[linux-2.6-block.git] / net / smc / smc_core.c

/*
 *  Shared Memory Communications over RDMA (SMC-R) and RoCE
 *
 *  Basic Transport Functions exploiting Infiniband API
 *
 *  Copyright IBM Corp. 2016
 *
 *  Author(s):  Ursula Braun <ubraun@linux.vnet.ibm.com>
 */

#include <linux/socket.h>
#include <linux/if_vlan.h>
#include <linux/random.h>
#include <linux/workqueue.h>
#include <net/tcp.h>
#include <net/sock.h>
#include <rdma/ib_verbs.h>

#include "smc.h"
#include "smc_clc.h"
#include "smc_core.h"
#include "smc_ib.h"
#include "smc_wr.h"
#include "smc_llc.h"
#include "smc_cdc.h"

#define SMC_LGR_NUM_INCR	256
#define SMC_LGR_FREE_DELAY	(600 * HZ)

static u32 smc_lgr_num;			/* unique link group number */

/* Register connection's alert token in our lookup structure.
 * To use rbtrees we have to implement our own insert core.
 * Requires @conns_lock
 * @smc		connection to register
 * Returns 0 on success, != otherwise.
 */
static void smc_lgr_add_alert_token(struct smc_connection *conn)
{
	struct rb_node **link, *parent = NULL;
	u32 token = conn->alert_token_local;

	link = &conn->lgr->conns_all.rb_node;
	while (*link) {
		struct smc_connection *cur = rb_entry(*link,
					struct smc_connection, alert_node);

		parent = *link;
		if (cur->alert_token_local > token)
			link = &parent->rb_left;
		else
			link = &parent->rb_right;
	}
	/* Put the new node there */
	rb_link_node(&conn->alert_node, parent, link);
	rb_insert_color(&conn->alert_node, &conn->lgr->conns_all);
}

/* Register connection in link group by assigning an alert token
 * registered in a search tree.
 * Requires @conns_lock
 * Note that '0' is a reserved value and not assigned.
 */
static void smc_lgr_register_conn(struct smc_connection *conn)
{
	struct smc_sock *smc = container_of(conn, struct smc_sock, conn);
	static atomic_t nexttoken = ATOMIC_INIT(0);

	/* find a new alert_token_local value not yet used by some connection
	 * in this link group
	 */
	sock_hold(&smc->sk); /* sock_put in smc_lgr_unregister_conn() */
	while (!conn->alert_token_local) {
		conn->alert_token_local = atomic_inc_return(&nexttoken);
		if (smc_lgr_find_conn(conn->alert_token_local, conn->lgr))
			conn->alert_token_local = 0;
	}
	smc_lgr_add_alert_token(conn);
	conn->lgr->conns_num++;
}

/* Unregister connection and reset the alert token of the given connection<
 */
static void __smc_lgr_unregister_conn(struct smc_connection *conn)
{
	struct smc_sock *smc = container_of(conn, struct smc_sock, conn);
	struct smc_link_group *lgr = conn->lgr;

	rb_erase(&conn->alert_node, &lgr->conns_all);
	lgr->conns_num--;
	conn->alert_token_local = 0;
	conn->lgr = NULL;
	sock_put(&smc->sk); /* sock_hold in smc_lgr_register_conn() */
}

/* Unregister connection and trigger lgr freeing if applicable
 */
static void smc_lgr_unregister_conn(struct smc_connection *conn)
{
	struct smc_link_group *lgr = conn->lgr;
	int reduced = 0;

	write_lock_bh(&lgr->conns_lock);
	if (conn->alert_token_local) {
		reduced = 1;
		__smc_lgr_unregister_conn(conn);
	}
	write_unlock_bh(&lgr->conns_lock);
	if (reduced && !lgr->conns_num)
		schedule_delayed_work(&lgr->free_work, SMC_LGR_FREE_DELAY);
}

static void smc_lgr_free_work(struct work_struct *work)
{
	struct smc_link_group *lgr = container_of(to_delayed_work(work),
						  struct smc_link_group,
						  free_work);
	bool conns;

	spin_lock_bh(&smc_lgr_list.lock);
	read_lock_bh(&lgr->conns_lock);
	conns = RB_EMPTY_ROOT(&lgr->conns_all);
	read_unlock_bh(&lgr->conns_lock);
	if (!conns) { /* number of lgr connections is no longer zero */
		spin_unlock_bh(&smc_lgr_list.lock);
		return;
	}
	list_del_init(&lgr->list); /* remove from smc_lgr_list */
	spin_unlock_bh(&smc_lgr_list.lock);
	smc_lgr_free(lgr);
}

/* create a new SMC link group */
static int smc_lgr_create(struct smc_sock *smc, __be32 peer_in_addr,
			  struct smc_ib_device *smcibdev, u8 ibport,
			  char *peer_systemid, unsigned short vlan_id)
{
	struct smc_link_group *lgr;
	struct smc_link *lnk;
	u8 rndvec[3];
	int rc = 0;
	int i;

	lgr = kzalloc(sizeof(*lgr), GFP_KERNEL);
	if (!lgr) {
		rc = -ENOMEM;
		goto out;
	}
	lgr->role = smc->listen_smc ? SMC_SERV : SMC_CLNT;
	lgr->sync_err = false;
	lgr->daddr = peer_in_addr;
	memcpy(lgr->peer_systemid, peer_systemid, SMC_SYSTEMID_LEN);
	lgr->vlan_id = vlan_id;
	rwlock_init(&lgr->sndbufs_lock);
	rwlock_init(&lgr->rmbs_lock);
	for (i = 0; i < SMC_RMBE_SIZES; i++) {
		INIT_LIST_HEAD(&lgr->sndbufs[i]);
		INIT_LIST_HEAD(&lgr->rmbs[i]);
	}
	smc_lgr_num += SMC_LGR_NUM_INCR;
	memcpy(&lgr->id, (u8 *)&smc_lgr_num, SMC_LGR_ID_SIZE);
	INIT_DELAYED_WORK(&lgr->free_work, smc_lgr_free_work);
	lgr->conns_all = RB_ROOT;

	lnk = &lgr->lnk[SMC_SINGLE_LINK];
	/* initialize link */
	lnk->smcibdev = smcibdev;
	lnk->ibport = ibport;
	lnk->path_mtu = smcibdev->pattr[ibport - 1].active_mtu;
	if (!smcibdev->initialized)
		smc_ib_setup_per_ibdev(smcibdev);
	get_random_bytes(rndvec, sizeof(rndvec));
	lnk->psn_initial = rndvec[0] + (rndvec[1] << 8) + (rndvec[2] << 16);
	rc = smc_wr_alloc_link_mem(lnk);
	if (rc)
		goto free_lgr;
	init_waitqueue_head(&lnk->wr_tx_wait);
	rc = smc_ib_create_protection_domain(lnk);
	if (rc)
		goto free_link_mem;
	rc = smc_ib_create_queue_pair(lnk);
	if (rc)
		goto dealloc_pd;
	rc = smc_wr_create_link(lnk);
	if (rc)
		goto destroy_qp;
	init_completion(&lnk->llc_confirm);
	init_completion(&lnk->llc_confirm_resp);

