-/* -*- mode: c; c-basic-offset: 8; indent-tabs-mode: nil; -*-
- * vim:expandtab:shiftwidth=8:tabstop=8:
+/*
+ * GPL HEADER START
*
- * Copyright (C) 2006 Cluster File Systems, Inc.
- * Author: Eric Barton <eric@bartonsoftware.com>
+ * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
*
- * This file is part of Lustre, http://www.lustre.org.
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 only,
+ * as published by the Free Software Foundation.
*
- * Lustre is free software; you can redistribute it and/or
- * modify it under the terms of version 2 of the GNU General Public
- * License as published by the Free Software Foundation.
+ * This program is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ * General Public License version 2 for more details (a copy is included
+ * in the LICENSE file that accompanied this code).
*
- * Lustre is distributed in the hope that it will be useful,
- * but WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
- * GNU General Public License for more details.
+ * You should have received a copy of the GNU General Public License
+ * version 2 along with this program; If not, see
+ * http://www.sun.com/software/products/lustre/docs/GPLv2.pdf
*
- * You should have received a copy of the GNU General Public License
- * along with Lustre; if not, write to the Free Software
- * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ * Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara,
+ * CA 95054 USA or visit www.sun.com if you need additional information or
+ * have any questions.
*
+ * GPL HEADER END
+ */
+/*
+ * Copyright (c) 2007, 2010, Oracle and/or its affiliates. All rights reserved.
+ * Use is subject to license terms.
+ *
+ * Copyright (c) 2011, 2015, Intel Corporation.
+ */
+/*
+ * This file is part of Lustre, http://www.lustre.org/
+ * Lustre is a trademark of Sun Microsystems, Inc.
+ *
+ * lnet/klnds/o2iblnd/o2iblnd.h
+ *
+ * Author: Eric Barton <eric@bartonsoftware.com>
*/
-#ifndef EXPORT_SYMTAB
-# define EXPORT_SYMTAB
-#endif
-#ifdef HAVE_KERNEL_CONFIG_H
-#include <linux/config.h>
+#ifdef HAVE_COMPAT_RDMA
+#include <linux/compat-2.6.h>
#endif
+
+#include <linux/version.h>
#include <linux/module.h>
#include <linux/kernel.h>
+#include <linux/kthread.h>
#include <linux/mm.h>
#include <linux/string.h>
#include <linux/stat.h>
#include <linux/errno.h>
-#include <linux/smp_lock.h>
#include <linux/unistd.h>
#include <linux/uio.h>
-#include <asm/system.h>
#include <asm/uaccess.h>
#include <asm/io.h>
#include <linux/list.h>
#include <linux/kmod.h>
#include <linux/sysctl.h>
-#include <linux/random.h>
+#include <linux/pci.h>
+#if LINUX_VERSION_CODE > KERNEL_VERSION(2,6,32)
+#include <linux/pci-dma.h>
+#endif
#include <net/sock.h>
#include <linux/in.h>
+#include <rdma/rdma_cm.h>
+#include <rdma/ib_cm.h>
+#include <rdma/ib_verbs.h>
+#include <rdma/ib_fmr_pool.h>
+
#define DEBUG_SUBSYSTEM S_LND
-#include <libcfs/kp30.h>
+#include <libcfs/libcfs.h>
#include <lnet/lnet.h>
#include <lnet/lib-lnet.h>
-#if !HAVE_GFP_T
-typedef int gfp_t;
-#endif
+#define IBLND_PEER_HASH_SIZE 101 /* # peer lists */
+/* # scheduler loops before reschedule */
+#define IBLND_RESCHED 100
-#include <rdma/rdma_cm.h>
-#include <rdma/ib_cm.h>
-#include <rdma/ib_verbs.h>
-#include <rdma/ib_fmr_pool.h>
+#define IBLND_N_SCHED 2
+#define IBLND_N_SCHED_HIGH 4
-/* tunables fixed at compile time */
-#ifdef CONFIG_SMP
-# define IBLND_N_SCHED num_online_cpus() /* # schedulers */
-#else
-# define IBLND_N_SCHED 1 /* # schedulers */
+typedef struct
+{
+ int *kib_dev_failover; /* HCA failover */
+ unsigned int *kib_service; /* IB service number */
+ int *kib_min_reconnect_interval; /* first failed connection retry... */
+ int *kib_max_reconnect_interval; /* ...exponentially increasing to this */
+ int *kib_cksum; /* checksum kib_msg_t? */
+ int *kib_timeout; /* comms timeout (seconds) */
+ int *kib_keepalive; /* keepalive timeout (seconds) */
+ int *kib_ntx; /* # tx descs */
+ int *kib_credits; /* # concurrent sends */
+ int *kib_peertxcredits; /* # concurrent sends to 1 peer */
+ int *kib_peerrtrcredits; /* # per-peer router buffer credits */
+ int *kib_peercredits_hiw; /* # when eagerly to return credits */
+ int *kib_peertimeout; /* seconds to consider peer dead */
+ char **kib_default_ipif; /* default IPoIB interface */
+ int *kib_retry_count;
+ int *kib_rnr_retry_count;
+ int *kib_concurrent_sends; /* send work queue sizing */
+ int *kib_ib_mtu; /* IB MTU */
+ int *kib_map_on_demand; /* map-on-demand if RD has more fragments
+ * than this value, 0 disable map-on-demand */
+ int *kib_fmr_pool_size; /* # FMRs in pool */
+ int *kib_fmr_flush_trigger; /* When to trigger FMR flush */
+ int *kib_fmr_cache; /* enable FMR pool cache? */
+#if defined(CONFIG_SYSCTL) && !CFS_SYSFS_MODULE_PARM
+ struct ctl_table_header *kib_sysctl; /* sysctl interface */
#endif
+ int *kib_require_priv_port;/* accept only privileged ports */
+ int *kib_use_priv_port; /* use privileged port for active connect */
+ /* # threads on each CPT */
+ int *kib_nscheds;
+} kib_tunables_t;
-#define IBLND_PEER_HASH_SIZE 101 /* # peer lists */
-#define IBLND_RESCHED 100 /* # scheduler loops before reschedule */
-#define IBLND_MSG_QUEUE_SIZE 8 /* # messages/RDMAs in-flight */
