/*
 * GPL HEADER START
 *
 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
 *
 * 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.
 *
 * 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).
 *
 * 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
 *
 * 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, 2012, Intel Corporation.
 */
/*
 * This file is part of Lustre, http://www.lustre.org/
 * Lustre is a trademark of Sun Microsystems, Inc.
 *
 * lustre/mgs/mgs_nids.c
 *
 * NID table management for lustre.
 *
 * Author: Jinshan Xiong <jinshan.xiong@whamcloud.com>
 */

#define DEBUG_SUBSYSTEM S_MGS
#define D_MGS D_CONFIG

#ifdef __KERNEL__
#include <linux/pagemap.h>
#endif

#include <obd.h>
#include <obd_class.h>
#include <lustre_disk.h>

#include "mgs_internal.h"

static unsigned int ir_timeout;

static int nidtbl_is_sane(struct mgs_nidtbl *tbl)
{
        struct mgs_nidtbl_target *tgt;
        int version = 0;

	LASSERT(mutex_is_locked(&tbl->mn_lock));
        cfs_list_for_each_entry(tgt, &tbl->mn_targets, mnt_list) {
                if (!tgt->mnt_version)
                        continue;

                if (version >= tgt->mnt_version)
                        return 0;

                version = tgt->mnt_version;
        }
        return 1;
}

/**
 * Fetch nidtbl entries whose version are not less than @version
 * nidtbl entries will be packed in @pages by @unit_size units - entries
 * shouldn't cross unit boundaries.
 */
static int mgs_nidtbl_read(struct obd_export *exp, struct mgs_nidtbl *tbl,
                           struct mgs_config_res *res, cfs_page_t **pages,
                           int nrpages, int units_total, int unit_size)
{
        struct mgs_nidtbl_target *tgt;
        struct mgs_nidtbl_entry  *entry;
        struct mgs_nidtbl_entry  *last_in_unit = NULL;
        struct mgs_target_info   *mti;
        __u64 version = res->mcr_offset;
        bool nobuf = false;
        void *buf = NULL;
        int bytes_in_unit = 0;
        int units_in_page = 0;
        int index = 0;
        int rc = 0;
        ENTRY;

        /* make sure unit_size is power 2 */
        LASSERT((unit_size & (unit_size - 1)) == 0);
        LASSERT(nrpages << CFS_PAGE_SHIFT >= units_total * unit_size);

	mutex_lock(&tbl->mn_lock);
        LASSERT(nidtbl_is_sane(tbl));

        /* no more entries ? */
        if (version > tbl->mn_version) {
                version = tbl->mn_version;
                goto out;
        }

        /* iterate over all targets to compose a bitmap by the type of llog.
         * If the llog is for MDTs, llog entries for OSTs will be returned;
         * otherwise, it's for clients, then llog entries for both OSTs and
         * MDTs will be returned.
         */
        cfs_list_for_each_entry(tgt, &tbl->mn_targets, mnt_list) {
                int entry_len = sizeof(*entry);

                if (tgt->mnt_version < version)
                        continue;

                /* write target recover information */
                mti  = &tgt->mnt_mti;
                LASSERT(mti->mti_nid_count < MTI_NIDS_MAX);
                entry_len += mti->mti_nid_count * sizeof(lnet_nid_t);

                if (entry_len > unit_size) {
                        CWARN("nidtbl: too large entry: entry length %d,"
                              "unit size: %d\n", entry_len, unit_size);
                        GOTO(out, rc = -EOVERFLOW);
                }

                if (bytes_in_unit < entry_len) {
                        if (units_total == 0) {
                                nobuf = true;
                                break;
                        }

                        /* check if we need to consume remaining bytes. */
                        if (last_in_unit != NULL && bytes_in_unit) {
#if LUSTRE_VERSION_CODE < OBD_OCD_VERSION(2, 6, 50, 0)
				/* May need to swab back to update the length.*/
				if (exp->exp_need_mne_swab)
					lustre_swab_mgs_nidtbl_entry(last_in_unit);
#endif
				last_in_unit->mne_length += bytes_in_unit;
#if LUSTRE_VERSION_CODE < OBD_OCD_VERSION(2, 6, 50, 0)
				if (exp->exp_need_mne_swab)
					lustre_swab_mgs_nidtbl_entry(last_in_unit);
#endif
                                rc  += bytes_in_unit;
                                buf += bytes_in_unit;
                                last_in_unit = NULL;
                        }
                        LASSERT((rc & (unit_size - 1)) == 0);