	smc->conn.lgr = lgr;
	rwlock_init(&lgr->conns_lock);
	spin_lock_bh(&smc_lgr_list.lock);
	list_add(&lgr->list, &smc_lgr_list.list);
	spin_unlock_bh(&smc_lgr_list.lock);
	return 0;

destroy_qp:
	smc_ib_destroy_queue_pair(lnk);
dealloc_pd:
	smc_ib_dealloc_protection_domain(lnk);
free_link_mem:
	smc_wr_free_link_mem(lnk);
free_lgr:
	kfree(lgr);
out:
	return rc;
}

static void smc_sndbuf_unuse(struct smc_connection *conn)
{
	if (conn->sndbuf_desc) {
		conn->sndbuf_desc->used = 0;
		conn->sndbuf_size = 0;
	}
}

static void smc_rmb_unuse(struct smc_connection *conn)
{
	if (conn->rmb_desc) {
		conn->rmb_desc->used = 0;
		conn->rmbe_size = 0;
	}
}

/* remove a finished connection from its link group */
void smc_conn_free(struct smc_connection *conn)
{
	struct smc_link_group *lgr = conn->lgr;

	if (!lgr)
		return;
	smc_cdc_tx_dismiss_slots(conn);
	smc_lgr_unregister_conn(conn);
	smc_rmb_unuse(conn);
	smc_sndbuf_unuse(conn);
}

static void smc_link_clear(struct smc_link *lnk)
{
	lnk->peer_qpn = 0;
	smc_ib_modify_qp_reset(lnk);
	smc_wr_free_link(lnk);
	smc_ib_destroy_queue_pair(lnk);
	smc_ib_dealloc_protection_domain(lnk);
	smc_wr_free_link_mem(lnk);
}

static void smc_lgr_free_sndbufs(struct smc_link_group *lgr)
{
	struct smc_buf_desc *sndbuf_desc, *bf_desc;
	int i;

	for (i = 0; i < SMC_RMBE_SIZES; i++) {
		list_for_each_entry_safe(sndbuf_desc, bf_desc, &lgr->sndbufs[i],
					 list) {
			list_del(&sndbuf_desc->list);
			smc_ib_buf_unmap(lgr->lnk[SMC_SINGLE_LINK].smcibdev,
					 smc_uncompress_bufsize(i),
					 sndbuf_desc, DMA_TO_DEVICE);
			kfree(sndbuf_desc->cpu_addr);
			kfree(sndbuf_desc);
		}
	}
}

static void smc_lgr_free_rmbs(struct smc_link_group *lgr)
{
	struct smc_buf_desc *rmb_desc, *bf_desc;
	struct smc_link *lnk = &lgr->lnk[SMC_SINGLE_LINK];
	int i;

	for (i = 0; i < SMC_RMBE_SIZES; i++) {
		list_for_each_entry_safe(rmb_desc, bf_desc, &lgr->rmbs[i],
					 list) {
			list_del(&rmb_desc->list);
			smc_ib_buf_unmap(lnk->smcibdev,
					 smc_uncompress_bufsize(i),
					 rmb_desc, DMA_FROM_DEVICE);
			kfree(rmb_desc->cpu_addr);
			kfree(rmb_desc);
		}
	}
}

/* remove a link group */
void smc_lgr_free(struct smc_link_group *lgr)
{
	smc_lgr_free_rmbs(lgr);
	smc_lgr_free_sndbufs(lgr);
	smc_link_clear(&lgr->lnk[SMC_SINGLE_LINK]);
	kfree(lgr);
}

/* terminate linkgroup abnormally */
void smc_lgr_terminate(struct smc_link_group *lgr)
{
	struct smc_connection *conn;
	struct rb_node *node;

	spin_lock_bh(&smc_lgr_list.lock);
	if (list_empty(&lgr->list)) {
		/* termination already triggered */
		spin_unlock_bh(&smc_lgr_list.lock);
		return;
	}
	/* do not use this link group for new connections */
	list_del_init(&lgr->list);
	spin_unlock_bh(&smc_lgr_list.lock);

	write_lock_bh(&lgr->conns_lock);
	node = rb_first(&lgr->conns_all);
	while (node) {
		conn = rb_entry(node, struct smc_connection, alert_node);
		__smc_lgr_unregister_conn(conn);
		node = rb_first(&lgr->conns_all);
	}
	write_unlock_bh(&lgr->conns_lock);
	schedule_delayed_work(&lgr->free_work, SMC_LGR_FREE_DELAY);
}

/* Determine vlan of internal TCP socket.
 * @vlan_id: address to store the determined vlan id into
 */
static int smc_vlan_by_tcpsk(struct socket *clcsock, unsigned short *vlan_id)
{
	struct dst_entry *dst = sk_dst_get(clcsock->sk);
	int rc = 0;

	*vlan_id = 0;
	if (!dst) {
		rc = -ENOTCONN;
		goto out;
	}
	if (!dst->dev) {
		rc = -ENODEV;
		goto out_rel;
	}

	if (is_vlan_dev(dst->dev))
		*vlan_id = vlan_dev_vlan_id(dst->dev);

out_rel:
	dst_release(dst);
out:
	return rc;
}

/* determine the link gid matching the vlan id of the link group */
static int smc_link_determine_gid(struct smc_link_group *lgr)
{
	struct smc_link *lnk = &lgr->lnk[SMC_SINGLE_LINK];
	struct ib_gid_attr gattr;
	union ib_gid gid;
	int i;

	if (!lgr->vlan_id) {
		lnk->gid = lnk->smcibdev->gid[lnk->ibport - 1];
		return 0;
	}

	for (i = 0; i < lnk->smcibdev->pattr[lnk->ibport - 1].gid_tbl_len;
	     i++) {
		if (ib_query_gid(lnk->smcibdev->ibdev, lnk->ibport, i, &gid,
				 &gattr))
			continue;
		if (gattr.ndev &&
		    (vlan_dev_vlan_id(gattr.ndev) == lgr->vlan_id)) {
			lnk->gid = gid;
			return 0;
		}
	}
	return -ENODEV;
}

/* create a new SMC connection (and a new link group if necessary) */
int smc_conn_create(struct smc_sock *smc, __be32 peer_in_addr,
		    struct smc_ib_device *smcibdev, u8 ibport,
		    struct smc_clc_msg_local *lcl, int srv_first_contact)
{
	struct smc_connection *conn = &smc->conn;
	struct smc_link_group *lgr;
	unsigned short vlan_id;
	enum smc_lgr_role role;
	int local_contact = SMC_FIRST_CONTACT;
	int rc = 0;

	role = smc->listen_smc ? SMC_SERV : SMC_CLNT;
	rc = smc_vlan_by_tcpsk(smc->clcsock, &vlan_id);
	if (rc)
		return rc;

	if ((role == SMC_CLNT) && srv_first_contact)
		/* create new link group as well */
		goto create;

