[mirror_ubuntu-jammy-kernel.git] / net / mptcp / pm.c

// SPDX-License-Identifier: GPL-2.0
/* Multipath TCP
 *
 * Copyright (c) 2019, Intel Corporation.
 */
#define pr_fmt(fmt) "MPTCP: " fmt

#include <linux/kernel.h>
#include <net/tcp.h>
#include <net/mptcp.h>
#include "protocol.h"

/* path manager command handlers */

int mptcp_pm_announce_addr(struct mptcp_sock *msk,
			   const struct mptcp_addr_info *addr,
			   bool echo)
{
	u8 add_addr = READ_ONCE(msk->pm.add_addr_signal);

	pr_debug("msk=%p, local_id=%d", msk, addr->id);

	msk->pm.local = *addr;
	add_addr |= BIT(MPTCP_ADD_ADDR_SIGNAL);
	if (echo)
		add_addr |= BIT(MPTCP_ADD_ADDR_ECHO);
	WRITE_ONCE(msk->pm.add_addr_signal, add_addr);
	return 0;
}

int mptcp_pm_remove_addr(struct mptcp_sock *msk, u8 local_id)
{
	pr_debug("msk=%p, local_id=%d", msk, local_id);

	msk->pm.rm_id = local_id;
	WRITE_ONCE(msk->pm.rm_addr_signal, true);
	return 0;
}

int mptcp_pm_remove_subflow(struct mptcp_sock *msk, u8 local_id)
{
	pr_debug("msk=%p, local_id=%d", msk, local_id);

	spin_lock_bh(&msk->pm.lock);
	mptcp_pm_nl_rm_subflow_received(msk, local_id);
	spin_unlock_bh(&msk->pm.lock);
	return 0;
}

/* path manager event handlers */

void mptcp_pm_new_connection(struct mptcp_sock *msk, int server_side)
{
	struct mptcp_pm_data *pm = &msk->pm;

	pr_debug("msk=%p, token=%u side=%d", msk, msk->token, server_side);

	WRITE_ONCE(pm->server_side, server_side);
}

bool mptcp_pm_allow_new_subflow(struct mptcp_sock *msk)
{
	struct mptcp_pm_data *pm = &msk->pm;
	int ret = 0;

	pr_debug("msk=%p subflows=%d max=%d allow=%d", msk, pm->subflows,
		 pm->subflows_max, READ_ONCE(pm->accept_subflow));

	/* try to avoid acquiring the lock below */
	if (!READ_ONCE(pm->accept_subflow))
		return false;

	spin_lock_bh(&pm->lock);
	if (READ_ONCE(pm->accept_subflow)) {
		ret = pm->subflows < pm->subflows_max;
		if (ret && ++pm->subflows == pm->subflows_max)
			WRITE_ONCE(pm->accept_subflow, false);
	}
	spin_unlock_bh(&pm->lock);

	return ret;
}

/* return true if the new status bit is currently cleared, that is, this event
 * can be server, eventually by an already scheduled work
 */
static bool mptcp_pm_schedule_work(struct mptcp_sock *msk,
				   enum mptcp_pm_status new_status)
{
	pr_debug("msk=%p status=%x new=%lx", msk, msk->pm.status,
		 BIT(new_status));
	if (msk->pm.status & BIT(new_status))
		return false;

	msk->pm.status |= BIT(new_status);
	mptcp_schedule_work((struct sock *)msk);
	return true;
}

void mptcp_pm_fully_established(struct mptcp_sock *msk)
{
	struct mptcp_pm_data *pm = &msk->pm;

	pr_debug("msk=%p", msk);

