LU-4942 at: per-export lock callback timeout

[fs/lustre-release.git] / lustre / ofd / ofd_dev.c

/*
 * 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.gnu.org/licenses/gpl-2.0.html
 *
 * GPL HEADER END
 */
/*
 * Copyright (c) 2009, 2010, Oracle and/or its affiliates. All rights reserved.
 * Use is subject to license terms.
 *
 * Copyright (c) 2012, 2014 Intel Corporation.
 */
/*
 * This file is part of Lustre, http://www.lustre.org/
 * Lustre is a trademark of Sun Microsystems, Inc.
 *
 * lustre/ofd/ofd_dev.c
 *
 * This file contains OSD API methods for OBD Filter Device (OFD),
 * request handlers and supplemental functions to set OFD up and clean it up.
 *
 * Author: Alex Zhuravlev <alexey.zhuravlev@intel.com>
 * Author: Mike Pershin <mike.pershin@intel.com>
 * Author: Johann Lombardi <johann.lombardi@intel.com>
 */
/*
 * The OBD Filter Device (OFD) module belongs to the Object Storage
 * Server stack and connects the RPC oriented Unified Target (TGT)
 * layer (see lustre/include/lu_target.h) to the storage oriented OSD
 * layer (see lustre/doc/osd-api.txt).
 *
 *     TGT
 *      |      DT and OBD APIs
 *     OFD
 *      |      DT API
 *     OSD
 *
 * OFD implements the LU and OBD device APIs and is responsible for:
 *
 * - Handling client requests (create, destroy, bulk IO, setattr,
 *   get_info, set_info, statfs) for the objects belonging to the OST
 *   (together with TGT).
 *
 * - Providing grant space management which allows clients to reserve
 *   disk space for data writeback. OFD tracks grants on global and
 *   per client levels.
 *
 * - Handling object precreation requests from MDTs.
 *
 * - Operating the LDLM service that allows clients to maintain object
 *   data cache coherence.
 */

#define DEBUG_SUBSYSTEM S_FILTER

#include <obd_class.h>
#include <lustre_param.h>
#include <lustre_fid.h>
#include <lustre_lfsck.h>
#include <lustre/lustre_idl.h>
#include <lustre_dlm.h>
#include <lustre_quota.h>

#include "ofd_internal.h"

/* Slab for OFD object allocation */
static struct kmem_cache *ofd_object_kmem;

static struct lu_kmem_descr ofd_caches[] = {
        {
                .ckd_cache = &ofd_object_kmem,
                .ckd_name  = "ofd_obj",
                .ckd_size  = sizeof(struct ofd_object)
        },
        {
                .ckd_cache = NULL
        }
};

/**
 * Connect OFD to the next device in the stack.
 *
 * This function is used for device stack configuration and links OFD
 * device with bottom OSD device.
 *
 * \param[in]  env      execution environment
 * \param[in]  m        OFD device
 * \param[in]  next     name of next device in the stack
 * \param[out] exp      export to return
 *
 * \retval              0 and export in \a exp if successful
 * \retval              negative value on error
 */
static int ofd_connect_to_next(const struct lu_env *env, struct ofd_device *m,
                               const char *next, struct obd_export **exp)
{
        struct obd_connect_data *data = NULL;
        struct obd_device       *obd;
        int                      rc;
        ENTRY;

        OBD_ALLOC_PTR(data);
        if (data == NULL)
                GOTO(out, rc = -ENOMEM);

        obd = class_name2obd(next);
        if (obd == NULL) {
                CERROR("%s: can't locate next device: %s\n",
                       ofd_name(m), next);
                GOTO(out, rc = -ENOTCONN);
        }

        data->ocd_connect_flags = OBD_CONNECT_VERSION;
        data->ocd_version = LUSTRE_VERSION_CODE;

        rc = obd_connect(NULL, exp, obd, &obd->obd_uuid, data, NULL);
        if (rc) {
                CERROR("%s: cannot connect to next dev %s: rc = %d\n",
                       ofd_name(m), next, rc);
                GOTO(out, rc);
        }

        m->ofd_dt_dev.dd_lu_dev.ld_site =
                m->ofd_osd_exp->exp_obd->obd_lu_dev->ld_site;
        LASSERT(m->ofd_dt_dev.dd_lu_dev.ld_site);
        m->ofd_osd = lu2dt_dev(m->ofd_osd_exp->exp_obd->obd_lu_dev);
        m->ofd_dt_dev.dd_lu_dev.ld_site->ls_top_dev = &m->ofd_dt_dev.dd_lu_dev;

out:
        if (data)
                OBD_FREE_PTR(data);
        RETURN(rc);
}

/**
 * Initialize stack of devices.
 *
 * This function initializes OFD-OSD device stack to serve OST requests
 *
 * \param[in] env       execution environment
 * \param[in] m         OFD device
 * \param[in] cfg       Lustre config for this server
 *
 * \retval              0 if successful
 * \retval              negative value on error
 */
static int ofd_stack_init(const struct lu_env *env,
                          struct ofd_device *m, struct lustre_cfg *cfg)
{
        const char              *dev = lustre_cfg_string(cfg, 0);
        struct lu_device        *d;
        struct ofd_thread_info  *info = ofd_info(env);
        struct lustre_mount_info *lmi;
        int                      rc;
        char                    *osdname;

        ENTRY;

        lmi = server_get_mount(dev);
        if (lmi == NULL) {
                CERROR("Cannot get mount info for %s!\n", dev);
                RETURN(-ENODEV);
        }

        /* find bottom osd */
        OBD_ALLOC(osdname, MTI_NAME_MAXLEN);
        if (osdname == NULL)
                RETURN(-ENOMEM);

        snprintf(osdname, MTI_NAME_MAXLEN, "%s-osd", dev);
        rc = ofd_connect_to_next(env, m, osdname, &m->ofd_osd_exp);
        OBD_FREE(osdname, MTI_NAME_MAXLEN);
        if (rc)
                RETURN(rc);

        d = m->ofd_osd_exp->exp_obd->obd_lu_dev;
        LASSERT(d);
        m->ofd_osd = lu2dt_dev(d);

        snprintf(info->fti_u.name, sizeof(info->fti_u.name),
                 "%s-osd", lustre_cfg_string(cfg, 0));

        RETURN(rc);
}

/**
 * Finalize the device stack OFD-OSD.
 *
 * This function cleans OFD-OSD device stack and
 * disconnects OFD from the OSD.
 *
 * \param[in] env       execution environment
 * \param[in] m         OFD device
 * \param[in] top       top device of stack
 *
 * \retval              0 if successful
 * \retval              negative value on error
 */
static void ofd_stack_fini(const struct lu_env *env, struct ofd_device *m,
                           struct lu_device *top)
{
        struct obd_device       *obd = ofd_obd(m);
        struct lustre_cfg_bufs   bufs;
        struct lustre_cfg       *lcfg;
        char                     flags[3] = "";

        ENTRY;

        lu_site_purge(env, top->ld_site, ~0);
        /* process cleanup, pass mdt obd name to get obd umount flags */
        lustre_cfg_bufs_reset(&bufs, obd->obd_name);
        if (obd->obd_force)
                strcat(flags, "F");
        if (obd->obd_fail)
                strcat(flags, "A");
        lustre_cfg_bufs_set_string(&bufs, 1, flags);
        lcfg = lustre_cfg_new(LCFG_CLEANUP, &bufs);
        if (!lcfg) {
                CERROR("Cannot alloc lcfg!\n");
                RETURN_EXIT;
        }

        LASSERT(top);
        top->ld_ops->ldo_process_config(env, top, lcfg);
        lustre_cfg_free(lcfg);

        lu_site_purge(env, top->ld_site, ~0);
        if (!cfs_hash_is_empty(top->ld_site->ls_obj_hash)) {
                LIBCFS_DEBUG_MSG_DATA_DECL(msgdata, D_ERROR, NULL);
                lu_site_print(env, top->ld_site, &msgdata, lu_cdebug_printer);
        }

        LASSERT(m->ofd_osd_exp);
        obd_disconnect(m->ofd_osd_exp);

        EXIT;
}

/* For interoperability, see mdt_interop_param[]. */
static struct cfg_interop_param ofd_interop_param[] = {
        { "ost.quota_type",     NULL },
        { NULL }
};

/**
 * Check if parameters are symlinks to the OSD.
 *
 * Some parameters were moved from ofd to osd and only their
 * symlinks were kept in ofd by LU-3106. They are:
 * -writehthrough_cache_enable
 * -readcache_max_filesize
 * -read_cache_enable
 * -brw_stats
 *
 * Since they are not included by the static lprocfs var list, a pre-check
 * is added for them to avoid "unknown param" errors. If they are matched
 * in this check, they will be passed to the OSD directly.
 *
 * \param[in] param     parameters to check
 *
 * \retval              true if param is symlink to OSD param
 *                      false otherwise
 */
static bool match_symlink_param(char *param)
{
        char *sval;
        int paramlen;

        if (class_match_param(param, PARAM_OST, &param) == 0) {
                sval = strchr(param, '=');
                if (sval != NULL) {
                        paramlen = sval - param;
                        if (strncmp(param, "writethrough_cache_enable",
                                    paramlen) == 0 ||
                            strncmp(param, "readcache_max_filesize",
                                    paramlen) == 0 ||
                            strncmp(param, "read_cache_enable",
                                    paramlen) == 0 ||
                            strncmp(param, "brw_stats", paramlen) == 0)
                                return true;
                }
        }

        return false;
}

/**
 * Process various configuration parameters.
 *
 * This function is used by MGS to process specific configurations and
 * pass them through to the next device in server stack, i.e. the OSD.
 *
 * \param[in] env       execution environment
 * \param[in] d         LU device of OFD
 * \param[in] cfg       parameters to process
 *
 * \retval              0 if successful
 * \retval              negative value on error
 */
static int ofd_process_config(const struct lu_env *env, struct lu_device *d,
                              struct lustre_cfg *cfg)
{
        struct ofd_device       *m = ofd_dev(d);
        struct dt_device        *dt_next = m->ofd_osd;
        struct lu_device        *next = &dt_next->dd_lu_dev;
        int                      rc;

        ENTRY;

        switch (cfg->lcfg_command) {
        case LCFG_PARAM: {
                struct obd_device       *obd = ofd_obd(m);
                /* For interoperability */
                struct cfg_interop_param   *ptr = NULL;
                struct lustre_cfg          *old_cfg = NULL;
                char                       *param = NULL;

                param = lustre_cfg_string(cfg, 1);
                if (param == NULL) {
                        CERROR("param is empty\n");
                        rc = -EINVAL;
                        break;
                }

                ptr = class_find_old_param(param, ofd_interop_param);
                if (ptr != NULL) {
                        if (ptr->new_param == NULL) {
                                rc = 0;
                                CWARN("For interoperability, skip this %s."
                                      " It is obsolete.\n", ptr->old_param);
                                break;
                        }

                        CWARN("Found old param %s, changed it to %s.\n",
                              ptr->old_param, ptr->new_param);

                        old_cfg = cfg;
                        cfg = lustre_cfg_rename(old_cfg, ptr->new_param);
                        if (IS_ERR(cfg)) {
                                rc = PTR_ERR(cfg);
                                break;
                        }
                }

                if (match_symlink_param(param)) {
                        rc = next->ld_ops->ldo_process_config(env, next, cfg);
                        break;
                }

                rc = class_process_proc_param(PARAM_OST, obd->obd_vars, cfg,
                                              d->ld_obd);
                if (rc > 0 || rc == -ENOSYS) {
                        CDEBUG(D_CONFIG, "pass param %s down the stack.\n",
                               param);
                        /* we don't understand; pass it on */
                        rc = next->ld_ops->ldo_process_config(env, next, cfg);
                }
                break;
        }
        case LCFG_SPTLRPC_CONF: {
                rc = -ENOTSUPP;
                break;
        }
        default:
                /* others are passed further */
                rc = next->ld_ops->ldo_process_config(env, next, cfg);
                break;
        }
        RETURN(rc);
}

/**
 * Implementation of lu_object_operations::loo_object_init for OFD
 *
 * Allocate just the next object (OSD) in stack.
 *
 * \param[in] env       execution environment
 * \param[in] o         lu_object of OFD object
 * \param[in] conf      additional configuration parameters, not used here
 *
 * \retval              0 if successful
 * \retval              negative value on error
 */
static int ofd_object_init(const struct lu_env *env, struct lu_object *o,
                           const struct lu_object_conf *conf)
{
        struct ofd_device       *d = ofd_dev(o->lo_dev);
        struct lu_device        *under;
        struct lu_object        *below;
        int                      rc = 0;

        ENTRY;

        CDEBUG(D_INFO, "object init, fid = "DFID"\n",
               PFID(lu_object_fid(o)));

        under = &d->ofd_osd->dd_lu_dev;
        below = under->ld_ops->ldo_object_alloc(env, o->lo_header, under);
        if (below != NULL)
                lu_object_add(o, below);
        else
                rc = -ENOMEM;

        RETURN(rc);
}

/**
 * Implementation of lu_object_operations::loo_object_free.
 *
 * Finish OFD object lifecycle and free its memory.
 *
 * \param[in] env       execution environment
 * \param[in] o         LU object of OFD object
 */
static void ofd_object_free(const struct lu_env *env, struct lu_object *o)
{
        struct ofd_object       *of = ofd_obj(o);
        struct lu_object_header *h;

        ENTRY;

        h = o->lo_header;
        CDEBUG(D_INFO, "object free, fid = "DFID"\n",
               PFID(lu_object_fid(o)));

        lu_object_fini(o);
        lu_object_header_fini(h);
        OBD_SLAB_FREE_PTR(of, ofd_object_kmem);
        EXIT;
}

