linux_dsm_epyc7002/fs/xfs/scrub/common.c

/*
 * Copyright (C) 2017 Oracle.  All Rights Reserved.
 *
 * Author: Darrick J. Wong <darrick.wong@oracle.com>
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public License
 * as published by the Free Software Foundation; either version 2
 * of the License, or (at your option) any later version.
 *
 * This program is distributed in the hope that it would 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
 * along with this program; if not, write the Free Software Foundation,
 * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301, USA.
 */
#include "xfs.h"
#include "xfs_fs.h"
#include "xfs_shared.h"
#include "xfs_format.h"
#include "xfs_trans_resv.h"
#include "xfs_mount.h"
#include "xfs_defer.h"
#include "xfs_btree.h"
#include "xfs_bit.h"
#include "xfs_log_format.h"
#include "xfs_trans.h"
#include "xfs_sb.h"
#include "xfs_inode.h"
#include "xfs_icache.h"
#include "xfs_itable.h"
#include "xfs_alloc.h"
#include "xfs_alloc_btree.h"
#include "xfs_bmap.h"
#include "xfs_bmap_btree.h"
#include "xfs_ialloc.h"
#include "xfs_ialloc_btree.h"
#include "xfs_refcount.h"
#include "xfs_refcount_btree.h"
#include "xfs_rmap.h"
#include "xfs_rmap_btree.h"
#include "xfs_log.h"
#include "xfs_trans_priv.h"
#include "scrub/xfs_scrub.h"
#include "scrub/scrub.h"
#include "scrub/common.h"
#include "scrub/trace.h"
#include "scrub/btree.h"

/* Common code for the metadata scrubbers. */

/*
 * Handling operational errors.
 *
 * The *_process_error() family of functions are used to process error return
 * codes from functions called as part of a scrub operation.
 *
 * If there's no error, we return true to tell the caller that it's ok
 * to move on to the next check in its list.
 *
 * For non-verifier errors (e.g. ENOMEM) we return false to tell the
 * caller that something bad happened, and we preserve *error so that
 * the caller can return the *error up the stack to userspace.
 *
 * Verifier errors (EFSBADCRC/EFSCORRUPTED) are recorded by setting
 * OFLAG_CORRUPT in sm_flags and the *error is cleared.  In other words,
 * we track verifier errors (and failed scrub checks) via OFLAG_CORRUPT,
 * not via return codes.  We return false to tell the caller that
 * something bad happened.  Since the error has been cleared, the caller
 * will (presumably) return that zero and scrubbing will move on to
 * whatever's next.
 *
 * ftrace can be used to record the precise metadata location and the
 * approximate code location of the failed operation.
 */

/* Check for operational errors. */
bool
xfs_scrub_process_error(
	struct xfs_scrub_context	*sc,
	xfs_agnumber_t			agno,
	xfs_agblock_t			bno,
	int				*error)
{
	switch (*error) {
	case 0:
		return true;
	case -EDEADLOCK:
		/* Used to restart an op with deadlock avoidance. */
		trace_xfs_scrub_deadlock_retry(sc->ip, sc->sm, *error);
		break;
	case -EFSBADCRC:
	case -EFSCORRUPTED:
		/* Note the badness but don't abort. */
		sc->sm->sm_flags |= XFS_SCRUB_OFLAG_CORRUPT;
		*error = 0;
		/* fall through */
	default:
		trace_xfs_scrub_op_error(sc, agno, bno, *error,
				__return_address);
		break;
	}
	return false;
}

/* Check for operational errors for a file offset. */
bool
xfs_scrub_fblock_process_error(
	struct xfs_scrub_context	*sc,
	int				whichfork,
	xfs_fileoff_t			offset,
	int				*error)
{
	switch (*error) {
	case 0:
		return true;
	case -EDEADLOCK:
		/* Used to restart an op with deadlock avoidance. */
		trace_xfs_scrub_deadlock_retry(sc->ip, sc->sm, *error);
		break;
	case -EFSBADCRC:
	case -EFSCORRUPTED:
		/* Note the badness but don't abort. */
		sc->sm->sm_flags |= XFS_SCRUB_OFLAG_CORRUPT;
		*error = 0;
		/* fall through */
	default:
		trace_xfs_scrub_file_op_error(sc, whichfork, offset, *error,
				__return_address);
		break;
	}
	return false;
}

/*
 * Handling scrub corruption/optimization/warning checks.
 *
 * The *_set_{corrupt,preen,warning}() family of functions are used to
 * record the presence of metadata that is incorrect (corrupt), could be
 * optimized somehow (preen), or should be flagged for administrative
 * review but is not incorrect (warn).
 *
 * ftrace can be used to record the precise metadata location and
 * approximate code location of the failed check.
 */

/* Record a block which could be optimized. */
void
xfs_scrub_block_set_preen(
	struct xfs_scrub_context	*sc,
	struct xfs_buf			*bp)
{
	sc->sm->sm_flags |= XFS_SCRUB_OFLAG_PREEN;
	trace_xfs_scrub_block_preen(sc, bp->b_bn, __return_address);
}