	/* determine if an existing link group can be reused */
	spin_lock_bh(&smc_lgr_list.lock);
	list_for_each_entry(lgr, &smc_lgr_list.list, list) {
		write_lock_bh(&lgr->conns_lock);
		if (!memcmp(lgr->peer_systemid, lcl->id_for_peer,
			    SMC_SYSTEMID_LEN) &&
		    !memcmp(lgr->lnk[SMC_SINGLE_LINK].peer_gid, &lcl->gid,
			    SMC_GID_SIZE) &&
		    !memcmp(lgr->lnk[SMC_SINGLE_LINK].peer_mac, lcl->mac,
			    sizeof(lcl->mac)) &&
		    !lgr->sync_err &&
		    (lgr->role == role) &&
		    (lgr->vlan_id == vlan_id) &&
		    ((role == SMC_CLNT) ||
		     (lgr->conns_num < SMC_RMBS_PER_LGR_MAX))) {
			/* link group found */
			local_contact = SMC_REUSE_CONTACT;
			conn->lgr = lgr;
			smc_lgr_register_conn(conn); /* add smc conn to lgr */
			write_unlock_bh(&lgr->conns_lock);
			break;
		}
		write_unlock_bh(&lgr->conns_lock);
	}
	spin_unlock_bh(&smc_lgr_list.lock);

	if (role == SMC_CLNT && !srv_first_contact &&
	    (local_contact == SMC_FIRST_CONTACT)) {
		/* Server reuses a link group, but Client wants to start
		 * a new one
		 * send out_of_sync decline, reason synchr. error
		 */
		return -ENOLINK;
	}

create:
	if (local_contact == SMC_FIRST_CONTACT) {
		rc = smc_lgr_create(smc, peer_in_addr, smcibdev, ibport,
				    lcl->id_for_peer, vlan_id);
		if (rc)
			goto out;
		smc_lgr_register_conn(conn); /* add smc conn to lgr */
		rc = smc_link_determine_gid(conn->lgr);
	}
	conn->local_tx_ctrl.common.type = SMC_CDC_MSG_TYPE;
	conn->local_tx_ctrl.len = sizeof(struct smc_cdc_msg);
#ifndef KERNEL_HAS_ATOMIC64
	spin_lock_init(&conn->acurs_lock);
#endif

out:
	return rc ? rc : local_contact;
}

/* try to reuse a sndbuf description slot of the sndbufs list for a certain
 * buf_size; if not available, return NULL
 */
static inline
struct smc_buf_desc *smc_sndbuf_get_slot(struct smc_link_group *lgr,
					 int compressed_bufsize)
{
	struct smc_buf_desc *sndbuf_slot;

	read_lock_bh(&lgr->sndbufs_lock);
	list_for_each_entry(sndbuf_slot, &lgr->sndbufs[compressed_bufsize],
			    list) {
		if (cmpxchg(&sndbuf_slot->used, 0, 1) == 0) {
			read_unlock_bh(&lgr->sndbufs_lock);
			return sndbuf_slot;
		}
	}
	read_unlock_bh(&lgr->sndbufs_lock);
	return NULL;
}

/* try to reuse an rmb description slot of the rmbs list for a certain
 * rmbe_size; if not available, return NULL
 */
static inline
struct smc_buf_desc *smc_rmb_get_slot(struct smc_link_group *lgr,
				      int compressed_bufsize)
{
	struct smc_buf_desc *rmb_slot;

	read_lock_bh(&lgr->rmbs_lock);
	list_for_each_entry(rmb_slot, &lgr->rmbs[compressed_bufsize],
			    list) {
		if (cmpxchg(&rmb_slot->used, 0, 1) == 0) {
			read_unlock_bh(&lgr->rmbs_lock);
			return rmb_slot;
		}
	}
	read_unlock_bh(&lgr->rmbs_lock);
	return NULL;
}

/* create the tx buffer for an SMC socket */
int smc_sndbuf_create(struct smc_sock *smc)
{
	struct smc_connection *conn = &smc->conn;
	struct smc_link_group *lgr = conn->lgr;
	int tmp_bufsize, tmp_bufsize_short;
	struct smc_buf_desc *sndbuf_desc;
	int rc;

	/* use socket send buffer size (w/o overhead) as start value */
	for (tmp_bufsize_short = smc_compress_bufsize(smc->sk.sk_sndbuf / 2);
	     tmp_bufsize_short >= 0; tmp_bufsize_short--) {
		tmp_bufsize = smc_uncompress_bufsize(tmp_bufsize_short);
		/* check for reusable sndbuf_slot in the link group */
		sndbuf_desc = smc_sndbuf_get_slot(lgr, tmp_bufsize_short);
		if (sndbuf_desc) {
			memset(sndbuf_desc->cpu_addr, 0, tmp_bufsize);
			break; /* found reusable slot */
		}
		/* try to alloc a new send buffer */
		sndbuf_desc = kzalloc(sizeof(*sndbuf_desc), GFP_KERNEL);
		if (!sndbuf_desc)
			break; /* give up with -ENOMEM */
		sndbuf_desc->cpu_addr = kzalloc(tmp_bufsize,
						GFP_KERNEL | __GFP_NOWARN |
						__GFP_NOMEMALLOC |
						__GFP_NORETRY);
		if (!sndbuf_desc->cpu_addr) {
			kfree(sndbuf_desc);
			/* if send buffer allocation has failed,
			 * try a smaller one
			 */
			continue;
		}
		rc = smc_ib_buf_map(lgr->lnk[SMC_SINGLE_LINK].smcibdev,
				    tmp_bufsize, sndbuf_desc,
				    DMA_TO_DEVICE);
		if (rc) {
			kfree(sndbuf_desc->cpu_addr);
			kfree(sndbuf_desc);
			continue; /* if mapping failed, try smaller one */
		}
		sndbuf_desc->used = 1;
		write_lock_bh(&lgr->sndbufs_lock);
		list_add(&sndbuf_desc->list,
			 &lgr->sndbufs[tmp_bufsize_short]);
		write_unlock_bh(&lgr->sndbufs_lock);
		break;
	}
	if (sndbuf_desc && sndbuf_desc->cpu_addr) {
		conn->sndbuf_desc = sndbuf_desc;
		conn->sndbuf_size = tmp_bufsize;
		smc->sk.sk_sndbuf = tmp_bufsize * 2;
		atomic_set(&conn->sndbuf_space, tmp_bufsize);
		return 0;
	} else {
		return -ENOMEM;
	}
}

/* create the RMB for an SMC socket (even though the SMC protocol
 * allows more than one RMB-element per RMB, the Linux implementation
 * uses just one RMB-element per RMB, i.e. uses an extra RMB for every
 * connection in a link group
 */
int smc_rmb_create(struct smc_sock *smc)
{
	struct smc_connection *conn = &smc->conn;
	struct smc_link_group *lgr = conn->lgr;
	int tmp_bufsize, tmp_bufsize_short;
	struct smc_buf_desc *rmb_desc;
	int rc;

