[linux.git] / net / mptcp / pm.c

// SPDX-License-Identifier: GPL-2.0
/* Multipath TCP
 *
 * Copyright (c) 2019, Intel Corporation.
 */
#include <linux/kernel.h>
#include <net/tcp.h>
#include <net/mptcp.h>
#include "protocol.h"

static struct workqueue_struct *pm_wq;

/* path manager command handlers */

int mptcp_pm_announce_addr(struct mptcp_sock *msk,
			   const struct mptcp_addr_info *addr)
{
	pr_debug("msk=%p, local_id=%d", msk, addr->id);

	msk->pm.local = *addr;
	WRITE_ONCE(msk->pm.addr_signal, true);
	return 0;
}

int mptcp_pm_remove_addr(struct mptcp_sock *msk, u8 local_id)
{
	return -ENOTSUPP;
}

int mptcp_pm_remove_subflow(struct mptcp_sock *msk, u8 remote_id)
{
	return -ENOTSUPP;
}

/* 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;

	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);
	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);
	if (queue_work(pm_wq, &msk->pm.work))
		sock_hold((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));

	/* avoid acquiring the lock if there is no room for fouther addresses */
	if (!READ_ONCE(pm->accept_addr))
		return;

	spin_lock_bh(&pm->lock);

	/* be sure there is something to signal re-checking under PM lock */
	if (READ_ONCE(pm->accept_addr) &&
	    mptcp_pm_schedule_work(msk, MPTCP_PM_ADD_ADDR_RECEIVED))
		pm->remote = *addr;

	spin_unlock_bh(&pm->lock);
}

/* path manager helpers */

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

	spin_lock_bh(&msk->pm.lock);

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

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

	*saddr = msk->pm.local;
	WRITE_ONCE(msk->pm.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 0;
}

static void pm_worker(struct work_struct *work)
{
	struct mptcp_pm_data *pm = container_of(work, struct mptcp_pm_data,
						work);
	struct mptcp_sock *msk = container_of(pm, struct mptcp_sock, pm);
	struct sock *sk = (struct sock *)msk;

	lock_sock(sk);
	spin_lock_bh(&msk->pm.lock);

	pr_debug("msk=%p status=%x", msk, pm->status);
	if (pm->status & BIT(MPTCP_PM_ADD_ADDR_RECEIVED)) {
		pm->status &= ~BIT(MPTCP_PM_ADD_ADDR_RECEIVED);
	}
	if (pm->status & BIT(MPTCP_PM_ESTABLISHED)) {
		pm->status &= ~BIT(MPTCP_PM_ESTABLISHED);
	}
	if (pm->status & BIT(MPTCP_PM_SUBFLOW_ESTABLISHED)) {
		pm->status &= ~BIT(MPTCP_PM_SUBFLOW_ESTABLISHED);
	}

	spin_unlock_bh(&msk->pm.lock);
	release_sock(sk);
	sock_put(sk);
}

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;
	WRITE_ONCE(msk->pm.work_pending, false);
	WRITE_ONCE(msk->pm.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_WORK(&msk->pm.work, pm_worker);
}

void mptcp_pm_close(struct mptcp_sock *msk)
{
	if (cancel_work_sync(&msk->pm.work))
		sock_put((struct sock *)msk);
}

