[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"
#include "smc_close.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 smc_sock *smc;
	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 = container_of(conn, struct smc_sock, conn);
		sock_hold(&smc->sk);
		__smc_lgr_unregister_conn(conn);
		schedule_work(&conn->close_work);
		sock_put(&smc->sk);
		node = rb_first(&lgr->conns_all);
	}
	write_unlock_bh(&lgr->conns_lock);
}

/* 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;
}

/* one of the conditions for announcing a receiver's current window size is
 * that it "results in a minimum increase in the window size of 10% of the
 * receive buffer space" [RFC7609]
 */
static inline int smc_rmb_wnd_update_limit(int rmbe_size)
{
	return min_t(int, rmbe_size / 10, SOCK_MIN_SNDBUF / 2);
}

/* 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;
	struct smc_buf_desc *sndbuf_desc;
	int bufsize, bufsize_short;
	int rc;

	/* use socket send buffer size (w/o overhead) as start value */
	for (bufsize_short = smc_compress_bufsize(smc->sk.sk_sndbuf / 2);
	     bufsize_short >= 0; bufsize_short--) {
		bufsize = smc_uncompress_bufsize(bufsize_short);
		/* check for reusable sndbuf_slot in the link group */
		sndbuf_desc = smc_sndbuf_get_slot(lgr, bufsize_short);
		if (sndbuf_desc) {
			memset(sndbuf_desc->cpu_addr, 0, 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(bufsize,
						GFP_KERNEL | __GFP_NOWARN |
						__GFP_NOMEMALLOC |
						__GFP_NORETRY);
		if (!sndbuf_desc->cpu_addr) {
			kfree(sndbuf_desc);
			sndbuf_desc = NULL;
			/* if send buffer allocation has failed,
			 * try a smaller one
			 */
			continue;
		}
		rc = smc_ib_buf_map(lgr->lnk[SMC_SINGLE_LINK].smcibdev,
				    bufsize, sndbuf_desc, DMA_TO_DEVICE);
		if (rc) {
			kfree(sndbuf_desc->cpu_addr);
			kfree(sndbuf_desc);
			sndbuf_desc = NULL;
			continue; /* if mapping failed, try smaller one */
		}
		sndbuf_desc->used = 1;
		write_lock_bh(&lgr->sndbufs_lock);
		list_add(&sndbuf_desc->list, &lgr->sndbufs[bufsize_short]);
		write_unlock_bh(&lgr->sndbufs_lock);
		break;
	}
	if (sndbuf_desc && sndbuf_desc->cpu_addr) {
		conn->sndbuf_desc = sndbuf_desc;
		conn->sndbuf_size = bufsize;
		smc->sk.sk_sndbuf = bufsize * 2;
		atomic_set(&conn->sndbuf_space, 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;
	struct smc_buf_desc *rmb_desc;
	int bufsize, bufsize_short;
	int rc;

	/* use socket recv buffer size (w/o overhead) as start value */
	for (bufsize_short = smc_compress_bufsize(smc->sk.sk_rcvbuf / 2);
	     bufsize_short >= 0; bufsize_short--) {
		bufsize = smc_uncompress_bufsize(bufsize_short);
		/* check for reusable rmb_slot in the link group */
		rmb_desc = smc_rmb_get_slot(lgr, bufsize_short);
		if (rmb_desc) {
			memset(rmb_desc->cpu_addr, 0, 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(bufsize,
					     GFP_KERNEL | __GFP_NOWARN |
					     __GFP_NOMEMALLOC |
					     __GFP_NORETRY);
		if (!rmb_desc->cpu_addr) {
			kfree(rmb_desc);
			rmb_desc = NULL;
			/* if RMB allocation has failed,
			 * try a smaller one
			 */
			continue;
		}
		rc = smc_ib_buf_map(lgr->lnk[SMC_SINGLE_LINK].smcibdev,
				    bufsize, rmb_desc, DMA_FROM_DEVICE);
		if (rc) {
			kfree(rmb_desc->cpu_addr);
			kfree(rmb_desc);
			rmb_desc = NULL;
			continue; /* if mapping failed, try smaller one */
		}
		rmb_desc->rkey[SMC_SINGLE_LINK] =
			lgr->lnk[SMC_SINGLE_LINK].roce_pd->unsafe_global_rkey;
		rmb_desc->used = 1;
		write_lock_bh(&lgr->rmbs_lock);
		list_add(&rmb_desc->list, &lgr->rmbs[bufsize_short]);
		write_unlock_bh(&lgr->rmbs_lock);
		break;
	}
	if (rmb_desc && rmb_desc->cpu_addr) {
		conn->rmb_desc = rmb_desc;
		conn->rmbe_size = bufsize;
		conn->rmbe_size_short = bufsize_short;
		smc->sk.sk_rcvbuf = bufsize * 2;
		atomic_set(&conn->bytes_to_rcv, 0);
		conn->rmbe_update_limit = smc_rmb_wnd_update_limit(bufsize);
		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) &&
		    (lgr->rtokens[i][SMC_SINGLE_LINK].dma_addr == dma_addr) &&
		    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"
b38d7324	26	#include "smc_close.h"
0cfdd8f9	27
9bf9abea	28	#define SMC_LGR_NUM_INCR 256
0cfdd8f9 UB	29	#define SMC_LGR_FREE_DELAY (600 * HZ)
0cfdd8f9 UB	30
9bf9abea UB	31	static u32 smc_lgr_num; /* unique link group number */
9bf9abea UB	32
0cfdd8f9 UB	33	/* Register connection's alert token in our lookup structure.