	/* use socket recv buffer size (w/o overhead) as start value */
	for (tmp_bufsize_short = smc_compress_bufsize(smc->sk.sk_rcvbuf / 2);
	     tmp_bufsize_short >= 0; tmp_bufsize_short--) {
		tmp_bufsize = smc_uncompress_bufsize(tmp_bufsize_short);
		/* check for reusable rmb_slot in the link group */
		rmb_desc = smc_rmb_get_slot(lgr, tmp_bufsize_short);
		if (rmb_desc) {
			memset(rmb_desc->cpu_addr, 0, tmp_bufsize);
			break; /* found reusable slot */
		}
		/* try to alloc a new RMB */
		rmb_desc = kzalloc(sizeof(*rmb_desc), GFP_KERNEL);
		if (!rmb_desc)
			break; /* give up with -ENOMEM */
		rmb_desc->cpu_addr = kzalloc(tmp_bufsize,
					     GFP_KERNEL | __GFP_NOWARN |
					     __GFP_NOMEMALLOC |
					     __GFP_NORETRY);
		if (!rmb_desc->cpu_addr) {
			kfree(rmb_desc);
			/* if RMB allocation has failed,
			 * try a smaller one
			 */
			continue;
		}
		rc = smc_ib_buf_map(lgr->lnk[SMC_SINGLE_LINK].smcibdev,
				    tmp_bufsize, rmb_desc,
				    DMA_FROM_DEVICE);
		if (rc) {
			kfree(rmb_desc->cpu_addr);
			kfree(rmb_desc);
			continue; /* if mapping failed, try smaller one */
		}
		rc = smc_ib_get_memory_region(lgr->lnk[SMC_SINGLE_LINK].roce_pd,
					      IB_ACCESS_REMOTE_WRITE |
					      IB_ACCESS_LOCAL_WRITE,
					     &rmb_desc->mr_rx[SMC_SINGLE_LINK]);
		if (rc) {
			smc_ib_buf_unmap(lgr->lnk[SMC_SINGLE_LINK].smcibdev,
					 tmp_bufsize, rmb_desc,
					 DMA_FROM_DEVICE);
			kfree(rmb_desc->cpu_addr);
			kfree(rmb_desc);
			continue;
		}
		rmb_desc->used = 1;
		write_lock_bh(&lgr->rmbs_lock);
		list_add(&rmb_desc->list,
			 &lgr->rmbs[tmp_bufsize_short]);
		write_unlock_bh(&lgr->rmbs_lock);
		break;
	}
	if (rmb_desc && rmb_desc->cpu_addr) {
		conn->rmb_desc = rmb_desc;
		conn->rmbe_size = tmp_bufsize;
		conn->rmbe_size_short = tmp_bufsize_short;
		smc->sk.sk_rcvbuf = tmp_bufsize * 2;
		atomic_set(&conn->bytes_to_rcv, 0);
		return 0;
	} else {
		return -ENOMEM;
	}
}

static inline int smc_rmb_reserve_rtoken_idx(struct smc_link_group *lgr)
{
	int i;

	for_each_clear_bit(i, lgr->rtokens_used_mask, SMC_RMBS_PER_LGR_MAX) {
		if (!test_and_set_bit(i, lgr->rtokens_used_mask))
			return i;
	}
	return -ENOSPC;
}

/* save rkey and dma_addr received from peer during clc handshake */
int smc_rmb_rtoken_handling(struct smc_connection *conn,
			    struct smc_clc_msg_accept_confirm *clc)
{
	u64 dma_addr = be64_to_cpu(clc->rmb_dma_addr);
	struct smc_link_group *lgr = conn->lgr;
	u32 rkey = ntohl(clc->rmb_rkey);
	int i;

