Whamcloud - gitweb
LU-1194 llog: fix for not sync llcd at thread stop
[fs/lustre-release.git] / lustre / obdclass / llog_cat.c
index 4c49a10..f0fa468 100644 (file)
@@ -1,46 +1,56 @@
-/* -*- mode: c; c-basic-offset: 8; indent-tabs-mode: nil; -*-
- * vim:expandtab:shiftwidth=8:tabstop=8:
+/*
+ * GPL HEADER START
  *
- *  Copyright (C) 2001-2003 Cluster File Systems, Inc.
- *   Author: Andreas Dilger <adilger@clusterfs.com>
+ * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
  *
- *   This file is part of Lustre, http://www.lustre.org.
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 only,
+ * as published by the Free Software Foundation.
  *
- *   Lustre is free software; you can redistribute it and/or
- *   modify it under the terms of version 2 of the GNU General Public
- *   License as published by the Free Software Foundation.
+ * This program is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * General Public License version 2 for more details (a copy is included
+ * in the LICENSE file that accompanied this code).
  *
- *   Lustre is distributed in the hope that it will be useful,
- *   but WITHOUT ANY WARRANTY; without even the implied warranty of
- *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
- *   GNU General Public License for more details.
+ * You should have received a copy of the GNU General Public License
+ * version 2 along with this program; If not, see
+ * http://www.sun.com/software/products/lustre/docs/GPLv2.pdf
  *
- *   You should have received a copy of the GNU General Public License
- *   along with Lustre; if not, write to the Free Software
- *   Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ * Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara,
+ * CA 95054 USA or visit www.sun.com if you need additional information or
+ * have any questions.
+ *
+ * GPL HEADER END
+ */
+/*
+ * Copyright (c) 2003, 2010, Oracle and/or its affiliates. All rights reserved.
+ * Use is subject to license terms.
+ */
+/*
+ * This file is part of Lustre, http://www.lustre.org/
+ * Lustre is a trademark of Sun Microsystems, Inc.
+ *
+ * lustre/obdclass/llog_cat.c
  *
  * OST<->MDS recovery logging infrastructure.
  *
  * Invariants in implementation:
  * - we do not share logs among different OST<->MDS connections, so that
  *   if an OST or MDS fails it need only look at log(s) relevant to itself
+ *
+ * Author: Andreas Dilger <adilger@clusterfs.com>
  */
 
 #define DEBUG_SUBSYSTEM S_LOG
 
-#ifndef EXPORT_SYMTAB
-#define EXPORT_SYMTAB
-#endif
-
-#ifdef __KERNEL__
-#include <linux/fs.h>
-#else
+#ifndef __KERNEL__
 #include <liblustre.h>
 #endif
 
-#include <linux/obd_class.h>
-#include <linux/lustre_log.h>
-#include <portals/list.h>
+#include <obd_class.h>
+#include <lustre_log.h>
+#include <libcfs/list.h>
 
 /* Create a new log handle and add it to the open list.
  * This log handle will be closed when all of the records in it are removed.
@@ -51,71 +61,80 @@ static struct llog_handle *llog_cat_new_log(struct llog_handle *cathandle)
 {
         struct llog_handle *loghandle;
         struct llog_log_hdr *llh;
-        struct llog_logid_rec rec;
-        int rc, index, bitmap_size, i;
+        struct llog_logid_rec rec = { { 0 }, };
+        int rc, index, bitmap_size;
         ENTRY;
 
+        llh = cathandle->lgh_hdr;
+        bitmap_size = LLOG_BITMAP_SIZE(llh);
+
+        index = (cathandle->lgh_last_idx + 1) % bitmap_size;
+
+        /* maximum number of available slots in catlog is bitmap_size - 2 */
+        if (llh->llh_cat_idx == index) {
+                CERROR("no free catalog slots for log...\n");
+                RETURN(ERR_PTR(-ENOSPC));
+        }
+
+        if (OBD_FAIL_CHECK(OBD_FAIL_MDS_LLOG_CREATE_FAILED))
+                RETURN(ERR_PTR(-ENOSPC));
+
         rc = llog_create(cathandle->lgh_ctxt, &loghandle, NULL, NULL);
         if (rc)
                 RETURN(ERR_PTR(rc));
 