-#define IBLND_CREDIT_HIGHWATER 7 /* when eagerly to return credits */
-#define IBLND_MSG_SIZE (4<<10) /* max size of queued messages (inc hdr) */
+extern kib_tunables_t kiblnd_tunables;
+
+#define IBLND_MSG_QUEUE_SIZE_V1 8 /* V1 only : # messages/RDMAs in-flight */
+#define IBLND_CREDIT_HIGHWATER_V1 7 /* V1 only : when eagerly to return credits */
+
+#define IBLND_CREDITS_DEFAULT 8 /* default # of peer credits */
+#define IBLND_CREDITS_MAX ((typeof(((kib_msg_t*) 0)->ibm_credits)) - 1) /* Max # of peer credits */
-#define IBLND_MAP_ON_DEMAND 0
-#if IBLND_MAP_ON_DEMAND
-# define IBLND_MAX_RDMA_FRAGS 1
+#define IBLND_MSG_QUEUE_SIZE(v) ((v) == IBLND_MSG_VERSION_1 ? \
+ IBLND_MSG_QUEUE_SIZE_V1 : \
+ *kiblnd_tunables.kib_peertxcredits) /* # messages/RDMAs in-flight */
+#define IBLND_CREDITS_HIGHWATER(v) ((v) == IBLND_MSG_VERSION_1 ? \
+ IBLND_CREDIT_HIGHWATER_V1 : \
+ *kiblnd_tunables.kib_peercredits_hiw) /* when eagerly to return credits */
+
+#ifdef HAVE_RDMA_CREATE_ID_4ARG
+#define kiblnd_rdma_create_id(cb, dev, ps, qpt) rdma_create_id(cb, dev, ps, qpt)
#else
-# define IBLND_MAX_RDMA_FRAGS LNET_MAX_IOV
+#define kiblnd_rdma_create_id(cb, dev, ps, qpt) rdma_create_id(cb, dev, ps)
#endif
+static inline int
+kiblnd_concurrent_sends_v1(void)
+{
+ if (*kiblnd_tunables.kib_concurrent_sends > IBLND_MSG_QUEUE_SIZE_V1 * 2)
+ return IBLND_MSG_QUEUE_SIZE_V1 * 2;
+
+ if (*kiblnd_tunables.kib_concurrent_sends < IBLND_MSG_QUEUE_SIZE_V1 / 2)
+ return IBLND_MSG_QUEUE_SIZE_V1 / 2;
+
+ return *kiblnd_tunables.kib_concurrent_sends;
+}
+
+#define IBLND_CONCURRENT_SENDS(v) ((v) == IBLND_MSG_VERSION_1 ? \
+ kiblnd_concurrent_sends_v1() : \
+ *kiblnd_tunables.kib_concurrent_sends)
+/* 2 OOB shall suffice for 1 keepalive and 1 returning credits */
+#define IBLND_OOB_CAPABLE(v) ((v) != IBLND_MSG_VERSION_1)
+#define IBLND_OOB_MSGS(v) (IBLND_OOB_CAPABLE(v) ? 2 : 0)
+
+#define IBLND_MSG_SIZE (4<<10) /* max size of queued messages (inc hdr) */
+#define IBLND_MAX_RDMA_FRAGS LNET_MAX_IOV /* max # of fragments supported */
+#define IBLND_CFG_RDMA_FRAGS (*kiblnd_tunables.kib_map_on_demand != 0 ? \
+ *kiblnd_tunables.kib_map_on_demand : \
+ IBLND_MAX_RDMA_FRAGS) /* max # of fragments configured by user */
+#define IBLND_RDMA_FRAGS(v) ((v) == IBLND_MSG_VERSION_1 ? \
+ IBLND_MAX_RDMA_FRAGS : IBLND_CFG_RDMA_FRAGS)
+
/************************/
/* derived constants... */
-
-/* TX messages (shared by all connections) */
-#define IBLND_TX_MSGS() (*kiblnd_tunables.kib_ntx)
-#define IBLND_TX_MSG_BYTES() (IBLND_TX_MSGS() * IBLND_MSG_SIZE)
-#define IBLND_TX_MSG_PAGES() ((IBLND_TX_MSG_BYTES() + PAGE_SIZE - 1)/PAGE_SIZE)
+/* Pools (shared by connections on each CPT) */
+/* These pools can grow at runtime, so don't need give a very large value */
+#define IBLND_TX_POOL 256
+#define IBLND_FMR_POOL 256
+#define IBLND_FMR_POOL_FLUSH 192
/* RX messages (per connection) */
-#define IBLND_RX_MSGS (IBLND_MSG_QUEUE_SIZE*2)
-#define IBLND_RX_MSG_BYTES (IBLND_RX_MSGS * IBLND_MSG_SIZE)
-#define IBLND_RX_MSG_PAGES ((IBLND_RX_MSG_BYTES + PAGE_SIZE - 1)/PAGE_SIZE)
-
-#define IBLND_CQ_ENTRIES() (IBLND_RX_MSGS + \
- (*kiblnd_tunables.kib_concurrent_sends) * \
- (1 + IBLND_MAX_RDMA_FRAGS))
+#define IBLND_RX_MSGS(c) \
+ ((c->ibc_queue_depth) * 2 + IBLND_OOB_MSGS(c->ibc_version))
+#define IBLND_RX_MSG_BYTES(c) (IBLND_RX_MSGS(c) * IBLND_MSG_SIZE)
+#define IBLND_RX_MSG_PAGES(c) \
+ ((IBLND_RX_MSG_BYTES(c) + PAGE_SIZE - 1) / PAGE_SIZE)
+
+/* WRs and CQEs (per connection) */
+#define IBLND_RECV_WRS(c) IBLND_RX_MSGS(c)
+#define IBLND_SEND_WRS(c) \
+ ((c->ibc_max_frags + 1) * IBLND_CONCURRENT_SENDS(c->ibc_version))
+#define IBLND_CQ_ENTRIES(c) (IBLND_RECV_WRS(c) + IBLND_SEND_WRS(c))
+
+struct kib_hca_dev;
+
+/* o2iblnd can run over aliased interface */
+#ifdef IFALIASZ
+#define KIB_IFNAME_SIZE IFALIASZ
+#else
+#define KIB_IFNAME_SIZE 256
+#endif
typedef struct
{
- unsigned int *kib_service; /* IB service number */
- int *kib_min_reconnect_interval; /* first failed connection retry... */
- int *kib_max_reconnect_interval; /* ...exponentially increasing to this */
- int *kib_cksum; /* checksum kib_msg_t? */
- int *kib_timeout; /* comms timeout (seconds) */
- int *kib_keepalive; /* keepalive timeout (seconds) */
- int *kib_ntx; /* # tx descs */
- int *kib_credits; /* # concurrent sends */
- int *kib_peercredits; /* # concurrent sends to 1 peer */
- char **kib_default_ipif; /* default IPoIB interface */
- int *kib_retry_count;
- int *kib_rnr_retry_count;
- int *kib_concurrent_sends; /* send work queue sizing */
- int *kib_ib_mtu; /* IB MTU */
-#if IBLND_MAP_ON_DEMAND
- int *kib_fmr_pool_size; /* # FMRs in pool */
- int *kib_fmr_flush_trigger; /* When to trigger FMR flush */
- int *kib_fmr_cache; /* enable FMR pool cache? */
-#endif
-#if CONFIG_SYSCTL && !CFS_SYSFS_MODULE_PARM
- struct ctl_table_header *kib_sysctl; /* sysctl interface */
-#endif
-} kib_tunables_t;
+ struct list_head ibd_list; /* chain on kib_devs */
+ struct list_head ibd_fail_list; /* chain on kib_failed_devs */
+ __u32 ibd_ifip; /* IPoIB interface IP */
+ /** IPoIB interface name */
+ char ibd_ifname[KIB_IFNAME_SIZE];
+ int ibd_nnets; /* # nets extant */
+
+ cfs_time_t ibd_next_failover;
+ /* # failover failures */
+ int ibd_failed_failover;
+ /* failover in progress */
+ unsigned int ibd_failover;
