[linux.git] / net / smc / smc_cdc.c

// SPDX-License-Identifier: GPL-2.0
/*
 * Shared Memory Communications over RDMA (SMC-R) and RoCE
 *
 * Connection Data Control (CDC)
 * handles flow control
 *
 * Copyright IBM Corp. 2016
 *
 * Author(s):  Ursula Braun <[email protected]>
 */

#include <linux/spinlock.h>

#include "smc.h"
#include "smc_wr.h"
#include "smc_cdc.h"
#include "smc_tx.h"
#include "smc_rx.h"
#include "smc_close.h"

/********************************** send *************************************/

/* handler for send/transmission completion of a CDC msg */
static void smc_cdc_tx_handler(struct smc_wr_tx_pend_priv *pnd_snd,
			       struct smc_link *link,
			       enum ib_wc_status wc_status)
{
	struct smc_cdc_tx_pend *cdcpend = (struct smc_cdc_tx_pend *)pnd_snd;
	struct smc_connection *conn = cdcpend->conn;
	struct smc_sock *smc;
	int diff;

	if (!conn)
		/* already dismissed */
		return;

	smc = container_of(conn, struct smc_sock, conn);
	bh_lock_sock(&smc->sk);
	if (!wc_status) {
		diff = smc_curs_diff(cdcpend->conn->sndbuf_desc->len,
				     &cdcpend->conn->tx_curs_fin,
				     &cdcpend->cursor);
		/* sndbuf_space is decreased in smc_sendmsg */
		smp_mb__before_atomic();
		atomic_add(diff, &cdcpend->conn->sndbuf_space);
		/* guarantee 0 <= sndbuf_space <= sndbuf_desc->len */
		smp_mb__after_atomic();
		smc_curs_copy(&conn->tx_curs_fin, &cdcpend->cursor, conn);
		smc_curs_copy(&conn->local_tx_ctrl_fin, &cdcpend->p_cursor,
			      conn);
		conn->tx_cdc_seq_fin = cdcpend->ctrl_seq;
	}
	smc_tx_sndbuf_nonfull(smc);
	bh_unlock_sock(&smc->sk);
}

int smc_cdc_get_free_slot(struct smc_connection *conn,
			  struct smc_link *link,
			  struct smc_wr_buf **wr_buf,
			  struct smc_rdma_wr **wr_rdma_buf,
			  struct smc_cdc_tx_pend **pend)
{
	int rc;

	rc = smc_wr_tx_get_free_slot(link, smc_cdc_tx_handler, wr_buf,
				     wr_rdma_buf,
				     (struct smc_wr_tx_pend_priv **)pend);
	if (conn->killed) {
		/* abnormal termination */
		if (!rc)
			smc_wr_tx_put_slot(link,
					   (struct smc_wr_tx_pend_priv *)pend);
		rc = -EPIPE;
	}
	return rc;
}

static inline void smc_cdc_add_pending_send(struct smc_connection *conn,
					    struct smc_cdc_tx_pend *pend)
{
	BUILD_BUG_ON_MSG(
		sizeof(struct smc_cdc_msg) > SMC_WR_BUF_SIZE,
		"must increase SMC_WR_BUF_SIZE to at least sizeof(struct smc_cdc_msg)");
	BUILD_BUG_ON_MSG(
		offsetofend(struct smc_cdc_msg, reserved) > SMC_WR_TX_SIZE,
		"must adapt SMC_WR_TX_SIZE to sizeof(struct smc_cdc_msg); if not all smc_wr upper layer protocols use the same message size any more, must start to set link->wr_tx_sges[i].length on each individual smc_wr_tx_send()");
	BUILD_BUG_ON_MSG(
		sizeof(struct smc_cdc_tx_pend) > SMC_WR_TX_PEND_PRIV_SIZE,
		"must increase SMC_WR_TX_PEND_PRIV_SIZE to at least sizeof(struct smc_cdc_tx_pend)");
	pend->conn = conn;
	pend->cursor = conn->tx_curs_sent;
	pend->p_cursor = conn->local_tx_ctrl.prod;
	pend->ctrl_seq = conn->tx_cdc_seq;
}

int smc_cdc_msg_send(struct smc_connection *conn,
		     struct smc_wr_buf *wr_buf,
		     struct smc_cdc_tx_pend *pend)
{
	struct smc_link *link = conn->lnk;
	union smc_host_cursor cfed;
	int rc;

	smc_cdc_add_pending_send(conn, pend);

	conn->tx_cdc_seq++;
	conn->local_tx_ctrl.seqno = conn->tx_cdc_seq;
	smc_host_msg_to_cdc((struct smc_cdc_msg *)wr_buf, conn, &cfed);
	rc = smc_wr_tx_send(link, (struct smc_wr_tx_pend_priv *)pend);
	if (!rc) {
		smc_curs_copy(&conn->rx_curs_confirmed, &cfed, conn);
		conn->local_rx_ctrl.prod_flags.cons_curs_upd_req = 0;
	} else {
		conn->tx_cdc_seq--;
		conn->local_tx_ctrl.seqno = conn->tx_cdc_seq;
	}

	return rc;
}

/* send a validation msg indicating the move of a conn to an other QP link */
int smcr_cdc_msg_send_validation(struct smc_connection *conn,
				 struct smc_cdc_tx_pend *pend,
				 struct smc_wr_buf *wr_buf)
{
	struct smc_host_cdc_msg *local = &conn->local_tx_ctrl;
	struct smc_link *link = conn->lnk;
	struct smc_cdc_msg *peer;
	int rc;