	/* try to avoid acquiring the lock below */
	if (!READ_ONCE(pm->work_pending))
		return;

	spin_lock_bh(&pm->lock);

	if (READ_ONCE(pm->work_pending))
		mptcp_pm_schedule_work(msk, MPTCP_PM_ESTABLISHED);

	spin_unlock_bh(&pm->lock);
}

void mptcp_pm_connection_closed(struct mptcp_sock *msk)
{
	pr_debug("msk=%p", msk);
}

void mptcp_pm_subflow_established(struct mptcp_sock *msk,
				  struct mptcp_subflow_context *subflow)
{
	struct mptcp_pm_data *pm = &msk->pm;

	pr_debug("msk=%p", msk);

	if (!READ_ONCE(pm->work_pending))
		return;

	spin_lock_bh(&pm->lock);

	if (READ_ONCE(pm->work_pending))
		mptcp_pm_schedule_work(msk, MPTCP_PM_SUBFLOW_ESTABLISHED);

	spin_unlock_bh(&pm->lock);
}

void mptcp_pm_subflow_closed(struct mptcp_sock *msk, u8 id)
{
	pr_debug("msk=%p", msk);
}

void mptcp_pm_add_addr_received(struct mptcp_sock *msk,
				const struct mptcp_addr_info *addr)
{
	struct mptcp_pm_data *pm = &msk->pm;

	pr_debug("msk=%p remote_id=%d accept=%d", msk, addr->id,
		 READ_ONCE(pm->accept_addr));

	spin_lock_bh(&pm->lock);

	if (!READ_ONCE(pm->accept_addr))
		mptcp_pm_announce_addr(msk, addr, true);
	else if (mptcp_pm_schedule_work(msk, MPTCP_PM_ADD_ADDR_RECEIVED))
		pm->remote = *addr;

	spin_unlock_bh(&pm->lock);
}

void mptcp_pm_rm_addr_received(struct mptcp_sock *msk, u8 rm_id)
{
	struct mptcp_pm_data *pm = &msk->pm;

	pr_debug("msk=%p remote_id=%d", msk, rm_id);

	spin_lock_bh(&pm->lock);
	mptcp_pm_schedule_work(msk, MPTCP_PM_RM_ADDR_RECEIVED);
	pm->rm_id = rm_id;
	spin_unlock_bh(&pm->lock);
}

/* path manager helpers */

bool mptcp_pm_add_addr_signal(struct mptcp_sock *msk, unsigned int remaining,
			      struct mptcp_addr_info *saddr, bool *echo)
{
	int ret = false;

	spin_lock_bh(&msk->pm.lock);

	/* double check after the lock is acquired */
	if (!mptcp_pm_should_add_signal(msk))
		goto out_unlock;

	*echo = mptcp_pm_should_add_signal_echo(msk);

	if (remaining < mptcp_add_addr_len(msk->pm.local.family, *echo))
		goto out_unlock;

	*saddr = msk->pm.local;
	WRITE_ONCE(msk->pm.add_addr_signal, 0);
	ret = true;

out_unlock:
	spin_unlock_bh(&msk->pm.lock);
	return ret;
}

bool mptcp_pm_rm_addr_signal(struct mptcp_sock *msk, unsigned int remaining,
			     u8 *rm_id)
{
	int ret = false;

	spin_lock_bh(&msk->pm.lock);

	/* double check after the lock is acquired */
	if (!mptcp_pm_should_rm_signal(msk))
		goto out_unlock;

	if (remaining < TCPOLEN_MPTCP_RM_ADDR_BASE)
		goto out_unlock;

	*rm_id = msk->pm.rm_id;
	WRITE_ONCE(msk->pm.rm_addr_signal, false);
	ret = true;

out_unlock:
	spin_unlock_bh(&msk->pm.lock);
	return ret;
}

int mptcp_pm_get_local_id(struct mptcp_sock *msk, struct sock_common *skc)
{
	return mptcp_pm_nl_get_local_id(msk, skc);
}

void mptcp_pm_data_init(struct mptcp_sock *msk)
{
	msk->pm.add_addr_signaled = 0;
	msk->pm.add_addr_accepted = 0;
	msk->pm.local_addr_used = 0;
	msk->pm.subflows = 0;
	msk->pm.rm_id = 0;
	WRITE_ONCE(msk->pm.work_pending, false);
	WRITE_ONCE(msk->pm.add_addr_signal, 0);
	WRITE_ONCE(msk->pm.rm_addr_signal, false);
	WRITE_ONCE(msk->pm.accept_addr, false);
	WRITE_ONCE(msk->pm.accept_subflow, false);
	msk->pm.status = 0;