/**
 * Implementation of lu_object_operations::loo_object_print.
 *
 * Print OFD part of compound OFD-OSD object. See lu_object_print() and
 * LU_OBJECT_DEBUG() for more details about the compound object printing.
 *
 * \param[in] env       execution environment
 * \param[in] cookie    opaque data passed to the printer function
 * \param[in] p         printer function to use
 * \param[in] o         LU object of OFD object
 *
 * \retval              0 if successful
 * \retval              negative value on error
 */
static int ofd_object_print(const struct lu_env *env, void *cookie,
                            lu_printer_t p, const struct lu_object *o)
{
        return (*p)(env, cookie, LUSTRE_OST_NAME"-object@%p", o);
}

struct lu_object_operations ofd_obj_ops = {
        .loo_object_init        = ofd_object_init,
        .loo_object_free        = ofd_object_free,
        .loo_object_print       = ofd_object_print
};

/**
 * Implementation of lu_device_operations::lod_object_alloc.
 *
 * This function allocates OFD part of compound OFD-OSD object and
 * initializes its header, because OFD is the top device in stack
 *
 * \param[in] env       execution environment
 * \param[in] hdr       object header, NULL for OFD
 * \param[in] d         lu_device
 *
 * \retval              allocated object if successful
 * \retval              NULL value on failed allocation
 */
static struct lu_object *ofd_object_alloc(const struct lu_env *env,
                                          const struct lu_object_header *hdr,
                                          struct lu_device *d)
{
        struct ofd_object *of;

        ENTRY;

        OBD_SLAB_ALLOC_PTR_GFP(of, ofd_object_kmem, GFP_NOFS);
        if (of != NULL) {
                struct lu_object        *o;
                struct lu_object_header *h;

                o = &of->ofo_obj.do_lu;
                h = &of->ofo_header;
                lu_object_header_init(h);
                lu_object_init(o, h, d);
                lu_object_add_top(h, o);
                o->lo_ops = &ofd_obj_ops;
                RETURN(o);
        } else {
                RETURN(NULL);
        }
}

/**
 * Return the result of LFSCK run to the OFD.
 *
 * Notify OFD about result of LFSCK run. That may block the new object
 * creation until problem is fixed by LFSCK.
 *
 * \param[in] env       execution environment
 * \param[in] data      pointer to the OFD device
 * \param[in] event     LFSCK event type
 *
 * \retval              0 if successful
 * \retval              negative value on unknown event
 */
static int ofd_lfsck_out_notify(const struct lu_env *env, void *data,
                                enum lfsck_events event)
{
        struct ofd_device *ofd = data;
        struct obd_device *obd = ofd_obd(ofd);

        switch (event) {
        case LE_LASTID_REBUILDING:
                CWARN("%s: Found crashed LAST_ID, deny creating new OST-object "
                      "on the device until the LAST_ID rebuilt successfully.\n",
                      obd->obd_name);
                down_write(&ofd->ofd_lastid_rwsem);
                ofd->ofd_lastid_rebuilding = 1;
                up_write(&ofd->ofd_lastid_rwsem);
                break;
        case LE_LASTID_REBUILT: {
                down_write(&ofd->ofd_lastid_rwsem);
                ofd_seqs_free(env, ofd);
                ofd->ofd_lastid_rebuilding = 0;
                ofd->ofd_lastid_gen++;
                up_write(&ofd->ofd_lastid_rwsem);
                CWARN("%s: Rebuilt crashed LAST_ID files successfully.\n",
                      obd->obd_name);
                break;
        }
        default:
                CERROR("%s: unknown lfsck event: rc = %d\n",
                       ofd_name(ofd), event);
                return -EINVAL;
        }

        return 0;
}

/**
 * Implementation of lu_device_operations::ldo_prepare.
 *
 * This method is called after layer has been initialized and before it starts
 * serving user requests. In OFD it starts lfsk check routines and initializes
 * recovery.
 *
 * \param[in] env       execution environment
 * \param[in] pdev      higher device in stack, NULL for OFD
 * \param[in] dev       lu_device of OFD device
 *
 * \retval              0 if successful
 * \retval              negative value on error
 */
static int ofd_prepare(const struct lu_env *env, struct lu_device *pdev,
                       struct lu_device *dev)
{
        struct ofd_thread_info          *info;
        struct ofd_device               *ofd = ofd_dev(dev);
        struct obd_device               *obd = ofd_obd(ofd);
        struct lu_device                *next = &ofd->ofd_osd->dd_lu_dev;
        int                              rc;

        ENTRY;

        info = ofd_info_init(env, NULL);
        if (info == NULL)
                RETURN(-EFAULT);

        /* initialize lower device */
        rc = next->ld_ops->ldo_prepare(env, dev, next);
        if (rc != 0)
                RETURN(rc);

        rc = lfsck_register(env, ofd->ofd_osd, ofd->ofd_osd, obd,
                            ofd_lfsck_out_notify, ofd, false);
        if (rc != 0) {
                CERROR("%s: failed to initialize lfsck: rc = %d\n",
                       obd->obd_name, rc);
                RETURN(rc);
        }

        rc = lfsck_register_namespace(env, ofd->ofd_osd, ofd->ofd_namespace);
        /* The LFSCK instance is registered just now, so it must be there when
         * register the namespace to such instance. */
        LASSERTF(rc == 0, "register namespace failed: rc = %d\n", rc);

        target_recovery_init(&ofd->ofd_lut, tgt_request_handle);
        LASSERT(obd->obd_no_conn);
        spin_lock(&obd->obd_dev_lock);
        obd->obd_no_conn = 0;
        spin_unlock(&obd->obd_dev_lock);

        if (obd->obd_recovering == 0)
                ofd_postrecov(env, ofd);

        RETURN(rc);
}

/**
 * Implementation of lu_device_operations::ldo_recovery_complete.
 *
 * This method notifies all layers about 'recovery complete' event. That means
 * device is in full state and consistent. An OFD calculates available grant
 * space upon this event.
 *
 * \param[in] env       execution environment
 * \param[in] dev       lu_device of OFD device
 *
 * \retval              0 if successful
 * \retval              negative value on error
 */
static int ofd_recovery_complete(const struct lu_env *env,
                                 struct lu_device *dev)
{
        struct ofd_device       *ofd = ofd_dev(dev);
        struct lu_device        *next = &ofd->ofd_osd->dd_lu_dev;
        int                      rc = 0, max_precreate;

        ENTRY;

        /*
         * Grant space for object precreation on the self export.
         * This initial reserved space (i.e. 10MB for zfs and 280KB for ldiskfs)
         * is enough to create 10k objects. More space is then acquired for
         * precreation in ofd_grant_create().
         */
        max_precreate = OST_MAX_PRECREATE * ofd->ofd_dt_conf.ddp_inodespace / 2;
        ofd_grant_connect(env, dev->ld_obd->obd_self_export, max_precreate,
                          false);
        rc = next->ld_ops->ldo_recovery_complete(env, next);
        RETURN(rc);
}

/**
 * lu_device_operations matrix for OFD device.
 */
static struct lu_device_operations ofd_lu_ops = {
        .ldo_object_alloc       = ofd_object_alloc,
        .ldo_process_config     = ofd_process_config,
        .ldo_recovery_complete  = ofd_recovery_complete,
        .ldo_prepare            = ofd_prepare,
};

LPROC_SEQ_FOPS(lprocfs_nid_stats_clear);

/**
 * Initialize all needed procfs entries for OFD device.
 *
 * \param[in] ofd       OFD device
 *
 * \retval              0 if successful
 * \retval              negative value on error
 */
static int ofd_procfs_init(struct ofd_device *ofd)
{
        struct obd_device               *obd = ofd_obd(ofd);
        struct proc_dir_entry           *entry;
        int                              rc = 0;

        ENTRY;

        /* lprocfs must be setup before the ofd so state can be safely added
         * to /proc incrementally as the ofd is setup */
        obd->obd_vars = lprocfs_ofd_obd_vars;
        rc = lprocfs_obd_setup(obd);
        if (rc) {
                CERROR("%s: lprocfs_obd_setup failed: %d.\n",
                       obd->obd_name, rc);
                RETURN(rc);
        }

        rc = lprocfs_alloc_obd_stats(obd, LPROC_OFD_STATS_LAST);
        if (rc) {
                CERROR("%s: lprocfs_alloc_obd_stats failed: %d.\n",
                       obd->obd_name, rc);
                GOTO(obd_cleanup, rc);
        }

        obd->obd_uses_nid_stats = 1;

        entry = lprocfs_seq_register("exports", obd->obd_proc_entry, NULL,
                                     NULL);
        if (IS_ERR(entry)) {
                rc = PTR_ERR(entry);
                CERROR("%s: error %d setting up lprocfs for %s\n",
                       obd->obd_name, rc, "exports");
                GOTO(obd_cleanup, rc);
        }
        obd->obd_proc_exports_entry = entry;

        entry = lprocfs_add_simple(obd->obd_proc_exports_entry, "clear",
                                   obd, &lprocfs_nid_stats_clear_fops);
        if (IS_ERR(entry)) {
                rc = PTR_ERR(entry);
                CERROR("%s: add proc entry 'clear' failed: %d.\n",
                       obd->obd_name, rc);
                GOTO(obd_cleanup, rc);
        }

        ofd_stats_counter_init(obd->obd_stats);

        rc = lprocfs_job_stats_init(obd, LPROC_OFD_STATS_LAST,
                                    ofd_stats_counter_init);
        if (rc)
                GOTO(obd_cleanup, rc);
        RETURN(0);
obd_cleanup:
        lprocfs_obd_cleanup(obd);
        lprocfs_free_obd_stats(obd);

        return rc;
}

/**
 * Expose OSD statistics to OFD layer.
 *
 * The osd interfaces to the backend file system exposes useful data
 * such as brw_stats and read or write cache states. This same data
 * needs to be exposed into the obdfilter (ofd) layer to maintain
 * backwards compatibility. This function creates the symlinks in the
 * proc layer to enable this.
 *
 * \param[in] ofd       OFD device
 */
static void ofd_procfs_add_brw_stats_symlink(struct ofd_device *ofd)
{
        struct obd_device       *obd = ofd_obd(ofd);
        struct obd_device       *osd_obd = ofd->ofd_osd_exp->exp_obd;

        if (obd->obd_proc_entry == NULL)
                return;

        lprocfs_add_symlink("brw_stats", obd->obd_proc_entry,
                            "../../%s/%s/brw_stats",
                            osd_obd->obd_type->typ_name, obd->obd_name);

        lprocfs_add_symlink("read_cache_enable", obd->obd_proc_entry,
                            "../../%s/%s/read_cache_enable",
                            osd_obd->obd_type->typ_name, obd->obd_name);

        lprocfs_add_symlink("readcache_max_filesize",
                            obd->obd_proc_entry,
                            "../../%s/%s/readcache_max_filesize",
                            osd_obd->obd_type->typ_name, obd->obd_name);

        lprocfs_add_symlink("writethrough_cache_enable",
                            obd->obd_proc_entry,
                            "../../%s/%s/writethrough_cache_enable",
                            osd_obd->obd_type->typ_name, obd->obd_name);
}

/**
 * Cleanup all procfs entries in OFD.
 *
 * \param[in] ofd       OFD device
 */
static void ofd_procfs_fini(struct ofd_device *ofd)
{
        struct obd_device *obd = ofd_obd(ofd);

        lprocfs_free_per_client_stats(obd);
        lprocfs_obd_cleanup(obd);
        lprocfs_free_obd_stats(obd);
        lprocfs_job_stats_fini(obd);
}

/**
 * Stop SEQ/FID server on OFD.
 *
 * \param[in] env       execution environment
 * \param[in] ofd       OFD device
 *
 * \retval              0 if successful
 * \retval              negative value on error
 */
int ofd_fid_fini(const struct lu_env *env, struct ofd_device *ofd)
{
        return seq_site_fini(env, &ofd->ofd_seq_site);
}

/**
 * Start SEQ/FID server on OFD.
 *
 * The SEQ/FID server on OFD is needed to allocate FIDs for new objects.
 * It also connects to the master server to get own FID sequence (SEQ) range
 * to this particular OFD. Typically that happens when the OST is first
 * formatted or in the rare case that it exhausts the local sequence range.
 *
 * The sequence range is allocated out to the MDTs for OST object allocations,
 * and not directly to the clients.
 *
 * \param[in] env       execution environment
 * \param[in] ofd       OFD device
 *
 * \retval              0 if successful
 * \retval              negative value on error
 */
int ofd_fid_init(const struct lu_env *env, struct ofd_device *ofd)
{
        struct seq_server_site  *ss = &ofd->ofd_seq_site;
        struct lu_device        *lu = &ofd->ofd_dt_dev.dd_lu_dev;
        char                    *obd_name = ofd_name(ofd);
        char                    *name = NULL;
        int                     rc = 0;

        ss = &ofd->ofd_seq_site;
        lu->ld_site->ld_seq_site = ss;
        ss->ss_lu = lu->ld_site;
        ss->ss_node_id = ofd->ofd_lut.lut_lsd.lsd_osd_index;

        OBD_ALLOC_PTR(ss->ss_server_seq);
        if (ss->ss_server_seq == NULL)
                GOTO(out_free, rc = -ENOMEM);

        OBD_ALLOC(name, strlen(obd_name) + 10);
        if (!name) {
                OBD_FREE_PTR(ss->ss_server_seq);
                ss->ss_server_seq = NULL;
                GOTO(out_free, rc = -ENOMEM);
        }

        rc = seq_server_init(env, ss->ss_server_seq, ofd->ofd_osd, obd_name,
                             LUSTRE_SEQ_SERVER, ss);
        if (rc) {
                CERROR("%s : seq server init error %d\n", obd_name, rc);
                GOTO(out_free, rc);
        }
        ss->ss_server_seq->lss_space.lsr_index = ss->ss_node_id;