+ /* IPoIB interface is a bonding master */
+ unsigned int ibd_can_failover;
+ struct list_head ibd_nets;
+ struct kib_hca_dev *ibd_hdev;
+} kib_dev_t;
+
+typedef struct kib_hca_dev
+{
+ struct rdma_cm_id *ibh_cmid; /* listener cmid */
+ struct ib_device *ibh_ibdev; /* IB device */
+ int ibh_page_shift; /* page shift of current HCA */
+ int ibh_page_size; /* page size of current HCA */
+ __u64 ibh_page_mask; /* page mask of current HCA */
+ int ibh_mr_shift; /* bits shift of max MR size */
+ __u64 ibh_mr_size; /* size of MR */
+ struct ib_mr *ibh_mrs; /* global MR */
+ struct ib_pd *ibh_pd; /* PD */
+ kib_dev_t *ibh_dev; /* owner */
+ atomic_t ibh_ref; /* refcount */
+} kib_hca_dev_t;
+
+/** # of seconds to keep pool alive */
+#define IBLND_POOL_DEADLINE 300
+/** # of seconds to retry if allocation failed */
+#define IBLND_POOL_RETRY 1
typedef struct
{
- int ibp_npages; /* # pages */
- struct page *ibp_pages[0];
+ int ibp_npages; /* # pages */
+ struct page *ibp_pages[0]; /* page array */
} kib_pages_t;
-typedef struct
-{
- struct list_head ibd_list; /* chain on kib_devs */
- __u32 ibd_ifip; /* IPoIB interface IP */
- char ibd_ifname[32]; /* IPoIB interface name */
- int ibd_nnets; /* # nets extant */
+struct kib_pool;
+struct kib_poolset;
- struct rdma_cm_id *ibd_cmid; /* IB listener (bound to 1 device) */
- struct ib_pd *ibd_pd; /* PD for the device */
- struct ib_mr *ibd_mr; /* MR for non RDMA I/O */
-} kib_dev_t;
+typedef int (*kib_ps_pool_create_t)(struct kib_poolset *ps,
+ int inc, struct kib_pool **pp_po);
+typedef void (*kib_ps_pool_destroy_t)(struct kib_pool *po);
+typedef void (*kib_ps_node_init_t)(struct kib_pool *po, struct list_head *node);
+typedef void (*kib_ps_node_fini_t)(struct kib_pool *po, struct list_head *node);
-typedef struct
-{
- __u64 ibn_incarnation; /* my epoch */
- int ibn_init; /* initialisation state */
- int ibn_shutdown; /* shutting down? */
+struct kib_net;
- atomic_t ibn_npeers; /* # peers extant */
- atomic_t ibn_nconns; /* # connections extant */
+#define IBLND_POOL_NAME_LEN 32
- struct kib_tx *ibn_tx_descs; /* all the tx descriptors */
- kib_pages_t *ibn_tx_pages; /* premapped tx msg pages */
- struct list_head ibn_idle_txs; /* idle tx descriptors */
- spinlock_t ibn_tx_lock; /* serialise */
+typedef struct kib_poolset
+{
+ /* serialize */
+ spinlock_t ps_lock;
+ /* network it belongs to */
+ struct kib_net *ps_net;
+ /* pool set name */
+ char ps_name[IBLND_POOL_NAME_LEN];
+ /* list of pools */
+ struct list_head ps_pool_list;
+ /* failed pool list */
+ struct list_head ps_failed_pool_list;
+ /* time stamp for retry if failed to allocate */
+ cfs_time_t ps_next_retry;
+ /* is allocating new pool */
+ int ps_increasing;
+ /* new pool size */
+ int ps_pool_size;
+ /* CPT id */
+ int ps_cpt;
+
+ /* create a new pool */
+ kib_ps_pool_create_t ps_pool_create;
+ /* destroy a pool */
+ kib_ps_pool_destroy_t ps_pool_destroy;
+ /* initialize new allocated node */
+ kib_ps_node_init_t ps_node_init;
+ /* finalize node */
+ kib_ps_node_fini_t ps_node_fini;
+} kib_poolset_t;
+
+typedef struct kib_pool
+{
+ /* chain on pool list */
+ struct list_head po_list;
+ /* pre-allocated node */
+ struct list_head po_free_list;
+ /* pool_set of this pool */
+ kib_poolset_t *po_owner;
+ /* deadline of this pool */
+ cfs_time_t po_deadline;
+ /* # of elements in use */
+ int po_allocated;
+ /* pool is created on failed HCA */
+ int po_failed;
+ /* # of pre-allocated elements */
+ int po_size;
+} kib_pool_t;
-#if IBLND_MAP_ON_DEMAND
- struct ib_fmr_pool *ibn_fmrpool; /* FMR pool for RDMA I/O */
-#endif
+typedef struct {
+ kib_poolset_t tps_poolset; /* pool-set */
+ __u64 tps_next_tx_cookie; /* cookie of TX */
+} kib_tx_poolset_t;
- kib_dev_t *ibn_dev; /* underlying IB device */
-} kib_net_t;
+typedef struct {
+ kib_pool_t tpo_pool; /* pool */
+ struct kib_hca_dev *tpo_hdev; /* device for this pool */
+ struct kib_tx *tpo_tx_descs; /* all the tx descriptors */
+ kib_pages_t *tpo_tx_pages; /* premapped tx msg pages */
+} kib_tx_pool_t;
+
+typedef struct
+{
+ spinlock_t fps_lock; /* serialize */
+ struct kib_net *fps_net; /* IB network */
+ struct list_head fps_pool_list; /* FMR pool list */
+ struct list_head fps_failed_pool_list; /* FMR pool list */
+ __u64 fps_version; /* validity stamp */
+ int fps_cpt; /* CPT id */
+ int fps_pool_size;
+ int fps_flush_trigger;
+ /* is allocating new pool */
+ int fps_increasing;
+ /* time stamp for retry if failed to allocate */
+ cfs_time_t fps_next_retry;
+} kib_fmr_poolset_t;
typedef struct
{
- int kib_init; /* initialisation state */
- int kib_shutdown; /* shut down? */
- struct list_head kib_devs; /* IB devices extant */
- atomic_t kib_nthreads; /* # live threads */
- rwlock_t kib_global_lock; /* stabilize net/dev/peer/conn ops */
+ struct list_head fpo_list; /* chain on pool list */
+ struct kib_hca_dev *fpo_hdev; /* device for this pool */
+ kib_fmr_poolset_t *fpo_owner; /* owner of this pool */
+ struct ib_fmr_pool *fpo_fmr_pool; /* IB FMR pool */
+ cfs_time_t fpo_deadline; /* deadline of this pool */
+ int fpo_failed; /* fmr pool is failed */
+ int fpo_map_count; /* # of mapped FMR */
+} kib_fmr_pool_t;
+
+typedef struct {
+ struct ib_pool_fmr *fmr_pfmr; /* IB pool fmr */
+ kib_fmr_pool_t *fmr_pool; /* pool of FMR */
