Whamcloud - gitweb
LU-14462 gss: fix support for namespace in lgss_keyring
[fs/lustre-release.git] / lustre / mdd / mdd_lproc.c
index 026f43e..e80a90c 100644 (file)
-/* -*- MODE: c; c-basic-offset: 8; indent-tabs-mode: nil; -*-
- * vim:expandtab:shiftwidth=8:tabstop=8:
+/*
+ * GPL HEADER START
  *
- *  mdd/mdd_lproc.c
- *  Lustre Metadata Server (mdd) routines
+ * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
  *
- *  Copyright (C) 2006 Cluster File Systems, Inc.
- *   Author: Wang Di <wangdi@clusterfs.com>
+ * 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 file is part of the Lustre file system, http://www.lustre.org
- *   Lustre is a trademark of Cluster File Systems, Inc.
+ * 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 may have signed or agreed to another license before downloading
- *   this software.  If so, you are bound by the terms and conditions
- *   of that agreement, and the following does not apply to you.  See the
- *   LICENSE file included with this distribution for more information.
+ * 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
  *
- *   If you did not agree to a different license, then this copy of Lustre
- *   is open source software; you can redistribute it and/or modify it
- *   under the terms of version 2 of the GNU General Public License as
- *   published by the Free Software Foundation.
+ * GPL HEADER END
+ */
+/*
+ * Copyright (c) 2007, 2010, Oracle and/or its affiliates. All rights reserved.
+ * Use is subject to license terms.
+ *
+ * Copyright (c) 2012, 2017, Intel Corporation.
+ */
+/*
+ * This file is part of Lustre, http://www.lustre.org/
+ * Lustre is a trademark of Sun Microsystems, Inc.
  *
- *   In either case, Lustre is distributed in the hope that it will be
- *   useful, but WITHOUT ANY WARRANTY; without even the implied warranty
- *   of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
- *   license text for more details.
+ * lustre/mdd/mdd_lproc.c
+ *
+ * Lustre Metadata Server (mdd) routines
+ *
+ * Author: Wang Di <wangdi@clusterfs.com>
  */
-#ifndef EXPORT_SYMTAB
-# define EXPORT_SYMTAB
-#endif
+
 #define DEBUG_SUBSYSTEM S_MDS
 
-#include <linux/module.h>
 #include <obd.h>
 #include <obd_class.h>
-#include <lustre_ver.h>
 #include <obd_support.h>
 #include <lprocfs_status.h>
-#include <lu_time.h>
+#include <libcfs/libcfs_string.h>
+#include "mdd_internal.h"
 
-#include <lustre/lustre_idl.h>
+static ssize_t uuid_show(struct kobject *kobj, struct attribute *attr,
+                        char *buf)
+{
+       struct mdd_device *mdd = container_of(kobj, struct mdd_device,
+                                             mdd_kobj);
+       struct obd_device *obd = mdd2obd_dev(mdd);
 
-#include "mdd_internal.h"
+       return sprintf(buf, "%s\n", obd->obd_uuid.uuid);
+}
+LUSTRE_RO_ATTR(uuid);
 
-static const char *mdd_counter_names[LPROC_MDD_NR] = {
-};
+static ssize_t atime_diff_show(struct kobject *kobj, struct attribute *attr,
+                              char *buf)
+{
+       struct mdd_device *mdd = container_of(kobj, struct mdd_device,
+                                             mdd_kobj);
 