	peer = (struct smc_cdc_msg *)wr_buf;
	peer->common.type = local->common.type;
	peer->len = local->len;
	peer->seqno = htons(conn->tx_cdc_seq_fin); /* seqno last compl. tx */
	peer->token = htonl(local->token);
	peer->prod_flags.failover_validation = 1;

	rc = smc_wr_tx_send(link, (struct smc_wr_tx_pend_priv *)pend);
	return rc;
}

static int smcr_cdc_get_slot_and_msg_send(struct smc_connection *conn)
{
	struct smc_cdc_tx_pend *pend;
	struct smc_wr_buf *wr_buf;
	struct smc_link *link;
	bool again = false;
	int rc;

again:
	link = conn->lnk;
	rc = smc_cdc_get_free_slot(conn, link, &wr_buf, NULL, &pend);
	if (rc)
		return rc;

	spin_lock_bh(&conn->send_lock);
	if (link != conn->lnk) {
		/* link of connection changed, try again one time*/
		spin_unlock_bh(&conn->send_lock);
		smc_wr_tx_put_slot(link,
				   (struct smc_wr_tx_pend_priv *)pend);
		if (again)
			return -ENOLINK;
		again = true;
		goto again;
	}
	rc = smc_cdc_msg_send(conn, wr_buf, pend);
	spin_unlock_bh(&conn->send_lock);
	return rc;
}

int smc_cdc_get_slot_and_msg_send(struct smc_connection *conn)
{
	int rc;

	if (!conn->lgr || (conn->lgr->is_smcd && conn->lgr->peer_shutdown))
		return -EPIPE;

	if (conn->lgr->is_smcd) {
		spin_lock_bh(&conn->send_lock);
		rc = smcd_cdc_msg_send(conn);
		spin_unlock_bh(&conn->send_lock);
	} else {
		rc = smcr_cdc_get_slot_and_msg_send(conn);
	}

	return rc;
}

static bool smc_cdc_tx_filter(struct smc_wr_tx_pend_priv *tx_pend,
			      unsigned long data)
{
	struct smc_connection *conn = (struct smc_connection *)data;
	struct smc_cdc_tx_pend *cdc_pend =
		(struct smc_cdc_tx_pend *)tx_pend;

	return cdc_pend->conn == conn;
}

static void smc_cdc_tx_dismisser(struct smc_wr_tx_pend_priv *tx_pend)
{
	struct smc_cdc_tx_pend *cdc_pend =
		(struct smc_cdc_tx_pend *)tx_pend;

	cdc_pend->conn = NULL;
}

void smc_cdc_tx_dismiss_slots(struct smc_connection *conn)
{
	struct smc_link *link = conn->lnk;

	smc_wr_tx_dismiss_slots(link, SMC_CDC_MSG_TYPE,
				smc_cdc_tx_filter, smc_cdc_tx_dismisser,
				(unsigned long)conn);
}

/* Send a SMC-D CDC header.
 * This increments the free space available in our send buffer.
 * Also update the confirmed receive buffer with what was sent to the peer.
 */
int smcd_cdc_msg_send(struct smc_connection *conn)
{
	struct smc_sock *smc = container_of(conn, struct smc_sock, conn);
	union smc_host_cursor curs;
	struct smcd_cdc_msg cdc;
	int rc, diff;

	memset(&cdc, 0, sizeof(cdc));
	cdc.common.type = SMC_CDC_MSG_TYPE;
	curs.acurs.counter = atomic64_read(&conn->local_tx_ctrl.prod.acurs);
	cdc.prod.wrap = curs.wrap;
	cdc.prod.count = curs.count;
	curs.acurs.counter = atomic64_read(&conn->local_tx_ctrl.cons.acurs);
	cdc.cons.wrap = curs.wrap;
	cdc.cons.count = curs.count;
	cdc.cons.prod_flags = conn->local_tx_ctrl.prod_flags;
	cdc.cons.conn_state_flags = conn->local_tx_ctrl.conn_state_flags;
	rc = smcd_tx_ism_write(conn, &cdc, sizeof(cdc), 0, 1);
	if (rc)
		return rc;
	smc_curs_copy(&conn->rx_curs_confirmed, &curs, conn);
	conn->local_rx_ctrl.prod_flags.cons_curs_upd_req = 0;
	/* Calculate transmitted data and increment free send buffer space */
	diff = smc_curs_diff(conn->sndbuf_desc->len, &conn->tx_curs_fin,
			     &conn->tx_curs_sent);
	/* increased by confirmed number of bytes */
	smp_mb__before_atomic();
	atomic_add(diff, &conn->sndbuf_space);
	/* guarantee 0 <= sndbuf_space <= sndbuf_desc->len */
	smp_mb__after_atomic();
	smc_curs_copy(&conn->tx_curs_fin, &conn->tx_curs_sent, conn);

	smc_tx_sndbuf_nonfull(smc);
	return rc;
}

/********************************* receive ***********************************/

static inline bool smc_cdc_before(u16 seq1, u16 seq2)
{
	return (s16)(seq1 - seq2) < 0;
}

static void smc_cdc_handle_urg_data_arrival(struct smc_sock *smc,
					    int *diff_prod)
{
	struct smc_connection *conn = &smc->conn;
	char *base;