+} kib_fmr_t;
+
+typedef struct kib_net
+{
+ /* chain on kib_dev_t::ibd_nets */
+ struct list_head ibn_list;
+ __u64 ibn_incarnation;/* my epoch */
+ int ibn_init; /* initialisation state */
+ int ibn_shutdown; /* shutting down? */
- struct list_head *kib_peers; /* hash table of all my known peers */
- int kib_peer_hash_size; /* size of kib_peers */
+ atomic_t ibn_npeers; /* # peers extant */
+ atomic_t ibn_nconns; /* # connections extant */
- void *kib_connd; /* the connd task (serialisation assertions) */
- struct list_head kib_connd_conns; /* connections to setup/teardown */
- struct list_head kib_connd_zombies; /* connections with zero refcount */
- wait_queue_head_t kib_connd_waitq; /* connection daemon sleeps here */
- spinlock_t kib_connd_lock; /* serialise */
+ kib_tx_poolset_t **ibn_tx_ps; /* tx pool-set */
+ kib_fmr_poolset_t **ibn_fmr_ps; /* fmr pool-set */
- wait_queue_head_t kib_sched_waitq; /* schedulers sleep here */
- struct list_head kib_sched_conns; /* conns to check for rx completions */
- spinlock_t kib_sched_lock; /* serialise */
+ kib_dev_t *ibn_dev; /* underlying IB device */
+} kib_net_t;
- __u64 kib_next_tx_cookie; /* RDMA completion cookie */
- struct ib_qp_attr kib_error_qpa; /* QP->ERROR */
+#define KIB_THREAD_SHIFT 16
+#define KIB_THREAD_ID(cpt, tid) ((cpt) << KIB_THREAD_SHIFT | (tid))
+#define KIB_THREAD_CPT(id) ((id) >> KIB_THREAD_SHIFT)
+#define KIB_THREAD_TID(id) ((id) & ((1UL << KIB_THREAD_SHIFT) - 1))
+
+struct kib_sched_info {
+ /* serialise */
+ spinlock_t ibs_lock;
+ /* schedulers sleep here */
+ wait_queue_head_t ibs_waitq;
+ /* conns to check for rx completions */
+ struct list_head ibs_conns;
+ /* number of scheduler threads */
+ int ibs_nthreads;
+ /* max allowed scheduler threads */
+ int ibs_nthreads_max;
+ int ibs_cpt; /* CPT id */
+};
+
+typedef struct
+{
+ int kib_init; /* initialisation state */
+ int kib_shutdown; /* shut down? */
+ struct list_head kib_devs; /* IB devices extant */
+ /* list head of failed devices */
+ struct list_head kib_failed_devs;
+ /* schedulers sleep here */
+ wait_queue_head_t kib_failover_waitq;
+ atomic_t kib_nthreads; /* # live threads */
+ /* stabilize net/dev/peer/conn ops */
+ rwlock_t kib_global_lock;
+ /* hash table of all my known peers */
+ struct list_head *kib_peers;
+ /* size of kib_peers */
+ int kib_peer_hash_size;
+ /* the connd task (serialisation assertions) */
+ void *kib_connd;
+ /* connections to setup/teardown */
+ struct list_head kib_connd_conns;
+ /* connections with zero refcount */
+ struct list_head kib_connd_zombies;
+ /* connection daemon sleeps here */
+ wait_queue_head_t kib_connd_waitq;
+ spinlock_t kib_connd_lock; /* serialise */
+ struct ib_qp_attr kib_error_qpa; /* QP->ERROR */
+ /* percpt data for schedulers */
+ struct kib_sched_info **kib_scheds;
} kib_data_t;
#define IBLND_INIT_NOTHING 0
char ibim_payload[0]; /* piggy-backed payload */
} WIRE_ATTR kib_immediate_msg_t;
-#if IBLND_MAP_ON_DEMAND
-typedef struct
-{
- __u64 rd_addr; /* IO VMA address */
- __u32 rd_nob; /* # of bytes */
- __u32 rd_key; /* remote key */
-} WIRE_ATTR kib_rdma_desc_t;
-#else
typedef struct
{
__u32 rf_nob; /* # bytes this frag */
__u32 rd_nfrags; /* # fragments */
kib_rdma_frag_t rd_frags[0]; /* buffer frags */
} WIRE_ATTR kib_rdma_desc_t;
-#endif
-
+
typedef struct
{
lnet_hdr_t ibprm_hdr; /* portals header */
typedef struct
{
/* First 2 fields fixed FOR ALL TIME */
- __u32 ibm_magic; /* I'm an openibnal message */
+ __u32 ibm_magic; /* I'm an ibnal message */
__u16 ibm_version; /* this is my version number */
__u8 ibm_type; /* msg type */
#define IBLND_MSG_MAGIC LNET_PROTO_IB_MAGIC /* unique magic */
-#define IBLND_MSG_VERSION 0x11
+#define IBLND_MSG_VERSION_1 0x11
+#define IBLND_MSG_VERSION_2 0x12
+#define IBLND_MSG_VERSION IBLND_MSG_VERSION_2
#define IBLND_MSG_CONNREQ 0xc0 /* connection request */
#define IBLND_MSG_CONNACK 0xc1 /* connection acknowledge */
__u32 ibr_magic; /* sender's magic */
__u16 ibr_version; /* sender's version */
__u8 ibr_why; /* reject reason */
+ __u8 ibr_padding; /* padding */
+ __u64 ibr_incarnation; /* incarnation of peer */
+ kib_connparams_t ibr_cp; /* connection parameters */
} WIRE_ATTR kib_rej_t;
-
/* connection rejection reasons */
#define IBLND_REJECT_CONN_RACE 1 /* You lost connection race */
#define IBLND_REJECT_NO_RESOURCES 2 /* Out of memory/conns etc */
#define IBLND_REJECT_FATAL 3 /* Anything else */
+#define IBLND_REJECT_CONN_UNCOMPAT 4 /* incompatible version peer */
+#define IBLND_REJECT_CONN_STALE 5 /* stale peer */
+
+/* peer's rdma frags doesn't match mine */
+#define IBLND_REJECT_RDMA_FRAGS 6
+/* peer's msg queue size doesn't match mine */
+#define IBLND_REJECT_MSG_QUEUE_SIZE 7
+
/***********************************************************************/
typedef struct kib_rx /* receive message */
{
- struct list_head rx_list; /* queue for attention */
- struct kib_conn *rx_conn; /* owning conn */
- int rx_nob; /* # bytes received (-1 while posted) */
- enum ib_wc_status rx_status; /* completion status */
- kib_msg_t *rx_msg; /* message buffer (host vaddr) */
- __u64 rx_msgaddr; /* message buffer (I/O addr) */
- DECLARE_PCI_UNMAP_ADDR (rx_msgunmap); /* for dma_unmap_single() */
- struct ib_recv_wr rx_wrq; /* receive work item... */