        OBD_ALLOC_PTR(ss->ss_client_seq);
        if (ss->ss_client_seq == NULL)
                GOTO(out_free, rc = -ENOMEM);

        snprintf(name, strlen(obd_name) + 6, "%p-super", obd_name);
        rc = seq_client_init(ss->ss_client_seq, NULL, LUSTRE_SEQ_DATA,
                             name, NULL);
        if (rc) {
                CERROR("%s : seq client init error %d\n", obd_name, rc);
                GOTO(out_free, rc);
        }
        OBD_FREE(name, strlen(obd_name) + 10);
        name = NULL;

        rc = seq_server_set_cli(env, ss->ss_server_seq, ss->ss_client_seq);

out_free:
        if (rc) {
                if (ss->ss_server_seq) {
                        seq_server_fini(ss->ss_server_seq, env);
                        OBD_FREE_PTR(ss->ss_server_seq);
                        ss->ss_server_seq = NULL;
                }

                if (ss->ss_client_seq) {
                        seq_client_fini(ss->ss_client_seq);
                        OBD_FREE_PTR(ss->ss_client_seq);
                        ss->ss_client_seq = NULL;
                }

                if (name) {
                        OBD_FREE(name, strlen(obd_name) + 10);
                        name = NULL;
                }
        }

        return rc;
}

/**
 * OFD request handler for OST_SET_INFO RPC.
 *
 * This is OFD-specific part of request handling
 *
 * \param[in] tsi       target session environment for this request
 *
 * \retval              0 if successful
 * \retval              negative value on error
 */
int ofd_set_info_hdl(struct tgt_session_info *tsi)
{
        struct ptlrpc_request   *req = tgt_ses_req(tsi);
        struct ost_body         *body = NULL, *repbody;
        void                    *key, *val = NULL;
        int                      keylen, vallen, rc = 0;
        bool                     is_grant_shrink;
        struct ofd_device       *ofd = ofd_exp(tsi->tsi_exp);

        ENTRY;

        key = req_capsule_client_get(tsi->tsi_pill, &RMF_SETINFO_KEY);
        if (key == NULL) {
                DEBUG_REQ(D_HA, req, "no set_info key");
                RETURN(err_serious(-EFAULT));
        }
        keylen = req_capsule_get_size(tsi->tsi_pill, &RMF_SETINFO_KEY,
                                      RCL_CLIENT);

        val = req_capsule_client_get(tsi->tsi_pill, &RMF_SETINFO_VAL);
        if (val == NULL) {
                DEBUG_REQ(D_HA, req, "no set_info val");
                RETURN(err_serious(-EFAULT));
        }
        vallen = req_capsule_get_size(tsi->tsi_pill, &RMF_SETINFO_VAL,
                                      RCL_CLIENT);

        is_grant_shrink = KEY_IS(KEY_GRANT_SHRINK);
        if (is_grant_shrink)
                /* In this case the value is actually an RMF_OST_BODY, so we
                 * transmutate the type of this PTLRPC */
                req_capsule_extend(tsi->tsi_pill, &RQF_OST_SET_GRANT_INFO);

        rc = req_capsule_server_pack(tsi->tsi_pill);
        if (rc < 0)
                RETURN(rc);

        if (is_grant_shrink) {
                body = req_capsule_client_get(tsi->tsi_pill, &RMF_OST_BODY);

                repbody = req_capsule_server_get(tsi->tsi_pill, &RMF_OST_BODY);
                *repbody = *body;

                /** handle grant shrink, similar to a read request */
                ofd_grant_prepare_read(tsi->tsi_env, tsi->tsi_exp,
                                       &repbody->oa);
        } else if (KEY_IS(KEY_EVICT_BY_NID)) {
                if (vallen > 0)
                        obd_export_evict_by_nid(tsi->tsi_exp->exp_obd, val);
                rc = 0;
        } else if (KEY_IS(KEY_CAPA_KEY)) {
                rc = ofd_update_capa_key(ofd, val);
        } else if (KEY_IS(KEY_SPTLRPC_CONF)) {
                rc = tgt_adapt_sptlrpc_conf(tsi->tsi_tgt, 0);
        } else {
                CERROR("%s: Unsupported key %s\n",
                       tgt_name(tsi->tsi_tgt), (char *)key);
                rc = -EOPNOTSUPP;
        }
        ofd_counter_incr(tsi->tsi_exp, LPROC_OFD_STATS_SET_INFO,
                         tsi->tsi_jobid, 1);

        RETURN(rc);
}

/**
 * Get FIEMAP (FIle Extent MAPping) for object with the given FID.
 *
 * This function returns a list of extents which describes how a file's
 * blocks are laid out on the disk.
 *
 * \param[in] env       execution environment
 * \param[in] ofd       OFD device
 * \param[in] fid       FID of object
 * \param[in] fiemap    fiemap structure to fill with data
 *
 * \retval              0 if \a fiemap is filled with data successfully
 * \retval              negative value on error
 */
int ofd_fiemap_get(const struct lu_env *env, struct ofd_device *ofd,
                   struct lu_fid *fid, struct ll_user_fiemap *fiemap)
{
        struct ofd_object       *fo;
        int                      rc;

        fo = ofd_object_find(env, ofd, fid);
        if (IS_ERR(fo)) {
                CERROR("%s: error finding object "DFID"\n",
                       ofd_name(ofd), PFID(fid));
                return PTR_ERR(fo);
        }

        ofd_read_lock(env, fo);
        if (ofd_object_exists(fo))
                rc = dt_fiemap_get(env, ofd_object_child(fo), fiemap);
        else
                rc = -ENOENT;
        ofd_read_unlock(env, fo);
        ofd_object_put(env, fo);
        return rc;
}

struct locked_region {
        struct list_head        list;
        struct lustre_handle    lh;
};

/**
 * Lock single extent and save lock handle in the list.
 *
 * This is supplemental function for lock_zero_regions(). It allocates
 * new locked_region structure and locks it with extent lock, then adds
 * it to the list of all such regions.
 *
 * \param[in] ns        LDLM namespace
 * \param[in] res_id    resource ID
 * \param[in] begin     start of region
 * \param[in] end       end of region
 * \param[in] locked    list head of regions list
 *
 * \retval              0 if successful locking
 * \retval              negative value on error
 */
static int lock_region(struct ldlm_namespace *ns, struct ldlm_res_id *res_id,
                       unsigned long long begin, unsigned long long end,
                       struct list_head *locked)
{
        struct locked_region    *region = NULL;
        __u64                    flags = 0;
        int                      rc;

        LASSERT(begin <= end);
        OBD_ALLOC_PTR(region);
        if (region == NULL)
                return -ENOMEM;

        rc = tgt_extent_lock(ns, res_id, begin, end, &region->lh,
                             LCK_PR, &flags);
        if (rc != 0)
                return rc;

        CDEBUG(D_OTHER, "ost lock [%llu,%llu], lh=%p\n", begin, end,
               &region->lh);
        list_add(&region->list, locked);

        return 0;
}

/**
 * Lock the sparse areas of given resource.
 *
 * The locking of sparse areas will cause dirty data to be flushed back from
 * clients. This is used when getting the FIEMAP of an object to make sure
 * there is no unaccounted cached data on clients.
 *
 * This function goes through \a fiemap list of extents and locks only sparse
 * areas between extents.
 *
 * \param[in] ns        LDLM namespace
 * \param[in] res_id    resource ID
 * \param[in] fiemap    file extents mapping on disk
 * \param[in] locked    list head of regions list
 *
 * \retval              0 if successful
 * \retval              negative value on error
 */
static int lock_zero_regions(struct ldlm_namespace *ns,
                             struct ldlm_res_id *res_id,
                             struct ll_user_fiemap *fiemap,
                             struct list_head *locked)
{
        __u64 begin = fiemap->fm_start;
        unsigned int i;
        int rc = 0;
        struct ll_fiemap_extent *fiemap_start = fiemap->fm_extents;

        ENTRY;

        CDEBUG(D_OTHER, "extents count %u\n", fiemap->fm_mapped_extents);
        for (i = 0; i < fiemap->fm_mapped_extents; i++) {
                if (fiemap_start[i].fe_logical > begin) {
                        CDEBUG(D_OTHER, "ost lock [%llu,%llu]\n",
                               begin, fiemap_start[i].fe_logical);
                        rc = lock_region(ns, res_id, begin,
                                         fiemap_start[i].fe_logical, locked);
                        if (rc)
                                RETURN(rc);
                }

                begin = fiemap_start[i].fe_logical + fiemap_start[i].fe_length;
        }

        if (begin < (fiemap->fm_start + fiemap->fm_length)) {
                CDEBUG(D_OTHER, "ost lock [%llu,%llu]\n",
                       begin, fiemap->fm_start + fiemap->fm_length);
                rc = lock_region(ns, res_id, begin,
                                 fiemap->fm_start + fiemap->fm_length, locked);
        }

        RETURN(rc);
}

/**
 * Unlock all previously locked sparse areas for given resource.
 *
 * This function goes through list of locked regions, unlocking and freeing
 * them one-by-one.
 *
 * \param[in] ns        LDLM namespace
 * \param[in] locked    list head of regions list
 */
static void
unlock_zero_regions(struct ldlm_namespace *ns, struct list_head *locked)
{
        struct locked_region *entry, *temp;

        list_for_each_entry_safe(entry, temp, locked, list) {
                CDEBUG(D_OTHER, "ost unlock lh=%p\n", &entry->lh);
                tgt_extent_unlock(&entry->lh, LCK_PR);
                list_del(&entry->list);
                OBD_FREE_PTR(entry);
        }
}

/**
 * OFD request handler for OST_GET_INFO RPC.
 *
 * This is OFD-specific part of request handling. The OFD-specific keys are:
 * - KEY_LAST_ID (obsolete)
 * - KEY_FIEMAP
 * - KEY_LAST_FID
 *
 * This function reads needed data from storage and fills reply with it.
 *
 * Note: the KEY_LAST_ID is obsolete, replaced by KEY_LAST_FID on newer MDTs,
 * and is kept for compatibility.
 *
 * \param[in] tsi       target session environment for this request
 *
 * \retval              0 if successful
 * \retval              negative value on error
 */
int ofd_get_info_hdl(struct tgt_session_info *tsi)
{
        struct obd_export               *exp = tsi->tsi_exp;
        struct ofd_device               *ofd = ofd_exp(exp);
        struct ofd_thread_info          *fti = tsi2ofd_info(tsi);
        void                            *key;
        int                              keylen;
        int                              replylen, rc = 0;

        ENTRY;

        /* this common part for get_info rpc */
        key = req_capsule_client_get(tsi->tsi_pill, &RMF_GETINFO_KEY);
        if (key == NULL) {
                DEBUG_REQ(D_HA, tgt_ses_req(tsi), "no get_info key");
                RETURN(err_serious(-EPROTO));
        }
        keylen = req_capsule_get_size(tsi->tsi_pill, &RMF_GETINFO_KEY,
                                      RCL_CLIENT);

        if (KEY_IS(KEY_LAST_ID)) {
                obd_id          *last_id;
                struct ofd_seq  *oseq;

                req_capsule_extend(tsi->tsi_pill, &RQF_OST_GET_INFO_LAST_ID);
                rc = req_capsule_server_pack(tsi->tsi_pill);
                if (rc)
                        RETURN(err_serious(rc));

                last_id = req_capsule_server_get(tsi->tsi_pill, &RMF_OBD_ID);

                oseq = ofd_seq_load(tsi->tsi_env, ofd,
                                    (obd_seq)exp->exp_filter_data.fed_group);
                if (IS_ERR(oseq))
                        rc = -EFAULT;
                else
                        *last_id = ofd_seq_last_oid(oseq);
                ofd_seq_put(tsi->tsi_env, oseq);
        } else if (KEY_IS(KEY_FIEMAP)) {
                struct ll_fiemap_info_key       *fm_key;
                struct ll_user_fiemap           *fiemap;
                struct lu_fid                   *fid;

                req_capsule_extend(tsi->tsi_pill, &RQF_OST_GET_INFO_FIEMAP);

                fm_key = req_capsule_client_get(tsi->tsi_pill, &RMF_FIEMAP_KEY);
                rc = tgt_validate_obdo(tsi, &fm_key->oa);
                if (rc)
                        RETURN(err_serious(rc));

                fid = &fm_key->oa.o_oi.oi_fid;

                CDEBUG(D_INODE, "get FIEMAP of object "DFID"\n", PFID(fid));

                replylen = fiemap_count_to_size(fm_key->fiemap.fm_extent_count);
                req_capsule_set_size(tsi->tsi_pill, &RMF_FIEMAP_VAL,
                                     RCL_SERVER, replylen);

                rc = req_capsule_server_pack(tsi->tsi_pill);
                if (rc)
                        RETURN(err_serious(rc));

                fiemap = req_capsule_server_get(tsi->tsi_pill, &RMF_FIEMAP_VAL);
                if (fiemap == NULL)
                        RETURN(-ENOMEM);

                *fiemap = fm_key->fiemap;
                rc = ofd_fiemap_get(tsi->tsi_env, ofd, fid, fiemap);

                /* LU-3219: Lock the sparse areas to make sure dirty
                 * flushed back from client, then call fiemap again. */
                if (fm_key->oa.o_valid & OBD_MD_FLFLAGS &&
                    fm_key->oa.o_flags & OBD_FL_SRVLOCK) {
                        struct list_head locked;