/*
 * Record an inode which could be optimized.  The trace data will
 * include the block given by bp if bp is given; otherwise it will use
 * the block location of the inode record itself.
 */
void
xfs_scrub_ino_set_preen(
	struct xfs_scrub_context	*sc,
	struct xfs_buf			*bp)
{
	sc->sm->sm_flags |= XFS_SCRUB_OFLAG_PREEN;
	trace_xfs_scrub_ino_preen(sc, sc->ip->i_ino, bp ? bp->b_bn : 0,
			__return_address);
}

/* Record a corrupt block. */
void
xfs_scrub_block_set_corrupt(
	struct xfs_scrub_context	*sc,
	struct xfs_buf			*bp)
{
	sc->sm->sm_flags |= XFS_SCRUB_OFLAG_CORRUPT;
	trace_xfs_scrub_block_error(sc, bp->b_bn, __return_address);
}

/*
 * Record a corrupt inode.  The trace data will include the block given
 * by bp if bp is given; otherwise it will use the block location of the
 * inode record itself.
 */
void
xfs_scrub_ino_set_corrupt(
	struct xfs_scrub_context	*sc,
	xfs_ino_t			ino,
	struct xfs_buf			*bp)
{
	sc->sm->sm_flags |= XFS_SCRUB_OFLAG_CORRUPT;
	trace_xfs_scrub_ino_error(sc, ino, bp ? bp->b_bn : 0, __return_address);
}

/* Record corruption in a block indexed by a file fork. */
void
xfs_scrub_fblock_set_corrupt(
	struct xfs_scrub_context	*sc,
	int				whichfork,
	xfs_fileoff_t			offset)
{
	sc->sm->sm_flags |= XFS_SCRUB_OFLAG_CORRUPT;
	trace_xfs_scrub_fblock_error(sc, whichfork, offset, __return_address);
}

/*
 * Warn about inodes that need administrative review but is not
 * incorrect.
 */
void
xfs_scrub_ino_set_warning(
	struct xfs_scrub_context	*sc,
	struct xfs_buf			*bp)
{
	sc->sm->sm_flags |= XFS_SCRUB_OFLAG_WARNING;
	trace_xfs_scrub_ino_warning(sc, sc->ip->i_ino, bp ? bp->b_bn : 0,
			__return_address);
}

/* Warn about a block indexed by a file fork that needs review. */
void
xfs_scrub_fblock_set_warning(
	struct xfs_scrub_context	*sc,
	int				whichfork,
	xfs_fileoff_t			offset)
{
	sc->sm->sm_flags |= XFS_SCRUB_OFLAG_WARNING;
	trace_xfs_scrub_fblock_warning(sc, whichfork, offset, __return_address);
}

/* Signal an incomplete scrub. */
void
xfs_scrub_set_incomplete(
	struct xfs_scrub_context	*sc)
{
	sc->sm->sm_flags |= XFS_SCRUB_OFLAG_INCOMPLETE;
	trace_xfs_scrub_incomplete(sc, __return_address);
}

/*
 * AG scrubbing
 *
 * These helpers facilitate locking an allocation group's header
 * buffers, setting up cursors for all btrees that are present, and
 * cleaning everything up once we're through.
 */

/* Decide if we want to return an AG header read failure. */
static inline bool
want_ag_read_header_failure(
	struct xfs_scrub_context	*sc,
	unsigned int			type)
{
	/* Return all AG header read failures when scanning btrees. */
	if (sc->sm->sm_type != XFS_SCRUB_TYPE_AGF &&
	    sc->sm->sm_type != XFS_SCRUB_TYPE_AGFL &&
	    sc->sm->sm_type != XFS_SCRUB_TYPE_AGI)
		return true;
	/*
	 * If we're scanning a given type of AG header, we only want to
	 * see read failures from that specific header.  We'd like the
	 * other headers to cross-check them, but this isn't required.
	 */
	if (sc->sm->sm_type == type)
		return true;
	return false;
}

/*
 * Grab all the headers for an AG.
 *
 * The headers should be released by xfs_scrub_ag_free, but as a fail
 * safe we attach all the buffers we grab to the scrub transaction so
 * they'll all be freed when we cancel it.
 */
int
xfs_scrub_ag_read_headers(
	struct xfs_scrub_context	*sc,
	xfs_agnumber_t			agno,
	struct xfs_buf			**agi,
	struct xfs_buf			**agf,
	struct xfs_buf			**agfl)
{
	struct xfs_mount		*mp = sc->mp;
	int				error;

	error = xfs_ialloc_read_agi(mp, sc->tp, agno, agi);
	if (error && want_ag_read_header_failure(sc, XFS_SCRUB_TYPE_AGI))
		goto out;

	error = xfs_alloc_read_agf(mp, sc->tp, agno, 0, agf);
	if (error && want_ag_read_header_failure(sc, XFS_SCRUB_TYPE_AGF))
		goto out;