	/* new data included urgent business */
	smc_curs_copy(&conn->urg_curs, &conn->local_rx_ctrl.prod, conn);
	conn->urg_state = SMC_URG_VALID;
	if (!sock_flag(&smc->sk, SOCK_URGINLINE))
		/* we'll skip the urgent byte, so don't account for it */
		(*diff_prod)--;
	base = (char *)conn->rmb_desc->cpu_addr + conn->rx_off;
	if (conn->urg_curs.count)
		conn->urg_rx_byte = *(base + conn->urg_curs.count - 1);
	else
		conn->urg_rx_byte = *(base + conn->rmb_desc->len - 1);
	sk_send_sigurg(&smc->sk);
}

static void smc_cdc_msg_validate(struct smc_sock *smc, struct smc_cdc_msg *cdc,
				 struct smc_link *link)
{
	struct smc_connection *conn = &smc->conn;
	u16 recv_seq = ntohs(cdc->seqno);
	s16 diff;

	/* check that seqnum was seen before */
	diff = conn->local_rx_ctrl.seqno - recv_seq;
	if (diff < 0) { /* diff larger than 0x7fff */
		/* drop connection */
		conn->out_of_sync = 1;	/* prevent any further receives */
		spin_lock_bh(&conn->send_lock);
		conn->local_tx_ctrl.conn_state_flags.peer_conn_abort = 1;
		conn->lnk = link;
		spin_unlock_bh(&conn->send_lock);
		sock_hold(&smc->sk); /* sock_put in abort_work */
		if (!queue_work(smc_close_wq, &conn->abort_work))
			sock_put(&smc->sk);
	}
}

static void smc_cdc_msg_recv_action(struct smc_sock *smc,
				    struct smc_cdc_msg *cdc)
{
	union smc_host_cursor cons_old, prod_old;
	struct smc_connection *conn = &smc->conn;
	int diff_cons, diff_prod;

	smc_curs_copy(&prod_old, &conn->local_rx_ctrl.prod, conn);
	smc_curs_copy(&cons_old, &conn->local_rx_ctrl.cons, conn);
	smc_cdc_msg_to_host(&conn->local_rx_ctrl, cdc, conn);

	diff_cons = smc_curs_diff(conn->peer_rmbe_size, &cons_old,
				  &conn->local_rx_ctrl.cons);
	if (diff_cons) {
		/* peer_rmbe_space is decreased during data transfer with RDMA
		 * write
		 */
		smp_mb__before_atomic();
		atomic_add(diff_cons, &conn->peer_rmbe_space);
		/* guarantee 0 <= peer_rmbe_space <= peer_rmbe_size */
		smp_mb__after_atomic();
	}

	diff_prod = smc_curs_diff(conn->rmb_desc->len, &prod_old,
				  &conn->local_rx_ctrl.prod);
	if (diff_prod) {
		if (conn->local_rx_ctrl.prod_flags.urg_data_present)
			smc_cdc_handle_urg_data_arrival(smc, &diff_prod);
		/* bytes_to_rcv is decreased in smc_recvmsg */
		smp_mb__before_atomic();
		atomic_add(diff_prod, &conn->bytes_to_rcv);
		/* guarantee 0 <= bytes_to_rcv <= rmb_desc->len */
		smp_mb__after_atomic();
		smc->sk.sk_data_ready(&smc->sk);
	} else {
		if (conn->local_rx_ctrl.prod_flags.write_blocked)
			smc->sk.sk_data_ready(&smc->sk);
		if (conn->local_rx_ctrl.prod_flags.urg_data_pending)
			conn->urg_state = SMC_URG_NOTYET;
	}

	/* trigger sndbuf consumer: RDMA write into peer RMBE and CDC */
	if ((diff_cons && smc_tx_prepared_sends(conn)) ||
	    conn->local_rx_ctrl.prod_flags.cons_curs_upd_req ||
	    conn->local_rx_ctrl.prod_flags.urg_data_pending)
		smc_tx_sndbuf_nonempty(conn);

	if (diff_cons && conn->urg_tx_pend &&
	    atomic_read(&conn->peer_rmbe_space) == conn->peer_rmbe_size) {
		/* urg data confirmed by peer, indicate we're ready for more */
		conn->urg_tx_pend = false;
		smc->sk.sk_write_space(&smc->sk);
	}

	if (conn->local_rx_ctrl.conn_state_flags.peer_conn_abort) {
		smc->sk.sk_err = ECONNRESET;
		conn->local_tx_ctrl.conn_state_flags.peer_conn_abort = 1;
	}
	if (smc_cdc_rxed_any_close_or_senddone(conn)) {
		smc->sk.sk_shutdown |= RCV_SHUTDOWN;
		if (smc->clcsock && smc->clcsock->sk)
			smc->clcsock->sk->sk_shutdown |= RCV_SHUTDOWN;
		sock_set_flag(&smc->sk, SOCK_DONE);
		sock_hold(&smc->sk); /* sock_put in close_work */
		if (!queue_work(smc_close_wq, &conn->close_work))
			sock_put(&smc->sk);
	}
}

/* called under tasklet context */
static void smc_cdc_msg_recv(struct smc_sock *smc, struct smc_cdc_msg *cdc)
{
	sock_hold(&smc->sk);
	bh_lock_sock(&smc->sk);
	smc_cdc_msg_recv_action(smc, cdc);
	bh_unlock_sock(&smc->sk);
	sock_put(&smc->sk); /* no free sk in softirq-context */
}

