spa_errlog.c

Go to the documentation of this file.

/*
 * CDDL HEADER START
 *
 * The contents of this file are subject to the terms of the
 * Common Development and Distribution License (the "License").
 * You may not use this file except in compliance with the License.
 *
 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
 * or http://www.opensolaris.org/os/licensing.
 * See the License for the specific language governing permissions
 * and limitations under the License.
 *
 * When distributing Covered Code, include this CDDL HEADER in each
 * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
 * If applicable, add the following below this CDDL HEADER, with the
 * fields enclosed by brackets "[]" replaced with your own identifying
 * information: Portions Copyright [yyyy] [name of copyright owner]
 *
 * CDDL HEADER END
 */
/*
 * Copyright (c) 2006, 2010, Oracle and/or its affiliates. All rights reserved.
 */

#include <sys/dmu_tx.h>
#include <sys/spa.h>
#include <sys/spa_impl.h>
#include <sys/zap.h>
#include <sys/zio.h>


static void
bookmark_to_name(zbookmark_t *zb, char *buf, size_t len)
{
        (void) snprintf(buf, len, "%llx:%llx:%llx:%llx",
            (u_longlong_t)zb->zb_objset, (u_longlong_t)zb->zb_object,
            (u_longlong_t)zb->zb_level, (u_longlong_t)zb->zb_blkid);
}

#ifdef _KERNEL
static void
name_to_bookmark(char *buf, zbookmark_t *zb)
{
        zb->zb_objset = strtonum(buf, &buf);
        ASSERT(*buf == ':');
        zb->zb_object = strtonum(buf + 1, &buf);
        ASSERT(*buf == ':');
        zb->zb_level = (int)strtonum(buf + 1, &buf);
        ASSERT(*buf == ':');
        zb->zb_blkid = strtonum(buf + 1, &buf);
        ASSERT(*buf == '\0');
}
#endif

void
spa_log_error(spa_t *spa, zio_t *zio)
{
        zbookmark_t *zb = &zio->io_logical->io_bookmark;
        spa_error_entry_t search;
        spa_error_entry_t *new;
        avl_tree_t *tree;
        avl_index_t where;

        /*
         * If we are trying to import a pool, ignore any errors, as we won't be
         * writing to the pool any time soon.
         */
        if (spa_load_state(spa) == SPA_LOAD_TRYIMPORT)
                return;

        mutex_enter(&spa->spa_errlist_lock);

        /*
         * If we have had a request to rotate the log, log it to the next list
         * instead of the current one.
         */
        if (spa->spa_scrub_active || spa->spa_scrub_finished)
                tree = &spa->spa_errlist_scrub;
        else
                tree = &spa->spa_errlist_last;

        search.se_bookmark = *zb;
        if (avl_find(tree, &search, &where) != NULL) {
                mutex_exit(&spa->spa_errlist_lock);
                return;
        }

        new = kmem_zalloc(sizeof (spa_error_entry_t), KM_SLEEP);
        new->se_bookmark = *zb;
        avl_insert(tree, new, where);

        mutex_exit(&spa->spa_errlist_lock);
}

uint64_t
spa_get_errlog_size(spa_t *spa)
{
        uint64_t total = 0, count;

        mutex_enter(&spa->spa_errlog_lock);
        if (spa->spa_errlog_scrub != 0 &&
            zap_count(spa->spa_meta_objset, spa->spa_errlog_scrub,
            &count) == 0)
                total += count;

        if (spa->spa_errlog_last != 0 && !spa->spa_scrub_finished &&
            zap_count(spa->spa_meta_objset, spa->spa_errlog_last,
            &count) == 0)
                total += count;
        mutex_exit(&spa->spa_errlog_lock);

        mutex_enter(&spa->spa_errlist_lock);
        total += avl_numnodes(&spa->spa_errlist_last);
        total += avl_numnodes(&spa->spa_errlist_scrub);
        mutex_exit(&spa->spa_errlist_lock);

        return (total);
}

#ifdef _KERNEL
static int
process_error_log(spa_t *spa, uint64_t obj, void *addr, size_t *count)
{
        zap_cursor_t zc;
        zap_attribute_t za;
        zbookmark_t zb;

        if (obj == 0)
                return (0);

        for (zap_cursor_init(&zc, spa->spa_meta_objset, obj);
            zap_cursor_retrieve(&zc, &za) == 0;
            zap_cursor_advance(&zc)) {

                if (*count == 0) {
                        zap_cursor_fini(&zc);
                        return (ENOMEM);
                }

                name_to_bookmark(za.za_name, &zb);

                if (copyout(&zb, (char *)addr +
                    (*count - 1) * sizeof (zbookmark_t),
                    sizeof (zbookmark_t)) != 0)
                        return (EFAULT);

                *count -= 1;
        }

        zap_cursor_fini(&zc);

        return (0);
}

static int
process_error_list(avl_tree_t *list, void *addr, size_t *count)
{
        spa_error_entry_t *se;

        for (se = avl_first(list); se != NULL; se = AVL_NEXT(list, se)) {

                if (*count == 0)
                        return (ENOMEM);

                if (copyout(&se->se_bookmark, (char *)addr +
                    (*count - 1) * sizeof (zbookmark_t),
                    sizeof (zbookmark_t)) != 0)
                        return (EFAULT);

