xref: /linux/fs/gfs2/recovery.c (revision 277536c882467e348173238e4ff701a4e4d2fb9e)

// SPDX-License-Identifier: GPL-2.0-only
/*
 * Copyright (C) Sistina Software, Inc.  1997-2003 All rights reserved.
 * Copyright (C) 2004-2006 Red Hat, Inc.  All rights reserved.
 */

#include <linux/module.h>
#include <linux/slab.h>
#include <linux/spinlock.h>
#include <linux/completion.h>
#include <linux/buffer_head.h>
#include <linux/gfs2_ondisk.h>
#include <linux/crc32.h>
#include <linux/crc32c.h>
#include <linux/ktime.h>

#include "gfs2.h"
#include "incore.h"
#include "bmap.h"
#include "glock.h"
#include "glops.h"
#include "log.h"
#include "lops.h"
#include "meta_io.h"
#include "recovery.h"
#include "super.h"
#include "util.h"
#include "dir.h"

struct workqueue_struct *gfs2_recovery_wq;

int gfs2_replay_read_block(struct gfs2_jdesc *jd, unsigned int blk,
			   struct buffer_head **bh)
{
	struct inode *inode = jd->jd_inode;
	struct gfs2_glock *gl = gfs2_inode_glock(inode);
	struct gfs2_inode *ip = GFS2_I(inode);
	u64 dblock;
	u32 extlen;
	int error;

	extlen = 32;
	error = gfs2_get_extent(inode, blk, &dblock, &extlen);
	if (error)
		return error;
	if (!dblock) {
		gfs2_consist_inode(ip);
		return -EIO;
	}

	*bh = gfs2_meta_ra(gl, dblock, extlen);

	return error;
}

int gfs2_revoke_add(struct gfs2_jdesc *jd, u64 blkno, unsigned int where)
{
	struct list_head *head = &jd->jd_revoke_list;
	struct gfs2_revoke_replay *rr = NULL, *iter;

	list_for_each_entry(iter, head, rr_list) {
		if (iter->rr_blkno == blkno) {
			rr = iter;
			break;
		}
	}

	if (rr) {
		rr->rr_where = where;
		return 0;
	}

	rr = kmalloc_obj(struct gfs2_revoke_replay, GFP_NOFS);
	if (!rr)
		return -ENOMEM;

	rr->rr_blkno = blkno;
	rr->rr_where = where;
	list_add(&rr->rr_list, head);

	return 1;
}

int gfs2_revoke_check(struct gfs2_jdesc *jd, u64 blkno, unsigned int where)
{
	struct gfs2_revoke_replay *rr = NULL, *iter;
	int wrap, a, b, revoke;

	list_for_each_entry(iter, &jd->jd_revoke_list, rr_list) {
		if (iter->rr_blkno == blkno) {
			rr = iter;
			break;
		}
	}

	if (!rr)
		return 0;

	wrap = (rr->rr_where < jd->jd_replay_tail);
	a = (jd->jd_replay_tail < where);
	b = (where < rr->rr_where);
	revoke = (wrap) ? (a || b) : (a && b);

	return revoke;
}

void gfs2_revoke_clean(struct gfs2_jdesc *jd)
{
	struct list_head *head = &jd->jd_revoke_list;
	struct gfs2_revoke_replay *rr;

	while (!list_empty(head)) {
		rr = list_first_entry(head, struct gfs2_revoke_replay, rr_list);
		list_del(&rr->rr_list);
		kfree(rr);
	}
}

int __get_log_header(struct gfs2_sbd *sdp, const struct gfs2_log_header *lh,
		     unsigned int blkno, struct gfs2_log_header_host *head)
{
	const u32 zero = 0;
	u32 hash, crc;

	if (lh->lh_header.mh_magic != cpu_to_be32(GFS2_MAGIC) ||
	    lh->lh_header.mh_type != cpu_to_be32(GFS2_METATYPE_LH) ||
	    (blkno && be32_to_cpu(lh->lh_blkno) != blkno))
		return 1;

	hash = crc32(~0, lh, LH_V1_SIZE - 4);
	hash = ~crc32(hash, &zero, 4); /* assume lh_hash is zero */

	if (be32_to_cpu(lh->lh_hash) != hash)
		return 1;