/* Schedule a tasklet for this connection. Triggered from the ISM device IRQ
 * handler to indicate update in the DMBE.
 *
 * Context:
 * - tasklet context
 */
static void smcd_cdc_rx_tsklet(struct tasklet_struct *t)
{
	struct smc_connection *conn = from_tasklet(conn, t, rx_tsklet);
	struct smcd_cdc_msg *data_cdc;
	struct smcd_cdc_msg cdc;
	struct smc_sock *smc;

	if (!conn || conn->killed)
		return;

	data_cdc = (struct smcd_cdc_msg *)conn->rmb_desc->cpu_addr;
	smcd_curs_copy(&cdc.prod, &data_cdc->prod, conn);
	smcd_curs_copy(&cdc.cons, &data_cdc->cons, conn);
	smc = container_of(conn, struct smc_sock, conn);
	smc_cdc_msg_recv(smc, (struct smc_cdc_msg *)&cdc);
}

/* Initialize receive tasklet. Called from ISM device IRQ handler to start
 * receiver side.
 */
void smcd_cdc_rx_init(struct smc_connection *conn)
{
	tasklet_setup(&conn->rx_tsklet, smcd_cdc_rx_tsklet);
}

/***************************** init, exit, misc ******************************/

static void smc_cdc_rx_handler(struct ib_wc *wc, void *buf)
{
	struct smc_link *link = (struct smc_link *)wc->qp->qp_context;
	struct smc_cdc_msg *cdc = buf;
	struct smc_connection *conn;
	struct smc_link_group *lgr;
	struct smc_sock *smc;

	if (wc->byte_len < offsetof(struct smc_cdc_msg, reserved))
		return; /* short message */
	if (cdc->len != SMC_WR_TX_SIZE)
		return; /* invalid message */

	/* lookup connection */
	lgr = smc_get_lgr(link);
	read_lock_bh(&lgr->conns_lock);
	conn = smc_lgr_find_conn(ntohl(cdc->token), lgr);
	read_unlock_bh(&lgr->conns_lock);
	if (!conn || conn->out_of_sync)
		return;
	smc = container_of(conn, struct smc_sock, conn);

	if (cdc->prod_flags.failover_validation) {
		smc_cdc_msg_validate(smc, cdc, link);
		return;
	}
	if (smc_cdc_before(ntohs(cdc->seqno),
			   conn->local_rx_ctrl.seqno))
		/* received seqno is old */
		return;

	smc_cdc_msg_recv(smc, cdc);
}

static struct smc_wr_rx_handler smc_cdc_rx_handlers[] = {
	{
		.handler	= smc_cdc_rx_handler,
		.type		= SMC_CDC_MSG_TYPE
	},
	{
		.handler	= NULL,
	}
};

int __init smc_cdc_init(void)
{
	struct smc_wr_rx_handler *handler;
	int rc = 0;