-int mdd_procfs_init(struct mdd_device *mdd, const char *name)
+       return sprintf(buf, "%lld\n", mdd->mdd_atime_diff);
+}
+
+static ssize_t atime_diff_store(struct kobject *kobj,
+                               struct attribute *attr,
+                               const char *buffer, size_t count)
+{
+       struct mdd_device *mdd = container_of(kobj, struct mdd_device,
+                                             mdd_kobj);
+       time64_t diff = 0;
+       int rc;
+
+       rc = kstrtoll(buffer, 10, &diff);
+       if (rc)
+               return rc;
+
+        mdd->mdd_atime_diff = diff;
+        return count;
+}
+LUSTRE_RW_ATTR(atime_diff);
+
+/**** changelogs ****/
+static int mdd_changelog_mask_seq_show(struct seq_file *m, void *data)
+{
+       struct mdd_device *mdd = m->private;
+       int i = 0;
+
+       while (i < CL_LAST) {
+               if (mdd->mdd_cl.mc_mask & BIT(i))
+                       seq_printf(m, "%s ", changelog_type2str(i));
+               i++;
+       }
+       seq_putc(m, '\n');
+       return 0;
+}
+
+static ssize_t
+mdd_changelog_mask_seq_write(struct file *file, const char __user *buffer,
+                            size_t count, loff_t *off)
 {
-        struct lprocfs_static_vars lvars;
-        struct lu_device    *ld = &mdd->mdd_md_dev.md_lu_dev;
-        struct obd_type     *type;
-        int                  rc;
-        ENTRY;
-
-        type = ld->ld_type->ldt_obd_type;
-
-        LASSERT(name != NULL);
-        LASSERT(type != NULL);
-
-        /* Find the type procroot and add the proc entry for this device */
-        lprocfs_mdd_init_vars(&lvars);
-        mdd->mdd_proc_entry = lprocfs_register(name, type->typ_procroot,
-                                               lvars.obd_vars, mdd);
-        if (IS_ERR(mdd->mdd_proc_entry)) {
-                rc = PTR_ERR(mdd->mdd_proc_entry);
-                CERROR("Error %d setting up lprocfs for %s\n",
-                       rc, name);
-                mdd->mdd_proc_entry = NULL;
-                GOTO(out, rc);
-        }
-
-        rc = lu_time_init(&mdd->mdd_stats,
-                          mdd->mdd_proc_entry,
-                          mdd_counter_names, ARRAY_SIZE(mdd_counter_names));
-        EXIT;
+       struct seq_file *m = file->private_data;
+       struct mdd_device *mdd = m->private;
+       char *kernbuf;
+       int rc;
+       ENTRY;
+
+       if (count >= PAGE_SIZE)
+               RETURN(-EINVAL);
+       OBD_ALLOC(kernbuf, PAGE_SIZE);
+       if (kernbuf == NULL)
+               RETURN(-ENOMEM);
+       if (copy_from_user(kernbuf, buffer, count))
+               GOTO(out, rc = -EFAULT);
+       kernbuf[count] = 0;
+
+       rc = cfs_str2mask(kernbuf, changelog_type2str, &mdd->mdd_cl.mc_mask,
+                         CHANGELOG_MINMASK, CHANGELOG_ALLMASK);
+       if (rc == 0)
+               rc = count;
 out:
-        if (rc)
-               mdd_procfs_fini(mdd);
+       OBD_FREE(kernbuf, PAGE_SIZE);
        return rc;
 }
+LDEBUGFS_SEQ_FOPS(mdd_changelog_mask);
 
-int mdd_procfs_fini(struct mdd_device *mdd)
+static int lprocfs_changelog_users_cb(const struct lu_env *env,
+                                     struct llog_handle *llh,
+                                     struct llog_rec_hdr *hdr, void *data)
 {
-        if (mdd->mdd_stats)
-                lu_time_fini(&mdd->mdd_stats);
+       struct llog_changelog_user_rec *rec;
+       struct seq_file *m = data;
+
+       LASSERT(llh->lgh_hdr->llh_flags & LLOG_F_IS_PLAIN);
+
+       rec = (struct llog_changelog_user_rec *)hdr;
 
-        if (mdd->mdd_proc_entry) {
-                 lprocfs_remove(&mdd->mdd_proc_entry);
-                 mdd->mdd_proc_entry = NULL;
-        }
-        RETURN(0);
+       seq_printf(m, CHANGELOG_USER_PREFIX"%-3d %llu (%u)\n",
+                  rec->cur_id, rec->cur_endrec, (__u32)get_seconds() -
+                                                rec->cur_time);
+       return 0;
 }
 