                        if (units_in_page == 0) {
                                /* allocate a new page */
                                pages[index] = cfs_alloc_page(CFS_ALLOC_STD);
                                if (pages[index] == NULL) {
                                        rc = -ENOMEM;
                                        break;
                                }

                                /* destroy previous map */
                                if (index > 0)
                                        cfs_kunmap(pages[index - 1]);

                                /* reassign buffer */
                                buf = cfs_kmap(pages[index]);
                                ++index;

                                units_in_page = CFS_PAGE_SIZE / unit_size;
                                LASSERT(units_in_page > 0);
                        }

                        /* allocate an unit */
                        LASSERT(((long)buf & (unit_size - 1)) == 0);
                        bytes_in_unit = unit_size;
                        --units_in_page;
                        --units_total;
                }

                /* fill in entry. */
                entry = (struct mgs_nidtbl_entry *)buf;
                entry->mne_version   = tgt->mnt_version;
                entry->mne_instance  = mti->mti_instance;
                entry->mne_index     = mti->mti_stripe_index;
                entry->mne_length    = entry_len;
                entry->mne_type      = tgt->mnt_type;
                entry->mne_nid_type  = 0;
                entry->mne_nid_size  = sizeof(lnet_nid_t);
                entry->mne_nid_count = mti->mti_nid_count;
                memcpy(entry->u.nids, mti->mti_nids,
                       mti->mti_nid_count * sizeof(lnet_nid_t));

#if LUSTRE_VERSION_CODE < OBD_OCD_VERSION(2, 6, 50, 0)
		/* For LU-1644, swab entry for 2.2 clients. */
		if (exp->exp_need_mne_swab)
			lustre_swab_mgs_nidtbl_entry(entry);
#endif

                version = tgt->mnt_version;
                rc     += entry_len;
                buf    += entry_len;

                bytes_in_unit -= entry_len;
                last_in_unit   = entry;

                CDEBUG(D_MGS, "fsname %s, entry size %d, pages %d/%d/%d/%d.\n",
                       tbl->mn_fsdb->fsdb_name, entry_len,
                       bytes_in_unit, index, nrpages, units_total);
        }
        if (index > 0)
                cfs_kunmap(pages[index - 1]);
out:
        LASSERT(version <= tbl->mn_version);
        res->mcr_size = tbl->mn_version;
        res->mcr_offset = nobuf ? version : tbl->mn_version;
	mutex_unlock(&tbl->mn_lock);
        LASSERT(ergo(version == 1, rc == 0)); /* get the log first time */

        CDEBUG(D_MGS, "Read IR logs %s return with %d, version %llu\n",
               tbl->mn_fsdb->fsdb_name, rc, version);
        RETURN(rc);
}

static int nidtbl_update_version(const struct lu_env *env,
				 struct mgs_device *mgs,
				 struct mgs_nidtbl *tbl)
{
	struct dt_object *fsdb;
	struct thandle   *th;
	u64		  version;
	struct lu_buf	  buf = {
				.lb_buf = &version,
				.lb_len = sizeof(version)
			  };
	loff_t		  off = 0;
	int		  rc;
        ENTRY;

	LASSERT(mutex_is_locked(&tbl->mn_lock));

	fsdb = local_file_find_or_create(env, mgs->mgs_los, mgs->mgs_nidtbl_dir,
					 tbl->mn_fsdb->fsdb_name,
					 S_IFREG | S_IRUGO | S_IWUSR);
	if (IS_ERR(fsdb))
		RETURN(PTR_ERR(fsdb));

	th = dt_trans_create(env, mgs->mgs_bottom);
	if (IS_ERR(th))
		GOTO(out_put, rc = PTR_ERR(th));

	th->th_sync = 1; /* update table synchronously */
	rc = dt_declare_record_write(env, fsdb, buf.lb_len, off, th);
	if (rc)
		GOTO(out, rc);

	rc = dt_trans_start_local(env, mgs->mgs_bottom, th);
	if (rc)
		GOTO(out, rc);

	version = cpu_to_le64(tbl->mn_version);
	rc = dt_record_write(env, fsdb, &buf, &off, th);

out:
	dt_trans_stop(env, mgs->mgs_bottom, th);
out_put:
	lu_object_put(env, &fsdb->do_lu);
	RETURN(rc);
}