void mptcp_pm_init(void)
{
	pm_wq = alloc_workqueue("pm_wq", WQ_UNBOUND | WQ_MEM_RECLAIM, 8);
	if (!pm_wq)
		panic("Failed to allocate workqueue");
}
Commit	Line	Data
1b1c7a0e PK	1	// SPDX-License-Identifier: GPL-2.0
	2	/* Multipath TCP
	3	*
	4	* Copyright (c) 2019, Intel Corporation.
	5	*/
	6	#include <linux/kernel.h>
	7	#include <net/tcp.h>
	8	#include <net/mptcp.h>
	9	#include "protocol.h"
	10
	11	static struct workqueue_struct *pm_wq;
	12
	13	/* path manager command handlers */
	14
	15	int mptcp_pm_announce_addr(struct mptcp_sock *msk,
	16	const struct mptcp_addr_info *addr)
	17	{
926bdeab PK	18	pr_debug("msk=%p, local_id=%d", msk, addr->id);
	19
	20	msk->pm.local = *addr;
	21	WRITE_ONCE(msk->pm.addr_signal, true);
	22	return 0;
1b1c7a0e PK	23	}
	24
	25	int mptcp_pm_remove_addr(struct mptcp_sock *msk, u8 local_id)
	26	{
	27	return -ENOTSUPP;
	28	}
	29
	30	int mptcp_pm_remove_subflow(struct mptcp_sock *msk, u8 remote_id)
	31	{
	32	return -ENOTSUPP;
	33	}
	34
	35	/* path manager event handlers */
	36
	37	void mptcp_pm_new_connection(struct mptcp_sock *msk, int server_side)
	38	{
	39	struct mptcp_pm_data *pm = &msk->pm;
	40
	41	pr_debug("msk=%p, token=%u side=%d", msk, msk->token, server_side);
	42
	43	WRITE_ONCE(pm->server_side, server_side);
	44	}
	45
	46	bool mptcp_pm_allow_new_subflow(struct mptcp_sock *msk)
	47	{
926bdeab PK	48	struct mptcp_pm_data *pm = &msk->pm;
	49	int ret;
	50
	51	pr_debug("msk=%p subflows=%d max=%d allow=%d", msk, pm->subflows,
	52	pm->subflows_max, READ_ONCE(pm->accept_subflow));
	53
	54	/* try to avoid acquiring the lock below */
	55	if (!READ_ONCE(pm->accept_subflow))
	56	return false;
	57
	58	spin_lock_bh(&pm->lock);
	59	ret = pm->subflows < pm->subflows_max;
	60	if (ret && ++pm->subflows == pm->subflows_max)
	61	WRITE_ONCE(pm->accept_subflow, false);
	62	spin_unlock_bh(&pm->lock);
	63
	64	return ret;
	65	}
	66
	67	/* return true if the new status bit is currently cleared, that is, this event
	68	* can be server, eventually by an already scheduled work
	69	*/
	70	static bool mptcp_pm_schedule_work(struct mptcp_sock *msk,
	71	enum mptcp_pm_status new_status)
	72	{
	73	pr_debug("msk=%p status=%x new=%lx", msk, msk->pm.status,
	74	BIT(new_status));
	75	if (msk->pm.status & BIT(new_status))
	76	return false;
	77
	78	msk->pm.status \|= BIT(new_status);
	79	if (queue_work(pm_wq, &msk->pm.work))
	80	sock_hold((struct sock *)msk);
	81	return true;
1b1c7a0e PK	82	}
	83
	84	void mptcp_pm_fully_established(struct mptcp_sock *msk)
	85	{
926bdeab PK	86	struct mptcp_pm_data *pm = &msk->pm;
926bdeab PK	87
1b1c7a0e	88	pr_debug("msk=%p", msk);
926bdeab PK	89
	90	/* try to avoid acquiring the lock below */
	91	if (!READ_ONCE(pm->work_pending))
	92	return;
	93
	94	spin_lock_bh(&pm->lock);
	95
	96	if (READ_ONCE(pm->work_pending))
	97	mptcp_pm_schedule_work(msk, MPTCP_PM_ESTABLISHED);
	98
	99	spin_unlock_bh(&pm->lock);
1b1c7a0e PK	100	}
	101
	102	void mptcp_pm_connection_closed(struct mptcp_sock *msk)
	103	{
	104	pr_debug("msk=%p", msk);
	105	}
	106
	107	void mptcp_pm_subflow_established(struct mptcp_sock *msk,
	108	struct mptcp_subflow_context *subflow)
	109	{
926bdeab PK	110	struct mptcp_pm_data *pm = &msk->pm;
926bdeab PK	111
1b1c7a0e	112	pr_debug("msk=%p", msk);
926bdeab PK	113
	114	if (!READ_ONCE(pm->work_pending))
	115	return;
	116
	117	spin_lock_bh(&pm->lock);
	118
	119	if (READ_ONCE(pm->work_pending))