	crc = crc32c(~0, (void *)lh + LH_V1_SIZE + 4,
		     sdp->sd_sb.sb_bsize - LH_V1_SIZE - 4);

	if ((lh->lh_crc != 0 && be32_to_cpu(lh->lh_crc) != crc))
		return 1;

	head->lh_sequence = be64_to_cpu(lh->lh_sequence);
	head->lh_flags = be32_to_cpu(lh->lh_flags);
	head->lh_tail = be32_to_cpu(lh->lh_tail);
	head->lh_blkno = be32_to_cpu(lh->lh_blkno);

	head->lh_local_total = be64_to_cpu(lh->lh_local_total);
	head->lh_local_free = be64_to_cpu(lh->lh_local_free);
	head->lh_local_dinodes = be64_to_cpu(lh->lh_local_dinodes);

	return 0;
}
/**
 * get_log_header - read the log header for a given segment
 * @jd: the journal
 * @blk: the block to look at
 * @head: the log header to return
 *
 * Read the log header for a given segment in a given journal.  Do a few
 * sanity checks on it.
 *
 * Returns: 0 on success,
 *          1 if the header was invalid or incomplete,
 *          errno on error
 */

static int get_log_header(struct gfs2_jdesc *jd, unsigned int blk,
			  struct gfs2_log_header_host *head)
{
	struct gfs2_sbd *sdp = GFS2_SB(jd->jd_inode);
	struct buffer_head *bh;
	int error;

	error = gfs2_replay_read_block(jd, blk, &bh);
	if (error)
		return error;

	error = __get_log_header(sdp, (const struct gfs2_log_header *)bh->b_data,
				 blk, head);
	brelse(bh);

	return error;
}

/**
 * foreach_descriptor - go through the active part of the log
 * @jd: the journal
 * @start: the first log header in the active region
 * @end: the last log header (don't process the contents of this entry))
 * @pass: iteration number (foreach_descriptor() is called in a for() loop)
 *
 * Call a given function once for every log descriptor in the active
 * portion of the log.
 *
 * Returns: errno
 */

static int foreach_descriptor(struct gfs2_jdesc *jd, u32 start,
			      unsigned int end, int pass)
{
	struct gfs2_sbd *sdp = GFS2_SB(jd->jd_inode);
	struct buffer_head *bh;
	struct gfs2_log_descriptor *ld;
	int error = 0;
	u32 length;
	__be64 *ptr;
	unsigned int offset = sizeof(struct gfs2_log_descriptor);
	offset += sizeof(__be64) - 1;
	offset &= ~(sizeof(__be64) - 1);

	while (start != end) {
		error = gfs2_replay_read_block(jd, start, &bh);
		if (error)
			return error;
		if (gfs2_meta_check(sdp, bh)) {
			brelse(bh);
			return -EIO;
		}
		ld = (struct gfs2_log_descriptor *)bh->b_data;
		length = be32_to_cpu(ld->ld_length);

		if (be32_to_cpu(ld->ld_header.mh_type) == GFS2_METATYPE_LH) {
			struct gfs2_log_header_host lh;
			error = get_log_header(jd, start, &lh);
			if (!error) {
				gfs2_replay_incr_blk(jd, &start);
				brelse(bh);
				continue;
			}
			if (error == 1) {
				gfs2_consist_inode(GFS2_I(jd->jd_inode));
				error = -EIO;
			}
			brelse(bh);
			return error;
		} else if (gfs2_metatype_check(sdp, bh, GFS2_METATYPE_LD)) {
			brelse(bh);
			return -EIO;
		}
		ptr = (__be64 *)(bh->b_data + offset);
		error = lops_scan_elements(jd, start, ld, ptr, pass);
		if (error) {
			brelse(bh);
			return error;
		}

		while (length--)
			gfs2_replay_incr_blk(jd, &start);

		brelse(bh);
	}

	return 0;
}

/**
 * clean_journal - mark a dirty journal as being clean
 * @jd: the journal
 * @head: the head journal to start from
 *
 * Returns: errno
 */

static void clean_journal(struct gfs2_jdesc *jd,
			  struct gfs2_log_header_host *head)
{
	struct gfs2_sbd *sdp = GFS2_SB(jd->jd_inode);