#define MGS_NIDTBL_VERSION_INIT 2

static int nidtbl_read_version(const struct lu_env *env,
			       struct mgs_device *mgs, struct mgs_nidtbl *tbl,
			       u64 *version)
{
	struct dt_object *fsdb;
	struct lu_fid     fid;
	u64               tmpver;
	struct lu_buf     buf = {
				.lb_buf = &tmpver,
				.lb_len = sizeof(tmpver)
			  };
        loff_t               off = 0;
        int                  rc;
        ENTRY;

	LASSERT(mutex_is_locked(&tbl->mn_lock));

	LASSERT(mgs->mgs_nidtbl_dir);
	rc = dt_lookup_dir(env, mgs->mgs_nidtbl_dir, tbl->mn_fsdb->fsdb_name,
			   &fid);
	if (rc == -ENOENT) {
		*version = MGS_NIDTBL_VERSION_INIT;
		RETURN(0);
	} else if (rc < 0) {
		RETURN(rc);
	}

	fsdb = dt_locate_at(env, mgs->mgs_bottom, &fid,
			    &mgs->mgs_dt_dev.dd_lu_dev);
	if (IS_ERR(fsdb))
		RETURN(PTR_ERR(fsdb));

	rc = dt_read(env, fsdb, &buf, &off);
	if (rc == buf.lb_len) {
		*version = le64_to_cpu(tmpver);
		rc = 0;
	} else if (rc == 0) {
		*version = MGS_NIDTBL_VERSION_INIT;
	} else {
		CERROR("%s: read version file %s error %d\n",
		       mgs->mgs_obd->obd_name, tbl->mn_fsdb->fsdb_name, rc);
	}
	lu_object_put(env, &fsdb->do_lu);
	RETURN(rc);
}

static int mgs_nidtbl_write(const struct lu_env *env, struct fs_db *fsdb,
			    struct mgs_target_info *mti)
{
        struct mgs_nidtbl        *tbl;
        struct mgs_nidtbl_target *tgt;
        bool found = false;
        int type   = mti->mti_flags & LDD_F_SV_TYPE_MASK;
        int rc     = 0;
        ENTRY;

        type &= ~LDD_F_SV_TYPE_MGS;
        LASSERT(type != 0);

        tbl = &fsdb->fsdb_nidtbl;
	mutex_lock(&tbl->mn_lock);
        cfs_list_for_each_entry(tgt, &tbl->mn_targets, mnt_list) {
                struct mgs_target_info *info = &tgt->mnt_mti;
                if (type == tgt->mnt_type &&
                    mti->mti_stripe_index == info->mti_stripe_index) {
                        found = true;
                        break;
                }
        }
        if (!found) {
                OBD_ALLOC_PTR(tgt);
                if (tgt == NULL)
                        GOTO(out, rc = -ENOMEM);

                CFS_INIT_LIST_HEAD(&tgt->mnt_list);
                tgt->mnt_fs      = tbl;
                tgt->mnt_version = 0;       /* 0 means invalid */
                tgt->mnt_type    = type;

                ++tbl->mn_nr_targets;
        }

        tgt->mnt_version = ++tbl->mn_version;
        tgt->mnt_mti     = *mti;

        cfs_list_move_tail(&tgt->mnt_list, &tbl->mn_targets);

	rc = nidtbl_update_version(env, fsdb->fsdb_mgs, tbl);
        EXIT;

out:
	mutex_unlock(&tbl->mn_lock);
        if (rc)
                CERROR("Write NID table version for file system %s error %d\n",
                       fsdb->fsdb_name, rc);
        return rc;
}

static void mgs_nidtbl_fini_fs(struct fs_db *fsdb)
{
        struct mgs_nidtbl *tbl = &fsdb->fsdb_nidtbl;
        CFS_LIST_HEAD(head);

	mutex_lock(&tbl->mn_lock);
        tbl->mn_nr_targets = 0;
        cfs_list_splice_init(&tbl->mn_targets, &head);
	mutex_unlock(&tbl->mn_lock);

        while (!cfs_list_empty(&head)) {
                struct mgs_nidtbl_target *tgt;
                tgt = list_entry(head.next, struct mgs_nidtbl_target, mnt_list);
                cfs_list_del(&tgt->mnt_list);
                OBD_FREE_PTR(tgt);
        }
}

static int mgs_nidtbl_init_fs(const struct lu_env *env, struct fs_db *fsdb)
{
        struct mgs_nidtbl *tbl = &fsdb->fsdb_nidtbl;
	int rc;