	spin_lock_init(&msk->pm.lock);
	INIT_LIST_HEAD(&msk->pm.anno_list);

	mptcp_pm_nl_data_init(msk);
}

void __init mptcp_pm_init(void)
{
	mptcp_pm_nl_init();
}
Commit	Line	Data
1b1c7a0e PK	1	// SPDX-License-Identifier: GPL-2.0
	2	/* Multipath TCP
	3	*
	4	* Copyright (c) 2019, Intel Corporation.
	5	*/
c85adced GT	6	#define pr_fmt(fmt) "MPTCP: " fmt
c85adced GT	7
1b1c7a0e PK	8	#include <linux/kernel.h>
	9	#include <net/tcp.h>
	10	#include <net/mptcp.h>
	11	#include "protocol.h"
	12
1b1c7a0e PK	13	/* path manager command handlers */
	14
	15	int mptcp_pm_announce_addr(struct mptcp_sock *msk,
6a6c05a8 GT	16	const struct mptcp_addr_info *addr,
6a6c05a8 GT	17	bool echo)
1b1c7a0e	18	{
d91d322a GT	19	u8 add_addr = READ_ONCE(msk->pm.add_addr_signal);
d91d322a GT	20
926bdeab PK	21	pr_debug("msk=%p, local_id=%d", msk, addr->id);
	22
	23	msk->pm.local = *addr;
d91d322a GT	24	add_addr \|= BIT(MPTCP_ADD_ADDR_SIGNAL);
	25	if (echo)
	26	add_addr \|= BIT(MPTCP_ADD_ADDR_ECHO);
	27	WRITE_ONCE(msk->pm.add_addr_signal, add_addr);
926bdeab	28	return 0;
1b1c7a0e PK	29	}
	30
	31	int mptcp_pm_remove_addr(struct mptcp_sock *msk, u8 local_id)
	32	{
b6c08380 GT	33	pr_debug("msk=%p, local_id=%d", msk, local_id);
	34
	35	msk->pm.rm_id = local_id;
	36	WRITE_ONCE(msk->pm.rm_addr_signal, true);
	37	return 0;
1b1c7a0e PK	38	}
1b1c7a0e PK	39
0ee4261a	40	int mptcp_pm_remove_subflow(struct mptcp_sock *msk, u8 local_id)
1b1c7a0e	41	{
0ee4261a GT	42	pr_debug("msk=%p, local_id=%d", msk, local_id);
	43
	44	spin_lock_bh(&msk->pm.lock);
	45	mptcp_pm_nl_rm_subflow_received(msk, local_id);
	46	spin_unlock_bh(&msk->pm.lock);
	47	return 0;
1b1c7a0e PK	48	}
	49
	50	/* path manager event handlers */
	51
	52	void mptcp_pm_new_connection(struct mptcp_sock *msk, int server_side)
	53	{
	54	struct mptcp_pm_data *pm = &msk->pm;
	55
	56	pr_debug("msk=%p, token=%u side=%d", msk, msk->token, server_side);
	57
	58	WRITE_ONCE(pm->server_side, server_side);
	59	}
	60
	61	bool mptcp_pm_allow_new_subflow(struct mptcp_sock *msk)
	62	{
926bdeab	63	struct mptcp_pm_data *pm = &msk->pm;
f58f065a	64	int ret = 0;
926bdeab PK	65
	66	pr_debug("msk=%p subflows=%d max=%d allow=%d", msk, pm->subflows,
	67	pm->subflows_max, READ_ONCE(pm->accept_subflow));
	68
	69	/* try to avoid acquiring the lock below */
	70	if (!READ_ONCE(pm->accept_subflow))
	71	return false;
	72
	73	spin_lock_bh(&pm->lock);
f58f065a GT	74	if (READ_ONCE(pm->accept_subflow)) {
	75	ret = pm->subflows < pm->subflows_max;
	76	if (ret && ++pm->subflows == pm->subflows_max)
	77	WRITE_ONCE(pm->accept_subflow, false);
	78	}
926bdeab PK	79	spin_unlock_bh(&pm->lock);