-        rc = llog_init_handle(loghandle, 
-                              LLOG_F_IS_PLAIN | LLOG_F_ZAP_WHEN_EMPTY, 
+        rc = llog_init_handle(loghandle,
+                              LLOG_F_IS_PLAIN | LLOG_F_ZAP_WHEN_EMPTY,
                               &cathandle->lgh_hdr->llh_tgtuuid);
         if (rc)
                 GOTO(out_destroy, rc);
 
-        /* Find first free entry */
-        llh = cathandle->lgh_hdr;
-        bitmap_size = sizeof(llh->llh_bitmap) * 8;
-        for (i = 0, index = le32_to_cpu(llh->llh_count); i < bitmap_size; 
-             i++, index++) {
-                index %= bitmap_size;
-                if (ext2_set_bit(index, llh->llh_bitmap)) {
-                        /* XXX This should trigger log clean up or similar */
-                        CERROR("catalog index %d is still in use\n", index);
-                } else {
-                        cathandle->lgh_last_idx = index;
-                        llh->llh_count = cpu_to_le32(le32_to_cpu(llh->llh_count) + 1);
-                        break;
-                }
-        }
-        if (i == bitmap_size) {
-                CERROR("no free catalog slots for log...\n");
-                GOTO(out_destroy, rc = -ENOSPC);
+        if (index == 0)
+                index = 1;
+        if (ext2_set_bit(index, llh->llh_bitmap)) {
+                CERROR("argh, index %u already set in log bitmap?\n",
+                       index);
+                LBUG(); /* should never happen */
         }
-        CWARN("new recovery log "LPX64":%x for index %u of catalog "LPX64"\n",
-               loghandle->lgh_id.lgl_oid, loghandle->lgh_id.lgl_ogen, index,
-               cathandle->lgh_id.lgl_oid);
+        cathandle->lgh_last_idx = index;
+        llh->llh_count++;
+        llh->llh_tail.lrt_index = index;
+
+        CDEBUG(D_RPCTRACE,"new recovery log "LPX64":%x for index %u of catalog "
+               LPX64"\n", loghandle->lgh_id.lgl_oid, loghandle->lgh_id.lgl_ogen,
+               index, cathandle->lgh_id.lgl_oid);
         /* build the record for this log in the catalog */
-        rec.lid_hdr.lrh_len = cpu_to_le32(sizeof(rec));
-        rec.lid_hdr.lrh_index = cpu_to_le32(index);
-        rec.lid_hdr.lrh_type = cpu_to_le32(LLOG_LOGID_MAGIC);
+        rec.lid_hdr.lrh_len = sizeof(rec);
+        rec.lid_hdr.lrh_index = index;
+        rec.lid_hdr.lrh_type = LLOG_LOGID_MAGIC;
         rec.lid_id = loghandle->lgh_id;
-        rec.lid_tail.lrt_len = cpu_to_le32(sizeof(rec));
-        rec.lid_tail.lrt_index = cpu_to_le32(index);
+        rec.lid_tail.lrt_len = sizeof(rec);
+        rec.lid_tail.lrt_index = index;
 
         /* update the catalog: header and record */
-        rc = llog_write_rec(cathandle, &rec.lid_hdr, 
+        rc = llog_write_rec(cathandle, &rec.lid_hdr,
                             &loghandle->u.phd.phd_cookie, 1, NULL, index);
         if (rc < 0) {
                 GOTO(out_destroy, rc);
         }
 