                        INIT_LIST_HEAD(&locked);
                        ost_fid_build_resid(fid, &fti->fti_resid);
                        rc = lock_zero_regions(ofd->ofd_namespace,
                                               &fti->fti_resid, fiemap,
                                               &locked);
                        if (rc == 0 && !list_empty(&locked)) {
                                rc = ofd_fiemap_get(tsi->tsi_env, ofd, fid,
                                                    fiemap);
                                unlock_zero_regions(ofd->ofd_namespace,
                                                    &locked);
                        }
                }
        } else if (KEY_IS(KEY_LAST_FID)) {
                struct ofd_device       *ofd = ofd_exp(exp);
                struct ofd_seq          *oseq;
                struct lu_fid           *fid;
                int                      rc;

                req_capsule_extend(tsi->tsi_pill, &RQF_OST_GET_INFO_LAST_FID);
                rc = req_capsule_server_pack(tsi->tsi_pill);
                if (rc)
                        RETURN(err_serious(rc));

                fid = req_capsule_client_get(tsi->tsi_pill, &RMF_FID);
                if (fid == NULL)
                        RETURN(err_serious(-EPROTO));

                fid_le_to_cpu(&fti->fti_ostid.oi_fid, fid);

                fid = req_capsule_server_get(tsi->tsi_pill, &RMF_FID);
                if (fid == NULL)
                        RETURN(-ENOMEM);

                oseq = ofd_seq_load(tsi->tsi_env, ofd,
                                    ostid_seq(&fti->fti_ostid));
                if (IS_ERR(oseq))
                        RETURN(PTR_ERR(oseq));

                rc = ostid_to_fid(fid, &oseq->os_oi,
                                  ofd->ofd_lut.lut_lsd.lsd_osd_index);
                if (rc != 0)
                        GOTO(out_put, rc);

                CDEBUG(D_HA, "%s: LAST FID is "DFID"\n", ofd_name(ofd),
                       PFID(fid));
out_put:
                ofd_seq_put(tsi->tsi_env, oseq);
        } else {
                CERROR("%s: not supported key %s\n", tgt_name(tsi->tsi_tgt),
                       (char *)key);
                rc = -EOPNOTSUPP;
        }
        ofd_counter_incr(tsi->tsi_exp, LPROC_OFD_STATS_GET_INFO,
                         tsi->tsi_jobid, 1);

        RETURN(rc);
}

/**
 * OFD request handler for OST_GETATTR RPC.
 *
 * This is OFD-specific part of request handling. It finds the OFD object
 * by its FID, gets attributes from storage and packs result to the reply.
 *
 * \param[in] tsi       target session environment for this request
 *
 * \retval              0 if successful
 * \retval              negative value on error
 */
static int ofd_getattr_hdl(struct tgt_session_info *tsi)
{
        struct ofd_thread_info  *fti = tsi2ofd_info(tsi);
        struct ofd_device       *ofd = ofd_exp(tsi->tsi_exp);
        struct ost_body         *repbody;
        struct lustre_handle     lh = { 0 };
        struct ofd_object       *fo;
        __u64                    flags = 0;
        ldlm_mode_t              lock_mode = LCK_PR;
        bool                     srvlock;
        int                      rc;
        ENTRY;

        LASSERT(tsi->tsi_ost_body != NULL);

        repbody = req_capsule_server_get(tsi->tsi_pill, &RMF_OST_BODY);
        if (repbody == NULL)
                RETURN(-ENOMEM);

        repbody->oa.o_oi = tsi->tsi_ost_body->oa.o_oi;
        repbody->oa.o_valid = OBD_MD_FLID | OBD_MD_FLGROUP;

        srvlock = tsi->tsi_ost_body->oa.o_valid & OBD_MD_FLFLAGS &&
                  tsi->tsi_ost_body->oa.o_flags & OBD_FL_SRVLOCK;

        if (srvlock) {
                if (unlikely(tsi->tsi_ost_body->oa.o_flags & OBD_FL_FLUSH))
                        lock_mode = LCK_PW;

                rc = tgt_extent_lock(tsi->tsi_tgt->lut_obd->obd_namespace,
                                     &tsi->tsi_resid, 0, OBD_OBJECT_EOF, &lh,
                                     lock_mode, &flags);
                if (rc != 0)
                        RETURN(rc);
        }

        fo = ofd_object_find_exists(tsi->tsi_env, ofd, &tsi->tsi_fid);
        if (IS_ERR(fo))
                GOTO(out, rc = PTR_ERR(fo));

        rc = ofd_attr_get(tsi->tsi_env, fo, &fti->fti_attr);
        if (rc == 0) {
                __u64    curr_version;

                obdo_from_la(&repbody->oa, &fti->fti_attr,
                             OFD_VALID_FLAGS | LA_UID | LA_GID);
                tgt_drop_id(tsi->tsi_exp, &repbody->oa);

                /* Store object version in reply */
                curr_version = dt_version_get(tsi->tsi_env,
                                              ofd_object_child(fo));
                if ((__s64)curr_version != -EOPNOTSUPP) {
                        repbody->oa.o_valid |= OBD_MD_FLDATAVERSION;
                        repbody->oa.o_data_version = curr_version;
                }
        }

        ofd_object_put(tsi->tsi_env, fo);
out:
        if (srvlock)
                tgt_extent_unlock(&lh, lock_mode);

        ofd_counter_incr(tsi->tsi_exp, LPROC_OFD_STATS_GETATTR,
                         tsi->tsi_jobid, 1);

        repbody->oa.o_valid |= OBD_MD_FLFLAGS;
        repbody->oa.o_flags = OBD_FL_FLUSH;

        RETURN(rc);
}

/**
 * OFD request handler for OST_SETATTR RPC.
 *
 * This is OFD-specific part of request handling. It finds the OFD object
 * by its FID, sets attributes from request and packs result to the reply.
 *
 * \param[in] tsi       target session environment for this request
 *
 * \retval              0 if successful
 * \retval              negative value on error
 */
static int ofd_setattr_hdl(struct tgt_session_info *tsi)
{
        struct ofd_thread_info  *fti = tsi2ofd_info(tsi);
        struct ofd_device       *ofd = ofd_exp(tsi->tsi_exp);
        struct ost_body         *body = tsi->tsi_ost_body;
        struct ost_body         *repbody;
        struct ldlm_resource    *res;
        struct ofd_object       *fo;
        struct filter_fid       *ff = NULL;
        int                      rc = 0;

        ENTRY;

        LASSERT(body != NULL);

        repbody = req_capsule_server_get(tsi->tsi_pill, &RMF_OST_BODY);
        if (repbody == NULL)
                RETURN(-ENOMEM);

        repbody->oa.o_oi = body->oa.o_oi;
        repbody->oa.o_valid = OBD_MD_FLID | OBD_MD_FLGROUP;

        /* This would be very bad - accidentally truncating a file when
         * changing the time or similar - bug 12203. */
        if (body->oa.o_valid & OBD_MD_FLSIZE &&
            body->oa.o_size != OBD_OBJECT_EOF) {
                static char mdsinum[48];

                if (body->oa.o_valid & OBD_MD_FLFID)
                        snprintf(mdsinum, sizeof(mdsinum) - 1,
                                 "of parent "DFID, body->oa.o_parent_seq,
                                 body->oa.o_parent_oid, 0);
                else
                        mdsinum[0] = '\0';

                CERROR("%s: setattr from %s is trying to truncate object "DFID
                       " %s\n", ofd_name(ofd), obd_export_nid2str(tsi->tsi_exp),
                       PFID(&tsi->tsi_fid), mdsinum);
                RETURN(-EPERM);
        }

        fo = ofd_object_find_exists(tsi->tsi_env, ofd, &tsi->tsi_fid);
        if (IS_ERR(fo))
                GOTO(out, rc = PTR_ERR(fo));

        la_from_obdo(&fti->fti_attr, &body->oa, body->oa.o_valid);
        fti->fti_attr.la_valid &= ~LA_TYPE;

        if (body->oa.o_valid & OBD_MD_FLFID) {
                ff = &fti->fti_mds_fid;
                ofd_prepare_fidea(ff, &body->oa);
        }

        /* setting objects attributes (including owner/group) */
        rc = ofd_attr_set(tsi->tsi_env, fo, &fti->fti_attr, ff);
        if (rc != 0)
                GOTO(out_put, rc);

        obdo_from_la(&repbody->oa, &fti->fti_attr,
                     OFD_VALID_FLAGS | LA_UID | LA_GID);
        tgt_drop_id(tsi->tsi_exp, &repbody->oa);

        ofd_counter_incr(tsi->tsi_exp, LPROC_OFD_STATS_SETATTR,
                         tsi->tsi_jobid, 1);
        EXIT;
out_put:
        ofd_object_put(tsi->tsi_env, fo);
out:
        if (rc == 0) {
                /* we do not call this before to avoid lu_object_find() in
                 *  ->lvbo_update() holding another reference on the object.
                 * otherwise concurrent destroy can make the object unavailable
                 * for 2nd lu_object_find() waiting for the first reference
                 * to go... deadlock! */
                res = ldlm_resource_get(ofd->ofd_namespace, NULL,
                                        &tsi->tsi_resid, LDLM_EXTENT, 0);
                if (!IS_ERR(res)) {
                        ldlm_res_lvbo_update(res, NULL, 0);
                        ldlm_resource_putref(res);
                }
        }
        return rc;
}

/**
 * Destroy OST orphans.
 *
 * This is part of OST_CREATE RPC handling. If there is flag OBD_FL_DELORPHAN
 * set then we must destroy possible orphaned objects.
 *
 * \param[in] env       execution environment
 * \param[in] exp       OBD export
 * \param[in] ofd       OFD device
 * \param[in] oa        obdo structure for reply
 *
 * \retval              0 if successful
 * \retval              negative value on error
 */
static int ofd_orphans_destroy(const struct lu_env *env,
                               struct obd_export *exp,
                               struct ofd_device *ofd, struct obdo *oa)
{
        struct ofd_thread_info  *info   = ofd_info(env);
        struct lu_fid           *fid    = &info->fti_fid;
        struct ost_id           *oi     = &oa->o_oi;
        struct ofd_seq          *oseq;
        obd_seq                  seq    = ostid_seq(oi);
        obd_id                   end_id = ostid_id(oi);
        obd_id                   last;
        obd_id                   oid;
        int                      skip_orphan;
        int                      rc     = 0;

        ENTRY;

        oseq = ofd_seq_get(ofd, seq);
        if (oseq == NULL) {
                CERROR("%s: Can not find seq for "DOSTID"\n",
                       ofd_name(ofd), POSTID(oi));
                RETURN(-EINVAL);
        }

        *fid = oi->oi_fid;
        last = ofd_seq_last_oid(oseq);
        oid = last;

        LASSERT(exp != NULL);
        skip_orphan = !!(exp_connect_flags(exp) & OBD_CONNECT_SKIP_ORPHAN);

        if (OBD_FAIL_CHECK(OBD_FAIL_OST_NODESTROY))
                goto done;

        LCONSOLE(D_INFO, "%s: deleting orphan objects from "DOSTID
                 " to "DOSTID"\n", ofd_name(ofd), seq, end_id + 1, seq, last);

        while (oid > end_id) {
                rc = fid_set_id(fid, oid);
                if (unlikely(rc != 0))
                        GOTO(out_put, rc);

                rc = ofd_destroy_by_fid(env, ofd, fid, 1);
                if (rc != 0 && rc != -ENOENT && rc != -ESTALE &&
                    likely(rc != -EREMCHG && rc != -EINPROGRESS))
                        /* this is pretty fatal... */
                        CEMERG("%s: error destroying precreated id "
                               DFID": rc = %d\n",
                               ofd_name(ofd), PFID(fid), rc);

                oid--;
                if (!skip_orphan) {
                        ofd_seq_last_oid_set(oseq, oid);
                        /* update last_id on disk periodically so that if we
                         * restart * we don't need to re-scan all of the just
                         * deleted objects. */
                        if ((oid & 511) == 0)
                                ofd_seq_last_oid_write(env, ofd, oseq);
                }
        }

        CDEBUG(D_HA, "%s: after destroy: set last_id to "DOSTID"\n",
               ofd_name(ofd), seq, oid);

done:
        if (!skip_orphan) {
                ofd_seq_last_oid_set(oseq, oid);
                rc = ofd_seq_last_oid_write(env, ofd, oseq);
        } else {
                /* don't reuse orphan object, return last used objid */
                ostid_set_id(oi, last);
                rc = 0;
        }

        GOTO(out_put, rc);

out_put:
        ofd_seq_put(env, oseq);
        return rc;
}

/**
 * OFD request handler for OST_CREATE RPC.
 *
 * This is OFD-specific part of request handling. Its main purpose is to
 * create new data objects on OST, but it also used to destroy orphans.
 *
 * \param[in] tsi       target session environment for this request
 *
 * \retval              0 if successful
 * \retval              negative value on error
 */
static int ofd_create_hdl(struct tgt_session_info *tsi)
{
        struct ptlrpc_request   *req = tgt_ses_req(tsi);
        struct ost_body         *repbody;
        const struct obdo       *oa = &tsi->tsi_ost_body->oa;
        struct obdo             *rep_oa;
        struct obd_export       *exp = tsi->tsi_exp;
        struct ofd_device       *ofd = ofd_exp(exp);
        obd_seq                  seq = ostid_seq(&oa->o_oi);
        obd_id                   oid = ostid_id(&oa->o_oi);
        struct ofd_seq          *oseq;
        int                      rc = 0, diff;
        int                      sync_trans = 0;

        ENTRY;

        if (OBD_FAIL_CHECK(OBD_FAIL_OST_EROFS))
                RETURN(-EROFS);

        repbody = req_capsule_server_get(tsi->tsi_pill, &RMF_OST_BODY);
        if (repbody == NULL)
                RETURN(-ENOMEM);

        down_read(&ofd->ofd_lastid_rwsem);
        /* Currently, for safe, we do not distinguish which LAST_ID is broken,
         * we may do that in the future.
         * Return -ENOSPC until the LAST_ID rebuilt. */
        if (unlikely(ofd->ofd_lastid_rebuilding))
                GOTO(out_sem, rc = -ENOSPC);

        rep_oa = &repbody->oa;
        rep_oa->o_oi = oa->o_oi;

        LASSERT(seq >= FID_SEQ_OST_MDT0);
        LASSERT(oa->o_valid & OBD_MD_FLGROUP);