	120	mptcp_pm_schedule_work(msk, MPTCP_PM_SUBFLOW_ESTABLISHED);
	121
	122	spin_unlock_bh(&pm->lock);
1b1c7a0e PK	123	}
	124
	125	void mptcp_pm_subflow_closed(struct mptcp_sock *msk, u8 id)
	126	{
	127	pr_debug("msk=%p", msk);
	128	}
	129
	130	void mptcp_pm_add_addr_received(struct mptcp_sock *msk,
	131	const struct mptcp_addr_info *addr)
	132	{
926bdeab PK	133	struct mptcp_pm_data *pm = &msk->pm;
	134
	135	pr_debug("msk=%p remote_id=%d accept=%d", msk, addr->id,
	136	READ_ONCE(pm->accept_addr));
	137
	138	/* avoid acquiring the lock if there is no room for fouther addresses */
	139	if (!READ_ONCE(pm->accept_addr))
	140	return;
	141
	142	spin_lock_bh(&pm->lock);
	143
	144	/* be sure there is something to signal re-checking under PM lock */
	145	if (READ_ONCE(pm->accept_addr) &&
	146	mptcp_pm_schedule_work(msk, MPTCP_PM_ADD_ADDR_RECEIVED))
	147	pm->remote = *addr;
	148
	149	spin_unlock_bh(&pm->lock);
1b1c7a0e PK	150	}
	151
	152	/* path manager helpers */
	153
	154	bool mptcp_pm_addr_signal(struct mptcp_sock *msk, unsigned int remaining,
	155	struct mptcp_addr_info *saddr)
	156	{
926bdeab PK	157	int ret = false;
	158
	159	spin_lock_bh(&msk->pm.lock);
	160
	161	/* double check after the lock is acquired */
	162	if (!mptcp_pm_should_signal(msk))
	163	goto out_unlock;
	164
	165	if (remaining < mptcp_add_addr_len(msk->pm.local.family))
	166	goto out_unlock;
	167
	168	*saddr = msk->pm.local;
	169	WRITE_ONCE(msk->pm.addr_signal, false);
	170	ret = true;
	171
	172	out_unlock:
	173	spin_unlock_bh(&msk->pm.lock);
	174	return ret;
1b1c7a0e PK	175	}
	176
	177	int mptcp_pm_get_local_id(struct mptcp_sock msk, struct sock_common skc)
	178	{
	179	return 0;
	180	}
	181
	182	static void pm_worker(struct work_struct *work)
	183	{
926bdeab PK	184	struct mptcp_pm_data *pm = container_of(work, struct mptcp_pm_data,
	185	work);
	186	struct mptcp_sock *msk = container_of(pm, struct mptcp_sock, pm);
	187	struct sock sk = (struct sock )msk;
	188
	189	lock_sock(sk);
	190	spin_lock_bh(&msk->pm.lock);
	191
	192	pr_debug("msk=%p status=%x", msk, pm->status);
	193	if (pm->status & BIT(MPTCP_PM_ADD_ADDR_RECEIVED)) {
	194	pm->status &= ~BIT(MPTCP_PM_ADD_ADDR_RECEIVED);
	195	}
	196	if (pm->status & BIT(MPTCP_PM_ESTABLISHED)) {
	197	pm->status &= ~BIT(MPTCP_PM_ESTABLISHED);
	198	}
	199	if (pm->status & BIT(MPTCP_PM_SUBFLOW_ESTABLISHED)) {
	200	pm->status &= ~BIT(MPTCP_PM_SUBFLOW_ESTABLISHED);
	201	}
	202
	203	spin_unlock_bh(&msk->pm.lock);
	204	release_sock(sk);
	205	sock_put(sk);
1b1c7a0e PK	206	}
	207
	208	void mptcp_pm_data_init(struct mptcp_sock *msk)
	209	{
	210	msk->pm.add_addr_signaled = 0;
	211	msk->pm.add_addr_accepted = 0;
	212	msk->pm.local_addr_used = 0;
	213	msk->pm.subflows = 0;
	214	WRITE_ONCE(msk->pm.work_pending, false);
	215	WRITE_ONCE(msk->pm.addr_signal, false);
	216	WRITE_ONCE(msk->pm.accept_addr, false);
	217	WRITE_ONCE(msk->pm.accept_subflow, false);
	218	msk->pm.status = 0;
	219
	220	spin_lock_init(&msk->pm.lock);
	221	INIT_WORK(&msk->pm.work, pm_worker);
	222	}
	223
926bdeab PK	224	void mptcp_pm_close(struct mptcp_sock *msk)
	225	{
	226	if (cancel_work_sync(&msk->pm.work))
	227	sock_put((struct sock *)msk);
	228	}
	229
1b1c7a0e PK	230	void mptcp_pm_init(void)
	231	{
	232	pm_wq = alloc_workqueue("pm_wq", WQ_UNBOUND \| WQ_MEM_RECLAIM, 8);
	233	if (!pm_wq)
	234	panic("Failed to allocate workqueue");
	235	}