-        loghandle->lgh_hdr->llh_cat_idx = cpu_to_le32(index);
+        loghandle->lgh_hdr->llh_cat_idx = index;
         cathandle->u.chd.chd_current_log = loghandle;
-        LASSERT(list_empty(&loghandle->u.phd.phd_entry));
-        list_add_tail(&loghandle->u.phd.phd_entry, &cathandle->u.chd.chd_head);
+        LASSERT(cfs_list_empty(&loghandle->u.phd.phd_entry));
+        cfs_list_add_tail(&loghandle->u.phd.phd_entry,
+                          &cathandle->u.chd.chd_head);
 
- out_destroy:
+out_destroy:
         if (rc < 0)
                 llog_destroy(loghandle);
 
         RETURN(loghandle);
 }
-EXPORT_SYMBOL(llog_cat_new_log);
 
-/* Assumes caller has already pushed us into the kernel context and is locking.
+/* Open an existent log handle and add it to the open list.
+ * This log handle will be closed when all of the records in it are removed.
+ *
+ * Assumes caller has already pushed us into the kernel context and is locking.
  * We return a lock on the handle to ensure nobody yanks it from us.
  */
 int llog_cat_id2handle(struct llog_handle *cathandle, struct llog_handle **res,
@@ -128,8 +147,8 @@ int llog_cat_id2handle(struct llog_handle *cathandle, struct llog_handle **res,
         if (cathandle == NULL)
                 RETURN(-EBADF);
 
-        list_for_each_entry(loghandle, &cathandle->u.chd.chd_head, 
-                            u.phd.phd_entry) {
+        cfs_list_for_each_entry(loghandle, &cathandle->u.chd.chd_head,
+                                u.phd.phd_entry) {
                 struct llog_logid *cgl = &loghandle->lgh_id;
                 if (cgl->lgl_oid == logid->lgl_oid) {
                         if (cgl->lgl_ogen != logid->lgl_ogen) {
@@ -139,7 +158,6 @@ int llog_cat_id2handle(struct llog_handle *cathandle, struct llog_handle **res,
                                 continue;
                         }
                         loghandle->u.phd.phd_cat_handle = cathandle;
-                        cathandle->u.chd.chd_current_log = loghandle;
                         GOTO(out, rc = 0);
                 }
         }
@@ -151,16 +169,15 @@ int llog_cat_id2handle(struct llog_handle *cathandle, struct llog_handle **res,
         } else {
                 rc = llog_init_handle(loghandle, LLOG_F_IS_PLAIN, NULL);
                 if (!rc) {
-                        list_add(&loghandle->u.phd.phd_entry, 
-                                 &cathandle->u.chd.chd_head);
-                        cathandle->u.chd.chd_current_log = loghandle;
+                        cfs_list_add(&loghandle->u.phd.phd_entry,
+                                     &cathandle->u.chd.chd_head);
                 }
         }
         if (!rc) {
                 loghandle->u.phd.phd_cat_handle = cathandle;
                 loghandle->u.phd.phd_cookie.lgc_lgl = cathandle->lgh_id;
-                loghandle->u.phd.phd_cookie.lgc_index = 
-                        le32_to_cpu(loghandle->lgh_hdr->llh_cat_idx);
+                loghandle->u.phd.phd_cookie.lgc_index =
+                        loghandle->lgh_hdr->llh_cat_idx;
         }
 
 out:
@@ -174,8 +191,8 @@ int llog_cat_put(struct llog_handle *cathandle)
         int rc;
         ENTRY;
 
-        list_for_each_entry_safe(loghandle, n, &cathandle->u.chd.chd_head, 
-                                 u.phd.phd_entry) {
+        cfs_list_for_each_entry_safe(loghandle, n, &cathandle->u.chd.chd_head,
+                                     u.phd.phd_entry) {
                 int err = llog_close(loghandle);
                 if (err)
                         CERROR("error closing loghandle\n");
@@ -185,7 +202,15 @@ int llog_cat_put(struct llog_handle *cathandle)
 }
 EXPORT_SYMBOL(llog_cat_put);
 