        CDEBUG(D_INFO, "ofd_create("DOSTID")\n", POSTID(&oa->o_oi));

        oseq = ofd_seq_load(tsi->tsi_env, ofd, seq);
        if (IS_ERR(oseq)) {
                CERROR("%s: Can't find FID Sequence "LPX64": rc = %ld\n",
                       ofd_name(ofd), seq, PTR_ERR(oseq));
                GOTO(out_sem, rc = -EINVAL);
        }

        if ((oa->o_valid & OBD_MD_FLFLAGS) &&
            (oa->o_flags & OBD_FL_RECREATE_OBJS)) {
                if (!ofd_obd(ofd)->obd_recovering ||
                    oid > ofd_seq_last_oid(oseq)) {
                        CERROR("%s: recreate objid "DOSTID" > last id "LPU64
                               "\n", ofd_name(ofd), POSTID(&oa->o_oi),
                               ofd_seq_last_oid(oseq));
                        GOTO(out_nolock, rc = -EINVAL);
                }
                /* Do nothing here, we re-create objects during recovery
                 * upon write replay, see ofd_preprw_write() */
                GOTO(out_nolock, rc = 0);
        }
        /* former ofd_handle_precreate */
        if ((oa->o_valid & OBD_MD_FLFLAGS) &&
            (oa->o_flags & OBD_FL_DELORPHAN)) {
                exp->exp_filter_data.fed_lastid_gen = ofd->ofd_lastid_gen;

                /* destroy orphans */
                if (lustre_msg_get_conn_cnt(tgt_ses_req(tsi)->rq_reqmsg) <
                    exp->exp_conn_cnt) {
                        CERROR("%s: dropping old orphan cleanup request\n",
                               ofd_name(ofd));
                        GOTO(out_nolock, rc = 0);
                }
                /* This causes inflight precreates to abort and drop lock */
                oseq->os_destroys_in_progress = 1;
                mutex_lock(&oseq->os_create_lock);
                if (!oseq->os_destroys_in_progress) {
                        CERROR("%s:["LPU64"] destroys_in_progress already"
                               " cleared\n", ofd_name(ofd), seq);
                        ostid_set_id(&rep_oa->o_oi, ofd_seq_last_oid(oseq));
                        GOTO(out, rc = 0);
                }
                diff = oid - ofd_seq_last_oid(oseq);
                CDEBUG(D_HA, "ofd_last_id() = "LPU64" -> diff = %d\n",
                        ofd_seq_last_oid(oseq), diff);
                if (-diff > OST_MAX_PRECREATE) {
                        /* FIXME: should reset precreate_next_id on MDS */
                        rc = 0;
                } else if (diff < 0) {
                        rc = ofd_orphans_destroy(tsi->tsi_env, exp,
                                                 ofd, rep_oa);
                        oseq->os_destroys_in_progress = 0;
                } else {
                        /* XXX: Used by MDS for the first time! */
                        oseq->os_destroys_in_progress = 0;
                }
        } else {
                if (unlikely(exp->exp_filter_data.fed_lastid_gen !=
                             ofd->ofd_lastid_gen)) {
                        /* Keep the export ref so we can send the reply. */
                        ofd_obd_disconnect(class_export_get(exp));
                        GOTO(out_nolock, rc = -ENOTCONN);
                }

                mutex_lock(&oseq->os_create_lock);
                if (lustre_msg_get_conn_cnt(tgt_ses_req(tsi)->rq_reqmsg) <
                    exp->exp_conn_cnt) {
                        CERROR("%s: dropping old precreate request\n",
                               ofd_name(ofd));
                        GOTO(out, rc = 0);
                }
                /* only precreate if seq is 0, IDIF or normal and also o_id
                 * must be specfied */
                if ((!fid_seq_is_mdt(seq) && !fid_seq_is_norm(seq) &&
                     !fid_seq_is_idif(seq)) || oid == 0) {
                        diff = 1; /* shouldn't we create this right now? */
                } else {
                        diff = oid - ofd_seq_last_oid(oseq);
                        /* Do sync create if the seq is about to used up */
                        if (fid_seq_is_idif(seq) || fid_seq_is_mdt0(seq)) {
                                if (unlikely(oid >= IDIF_MAX_OID - 1))
                                        sync_trans = 1;
                        } else if (fid_seq_is_norm(seq)) {
                                if (unlikely(oid >=
                                             LUSTRE_DATA_SEQ_MAX_WIDTH - 1))
                                        sync_trans = 1;
                        } else {
                                CERROR("%s : invalid o_seq "DOSTID"\n",
                                       ofd_name(ofd), POSTID(&oa->o_oi));
                                GOTO(out, rc = -EINVAL);
                        }
                }
        }
        if (diff > 0) {
                cfs_time_t       enough_time = cfs_time_shift(DISK_TIMEOUT);
                obd_id           next_id;
                int              created = 0;
                int              count;

                if (!(oa->o_valid & OBD_MD_FLFLAGS) ||
                    !(oa->o_flags & OBD_FL_DELORPHAN)) {
                        /* don't enforce grant during orphan recovery */
                        rc = ofd_grant_create(tsi->tsi_env,
                                              ofd_obd(ofd)->obd_self_export,
                                              &diff);
                        if (rc) {
                                CDEBUG(D_HA, "%s: failed to acquire grant "
                                       "space for precreate (%d): rc = %d\n",
                                       ofd_name(ofd), diff, rc);
                                diff = 0;
                        }
                }

                /* This can happen if a new OST is formatted and installed
                 * in place of an old one at the same index.  Instead of
                 * precreating potentially millions of deleted old objects
                 * (possibly filling the OST), only precreate the last batch.
                 * LFSCK will eventually clean up any orphans. LU-14 */
                if (diff > 5 * OST_MAX_PRECREATE) {
                        diff = OST_MAX_PRECREATE / 2;
                        LCONSOLE_WARN("%s: precreate FID "DOSTID" is over %u "
                                      "larger than the LAST_ID "DOSTID", only "
                                      "precreating the last %u objects.\n",
                                      ofd_name(ofd), POSTID(&oa->o_oi),
                                      5 * OST_MAX_PRECREATE,
                                      POSTID(&oseq->os_oi), diff);
                        ofd_seq_last_oid_set(oseq, ostid_id(&oa->o_oi) - diff);
                }

                while (diff > 0) {
                        next_id = ofd_seq_last_oid(oseq) + 1;
                        count = ofd_precreate_batch(ofd, diff);

                        CDEBUG(D_HA, "%s: reserve %d objects in group "LPX64
                               " at "LPU64"\n", ofd_name(ofd),
                               count, seq, next_id);

                        if (!(lustre_msg_get_flags(req->rq_reqmsg) & MSG_REPLAY)
                            && cfs_time_after(jiffies, enough_time)) {
                                CDEBUG(D_HA, "%s: Slow creates, %d/%d objects"
                                      " created at a rate of %d/s\n",
                                      ofd_name(ofd), created, diff + created,
                                      created / DISK_TIMEOUT);
                                break;
                        }

                        rc = ofd_precreate_objects(tsi->tsi_env, ofd, next_id,
                                                   oseq, count, sync_trans);
                        if (rc > 0) {
                                created += rc;
                                diff -= rc;
                        } else if (rc < 0) {
                                break;
                        }
                }

                if (diff > 0 &&
                    lustre_msg_get_flags(req->rq_reqmsg) & MSG_REPLAY)
                        LCONSOLE_WARN("%s: can't create the same count of"
                                      " objects when replaying the request"
                                      " (diff is %d). see LU-4621\n",
                                      ofd_name(ofd), diff);

                if (created > 0)
                        /* some objects got created, we can return
                         * them, even if last creation failed */
                        rc = 0;
                else
                        CERROR("%s: unable to precreate: rc = %d\n",
                               ofd_name(ofd), rc);

                if (!(oa->o_valid & OBD_MD_FLFLAGS) ||
                    !(oa->o_flags & OBD_FL_DELORPHAN))
                        ofd_grant_commit(tsi->tsi_env,
                                         ofd_obd(ofd)->obd_self_export, rc);

                ostid_set_id(&rep_oa->o_oi, ofd_seq_last_oid(oseq));
        }
        EXIT;
        ofd_counter_incr(exp, LPROC_OFD_STATS_CREATE,
                         tsi->tsi_jobid, 1);
out:
        mutex_unlock(&oseq->os_create_lock);
out_nolock:
        if (rc == 0) {
#if LUSTRE_VERSION_CODE < OBD_OCD_VERSION(2, 8, 53, 0)
                struct ofd_thread_info  *info = ofd_info(tsi->tsi_env);
                struct lu_fid           *fid = &info->fti_fid;

                /* For compatible purpose, it needs to convert back to
                 * OST ID before put it on wire. */
                *fid = rep_oa->o_oi.oi_fid;
                fid_to_ostid(fid, &rep_oa->o_oi);
#endif
                rep_oa->o_valid |= OBD_MD_FLID | OBD_MD_FLGROUP;
        }
        ofd_seq_put(tsi->tsi_env, oseq);

out_sem:
        up_read(&ofd->ofd_lastid_rwsem);
        return rc;
}

/**
 * OFD request handler for OST_DESTROY RPC.
 *
 * This is OFD-specific part of request handling. It destroys data objects
 * related to destroyed object on MDT.
 *
 * \param[in] tsi       target session environment for this request
 *
 * \retval              0 if successful
 * \retval              negative value on error
 */
static int ofd_destroy_hdl(struct tgt_session_info *tsi)
{
        const struct ost_body   *body = tsi->tsi_ost_body;
        struct ost_body         *repbody;
        struct ofd_device       *ofd = ofd_exp(tsi->tsi_exp);
        struct ofd_thread_info  *fti = tsi2ofd_info(tsi);
        struct lu_fid           *fid = &fti->fti_fid;
        obd_id                   oid;
        obd_count                count;
        int                      rc = 0;

        ENTRY;

        if (OBD_FAIL_CHECK(OBD_FAIL_OST_EROFS))
                RETURN(-EROFS);

        /* This is old case for clients before Lustre 2.4 */
        /* If there's a DLM request, cancel the locks mentioned in it */
        if (req_capsule_field_present(tsi->tsi_pill, &RMF_DLM_REQ,
                                      RCL_CLIENT)) {
                struct ldlm_request *dlm;

                dlm = req_capsule_client_get(tsi->tsi_pill, &RMF_DLM_REQ);
                if (dlm == NULL)
                        RETURN(-EFAULT);
                ldlm_request_cancel(tgt_ses_req(tsi), dlm, 0, LATF_SKIP);
        }

        *fid = body->oa.o_oi.oi_fid;
        oid = ostid_id(&body->oa.o_oi);
        LASSERT(oid != 0);

        repbody = req_capsule_server_get(tsi->tsi_pill, &RMF_OST_BODY);

        /* check that o_misc makes sense */
        if (body->oa.o_valid & OBD_MD_FLOBJCOUNT)
                count = body->oa.o_misc;
        else
                count = 1; /* default case - single destroy */

        CDEBUG(D_HA, "%s: Destroy object "DOSTID" count %d\n", ofd_name(ofd),
               POSTID(&body->oa.o_oi), count);

        while (count > 0) {
                int lrc;

                lrc = ofd_destroy_by_fid(tsi->tsi_env, ofd, fid, 0);
                if (lrc == -ENOENT) {
                        CDEBUG(D_INODE,
                               "%s: destroying non-existent object "DFID"\n",
                               ofd_name(ofd), PFID(fid));
                        /* rewrite rc with -ENOENT only if it is 0 */
                        if (rc == 0)
                                rc = lrc;
                } else if (lrc != 0) {
                        CERROR("%s: error destroying object "DFID": %d\n",
                               ofd_name(ofd), PFID(fid), lrc);
                        rc = lrc;
                }

                count--;
                oid++;
                lrc = fid_set_id(fid, oid);
                if (unlikely(lrc != 0 && count > 0))
                        GOTO(out, rc = lrc);
        }

        ofd_counter_incr(tsi->tsi_exp, LPROC_OFD_STATS_DESTROY,
                         tsi->tsi_jobid, 1);

        GOTO(out, rc);

out:
        fid_to_ostid(fid, &repbody->oa.o_oi);
        return rc;
}

/**
 * OFD request handler for OST_STATFS RPC.
 *
 * This function gets statfs data from storage as part of request
 * processing.
 *
 * \param[in] tsi       target session environment for this request
 *
 * \retval              0 if successful
 * \retval              negative value on error
 */
static int ofd_statfs_hdl(struct tgt_session_info *tsi)
{
        struct obd_statfs       *osfs;
        int                      rc;

        ENTRY;

        osfs = req_capsule_server_get(tsi->tsi_pill, &RMF_OBD_STATFS);

        rc = ofd_statfs(tsi->tsi_env, tsi->tsi_exp, osfs,
                        cfs_time_shift_64(-OBD_STATFS_CACHE_SECONDS), 0);
        if (rc != 0)
                CERROR("%s: statfs failed: rc = %d\n",
                       tgt_name(tsi->tsi_tgt), rc);

        if (OBD_FAIL_CHECK(OBD_FAIL_OST_STATFS_EINPROGRESS))
                rc = -EINPROGRESS;

        ofd_counter_incr(tsi->tsi_exp, LPROC_OFD_STATS_STATFS,
                         tsi->tsi_jobid, 1);

        RETURN(rc);
}

/**
 * OFD request handler for OST_SYNC RPC.
 *
 * Sync object data or all filesystem data to the disk and pack the
 * result in reply.
 *
 * \param[in] tsi       target session environment for this request
 *
 * \retval              0 if successful
 * \retval              negative value on error
 */
static int ofd_sync_hdl(struct tgt_session_info *tsi)
{
        struct ost_body         *body = tsi->tsi_ost_body;
        struct ost_body         *repbody;
        struct ofd_thread_info  *fti = tsi2ofd_info(tsi);
        struct ofd_device       *ofd = ofd_exp(tsi->tsi_exp);
        struct ofd_object       *fo = NULL;
        int                      rc = 0;