- struct ib_sge rx_sge; /* ...and its memory */
+ /* queue for attention */
+ struct list_head rx_list;
+ /* owning conn */
+ struct kib_conn *rx_conn;
+ /* # bytes received (-1 while posted) */
+ int rx_nob;
+ /* completion status */
+ enum ib_wc_status rx_status;
+ /* message buffer (host vaddr) */
+ kib_msg_t *rx_msg;
+ /* message buffer (I/O addr) */
+ __u64 rx_msgaddr;
+ /* for dma_unmap_single() */
+ DECLARE_PCI_UNMAP_ADDR(rx_msgunmap);
+ /* receive work item... */
+ struct ib_recv_wr rx_wrq;
+ /* ...and its memory */
+ struct ib_sge rx_sge;
} kib_rx_t;
#define IBLND_POSTRX_DONT_POST 0 /* don't post */
typedef struct kib_tx /* transmit message */
{
- struct list_head tx_list; /* queue on idle_txs ibc_tx_queue etc. */
- struct kib_conn *tx_conn; /* owning conn */
- int tx_sending; /* # tx callbacks outstanding */
- int tx_queued; /* queued for sending */
- int tx_waiting; /* waiting for peer */
- int tx_status; /* LNET completion status */
- unsigned long tx_deadline; /* completion deadline */
- __u64 tx_cookie; /* completion cookie */
- lnet_msg_t *tx_lntmsg[2]; /* lnet msgs to finalize on completion */
- kib_msg_t *tx_msg; /* message buffer (host vaddr) */
- __u64 tx_msgaddr; /* message buffer (I/O addr) */
- DECLARE_PCI_UNMAP_ADDR (tx_msgunmap); /* for dma_unmap_single() */
- int tx_nwrq; /* # send work items */
-#if IBLND_MAP_ON_DEMAND
- struct ib_send_wr tx_wrq[2]; /* send work items... */
- struct ib_sge tx_sge[2]; /* ...and their memory */
- kib_rdma_desc_t tx_rd[1]; /* rdma descriptor */
- __u64 *tx_pages; /* rdma phys page addrs */
- struct ib_pool_fmr *tx_fmr; /* rdma mapping (mapped if != NULL) */
-#else
- struct ib_send_wr *tx_wrq; /* send work items... */
- struct ib_sge *tx_sge; /* ...and their memory */
- kib_rdma_desc_t *tx_rd; /* rdma descriptor */
- int tx_nfrags; /* # entries in... */
- struct scatterlist *tx_frags; /* dma_map_sg descriptor */
- int tx_dmadir; /* dma direction */
-#endif
+ /* queue on idle_txs ibc_tx_queue etc. */
+ struct list_head tx_list;
+ /* pool I'm from */
+ kib_tx_pool_t *tx_pool;
+ /* owning conn */
+ struct kib_conn *tx_conn;
+ /* # tx callbacks outstanding */
+ short tx_sending;
+ /* queued for sending */
+ short tx_queued;
+ /* waiting for peer */
+ short tx_waiting;
+ /* LNET completion status */
+ int tx_status;
+ /* completion deadline */
+ unsigned long tx_deadline;
+ /* completion cookie */
+ __u64 tx_cookie;
+ /* lnet msgs to finalize on completion */
+ lnet_msg_t *tx_lntmsg[2];
+ /* message buffer (host vaddr) */
+ kib_msg_t *tx_msg;
+ /* message buffer (I/O addr) */
+ __u64 tx_msgaddr;
+ /* for dma_unmap_single() */
+ DECLARE_PCI_UNMAP_ADDR(tx_msgunmap);
+ /* # send work items */
+ int tx_nwrq;
+ /* send work items... */
+ struct ib_send_wr *tx_wrq;
+ /* ...and their memory */
+ struct ib_sge *tx_sge;
+ /* rdma descriptor */
+ kib_rdma_desc_t *tx_rd;
+ /* # entries in... */
+ int tx_nfrags;
+ /* dma_map_sg descriptor */
+ struct scatterlist *tx_frags;
+ /* rdma phys page addrs */
+ __u64 *tx_pages;
+ /* FMR */
+ kib_fmr_t fmr;
+ /* dma direction */
+ int tx_dmadir;
} kib_tx_t;
typedef struct kib_connvars
typedef struct kib_conn
{
- struct kib_peer *ibc_peer; /* owning peer */
- struct list_head ibc_list; /* stash on peer's conn list */
- struct list_head ibc_sched_list; /* schedule for attention */
- __u64 ibc_incarnation; /* which instance of the peer */
- atomic_t ibc_refcount; /* # users */
- int ibc_state; /* what's happening */
- int ibc_nsends_posted; /* # uncompleted sends */
- int ibc_credits; /* # credits I have */
- int ibc_outstanding_credits; /* # credits to return */
- int ibc_reserved_credits;/* # ACK/DONE msg credits */
- int ibc_comms_error; /* set on comms error */
- int ibc_nrx:8; /* receive buffers owned */
- int ibc_scheduled:1; /* scheduled for attention */
- int ibc_ready:1; /* CQ callback fired */
- unsigned long ibc_last_send; /* time of last send */
- struct list_head ibc_early_rxs; /* rxs completed before ESTABLISHED */
- struct list_head ibc_tx_queue; /* sends that need a credit */
- struct list_head ibc_tx_queue_nocred;/* sends that don't need a credit */
- struct list_head ibc_tx_queue_rsrvd; /* sends that need to reserve an ACK/DONE msg */
- struct list_head ibc_active_txs; /* active tx awaiting completion */
- spinlock_t ibc_lock; /* serialise */
- kib_rx_t *ibc_rxs; /* the rx descs */
- kib_pages_t *ibc_rx_pages; /* premapped rx msg pages */
-
- struct rdma_cm_id *ibc_cmid; /* CM id */
- struct ib_cq *ibc_cq; /* completion queue */
-
- kib_connvars_t *ibc_connvars; /* in-progress connection state */
+ /* scheduler information */
+ struct kib_sched_info *ibc_sched;
+ /* owning peer */
+ struct kib_peer *ibc_peer;
+ /* HCA bound on */
+ kib_hca_dev_t *ibc_hdev;
+ /* stash on peer's conn list */
+ struct list_head ibc_list;
+ /* schedule for attention */
+ struct list_head ibc_sched_list;
+ /* version of connection */
+ __u16 ibc_version;
+ /* which instance of the peer */
+ __u64 ibc_incarnation;
+ /* # users */
+ atomic_t ibc_refcount;
+ /* what's happening */
+ int ibc_state;
+ /* # uncompleted sends */
+ int ibc_nsends_posted;
+ /* # uncompleted NOOPs */
+ int ibc_noops_posted;
+ /* # credits I have */
+ int ibc_credits;
+ /* # credits to return */
+ int ibc_outstanding_credits;