	gfs2_replay_incr_blk(jd, &head->lh_blkno);
	head->lh_sequence++;
	gfs2_write_log_header(sdp, jd, head->lh_sequence, 0, head->lh_blkno,
			      GFS2_LOG_HEAD_UNMOUNT | GFS2_LOG_HEAD_RECOVERY,
			      REQ_PREFLUSH | REQ_FUA | REQ_META | REQ_SYNC);
}


static void gfs2_recovery_done(struct gfs2_sbd *sdp, unsigned int jid,
                               unsigned int message)
{
	char env_jid[20];
	char env_status[20];
	char *envp[] = { env_jid, env_status, NULL };
	struct lm_lockstruct *ls = &sdp->sd_lockstruct;

        ls->ls_recover_jid_done = jid;
        ls->ls_recover_jid_status = message;
	sprintf(env_jid, "JID=%u", jid);
	sprintf(env_status, "RECOVERY=%s",
		message == LM_RD_SUCCESS ? "Done" : "Failed");
        kobject_uevent_env(&sdp->sd_kobj, KOBJ_CHANGE, envp);

	if (sdp->sd_lockstruct.ls_ops->lm_recovery_result)
		sdp->sd_lockstruct.ls_ops->lm_recovery_result(sdp, jid, message);
}

/**
 * update_statfs_inode - Update the master statfs inode or zero out the local
 *			 statfs inode for a given journal.
 * @jd: The journal
 * @head: If NULL, @inode is the local statfs inode and we need to zero it out.
 *	  Otherwise, it @head contains the statfs change info that needs to be
 *	  synced to the master statfs inode (pointed to by @inode).
 * @inode: statfs inode to update.
 */
static int update_statfs_inode(struct gfs2_jdesc *jd,
			       struct gfs2_log_header_host *head,
			       struct inode *inode)
{
	struct gfs2_sbd *sdp = GFS2_SB(jd->jd_inode);
	struct buffer_head *bh;
	struct gfs2_statfs_change_host sc;
	int error = 0;

	BUG_ON(!inode);

	error = gfs2_meta_inode_buffer(GFS2_I(inode), &bh);
	if (error)
		goto out;

	spin_lock(&sdp->sd_statfs_spin);

	if (head) { /* Update the master statfs inode */
		gfs2_statfs_change_in(&sc, bh->b_data + sizeof(struct gfs2_dinode));
		sc.sc_total += head->lh_local_total;
		sc.sc_free += head->lh_local_free;
		sc.sc_dinodes += head->lh_local_dinodes;
		gfs2_statfs_change_out(&sc, bh->b_data + sizeof(struct gfs2_dinode));

		fs_info(sdp, "jid=%u: Updated master statfs Total:%lld, "
			"Free:%lld, Dinodes:%lld after change "
			"[%+lld,%+lld,%+lld]\n", jd->jd_jid, sc.sc_total,
			sc.sc_free, sc.sc_dinodes, head->lh_local_total,
			head->lh_local_free, head->lh_local_dinodes);
	} else { /* Zero out the local statfs inode */
		memset(bh->b_data + sizeof(struct gfs2_dinode), 0,
		       sizeof(struct gfs2_statfs_change));
		/* If it's our own journal, reset any in-memory changes too */
		if (jd->jd_jid == sdp->sd_lockstruct.ls_jid) {
			memset(&sdp->sd_statfs_local, 0,
			       sizeof(struct gfs2_statfs_change_host));
		}
	}
	spin_unlock(&sdp->sd_statfs_spin);

	mark_buffer_dirty(bh);
	brelse(bh);
	gfs2_inode_metasync(gfs2_inode_glock(inode));

out:
	return error;
}

/**
 * recover_local_statfs - Update the master and local statfs changes for this
 *			  journal.
 *
 * Previously, statfs updates would be read in from the local statfs inode and
 * synced to the master statfs inode during recovery.
 *
 * We now use the statfs updates in the journal head to update the master statfs
 * inode instead of reading in from the local statfs inode. To preserve backward
 * compatibility with kernels that can't do this, we still need to keep the
 * local statfs inode up to date by writing changes to it. At some point in the
 * future, we can do away with the local statfs inodes altogether and keep the
 * statfs changes solely in the journal.
 *
 * @jd: the journal
 * @head: the journal head
 *
 * Returns: errno
 */
static void recover_local_statfs(struct gfs2_jdesc *jd,
				 struct gfs2_log_header_host *head)
{
	int error;
	struct gfs2_sbd *sdp = GFS2_SB(jd->jd_inode);

	if (!head->lh_local_total && !head->lh_local_free
	    && !head->lh_local_dinodes) /* No change */
		goto zero_local;