        ENTRY;

        repbody = req_capsule_server_get(tsi->tsi_pill, &RMF_OST_BODY);

        /* if no objid is specified, it means "sync whole filesystem" */
        if (!fid_is_zero(&tsi->tsi_fid)) {
                fo = ofd_object_find_exists(tsi->tsi_env, ofd, &tsi->tsi_fid);
                if (IS_ERR(fo))
                        RETURN(PTR_ERR(fo));
        }

        rc = tgt_sync(tsi->tsi_env, tsi->tsi_tgt,
                      fo != NULL ? ofd_object_child(fo) : NULL,
                      repbody->oa.o_size, repbody->oa.o_blocks);
        if (rc)
                GOTO(put, rc);

        ofd_counter_incr(tsi->tsi_exp, LPROC_OFD_STATS_SYNC,
                         tsi->tsi_jobid, 1);
        if (fo == NULL)
                RETURN(0);

        repbody->oa.o_oi = body->oa.o_oi;
        repbody->oa.o_valid = OBD_MD_FLID | OBD_MD_FLGROUP;

        rc = ofd_attr_get(tsi->tsi_env, fo, &fti->fti_attr);
        if (rc == 0)
                obdo_from_la(&repbody->oa, &fti->fti_attr,
                             OFD_VALID_FLAGS);
        else
                /* don't return rc from getattr */
                rc = 0;
        EXIT;
put:
        if (fo != NULL)
                ofd_object_put(tsi->tsi_env, fo);
        return rc;
}

/**
 * OFD request handler for OST_PUNCH RPC.
 *
 * This is part of request processing. Validate request fields,
 * punch (truncate) the given OFD object and pack reply.
 *
 * \param[in] tsi       target session environment for this request
 *
 * \retval              0 if successful
 * \retval              negative value on error
 */
static int ofd_punch_hdl(struct tgt_session_info *tsi)
{
        const struct obdo       *oa = &tsi->tsi_ost_body->oa;
        struct ost_body         *repbody;
        struct ofd_thread_info  *info = tsi2ofd_info(tsi);
        struct ldlm_namespace   *ns = tsi->tsi_tgt->lut_obd->obd_namespace;
        struct ldlm_resource    *res;
        struct ofd_object       *fo;
        struct filter_fid       *ff = NULL;
        __u64                    flags = 0;
        struct lustre_handle     lh = { 0, };
        int                      rc;
        __u64                    start, end;
        bool                     srvlock;

        ENTRY;

        /* check that we do support OBD_CONNECT_TRUNCLOCK. */
        CLASSERT(OST_CONNECT_SUPPORTED & OBD_CONNECT_TRUNCLOCK);

        if ((oa->o_valid & (OBD_MD_FLSIZE | OBD_MD_FLBLOCKS)) !=
            (OBD_MD_FLSIZE | OBD_MD_FLBLOCKS))
                RETURN(err_serious(-EPROTO));

        repbody = req_capsule_server_get(tsi->tsi_pill, &RMF_OST_BODY);
        if (repbody == NULL)
                RETURN(err_serious(-ENOMEM));

        /* punch start,end are passed in o_size,o_blocks throught wire */
        start = oa->o_size;
        end = oa->o_blocks;

        if (end != OBD_OBJECT_EOF) /* Only truncate is supported */
                RETURN(-EPROTO);

        /* standard truncate optimization: if file body is completely
         * destroyed, don't send data back to the server. */
        if (start == 0)
                flags |= LDLM_FL_AST_DISCARD_DATA;

        repbody->oa.o_oi = oa->o_oi;
        repbody->oa.o_valid = OBD_MD_FLID;

        srvlock = oa->o_valid & OBD_MD_FLFLAGS &&
                  oa->o_flags & OBD_FL_SRVLOCK;

        if (srvlock) {
                rc = tgt_extent_lock(ns, &tsi->tsi_resid, start, end, &lh,
                                     LCK_PW, &flags);
                if (rc != 0)
                        RETURN(rc);
        }

        CDEBUG(D_INODE, "calling punch for object "DFID", valid = "LPX64
               ", start = "LPD64", end = "LPD64"\n", PFID(&tsi->tsi_fid),
               oa->o_valid, start, end);

        fo = ofd_object_find_exists(tsi->tsi_env, ofd_exp(tsi->tsi_exp),
                                    &tsi->tsi_fid);
        if (IS_ERR(fo))
                GOTO(out, rc = PTR_ERR(fo));

        la_from_obdo(&info->fti_attr, oa,
                     OBD_MD_FLMTIME | OBD_MD_FLATIME | OBD_MD_FLCTIME);
        info->fti_attr.la_size = start;
        info->fti_attr.la_valid |= LA_SIZE;

        if (oa->o_valid & OBD_MD_FLFID) {
                ff = &info->fti_mds_fid;
                ofd_prepare_fidea(ff, oa);
        }

        rc = ofd_object_punch(tsi->tsi_env, fo, start, end, &info->fti_attr,
                              ff, (struct obdo *)oa);
        if (rc)
                GOTO(out_put, rc);

        ofd_counter_incr(tsi->tsi_exp, LPROC_OFD_STATS_PUNCH,
                         tsi->tsi_jobid, 1);
        EXIT;
out_put:
        ofd_object_put(tsi->tsi_env, fo);
out:
        if (srvlock)
                tgt_extent_unlock(&lh, LCK_PW);
        if (rc == 0) {
                /* we do not call this before to avoid lu_object_find() in
                 *  ->lvbo_update() holding another reference on the object.
                 * otherwise concurrent destroy can make the object unavailable
                 * for 2nd lu_object_find() waiting for the first reference
                 * to go... deadlock! */
                res = ldlm_resource_get(ns, NULL, &tsi->tsi_resid,
                                        LDLM_EXTENT, 0);
                if (!IS_ERR(res)) {
                        ldlm_res_lvbo_update(res, NULL, 0);
                        ldlm_resource_putref(res);
                }
        }
        return rc;
}

/**
 * OFD request handler for OST_QUOTACTL RPC.
 *
 * This is part of request processing to validate incoming request fields,
 * get the requested data from OSD and pack reply.
 *
 * \param[in] tsi       target session environment for this request
 *
 * \retval              0 if successful
 * \retval              negative value on error
 */
static int ofd_quotactl(struct tgt_session_info *tsi)
{
        struct obd_quotactl     *oqctl, *repoqc;
        int                      rc;

        ENTRY;

        oqctl = req_capsule_client_get(tsi->tsi_pill, &RMF_OBD_QUOTACTL);
        if (oqctl == NULL)
                RETURN(err_serious(-EPROTO));

        repoqc = req_capsule_server_get(tsi->tsi_pill, &RMF_OBD_QUOTACTL);
        if (repoqc == NULL)
                RETURN(err_serious(-ENOMEM));

        /* report success for quota on/off for interoperability with current MDT
         * stack */
        if (oqctl->qc_cmd == Q_QUOTAON || oqctl->qc_cmd == Q_QUOTAOFF)
                RETURN(0);

        *repoqc = *oqctl;
        rc = lquotactl_slv(tsi->tsi_env, tsi->tsi_tgt->lut_bottom, repoqc);

        ofd_counter_incr(tsi->tsi_exp, LPROC_OFD_STATS_QUOTACTL,
                         tsi->tsi_jobid, 1);

        RETURN(rc);
}

/**
 * Calculate the amount of time for lock prolongation.
 *
 * This is helper for ofd_prolong_extent_locks() function to get
 * the timeout extra time.
 *
 * \param[in] req       current request
 *
 * \retval              amount of time to extend the timeout with
 */
static inline int prolong_timeout(struct ptlrpc_request *req,
                                  struct ldlm_lock *lock)
{
        struct ptlrpc_service_part *svcpt = req->rq_rqbd->rqbd_svcpt;

        if (AT_OFF)
                return obd_timeout / 2;

        /* We are in the middle of the process - BL AST is sent, CANCEL
          is ahead. Take half of AT + IO process time. */
        return at_est2timeout(at_get(&svcpt->scp_at_estimate)) +
                (ldlm_bl_timeout(lock) >> 1);
}

/**
 * Prolong single lock timeout.
 *
 * This is supplemental function to the ofd_prolong_locks(). It prolongs
 * a single lock.
 *
 * \param[in] tsi       target session environment for this request
 * \param[in] lock      LDLM lock to prolong
 * \param[in] extent    related extent
 * \param[in] timeout   timeout value to add
 *
 * \retval              0 if lock is not suitable for prolongation
 * \retval              1 if lock was prolonged successfully
 */
static int ofd_prolong_one_lock(struct tgt_session_info *tsi,
                                struct ldlm_lock *lock,
                                struct ldlm_extent *extent)
{
        int timeout = prolong_timeout(tgt_ses_req(tsi), lock);

        if (lock->l_flags & LDLM_FL_DESTROYED) /* lock already cancelled */
                return 0;

        /* XXX: never try to grab resource lock here because we're inside
         * exp_bl_list_lock; in ldlm_lockd.c to handle waiting list we take
         * res lock and then exp_bl_list_lock. */

        if (!(lock->l_flags & LDLM_FL_AST_SENT))
                /* ignore locks not being cancelled */
                return 0;

        LDLM_DEBUG(lock, "refreshed for req x"LPU64" ext("LPU64"->"LPU64") "
                         "to %ds.\n", tgt_ses_req(tsi)->rq_xid, extent->start,
                         extent->end, timeout);

        /* OK. this is a possible lock the user holds doing I/O
         * let's refresh eviction timer for it */
        ldlm_refresh_waiting_lock(lock, timeout);
        return 1;
}

/**
 * Prolong lock timeout for the given extent.
 *
 * This function finds all locks related with incoming request and
 * prolongs their timeout.
 *
 * If a client is holding a lock for a long time while it sends
 * read or write RPCs to the OST for the object under this lock,
 * then we don't want the OST to evict the client. Otherwise,
 * if the network or disk is very busy then the client may not
 * be able to make any progress to clear out dirty pages under
 * the lock and the application will fail.
 *
 * Every time a Bulk Read/Write (BRW) request arrives for the object
 * covered by the lock, extend the timeout on that lock. The RPC should
 * contain a lock handle for the lock it is using, but this
 * isn't handled correctly by all client versions, and the
 * request may cover multiple locks.
 *
 * \param[in] tsi       target session environment for this request
 * \param[in] start     start of extent
 * \param[in] end       end of extent
 *
 * \retval              number of prolonged locks
 */
static int ofd_prolong_extent_locks(struct tgt_session_info *tsi,
                                    __u64 start, __u64 end)
{
        struct obd_export       *exp = tsi->tsi_exp;
        struct obdo             *oa  = &tsi->tsi_ost_body->oa;
        struct ldlm_extent       extent = {
                .start = start,
                .end = end
        };
        struct ldlm_lock        *lock;
        int                      lock_count = 0;

        ENTRY;

        if (oa->o_valid & OBD_MD_FLHANDLE) {
                /* mostly a request should be covered by only one lock, try
                 * fast path. */
                lock = ldlm_handle2lock(&oa->o_handle);
                if (lock != NULL) {
                        /* Fast path to check if the lock covers the whole IO
                         * region exclusively. */
                        if (lock->l_granted_mode == LCK_PW &&
                            ldlm_extent_contain(&lock->l_policy_data.l_extent,
                                                &extent)) {
                                /* bingo */
                                LASSERT(lock->l_export == exp);
                                lock_count = ofd_prolong_one_lock(tsi, lock,
                                                                  &extent);
                                LDLM_LOCK_PUT(lock);
                                RETURN(lock_count);
                        }
                        LDLM_LOCK_PUT(lock);
                }
        }

        spin_lock_bh(&exp->exp_bl_list_lock);
        list_for_each_entry(lock, &exp->exp_bl_list, l_exp_list) {
                LASSERT(lock->l_flags & LDLM_FL_AST_SENT);
                LASSERT(lock->l_resource->lr_type == LDLM_EXTENT);

                if (!ldlm_res_eq(&tsi->tsi_resid, &lock->l_resource->lr_name))
                        continue;

                if (!ldlm_extent_overlap(&lock->l_policy_data.l_extent,
                                         &extent))
                        continue;

                lock_count += ofd_prolong_one_lock(tsi, lock, &extent);
        }
        spin_unlock_bh(&exp->exp_bl_list_lock);

        RETURN(lock_count);
}

/**
 * Implementation of ptlrpc_hpreq_ops::hpreq_lock_match for OFD RW requests.
 *
 * Determine if \a lock and the lock from request \a req are equivalent
 * by comparing their resource names, modes, and extents.
 *
 * It is used to give priority to read and write RPCs being done
 * under this lock so that the client can drop the contended
 * lock more quickly and let other clients use it. This improves
 * overall performance in the case where the first client gets a
 * very large lock extent that prevents other clients from
 * submitting their writes.
 *
 * \param[in] req       ptlrpc_request being processed
 * \param[in] lock      contended lock to match
 *
 * \retval              1 if lock is matched
 * \retval              0 otherwise
 */
static int ofd_rw_hpreq_lock_match(struct ptlrpc_request *req,
                                   struct ldlm_lock *lock)
{
        struct niobuf_remote    *rnb;
        struct obd_ioobj        *ioo;
        ldlm_mode_t              mode;
        struct ldlm_extent       ext;
        __u32                    opc = lustre_msg_get_opc(req->rq_reqmsg);

        ENTRY;

        ioo = req_capsule_client_get(&req->rq_pill, &RMF_OBD_IOOBJ);
        LASSERT(ioo != NULL);

        rnb = req_capsule_client_get(&req->rq_pill, &RMF_NIOBUF_REMOTE);
        LASSERT(rnb != NULL);

        ext.start = rnb->rnb_offset;
        rnb += ioo->ioo_bufcnt - 1;
        ext.end = rnb->rnb_offset + rnb->rnb_len - 1;