        CFS_INIT_LIST_HEAD(&tbl->mn_targets);
	mutex_init(&tbl->mn_lock);
        tbl->mn_nr_targets = 0;
        tbl->mn_fsdb = fsdb;
	mutex_lock(&tbl->mn_lock);
	rc = nidtbl_read_version(env, fsdb->fsdb_mgs, tbl, &tbl->mn_version);
	mutex_unlock(&tbl->mn_lock);
	if (rc < 0)
		CERROR("%s: IR: failed to read current version, rc = %d\n",
		       fsdb->fsdb_mgs->mgs_obd->obd_name, rc);
	else
		CDEBUG(D_MGS, "IR: current version is %llu\n",
		       tbl->mn_version);

	return rc;
}

/* --------- Imperative Recovery relies on nidtbl stuff ------- */
void mgs_ir_notify_complete(struct fs_db *fsdb)
{
        struct timeval tv;
        cfs_duration_t delta;

        cfs_atomic_set(&fsdb->fsdb_notify_phase, 0);

        /* do statistic */
        fsdb->fsdb_notify_count++;
        delta = cfs_time_sub(cfs_time_current(), fsdb->fsdb_notify_start);
        fsdb->fsdb_notify_total += delta;
        if (delta > fsdb->fsdb_notify_max)
                fsdb->fsdb_notify_max = delta;

        cfs_duration_usec(delta, &tv);
        CDEBUG(D_MGS, "Revoke recover lock of %s completed after %ld.%06lds\n",
               fsdb->fsdb_name, tv.tv_sec, tv.tv_usec);
}

static int mgs_ir_notify(void *arg)
{
        struct fs_db      *fsdb   = arg;
        struct ldlm_res_id resid;

        char name[sizeof(fsdb->fsdb_name) + 20];

        LASSERTF(sizeof(name) < 32, "name is too large to be in stack.\n");
        sprintf(name, "mgs_%s_notify", fsdb->fsdb_name);
        cfs_daemonize(name);

	complete(&fsdb->fsdb_notify_comp);

        set_user_nice(current, -2);

        mgc_fsname2resid(fsdb->fsdb_name, &resid, CONFIG_T_RECOVER);
        while (1) {
                struct l_wait_info   lwi = { 0 };

                l_wait_event(fsdb->fsdb_notify_waitq,
                             fsdb->fsdb_notify_stop ||
                             cfs_atomic_read(&fsdb->fsdb_notify_phase),
                             &lwi);
                if (fsdb->fsdb_notify_stop)
                        break;

                CDEBUG(D_MGS, "%s woken up, phase is %d\n",
                       name, cfs_atomic_read(&fsdb->fsdb_notify_phase));

                fsdb->fsdb_notify_start = cfs_time_current();
		mgs_revoke_lock(fsdb->fsdb_mgs, fsdb, CONFIG_T_RECOVER);
        }

	complete(&fsdb->fsdb_notify_comp);
        return 0;
}

int mgs_ir_init_fs(const struct lu_env *env, struct mgs_device *mgs,
		   struct fs_db *fsdb)
{
        int rc;

        if (!ir_timeout)
                ir_timeout = OBD_IR_MGS_TIMEOUT;

        fsdb->fsdb_ir_state = IR_FULL;
        if (cfs_time_before(cfs_time_current_sec(),
                            mgs->mgs_start_time + ir_timeout))
                fsdb->fsdb_ir_state = IR_STARTUP;
        fsdb->fsdb_nonir_clients = 0;
        CFS_INIT_LIST_HEAD(&fsdb->fsdb_clients);