	 /* First update the master statfs inode with the changes we
	  * found in the journal. */
	error = update_statfs_inode(jd, head, sdp->sd_statfs_inode);
	if (error)
		goto out;

zero_local:
	/* Zero out the local statfs inode so any changes in there
	 * are not re-recovered. */
	error = update_statfs_inode(jd, NULL,
				    find_local_statfs_inode(sdp, jd->jd_jid));
out:
	return;
}

void gfs2_recover_func(struct work_struct *work)
{
	struct gfs2_jdesc *jd = container_of(work, struct gfs2_jdesc, jd_work);
	struct gfs2_glock *gl = gfs2_inode_glock(jd->jd_inode);
	struct gfs2_sbd *sdp = GFS2_SB(jd->jd_inode);
	struct gfs2_log_header_host head;
	struct gfs2_holder j_gh, ji_gh;
	ktime_t t_start, t_jlck, t_jhd, t_tlck, t_rep;
	int ro = 0;
	unsigned int pass;
	int error = 0;
	int jlocked = 0;

	if (gfs2_withdrawn(sdp)) {
		fs_err(sdp, "jid=%u: Recovery not attempted due to withdraw.\n",
		       jd->jd_jid);
		goto fail;
	}
	t_start = ktime_get();
	if (sdp->sd_args.ar_spectator)
		goto fail;
	if (jd->jd_jid != sdp->sd_lockstruct.ls_jid) {
		fs_info(sdp, "jid=%u: Trying to acquire journal glock...\n",
			jd->jd_jid);
		jlocked = 1;
		/* Acquire the journal glock so we can do recovery */

		error = gfs2_glock_nq_num(sdp, jd->jd_jid, &gfs2_journal_glops,
					  LM_ST_EXCLUSIVE,
					  LM_FLAG_RECOVER | LM_FLAG_TRY |
					  GL_NOCACHE,
					  &j_gh);
		switch (error) {
		case 0:
			break;

		case GLR_TRYFAILED:
			fs_info(sdp, "jid=%u: Busy\n", jd->jd_jid);
			error = 0;
			goto fail;

		default:
			goto fail;
		}

		error = gfs2_glock_nq_init(gl, LM_ST_SHARED,
					   LM_FLAG_RECOVER | GL_NOCACHE,
					   &ji_gh);
		if (error)
			goto fail_gunlock_j;
	} else {
		fs_info(sdp, "jid=%u, already locked for use\n", jd->jd_jid);
	}

	t_jlck = ktime_get();
	fs_info(sdp, "jid=%u: Looking at journal...\n", jd->jd_jid);

	error = gfs2_jdesc_check(jd);
	if (error)
		goto fail_gunlock_ji;

	error = gfs2_find_jhead(jd, &head);
	if (error)
		goto fail_gunlock_ji;
	t_jhd = ktime_get();
	fs_info(sdp, "jid=%u: Journal head lookup took %lldms\n", jd->jd_jid,
		ktime_ms_delta(t_jhd, t_jlck));

	if (!(head.lh_flags & GFS2_LOG_HEAD_UNMOUNT)) {
		mutex_lock(&sdp->sd_freeze_mutex);

		if (test_bit(SDF_FROZEN, &sdp->sd_flags)) {
			mutex_unlock(&sdp->sd_freeze_mutex);
			fs_warn(sdp, "jid=%u: Can't replay: filesystem "
				"is frozen\n", jd->jd_jid);
			goto fail_gunlock_ji;
		}

		if (test_bit(SDF_RORECOVERY, &sdp->sd_flags)) {
			ro = 1;
		} else if (test_bit(SDF_JOURNAL_CHECKED, &sdp->sd_flags)) {
			if (!test_bit(SDF_JOURNAL_LIVE, &sdp->sd_flags))
				ro = 1;
		} else {
			if (sb_rdonly(sdp->sd_vfs)) {
				/* check if device itself is read-only */
				ro = bdev_read_only(sdp->sd_vfs->s_bdev);
				if (!ro) {
					fs_info(sdp, "recovery required on "
						"read-only filesystem.\n");
					fs_info(sdp, "write access will be "
						"enabled during recovery.\n");
				}
			}
		}

		if (ro) {
			fs_warn(sdp, "jid=%u: Can't replay: read-only block "
				"device\n", jd->jd_jid);
			error = -EROFS;
			goto fail_gunlock_nofreeze;
		}

		t_tlck = ktime_get();
		fs_info(sdp, "jid=%u: Replaying journal...0x%x to 0x%x\n",
			jd->jd_jid, head.lh_tail, head.lh_blkno);