	34	* To use rbtrees we have to implement our own insert core.
	35	* Requires @conns_lock
	36	* @smc connection to register
	37	* Returns 0 on success, != otherwise.
	38	*/
	39	static void smc_lgr_add_alert_token(struct smc_connection *conn)
	40	{
	41	struct rb_node *link, parent = NULL;
	42	u32 token = conn->alert_token_local;
	43
	44	link = &conn->lgr->conns_all.rb_node;
	45	while (*link) {
	46	struct smc_connection cur = rb_entry(link,
	47	struct smc_connection, alert_node);
	48
	49	parent = *link;
	50	if (cur->alert_token_local > token)
	51	link = &parent->rb_left;
	52	else
	53	link = &parent->rb_right;
	54	}
	55	/* Put the new node there */
	56	rb_link_node(&conn->alert_node, parent, link);
	57	rb_insert_color(&conn->alert_node, &conn->lgr->conns_all);
	58	}
	59
	60	/* Register connection in link group by assigning an alert token
	61	* registered in a search tree.
	62	* Requires @conns_lock
	63	* Note that '0' is a reserved value and not assigned.
	64	*/
	65	static void smc_lgr_register_conn(struct smc_connection *conn)
	66	{
	67	struct smc_sock *smc = container_of(conn, struct smc_sock, conn);
	68	static atomic_t nexttoken = ATOMIC_INIT(0);
	69
	70	/* find a new alert_token_local value not yet used by some connection
	71	* in this link group
	72	*/
	73	sock_hold(&smc->sk); /* sock_put in smc_lgr_unregister_conn() */
	74	while (!conn->alert_token_local) {
	75	conn->alert_token_local = atomic_inc_return(&nexttoken);
	76	if (smc_lgr_find_conn(conn->alert_token_local, conn->lgr))
	77	conn->alert_token_local = 0;
	78	}
	79	smc_lgr_add_alert_token(conn);
	80	conn->lgr->conns_num++;
	81	}
	82
	83	/* Unregister connection and reset the alert token of the given connection<
	84	*/
	85	static void __smc_lgr_unregister_conn(struct smc_connection *conn)
	86	{
	87	struct smc_sock *smc = container_of(conn, struct smc_sock, conn);
	88	struct smc_link_group *lgr = conn->lgr;
	89
	90	rb_erase(&conn->alert_node, &lgr->conns_all);
	91	lgr->conns_num--;
	92	conn->alert_token_local = 0;
	93	conn->lgr = NULL;
	94	sock_put(&smc->sk); /* sock_hold in smc_lgr_register_conn() */
	95	}
	96
97	/* Unregister connection and trigger lgr freeing if applicable
98	*/
99	static void smc_lgr_unregister_conn(struct smc_connection *conn)
100	{
101	struct smc_link_group *lgr = conn->lgr;
102	int reduced = 0;
103
104	write_lock_bh(&lgr->conns_lock);
105	if (conn->alert_token_local) {
106	reduced = 1;
107	__smc_lgr_unregister_conn(conn);
108	}
109	write_unlock_bh(&lgr->conns_lock);
110	if (reduced && !lgr->conns_num)
111	schedule_delayed_work(&lgr->free_work, SMC_LGR_FREE_DELAY);
112	}
113
114	static void smc_lgr_free_work(struct work_struct *work)
115	{
116	struct smc_link_group *lgr = container_of(to_delayed_work(work),
117	struct smc_link_group,
118	free_work);
119	bool conns;
120
121	spin_lock_bh(&smc_lgr_list.lock);
122	read_lock_bh(&lgr->conns_lock);
123	conns = RB_EMPTY_ROOT(&lgr->conns_all);
124	read_unlock_bh(&lgr->conns_lock);
125	if (!conns) { /* number of lgr connections is no longer zero */
126	spin_unlock_bh(&smc_lgr_list.lock);
127	return;
128	}