        LASSERT(lock->l_resource != NULL);
        if (!ostid_res_name_eq(&ioo->ioo_oid, &lock->l_resource->lr_name))
                RETURN(0);

        /* a bulk write can only hold a reference on a PW extent lock */
        mode = LCK_PW;
        if (opc == OST_READ)
                /* whereas a bulk read can be protected by either a PR or PW
                 * extent lock */
                mode |= LCK_PR;

        if (!(lock->l_granted_mode & mode))
                RETURN(0);

        RETURN(ldlm_extent_overlap(&lock->l_policy_data.l_extent, &ext));
}

/**
 * Implementation of ptlrpc_hpreq_ops::hpreq_lock_check for OFD RW requests.
 *
 * Check for whether the given PTLRPC request (\a req) is blocking
 * an LDLM lock cancel.
 *
 * \param[in] req       the incoming request
 *
 * \retval              1 if \a req is blocking an LDLM lock cancel
 * \retval              0 if it is not
 */
static int ofd_rw_hpreq_check(struct ptlrpc_request *req)
{
        struct tgt_session_info *tsi;
        struct obd_ioobj        *ioo;
        struct niobuf_remote    *rnb;
        __u64                    start, end;
        int                      lock_count;

        ENTRY;

        /* Don't use tgt_ses_info() to get session info, because lock_match()
         * can be called while request has no processing thread yet. */
        tsi = lu_context_key_get(&req->rq_session, &tgt_session_key);

        /*
         * Use LASSERT below because malformed RPCs should have
         * been filtered out in tgt_hpreq_handler().
         */
        ioo = req_capsule_client_get(&req->rq_pill, &RMF_OBD_IOOBJ);
        LASSERT(ioo != NULL);

        rnb = req_capsule_client_get(&req->rq_pill, &RMF_NIOBUF_REMOTE);
        LASSERT(rnb != NULL);
        LASSERT(!(rnb->rnb_flags & OBD_BRW_SRVLOCK));

        start = rnb->rnb_offset;
        rnb += ioo->ioo_bufcnt - 1;
        end = rnb->rnb_offset + rnb->rnb_len - 1;

        DEBUG_REQ(D_RPCTRACE, req, "%s %s: refresh rw locks: "DFID
                                   " ("LPU64"->"LPU64")\n",
                  tgt_name(tsi->tsi_tgt), current->comm,
                  PFID(&tsi->tsi_fid), start, end);

        lock_count = ofd_prolong_extent_locks(tsi, start, end);

        CDEBUG(D_DLMTRACE, "%s: refreshed %u locks timeout for req %p.\n",
               tgt_name(tsi->tsi_tgt), lock_count, req);

        RETURN(lock_count > 0);
}

/**
 * Implementation of ptlrpc_hpreq_ops::hpreq_lock_fini for OFD RW requests.
 *
 * Called after the request has been handled. It refreshes lock timeout again
 * so that client has more time to send lock cancel RPC.
 *
 * \param[in] req       request which is being processed.
 */
static void ofd_rw_hpreq_fini(struct ptlrpc_request *req)
{
        ofd_rw_hpreq_check(req);
}

/**
 * Implementation of ptlrpc_hpreq_ops::hpreq_lock_match for OST_PUNCH request.
 *
 * This function checks if the given lock is the same by its resname, mode
 * and extent as one taken from the request.
 * It is used to give priority to punch/truncate RPCs that might lead to
 * the fastest release of that lock when a lock is contended.
 *
 * \param[in] req       ptlrpc_request being processed
 * \param[in] lock      contended lock to match
 *
 * \retval              1 if lock is matched
 * \retval              0 otherwise
 */
static int ofd_punch_hpreq_lock_match(struct ptlrpc_request *req,
                                      struct ldlm_lock *lock)
{
        struct tgt_session_info *tsi;

        /* Don't use tgt_ses_info() to get session info, because lock_match()
         * can be called while request has no processing thread yet. */
        tsi = lu_context_key_get(&req->rq_session, &tgt_session_key);

        /*
         * Use LASSERT below because malformed RPCs should have
         * been filtered out in tgt_hpreq_handler().
         */
        LASSERT(tsi->tsi_ost_body != NULL);
        if (tsi->tsi_ost_body->oa.o_valid & OBD_MD_FLHANDLE &&
            tsi->tsi_ost_body->oa.o_handle.cookie == lock->l_handle.h_cookie)
                return 1;

        return 0;
}

/**
 * Implementation of ptlrpc_hpreq_ops::hpreq_lock_check for OST_PUNCH request.
 *
 * High-priority queue request check for whether the given punch request
 * (\a req) is blocking an LDLM lock cancel.
 *
 * \param[in] req       the incoming request
 *
 * \retval              1 if \a req is blocking an LDLM lock cancel
 * \retval              0 if it is not
 */
static int ofd_punch_hpreq_check(struct ptlrpc_request *req)
{
        struct tgt_session_info *tsi;
        struct obdo             *oa;
        int                      lock_count;

        ENTRY;

        /* Don't use tgt_ses_info() to get session info, because lock_match()
         * can be called while request has no processing thread yet. */
        tsi = lu_context_key_get(&req->rq_session, &tgt_session_key);
        LASSERT(tsi != NULL);
        oa = &tsi->tsi_ost_body->oa;

        LASSERT(!(oa->o_valid & OBD_MD_FLFLAGS &&
                  oa->o_flags & OBD_FL_SRVLOCK));

        CDEBUG(D_DLMTRACE,
               "%s: refresh locks: "LPU64"/"LPU64" ("LPU64"->"LPU64")\n",
               tgt_name(tsi->tsi_tgt), tsi->tsi_resid.name[0],
               tsi->tsi_resid.name[1], oa->o_size, oa->o_blocks);

        lock_count = ofd_prolong_extent_locks(tsi, oa->o_size, oa->o_blocks);

        CDEBUG(D_DLMTRACE, "%s: refreshed %u locks timeout for req %p.\n",
               tgt_name(tsi->tsi_tgt), lock_count, req);

        RETURN(lock_count > 0);
}

/**
 * Implementation of ptlrpc_hpreq_ops::hpreq_lock_fini for OST_PUNCH request.
 *
 * Called after the request has been handled. It refreshes lock timeout again
 * so that client has more time to send lock cancel RPC.
 *
 * \param[in] req       request which is being processed.
 */
static void ofd_punch_hpreq_fini(struct ptlrpc_request *req)
{
        ofd_punch_hpreq_check(req);
}

struct ptlrpc_hpreq_ops ofd_hpreq_rw = {
        .hpreq_lock_match       = ofd_rw_hpreq_lock_match,
        .hpreq_check            = ofd_rw_hpreq_check,
        .hpreq_fini             = ofd_rw_hpreq_fini
};

struct ptlrpc_hpreq_ops ofd_hpreq_punch = {
        .hpreq_lock_match       = ofd_punch_hpreq_lock_match,
        .hpreq_check            = ofd_punch_hpreq_check,
        .hpreq_fini             = ofd_punch_hpreq_fini
};

/**
 * Assign high priority operations to an IO request.
 *
 * Check if the incoming request is a candidate for
 * high-priority processing. If it is, assign it a high
 * priority operations table.
 *
 * \param[in] tsi       target session environment for this request
 */
static void ofd_hp_brw(struct tgt_session_info *tsi)
{
        struct niobuf_remote    *rnb;
        struct obd_ioobj        *ioo;

        ENTRY;

        ioo = req_capsule_client_get(tsi->tsi_pill, &RMF_OBD_IOOBJ);
        LASSERT(ioo != NULL); /* must exist after request preprocessing */
        if (ioo->ioo_bufcnt > 0) {
                rnb = req_capsule_client_get(tsi->tsi_pill, &RMF_NIOBUF_REMOTE);
                LASSERT(rnb != NULL); /* must exist after request preprocessing */

                /* no high priority if server lock is needed */
                if (rnb->rnb_flags & OBD_BRW_SRVLOCK)
                        return;
        }
        tgt_ses_req(tsi)->rq_ops = &ofd_hpreq_rw;
}

/**
 * Assign high priority operations to an punch request.
 *
 * Check if the incoming request is a candidate for
 * high-priority processing. If it is, assign it a high
 * priority operations table.
 *
 * \param[in] tsi       target session environment for this request
 */
static void ofd_hp_punch(struct tgt_session_info *tsi)
{
        LASSERT(tsi->tsi_ost_body != NULL); /* must exists if we are here */
        /* no high-priority if server lock is needed */
        if (tsi->tsi_ost_body->oa.o_valid & OBD_MD_FLFLAGS &&
            tsi->tsi_ost_body->oa.o_flags & OBD_FL_SRVLOCK)
                return;
        tgt_ses_req(tsi)->rq_ops = &ofd_hpreq_punch;
}

#define OBD_FAIL_OST_READ_NET   OBD_FAIL_OST_BRW_NET
#define OBD_FAIL_OST_WRITE_NET  OBD_FAIL_OST_BRW_NET
#define OST_BRW_READ    OST_READ
#define OST_BRW_WRITE   OST_WRITE

/**
 * Table of OFD-specific request handlers
 *
 * This table contains all opcodes accepted by OFD and
 * specifies handlers for them. The tgt_request_handler()
 * uses such table from each target to process incoming
 * requests.
 */
static struct tgt_handler ofd_tgt_handlers[] = {
TGT_RPC_HANDLER(OST_FIRST_OPC,
                0,                      OST_CONNECT,    tgt_connect,
                &RQF_CONNECT, LUSTRE_OBD_VERSION),
TGT_RPC_HANDLER(OST_FIRST_OPC,
                0,                      OST_DISCONNECT, tgt_disconnect,
                &RQF_OST_DISCONNECT, LUSTRE_OBD_VERSION),
TGT_RPC_HANDLER(OST_FIRST_OPC,
                0,                      OST_SET_INFO,   ofd_set_info_hdl,
                &RQF_OBD_SET_INFO, LUSTRE_OST_VERSION),
TGT_OST_HDL(0,                          OST_GET_INFO,   ofd_get_info_hdl),
TGT_OST_HDL(HABEO_CORPUS| HABEO_REFERO, OST_GETATTR,    ofd_getattr_hdl),
TGT_OST_HDL(HABEO_CORPUS| HABEO_REFERO | MUTABOR,
                                        OST_SETATTR,    ofd_setattr_hdl),
TGT_OST_HDL(0           | HABEO_REFERO | MUTABOR,
                                        OST_CREATE,     ofd_create_hdl),
TGT_OST_HDL(0           | HABEO_REFERO | MUTABOR,
                                        OST_DESTROY,    ofd_destroy_hdl),
TGT_OST_HDL(0           | HABEO_REFERO, OST_STATFS,     ofd_statfs_hdl),
TGT_OST_HDL_HP(HABEO_CORPUS| HABEO_REFERO,
                                        OST_BRW_READ,   tgt_brw_read,
                                                        ofd_hp_brw),
/* don't set CORPUS flag for brw_write because -ENOENT may be valid case */
TGT_OST_HDL_HP(HABEO_CORPUS| MUTABOR,   OST_BRW_WRITE,  tgt_brw_write,
                                                        ofd_hp_brw),
TGT_OST_HDL_HP(HABEO_CORPUS| HABEO_REFERO | MUTABOR,
                                        OST_PUNCH,      ofd_punch_hdl,
                                                        ofd_hp_punch),
TGT_OST_HDL(HABEO_CORPUS| HABEO_REFERO, OST_SYNC,       ofd_sync_hdl),
TGT_OST_HDL(0           | HABEO_REFERO, OST_QUOTACTL,   ofd_quotactl),
};

static struct tgt_opc_slice ofd_common_slice[] = {
        {
                .tos_opc_start  = OST_FIRST_OPC,
                .tos_opc_end    = OST_LAST_OPC,
                .tos_hs         = ofd_tgt_handlers
        },
        {
                .tos_opc_start  = OBD_FIRST_OPC,
                .tos_opc_end    = OBD_LAST_OPC,
                .tos_hs         = tgt_obd_handlers
        },
        {
                .tos_opc_start  = LDLM_FIRST_OPC,
                .tos_opc_end    = LDLM_LAST_OPC,
                .tos_hs         = tgt_dlm_handlers
        },
        {
                .tos_opc_start  = OUT_UPDATE_FIRST_OPC,
                .tos_opc_end    = OUT_UPDATE_LAST_OPC,
                .tos_hs         = tgt_out_handlers
        },
        {
                .tos_opc_start  = SEQ_FIRST_OPC,
                .tos_opc_end    = SEQ_LAST_OPC,
                .tos_hs         = seq_handlers
        },
        {
                .tos_opc_start  = LFSCK_FIRST_OPC,
                .tos_opc_end    = LFSCK_LAST_OPC,
                .tos_hs         = tgt_lfsck_handlers
        },
        {
                .tos_hs         = NULL
        }
};

/* context key constructor/destructor: ofd_key_init(), ofd_key_fini() */
LU_KEY_INIT_FINI(ofd, struct ofd_thread_info);

/**
 * Implementation of lu_context_key::lct_key_exit.
 *
 * Optional method called on lu_context_exit() for all allocated
 * keys.
 * It is used in OFD to sanitize context values which may be re-used
 * during another request processing by the same thread.
 *
 * \param[in] ctx       execution context
 * \param[in] key       context key
 * \param[in] data      ofd_thread_info
 */
static void ofd_key_exit(const struct lu_context *ctx,
                         struct lu_context_key *key, void *data)
{
        struct ofd_thread_info *info = data;

        info->fti_env = NULL;
        info->fti_exp = NULL;

        info->fti_xid = 0;
        info->fti_pre_version = 0;
        info->fti_used = 0;

        memset(&info->fti_attr, 0, sizeof info->fti_attr);
}

struct lu_context_key ofd_thread_key = {
        .lct_tags = LCT_DT_THREAD,
        .lct_init = ofd_key_init,
        .lct_fini = ofd_key_fini,
        .lct_exit = ofd_key_exit
};