                *count -= 1;
        }

        return (0);
}
#endif

int
spa_get_errlog(spa_t *spa, void *uaddr, size_t *count)
{
        int ret = 0;

#ifdef _KERNEL
        mutex_enter(&spa->spa_errlog_lock);

        ret = process_error_log(spa, spa->spa_errlog_scrub, uaddr, count);

        if (!ret && !spa->spa_scrub_finished)
                ret = process_error_log(spa, spa->spa_errlog_last, uaddr,
                    count);

        mutex_enter(&spa->spa_errlist_lock);
        if (!ret)
                ret = process_error_list(&spa->spa_errlist_scrub, uaddr,
                    count);
        if (!ret)
                ret = process_error_list(&spa->spa_errlist_last, uaddr,
                    count);
        mutex_exit(&spa->spa_errlist_lock);

        mutex_exit(&spa->spa_errlog_lock);
#endif

        return (ret);
}

void
spa_errlog_rotate(spa_t *spa)
{
        mutex_enter(&spa->spa_errlist_lock);
        spa->spa_scrub_finished = B_TRUE;
        mutex_exit(&spa->spa_errlist_lock);
}

void
spa_errlog_drain(spa_t *spa)
{
        spa_error_entry_t *se;
        void *cookie;

        mutex_enter(&spa->spa_errlist_lock);

        cookie = NULL;
        while ((se = avl_destroy_nodes(&spa->spa_errlist_last,
            &cookie)) != NULL)
                kmem_free(se, sizeof (spa_error_entry_t));
        cookie = NULL;
        while ((se = avl_destroy_nodes(&spa->spa_errlist_scrub,
            &cookie)) != NULL)
                kmem_free(se, sizeof (spa_error_entry_t));

        mutex_exit(&spa->spa_errlist_lock);
}

static void
sync_error_list(spa_t *spa, avl_tree_t *t, uint64_t *obj, dmu_tx_t *tx)
{
        spa_error_entry_t *se;
        char buf[64];
        void *cookie;

        if (avl_numnodes(t) != 0) {
                /* create log if necessary */
                if (*obj == 0)
                        *obj = zap_create(spa->spa_meta_objset,
                            DMU_OT_ERROR_LOG, DMU_OT_NONE,
                            0, tx);

                /* add errors to the current log */
                for (se = avl_first(t); se != NULL; se = AVL_NEXT(t, se)) {
                        char *name = se->se_name ? se->se_name : "";

                        bookmark_to_name(&se->se_bookmark, buf, sizeof (buf));

                        (void) zap_update(spa->spa_meta_objset,
                            *obj, buf, 1, strlen(name) + 1, name, tx);
                }

                /* purge the error list */
                cookie = NULL;
                while ((se = avl_destroy_nodes(t, &cookie)) != NULL)
                        kmem_free(se, sizeof (spa_error_entry_t));
        }
}

void
spa_errlog_sync(spa_t *spa, uint64_t txg)
{
        dmu_tx_t *tx;
        avl_tree_t scrub, last;
        int scrub_finished;

        mutex_enter(&spa->spa_errlist_lock);

        /*
         * Bail out early under normal circumstances.
         */
        if (avl_numnodes(&spa->spa_errlist_scrub) == 0 &&
            avl_numnodes(&spa->spa_errlist_last) == 0 &&
            !spa->spa_scrub_finished) {
                mutex_exit(&spa->spa_errlist_lock);
                return;
        }

        spa_get_errlists(spa, &last, &scrub);
        scrub_finished = spa->spa_scrub_finished;
        spa->spa_scrub_finished = B_FALSE;

        mutex_exit(&spa->spa_errlist_lock);
        mutex_enter(&spa->spa_errlog_lock);

        tx = dmu_tx_create_assigned(spa->spa_dsl_pool, txg);

        /*
         * Sync out the current list of errors.
         */
        sync_error_list(spa, &last, &spa->spa_errlog_last, tx);

        /*
         * Rotate the log if necessary.
         */
        if (scrub_finished) {
                if (spa->spa_errlog_last != 0)
                        VERIFY(dmu_object_free(spa->spa_meta_objset,
                            spa->spa_errlog_last, tx) == 0);
                spa->spa_errlog_last = spa->spa_errlog_scrub;
                spa->spa_errlog_scrub = 0;

                sync_error_list(spa, &scrub, &spa->spa_errlog_last, tx);
        }

        /*
         * Sync out any pending scrub errors.
         */
        sync_error_list(spa, &scrub, &spa->spa_errlog_scrub, tx);

        /*
         * Update the MOS to reflect the new values.
         */
        (void) zap_update(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT,
            DMU_POOL_ERRLOG_LAST, sizeof (uint64_t), 1,
            &spa->spa_errlog_last, tx);
        (void) zap_update(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT,
            DMU_POOL_ERRLOG_SCRUB, sizeof (uint64_t), 1,
            &spa->spa_errlog_scrub, tx);

        dmu_tx_commit(tx);

        mutex_exit(&spa->spa_errlog_lock);
}