		/* We take the sd_log_flush_lock here primarily to prevent log
		 * flushes and simultaneous journal replays from stomping on
		 * each other wrt jd_log_bio. */
		down_read(&sdp->sd_log_flush_lock);
		for (pass = 0; pass < 2; pass++) {
			lops_before_scan(jd, &head, pass);
			error = foreach_descriptor(jd, head.lh_tail,
						   head.lh_blkno, pass);
			lops_after_scan(jd, error, pass);
			if (error) {
				up_read(&sdp->sd_log_flush_lock);
				goto fail_gunlock_nofreeze;
			}
		}

		recover_local_statfs(jd, &head);
		clean_journal(jd, &head);
		up_read(&sdp->sd_log_flush_lock);

		mutex_unlock(&sdp->sd_freeze_mutex);
		t_rep = ktime_get();
		fs_info(sdp, "jid=%u: Journal replayed in %lldms [jlck:%lldms, "
			"jhead:%lldms, tlck:%lldms, replay:%lldms]\n",
			jd->jd_jid, ktime_ms_delta(t_rep, t_start),
			ktime_ms_delta(t_jlck, t_start),
			ktime_ms_delta(t_jhd, t_jlck),
			ktime_ms_delta(t_tlck, t_jhd),
			ktime_ms_delta(t_rep, t_tlck));
	}

	if (jd->jd_jid == sdp->sd_lockstruct.ls_jid)
		gfs2_log_pointers_init(sdp, &head);

	gfs2_recovery_done(sdp, jd->jd_jid, LM_RD_SUCCESS);

	if (jlocked) {
		gfs2_glock_dq_uninit(&ji_gh);
		gfs2_glock_dq_uninit(&j_gh);
	}

	fs_info(sdp, "jid=%u: Done\n", jd->jd_jid);
	goto done;

fail_gunlock_nofreeze:
	mutex_unlock(&sdp->sd_freeze_mutex);
fail_gunlock_ji:
	if (jlocked) {
		gfs2_glock_dq_uninit(&ji_gh);
fail_gunlock_j:
		gfs2_glock_dq_uninit(&j_gh);
	}

	fs_info(sdp, "jid=%u: %s\n", jd->jd_jid, (error) ? "Failed" : "Done");
fail:
	jd->jd_recover_error = error;
	gfs2_recovery_done(sdp, jd->jd_jid, LM_RD_GAVEUP);
done:
	clear_bit(JDF_RECOVERY, &jd->jd_flags);
	smp_mb__after_atomic();
	wake_up_bit(&jd->jd_flags, JDF_RECOVERY);
}

int gfs2_recover_journal(struct gfs2_jdesc *jd, bool wait)
{
	int rv;

	if (test_and_set_bit(JDF_RECOVERY, &jd->jd_flags))
		return -EBUSY;

	/* we have JDF_RECOVERY, queue should always succeed */
	rv = queue_work(gfs2_recovery_wq, &jd->jd_work);
	BUG_ON(!rv);

	if (wait)
		wait_on_bit(&jd->jd_flags, JDF_RECOVERY,
			    TASK_UNINTERRUPTIBLE);

	return wait ? jd->jd_recover_error : 0;
}

void gfs2_log_pointers_init(struct gfs2_sbd *sdp,
			    struct gfs2_log_header_host *head)
{
	sdp->sd_log_sequence = head->lh_sequence + 1;
	gfs2_replay_incr_blk(sdp->sd_jdesc, &head->lh_blkno);
	sdp->sd_log_tail = head->lh_blkno;
	sdp->sd_log_flush_head = head->lh_blkno;
	sdp->sd_log_flush_tail = head->lh_blkno;
	sdp->sd_log_head = head->lh_blkno;
}