129	list_del_init(&lgr->list); /* remove from smc_lgr_list */
130	spin_unlock_bh(&smc_lgr_list.lock);
131	smc_lgr_free(lgr);
132	}
133
134	/* create a new SMC link group */
135	static int smc_lgr_create(struct smc_sock *smc, __be32 peer_in_addr,
136	struct smc_ib_device *smcibdev, u8 ibport,
137	char *peer_systemid, unsigned short vlan_id)
138	{
139	struct smc_link_group *lgr;
140	struct smc_link *lnk;
141	u8 rndvec[3];
142	int rc = 0;
cd6851f3	143	int i;
0cfdd8f9 UB	144
	145	lgr = kzalloc(sizeof(*lgr), GFP_KERNEL);
	146	if (!lgr) {
	147	rc = -ENOMEM;
	148	goto out;
	149	}
	150	lgr->role = smc->listen_smc ? SMC_SERV : SMC_CLNT;
	151	lgr->sync_err = false;
	152	lgr->daddr = peer_in_addr;
	153	memcpy(lgr->peer_systemid, peer_systemid, SMC_SYSTEMID_LEN);
	154	lgr->vlan_id = vlan_id;
cd6851f3 UB	155	rwlock_init(&lgr->sndbufs_lock);
	156	rwlock_init(&lgr->rmbs_lock);
	157	for (i = 0; i < SMC_RMBE_SIZES; i++) {
	158	INIT_LIST_HEAD(&lgr->sndbufs[i]);
	159	INIT_LIST_HEAD(&lgr->rmbs[i]);
	160	}
9bf9abea UB	161	smc_lgr_num += SMC_LGR_NUM_INCR;
9bf9abea UB	162	memcpy(&lgr->id, (u8 *)&smc_lgr_num, SMC_LGR_ID_SIZE);
0cfdd8f9 UB	163	INIT_DELAYED_WORK(&lgr->free_work, smc_lgr_free_work);
	164	lgr->conns_all = RB_ROOT;
	165
	166	lnk = &lgr->lnk[SMC_SINGLE_LINK];
	167	/* initialize link */
	168	lnk->smcibdev = smcibdev;
	169	lnk->ibport = ibport;
	170	lnk->path_mtu = smcibdev->pattr[ibport - 1].active_mtu;
bd4ad577 UB	171	if (!smcibdev->initialized)
bd4ad577 UB	172	smc_ib_setup_per_ibdev(smcibdev);
0cfdd8f9 UB	173	get_random_bytes(rndvec, sizeof(rndvec));
0cfdd8f9 UB	174	lnk->psn_initial = rndvec[0] + (rndvec[1] << 8) + (rndvec[2] << 16);
f38ba179 UB	175	rc = smc_wr_alloc_link_mem(lnk);
	176	if (rc)
	177	goto free_lgr;
	178	init_waitqueue_head(&lnk->wr_tx_wait);
bd4ad577 UB	179	rc = smc_ib_create_protection_domain(lnk);
	180	if (rc)
	181	goto free_link_mem;
	182	rc = smc_ib_create_queue_pair(lnk);
	183	if (rc)
	184	goto dealloc_pd;
	185	rc = smc_wr_create_link(lnk);
	186	if (rc)
	187	goto destroy_qp;
9bf9abea UB	188	init_completion(&lnk->llc_confirm);
9bf9abea UB	189	init_completion(&lnk->llc_confirm_resp);
0cfdd8f9 UB	190
	191	smc->conn.lgr = lgr;
	192	rwlock_init(&lgr->conns_lock);
	193	spin_lock_bh(&smc_lgr_list.lock);
	194	list_add(&lgr->list, &smc_lgr_list.list);
	195	spin_unlock_bh(&smc_lgr_list.lock);
f38ba179 UB	196	return 0;
f38ba179 UB	197
bd4ad577 UB	198	destroy_qp:
	199	smc_ib_destroy_queue_pair(lnk);
	200	dealloc_pd:
	201	smc_ib_dealloc_protection_domain(lnk);
	202	free_link_mem:
	203	smc_wr_free_link_mem(lnk);
f38ba179 UB	204	free_lgr:
f38ba179 UB	205	kfree(lgr);
0cfdd8f9 UB	206	out:
	207	return rc;
	208	}
	209
cd6851f3 UB	210	static void smc_sndbuf_unuse(struct smc_connection *conn)
	211	{
	212	if (conn->sndbuf_desc) {
	213	conn->sndbuf_desc->used = 0;
	214	conn->sndbuf_size = 0;
	215	}
	216	}
	217
	218	static void smc_rmb_unuse(struct smc_connection *conn)
	219	{
	220	if (conn->rmb_desc) {
	221	conn->rmb_desc->used = 0;
	222	conn->rmbe_size = 0;
	223	}
	224	}
	225
0cfdd8f9 UB	226	/* remove a finished connection from its link group */
	227	void smc_conn_free(struct smc_connection *conn)
