
317a76f9 SH	226	list_move(&ca->list, &tcp_cong_list);
	227	ret = 0;
	228	}
	229	spin_unlock(&tcp_cong_list_lock);
	230
	231	return ret;
	232	}
	233
b1736a71 SH	234	/* Set default value from kernel configuration at bootup */
	235	static int __init tcp_congestion_default(void)
	236	{
	237	return tcp_set_default_congestion_control(CONFIG_DEFAULT_TCP_CONG);
	238	}
	239	late_initcall(tcp_congestion_default);
	240
3ff825b2 SH	241	/* Build string with list of available congestion control values */
	242	void tcp_get_available_congestion_control(char *buf, size_t maxlen)
	243	{
	244	struct tcp_congestion_ops *ca;
	245	size_t offs = 0;
	246
	247	rcu_read_lock();
	248	list_for_each_entry_rcu(ca, &tcp_cong_list, list) {
	249	offs += snprintf(buf + offs, maxlen - offs,
	250	"%s%s",
	251	offs == 0 ? "" : " ", ca->name);
3ff825b2 SH	252	}
	253	rcu_read_unlock();
	254	}
	255
317a76f9 SH	256	/* Get current default congestion control */
	257	void tcp_get_default_congestion_control(char *name)
	258	{
	259	struct tcp_congestion_ops *ca;
	260	/* We will always have reno... */
	261	BUG_ON(list_empty(&tcp_cong_list));
	262
	263	rcu_read_lock();
	264	ca = list_entry(tcp_cong_list.next, struct tcp_congestion_ops, list);
	265	strncpy(name, ca->name, TCP_CA_NAME_MAX);
	266	rcu_read_unlock();
	267	}
	268
ce7bc3bf SH	269	/* Built list of non-restricted congestion control values */
	270	void tcp_get_allowed_congestion_control(char *buf, size_t maxlen)
	271	{
	272	struct tcp_congestion_ops *ca;
	273	size_t offs = 0;
	274
	275	*buf = '\0';
	276	rcu_read_lock();
	277	list_for_each_entry_rcu(ca, &tcp_cong_list, list) {
164891aa	278	if (!(ca->flags & TCP_CONG_NON_RESTRICTED))
ce7bc3bf SH	279	continue;
	280	offs += snprintf(buf + offs, maxlen - offs,
	281	"%s%s",
	282	offs == 0 ? "" : " ", ca->name);
ce7bc3bf SH	283	}
	284	rcu_read_unlock();
	285	}
	286
	287	/* Change list of non-restricted congestion control */
	288	int tcp_set_allowed_congestion_control(char *val)
	289	{
	290	struct tcp_congestion_ops *ca;
c34186ed	291	char saved_clone, clone, *name;
ce7bc3bf SH	292	int ret = 0;
ce7bc3bf SH	293
c34186ed	294	saved_clone = clone = kstrdup(val, GFP_USER);
ce7bc3bf SH	295	if (!clone)
	296	return -ENOMEM;
	297
	298	spin_lock(&tcp_cong_list_lock);
	299	/* pass 1 check for bad entries */
	300	while ((name = strsep(&clone, " ")) && *name) {
	301	ca = tcp_ca_find(name);
	302	if (!ca) {
	303	ret = -ENOENT;
	304	goto out;
	305	}
	306	}
	307
164891aa	308	/* pass 2 clear old values */
ce7bc3bf	309	list_for_each_entry_rcu(ca, &tcp_cong_list, list)
164891aa	310	ca->flags &= ~TCP_CONG_NON_RESTRICTED;
ce7bc3bf SH	311
	312	/* pass 3 mark as allowed */
	313	while ((name = strsep(&val, " ")) && *name) {
	314	ca = tcp_ca_find(name);
	315	WARN_ON(!ca);
	316	if (ca)
164891aa	317	ca->flags \|= TCP_CONG_NON_RESTRICTED;
ce7bc3bf SH	318	}
	319	out:
	320	spin_unlock(&tcp_cong_list_lock);
c34186ed	321	kfree(saved_clone);
ce7bc3bf SH	322
	323	return ret;
	324	}
	325
5f8ef48d	326	/* Change congestion control for socket */
6687e988	327	int tcp_set_congestion_control(struct sock sk, const char name)
5f8ef48d	328	{
6687e988	329	struct inet_connection_sock *icsk = inet_csk(sk);
c5c6a8ab	330	const struct tcp_congestion_ops *ca;
5f8ef48d SH	331	int err = 0;
5f8ef48d SH	332
c5c6a8ab DB	333	if (icsk->icsk_ca_dst_locked)
c5c6a8ab DB	334	return -EPERM;
4d4d3d1e	335
c5c6a8ab DB	336	rcu_read_lock();
	337	ca = __tcp_ca_find_autoload(name);
	338	/* No change asking for existing value */
9f950415 NC	339	if (ca == icsk->icsk_ca_ops) {
9f950415 NC	340	icsk->icsk_ca_setsockopt = 1;
5f8ef48d	341	goto out;
9f950415	342	}
5f8ef48d SH	343	if (!ca)
5f8ef48d SH	344	err = -ENOENT;
52e804c6 EB	345	else if (!((ca->flags & TCP_CONG_NON_RESTRICTED) \|\|
52e804c6 EB	346	ns_capable(sock_net(sk)->user_ns, CAP_NET_ADMIN)))
ce7bc3bf	347	err = -EPERM;
5f8ef48d SH	348	else if (!try_module_get(ca->owner))
5f8ef48d SH	349	err = -EBUSY;
29ba4fff DB	350	else
29ba4fff DB	351	tcp_reinit_congestion_control(sk, ca);
5f8ef48d SH	352	out:
	353	rcu_read_unlock();
	354	return err;
	355	}
	356
9f9843a7 YC	357	/* Slow start is used when congestion window is no greater than the slow start
	358	* threshold. We base on RFC2581 and also handle stretch ACKs properly.