	for (i = 0; i < SMC_RMBS_PER_LGR_MAX; i++) {
		if ((lgr->rtokens[i][SMC_SINGLE_LINK].rkey == rkey) &&
		    test_bit(i, lgr->rtokens_used_mask)) {
			conn->rtoken_idx = i;
			return 0;
		}
	}
	conn->rtoken_idx = smc_rmb_reserve_rtoken_idx(lgr);
	if (conn->rtoken_idx < 0)
		return conn->rtoken_idx;
	lgr->rtokens[conn->rtoken_idx][SMC_SINGLE_LINK].rkey = rkey;
	lgr->rtokens[conn->rtoken_idx][SMC_SINGLE_LINK].dma_addr = dma_addr;
	return 0;
}
Commit	Line	Data
0cfdd8f9 UB	1	/*
	2	* Shared Memory Communications over RDMA (SMC-R) and RoCE
	3	*
	4	* Basic Transport Functions exploiting Infiniband API
	5	*
	6	* Copyright IBM Corp. 2016
	7	*
	8	* Author(s): Ursula Braun <ubraun@linux.vnet.ibm.com>
	9	*/
	10
	11	#include <linux/socket.h>
	12	#include <linux/if_vlan.h>
	13	#include <linux/random.h>
	14	#include <linux/workqueue.h>
	15	#include <net/tcp.h>
	16	#include <net/sock.h>
	17	#include <rdma/ib_verbs.h>
	18
	19	#include "smc.h"
	20	#include "smc_clc.h"
	21	#include "smc_core.h"
	22	#include "smc_ib.h"
f38ba179	23	#include "smc_wr.h"
9bf9abea	24	#include "smc_llc.h"
5f08318f	25	#include "smc_cdc.h"
0cfdd8f9	26
9bf9abea	27	#define SMC_LGR_NUM_INCR 256
0cfdd8f9 UB	28	#define SMC_LGR_FREE_DELAY (600 * HZ)
0cfdd8f9 UB	29
9bf9abea UB	30	static u32 smc_lgr_num; /* unique link group number */
9bf9abea UB	31
0cfdd8f9 UB	32	/* Register connection's alert token in our lookup structure.
	33	* To use rbtrees we have to implement our own insert core.
	34	* Requires @conns_lock
	35	* @smc connection to register
	36	* Returns 0 on success, != otherwise.
	37	*/
	38	static void smc_lgr_add_alert_token(struct smc_connection *conn)
	39	{
	40	struct rb_node *link, parent = NULL;
	41	u32 token = conn->alert_token_local;
	42
	43	link = &conn->lgr->conns_all.rb_node;
	44	while (*link) {
	45	struct smc_connection cur = rb_entry(link,
	46	struct smc_connection, alert_node);
	47
	48	parent = *link;
	49	if (cur->alert_token_local > token)
	50	link = &parent->rb_left;
	51	else
	52	link = &parent->rb_right;
	53	}
	54	/* Put the new node there */
	55	rb_link_node(&conn->alert_node, parent, link);
	56	rb_insert_color(&conn->alert_node, &conn->lgr->conns_all);
	57	}
	58
	59	/* Register connection in link group by assigning an alert token
	60	* registered in a search tree.
	61	* Requires @conns_lock
	62	* Note that '0' is a reserved value and not assigned.
	63	*/
	64	static void smc_lgr_register_conn(struct smc_connection *conn)
	65	{
	66	struct smc_sock *smc = container_of(conn, struct smc_sock, conn);
	67	static atomic_t nexttoken = ATOMIC_INIT(0);
	68
	69	/* find a new alert_token_local value not yet used by some connection
	70	* in this link group
	71	*/
	72	sock_hold(&smc->sk); /* sock_put in smc_lgr_unregister_conn() */
	73	while (!conn->alert_token_local) {
	74	conn->alert_token_local = atomic_inc_return(&nexttoken);
	75	if (smc_lgr_find_conn(conn->alert_token_local, conn->lgr))
	76	conn->alert_token_local = 0;
	77	}
	78	smc_lgr_add_alert_token(conn);
	79	conn->lgr->conns_num++;
	80	}
	81
	82	/* Unregister connection and reset the alert token of the given connection<
	83	*/
	84	static void __smc_lgr_unregister_conn(struct smc_connection *conn)
	85	{
	86	struct smc_sock *smc = container_of(conn, struct smc_sock, conn);
	87	struct smc_link_group *lgr = conn->lgr;
	88
	89	rb_erase(&conn->alert_node, &lgr->conns_all);
	90	lgr->conns_num--;
	91	conn->alert_token_local = 0;
	92	conn->lgr = NULL;
	93	sock_put(&smc->sk); /* sock_hold in smc_lgr_register_conn() */
	94	}
	95
96	/* Unregister connection and trigger lgr freeing if applicable
97	*/
98	static void smc_lgr_unregister_conn(struct smc_connection *conn)
99	{
100	struct smc_link_group *lgr = conn->lgr;
101	int reduced = 0;
102
103	write_lock_bh(&lgr->conns_lock);
104	if (conn->alert_token_local) {
105	reduced = 1;
106	__smc_lgr_unregister_conn(conn);
107	}
108	write_unlock_bh(&lgr->conns_lock);
109	if (reduced && !lgr->conns_num)
110	schedule_delayed_work(&lgr->free_work, SMC_LGR_FREE_DELAY);
111	}
112
113	static void smc_lgr_free_work(struct work_struct *work)
114	{
115	struct smc_link_group *lgr = container_of(to_delayed_work(work),
116	struct smc_link_group,
117	free_work);
118	bool conns;
119
120	spin_lock_bh(&smc_lgr_list.lock);
121	read_lock_bh(&lgr->conns_lock);
122	conns = RB_EMPTY_ROOT(&lgr->conns_all);
123	read_unlock_bh(&lgr->conns_lock);
124	if (!conns) { /* number of lgr connections is no longer zero */
125	spin_unlock_bh(&smc_lgr_list.lock);
126	return;
127	}
128	list_del_init(&lgr->list); /* remove from smc_lgr_list */
129	spin_unlock_bh(&smc_lgr_list.lock);
130	smc_lgr_free(lgr);
131	}
132
133	/* create a new SMC link group */
134	static int smc_lgr_create(struct smc_sock *smc, __be32 peer_in_addr,
135	struct smc_ib_device *smcibdev, u8 ibport,
136	char *peer_systemid, unsigned short vlan_id)
137	{
138	struct smc_link_group *lgr;
139	struct smc_link *lnk;
140	u8 rndvec[3];
141	int rc = 0;
cd6851f3	142	int i;
0cfdd8f9 UB	143
	144	lgr = kzalloc(sizeof(*lgr), GFP_KERNEL);
	145	if (!lgr) {
	146	rc = -ENOMEM;
	147	goto out;
	148	}
	149	lgr->role = smc->listen_smc ? SMC_SERV : SMC_CLNT;
	150	lgr->sync_err = false;
	151	lgr->daddr = peer_in_addr;
	152	memcpy(lgr->peer_systemid, peer_systemid, SMC_SYSTEMID_LEN);
	153	lgr->vlan_id = vlan_id;