	228	{
	229	struct smc_link_group *lgr = conn->lgr;
	230
	231	if (!lgr)
	232	return;
5f08318f	233	smc_cdc_tx_dismiss_slots(conn);
0cfdd8f9	234	smc_lgr_unregister_conn(conn);
cd6851f3 UB	235	smc_rmb_unuse(conn);
cd6851f3 UB	236	smc_sndbuf_unuse(conn);
0cfdd8f9 UB	237	}
	238
	239	static void smc_link_clear(struct smc_link *lnk)
	240	{
	241	lnk->peer_qpn = 0;
bd4ad577	242	smc_ib_modify_qp_reset(lnk);
f38ba179	243	smc_wr_free_link(lnk);
bd4ad577 UB	244	smc_ib_destroy_queue_pair(lnk);
bd4ad577 UB	245	smc_ib_dealloc_protection_domain(lnk);
f38ba179	246	smc_wr_free_link_mem(lnk);
0cfdd8f9 UB	247	}
0cfdd8f9 UB	248
cd6851f3 UB	249	static void smc_lgr_free_sndbufs(struct smc_link_group *lgr)
	250	{
	251	struct smc_buf_desc sndbuf_desc, bf_desc;
	252	int i;
	253
	254	for (i = 0; i < SMC_RMBE_SIZES; i++) {
	255	list_for_each_entry_safe(sndbuf_desc, bf_desc, &lgr->sndbufs[i],
	256	list) {
bd4ad577 UB	257	list_del(&sndbuf_desc->list);
	258	smc_ib_buf_unmap(lgr->lnk[SMC_SINGLE_LINK].smcibdev,
	259	smc_uncompress_bufsize(i),
	260	sndbuf_desc, DMA_TO_DEVICE);
cd6851f3 UB	261	kfree(sndbuf_desc->cpu_addr);
	262	kfree(sndbuf_desc);
	263	}
	264	}
	265	}
	266
	267	static void smc_lgr_free_rmbs(struct smc_link_group *lgr)
	268	{
	269	struct smc_buf_desc rmb_desc, bf_desc;
bd4ad577	270	struct smc_link *lnk = &lgr->lnk[SMC_SINGLE_LINK];
cd6851f3 UB	271	int i;
	272
	273	for (i = 0; i < SMC_RMBE_SIZES; i++) {
	274	list_for_each_entry_safe(rmb_desc, bf_desc, &lgr->rmbs[i],
	275	list) {
bd4ad577 UB	276	list_del(&rmb_desc->list);
	277	smc_ib_buf_unmap(lnk->smcibdev,
	278	smc_uncompress_bufsize(i),
	279	rmb_desc, DMA_FROM_DEVICE);
cd6851f3 UB	280	kfree(rmb_desc->cpu_addr);
	281	kfree(rmb_desc);
	282	}
	283	}
	284	}
	285
0cfdd8f9 UB	286	/* remove a link group */
	287	void smc_lgr_free(struct smc_link_group *lgr)
	288	{
cd6851f3 UB	289	smc_lgr_free_rmbs(lgr);
cd6851f3 UB	290	smc_lgr_free_sndbufs(lgr);
0cfdd8f9 UB	291	smc_link_clear(&lgr->lnk[SMC_SINGLE_LINK]);
	292	kfree(lgr);
	293	}
	294
	295	/* terminate linkgroup abnormally */
	296	void smc_lgr_terminate(struct smc_link_group *lgr)
	297	{
	298	struct smc_connection *conn;
b38d7324	299	struct smc_sock *smc;
0cfdd8f9 UB	300	struct rb_node *node;
	301
	302	spin_lock_bh(&smc_lgr_list.lock);
	303	if (list_empty(&lgr->list)) {
	304	/* termination already triggered */
	305	spin_unlock_bh(&smc_lgr_list.lock);
	306	return;
	307	}
	308	/* do not use this link group for new connections */
	309	list_del_init(&lgr->list);
	310	spin_unlock_bh(&smc_lgr_list.lock);
	311
	312	write_lock_bh(&lgr->conns_lock);
	313	node = rb_first(&lgr->conns_all);
	314	while (node) {
	315	conn = rb_entry(node, struct smc_connection, alert_node);
b38d7324 UB	316	smc = container_of(conn, struct smc_sock, conn);
b38d7324 UB	317	sock_hold(&smc->sk);
0cfdd8f9	318	__smc_lgr_unregister_conn(conn);
46c28dbd	319	schedule_work(&conn->close_work);
b38d7324	320	sock_put(&smc->sk);
0cfdd8f9 UB	321	node = rb_first(&lgr->conns_all);
	322	}
	323	write_unlock_bh(&lgr->conns_lock);
0cfdd8f9 UB	324	}
	325
	326	/* Determine vlan of internal TCP socket.