-/* Return the currently active log handle.  If the current log handle doesn't
+/**
+ * lockdep markers for nested struct llog_handle::lgh_lock locking.
+ */
+enum {
+        LLOGH_CAT,
+        LLOGH_LOG
+};
+
+/** Return the currently active log handle.  If the current log handle doesn't
  * have enough space left for the current record, start a new one.
  *
  * If reclen is 0, we only want to know what the currently active log is,
@@ -195,49 +220,53 @@ EXPORT_SYMBOL(llog_cat_put);
  *
  * NOTE: loghandle is write-locked upon successful return
  */
-static struct llog_handle *llog_cat_current_log(struct llog_handle *cathandle, 
+static struct llog_handle *llog_cat_current_log(struct llog_handle *cathandle,
                                                 int create)
 {
         struct llog_handle *loghandle = NULL;
         ENTRY;
 
-        down_read(&cathandle->lgh_lock);
+        cfs_down_read_nested(&cathandle->lgh_lock, LLOGH_CAT);
         loghandle = cathandle->u.chd.chd_current_log;
         if (loghandle) {
                 struct llog_log_hdr *llh = loghandle->lgh_hdr;
-                if (loghandle->lgh_last_idx < (sizeof(llh->llh_bitmap) * 8) - 1) {
-                        down_write(&loghandle->lgh_lock);
-                        up_read(&cathandle->lgh_lock);
+                cfs_down_write_nested(&loghandle->lgh_lock, LLOGH_LOG);
+                if (loghandle->lgh_last_idx < LLOG_BITMAP_SIZE(llh) - 1) {
+                        cfs_up_read(&cathandle->lgh_lock);
                         RETURN(loghandle);
+                } else {
+                        cfs_up_write(&loghandle->lgh_lock);
                 }
         }
         if (!create) {
                 if (loghandle)
-                        down_write(&loghandle->lgh_lock);
-                up_read(&cathandle->lgh_lock);
+                        cfs_down_write(&loghandle->lgh_lock);
+                cfs_up_read(&cathandle->lgh_lock);
                 RETURN(loghandle);
         }
-        up_read(&cathandle->lgh_lock);
+        cfs_up_read(&cathandle->lgh_lock);
 
         /* time to create new log */
 
         /* first, we have to make sure the state hasn't changed */
-        down_write(&cathandle->lgh_lock);
+        cfs_down_write_nested(&cathandle->lgh_lock, LLOGH_CAT);
         loghandle = cathandle->u.chd.chd_current_log;
         if (loghandle) {
                 struct llog_log_hdr *llh = loghandle->lgh_hdr;
-                if (loghandle->lgh_last_idx < (sizeof(llh->llh_bitmap) * 8) - 1) {
-                        down_write(&loghandle->lgh_lock);
-                        up_write(&cathandle->lgh_lock);
+                cfs_down_write_nested(&loghandle->lgh_lock, LLOGH_LOG);
+                if (loghandle->lgh_last_idx < LLOG_BITMAP_SIZE(llh) - 1) {
+                        cfs_up_write(&cathandle->lgh_lock);
                         RETURN(loghandle);
+                } else {
+                        cfs_up_write(&loghandle->lgh_lock);
                 }
         }
 
         CDEBUG(D_INODE, "creating new log\n");
         loghandle = llog_cat_new_log(cathandle);
-        if (loghandle)
-                down_write(&loghandle->lgh_lock);
-        up_write(&cathandle->lgh_lock);
+        if (!IS_ERR(loghandle))
+                cfs_down_write_nested(&loghandle->lgh_lock, LLOGH_LOG);
+        cfs_up_write(&cathandle->lgh_lock);
         RETURN(loghandle);
 }
 
@@ -247,19 +276,30 @@ static struct llog_handle *llog_cat_current_log(struct llog_handle *cathandle,
  * Assumes caller has already pushed us into the kernel context.
  */
 int llog_cat_add_rec(struct llog_handle *cathandle, struct llog_rec_hdr *rec,
-                    struct llog_cookie *reccookie, void *buf)
+                     struct llog_cookie *reccookie, void *buf)
 {
         struct llog_handle *loghandle;
         int rc;
         ENTRY;
 