/**
 * Initialize OFD device according to parameters in the config log \a cfg.
 *
 * This is the main starting point of OFD initialization. It fills all OFD
 * parameters with their initial values and calls other initializing functions
 * to set up all OFD subsystems.
 *
 * \param[in] env       execution environment
 * \param[in] m         OFD device
 * \param[in] ldt       LU device type of OFD
 * \param[in] cfg       configuration log
 *
 * \retval              0 if successful
 * \retval              negative value on error
 */
static int ofd_init0(const struct lu_env *env, struct ofd_device *m,
                     struct lu_device_type *ldt, struct lustre_cfg *cfg)
{
        const char              *dev = lustre_cfg_string(cfg, 0);
        struct ofd_thread_info  *info = NULL;
        struct obd_device       *obd;
        struct obd_statfs       *osfs;
        int                      rc;

        ENTRY;

        obd = class_name2obd(dev);
        if (obd == NULL) {
                CERROR("Cannot find obd with name %s\n", dev);
                RETURN(-ENODEV);
        }

        rc = lu_env_refill((struct lu_env *)env);
        if (rc != 0)
                RETURN(rc);

        obd->u.obt.obt_magic = OBT_MAGIC;

        m->ofd_fmd_max_num = OFD_FMD_MAX_NUM_DEFAULT;
        m->ofd_fmd_max_age = OFD_FMD_MAX_AGE_DEFAULT;

        spin_lock_init(&m->ofd_flags_lock);
        m->ofd_raid_degraded = 0;
        m->ofd_syncjournal = 0;
        ofd_slc_set(m);
        m->ofd_grant_compat_disable = 0;
        m->ofd_soft_sync_limit = OFD_SOFT_SYNC_LIMIT_DEFAULT;

        /* statfs data */
        spin_lock_init(&m->ofd_osfs_lock);
        m->ofd_osfs_age = cfs_time_shift_64(-1000);
        m->ofd_osfs_unstable = 0;
        m->ofd_statfs_inflight = 0;
        m->ofd_osfs_inflight = 0;

        /* grant data */
        spin_lock_init(&m->ofd_grant_lock);
        m->ofd_tot_dirty = 0;
        m->ofd_tot_granted = 0;
        m->ofd_tot_pending = 0;
        m->ofd_seq_count = 0;
        init_waitqueue_head(&m->ofd_inconsistency_thread.t_ctl_waitq);
        INIT_LIST_HEAD(&m->ofd_inconsistency_list);
        spin_lock_init(&m->ofd_inconsistency_lock);

        spin_lock_init(&m->ofd_batch_lock);
        init_rwsem(&m->ofd_lastid_rwsem);

        obd->u.filter.fo_fl_oss_capa = 0;
        INIT_LIST_HEAD(&obd->u.filter.fo_capa_keys);
        obd->u.filter.fo_capa_hash = init_capa_hash();
        if (obd->u.filter.fo_capa_hash == NULL)
                RETURN(-ENOMEM);

        m->ofd_dt_dev.dd_lu_dev.ld_ops = &ofd_lu_ops;
        m->ofd_dt_dev.dd_lu_dev.ld_obd = obd;
        /* set this lu_device to obd, because error handling need it */
        obd->obd_lu_dev = &m->ofd_dt_dev.dd_lu_dev;

        rc = ofd_procfs_init(m);
        if (rc) {
                CERROR("Can't init ofd lprocfs, rc %d\n", rc);
                RETURN(rc);
        }

        /* No connection accepted until configurations will finish */
        spin_lock(&obd->obd_dev_lock);
        obd->obd_no_conn = 1;
        spin_unlock(&obd->obd_dev_lock);
        obd->obd_replayable = 1;
        if (cfg->lcfg_bufcount > 4 && LUSTRE_CFG_BUFLEN(cfg, 4) > 0) {
                char *str = lustre_cfg_string(cfg, 4);

                if (strchr(str, 'n')) {
                        CWARN("%s: recovery disabled\n", obd->obd_name);
                        obd->obd_replayable = 0;
                }
        }

        info = ofd_info_init(env, NULL);
        if (info == NULL)
                RETURN(-EFAULT);

        rc = ofd_stack_init(env, m, cfg);
        if (rc) {
                CERROR("Can't init device stack, rc %d\n", rc);
                GOTO(err_fini_proc, rc);
        }

        ofd_procfs_add_brw_stats_symlink(m);

        /* populate cached statfs data */
        osfs = &ofd_info(env)->fti_u.osfs;
        rc = ofd_statfs_internal(env, m, osfs, 0, NULL);
        if (rc != 0) {
                CERROR("%s: can't get statfs data, rc %d\n", obd->obd_name, rc);
                GOTO(err_fini_stack, rc);
        }
        if (!IS_PO2(osfs->os_bsize)) {
                CERROR("%s: blocksize (%d) is not a power of 2\n",
                                obd->obd_name, osfs->os_bsize);
                GOTO(err_fini_stack, rc = -EPROTO);
        }
        m->ofd_blockbits = fls(osfs->os_bsize) - 1;

        m->ofd_precreate_batch = OFD_PRECREATE_BATCH_DEFAULT;
        if (osfs->os_bsize * osfs->os_blocks < OFD_PRECREATE_SMALL_FS)
                m->ofd_precreate_batch = OFD_PRECREATE_BATCH_SMALL;

        snprintf(info->fti_u.name, sizeof(info->fti_u.name), "%s-%s",
                 "filter"/*LUSTRE_OST_NAME*/, obd->obd_uuid.uuid);
        m->ofd_namespace = ldlm_namespace_new(obd, info->fti_u.name,
                                              LDLM_NAMESPACE_SERVER,
                                              LDLM_NAMESPACE_GREEDY,
                                              LDLM_NS_TYPE_OST);
        if (m->ofd_namespace == NULL)
                GOTO(err_fini_stack, rc = -ENOMEM);
        /* set obd_namespace for compatibility with old code */
        obd->obd_namespace = m->ofd_namespace;
        ldlm_register_intent(m->ofd_namespace, ofd_intent_policy);
        m->ofd_namespace->ns_lvbo = &ofd_lvbo;
        m->ofd_namespace->ns_lvbp = m;

        ptlrpc_init_client(LDLM_CB_REQUEST_PORTAL, LDLM_CB_REPLY_PORTAL,
                           "filter_ldlm_cb_client", &obd->obd_ldlm_client);

        dt_conf_get(env, m->ofd_osd, &m->ofd_dt_conf);

        /* Allow at most ddp_grant_reserved% of the available filesystem space
         * to be granted to clients, so that any errors in the grant overhead
         * calculations do not allow granting more space to clients than can be
         * written. Assumes that in aggregate the grant overhead calculations do
         * not have more than ddp_grant_reserved% estimation error in them. */
        m->ofd_grant_ratio =
                ofd_grant_ratio_conv(m->ofd_dt_conf.ddp_grant_reserved);

        rc = tgt_init(env, &m->ofd_lut, obd, m->ofd_osd, ofd_common_slice,
                      OBD_FAIL_OST_ALL_REQUEST_NET,
                      OBD_FAIL_OST_ALL_REPLY_NET);
        if (rc)
                GOTO(err_free_ns, rc);

        rc = ofd_fs_setup(env, m, obd);
        if (rc)
                GOTO(err_fini_lut, rc);

        rc = ofd_start_inconsistency_verification_thread(m);
        if (rc != 0)
                GOTO(err_fini_fs, rc);

        RETURN(0);

err_fini_fs:
        ofd_fs_cleanup(env, m);
err_fini_lut:
        tgt_fini(env, &m->ofd_lut);
err_free_ns:
        ldlm_namespace_free(m->ofd_namespace, 0, obd->obd_force);
        obd->obd_namespace = m->ofd_namespace = NULL;
err_fini_stack:
        ofd_stack_fini(env, m, &m->ofd_osd->dd_lu_dev);
err_fini_proc:
        ofd_procfs_fini(m);
        return rc;
}

/**
 * Stop the OFD device
 *
 * This function stops the OFD device and all its subsystems.
 * This is the end of OFD lifecycle.
 *
 * \param[in] env       execution environment
 * \param[in] m         OFD device
 */
static void ofd_fini(const struct lu_env *env, struct ofd_device *m)
{
        struct obd_device       *obd = ofd_obd(m);
        struct lu_device        *d   = &m->ofd_dt_dev.dd_lu_dev;
        struct lfsck_stop        stop;

        stop.ls_status = LS_PAUSED;
        stop.ls_flags = 0;
        lfsck_stop(env, m->ofd_osd, &stop);
        target_recovery_fini(obd);
        obd_exports_barrier(obd);
        obd_zombie_barrier();

        tgt_fini(env, &m->ofd_lut);
        ofd_stop_inconsistency_verification_thread(m);
        lfsck_degister(env, m->ofd_osd);
        ofd_fs_cleanup(env, m);

        ofd_free_capa_keys(m);
        cleanup_capa_hash(obd->u.filter.fo_capa_hash);

        if (m->ofd_namespace != NULL) {
                ldlm_namespace_free(m->ofd_namespace, NULL,
                                    d->ld_obd->obd_force);
                d->ld_obd->obd_namespace = m->ofd_namespace = NULL;
        }

        ofd_stack_fini(env, m, &m->ofd_dt_dev.dd_lu_dev);
        ofd_procfs_fini(m);
        LASSERT(atomic_read(&d->ld_ref) == 0);
        server_put_mount(obd->obd_name, true);
        EXIT;
}

/**
 * Implementation of lu_device_type_operations::ldto_device_fini.
 *
 * Finalize device. Dual to ofd_device_init(). It is called from
 * obd_precleanup() and stops the current device.
 *
 * \param[in] env       execution environment
 * \param[in] d         LU device of OFD
 *
 * \retval              NULL
 */
static struct lu_device *ofd_device_fini(const struct lu_env *env,
                                         struct lu_device *d)
{
        ENTRY;
        ofd_fini(env, ofd_dev(d));
        RETURN(NULL);
}

/**
 * Implementation of lu_device_type_operations::ldto_device_free.
 *
 * Free OFD device. Dual to ofd_device_alloc().
 *
 * \param[in] env       execution environment
 * \param[in] d         LU device of OFD
 *
 * \retval              NULL
 */
static struct lu_device *ofd_device_free(const struct lu_env *env,
                                         struct lu_device *d)
{
        struct ofd_device *m = ofd_dev(d);

        dt_device_fini(&m->ofd_dt_dev);
        OBD_FREE_PTR(m);
        RETURN(NULL);
}

/**
 * Implementation of lu_device_type_operations::ldto_device_alloc.
 *
 * This function allocates the new OFD device. It is called from
 * obd_setup() if OBD device had lu_device_type defined.
 *
 * \param[in] env       execution environment
 * \param[in] t         lu_device_type of OFD device
 * \param[in] cfg       configuration log
 *
 * \retval              pointer to the lu_device of just allocated OFD
 * \retval              ERR_PTR of return value on error
 */
static struct lu_device *ofd_device_alloc(const struct lu_env *env,
                                          struct lu_device_type *t,
                                          struct lustre_cfg *cfg)
{
        struct ofd_device *m;
        struct lu_device  *l;
        int                rc;

        OBD_ALLOC_PTR(m);
        if (m == NULL)
                return ERR_PTR(-ENOMEM);

        l = &m->ofd_dt_dev.dd_lu_dev;
        dt_device_init(&m->ofd_dt_dev, t);
        rc = ofd_init0(env, m, t, cfg);
        if (rc != 0) {
                ofd_device_free(env, l);
                l = ERR_PTR(rc);
        }

        return l;
}

/* type constructor/destructor: ofd_type_init(), ofd_type_fini() */
LU_TYPE_INIT_FINI(ofd, &ofd_thread_key);

static struct lu_device_type_operations ofd_device_type_ops = {
        .ldto_init              = ofd_type_init,
        .ldto_fini              = ofd_type_fini,

        .ldto_start             = ofd_type_start,
        .ldto_stop              = ofd_type_stop,

        .ldto_device_alloc      = ofd_device_alloc,
        .ldto_device_free       = ofd_device_free,
        .ldto_device_fini       = ofd_device_fini
};

static struct lu_device_type ofd_device_type = {
        .ldt_tags       = LU_DEVICE_DT,
        .ldt_name       = LUSTRE_OST_NAME,
        .ldt_ops        = &ofd_device_type_ops,
        .ldt_ctx_tags   = LCT_DT_THREAD
};

/**
 * Initialize OFD module.
 *
 * This function is called upon module loading. It registers OFD device type
 * and prepares all in-memory structures used by all OFD devices.
 *
 * \retval              0 if successful
 * \retval              negative value on error
 */
int __init ofd_init(void)
{
        int                             rc;

        rc = lu_kmem_init(ofd_caches);
        if (rc)
                return rc;

        rc = ofd_fmd_init();
        if (rc) {
                lu_kmem_fini(ofd_caches);
                return(rc);
        }

        rc = class_register_type(&ofd_obd_ops, NULL, true, NULL,
                                 LUSTRE_OST_NAME, &ofd_device_type);
        return rc;
}

/**
 * Stop OFD module.
 *
 * This function is called upon OFD module unloading.
 * It frees all related structures and unregisters OFD device type.
 */
void __exit ofd_exit(void)
{
        ofd_fmd_exit();
        lu_kmem_fini(ofd_caches);
        class_unregister_type(LUSTRE_OST_NAME);
}

MODULE_AUTHOR("Whamcloud, Inc. <http://www.whamcloud.com/>");
MODULE_DESCRIPTION("Lustre Object Filtering Device");
MODULE_LICENSE("GPL");

module_init(ofd_init);
module_exit(ofd_exit);