	327	* @vlan_id: address to store the determined vlan id into
	328	*/
	329	static int smc_vlan_by_tcpsk(struct socket clcsock, unsigned short vlan_id)
	330	{
	331	struct dst_entry *dst = sk_dst_get(clcsock->sk);
	332	int rc = 0;
	333
	334	*vlan_id = 0;
	335	if (!dst) {
	336	rc = -ENOTCONN;
	337	goto out;
	338	}
	339	if (!dst->dev) {
	340	rc = -ENODEV;
	341	goto out_rel;
	342	}
	343
	344	if (is_vlan_dev(dst->dev))
	345	*vlan_id = vlan_dev_vlan_id(dst->dev);
	346
	347	out_rel:
	348	dst_release(dst);
	349	out:
	350	return rc;
	351	}
	352
	353	/* determine the link gid matching the vlan id of the link group */
	354	static int smc_link_determine_gid(struct smc_link_group *lgr)
	355	{
	356	struct smc_link *lnk = &lgr->lnk[SMC_SINGLE_LINK];
	357	struct ib_gid_attr gattr;
	358	union ib_gid gid;
	359	int i;
	360
	361	if (!lgr->vlan_id) {
	362	lnk->gid = lnk->smcibdev->gid[lnk->ibport - 1];
	363	return 0;
	364	}
	365
	366	for (i = 0; i < lnk->smcibdev->pattr[lnk->ibport - 1].gid_tbl_len;
	367	i++) {
	368	if (ib_query_gid(lnk->smcibdev->ibdev, lnk->ibport, i, &gid,
	369	&gattr))
	370	continue;
	371	if (gattr.ndev &&
	372	(vlan_dev_vlan_id(gattr.ndev) == lgr->vlan_id)) {
	373	lnk->gid = gid;
	374	return 0;
	375	}
	376	}
	377	return -ENODEV;
	378	}
	379
	380	/* create a new SMC connection (and a new link group if necessary) */
	381	int smc_conn_create(struct smc_sock *smc, __be32 peer_in_addr,
	382	struct smc_ib_device *smcibdev, u8 ibport,
	383	struct smc_clc_msg_local *lcl, int srv_first_contact)
	384	{
	385	struct smc_connection *conn = &smc->conn;
	386	struct smc_link_group *lgr;
	387	unsigned short vlan_id;
388	enum smc_lgr_role role;
389	int local_contact = SMC_FIRST_CONTACT;
390	int rc = 0;
391
392	role = smc->listen_smc ? SMC_SERV : SMC_CLNT;
393	rc = smc_vlan_by_tcpsk(smc->clcsock, &vlan_id);
394	if (rc)
395	return rc;
396
397	if ((role == SMC_CLNT) && srv_first_contact)
398	/* create new link group as well */
399	goto create;
400
401	/* determine if an existing link group can be reused */
402	spin_lock_bh(&smc_lgr_list.lock);
403	list_for_each_entry(lgr, &smc_lgr_list.list, list) {
404	write_lock_bh(&lgr->conns_lock);
405	if (!memcmp(lgr->peer_systemid, lcl->id_for_peer,
406	SMC_SYSTEMID_LEN) &&
407	!memcmp(lgr->lnk[SMC_SINGLE_LINK].peer_gid, &lcl->gid,
408	SMC_GID_SIZE) &&
409	!memcmp(lgr->lnk[SMC_SINGLE_LINK].peer_mac, lcl->mac,
410	sizeof(lcl->mac)) &&
411	!lgr->sync_err &&
412	(lgr->role == role) &&
cd6851f3 UB	413	(lgr->vlan_id == vlan_id) &&
	414	((role == SMC_CLNT) \|\|
	415	(lgr->conns_num < SMC_RMBS_PER_LGR_MAX))) {
0cfdd8f9 UB	416	/* link group found */
	417	local_contact = SMC_REUSE_CONTACT;
	418	conn->lgr = lgr;
	419	smc_lgr_register_conn(conn); /* add smc conn to lgr */
	420	write_unlock_bh(&lgr->conns_lock);
	421	break;
	422	}
	423	write_unlock_bh(&lgr->conns_lock);