	80
	81	return ret;
	82	}
	83
	84	/* return true if the new status bit is currently cleared, that is, this event
	85	* can be server, eventually by an already scheduled work
	86	*/
	87	static bool mptcp_pm_schedule_work(struct mptcp_sock *msk,
	88	enum mptcp_pm_status new_status)
	89	{
	90	pr_debug("msk=%p status=%x new=%lx", msk, msk->pm.status,
	91	BIT(new_status));
	92	if (msk->pm.status & BIT(new_status))
	93	return false;
	94
	95	msk->pm.status \|= BIT(new_status);
ba8f48f7	96	mptcp_schedule_work((struct sock *)msk);
926bdeab	97	return true;
1b1c7a0e PK	98	}
	99
	100	void mptcp_pm_fully_established(struct mptcp_sock *msk)
	101	{
926bdeab PK	102	struct mptcp_pm_data *pm = &msk->pm;
926bdeab PK	103
1b1c7a0e	104	pr_debug("msk=%p", msk);
926bdeab PK	105
	106	/* try to avoid acquiring the lock below */
	107	if (!READ_ONCE(pm->work_pending))
	108	return;
	109
	110	spin_lock_bh(&pm->lock);
	111
	112	if (READ_ONCE(pm->work_pending))
	113	mptcp_pm_schedule_work(msk, MPTCP_PM_ESTABLISHED);
	114
	115	spin_unlock_bh(&pm->lock);
1b1c7a0e PK	116	}
	117
	118	void mptcp_pm_connection_closed(struct mptcp_sock *msk)
	119	{
	120	pr_debug("msk=%p", msk);
	121	}
	122
	123	void mptcp_pm_subflow_established(struct mptcp_sock *msk,
	124	struct mptcp_subflow_context *subflow)
	125	{
926bdeab PK	126	struct mptcp_pm_data *pm = &msk->pm;
926bdeab PK	127
1b1c7a0e	128	pr_debug("msk=%p", msk);
926bdeab PK	129
	130	if (!READ_ONCE(pm->work_pending))
	131	return;
	132
	133	spin_lock_bh(&pm->lock);
	134
	135	if (READ_ONCE(pm->work_pending))
	136	mptcp_pm_schedule_work(msk, MPTCP_PM_SUBFLOW_ESTABLISHED);
	137
	138	spin_unlock_bh(&pm->lock);
1b1c7a0e PK	139	}
	140
	141	void mptcp_pm_subflow_closed(struct mptcp_sock *msk, u8 id)
	142	{
	143	pr_debug("msk=%p", msk);
	144	}
	145
	146	void mptcp_pm_add_addr_received(struct mptcp_sock *msk,
	147	const struct mptcp_addr_info *addr)
	148	{
926bdeab PK	149	struct mptcp_pm_data *pm = &msk->pm;
	150
	151	pr_debug("msk=%p remote_id=%d accept=%d", msk, addr->id,
	152	READ_ONCE(pm->accept_addr));
	153
926bdeab PK	154	spin_lock_bh(&pm->lock);
926bdeab PK	155
6a6c05a8 GT	156	if (!READ_ONCE(pm->accept_addr))
	157	mptcp_pm_announce_addr(msk, addr, true);
	158	else if (mptcp_pm_schedule_work(msk, MPTCP_PM_ADD_ADDR_RECEIVED))
926bdeab PK	159	pm->remote = *addr;
	160
	161	spin_unlock_bh(&pm->lock);
1b1c7a0e PK	162	}
1b1c7a0e PK	163
d0876b22 GT	164	void mptcp_pm_rm_addr_received(struct mptcp_sock *msk, u8 rm_id)
	165	{
	166	struct mptcp_pm_data *pm = &msk->pm;
	167
	168	pr_debug("msk=%p remote_id=%d", msk, rm_id);
	169