	for (handler = smc_cdc_rx_handlers; handler->handler; handler++) {
		INIT_HLIST_NODE(&handler->list);
		rc = smc_wr_rx_register_handler(handler);
		if (rc)
			break;
	}
	return rc;
}
Commit	Line	Data
b2441318	1	// SPDX-License-Identifier: GPL-2.0
5f08318f UB	2	/*
	3	* Shared Memory Communications over RDMA (SMC-R) and RoCE
	4	*
	5	* Connection Data Control (CDC)
	6	* handles flow control
	7	*
	8	* Copyright IBM Corp. 2016
	9	*
	10	* Author(s): Ursula Braun <[email protected]>
	11	*/
	12
	13	#include <linux/spinlock.h>
	14
	15	#include "smc.h"
	16	#include "smc_wr.h"
	17	#include "smc_cdc.h"
e6727f39	18	#include "smc_tx.h"
952310cc	19	#include "smc_rx.h"
b38d7324	20	#include "smc_close.h"
5f08318f UB	21
	22	/******************************** send ***********************************/
	23
5f08318f UB	24	/* handler for send/transmission completion of a CDC msg */
	25	static void smc_cdc_tx_handler(struct smc_wr_tx_pend_priv *pnd_snd,
	26	struct smc_link *link,
	27	enum ib_wc_status wc_status)
	28	{
	29	struct smc_cdc_tx_pend cdcpend = (struct smc_cdc_tx_pend )pnd_snd;
bac6de7b	30	struct smc_connection *conn = cdcpend->conn;
5f08318f UB	31	struct smc_sock *smc;
	32	int diff;
	33
bac6de7b	34	if (!conn)
5f08318f UB	35	/* already dismissed */
	36	return;
	37
bac6de7b	38	smc = container_of(conn, struct smc_sock, conn);
5f08318f UB	39	bh_lock_sock(&smc->sk);
5f08318f UB	40	if (!wc_status) {
69cb7dc0	41	diff = smc_curs_diff(cdcpend->conn->sndbuf_desc->len,
5f08318f UB	42	&cdcpend->conn->tx_curs_fin,
	43	&cdcpend->cursor);
	44	/* sndbuf_space is decreased in smc_sendmsg */
	45	smp_mb__before_atomic();
	46	atomic_add(diff, &cdcpend->conn->sndbuf_space);
69cb7dc0	47	/* guarantee 0 <= sndbuf_space <= sndbuf_desc->len */
5f08318f	48	smp_mb__after_atomic();
bac6de7b	49	smc_curs_copy(&conn->tx_curs_fin, &cdcpend->cursor, conn);
f0ec4f1d KG	50	smc_curs_copy(&conn->local_tx_ctrl_fin, &cdcpend->p_cursor,
	51	conn);
	52	conn->tx_cdc_seq_fin = cdcpend->ctrl_seq;
5f08318f	53	}
e6727f39	54	smc_tx_sndbuf_nonfull(smc);
5f08318f UB	55	bh_unlock_sock(&smc->sk);
	56	}
	57
51957bc5	58	int smc_cdc_get_free_slot(struct smc_connection *conn,
c6f02ebe	59	struct smc_link *link,
5f08318f	60	struct smc_wr_buf **wr_buf,
ad6f317f	61	struct smc_rdma_wr **wr_rdma_buf,
5f08318f UB	62	struct smc_cdc_tx_pend **pend)
5f08318f UB	63	{
1a0a04c7	64	int rc;
51957bc5	65
1a0a04c7	66	rc = smc_wr_tx_get_free_slot(link, smc_cdc_tx_handler, wr_buf,
ad6f317f	67	wr_rdma_buf,
1a0a04c7	68	(struct smc_wr_tx_pend_priv **)pend);
2bced6ae	69	if (conn->killed) {
1a0a04c7	70	/* abnormal termination */
2bced6ae KG	71	if (!rc)
	72	smc_wr_tx_put_slot(link,
	73	(struct smc_wr_tx_pend_priv *)pend);
1a0a04c7	74	rc = -EPIPE;
2bced6ae	75	}
1a0a04c7	76	return rc;
5f08318f UB	77	}
	78
	79	static inline void smc_cdc_add_pending_send(struct smc_connection *conn,
	80	struct smc_cdc_tx_pend *pend)
	81	{
	82	BUILD_BUG_ON_MSG(
	83	sizeof(struct smc_cdc_msg) > SMC_WR_BUF_SIZE,
	84	"must increase SMC_WR_BUF_SIZE to at least sizeof(struct smc_cdc_msg)");
	85	BUILD_BUG_ON_MSG(
b9a22dd9	86	offsetofend(struct smc_cdc_msg, reserved) > SMC_WR_TX_SIZE,
5f08318f UB	87	"must adapt SMC_WR_TX_SIZE to sizeof(struct smc_cdc_msg); if not all smc_wr upper layer protocols use the same message size any more, must start to set link->wr_tx_sges[i].length on each individual smc_wr_tx_send()");
	88	BUILD_BUG_ON_MSG(
	89	sizeof(struct smc_cdc_tx_pend) > SMC_WR_TX_PEND_PRIV_SIZE,
	90	"must increase SMC_WR_TX_PEND_PRIV_SIZE to at least sizeof(struct smc_cdc_tx_pend)");
	91	pend->conn = conn;
	92	pend->cursor = conn->tx_curs_sent;
	93	pend->p_cursor = conn->local_tx_ctrl.prod;
	94	pend->ctrl_seq = conn->tx_cdc_seq;
	95	}
	96
	97	int smc_cdc_msg_send(struct smc_connection *conn,
	98	struct smc_wr_buf *wr_buf,
	99	struct smc_cdc_tx_pend *pend)
	100	{
387707fd	101	struct smc_link *link = conn->lnk;
b8649efa	102	union smc_host_cursor cfed;
5f08318f UB	103	int rc;
5f08318f UB	104
5f08318f UB	105	smc_cdc_add_pending_send(conn, pend);
	106
	107	conn->tx_cdc_seq++;
	108	conn->local_tx_ctrl.seqno = conn->tx_cdc_seq;
ccc8ca9b	109	smc_host_msg_to_cdc((struct smc_cdc_msg *)wr_buf, conn, &cfed);
5f08318f	110	rc = smc_wr_tx_send(link, (struct smc_wr_tx_pend_priv *)pend);
4dff63c2	111	if (!rc) {
b8649efa	112	smc_curs_copy(&conn->rx_curs_confirmed, &cfed, conn);