+ /* # ACK/DONE msg credits */
+ int ibc_reserved_credits;
+ /* set on comms error */
+ int ibc_comms_error;
+ /* connections queue depth */
+ __u16 ibc_queue_depth;
+ /* connections max frags */
+ __u16 ibc_max_frags;
+ /* receive buffers owned */
+ unsigned short ibc_nrx;
+ /** rejected by connection race */
+ unsigned short ibc_conn_race:1;
+ /* scheduled for attention */
+ unsigned short ibc_scheduled:1;
+ /* CQ callback fired */
+ unsigned short ibc_ready:1;
+ /* time of last send */
+ unsigned long ibc_last_send;
+ /** link chain for kiblnd_check_conns only */
+ struct list_head ibc_connd_list;
+ /** rxs completed before ESTABLISHED */
+ struct list_head ibc_early_rxs;
+ /** IBLND_MSG_NOOPs for IBLND_MSG_VERSION_1 */
+ struct list_head ibc_tx_noops;
+ /* sends that need a credit */
+ struct list_head ibc_tx_queue;
+ /* sends that don't need a credit */
+ struct list_head ibc_tx_queue_nocred;
+ /* sends that need to reserve an ACK/DONE msg */
+ struct list_head ibc_tx_queue_rsrvd;
+ /* active tx awaiting completion */
+ struct list_head ibc_active_txs;
+ /* serialise */
+ spinlock_t ibc_lock;
+ /* the rx descs */
+ kib_rx_t *ibc_rxs;
+ /* premapped rx msg pages */
+ kib_pages_t *ibc_rx_pages;
+
+ /* CM id */
+ struct rdma_cm_id *ibc_cmid;
+ /* completion queue */
+ struct ib_cq *ibc_cq;
+
+ /* in-progress connection state */
+ kib_connvars_t *ibc_connvars;
} kib_conn_t;
-#define IBLND_CONN_INIT 0 /* being intialised */
+#define IBLND_CONN_INIT 0 /* being initialised */
#define IBLND_CONN_ACTIVE_CONNECT 1 /* active sending req */
#define IBLND_CONN_PASSIVE_WAIT 2 /* passive waiting for rtu */
#define IBLND_CONN_ESTABLISHED 3 /* connection established */
typedef struct kib_peer
{
- struct list_head ibp_list; /* stash on global peer list */
- lnet_nid_t ibp_nid; /* who's on the other end(s) */
- lnet_ni_t *ibp_ni; /* LNet interface */
- atomic_t ibp_refcount; /* # users */
- struct list_head ibp_conns; /* all active connections */
- struct list_head ibp_tx_queue; /* msgs waiting for a conn */
- int ibp_connecting; /* current active connection attempts */
- int ibp_accepting; /* current passive connection attempts */
- int ibp_error; /* errno on closing this peer */
- cfs_time_t ibp_last_alive; /* when (in jiffies) I was last alive */
+ /* stash on global peer list */
+ struct list_head ibp_list;
+ /* who's on the other end(s) */
+ lnet_nid_t ibp_nid;
+ /* LNet interface */
+ lnet_ni_t *ibp_ni;
+ /* # users */
+ atomic_t ibp_refcount;
+ /* all active connections */
+ struct list_head ibp_conns;
+ /* msgs waiting for a conn */
+ struct list_head ibp_tx_queue;
+ /* version of peer */
+ __u16 ibp_version;
+ /* incarnation of peer */
+ __u64 ibp_incarnation;
+ /* current active connection attempts */
+ int ibp_connecting;
+ /* current passive connection attempts */
+ int ibp_accepting;
+ /* errno on closing this peer */
+ int ibp_error;
+ /* when (in jiffies) I was last alive */
+ cfs_time_t ibp_last_alive;
+ /* max map_on_demand */
+ __u16 ibp_max_frags;
+ /* max_peer_credits */
+ __u16 ibp_queue_depth;
} kib_peer_t;
-
extern kib_data_t kiblnd_data;
-extern kib_tunables_t kiblnd_tunables;
+
+extern void kiblnd_hdev_destroy(kib_hca_dev_t *hdev);
+
+static inline void
+kiblnd_hdev_addref_locked(kib_hca_dev_t *hdev)
+{
+ LASSERT(atomic_read(&hdev->ibh_ref) > 0);
+ atomic_inc(&hdev->ibh_ref);
+}
+
+static inline void
+kiblnd_hdev_decref(kib_hca_dev_t *hdev)
+{
+ LASSERT(atomic_read(&hdev->ibh_ref) > 0);
+ if (atomic_dec_and_test(&hdev->ibh_ref))
+ kiblnd_hdev_destroy(hdev);
+}
+
+static inline int
+kiblnd_dev_can_failover(kib_dev_t *dev)
+{
+ if (!list_empty(&dev->ibd_fail_list)) /* already scheduled */
+ return 0;
+
+ if (*kiblnd_tunables.kib_dev_failover == 0) /* disabled */
+ return 0;
+
+ if (*kiblnd_tunables.kib_dev_failover > 1) /* force failover */
+ return 1;
+
+ return dev->ibd_can_failover;
+}
#define kiblnd_conn_addref(conn) \
do { \
CDEBUG(D_NET, "conn[%p] (%d)++\n", \
- (conn), atomic_read(&(conn)->ibc_refcount)); \
- LASSERT(atomic_read(&(conn)->ibc_refcount) > 0); \
- atomic_inc(&(conn)->ibc_refcount); \
+ (conn), atomic_read(&(conn)->ibc_refcount)); \
+ atomic_inc(&(conn)->ibc_refcount); \
} while (0)
-#define kiblnd_conn_decref(conn) \
-do { \
- unsigned long flags; \
- \
- CDEBUG(D_NET, "conn[%p] (%d)--\n", \
- (conn), atomic_read(&(conn)->ibc_refcount)); \
- LASSERT(atomic_read(&(conn)->ibc_refcount) > 0); \
- if (atomic_dec_and_test(&(conn)->ibc_refcount)) { \
- spin_lock_irqsave(&kiblnd_data.kib_connd_lock, flags); \
- list_add_tail(&(conn)->ibc_list, \
- &kiblnd_data.kib_connd_zombies); \
- wake_up(&kiblnd_data.kib_connd_waitq); \
- spin_unlock_irqrestore(&kiblnd_data.kib_connd_lock, flags); \
- } \
+#define kiblnd_conn_decref(conn) \
+do { \
+ unsigned long flags; \
+ \
+ CDEBUG(D_NET, "conn[%p] (%d)--\n", \
+ (conn), atomic_read(&(conn)->ibc_refcount)); \
+ LASSERT_ATOMIC_POS(&(conn)->ibc_refcount); \
+ if (atomic_dec_and_test(&(conn)->ibc_refcount)) { \
+ spin_lock_irqsave(&kiblnd_data.kib_connd_lock, flags); \
+ list_add_tail(&(conn)->ibc_list, \
+ &kiblnd_data.kib_connd_zombies); \
+ wake_up(&kiblnd_data.kib_connd_waitq); \
+ spin_unlock_irqrestore(&kiblnd_data.kib_connd_lock, flags);\
+ } \
} while (0)
#define kiblnd_peer_addref(peer) \
do { \