-void mdd_lprocfs_time_start(const struct lu_env *env)
+static int mdd_changelog_users_seq_show(struct seq_file *m, void *data)
 {
-        lu_lprocfs_time_start(env);
+       struct lu_env            env;
+       struct mdd_device       *mdd = m->private;
+       struct llog_ctxt        *ctxt;
+       __u64                    cur;
+       int                      rc;
+
+        ctxt = llog_get_context(mdd2obd_dev(mdd),
+                               LLOG_CHANGELOG_USER_ORIG_CTXT);
+        if (ctxt == NULL)
+                return -ENXIO;
+        LASSERT(ctxt->loc_handle->lgh_hdr->llh_flags & LLOG_F_IS_CAT);
+
+       rc = lu_env_init(&env, LCT_LOCAL);
+       if (rc) {
+               llog_ctxt_put(ctxt);
+               return rc;
+       }
+
+       spin_lock(&mdd->mdd_cl.mc_lock);
+       cur = mdd->mdd_cl.mc_index;
+       spin_unlock(&mdd->mdd_cl.mc_lock);
+
+       seq_printf(m, "current index: %llu\n", cur);
+       seq_printf(m, "%-5s %s %s\n", "ID", "index", "(idle seconds)");
+
+       llog_cat_process(&env, ctxt->loc_handle, lprocfs_changelog_users_cb,
+                        m, 0, 0);
+
+       lu_env_fini(&env);
+       llog_ctxt_put(ctxt);
+       return 0;
 }
+LDEBUGFS_SEQ_FOPS_RO(mdd_changelog_users);
 
-void mdd_lprocfs_time_end(const struct lu_env *env, struct mdd_device *mdd,
-                          int idx)
+static int mdd_changelog_size_ctxt(const struct lu_env *env,
+                                  struct mdd_device *mdd,
+                                  int index, __u64 *val)
 {
-        lu_lprocfs_time_end(env, mdd->mdd_stats, idx);
+       struct llog_ctxt        *ctxt;
+
+       ctxt = llog_get_context(mdd2obd_dev(mdd),
+                               index);
+       if (ctxt == NULL)
+               return -ENXIO;
+
+       if (!(ctxt->loc_handle->lgh_hdr->llh_flags & LLOG_F_IS_CAT)) {
+               CERROR("%s: ChangeLog has wrong flags: rc = %d\n",
+                      ctxt->loc_obd->obd_name, -EINVAL);
+               llog_ctxt_put(ctxt);
+               return -EINVAL;
+       }
+
+       *val += llog_cat_size(env, ctxt->loc_handle);
+
+       llog_ctxt_put(ctxt);
+
+       return 0;
 }
 
-static int lprocfs_wr_atime_diff(struct file *file, const char *buffer,
-                                 unsigned long count, void *data)
+static ssize_t changelog_size_show(struct kobject *kobj,
+                                  struct attribute *attr,
+                                  char *buf)
 {
-        struct mdd_device *mdd = data;
-        char kernbuf[20], *end;
-        unsigned long diff = 0;
+       struct mdd_device *mdd = container_of(kobj, struct mdd_device,
+                                             mdd_kobj);
+       struct lu_env env;
+       u64 tmp = 0;
+       int rc;
+
+       rc = lu_env_init(&env, LCT_LOCAL);
+       if (rc)
+               return rc;
 
-        if (count > (sizeof(kernbuf) - 1))
-                return -EINVAL;
+       rc = mdd_changelog_size_ctxt(&env, mdd, LLOG_CHANGELOG_ORIG_CTXT, &tmp);
+       if (rc) {
+               lu_env_fini(&env);
+               return rc;
+       }
 
-        if (copy_from_user(kernbuf, buffer, count))
-                return -EFAULT;
+       rc = mdd_changelog_size_ctxt(&env, mdd, LLOG_CHANGELOG_USER_ORIG_CTXT,
+                                    &tmp);
 
-        kernbuf[count] = '\0';
+       rc = sprintf(buf, "%llu\n", tmp);
+       lu_env_fini(&env);
+       return rc;
+}
+LUSTRE_RO_ATTR(changelog_size);
 
-        diff = simple_strtoul(kernbuf, &end, 0);
-        if (kernbuf == end)
-                return -EINVAL;
+static ssize_t changelog_gc_show(struct kobject *kobj,
+                                struct attribute *attr,
+                                char *buf)
+{
+       struct mdd_device *mdd = container_of(kobj, struct mdd_device,
+                                             mdd_kobj);
 