4dff63c2	113	conn->local_rx_ctrl.prod_flags.cons_curs_upd_req = 0;
f0ec4f1d KG	114	} else {
	115	conn->tx_cdc_seq--;
	116	conn->local_tx_ctrl.seqno = conn->tx_cdc_seq;
4dff63c2	117	}
5f08318f UB	118
	119	return rc;
	120	}
	121
29bd73db	122	/* send a validation msg indicating the move of a conn to an other QP link */
b8ded9de KG	123	int smcr_cdc_msg_send_validation(struct smc_connection *conn,
	124	struct smc_cdc_tx_pend *pend,
	125	struct smc_wr_buf *wr_buf)
29bd73db KG	126	{
	127	struct smc_host_cdc_msg *local = &conn->local_tx_ctrl;
	128	struct smc_link *link = conn->lnk;
29bd73db KG	129	struct smc_cdc_msg *peer;
	130	int rc;
	131
29bd73db KG	132	peer = (struct smc_cdc_msg *)wr_buf;
	133	peer->common.type = local->common.type;
	134	peer->len = local->len;
	135	peer->seqno = htons(conn->tx_cdc_seq_fin); /* seqno last compl. tx */
	136	peer->token = htonl(local->token);
	137	peer->prod_flags.failover_validation = 1;
	138
	139	rc = smc_wr_tx_send(link, (struct smc_wr_tx_pend_priv *)pend);
	140	return rc;
	141	}
	142
be244f28	143	static int smcr_cdc_get_slot_and_msg_send(struct smc_connection *conn)
5f08318f UB	144	{
	145	struct smc_cdc_tx_pend *pend;
	146	struct smc_wr_buf *wr_buf;
c6f02ebe KG	147	struct smc_link *link;
c6f02ebe KG	148	bool again = false;
5f08318f UB	149	int rc;
5f08318f UB	150
c6f02ebe KG	151	again:
	152	link = conn->lnk;
	153	rc = smc_cdc_get_free_slot(conn, link, &wr_buf, NULL, &pend);
5f08318f UB	154	if (rc)
	155	return rc;
	156
2dee25af	157	spin_lock_bh(&conn->send_lock);
c6f02ebe KG	158	if (link != conn->lnk) {
	159	/* link of connection changed, try again one time*/
	160	spin_unlock_bh(&conn->send_lock);
	161	smc_wr_tx_put_slot(link,
	162	(struct smc_wr_tx_pend_priv *)pend);
	163	if (again)
	164	return -ENOLINK;
	165	again = true;
	166	goto again;
	167	}
2dee25af KG	168	rc = smc_cdc_msg_send(conn, wr_buf, pend);
	169	spin_unlock_bh(&conn->send_lock);
	170	return rc;
5f08318f UB	171	}
5f08318f UB	172
be244f28 HW	173	int smc_cdc_get_slot_and_msg_send(struct smc_connection *conn)
	174	{
	175	int rc;
	176
50c6b20e UB	177	if (!conn->lgr \|\| (conn->lgr->is_smcd && conn->lgr->peer_shutdown))
	178	return -EPIPE;
	179
be244f28 HW	180	if (conn->lgr->is_smcd) {
	181	spin_lock_bh(&conn->send_lock);
	182	rc = smcd_cdc_msg_send(conn);
	183	spin_unlock_bh(&conn->send_lock);
	184	} else {
	185	rc = smcr_cdc_get_slot_and_msg_send(conn);
	186	}
	187
	188	return rc;
	189	}
	190
5f08318f UB	191	static bool smc_cdc_tx_filter(struct smc_wr_tx_pend_priv *tx_pend,
	192	unsigned long data)
	193	{
	194	struct smc_connection conn = (struct smc_connection )data;
	195	struct smc_cdc_tx_pend *cdc_pend =
	196	(struct smc_cdc_tx_pend *)tx_pend;
	197
	198	return cdc_pend->conn == conn;
	199	}
	200
	201	static void smc_cdc_tx_dismisser(struct smc_wr_tx_pend_priv *tx_pend)
	202	{
	203	struct smc_cdc_tx_pend *cdc_pend =
	204	(struct smc_cdc_tx_pend *)tx_pend;
	205
	206	cdc_pend->conn = NULL;
	207	}
	208
	209	void smc_cdc_tx_dismiss_slots(struct smc_connection *conn)
	210	{
387707fd	211	struct smc_link *link = conn->lnk;
5f08318f UB	212
	213	smc_wr_tx_dismiss_slots(link, SMC_CDC_MSG_TYPE,
	214	smc_cdc_tx_filter, smc_cdc_tx_dismisser,
	215	(unsigned long)conn);
	216	}
	217
be244f28 HW	218	/* Send a SMC-D CDC header.
	219	* This increments the free space available in our send buffer.
	220	* Also update the confirmed receive buffer with what was sent to the peer.
	221	*/
	222	int smcd_cdc_msg_send(struct smc_connection *conn)
	223	{
	224	struct smc_sock *smc = container_of(conn, struct smc_sock, conn);
b9a22dd9	225	union smc_host_cursor curs;
be244f28 HW	226	struct smcd_cdc_msg cdc;
	227	int rc, diff;
	228
	229	memset(&cdc, 0, sizeof(cdc));
	230	cdc.common.type = SMC_CDC_MSG_TYPE;
b9a22dd9 UB	231	curs.acurs.counter = atomic64_read(&conn->local_tx_ctrl.prod.acurs);
	232	cdc.prod.wrap = curs.wrap;
	233	cdc.prod.count = curs.count;
	234	curs.acurs.counter = atomic64_read(&conn->local_tx_ctrl.cons.acurs);
	235	cdc.cons.wrap = curs.wrap;
	236	cdc.cons.count = curs.count;
	237	cdc.cons.prod_flags = conn->local_tx_ctrl.prod_flags;
	238	cdc.cons.conn_state_flags = conn->local_tx_ctrl.conn_state_flags;
be244f28 HW	239	rc = smcd_tx_ism_write(conn, &cdc, sizeof(cdc), 0, 1);
	240	if (rc)
	241	return rc;
b9a22dd9	242	smc_curs_copy(&conn->rx_curs_confirmed, &curs, conn);