- CDEBUG(D_NET, "peer[%p] -> %s (%d)++\n", \
- (peer), libcfs_nid2str((peer)->ibp_nid), \
- atomic_read (&(peer)->ibp_refcount)); \
- LASSERT(atomic_read(&(peer)->ibp_refcount) > 0); \
- atomic_inc(&(peer)->ibp_refcount); \
+ CDEBUG(D_NET, "peer[%p] -> %s (%d)++\n", \
+ (peer), libcfs_nid2str((peer)->ibp_nid), \
+ atomic_read (&(peer)->ibp_refcount)); \
+ atomic_inc(&(peer)->ibp_refcount); \
} while (0)
#define kiblnd_peer_decref(peer) \
do { \
- CDEBUG(D_NET, "peer[%p] -> %s (%d)--\n", \
- (peer), libcfs_nid2str((peer)->ibp_nid), \
- atomic_read (&(peer)->ibp_refcount)); \
- LASSERT(atomic_read(&(peer)->ibp_refcount) > 0); \
- if (atomic_dec_and_test(&(peer)->ibp_refcount)) \
- kiblnd_destroy_peer(peer); \
+ CDEBUG(D_NET, "peer[%p] -> %s (%d)--\n", \
+ (peer), libcfs_nid2str((peer)->ibp_nid), \
+ atomic_read (&(peer)->ibp_refcount)); \
+ LASSERT_ATOMIC_POS(&(peer)->ibp_refcount); \
+ if (atomic_dec_and_test(&(peer)->ibp_refcount)) \
+ kiblnd_destroy_peer(peer); \
} while (0)
static inline struct list_head *
kiblnd_nid2peerlist (lnet_nid_t nid)
{
- unsigned int hash = ((unsigned int)nid) % kiblnd_data.kib_peer_hash_size;
+ unsigned int hash =
+ ((unsigned int)nid) % kiblnd_data.kib_peer_hash_size;
- return (&kiblnd_data.kib_peers [hash]);
+ return &kiblnd_data.kib_peers[hash];
}
static inline int
kiblnd_peer_active (kib_peer_t *peer)
{
- /* Am I in the peer hash table? */
- return (!list_empty(&peer->ibp_list));
+ /* Am I in the peer hash table? */
+ return !list_empty(&peer->ibp_list);
}
static inline kib_conn_t *
kiblnd_get_conn_locked (kib_peer_t *peer)
{
- LASSERT (!list_empty(&peer->ibp_conns));
-
+ LASSERT(!list_empty(&peer->ibp_conns));
+
/* just return the first connection */
- return list_entry(peer->ibp_conns.next, kib_conn_t, ibc_list);
+ return list_entry(peer->ibp_conns.next, kib_conn_t, ibc_list);
}
static inline int
-kiblnd_send_keepalive(kib_conn_t *conn)
+kiblnd_send_keepalive(kib_conn_t *conn)
{
- return (*kiblnd_tunables.kib_keepalive > 0) &&
- time_after(jiffies, conn->ibc_last_send +
- *kiblnd_tunables.kib_keepalive*HZ);
+ return (*kiblnd_tunables.kib_keepalive > 0) &&
+ cfs_time_after(jiffies, conn->ibc_last_send +
+ msecs_to_jiffies(*kiblnd_tunables.kib_keepalive *
+ MSEC_PER_SEC));
+}
+
+static inline int
+kiblnd_need_noop(kib_conn_t *conn)
+{
+ LASSERT (conn->ibc_state >= IBLND_CONN_ESTABLISHED);
+
+ if (conn->ibc_outstanding_credits <
+ IBLND_CREDITS_HIGHWATER(conn->ibc_version) &&
+ !kiblnd_send_keepalive(conn))
+ return 0; /* No need to send NOOP */
+
+ if (IBLND_OOB_CAPABLE(conn->ibc_version)) {
+ if (!list_empty(&conn->ibc_tx_queue_nocred))
+ return 0; /* NOOP can be piggybacked */
+
+ /* No tx to piggyback NOOP onto or no credit to send a tx */
+ return (list_empty(&conn->ibc_tx_queue) ||
+ conn->ibc_credits == 0);
+ }
+
+ if (!list_empty(&conn->ibc_tx_noops) || /* NOOP already queued */
+ !list_empty(&conn->ibc_tx_queue_nocred) || /* piggyback NOOP */
+ conn->ibc_credits == 0) /* no credit */
+ return 0;
+
+ if (conn->ibc_credits == 1 && /* last credit reserved for */
+ conn->ibc_outstanding_credits == 0) /* giving back credits */
+ return 0;
+
+ /* No tx to piggyback NOOP onto or no credit to send a tx */
+ return (list_empty(&conn->ibc_tx_queue) || conn->ibc_credits == 1);
}
static inline void
&kiblnd_data.kib_error_qpa, IB_QP_STATE);
}
+static inline const char *
+kiblnd_queue2str(kib_conn_t *conn, struct list_head *q)
+{
+ if (q == &conn->ibc_tx_queue)
+ return "tx_queue";
+
+ if (q == &conn->ibc_tx_queue_rsrvd)
+ return "tx_queue_rsrvd";
+
+ if (q == &conn->ibc_tx_queue_nocred)
+ return "tx_queue_nocred";
+
+ if (q == &conn->ibc_active_txs)
+ return "active_txs";
+
+ LBUG();
+ return NULL;
+}
+
/* CAVEAT EMPTOR: We rely on descriptor alignment to allow us to use the
* lowest bits of the work request id to stash the work item type. */
static inline void
kiblnd_set_conn_state (kib_conn_t *conn, int state)
{
- conn->ibc_state = state;
- mb();
+ conn->ibc_state = state;
+ smp_mb();
}
-#if IBLND_MAP_ON_DEMAND
-static inline int
-kiblnd_rd_size (kib_rdma_desc_t *rd)
+static inline void
+kiblnd_init_msg (kib_msg_t *msg, int type, int body_nob)
{
- return rd->rd_nob;
+ msg->ibm_type = type;
+ msg->ibm_nob = offsetof(kib_msg_t, ibm_u) + body_nob;
}
-#else
+
static inline int
kiblnd_rd_size (kib_rdma_desc_t *rd)
{
int i;
int size;
-
+
for (i = size = 0; i < rd->rd_nfrags; i++)
size += rd->rd_frags[i].rf_nob;
-
+
return size;
}
-#endif
-#if (IBLND_OFED_VERSION == 102)
+static inline __u64
+kiblnd_rd_frag_addr(kib_rdma_desc_t *rd, int index)
+{
+ return rd->rd_frags[index].rf_addr;
+}
+
+static inline __u32
+kiblnd_rd_frag_size(kib_rdma_desc_t *rd, int index)
+{
+ return rd->rd_frags[index].rf_nob;
+}
+
+static inline __u32
+kiblnd_rd_frag_key(kib_rdma_desc_t *rd, int index)
+{
+ return rd->rd_key;
+}
+
+static inline int
+kiblnd_rd_consume_frag(kib_rdma_desc_t *rd, int index, __u32 nob)
+{
+ if (nob < rd->rd_frags[index].rf_nob) {
+ rd->rd_frags[index].rf_addr += nob;
+ rd->rd_frags[index].rf_nob -= nob;
+ } else {
+ index ++;
+ }
+
+ return index;
+}
+
+static inline int
+kiblnd_rd_msg_size(kib_rdma_desc_t *rd, int msgtype, int n)
+{
+ LASSERT (msgtype == IBLND_MSG_GET_REQ ||
+ msgtype == IBLND_MSG_PUT_ACK);
+
+ return msgtype == IBLND_MSG_GET_REQ ?