	424	}
	425	spin_unlock_bh(&smc_lgr_list.lock);
	426
	427	if (role == SMC_CLNT && !srv_first_contact &&
	428	(local_contact == SMC_FIRST_CONTACT)) {
	429	/* Server reuses a link group, but Client wants to start
	430	* a new one
	431	* send out_of_sync decline, reason synchr. error
	432	*/
	433	return -ENOLINK;
	434	}
	435
	436	create:
	437	if (local_contact == SMC_FIRST_CONTACT) {
	438	rc = smc_lgr_create(smc, peer_in_addr, smcibdev, ibport,
	439	lcl->id_for_peer, vlan_id);
	440	if (rc)
	441	goto out;
	442	smc_lgr_register_conn(conn); /* add smc conn to lgr */
	443	rc = smc_link_determine_gid(conn->lgr);
	444	}
5f08318f UB	445	conn->local_tx_ctrl.common.type = SMC_CDC_MSG_TYPE;
	446	conn->local_tx_ctrl.len = sizeof(struct smc_cdc_msg);
	447	#ifndef KERNEL_HAS_ATOMIC64
	448	spin_lock_init(&conn->acurs_lock);
	449	#endif
0cfdd8f9 UB	450
	451	out:
	452	return rc ? rc : local_contact;
	453	}
cd6851f3 UB	454
	455	/* try to reuse a sndbuf description slot of the sndbufs list for a certain
	456	* buf_size; if not available, return NULL
	457	*/
	458	static inline
	459	struct smc_buf_desc smc_sndbuf_get_slot(struct smc_link_group lgr,
	460	int compressed_bufsize)
	461	{
	462	struct smc_buf_desc *sndbuf_slot;
	463
	464	read_lock_bh(&lgr->sndbufs_lock);
	465	list_for_each_entry(sndbuf_slot, &lgr->sndbufs[compressed_bufsize],
	466	list) {
	467	if (cmpxchg(&sndbuf_slot->used, 0, 1) == 0) {
	468	read_unlock_bh(&lgr->sndbufs_lock);
	469	return sndbuf_slot;
	470	}
	471	}
	472	read_unlock_bh(&lgr->sndbufs_lock);
	473	return NULL;
	474	}
	475
	476	/* try to reuse an rmb description slot of the rmbs list for a certain
	477	* rmbe_size; if not available, return NULL
	478	*/
	479	static inline
	480	struct smc_buf_desc smc_rmb_get_slot(struct smc_link_group lgr,
	481	int compressed_bufsize)
	482	{
	483	struct smc_buf_desc *rmb_slot;
	484
	485	read_lock_bh(&lgr->rmbs_lock);
	486	list_for_each_entry(rmb_slot, &lgr->rmbs[compressed_bufsize],
	487	list) {
	488	if (cmpxchg(&rmb_slot->used, 0, 1) == 0) {
	489	read_unlock_bh(&lgr->rmbs_lock);
	490	return rmb_slot;
	491	}
	492	}
	493	read_unlock_bh(&lgr->rmbs_lock);
	494	return NULL;
	495	}
	496
952310cc UB	497	/* one of the conditions for announcing a receiver's current window size is
	498	* that it "results in a minimum increase in the window size of 10% of the
	499	* receive buffer space" [RFC7609]
	500	*/
	501	static inline int smc_rmb_wnd_update_limit(int rmbe_size)
	502	{
	503	return min_t(int, rmbe_size / 10, SOCK_MIN_SNDBUF / 2);
	504	}
	505
cd6851f3 UB	506	/* create the tx buffer for an SMC socket */
	507	int smc_sndbuf_create(struct smc_sock *smc)
	508	{
	509	struct smc_connection *conn = &smc->conn;
	510	struct smc_link_group *lgr = conn->lgr;
cd6851f3	511	struct smc_buf_desc *sndbuf_desc;
c45abf31	512	int bufsize, bufsize_short;
cd6851f3 UB	513	int rc;