-        LASSERT(le32_to_cpu(rec->lrh_len) <= LLOG_CHUNK_SIZE);
+        LASSERT(rec->lrh_len <= LLOG_CHUNK_SIZE);
         loghandle = llog_cat_current_log(cathandle, 1);
         if (IS_ERR(loghandle))
                 RETURN(PTR_ERR(loghandle));
         /* loghandle is already locked by llog_cat_current_log() for us */
         rc = llog_write_rec(loghandle, rec, reccookie, 1, buf, -1);
-        up_write(&loghandle->lgh_lock);
+        if (rc < 0)
+                CERROR("llog_write_rec %d: lh=%p\n", rc, loghandle);
+        cfs_up_write(&loghandle->lgh_lock);
+        if (rc == -ENOSPC) {
+                /* to create a new plain log */
+                loghandle = llog_cat_current_log(cathandle, 1);
+                if (IS_ERR(loghandle))
+                        RETURN(PTR_ERR(loghandle));
+                rc = llog_write_rec(loghandle, rec, reccookie, 1, buf, -1);
+                cfs_up_write(&loghandle->lgh_lock);
+        }
+
         RETURN(rc);
 }
 EXPORT_SYMBOL(llog_cat_add_rec);
@@ -274,12 +314,12 @@ EXPORT_SYMBOL(llog_cat_add_rec);
  * Assumes caller has already pushed us into the kernel context.
  */
 int llog_cat_cancel_records(struct llog_handle *cathandle, int count,
-                        struct llog_cookie *cookies)
+                            struct llog_cookie *cookies)
 {
         int i, index, rc = 0;
         ENTRY;
 
-        down_write(&cathandle->lgh_lock);
+        cfs_down_write_nested(&cathandle->lgh_lock, LLOGH_CAT);
         for (i = 0; i < count; i++, cookies++) {
                 struct llog_handle *loghandle;
                 struct llog_logid *lgl = &cookies->lgc_lgl;
@@ -290,107 +330,267 @@ int llog_cat_cancel_records(struct llog_handle *cathandle, int count,
                         break;
                 }
 
-                down_write(&loghandle->lgh_lock);
+                cfs_down_write_nested(&loghandle->lgh_lock, LLOGH_LOG);
                 rc = llog_cancel_rec(loghandle, cookies->lgc_index);
-                up_write(&loghandle->lgh_lock);
-                
+                cfs_up_write(&loghandle->lgh_lock);
+
                 if (rc == 1) {          /* log has been destroyed */
                         index = loghandle->u.phd.phd_cookie.lgc_index;
                         if (cathandle->u.chd.chd_current_log == loghandle)
                                 cathandle->u.chd.chd_current_log = NULL;
                         llog_free_handle(loghandle);
-                        
+
                         LASSERT(index);
+                        llog_cat_set_first_idx(cathandle, index);
                         rc = llog_cancel_rec(cathandle, index);
+                        if (rc == 0)
+                                CDEBUG(D_RPCTRACE,"cancel plain log at index %u"
+                                       " of catalog "LPX64"\n",
+                                       index, cathandle->lgh_id.lgl_oid);
                 }
         }
-        up_write(&cathandle->lgh_lock);
+        cfs_up_write(&cathandle->lgh_lock);
 
         RETURN(rc);
 }
 EXPORT_SYMBOL(llog_cat_cancel_records);
 
-int llog_cat_process_cb(struct llog_handle *cat_llh, struct llog_rec_hdr *rec, void *data)
+int llog_cat_process_cb(struct llog_handle *cat_llh, struct llog_rec_hdr *rec,
+                        void *data)
 {
         struct llog_process_data *d = data;
         struct llog_logid_rec *lir = (struct llog_logid_rec *)rec;
         struct llog_handle *llh;
         int rc;
 