        /* start notify thread */
	fsdb->fsdb_mgs = mgs;
        cfs_atomic_set(&fsdb->fsdb_notify_phase, 0);
        cfs_waitq_init(&fsdb->fsdb_notify_waitq);
	init_completion(&fsdb->fsdb_notify_comp);
        rc = cfs_create_thread(mgs_ir_notify, fsdb, CFS_DAEMON_FLAGS);
        if (rc > 0)
		wait_for_completion(&fsdb->fsdb_notify_comp);
        else
                CERROR("Start notify thread error %d\n", rc);

	mgs_nidtbl_init_fs(env, fsdb);
        return 0;
}

void mgs_ir_fini_fs(struct mgs_device *mgs, struct fs_db *fsdb)
{
	if (test_bit(FSDB_MGS_SELF, &fsdb->fsdb_flags))
                return;

        mgs_fsc_cleanup_by_fsdb(fsdb);

        mgs_nidtbl_fini_fs(fsdb);

        LASSERT(cfs_list_empty(&fsdb->fsdb_clients));

        fsdb->fsdb_notify_stop = 1;
        cfs_waitq_signal(&fsdb->fsdb_notify_waitq);
	wait_for_completion(&fsdb->fsdb_notify_comp);
}

/* caller must have held fsdb_mutex */
static inline void ir_state_graduate(struct fs_db *fsdb)
{
        if (fsdb->fsdb_ir_state == IR_STARTUP) {
		if (cfs_time_before(fsdb->fsdb_mgs->mgs_start_time + ir_timeout,
                                    cfs_time_current_sec())) {
                        fsdb->fsdb_ir_state = IR_FULL;
                        if (fsdb->fsdb_nonir_clients)
                                fsdb->fsdb_ir_state = IR_PARTIAL;
                }
        }
}

int mgs_ir_update(const struct lu_env *env, struct mgs_device *mgs,
		  struct mgs_target_info *mti)
{
        struct fs_db *fsdb;
        bool notify = true;
        int rc;

        if (mti->mti_instance == 0)
                return -EINVAL;

	rc = mgs_find_or_make_fsdb(env, mgs, mti->mti_fsname, &fsdb);
        if (rc)
                return rc;

	rc = mgs_nidtbl_write(env, fsdb, mti);
        if (rc)
                return rc;

        /* check ir state */
	mutex_lock(&fsdb->fsdb_mutex);
        ir_state_graduate(fsdb);
        switch (fsdb->fsdb_ir_state) {
        case IR_FULL:
                mti->mti_flags |= LDD_F_IR_CAPABLE;
                break;
        case IR_DISABLED:
                notify = false;
        case IR_STARTUP:
        case IR_PARTIAL:
                break;
        default:
                LBUG();
        }
	mutex_unlock(&fsdb->fsdb_mutex);

        LASSERT(ergo(mti->mti_flags & LDD_F_IR_CAPABLE, notify));
        if (notify) {
                CDEBUG(D_MGS, "Try to revoke recover lock of %s\n",
                       fsdb->fsdb_name);
                cfs_atomic_inc(&fsdb->fsdb_notify_phase);
                cfs_waitq_signal(&fsdb->fsdb_notify_waitq);
        }
        return 0;
}

/* NID table can be cached by two entities: Clients and MDTs */
enum {
        IR_CLIENT  = 1,
        IR_MDT     = 2
};

static int delogname(char *logname, char *fsname, int *typ)
{
        char *ptr;
        int   type;
        int   len;

        ptr = strrchr(logname, '-');
        if (ptr == NULL)
                return -EINVAL;

        /* decouple file system name. The llog name may be:
         * - "prefix-fsname", prefix is "cliir" or "mdtir"
         */
        if (strncmp(ptr, "-mdtir", 6) == 0)
                type = IR_MDT;
        else if (strncmp(ptr, "-cliir", 6) == 0)
                type = IR_CLIENT;
        else
                return -EINVAL;

        len = ptr - logname;
        if (len == 0)
                return -EINVAL;

        memcpy(fsname, logname, len);
        fsname[len] = 0;
        if (typ)
                *typ = type;
        return 0;
}

int mgs_get_ir_logs(struct ptlrpc_request *req)
{
	struct lu_env     *env = req->rq_svc_thread->t_env;
	struct mgs_device *mgs = exp2mgs_dev(req->rq_export);
        struct fs_db      *fsdb;
        struct mgs_config_body  *body;
        struct mgs_config_res   *res;
        struct ptlrpc_bulk_desc *desc;
        struct l_wait_info lwi;
        char               fsname[16];
        long               bufsize;
        int                unit_size;
        int                type;
        int                rc = 0;
        int                i;
        int                bytes;
        int                page_count;
        int                nrpages;
        cfs_page_t       **pages = NULL;
        ENTRY;