-        mdd->mdd_atime_diff = diff;
-        return count;
+       return sprintf(buf, "%u\n", mdd->mdd_changelog_gc);
+}
+
+static ssize_t changelog_gc_store(struct kobject *kobj,
+                                 struct attribute *attr,
+                                 const char *buffer, size_t count)
+{
+       struct mdd_device *mdd = container_of(kobj, struct mdd_device,
+                                             mdd_kobj);
+       bool val;
+       int rc;
+
+       rc = kstrtobool(buffer, &val);
+       if (rc)
+               return rc;
+
+       mdd->mdd_changelog_gc = val;
+
+       return count;
+}
+LUSTRE_RW_ATTR(changelog_gc);
+
+static ssize_t changelog_max_idle_time_show(struct kobject *kobj,
+                                           struct attribute *attr,
+                                           char *buf)
+{
+       struct mdd_device *mdd = container_of(kobj, struct mdd_device,
+                                             mdd_kobj);
+
+       return sprintf(buf, "%lld\n", mdd->mdd_changelog_max_idle_time);
+}
+
+static ssize_t changelog_max_idle_time_store(struct kobject *kobj,
+                                            struct attribute *attr,
+                                            const char *buffer, size_t count)
+{
+       struct mdd_device *mdd = container_of(kobj, struct mdd_device,
+                                             mdd_kobj);
+       time64_t val;
+       int rc;
+
+       rc = kstrtoll(buffer, 10, &val);
+       if (rc)
+               return rc;
+
+       /* as it sounds reasonable, do not allow a user to be idle since
+        * more than about 68 years, this will allow to use 32bits
+        * timestamps for comparison
+        */
+       if (val < 1 || val > INT_MAX)
+               return -ERANGE;
+
+       mdd->mdd_changelog_max_idle_time = val;
+
+       return count;
+}
+LUSTRE_RW_ATTR(changelog_max_idle_time);
+
+static ssize_t changelog_max_idle_indexes_show(struct kobject *kobj,
+                                              struct attribute *attr,
+                                              char *buf)
+{
+       struct mdd_device *mdd = container_of(kobj, struct mdd_device,
+                                             mdd_kobj);
+
+       return sprintf(buf, "%lu\n", mdd->mdd_changelog_max_idle_indexes);
+}
+
+static ssize_t changelog_max_idle_indexes_store(struct kobject *kobj,
+                                               struct attribute *attr,
+                                               const char *buffer,
+                                               size_t count)
+{
+       struct mdd_device *mdd = container_of(kobj, struct mdd_device,
+                                             mdd_kobj);
+       unsigned long val;
+       int rc;
+
+       LASSERT(mdd != NULL);
+       rc = kstrtoul(buffer, 0, &val);
+       if (rc)
+               return rc;
+
+       /* XXX may need to limit/check with reasonable elapsed/idle indexes */
+       /* XXX may better allow to specify a % of full ChangeLogs */
+
+       mdd->mdd_changelog_max_idle_indexes = val;
+
+       return count;
+}
+LUSTRE_RW_ATTR(changelog_max_idle_indexes);
+
+static ssize_t changelog_min_gc_interval_show(struct kobject *kobj,
+                                             struct attribute *attr,
+                                             char *buf)
+{
+       struct mdd_device *mdd = container_of(kobj, struct mdd_device,
+                                             mdd_kobj);
+
+       return sprintf(buf, "%lld\n", mdd->mdd_changelog_min_gc_interval);
+}
+
+static ssize_t changelog_min_gc_interval_store(struct kobject *kobj,
+                                              struct attribute *attr,
+                                              const char *buffer,
+                                              size_t count)
+{
+       struct mdd_device *mdd = container_of(kobj, struct mdd_device,
+                                             mdd_kobj);
+       time64_t val;
+       int rc;
+
+       rc = kstrtoll(buffer, 10, &val);
+       if (rc)
+               return rc;
+
+       /* XXX may need to limit with reasonable elapsed/interval times */
+       if (val < 1)
+               return -ERANGE;
+
+       mdd->mdd_changelog_min_gc_interval = val;
+
+       return count;
+}
+LUSTRE_RW_ATTR(changelog_min_gc_interval);
+
+static ssize_t changelog_min_free_cat_entries_show(struct kobject *kobj,
+                                                  struct attribute *attr,
+                                                  char *buf)
+{
+       struct mdd_device *mdd = container_of(kobj, struct mdd_device,
+                                             mdd_kobj);
+
+       return sprintf(buf, "%u\n", mdd->mdd_changelog_min_free_cat_entries);
+}
+
+static ssize_t changelog_min_free_cat_entries_store(struct kobject *kobj,
+                                                   struct attribute *attr,
+                                                   const char *buffer,
+                                                   size_t count)
+{
+       struct mdd_device *mdd = container_of(kobj, struct mdd_device,
+                                             mdd_kobj);
+       unsigned int val;
+       int rc;
+
+       rc = kstrtouint(buffer, 10, &val);
+       if (rc)
+               return rc;
+
+       /* XXX may need to limit with more reasonable number of free entries */
+       if (val < 1)
+               return -ERANGE;
+
+       mdd->mdd_changelog_min_free_cat_entries = val;
+
+       return count;
+}
+LUSTRE_RW_ATTR(changelog_min_free_cat_entries);
+
+static ssize_t changelog_deniednext_show(struct kobject *kobj,
+                                        struct attribute *attr,
+                                        char *buf)
+{
+       struct mdd_device *mdd = container_of(kobj, struct mdd_device,
+                                             mdd_kobj);
+
+       return sprintf(buf, "%u\n", mdd->mdd_cl.mc_deniednext);
+}
+
+static ssize_t changelog_deniednext_store(struct kobject *kobj,