-        if (le32_to_cpu(rec->lrh_type) != LLOG_LOGID_MAGIC) {
+        ENTRY;
+        if (rec->lrh_type != LLOG_LOGID_MAGIC) {
                 CERROR("invalid record in catalog\n");
                 RETURN(-EINVAL);
         }
-        CWARN("processing log "LPX64":%x at index %u of catalog "LPX64"\n", 
-               lir->lid_id.lgl_oid, lir->lid_id.lgl_ogen,
-               le32_to_cpu(rec->lrh_index), cat_llh->lgh_id.lgl_oid);
+        CDEBUG(D_HA, "processing log "LPX64":%x at index %u of catalog "
+               LPX64"\n", lir->lid_id.lgl_oid, lir->lid_id.lgl_ogen,
+               rec->lrh_index, cat_llh->lgh_id.lgl_oid);
 
         rc = llog_cat_id2handle(cat_llh, &llh, &lir->lid_id);
         if (rc) {
-                CERROR("Cannot find handle for log "LPX64"\n", lir->lid_id.lgl_oid);
+                CERROR("Cannot find handle for log "LPX64"\n",
+                       lir->lid_id.lgl_oid);
                 RETURN(rc);
-        }        
+        }
+
+        if (rec->lrh_index < d->lpd_startcat)
+                /* Skip processing of the logs until startcat */
+                RETURN(0);
+
+        if (d->lpd_startidx > 0) {
+                struct llog_process_cat_data cd;
+
+                cd.lpcd_first_idx = d->lpd_startidx;
+                cd.lpcd_last_idx = 0;
+                rc = llog_process_flags(llh, d->lpd_cb, d->lpd_data, &cd,
+                                        d->lpd_flags);
+                /* Continue processing the next log from idx 0 */
+                d->lpd_startidx = 0;
+        } else {
+                rc = llog_process_flags(llh, d->lpd_cb, d->lpd_data, NULL,
+                                        d->lpd_flags);
+        }
 
-        rc = llog_process(llh, d->lpd_cb, d->lpd_data);
         RETURN(rc);
 }
 
-int llog_cat_process(struct llog_handle *cat_llh, llog_cb_t cb, void *data)
+int llog_cat_process_flags(struct llog_handle *cat_llh, llog_cb_t cb,
+                           void *data, int flags, int startcat, int startidx)
 {
         struct llog_process_data d;
+        struct llog_log_hdr *llh = cat_llh->lgh_hdr;
         int rc;
         ENTRY;
+
+        LASSERT(llh->llh_flags & LLOG_F_IS_CAT);
         d.lpd_data = data;
         d.lpd_cb = cb;
+        d.lpd_startcat = startcat;
+        d.lpd_startidx = startidx;
+        d.lpd_flags = flags;
+
+        if (llh->llh_cat_idx > cat_llh->lgh_last_idx) {
+                struct llog_process_cat_data cd;
+
+                CWARN("catlog "LPX64" crosses index zero\n",
+                      cat_llh->lgh_id.lgl_oid);
+
+                cd.lpcd_first_idx = llh->llh_cat_idx;
+                cd.lpcd_last_idx = 0;
+                rc = llog_process_flags(cat_llh, llog_cat_process_cb, &d, &cd,
+                                        flags);
+                if (rc != 0)
+                        RETURN(rc);
+
+                cd.lpcd_first_idx = 0;
+                cd.lpcd_last_idx = cat_llh->lgh_last_idx;
+                rc = llog_process_flags(cat_llh, llog_cat_process_cb, &d, &cd,
+                                        flags);
+        } else {
+                rc = llog_process_flags(cat_llh, llog_cat_process_cb, &d, NULL,
+                                        flags);
+        }
 
-        rc = llog_process(cat_llh, llog_cat_process_cb, &d);
         RETURN(rc);
 }
-EXPORT_SYMBOL(llog_cat_process);
+EXPORT_SYMBOL(llog_cat_process_flags);
 