	514
	515	/* use socket send buffer size (w/o overhead) as start value */
c45abf31 UB	516	for (bufsize_short = smc_compress_bufsize(smc->sk.sk_sndbuf / 2);
	517	bufsize_short >= 0; bufsize_short--) {
	518	bufsize = smc_uncompress_bufsize(bufsize_short);
cd6851f3	519	/* check for reusable sndbuf_slot in the link group */
c45abf31	520	sndbuf_desc = smc_sndbuf_get_slot(lgr, bufsize_short);
cd6851f3	521	if (sndbuf_desc) {
c45abf31	522	memset(sndbuf_desc->cpu_addr, 0, bufsize);
cd6851f3 UB	523	break; /* found reusable slot */
	524	}
	525	/* try to alloc a new send buffer */
	526	sndbuf_desc = kzalloc(sizeof(*sndbuf_desc), GFP_KERNEL);
	527	if (!sndbuf_desc)
	528	break; /* give up with -ENOMEM */
c45abf31	529	sndbuf_desc->cpu_addr = kzalloc(bufsize,
cd6851f3 UB	530	GFP_KERNEL \| __GFP_NOWARN \|
	531	__GFP_NOMEMALLOC \|
	532	__GFP_NORETRY);
	533	if (!sndbuf_desc->cpu_addr) {
	534	kfree(sndbuf_desc);
cdaf25df	535	sndbuf_desc = NULL;
cd6851f3 UB	536	/* if send buffer allocation has failed,
	537	* try a smaller one
	538	*/
	539	continue;
	540	}
	541	rc = smc_ib_buf_map(lgr->lnk[SMC_SINGLE_LINK].smcibdev,
c45abf31	542	bufsize, sndbuf_desc, DMA_TO_DEVICE);
cd6851f3 UB	543	if (rc) {
	544	kfree(sndbuf_desc->cpu_addr);
	545	kfree(sndbuf_desc);
cdaf25df	546	sndbuf_desc = NULL;
cd6851f3 UB	547	continue; /* if mapping failed, try smaller one */
	548	}
	549	sndbuf_desc->used = 1;
	550	write_lock_bh(&lgr->sndbufs_lock);
c45abf31	551	list_add(&sndbuf_desc->list, &lgr->sndbufs[bufsize_short]);
cd6851f3 UB	552	write_unlock_bh(&lgr->sndbufs_lock);
	553	break;
	554	}
	555	if (sndbuf_desc && sndbuf_desc->cpu_addr) {
	556	conn->sndbuf_desc = sndbuf_desc;
c45abf31 UB	557	conn->sndbuf_size = bufsize;
	558	smc->sk.sk_sndbuf = bufsize * 2;
	559	atomic_set(&conn->sndbuf_space, bufsize);
cd6851f3 UB	560	return 0;
	561	} else {
	562	return -ENOMEM;
	563	}
	564	}
	565
	566	/* create the RMB for an SMC socket (even though the SMC protocol
	567	* allows more than one RMB-element per RMB, the Linux implementation
	568	* uses just one RMB-element per RMB, i.e. uses an extra RMB for every
	569	* connection in a link group
	570	*/
	571	int smc_rmb_create(struct smc_sock *smc)
	572	{
	573	struct smc_connection *conn = &smc->conn;
	574	struct smc_link_group *lgr = conn->lgr;
cd6851f3	575	struct smc_buf_desc *rmb_desc;
c45abf31	576	int bufsize, bufsize_short;
cd6851f3 UB	577	int rc;
	578
	579	/* use socket recv buffer size (w/o overhead) as start value */
c45abf31 UB	580	for (bufsize_short = smc_compress_bufsize(smc->sk.sk_rcvbuf / 2);
	581	bufsize_short >= 0; bufsize_short--) {
	582	bufsize = smc_uncompress_bufsize(bufsize_short);
cd6851f3	583	/* check for reusable rmb_slot in the link group */
c45abf31	584	rmb_desc = smc_rmb_get_slot(lgr, bufsize_short);
cd6851f3	585	if (rmb_desc) {