        body = req_capsule_client_get(&req->rq_pill, &RMF_MGS_CONFIG_BODY);
        if (body == NULL)
                RETURN(-EINVAL);

        if (body->mcb_type != CONFIG_T_RECOVER)
                RETURN(-EINVAL);

        rc = delogname(body->mcb_name, fsname, &type);
        if (rc)
                RETURN(rc);

	rc = mgs_find_or_make_fsdb(env, mgs, fsname, &fsdb);
        if (rc)
		RETURN(rc);

        bufsize = body->mcb_units << body->mcb_bits;
        nrpages = (bufsize + CFS_PAGE_SIZE - 1) >> CFS_PAGE_SHIFT;
        if (nrpages > PTLRPC_MAX_BRW_PAGES)
                RETURN(-EINVAL);

        CDEBUG(D_MGS, "Reading IR log %s bufsize %ld.\n",
               body->mcb_name, bufsize);

        OBD_ALLOC(pages, sizeof(*pages) * nrpages);
        if (pages == NULL)
                RETURN(-ENOMEM);

        rc = req_capsule_server_pack(&req->rq_pill);
        if (rc)
                GOTO(out, rc);

        res = req_capsule_server_get(&req->rq_pill, &RMF_MGS_CONFIG_RES);
        if (res == NULL)
                GOTO(out, rc = -EINVAL);

        res->mcr_offset = body->mcb_offset;
        unit_size = min_t(int, 1 << body->mcb_bits, CFS_PAGE_SIZE);
	bytes = mgs_nidtbl_read(req->rq_export, &fsdb->fsdb_nidtbl, res,
				pages, nrpages, bufsize / unit_size, unit_size);
        if (bytes < 0)
                GOTO(out, rc = bytes);

        /* start bulk transfer */
        page_count = (bytes + CFS_PAGE_SIZE - 1) >> CFS_PAGE_SHIFT;
        LASSERT(page_count <= nrpages);
        desc = ptlrpc_prep_bulk_exp(req, page_count,
                                    BULK_PUT_SOURCE, MGS_BULK_PORTAL);
        if (desc == NULL)
                GOTO(out, rc = -ENOMEM);

        for (i = 0; i < page_count && bytes > 0; i++) {
		ptlrpc_prep_bulk_page_pin(desc, pages[i], 0,
					  min_t(int, bytes, CFS_PAGE_SIZE));
                bytes -= CFS_PAGE_SIZE;
        }

        rc = target_bulk_io(req->rq_export, desc, &lwi);
	ptlrpc_free_bulk_pin(desc);

out:
	for (i = 0; i < nrpages; i++) {
		if (pages[i] == NULL)
			break;
		cfs_free_page(pages[i]);
	}
	OBD_FREE(pages, sizeof(*pages) * nrpages);
	return rc;
}

static int lprocfs_ir_set_state(struct fs_db *fsdb, const char *buf)
{
        const char *strings[] = IR_STRINGS;
        int         state = -1;
        int         i;

        for (i = 0; i < ARRAY_SIZE(strings); i++) {
                if (strcmp(strings[i], buf) == 0) {
                        state = i;
                        break;
                }
        }
        if (state < 0)
                return -EINVAL;

        CDEBUG(D_MGS, "change fsr state of %s from %s to %s\n",
               fsdb->fsdb_name, strings[fsdb->fsdb_ir_state], strings[state]);
	mutex_lock(&fsdb->fsdb_mutex);
        if (state == IR_FULL && fsdb->fsdb_nonir_clients)
                state = IR_PARTIAL;
        fsdb->fsdb_ir_state = state;
	mutex_unlock(&fsdb->fsdb_mutex);

        return 0;
}

static int lprocfs_ir_set_timeout(struct fs_db *fsdb, const char *buf)
{
        return -EINVAL;
}

static int lprocfs_ir_clear_stats(struct fs_db *fsdb, const char *buf)
{
        if (*buf)
                return -EINVAL;

        fsdb->fsdb_notify_total = 0;
        fsdb->fsdb_notify_max   = 0;
        fsdb->fsdb_notify_count = 0;
        return 0;
}

static struct lproc_ir_cmd {
        char *name;
        int   namelen;
        int (*handler)(struct fs_db *, const char *);
} ir_cmds[] = {
        { "state=",   6, lprocfs_ir_set_state },
        { "timeout=", 8, lprocfs_ir_set_timeout },
        { "0",        1, lprocfs_ir_clear_stats }
};

int lprocfs_wr_ir_state(struct file *file, const char *buffer,
                         unsigned long count, void *data)
{
        struct fs_db *fsdb = data;
        char *kbuf;
        char *ptr;
        int rc = 0;

        if (count > CFS_PAGE_SIZE)
                return -EINVAL;