+int llog_cat_process(struct llog_handle *cat_llh, llog_cb_t cb, void *data,
+                     int startcat, int startidx)
+{
+        return llog_cat_process_flags(cat_llh, cb, data, 0, startcat, startidx);
+}
+EXPORT_SYMBOL(llog_cat_process);
 
-#if 0
-/* Assumes caller has already pushed us into the kernel context. */
-int llog_cat_init(struct llog_handle *cathandle, struct obd_uuid *tgtuuid)
+#ifdef __KERNEL__
+int llog_cat_process_thread(void *data)
 {
-        struct llog_log_hdr *llh;
-        loff_t offset = 0;
-        int rc = 0;
+        struct llog_process_cat_args *args = data;
+        struct llog_ctxt *ctxt = args->lpca_ctxt;
+        struct llog_handle *llh = NULL;
+        llog_cb_t cb = args->lpca_cb;
+        struct llog_logid logid;
+        int rc;
         ENTRY;
 
-        LASSERT(sizeof(*llh) == LLOG_CHUNK_SIZE);
+        cfs_daemonize_ctxt("ll_log_process");
 
-        down(&cathandle->lgh_lock);
-        llh = cathandle->lgh_hdr;
+        logid = *(struct llog_logid *)(args->lpca_arg);
+        rc = llog_create(ctxt, &llh, &logid, NULL);
+        if (rc) {
+                CERROR("llog_create() failed %d\n", rc);
+                GOTO(out, rc);
+        }
+        rc = llog_init_handle(llh, LLOG_F_IS_CAT, NULL);
+        if (rc) {
+                CERROR("llog_init_handle failed %d\n", rc);
+                GOTO(release_llh, rc);
+        }
+
+        if (cb) {
+                rc = llog_cat_process(llh, cb, NULL, 0, 0);
+                if (rc != LLOG_PROC_BREAK && rc != 0)
+                        CERROR("llog_cat_process() failed %d\n", rc);
+                cb(llh, NULL, NULL);
+        } else {
+                CWARN("No callback function for recovery\n");
+        }
+
+        /*
+         * Make sure that all cached data is sent.
+         */
+       llog_sync(ctxt, NULL, 0);
+        GOTO(release_llh, rc);
+release_llh:
+        rc = llog_cat_put(llh);
+        if (rc)
+                CERROR("llog_cat_put() failed %d\n", rc);
+out:
+        llog_ctxt_put(ctxt);
+        OBD_FREE_PTR(args);
+        return rc;
+}
+EXPORT_SYMBOL(llog_cat_process_thread);
+#endif
+
+static int llog_cat_reverse_process_cb(struct llog_handle *cat_llh,
+                                       struct llog_rec_hdr *rec, void *data)
+{
+        struct llog_process_data *d = data;
+        struct llog_logid_rec *lir = (struct llog_logid_rec *)rec;
+        struct llog_handle *llh;
+        int rc;
+
+        if (le32_to_cpu(rec->lrh_type) != LLOG_LOGID_MAGIC) {
+                CERROR("invalid record in catalog\n");
+                RETURN(-EINVAL);
+        }
+        CDEBUG(D_HA, "processing log "LPX64":%x at index %u of catalog "
+               LPX64"\n", lir->lid_id.lgl_oid, lir->lid_id.lgl_ogen,
+               le32_to_cpu(rec->lrh_index), cat_llh->lgh_id.lgl_oid);
 