+ offsetof(kib_get_msg_t, ibgm_rd.rd_frags[n]) :
+ offsetof(kib_putack_msg_t, ibpam_rd.rd_frags[n]);
+}
+
+static inline __u64
+kiblnd_dma_mapping_error(struct ib_device *dev, u64 dma_addr)
+{
+ return ib_dma_mapping_error(dev, dma_addr);
+}
static inline __u64 kiblnd_dma_map_single(struct ib_device *dev,
void *msg, size_t size,
#define KIBLND_CONN_PARAM(e) ((e)->param.conn.private_data)
#define KIBLND_CONN_PARAM_LEN(e) ((e)->param.conn.private_data_len)
-#elif (IBLND_OFED_VERSION == 101)
-
-static inline dma_addr_t kiblnd_dma_map_single(struct ib_device *dev,
- void *msg, size_t size,
- enum dma_data_direction direction)
-{
- return dma_map_single(dev->dma_device, msg, size, direction);
-}
-
-static inline void kiblnd_dma_unmap_single(struct ib_device *dev,
- dma_addr_t addr, size_t size,
- enum dma_data_direction direction)
-{
- dma_unmap_single(dev->dma_device, addr, size, direction);
-}
-
-#define KIBLND_UNMAP_ADDR_SET(p, m, a) pci_unmap_addr_set(p, m, a)
-#define KIBLND_UNMAP_ADDR(p, m, a) pci_unmap_addr(p, m)
-
-static inline int kiblnd_dma_map_sg(struct ib_device *dev,
- struct scatterlist *sg, int nents,
- enum dma_data_direction direction)
-{
- return dma_map_sg(dev->dma_device, sg, nents, direction);
-}
-
-static inline void kiblnd_dma_unmap_sg(struct ib_device *dev,
- struct scatterlist *sg, int nents,
- enum dma_data_direction direction)
-{
- return dma_unmap_sg(dev->dma_device, sg, nents, direction);
-}
-
-
-static inline dma_addr_t kiblnd_sg_dma_address(struct ib_device *dev,
- struct scatterlist *sg)
-{
- return sg_dma_address(sg);
-}
-
+struct ib_mr *kiblnd_find_rd_dma_mr(kib_hca_dev_t *hdev,
+ kib_rdma_desc_t *rd,
+ int negotiated_nfrags);
+void kiblnd_map_rx_descs(kib_conn_t *conn);
+void kiblnd_unmap_rx_descs(kib_conn_t *conn);
+void kiblnd_pool_free_node(kib_pool_t *pool, struct list_head *node);
+struct list_head *kiblnd_pool_alloc_node(kib_poolset_t *ps);
-static inline unsigned int kiblnd_sg_dma_len(struct ib_device *dev,
- struct scatterlist *sg)
-{
- return sg_dma_len(sg);
-}
-
-#define KIBLND_CONN_PARAM(e) ((e)->private_data)
-#define KIBLND_CONN_PARAM_LEN(e) ((e)->private_data_len)
-
-#endif
-
-int kiblnd_startup (lnet_ni_t *ni);
-void kiblnd_shutdown (lnet_ni_t *ni);
-int kiblnd_ctl (lnet_ni_t *ni, unsigned int cmd, void *arg);
+int kiblnd_fmr_pool_map(kib_fmr_poolset_t *fps, __u64 *pages,
+ int npages, __u64 iov, kib_fmr_t *fmr);
+void kiblnd_fmr_pool_unmap(kib_fmr_t *fmr, int status);
int kiblnd_tunables_init(void);
void kiblnd_tunables_fini(void);
int kiblnd_connd (void *arg);
int kiblnd_scheduler(void *arg);
-int kiblnd_thread_start (int (*fn)(void *arg), void *arg);
+int kiblnd_thread_start(int (*fn)(void *arg), void *arg, char *name);
+int kiblnd_failover_thread (void *arg);
-int kiblnd_alloc_pages (kib_pages_t **pp, int npages);
-void kiblnd_free_pages (kib_pages_t *p);
+int kiblnd_alloc_pages(kib_pages_t **pp, int cpt, int npages);
int kiblnd_cm_callback(struct rdma_cm_id *cmid,
struct rdma_cm_event *event);
+int kiblnd_translate_mtu(int value);
-int kiblnd_create_peer (lnet_ni_t *ni, kib_peer_t **peerp, lnet_nid_t nid);
+int kiblnd_dev_failover(kib_dev_t *dev);
+int kiblnd_create_peer(lnet_ni_t *ni, kib_peer_t **peerp, lnet_nid_t nid);
void kiblnd_destroy_peer (kib_peer_t *peer);
+void kiblnd_connect_peer(kib_peer_t *peer);
void kiblnd_destroy_dev (kib_dev_t *dev);
void kiblnd_unlink_peer_locked (kib_peer_t *peer);
-void kiblnd_peer_alive (kib_peer_t *peer);
kib_peer_t *kiblnd_find_peer_locked (lnet_nid_t nid);
-void kiblnd_peer_connect_failed (kib_peer_t *peer, int active, int error);
-int kiblnd_close_stale_conns_locked (kib_peer_t *peer, __u64 incarnation);
+int kiblnd_close_stale_conns_locked (kib_peer_t *peer,
+ int version, __u64 incarnation);
+int kiblnd_close_peer_conns_locked (kib_peer_t *peer, int why);
-void kiblnd_connreq_done(kib_conn_t *conn, int status);
-kib_conn_t *kiblnd_create_conn (kib_peer_t *peer, struct rdma_cm_id *cmid,
- int state);
+kib_conn_t *kiblnd_create_conn(kib_peer_t *peer, struct rdma_cm_id *cmid,
+ int state, int version);
void kiblnd_destroy_conn (kib_conn_t *conn);
void kiblnd_close_conn (kib_conn_t *conn, int error);
void kiblnd_close_conn_locked (kib_conn_t *conn, int error);
-int kiblnd_init_rdma (lnet_ni_t *ni, kib_tx_t *tx, int type,
- int nob, kib_rdma_desc_t *dstrd, __u64 dstcookie);
-
-void kiblnd_queue_tx_locked (kib_tx_t *tx, kib_conn_t *conn);
-void kiblnd_queue_tx (kib_tx_t *tx, kib_conn_t *conn);
-void kiblnd_init_tx_msg (lnet_ni_t *ni, kib_tx_t *tx, int type, int body_nob);
-void kiblnd_txlist_done (lnet_ni_t *ni, struct list_head *txlist, int status);
-void kiblnd_check_sends (kib_conn_t *conn);
+void kiblnd_launch_tx (lnet_ni_t *ni, kib_tx_t *tx, lnet_nid_t nid);
+void kiblnd_txlist_done(lnet_ni_t *ni, struct list_head *txlist, int status);
void kiblnd_qp_event(struct ib_event *event, void *arg);
void kiblnd_cq_event(struct ib_event *event, void *arg);
void kiblnd_cq_completion(struct ib_cq *cq, void *arg);
-void kiblnd_init_msg (kib_msg_t *msg, int type, int body_nob);
-void kiblnd_pack_msg (lnet_ni_t *ni, kib_msg_t *msg,
+void kiblnd_pack_msg (lnet_ni_t *ni, kib_msg_t *msg, int version,
int credits, lnet_nid_t dstnid, __u64 dststamp);
int kiblnd_unpack_msg(kib_msg_t *msg, int nob);
int kiblnd_post_rx (kib_rx_t *rx, int credit);
int kiblnd_send(lnet_ni_t *ni, void *private, lnet_msg_t *lntmsg);
int kiblnd_recv(lnet_ni_t *ni, void *private, lnet_msg_t *lntmsg, int delayed,
- unsigned int niov, struct iovec *iov, lnet_kiov_t *kiov,
+ unsigned int niov, struct kvec *iov, lnet_kiov_t *kiov,
unsigned int offset, unsigned int mlen, unsigned int rlen);
-
-
git://git.whamcloud.com / fs / lustre-release.git / blobdiff