cd6851f3 UB	154	rwlock_init(&lgr->sndbufs_lock);
	155	rwlock_init(&lgr->rmbs_lock);
	156	for (i = 0; i < SMC_RMBE_SIZES; i++) {
	157	INIT_LIST_HEAD(&lgr->sndbufs[i]);
	158	INIT_LIST_HEAD(&lgr->rmbs[i]);
	159	}
9bf9abea UB	160	smc_lgr_num += SMC_LGR_NUM_INCR;
9bf9abea UB	161	memcpy(&lgr->id, (u8 *)&smc_lgr_num, SMC_LGR_ID_SIZE);
0cfdd8f9 UB	162	INIT_DELAYED_WORK(&lgr->free_work, smc_lgr_free_work);
	163	lgr->conns_all = RB_ROOT;
	164
	165	lnk = &lgr->lnk[SMC_SINGLE_LINK];
	166	/* initialize link */
	167	lnk->smcibdev = smcibdev;
	168	lnk->ibport = ibport;
	169	lnk->path_mtu = smcibdev->pattr[ibport - 1].active_mtu;
bd4ad577 UB	170	if (!smcibdev->initialized)
bd4ad577 UB	171	smc_ib_setup_per_ibdev(smcibdev);
0cfdd8f9 UB	172	get_random_bytes(rndvec, sizeof(rndvec));
0cfdd8f9 UB	173	lnk->psn_initial = rndvec[0] + (rndvec[1] << 8) + (rndvec[2] << 16);
f38ba179 UB	174	rc = smc_wr_alloc_link_mem(lnk);
	175	if (rc)
	176	goto free_lgr;
	177	init_waitqueue_head(&lnk->wr_tx_wait);
bd4ad577 UB	178	rc = smc_ib_create_protection_domain(lnk);
	179	if (rc)
	180	goto free_link_mem;
	181	rc = smc_ib_create_queue_pair(lnk);
	182	if (rc)
	183	goto dealloc_pd;
	184	rc = smc_wr_create_link(lnk);
	185	if (rc)
	186	goto destroy_qp;
9bf9abea UB	187	init_completion(&lnk->llc_confirm);
9bf9abea UB	188	init_completion(&lnk->llc_confirm_resp);
0cfdd8f9 UB	189
	190	smc->conn.lgr = lgr;
	191	rwlock_init(&lgr->conns_lock);
	192	spin_lock_bh(&smc_lgr_list.lock);
	193	list_add(&lgr->list, &smc_lgr_list.list);
	194	spin_unlock_bh(&smc_lgr_list.lock);
f38ba179 UB	195	return 0;
f38ba179 UB	196
bd4ad577 UB	197	destroy_qp:
	198	smc_ib_destroy_queue_pair(lnk);
	199	dealloc_pd:
	200	smc_ib_dealloc_protection_domain(lnk);
	201	free_link_mem:
	202	smc_wr_free_link_mem(lnk);
f38ba179 UB	203	free_lgr:
f38ba179 UB	204	kfree(lgr);
0cfdd8f9 UB	205	out:
	206	return rc;
	207	}
	208
cd6851f3 UB	209	static void smc_sndbuf_unuse(struct smc_connection *conn)
	210	{
	211	if (conn->sndbuf_desc) {
	212	conn->sndbuf_desc->used = 0;
	213	conn->sndbuf_size = 0;
	214	}
	215	}
	216
	217	static void smc_rmb_unuse(struct smc_connection *conn)
	218	{
	219	if (conn->rmb_desc) {
	220	conn->rmb_desc->used = 0;
	221	conn->rmbe_size = 0;
	222	}
	223	}
	224
0cfdd8f9 UB	225	/* remove a finished connection from its link group */
	226	void smc_conn_free(struct smc_connection *conn)
	227	{
	228	struct smc_link_group *lgr = conn->lgr;
	229
	230	if (!lgr)
	231	return;
5f08318f	232	smc_cdc_tx_dismiss_slots(conn);
0cfdd8f9	233	smc_lgr_unregister_conn(conn);
cd6851f3 UB	234	smc_rmb_unuse(conn);
cd6851f3 UB	235	smc_sndbuf_unuse(conn);
0cfdd8f9 UB	236	}
	237
	238	static void smc_link_clear(struct smc_link *lnk)
	239	{
	240	lnk->peer_qpn = 0;
bd4ad577	241	smc_ib_modify_qp_reset(lnk);
f38ba179	242	smc_wr_free_link(lnk);
bd4ad577 UB	243	smc_ib_destroy_queue_pair(lnk);
bd4ad577 UB	244	smc_ib_dealloc_protection_domain(lnk);
f38ba179	245	smc_wr_free_link_mem(lnk);
0cfdd8f9 UB	246	}
0cfdd8f9 UB	247
cd6851f3 UB	248	static void smc_lgr_free_sndbufs(struct smc_link_group *lgr)
	249	{
	250	struct smc_buf_desc sndbuf_desc, bf_desc;
	251	int i;
	252
	253	for (i = 0; i < SMC_RMBE_SIZES; i++) {
	254	list_for_each_entry_safe(sndbuf_desc, bf_desc, &lgr->sndbufs[i],
	255	list) {
bd4ad577 UB	256	list_del(&sndbuf_desc->list);
	257	smc_ib_buf_unmap(lgr->lnk[SMC_SINGLE_LINK].smcibdev,
	258	smc_uncompress_bufsize(i),
	259	sndbuf_desc, DMA_TO_DEVICE);
cd6851f3 UB	260	kfree(sndbuf_desc->cpu_addr);
	261	kfree(sndbuf_desc);
	262	}
	263	}
	264	}
	265
	266	static void smc_lgr_free_rmbs(struct smc_link_group *lgr)
	267	{
	268	struct smc_buf_desc rmb_desc, bf_desc;
bd4ad577	269	struct smc_link *lnk = &lgr->lnk[SMC_SINGLE_LINK];
cd6851f3 UB	270	int i;
	271
	272	for (i = 0; i < SMC_RMBE_SIZES; i++) {
	273	list_for_each_entry_safe(rmb_desc, bf_desc, &lgr->rmbs[i],
	274	list) {
bd4ad577 UB	275	list_del(&rmb_desc->list);
	276	smc_ib_buf_unmap(lnk->smcibdev,
	277	smc_uncompress_bufsize(i),
	278	rmb_desc, DMA_FROM_DEVICE);
cd6851f3 UB	279	kfree(rmb_desc->cpu_addr);
	280	kfree(rmb_desc);
	281	}
	282	}
	283	}
	284
0cfdd8f9 UB	285	/* remove a link group */
	286	void smc_lgr_free(struct smc_link_group *lgr)
	287	{
cd6851f3 UB	288	smc_lgr_free_rmbs(lgr);
cd6851f3 UB	289	smc_lgr_free_sndbufs(lgr);
0cfdd8f9 UB	290	smc_link_clear(&lgr->lnk[SMC_SINGLE_LINK]);
	291	kfree(lgr);
	292	}
	293
	294	/* terminate linkgroup abnormally */
	295	void smc_lgr_terminate(struct smc_link_group *lgr)
	296	{
	297	struct smc_connection *conn;
	298	struct rb_node *node;
	299
	300	spin_lock_bh(&smc_lgr_list.lock);
	301	if (list_empty(&lgr->list)) {
	302	/* termination already triggered */
	303	spin_unlock_bh(&smc_lgr_list.lock);
	304	return;
	305	}
	306	/* do not use this link group for new connections */
	307	list_del_init(&lgr->list);
	308	spin_unlock_bh(&smc_lgr_list.lock);
	309
	310	write_lock_bh(&lgr->conns_lock);
	311	node = rb_first(&lgr->conns_all);
	312	while (node) {
	313	conn = rb_entry(node, struct smc_connection, alert_node);
	314	__smc_lgr_unregister_conn(conn);
	315	node = rb_first(&lgr->conns_all);
	316	}
	317	write_unlock_bh(&lgr->conns_lock);
	318	schedule_delayed_work(&lgr->free_work, SMC_LGR_FREE_DELAY);
	319	}
	320
	321	/* Determine vlan of internal TCP socket.
	322	* @vlan_id: address to store the determined vlan id into
	323	*/