c45abf31	586	memset(rmb_desc->cpu_addr, 0, bufsize);
cd6851f3 UB	587	break; /* found reusable slot */
	588	}
	589	/* try to alloc a new RMB */
	590	rmb_desc = kzalloc(sizeof(*rmb_desc), GFP_KERNEL);
	591	if (!rmb_desc)
	592	break; /* give up with -ENOMEM */
c45abf31	593	rmb_desc->cpu_addr = kzalloc(bufsize,
cd6851f3 UB	594	GFP_KERNEL \| __GFP_NOWARN \|
	595	__GFP_NOMEMALLOC \|
	596	__GFP_NORETRY);
	597	if (!rmb_desc->cpu_addr) {
	598	kfree(rmb_desc);
cdaf25df	599	rmb_desc = NULL;
cd6851f3 UB	600	/* if RMB allocation has failed,
	601	* try a smaller one
	602	*/
	603	continue;
	604	}
	605	rc = smc_ib_buf_map(lgr->lnk[SMC_SINGLE_LINK].smcibdev,
c45abf31	606	bufsize, rmb_desc, DMA_FROM_DEVICE);
cd6851f3 UB	607	if (rc) {
	608	kfree(rmb_desc->cpu_addr);
	609	kfree(rmb_desc);
cdaf25df	610	rmb_desc = NULL;
cd6851f3 UB	611	continue; /* if mapping failed, try smaller one */
cd6851f3 UB	612	}
263eec9b UB	613	rmb_desc->rkey[SMC_SINGLE_LINK] =
263eec9b UB	614	lgr->lnk[SMC_SINGLE_LINK].roce_pd->unsafe_global_rkey;
cd6851f3 UB	615	rmb_desc->used = 1;
cd6851f3 UB	616	write_lock_bh(&lgr->rmbs_lock);
c45abf31	617	list_add(&rmb_desc->list, &lgr->rmbs[bufsize_short]);
cd6851f3 UB	618	write_unlock_bh(&lgr->rmbs_lock);
	619	break;
	620	}
	621	if (rmb_desc && rmb_desc->cpu_addr) {
	622	conn->rmb_desc = rmb_desc;
c45abf31 UB	623	conn->rmbe_size = bufsize;
	624	conn->rmbe_size_short = bufsize_short;
	625	smc->sk.sk_rcvbuf = bufsize * 2;
5f08318f	626	atomic_set(&conn->bytes_to_rcv, 0);
c45abf31	627	conn->rmbe_update_limit = smc_rmb_wnd_update_limit(bufsize);
cd6851f3 UB	628	return 0;
	629	} else {
	630	return -ENOMEM;
	631	}
	632	}
bd4ad577 UB	633
	634	static inline int smc_rmb_reserve_rtoken_idx(struct smc_link_group *lgr)
	635	{
	636	int i;
	637
	638	for_each_clear_bit(i, lgr->rtokens_used_mask, SMC_RMBS_PER_LGR_MAX) {
	639	if (!test_and_set_bit(i, lgr->rtokens_used_mask))
	640	return i;
	641	}
	642	return -ENOSPC;
	643	}
	644
	645	/* save rkey and dma_addr received from peer during clc handshake */
	646	int smc_rmb_rtoken_handling(struct smc_connection *conn,
	647	struct smc_clc_msg_accept_confirm *clc)
	648	{
	649	u64 dma_addr = be64_to_cpu(clc->rmb_dma_addr);
	650	struct smc_link_group *lgr = conn->lgr;
	651	u32 rkey = ntohl(clc->rmb_rkey);
	652	int i;
	653
	654	for (i = 0; i < SMC_RMBS_PER_LGR_MAX; i++) {
	655	if ((lgr->rtokens[i][SMC_SINGLE_LINK].rkey == rkey) &&
263eec9b	656	(lgr->rtokens[i][SMC_SINGLE_LINK].dma_addr == dma_addr) &&
bd4ad577 UB	657	test_bit(i, lgr->rtokens_used_mask)) {
	658	conn->rtoken_idx = i;
	659	return 0;
	660	}
	661	}
	662	conn->rtoken_idx = smc_rmb_reserve_rtoken_idx(lgr);
	663	if (conn->rtoken_idx < 0)
	664	return conn->rtoken_idx;
	665	lgr->rtokens[conn->rtoken_idx][SMC_SINGLE_LINK].rkey = rkey;
	666	lgr->rtokens[conn->rtoken_idx][SMC_SINGLE_LINK].dma_addr = dma_addr;
	667	return 0;
	668	}