        OBD_ALLOC(kbuf, count + 1);
        if (kbuf == NULL)
                return -ENOMEM;

        if (copy_from_user(kbuf, buffer, count)) {
                OBD_FREE(kbuf, count);
                return -EFAULT;
        }

        kbuf[count] = 0; /* buffer is supposed to end with 0 */
        if (kbuf[count - 1] == '\n')
                kbuf[count - 1] = 0;
        ptr = kbuf;

        /* fsname=<file system name> must be the 1st entry */
        while (ptr != NULL) {
                char *tmpptr;
                int i;

                tmpptr = strchr(ptr, ';');
                if (tmpptr)
                        *tmpptr++ = 0;

                rc = -EINVAL;
                for (i = 0; i < ARRAY_SIZE(ir_cmds); i++) {
                        struct lproc_ir_cmd *cmd;
                        int cmdlen;

                        cmd    = &ir_cmds[i];
                        cmdlen = cmd->namelen;
                        if (strncmp(cmd->name, ptr, cmdlen) == 0) {
                                ptr += cmdlen;
                                rc = cmd->handler(fsdb, ptr);
                                break;
                        }
                }
                if (rc)
                        break;

                ptr = tmpptr;
        }
        if (rc)
                CERROR("Unable to process command: %s(%d)\n", ptr, rc);
        OBD_FREE(kbuf, count + 1);
        return rc ?: count;
}

int lprocfs_rd_ir_state(struct seq_file *seq, void *data)
{
        struct fs_db      *fsdb = data;
        struct mgs_nidtbl *tbl  = &fsdb->fsdb_nidtbl;
        const char        *ir_strings[] = IR_STRINGS;
        struct timeval     tv_max;
        struct timeval     tv;

        /* mgs_live_seq_show() already holds fsdb_mutex. */
        ir_state_graduate(fsdb);

        seq_printf(seq, "\nimperative_recovery_state:\n");
        seq_printf(seq,
                   "    state: %s\n"
                   "    nonir_clients: %d\n"
                   "    nidtbl_version: %lld\n",
                   ir_strings[fsdb->fsdb_ir_state], fsdb->fsdb_nonir_clients,
                   tbl->mn_version);

        cfs_duration_usec(fsdb->fsdb_notify_total, &tv);
        cfs_duration_usec(fsdb->fsdb_notify_max, &tv_max);

        seq_printf(seq, "    notify_duration_total: %lu.%06lu\n"
                        "    notify_duation_max: %lu.%06lu\n"
                        "    notify_count: %u\n",
                   tv.tv_sec, tv.tv_usec,
                   tv_max.tv_sec, tv_max.tv_usec,
                   fsdb->fsdb_notify_count);

        return 0;
}

int lprocfs_rd_ir_timeout(char *page, char **start, off_t off, int count,
                          int *eof, void *data)
{
        *eof = 1;
        return snprintf(page, count, "%d\n", ir_timeout);
}

int lprocfs_wr_ir_timeout(struct file *file, const char *buffer,
                          unsigned long count, void *data)
{
        return lprocfs_wr_uint(file, buffer, count, &ir_timeout);
}

/* --------------- Handle non IR support clients --------------- */
/* attach a lustre file system to an export */
int mgs_fsc_attach(const struct lu_env *env, struct obd_export *exp,
		   char *fsname)
{
        struct mgs_export_data *data = &exp->u.eu_mgs_data;
	struct mgs_device *mgs = exp2mgs_dev(exp);
        struct fs_db      *fsdb;
        struct mgs_fsc    *fsc     = NULL;
        struct mgs_fsc    *new_fsc = NULL;
        bool               found   = false;
        int                rc;
        ENTRY;

	rc = mgs_find_or_make_fsdb(env, mgs, fsname, &fsdb);
        if (rc)
                RETURN(rc);