	324	static int smc_vlan_by_tcpsk(struct socket clcsock, unsigned short vlan_id)
	325	{
	326	struct dst_entry *dst = sk_dst_get(clcsock->sk);
	327	int rc = 0;
	328
	329	*vlan_id = 0;
	330	if (!dst) {
	331	rc = -ENOTCONN;
	332	goto out;
	333	}
	334	if (!dst->dev) {
	335	rc = -ENODEV;
	336	goto out_rel;
	337	}
	338
	339	if (is_vlan_dev(dst->dev))
	340	*vlan_id = vlan_dev_vlan_id(dst->dev);
	341
	342	out_rel:
	343	dst_release(dst);
	344	out:
	345	return rc;
	346	}
	347
	348	/* determine the link gid matching the vlan id of the link group */
	349	static int smc_link_determine_gid(struct smc_link_group *lgr)
	350	{
	351	struct smc_link *lnk = &lgr->lnk[SMC_SINGLE_LINK];
	352	struct ib_gid_attr gattr;
	353	union ib_gid gid;
354	int i;
355
356	if (!lgr->vlan_id) {
357	lnk->gid = lnk->smcibdev->gid[lnk->ibport - 1];
358	return 0;
359	}
360
361	for (i = 0; i < lnk->smcibdev->pattr[lnk->ibport - 1].gid_tbl_len;
362	i++) {
363	if (ib_query_gid(lnk->smcibdev->ibdev, lnk->ibport, i, &gid,
364	&gattr))
365	continue;
366	if (gattr.ndev &&
367	(vlan_dev_vlan_id(gattr.ndev) == lgr->vlan_id)) {
368	lnk->gid = gid;
369	return 0;
370	}
371	}
372	return -ENODEV;
373	}
374
375	/* create a new SMC connection (and a new link group if necessary) */
376	int smc_conn_create(struct smc_sock *smc, __be32 peer_in_addr,
377	struct smc_ib_device *smcibdev, u8 ibport,
378	struct smc_clc_msg_local *lcl, int srv_first_contact)
379	{
380	struct smc_connection *conn = &smc->conn;
381	struct smc_link_group *lgr;
382	unsigned short vlan_id;
383	enum smc_lgr_role role;
384	int local_contact = SMC_FIRST_CONTACT;
385	int rc = 0;
386
387	role = smc->listen_smc ? SMC_SERV : SMC_CLNT;
388	rc = smc_vlan_by_tcpsk(smc->clcsock, &vlan_id);
389	if (rc)
390	return rc;
391
392	if ((role == SMC_CLNT) && srv_first_contact)
393	/* create new link group as well */
394	goto create;
395
396	/* determine if an existing link group can be reused */
397	spin_lock_bh(&smc_lgr_list.lock);
398	list_for_each_entry(lgr, &smc_lgr_list.list, list) {
399	write_lock_bh(&lgr->conns_lock);
400	if (!memcmp(lgr->peer_systemid, lcl->id_for_peer,
401	SMC_SYSTEMID_LEN) &&
402	!memcmp(lgr->lnk[SMC_SINGLE_LINK].peer_gid, &lcl->gid,
403	SMC_GID_SIZE) &&
404	!memcmp(lgr->lnk[SMC_SINGLE_LINK].peer_mac, lcl->mac,
405	sizeof(lcl->mac)) &&
406	!lgr->sync_err &&
407	(lgr->role == role) &&
cd6851f3 UB	408	(lgr->vlan_id == vlan_id) &&
	409	((role == SMC_CLNT) \|\|
	410	(lgr->conns_num < SMC_RMBS_PER_LGR_MAX))) {
0cfdd8f9 UB	411	/* link group found */
	412	local_contact = SMC_REUSE_CONTACT;
	413	conn->lgr = lgr;
	414	smc_lgr_register_conn(conn); /* add smc conn to lgr */
	415	write_unlock_bh(&lgr->conns_lock);
	416	break;
	417	}
	418	write_unlock_bh(&lgr->conns_lock);
	419	}
	420	spin_unlock_bh(&smc_lgr_list.lock);
	421
	422	if (role == SMC_CLNT && !srv_first_contact &&
	423	(local_contact == SMC_FIRST_CONTACT)) {
	424	/* Server reuses a link group, but Client wants to start
	425	* a new one
	426	* send out_of_sync decline, reason synchr. error
	427	*/
	428	return -ENOLINK;
	429	}
	430
	431	create:
	432	if (local_contact == SMC_FIRST_CONTACT) {
	433	rc = smc_lgr_create(smc, peer_in_addr, smcibdev, ibport,
	434	lcl->id_for_peer, vlan_id);
	435	if (rc)
	436	goto out;
	437	smc_lgr_register_conn(conn); /* add smc conn to lgr */
	438	rc = smc_link_determine_gid(conn->lgr);
	439	}
5f08318f UB	440	conn->local_tx_ctrl.common.type = SMC_CDC_MSG_TYPE;
	441	conn->local_tx_ctrl.len = sizeof(struct smc_cdc_msg);
	442	#ifndef KERNEL_HAS_ATOMIC64
	443	spin_lock_init(&conn->acurs_lock);
	444	#endif
0cfdd8f9 UB	445
	446	out:
	447	return rc ? rc : local_contact;
	448	}
cd6851f3 UB	449
	450	/* try to reuse a sndbuf description slot of the sndbufs list for a certain
	451	* buf_size; if not available, return NULL
	452	*/
	453	static inline
	454	struct smc_buf_desc smc_sndbuf_get_slot(struct smc_link_group lgr,
	455	int compressed_bufsize)
	456	{
	457	struct smc_buf_desc *sndbuf_slot;
	458
	459	read_lock_bh(&lgr->sndbufs_lock);
	460	list_for_each_entry(sndbuf_slot, &lgr->sndbufs[compressed_bufsize],
	461	list) {
	462	if (cmpxchg(&sndbuf_slot->used, 0, 1) == 0) {
	463	read_unlock_bh(&lgr->sndbufs_lock);
	464	return sndbuf_slot;
	465	}
	466	}
	467	read_unlock_bh(&lgr->sndbufs_lock);
	468	return NULL;
	469	}
	470
	471	/* try to reuse an rmb description slot of the rmbs list for a certain
	472	* rmbe_size; if not available, return NULL
	473	*/
	474	static inline
	475	struct smc_buf_desc smc_rmb_get_slot(struct smc_link_group lgr,
	476	int compressed_bufsize)
	477	{
	478	struct smc_buf_desc *rmb_slot;
	479
	480	read_lock_bh(&lgr->rmbs_lock);
	481	list_for_each_entry(rmb_slot, &lgr->rmbs[compressed_bufsize],
	482	list) {
	483	if (cmpxchg(&rmb_slot->used, 0, 1) == 0) {
	484	read_unlock_bh(&lgr->rmbs_lock);
	485	return rmb_slot;
	486	}
	487	}
	488	read_unlock_bh(&lgr->rmbs_lock);
	489	return NULL;
	490	}
	491
	492	/* create the tx buffer for an SMC socket */
	493	int smc_sndbuf_create(struct smc_sock *smc)
	494	{
	495	struct smc_connection *conn = &smc->conn;
	496	struct smc_link_group *lgr = conn->lgr;
	497	int tmp_bufsize, tmp_bufsize_short;
	498	struct smc_buf_desc *sndbuf_desc;
	499	int rc;
	500
	501	/* use socket send buffer size (w/o overhead) as start value */