	359	* We do not implement RFC3465 Appropriate Byte Counting (ABC) per se but
	360	* something better;) a packet is only considered (s)acked in its entirety to
	361	* defend the ACK attacks described in the RFC. Slow start processes a stretch
	362	* ACK of degree N as if N acks of degree 1 are received back to back except
	363	* ABC caps N to 2. Slow start exits when cwnd grows over ssthresh and
	364	* returns the leftover acks to adjust cwnd in congestion avoidance mode.
40efc6fa	365	*/
e73ebb08	366	u32 tcp_slow_start(struct tcp_sock *tp, u32 acked)
40efc6fa	367	{
9f9843a7	368	u32 cwnd = tp->snd_cwnd + acked;
a02ba041	369
9f9843a7 YC	370	if (cwnd > tp->snd_ssthresh)
9f9843a7 YC	371	cwnd = tp->snd_ssthresh + 1;
e73ebb08	372	acked -= cwnd - tp->snd_cwnd;
9f9843a7	373	tp->snd_cwnd = min(cwnd, tp->snd_cwnd_clamp);
e73ebb08 NC	374
e73ebb08 NC	375	return acked;
40efc6fa SH	376	}
	377	EXPORT_SYMBOL_GPL(tcp_slow_start);
	378
814d488c NC	379	/* In theory this is tp->snd_cwnd += 1 / tp->snd_cwnd (or alternative w),
	380	* for every packet that was ACKed.
	381	*/
e73ebb08	382	void tcp_cong_avoid_ai(struct tcp_sock *tp, u32 w, u32 acked)
758ce5c8	383	{
9949afa4 NC	384	/* If credits accumulated at a higher w, apply them gently now. */
	385	if (tp->snd_cwnd_cnt >= w) {
	386	tp->snd_cwnd_cnt = 0;
	387	tp->snd_cwnd++;
	388	}
	389
814d488c	390	tp->snd_cwnd_cnt += acked;
758ce5c8	391	if (tp->snd_cwnd_cnt >= w) {
814d488c NC	392	u32 delta = tp->snd_cwnd_cnt / w;
	393
	394	tp->snd_cwnd_cnt -= delta * w;
	395	tp->snd_cwnd += delta;
758ce5c8	396	}
814d488c	397	tp->snd_cwnd = min(tp->snd_cwnd, tp->snd_cwnd_clamp);
758ce5c8 IJ	398	}
	399	EXPORT_SYMBOL_GPL(tcp_cong_avoid_ai);
	400
317a76f9 SH	401	/*
	402	* TCP Reno congestion control
	403	* This is special case used for fallback as well.
	404	*/
	405	/* This is Jacobson's slow start and congestion avoidance.
	406	* SIGCOMM '88, p. 328.
	407	*/
24901551	408	void tcp_reno_cong_avoid(struct sock *sk, u32 ack, u32 acked)
317a76f9	409	{
6687e988 ACM	410	struct tcp_sock *tp = tcp_sk(sk);
6687e988 ACM	411
24901551	412	if (!tcp_is_cwnd_limited(sk))
317a76f9 SH	413	return;
317a76f9 SH	414
7faffa1c	415	/* In "safe" area, increase. */
c22bdca9 NC	416	if (tp->snd_cwnd <= tp->snd_ssthresh) {
	417	acked = tcp_slow_start(tp, acked);
	418	if (!acked)
	419	return;
	420	}
e905a9ed	421	/* In dangerous area, increase slowly. */
c22bdca9	422	tcp_cong_avoid_ai(tp, tp->snd_cwnd, acked);
317a76f9 SH	423	}
	424	EXPORT_SYMBOL_GPL(tcp_reno_cong_avoid);
	425
	426	/* Slow start threshold is half the congestion window (min 2) */
6687e988	427	u32 tcp_reno_ssthresh(struct sock *sk)
317a76f9	428	{
6687e988	429	const struct tcp_sock *tp = tcp_sk(sk);
688d1945	430
317a76f9 SH	431	return max(tp->snd_cwnd >> 1U, 2U);
	432	}
	433	EXPORT_SYMBOL_GPL(tcp_reno_ssthresh);
	434
317a76f9	435	struct tcp_congestion_ops tcp_reno = {
164891aa	436	.flags = TCP_CONG_NON_RESTRICTED,
317a76f9 SH	437	.name = "reno",
	438	.owner = THIS_MODULE,
	439	.ssthresh = tcp_reno_ssthresh,
	440	.cong_avoid = tcp_reno_cong_avoid,
317a76f9	441	};