        /* allocate a new fsc in case we need it in spinlock. */
        OBD_ALLOC_PTR(new_fsc);
        if (new_fsc == NULL)
                RETURN(-ENOMEM);

	CFS_INIT_LIST_HEAD(&new_fsc->mfc_export_list);
	CFS_INIT_LIST_HEAD(&new_fsc->mfc_fsdb_list);
	new_fsc->mfc_fsdb       = fsdb;
	new_fsc->mfc_export     = class_export_get(exp);
	new_fsc->mfc_ir_capable = !!(exp_connect_flags(exp) &
				     OBD_CONNECT_IMP_RECOV);

	rc = -EEXIST;
	mutex_lock(&fsdb->fsdb_mutex);

	/* tend to find it in export list because this list is shorter. */
	spin_lock(&data->med_lock);
        cfs_list_for_each_entry(fsc, &data->med_clients, mfc_export_list) {
                if (strcmp(fsname, fsc->mfc_fsdb->fsdb_name) == 0) {
                        found = true;
                        break;
                }
        }
        if (!found) {
                fsc = new_fsc;
                new_fsc = NULL;

                /* add it into export list. */
                cfs_list_add(&fsc->mfc_export_list, &data->med_clients);

                /* add into fsdb list. */
                cfs_list_add(&fsc->mfc_fsdb_list, &fsdb->fsdb_clients);
                if (!fsc->mfc_ir_capable) {
                        ++fsdb->fsdb_nonir_clients;
                        if (fsdb->fsdb_ir_state == IR_FULL)
                                fsdb->fsdb_ir_state = IR_PARTIAL;
                }
                rc = 0;
        }
	spin_unlock(&data->med_lock);
	mutex_unlock(&fsdb->fsdb_mutex);

        if (new_fsc) {
                class_export_put(new_fsc->mfc_export);
                OBD_FREE_PTR(new_fsc);
        }
        RETURN(rc);
}

void mgs_fsc_cleanup(struct obd_export *exp)
{
	struct mgs_export_data *data = &exp->u.eu_mgs_data;
	struct mgs_fsc *fsc, *tmp;
	CFS_LIST_HEAD(head);

	spin_lock(&data->med_lock);
	cfs_list_splice_init(&data->med_clients, &head);
	spin_unlock(&data->med_lock);

	cfs_list_for_each_entry_safe(fsc, tmp, &head, mfc_export_list) {
		struct fs_db *fsdb = fsc->mfc_fsdb;

		LASSERT(fsc->mfc_export == exp);

		mutex_lock(&fsdb->fsdb_mutex);
		cfs_list_del_init(&fsc->mfc_fsdb_list);
		if (fsc->mfc_ir_capable == 0) {
			--fsdb->fsdb_nonir_clients;
			LASSERT(fsdb->fsdb_ir_state != IR_FULL);
			if (fsdb->fsdb_nonir_clients == 0 &&
			    fsdb->fsdb_ir_state == IR_PARTIAL)
				fsdb->fsdb_ir_state = IR_FULL;
		}
		mutex_unlock(&fsdb->fsdb_mutex);
		cfs_list_del_init(&fsc->mfc_export_list);
		class_export_put(fsc->mfc_export);
		OBD_FREE_PTR(fsc);
	}
}

/* must be called with fsdb->fsdb_mutex held */
void mgs_fsc_cleanup_by_fsdb(struct fs_db *fsdb)
{
        struct mgs_fsc *fsc, *tmp;

        cfs_list_for_each_entry_safe(fsc, tmp, &fsdb->fsdb_clients,
                                     mfc_fsdb_list) {
                struct mgs_export_data *data = &fsc->mfc_export->u.eu_mgs_data;

                LASSERT(fsdb == fsc->mfc_fsdb);
                cfs_list_del_init(&fsc->mfc_fsdb_list);

		spin_lock(&data->med_lock);
		cfs_list_del_init(&fsc->mfc_export_list);
		spin_unlock(&data->med_lock);
                class_export_put(fsc->mfc_export);
                OBD_FREE_PTR(fsc);
        }

        fsdb->fsdb_nonir_clients = 0;
        if (fsdb->fsdb_ir_state == IR_PARTIAL)
                fsdb->fsdb_ir_state = IR_FULL;
}