	502	for (tmp_bufsize_short = smc_compress_bufsize(smc->sk.sk_sndbuf / 2);
	503	tmp_bufsize_short >= 0; tmp_bufsize_short--) {
	504	tmp_bufsize = smc_uncompress_bufsize(tmp_bufsize_short);
	505	/* check for reusable sndbuf_slot in the link group */
	506	sndbuf_desc = smc_sndbuf_get_slot(lgr, tmp_bufsize_short);
	507	if (sndbuf_desc) {
	508	memset(sndbuf_desc->cpu_addr, 0, tmp_bufsize);
	509	break; /* found reusable slot */
	510	}
	511	/* try to alloc a new send buffer */
	512	sndbuf_desc = kzalloc(sizeof(*sndbuf_desc), GFP_KERNEL);
513	if (!sndbuf_desc)
514	break; /* give up with -ENOMEM */
515	sndbuf_desc->cpu_addr = kzalloc(tmp_bufsize,
516	GFP_KERNEL \| __GFP_NOWARN \|
517	__GFP_NOMEMALLOC \|
518	__GFP_NORETRY);
519	if (!sndbuf_desc->cpu_addr) {
520	kfree(sndbuf_desc);
521	/* if send buffer allocation has failed,
522	* try a smaller one
523	*/
524	continue;
525	}
526	rc = smc_ib_buf_map(lgr->lnk[SMC_SINGLE_LINK].smcibdev,
527	tmp_bufsize, sndbuf_desc,
528	DMA_TO_DEVICE);
529	if (rc) {
530	kfree(sndbuf_desc->cpu_addr);
531	kfree(sndbuf_desc);
532	continue; /* if mapping failed, try smaller one */
533	}
534	sndbuf_desc->used = 1;
535	write_lock_bh(&lgr->sndbufs_lock);
536	list_add(&sndbuf_desc->list,
537	&lgr->sndbufs[tmp_bufsize_short]);
538	write_unlock_bh(&lgr->sndbufs_lock);
539	break;
540	}
541	if (sndbuf_desc && sndbuf_desc->cpu_addr) {
542	conn->sndbuf_desc = sndbuf_desc;
543	conn->sndbuf_size = tmp_bufsize;
544	smc->sk.sk_sndbuf = tmp_bufsize * 2;
5f08318f	545	atomic_set(&conn->sndbuf_space, tmp_bufsize);
cd6851f3 UB	546	return 0;
	547	} else {
	548	return -ENOMEM;
	549	}
	550	}
	551
	552	/* create the RMB for an SMC socket (even though the SMC protocol
	553	* allows more than one RMB-element per RMB, the Linux implementation
	554	* uses just one RMB-element per RMB, i.e. uses an extra RMB for every
	555	* connection in a link group
	556	*/
	557	int smc_rmb_create(struct smc_sock *smc)
	558	{
	559	struct smc_connection *conn = &smc->conn;
	560	struct smc_link_group *lgr = conn->lgr;
	561	int tmp_bufsize, tmp_bufsize_short;
	562	struct smc_buf_desc *rmb_desc;
	563	int rc;
	564
	565	/* use socket recv buffer size (w/o overhead) as start value */
	566	for (tmp_bufsize_short = smc_compress_bufsize(smc->sk.sk_rcvbuf / 2);
	567	tmp_bufsize_short >= 0; tmp_bufsize_short--) {
	568	tmp_bufsize = smc_uncompress_bufsize(tmp_bufsize_short);
	569	/* check for reusable rmb_slot in the link group */
	570	rmb_desc = smc_rmb_get_slot(lgr, tmp_bufsize_short);
	571	if (rmb_desc) {
	572	memset(rmb_desc->cpu_addr, 0, tmp_bufsize);
	573	break; /* found reusable slot */
	574	}
	575	/* try to alloc a new RMB */
	576	rmb_desc = kzalloc(sizeof(*rmb_desc), GFP_KERNEL);
	577	if (!rmb_desc)
	578	break; /* give up with -ENOMEM */
	579	rmb_desc->cpu_addr = kzalloc(tmp_bufsize,
	580	GFP_KERNEL \| __GFP_NOWARN \|
	581	__GFP_NOMEMALLOC \|
	582	__GFP_NORETRY);
	583	if (!rmb_desc->cpu_addr) {
	584	kfree(rmb_desc);
	585	/* if RMB allocation has failed,
	586	* try a smaller one
	587	*/
	588	continue;
	589	}
	590	rc = smc_ib_buf_map(lgr->lnk[SMC_SINGLE_LINK].smcibdev,
	591	tmp_bufsize, rmb_desc,
	592	DMA_FROM_DEVICE);
	593	if (rc) {
	594	kfree(rmb_desc->cpu_addr);
	595	kfree(rmb_desc);
	596	continue; /* if mapping failed, try smaller one */
	597	}
bd4ad577 UB	598	rc = smc_ib_get_memory_region(lgr->lnk[SMC_SINGLE_LINK].roce_pd,
	599	IB_ACCESS_REMOTE_WRITE \|
	600	IB_ACCESS_LOCAL_WRITE,
	601	&rmb_desc->mr_rx[SMC_SINGLE_LINK]);
	602	if (rc) {
	603	smc_ib_buf_unmap(lgr->lnk[SMC_SINGLE_LINK].smcibdev,
	604	tmp_bufsize, rmb_desc,
	605	DMA_FROM_DEVICE);
	606	kfree(rmb_desc->cpu_addr);
	607	kfree(rmb_desc);
	608	continue;
	609	}
cd6851f3 UB	610	rmb_desc->used = 1;
	611	write_lock_bh(&lgr->rmbs_lock);
	612	list_add(&rmb_desc->list,
	613	&lgr->rmbs[tmp_bufsize_short]);
	614	write_unlock_bh(&lgr->rmbs_lock);
	615	break;
	616	}
	617	if (rmb_desc && rmb_desc->cpu_addr) {
	618	conn->rmb_desc = rmb_desc;
	619	conn->rmbe_size = tmp_bufsize;
	620	conn->rmbe_size_short = tmp_bufsize_short;
	621	smc->sk.sk_rcvbuf = tmp_bufsize * 2;
5f08318f	622	atomic_set(&conn->bytes_to_rcv, 0);
cd6851f3 UB	623	return 0;
	624	} else {
	625	return -ENOMEM;
	626	}
	627	}
bd4ad577 UB	628
	629	static inline int smc_rmb_reserve_rtoken_idx(struct smc_link_group *lgr)
	630	{
	631	int i;
	632
	633	for_each_clear_bit(i, lgr->rtokens_used_mask, SMC_RMBS_PER_LGR_MAX) {
	634	if (!test_and_set_bit(i, lgr->rtokens_used_mask))
	635	return i;
	636	}
	637	return -ENOSPC;
	638	}
	639
	640	/* save rkey and dma_addr received from peer during clc handshake */
	641	int smc_rmb_rtoken_handling(struct smc_connection *conn,
	642	struct smc_clc_msg_accept_confirm *clc)
	643	{
	644	u64 dma_addr = be64_to_cpu(clc->rmb_dma_addr);
	645	struct smc_link_group *lgr = conn->lgr;
	646	u32 rkey = ntohl(clc->rmb_rkey);
	647	int i;
	648
	649	for (i = 0; i < SMC_RMBS_PER_LGR_MAX; i++) {
	650	if ((lgr->rtokens[i][SMC_SINGLE_LINK].rkey == rkey) &&
	651	test_bit(i, lgr->rtokens_used_mask)) {
	652	conn->rtoken_idx = i;
	653	return 0;
	654	}
	655	}
	656	conn->rtoken_idx = smc_rmb_reserve_rtoken_idx(lgr);
	657	if (conn->rtoken_idx < 0)
	658	return conn->rtoken_idx;
	659	lgr->rtokens[conn->rtoken_idx][SMC_SINGLE_LINK].rkey = rkey;
	660	lgr->rtokens[conn->rtoken_idx][SMC_SINGLE_LINK].dma_addr = dma_addr;
	661	return 0;
	662	}