-        if (cathandle->lgh_file->f_dentry->d_inode->i_size == 0) {
-                llog_write_rec(cathandle, &llh->llh_hdr, NULL, 0, NULL, 0);
-
-write_hdr:    
-                rc = lustre_fwrite(cathandle->lgh_file, llh, LLOG_CHUNK_SIZE,
-                                   &offset);
-                if (rc != LLOG_CHUNK_SIZE) {
-                        CERROR("error writing catalog header: rc %d\n", rc);
-                        OBD_FREE(llh, sizeof(*llh));
-                        if (rc >= 0)
-                                rc = -ENOSPC;
-                } else
-                        rc = 0;
+        rc = llog_cat_id2handle(cat_llh, &llh, &lir->lid_id);
+        if (rc) {
+                CERROR("Cannot find handle for log "LPX64"\n",
+                       lir->lid_id.lgl_oid);
+                RETURN(rc);
+        }
+
+        rc = llog_reverse_process(llh, d->lpd_cb, d->lpd_data, NULL);
+        RETURN(rc);
+}
+
+int llog_cat_reverse_process(struct llog_handle *cat_llh,
+                             llog_cb_t cb, void *data)
+{
+        struct llog_process_data d;
+        struct llog_process_cat_data cd;
+        struct llog_log_hdr *llh = cat_llh->lgh_hdr;
+        int rc;
+        ENTRY;
+
+        LASSERT(llh->llh_flags & LLOG_F_IS_CAT);
+        d.lpd_data = data;
+        d.lpd_cb = cb;
+
+        if (llh->llh_cat_idx > cat_llh->lgh_last_idx) {
+                CWARN("catalog "LPX64" crosses index zero\n",
+                      cat_llh->lgh_id.lgl_oid);
+
+                cd.lpcd_first_idx = 0;
+                cd.lpcd_last_idx = cat_llh->lgh_last_idx;
+                rc = llog_reverse_process(cat_llh, llog_cat_reverse_process_cb,
+                                          &d, &cd);
+                if (rc != 0)
+                        RETURN(rc);
+
+                cd.lpcd_first_idx = le32_to_cpu(llh->llh_cat_idx);
+                cd.lpcd_last_idx = 0;
+                rc = llog_reverse_process(cat_llh, llog_cat_reverse_process_cb,
+                                          &d, &cd);
         } else {
-                rc = lustre_fread(cathandle->lgh_file, llh, LLOG_CHUNK_SIZE,
-                                  &offset);
-                if (rc != LLOG_CHUNK_SIZE) {
-                        CERROR("error reading catalog header: rc %d\n", rc);
-                        /* Can we do much else if the header is bad? */
-                        goto write_hdr;
-                } else
-                        rc = 0;
+                rc = llog_reverse_process(cat_llh, llog_cat_reverse_process_cb,
+                                          &d, NULL);
         }
 
-        cathandle->lgh_tgtuuid = &llh->llh_tgtuuid;
-        up(&cathandle->lgh_lock);
         RETURN(rc);
 }
-EXPORT_SYMBOL(llog_cat_init);
+EXPORT_SYMBOL(llog_cat_reverse_process);
 
-#endif
+int llog_cat_set_first_idx(struct llog_handle *cathandle, int index)
+{
+        struct llog_log_hdr *llh = cathandle->lgh_hdr;
+        int i, bitmap_size, idx;
+        ENTRY;
+
+        bitmap_size = LLOG_BITMAP_SIZE(llh);
+        if (llh->llh_cat_idx == (index - 1)) {
+                idx = llh->llh_cat_idx + 1;
+                llh->llh_cat_idx = idx;
+                if (idx == cathandle->lgh_last_idx)
+                        goto out;
+                for (i = (index + 1) % bitmap_size;
+                     i != cathandle->lgh_last_idx;
+                     i = (i + 1) % bitmap_size) {
+                        if (!ext2_test_bit(i, llh->llh_bitmap)) {
+                                idx = llh->llh_cat_idx + 1;
+                                llh->llh_cat_idx = idx;
+                        } else if (i == 0) {
+                                llh->llh_cat_idx = 0;
+                        } else {
+                                break;
+                        }
+                }
+out:
+                CDEBUG(D_RPCTRACE, "set catlog "LPX64" first idx %u\n",
+                       cathandle->lgh_id.lgl_oid, llh->llh_cat_idx);
+        }
+
+        RETURN(0);
+}