+                                         struct attribute *attr,
+                                         const char *buffer,
+                                         size_t count)
+{
+       struct mdd_device *mdd = container_of(kobj, struct mdd_device,
+                                             mdd_kobj);
+       unsigned int time = 0;
+       int rc;
+
+       rc = kstrtouint(buffer, 0, &time);
+       if (rc)
+               return rc;
+
+       mdd->mdd_cl.mc_deniednext = time;
+       return count;
+}
+LUSTRE_RW_ATTR(changelog_deniednext);
+
+static ssize_t sync_permission_show(struct kobject *kobj,
+                                   struct attribute *attr, char *buf)
+{
+       struct mdd_device *mdd = container_of(kobj, struct mdd_device,
+                                             mdd_kobj);
+
+       return sprintf(buf, "%d\n", mdd->mdd_sync_permission);
+}
+
+static ssize_t sync_permission_store(struct kobject *kobj,
+                                    struct attribute *attr,
+                                    const char *buffer, size_t count)
+{
+       struct mdd_device *mdd = container_of(kobj, struct mdd_device,
+                                             mdd_kobj);
+       bool val;
+       int rc;
+
+       rc = kstrtobool(buffer, &val);
+       if (rc)
+               return rc;
+
+       mdd->mdd_sync_permission = val;
+
+       return count;
+}
+LUSTRE_RW_ATTR(sync_permission);
+
+static ssize_t lfsck_speed_limit_show(struct kobject *kobj,
+                                     struct attribute *attr, char *buf)
+{
+       struct mdd_device *mdd = container_of(kobj, struct mdd_device,
+                                             mdd_kobj);
+
+       return lfsck_get_speed(buf, mdd->mdd_bottom);
+}
+
+static ssize_t lfsck_speed_limit_store(struct kobject *kobj,
+                                      struct attribute *attr,
+                                      const char *buffer, size_t count)
+{
+       struct mdd_device *mdd = container_of(kobj, struct mdd_device,
+                                             mdd_kobj);
+       unsigned int val;
+       int rc;
+
+       rc = kstrtouint(buffer, 10, &val);
+       if (rc != 0)
+               return rc;
+
+       rc = lfsck_set_speed(mdd->mdd_bottom, val);
+       return rc != 0 ? rc : count;
+}
+LUSTRE_RW_ATTR(lfsck_speed_limit);
+
+static ssize_t lfsck_async_windows_show(struct kobject *kobj,
+                                       struct attribute *attr, char *buf)
+{
+       struct mdd_device *mdd = container_of(kobj, struct mdd_device,
+                                             mdd_kobj);
+
+       return lfsck_get_windows(buf, mdd->mdd_bottom);
+}
+
+static ssize_t lfsck_async_windows_store(struct kobject *kobj,
+                                        struct attribute *attr,
+                                        const char *buffer, size_t count)
+{
+       struct mdd_device *mdd = container_of(kobj, struct mdd_device,
+                                             mdd_kobj);
+       unsigned int val;
+       int rc;
+
+       rc = kstrtouint(buffer, 10, &val);
+       if (rc)
+               return rc;
+
+       rc = lfsck_set_windows(mdd->mdd_bottom, val);
+
+       return rc != 0 ? rc : count;
+}
+LUSTRE_RW_ATTR(lfsck_async_windows);
+
+static int mdd_lfsck_namespace_seq_show(struct seq_file *m, void *data)
+{
+       struct mdd_device *mdd = m->private;
+
+       LASSERT(mdd != NULL);
+
+       return lfsck_dump(m, mdd->mdd_bottom, LFSCK_TYPE_NAMESPACE);
+}
+LDEBUGFS_SEQ_FOPS_RO(mdd_lfsck_namespace);
+
+static int mdd_lfsck_layout_seq_show(struct seq_file *m, void *data)
+{
+       struct mdd_device *mdd = m->private;
+
+       LASSERT(mdd != NULL);
+
+       return lfsck_dump(m, mdd->mdd_bottom, LFSCK_TYPE_LAYOUT);
+}
+LDEBUGFS_SEQ_FOPS_RO(mdd_lfsck_layout);
+
+/**
+ * Show default number of stripes for O_APPEND files.
+ *
+ * \param[in] m                seq file
+ * \param[in] v                unused for single entry
+ *
+ * \retval 0           on success,
+ * \retval negative    error code if failed
+ */
+static ssize_t append_stripe_count_show(struct kobject *kobj,
+                                       struct attribute *attr, char *buf)
+{
+       struct mdd_device *mdd = container_of(kobj, struct mdd_device,
+                                             mdd_kobj);
+
+       return scnprintf(buf, PAGE_SIZE, "%d\n", mdd->mdd_append_stripe_count);
+}
+
+/**
+ * Set default number of stripes for O_APPEND files.
+ *
+ * \param[in] file     proc file
+ * \param[in] buffer   string containing the default number of stripes
+ *                     for new files
+ * \param[in] count    @buffer length
+ * \param[in] off      unused for single entry
+ *
+ * \retval @count      on success
+ * \retval negative    error code otherwise
+ */
+static ssize_t append_stripe_count_store(struct kobject *kobj,
+                                        struct attribute *attr,
+                                        const char *buffer, size_t count)
+{
+       struct mdd_device *mdd = container_of(kobj, struct mdd_device,
+                                             mdd_kobj);
+       int stripe_count;
+       int rc;
+
+       rc = kstrtoint(buffer, 0, &stripe_count);
+       if (rc)
+               return rc;
+
+       if (stripe_count < -1)
+               return -ERANGE;
+
+       mdd->mdd_append_stripe_count = stripe_count;
+
+       return count;
+}
+LUSTRE_RW_ATTR(append_stripe_count);
+
+/**
+ * Show default OST pool for O_APPEND files.
+ *
+ * \param[in] kobject  proc object
+ * \param[in] attribute proc attribute
+ * \param[in] buf      output buffer
+ *
+ * \retval 0           on success,
+ * \retval negative    error code if failed
+ */
+static ssize_t append_pool_show(struct kobject *kobj,
+                               struct attribute *attr, char *buf)
+{
+       struct mdd_device *mdd = container_of(kobj, struct mdd_device,
+                                             mdd_kobj);
+
+       return scnprintf(buf, PAGE_SIZE, "%s\n", mdd->mdd_append_pool);
 }
 