4dff63c2	243	conn->local_rx_ctrl.prod_flags.cons_curs_upd_req = 0;
be244f28 HW	244	/* Calculate transmitted data and increment free send buffer space */
	245	diff = smc_curs_diff(conn->sndbuf_desc->len, &conn->tx_curs_fin,
	246	&conn->tx_curs_sent);
	247	/* increased by confirmed number of bytes */
	248	smp_mb__before_atomic();
	249	atomic_add(diff, &conn->sndbuf_space);
	250	/* guarantee 0 <= sndbuf_space <= sndbuf_desc->len */
	251	smp_mb__after_atomic();
bac6de7b	252	smc_curs_copy(&conn->tx_curs_fin, &conn->tx_curs_sent, conn);
be244f28 HW	253
	254	smc_tx_sndbuf_nonfull(smc);
	255	return rc;
	256	}
	257
5f08318f UB	258	/******************************* receive *********************************/
	259
	260	static inline bool smc_cdc_before(u16 seq1, u16 seq2)
	261	{
	262	return (s16)(seq1 - seq2) < 0;
	263	}
	264
de8474eb SR	265	static void smc_cdc_handle_urg_data_arrival(struct smc_sock *smc,
	266	int *diff_prod)
	267	{
	268	struct smc_connection *conn = &smc->conn;
	269	char *base;
	270
	271	/* new data included urgent business */
bac6de7b	272	smc_curs_copy(&conn->urg_curs, &conn->local_rx_ctrl.prod, conn);
de8474eb SR	273	conn->urg_state = SMC_URG_VALID;
	274	if (!sock_flag(&smc->sk, SOCK_URGINLINE))
	275	/* we'll skip the urgent byte, so don't account for it */
	276	(*diff_prod)--;
be244f28	277	base = (char *)conn->rmb_desc->cpu_addr + conn->rx_off;
de8474eb SR	278	if (conn->urg_curs.count)
	279	conn->urg_rx_byte = *(base + conn->urg_curs.count - 1);
	280	else
	281	conn->urg_rx_byte = *(base + conn->rmb_desc->len - 1);
	282	sk_send_sigurg(&smc->sk);
	283	}
	284
b286a065 KG	285	static void smc_cdc_msg_validate(struct smc_sock smc, struct smc_cdc_msg cdc,
	286	struct smc_link *link)
	287	{
	288	struct smc_connection *conn = &smc->conn;
	289	u16 recv_seq = ntohs(cdc->seqno);
	290	s16 diff;
	291
	292	/* check that seqnum was seen before */
	293	diff = conn->local_rx_ctrl.seqno - recv_seq;
	294	if (diff < 0) { /* diff larger than 0x7fff */
	295	/* drop connection */
	296	conn->out_of_sync = 1; /* prevent any further receives */
	297	spin_lock_bh(&conn->send_lock);
	298	conn->local_tx_ctrl.conn_state_flags.peer_conn_abort = 1;
	299	conn->lnk = link;
	300	spin_unlock_bh(&conn->send_lock);
	301	sock_hold(&smc->sk); /* sock_put in abort_work */
22ef473d	302	if (!queue_work(smc_close_wq, &conn->abort_work))
b286a065 KG	303	sock_put(&smc->sk);
	304	}
	305	}
	306
5f08318f	307	static void smc_cdc_msg_recv_action(struct smc_sock *smc,
5f08318f UB	308	struct smc_cdc_msg *cdc)
	309	{
	310	union smc_host_cursor cons_old, prod_old;
	311	struct smc_connection *conn = &smc->conn;
	312	int diff_cons, diff_prod;
	313
bac6de7b SR	314	smc_curs_copy(&prod_old, &conn->local_rx_ctrl.prod, conn);
bac6de7b SR	315	smc_curs_copy(&cons_old, &conn->local_rx_ctrl.cons, conn);
5f08318f UB	316	smc_cdc_msg_to_host(&conn->local_rx_ctrl, cdc, conn);
	317
	318	diff_cons = smc_curs_diff(conn->peer_rmbe_size, &cons_old,
	319	&conn->local_rx_ctrl.cons);
	320	if (diff_cons) {
	321	/* peer_rmbe_space is decreased during data transfer with RDMA
	322	* write
	323	*/
	324	smp_mb__before_atomic();
	325	atomic_add(diff_cons, &conn->peer_rmbe_space);
	326	/* guarantee 0 <= peer_rmbe_space <= peer_rmbe_size */
	327	smp_mb__after_atomic();
	328	}
	329
69cb7dc0	330	diff_prod = smc_curs_diff(conn->rmb_desc->len, &prod_old,
5f08318f UB	331	&conn->local_rx_ctrl.prod);
5f08318f UB	332	if (diff_prod) {
de8474eb SR	333	if (conn->local_rx_ctrl.prod_flags.urg_data_present)
de8474eb SR	334	smc_cdc_handle_urg_data_arrival(smc, &diff_prod);
5f08318f UB	335	/* bytes_to_rcv is decreased in smc_recvmsg */
	336	smp_mb__before_atomic();
	337	atomic_add(diff_prod, &conn->bytes_to_rcv);
69cb7dc0	338	/* guarantee 0 <= bytes_to_rcv <= rmb_desc->len */
5f08318f	339	smp_mb__after_atomic();
952310cc	340	smc->sk.sk_data_ready(&smc->sk);
de8474eb	341	} else {
cf0cfe53 KG	342	if (conn->local_rx_ctrl.prod_flags.write_blocked)
	343	smc->sk.sk_data_ready(&smc->sk);
	344	if (conn->local_rx_ctrl.prod_flags.urg_data_pending)
	345	conn->urg_state = SMC_URG_NOTYET;
5f08318f UB	346	}
5f08318f UB	347
51f1de79	348	/* trigger sndbuf consumer: RDMA write into peer RMBE and CDC */
cf0cfe53 KG	349	if ((diff_cons && smc_tx_prepared_sends(conn)) \|\|
	350	conn->local_rx_ctrl.prod_flags.cons_curs_upd_req \|\|