	170	spin_lock_bh(&pm->lock);
	171	mptcp_pm_schedule_work(msk, MPTCP_PM_RM_ADDR_RECEIVED);
	172	pm->rm_id = rm_id;
	173	spin_unlock_bh(&pm->lock);
	174	}
	175
1b1c7a0e PK	176	/* path manager helpers */
1b1c7a0e PK	177
f643b803	178	bool mptcp_pm_add_addr_signal(struct mptcp_sock *msk, unsigned int remaining,
6a6c05a8	179	struct mptcp_addr_info saddr, bool echo)
1b1c7a0e	180	{
926bdeab PK	181	int ret = false;
	182
	183	spin_lock_bh(&msk->pm.lock);
	184
	185	/* double check after the lock is acquired */
f643b803	186	if (!mptcp_pm_should_add_signal(msk))
926bdeab PK	187	goto out_unlock;
926bdeab PK	188
d91d322a	189	*echo = mptcp_pm_should_add_signal_echo(msk);
456afe01 MB	190
456afe01 MB	191	if (remaining < mptcp_add_addr_len(msk->pm.local.family, *echo))
926bdeab PK	192	goto out_unlock;
	193
	194	*saddr = msk->pm.local;
d91d322a	195	WRITE_ONCE(msk->pm.add_addr_signal, 0);
926bdeab PK	196	ret = true;
	197
	198	out_unlock:
	199	spin_unlock_bh(&msk->pm.lock);
	200	return ret;
1b1c7a0e PK	201	}
1b1c7a0e PK	202
5cb104ae GT	203	bool mptcp_pm_rm_addr_signal(struct mptcp_sock *msk, unsigned int remaining,
	204	u8 *rm_id)
	205	{
	206	int ret = false;
	207
	208	spin_lock_bh(&msk->pm.lock);
	209
	210	/* double check after the lock is acquired */
	211	if (!mptcp_pm_should_rm_signal(msk))
	212	goto out_unlock;
	213
	214	if (remaining < TCPOLEN_MPTCP_RM_ADDR_BASE)
	215	goto out_unlock;
	216
	217	*rm_id = msk->pm.rm_id;
	218	WRITE_ONCE(msk->pm.rm_addr_signal, false);
	219	ret = true;
	220
	221	out_unlock:
	222	spin_unlock_bh(&msk->pm.lock);
	223	return ret;
	224	}
	225
1b1c7a0e PK	226	int mptcp_pm_get_local_id(struct mptcp_sock msk, struct sock_common skc)
1b1c7a0e PK	227	{
01cacb00	228	return mptcp_pm_nl_get_local_id(msk, skc);
1b1c7a0e PK	229	}
1b1c7a0e PK	230
1b1c7a0e PK	231	void mptcp_pm_data_init(struct mptcp_sock *msk)
	232	{
	233	msk->pm.add_addr_signaled = 0;
	234	msk->pm.add_addr_accepted = 0;
	235	msk->pm.local_addr_used = 0;
	236	msk->pm.subflows = 0;
5cb104ae	237	msk->pm.rm_id = 0;
1b1c7a0e	238	WRITE_ONCE(msk->pm.work_pending, false);
d91d322a	239	WRITE_ONCE(msk->pm.add_addr_signal, 0);
5cb104ae	240	WRITE_ONCE(msk->pm.rm_addr_signal, false);
1b1c7a0e PK	241	WRITE_ONCE(msk->pm.accept_addr, false);
	242	WRITE_ONCE(msk->pm.accept_subflow, false);
	243	msk->pm.status = 0;
	244
	245	spin_lock_init(&msk->pm.lock);
b6c08380	246	INIT_LIST_HEAD(&msk->pm.anno_list);
01cacb00 PA	247
01cacb00 PA	248	mptcp_pm_nl_data_init(msk);
1b1c7a0e PK	249	}
1b1c7a0e PK	250
d39dceca	251	void __init mptcp_pm_init(void)
1b1c7a0e	252	{
01cacb00	253	mptcp_pm_nl_init();
1b1c7a0e	254	}