-static int lprocfs_rd_atime_diff(char *page, char **start, off_t off,
-                                 int count, int *eof, void *data)
+/**
+ * Set default OST pool for O_APPEND files.
+ *
+ * \param[in] kobject  proc object
+ * \param[in] attribute proc attribute
+ * \param[in] buffer   user inputted pool name
+ * \param[in] count    @buffer length
+ *
+ * \retval @count      on success
+ * \retval negative    error code otherwise
+ */
+static ssize_t append_pool_store(struct kobject *kobj, struct attribute *attr,
+                                const char *buffer, size_t count)
 {
-        struct mdd_device *mdd = data;
+       struct mdd_device *mdd = container_of(kobj, struct mdd_device,
+                                             mdd_kobj);
+
+       if (!count || count > LOV_MAXPOOLNAME + 1)
+               return -EINVAL;
+
+       /* clear previous value */
+       memset(mdd->mdd_append_pool, 0, LOV_MAXPOOLNAME + 1);
 
-        *eof = 1;
-        return snprintf(page, count, "%lu\n", mdd->mdd_atime_diff);
+       /* entering "none" clears the pool, otherwise copy the new pool */
+       if (strncmp("none", buffer, 4)) {
+               memcpy(mdd->mdd_append_pool, buffer, count);
+
+               /* Trim the trailing '\n' if any */
+               if (mdd->mdd_append_pool[count - 1] == '\n') {
+                       /* Don't echo just a newline */
+                       if (count == 1)
+                               return -EINVAL;
+                       mdd->mdd_append_pool[count - 1] = 0;
+               }
+       }
+
+       return count;
 }
+LUSTRE_RW_ATTR(append_pool);
 