	351	conn->local_rx_ctrl.prod_flags.urg_data_pending)
51f1de79	352	smc_tx_sndbuf_nonempty(conn);
cf0cfe53	353
de8474eb SR	354	if (diff_cons && conn->urg_tx_pend &&
	355	atomic_read(&conn->peer_rmbe_space) == conn->peer_rmbe_size) {
	356	/* urg data confirmed by peer, indicate we're ready for more */
	357	conn->urg_tx_pend = false;
	358	smc->sk.sk_write_space(&smc->sk);
	359	}
51f1de79	360
b38d7324	361	if (conn->local_rx_ctrl.conn_state_flags.peer_conn_abort) {
5f08318f	362	smc->sk.sk_err = ECONNRESET;
b38d7324 UB	363	conn->local_tx_ctrl.conn_state_flags.peer_conn_abort = 1;
b38d7324 UB	364	}
46c28dbd UB	365	if (smc_cdc_rxed_any_close_or_senddone(conn)) {
	366	smc->sk.sk_shutdown \|= RCV_SHUTDOWN;
	367	if (smc->clcsock && smc->clcsock->sk)
	368	smc->clcsock->sk->sk_shutdown \|= RCV_SHUTDOWN;
	369	sock_set_flag(&smc->sk, SOCK_DONE);
51f1de79	370	sock_hold(&smc->sk); /* sock_put in close_work */
22ef473d	371	if (!queue_work(smc_close_wq, &conn->close_work))
51f1de79	372	sock_put(&smc->sk);
b38d7324	373	}
5f08318f UB	374	}
	375
	376	/* called under tasklet context */
d7b0e37c	377	static void smc_cdc_msg_recv(struct smc_sock smc, struct smc_cdc_msg cdc)
5f08318f	378	{
5f08318f	379	sock_hold(&smc->sk);
5f08318f	380	bh_lock_sock(&smc->sk);
d7b0e37c	381	smc_cdc_msg_recv_action(smc, cdc);
5f08318f UB	382	bh_unlock_sock(&smc->sk);
	383	sock_put(&smc->sk); /* no free sk in softirq-context */
	384	}
	385
be244f28 HW	386	/* Schedule a tasklet for this connection. Triggered from the ISM device IRQ
	387	* handler to indicate update in the DMBE.
	388	*
	389	* Context:
	390	* - tasklet context
	391	*/
fcb8e3a3	392	static void smcd_cdc_rx_tsklet(struct tasklet_struct *t)
be244f28	393	{
fcb8e3a3	394	struct smc_connection *conn = from_tasklet(conn, t, rx_tsklet);
b9a22dd9	395	struct smcd_cdc_msg *data_cdc;
be244f28 HW	396	struct smcd_cdc_msg cdc;
	397	struct smc_sock *smc;
	398
b2900980	399	if (!conn \|\| conn->killed)
be244f28 HW	400	return;
be244f28 HW	401
b9a22dd9 UB	402	data_cdc = (struct smcd_cdc_msg *)conn->rmb_desc->cpu_addr;
	403	smcd_curs_copy(&cdc.prod, &data_cdc->prod, conn);
	404	smcd_curs_copy(&cdc.cons, &data_cdc->cons, conn);
be244f28 HW	405	smc = container_of(conn, struct smc_sock, conn);
	406	smc_cdc_msg_recv(smc, (struct smc_cdc_msg *)&cdc);
	407	}
	408
	409	/* Initialize receive tasklet. Called from ISM device IRQ handler to start
	410	* receiver side.
	411	*/
	412	void smcd_cdc_rx_init(struct smc_connection *conn)
	413	{
fcb8e3a3	414	tasklet_setup(&conn->rx_tsklet, smcd_cdc_rx_tsklet);
be244f28 HW	415	}
be244f28 HW	416
5f08318f UB	417	/*************************** init, exit, misc ****************************/
	418
	419	static void smc_cdc_rx_handler(struct ib_wc wc, void buf)
	420	{
	421	struct smc_link link = (struct smc_link )wc->qp->qp_context;
	422	struct smc_cdc_msg *cdc = buf;
d7b0e37c HW	423	struct smc_connection *conn;
	424	struct smc_link_group *lgr;
	425	struct smc_sock *smc;
5f08318f UB	426
	427	if (wc->byte_len < offsetof(struct smc_cdc_msg, reserved))
	428	return; /* short message */
cbba07a7	429	if (cdc->len != SMC_WR_TX_SIZE)
5f08318f	430	return; /* invalid message */
d7b0e37c HW	431
d7b0e37c HW	432	/* lookup connection */
00e5fb26	433	lgr = smc_get_lgr(link);
d7b0e37c HW	434	read_lock_bh(&lgr->conns_lock);
	435	conn = smc_lgr_find_conn(ntohl(cdc->token), lgr);
	436	read_unlock_bh(&lgr->conns_lock);
b286a065	437	if (!conn \|\| conn->out_of_sync)
d7b0e37c HW	438	return;
	439	smc = container_of(conn, struct smc_sock, conn);
	440
b286a065 KG	441	if (cdc->prod_flags.failover_validation) {
	442	smc_cdc_msg_validate(smc, cdc, link);
	443	return;
d7b0e37c	444	}
b286a065 KG	445	if (smc_cdc_before(ntohs(cdc->seqno),
	446	conn->local_rx_ctrl.seqno))
	447	/* received seqno is old */
	448	return;
	449
d7b0e37c	450	smc_cdc_msg_recv(smc, cdc);
5f08318f UB	451	}
	452
	453	static struct smc_wr_rx_handler smc_cdc_rx_handlers[] = {
	454	{
	455	.handler = smc_cdc_rx_handler,
	456	.type = SMC_CDC_MSG_TYPE
	457	},
	458	{
	459	.handler = NULL,
	460	}
	461	};
	462
	463	int __init smc_cdc_init(void)
	464	{
	465	struct smc_wr_rx_handler *handler;
	466	int rc = 0;
	467
	468	for (handler = smc_cdc_rx_handlers; handler->handler; handler++) {
	469	INIT_HLIST_NODE(&handler->list);
	470	rc = smc_wr_rx_register_handler(handler);
	471	if (rc)
	472	break;
	473	}
	474	return rc;
	475	}