-static struct lprocfs_vars lprocfs_mdd_obd_vars[] = {
-        { "atime_diff", lprocfs_rd_atime_diff, lprocfs_wr_atime_diff, 0 },
-        { 0 }
+static struct ldebugfs_vars ldebugfs_mdd_obd_vars[] = {
+       { .name =       "changelog_mask",
+         .fops =       &mdd_changelog_mask_fops        },
+       { .name =       "changelog_users",
+         .fops =       &mdd_changelog_users_fops       },
+       { .name =       "lfsck_namespace",
+         .fops =       &mdd_lfsck_namespace_fops       },
+       { .name =       "lfsck_layout",
+         .fops =       &mdd_lfsck_layout_fops          },
+       { NULL }
 };
 
-static struct lprocfs_vars lprocfs_mdd_module_vars[] = {
-        { "num_refs",   lprocfs_rd_numrefs, 0, 0 },
-        { 0 }
+static struct attribute *mdd_attrs[] = {
+       &lustre_attr_uuid.attr,
+       &lustre_attr_atime_diff.attr,
+       &lustre_attr_changelog_size.attr,
+       &lustre_attr_changelog_gc.attr,
+       &lustre_attr_changelog_max_idle_time.attr,
+       &lustre_attr_changelog_max_idle_indexes.attr,
+       &lustre_attr_changelog_min_gc_interval.attr,
+       &lustre_attr_changelog_min_free_cat_entries.attr,
+       &lustre_attr_changelog_deniednext.attr,
+       &lustre_attr_lfsck_async_windows.attr,
+       &lustre_attr_lfsck_speed_limit.attr,
+       &lustre_attr_sync_permission.attr,
+       &lustre_attr_append_stripe_count.attr,
+       &lustre_attr_append_pool.attr,
+       NULL,
 };
 
-void lprocfs_mdd_init_vars(struct lprocfs_static_vars *lvars)
+static void mdd_sysfs_release(struct kobject *kobj)
+{
+       struct mdd_device *mdd = container_of(kobj, struct mdd_device,
+                                             mdd_kobj);
+       struct obd_device *obd = mdd2obd_dev(mdd);
+
+       debugfs_remove_recursive(obd->obd_debugfs_entry);
+       obd->obd_debugfs_entry = NULL;
+
+       complete(&mdd->mdd_kobj_unregister);
+}
+
+int mdd_procfs_init(struct mdd_device *mdd, const char *name)
+{
+       struct obd_device *obd = mdd2obd_dev(mdd);
+       struct obd_type *type;
+       int rc;
+
+       ENTRY;
+       /* at the moment there is no linkage between lu_type
+        * and obd_type, so we lookup obd_type this way
+        */
+       type = class_search_type(LUSTRE_MDD_NAME);
+
+       LASSERT(name != NULL);
+       LASSERT(type != NULL);
+       LASSERT(obd  != NULL);
+
+       /* put reference taken by class_search_type */
+       kobject_put(&type->typ_kobj);
+
+       mdd->mdd_ktype.default_attrs = mdd_attrs;
+       mdd->mdd_ktype.release = mdd_sysfs_release;
+       mdd->mdd_ktype.sysfs_ops = &lustre_sysfs_ops;
+
+       init_completion(&mdd->mdd_kobj_unregister);
+       rc = kobject_init_and_add(&mdd->mdd_kobj, &mdd->mdd_ktype,
+                                 &type->typ_kobj, "%s", name);
+       if (rc)
+               return rc;
+
+       /* Find the type procroot and add the proc entry for this device */
+       obd->obd_debugfs_vars = ldebugfs_mdd_obd_vars;
+       obd->obd_debugfs_entry = debugfs_create_dir(name,
+                                                    type->typ_debugfs_entry);
+       ldebugfs_add_vars(obd->obd_debugfs_entry, obd->obd_debugfs_vars, mdd);
+
+       RETURN(rc);
+}
+
+void mdd_procfs_fini(struct mdd_device *mdd)
 {
-        lvars->module_vars  = lprocfs_mdd_module_vars;
-        lvars->obd_vars     = lprocfs_mdd_obd_vars;
+       kobject_put(&mdd->mdd_kobj);
+       wait_for_completion(&mdd->mdd_kobj_unregister);
 }