	error = xfs_alloc_read_agfl(mp, sc->tp, agno, agfl);
	if (error && want_ag_read_header_failure(sc, XFS_SCRUB_TYPE_AGFL))
		goto out;

out:
	return error;
}

/* Release all the AG btree cursors. */
void
xfs_scrub_ag_btcur_free(
	struct xfs_scrub_ag		*sa)
{
	if (sa->refc_cur)
		xfs_btree_del_cursor(sa->refc_cur, XFS_BTREE_ERROR);
	if (sa->rmap_cur)
		xfs_btree_del_cursor(sa->rmap_cur, XFS_BTREE_ERROR);
	if (sa->fino_cur)
		xfs_btree_del_cursor(sa->fino_cur, XFS_BTREE_ERROR);
	if (sa->ino_cur)
		xfs_btree_del_cursor(sa->ino_cur, XFS_BTREE_ERROR);
	if (sa->cnt_cur)
		xfs_btree_del_cursor(sa->cnt_cur, XFS_BTREE_ERROR);
	if (sa->bno_cur)
		xfs_btree_del_cursor(sa->bno_cur, XFS_BTREE_ERROR);

	sa->refc_cur = NULL;
	sa->rmap_cur = NULL;
	sa->fino_cur = NULL;
	sa->ino_cur = NULL;
	sa->bno_cur = NULL;
	sa->cnt_cur = NULL;
}

/* Initialize all the btree cursors for an AG. */
int
xfs_scrub_ag_btcur_init(
	struct xfs_scrub_context	*sc,
	struct xfs_scrub_ag		*sa)
{
	struct xfs_mount		*mp = sc->mp;
	xfs_agnumber_t			agno = sa->agno;

	if (sa->agf_bp) {
		/* Set up a bnobt cursor for cross-referencing. */
		sa->bno_cur = xfs_allocbt_init_cursor(mp, sc->tp, sa->agf_bp,
				agno, XFS_BTNUM_BNO);
		if (!sa->bno_cur)
			goto err;

		/* Set up a cntbt cursor for cross-referencing. */
		sa->cnt_cur = xfs_allocbt_init_cursor(mp, sc->tp, sa->agf_bp,
				agno, XFS_BTNUM_CNT);
		if (!sa->cnt_cur)
			goto err;
	}

	/* Set up a inobt cursor for cross-referencing. */
	if (sa->agi_bp) {
		sa->ino_cur = xfs_inobt_init_cursor(mp, sc->tp, sa->agi_bp,
					agno, XFS_BTNUM_INO);
		if (!sa->ino_cur)
			goto err;
	}

	/* Set up a finobt cursor for cross-referencing. */
	if (sa->agi_bp && xfs_sb_version_hasfinobt(&mp->m_sb)) {
		sa->fino_cur = xfs_inobt_init_cursor(mp, sc->tp, sa->agi_bp,
				agno, XFS_BTNUM_FINO);
		if (!sa->fino_cur)
			goto err;
	}

	/* Set up a rmapbt cursor for cross-referencing. */
	if (sa->agf_bp && xfs_sb_version_hasrmapbt(&mp->m_sb)) {
		sa->rmap_cur = xfs_rmapbt_init_cursor(mp, sc->tp, sa->agf_bp,
				agno);
		if (!sa->rmap_cur)
			goto err;
	}

	/* Set up a refcountbt cursor for cross-referencing. */
	if (sa->agf_bp && xfs_sb_version_hasreflink(&mp->m_sb)) {
		sa->refc_cur = xfs_refcountbt_init_cursor(mp, sc->tp,
				sa->agf_bp, agno, NULL);
		if (!sa->refc_cur)
			goto err;
	}

	return 0;
err:
	return -ENOMEM;
}

/* Release the AG header context and btree cursors. */
void
xfs_scrub_ag_free(
	struct xfs_scrub_context	*sc,
	struct xfs_scrub_ag		*sa)
{
	xfs_scrub_ag_btcur_free(sa);
	if (sa->agfl_bp) {
		xfs_trans_brelse(sc->tp, sa->agfl_bp);
		sa->agfl_bp = NULL;
	}
	if (sa->agf_bp) {
		xfs_trans_brelse(sc->tp, sa->agf_bp);
		sa->agf_bp = NULL;
	}
	if (sa->agi_bp) {
		xfs_trans_brelse(sc->tp, sa->agi_bp);
		sa->agi_bp = NULL;
	}
	sa->agno = NULLAGNUMBER;
}

/*
 * For scrub, grab the AGI and the AGF headers, in that order.  Locking
 * order requires us to get the AGI before the AGF.  We use the
 * transaction to avoid deadlocking on crosslinked metadata buffers;
 * either the caller passes one in (bmap scrub) or we have to create a
 * transaction ourselves.
 */
int
xfs_scrub_ag_init(
	struct xfs_scrub_context	*sc,
	xfs_agnumber_t			agno,
	struct xfs_scrub_ag		*sa)
{
	int				error;

	sa->agno = agno;
	error = xfs_scrub_ag_read_headers(sc, agno, &sa->agi_bp,
			&sa->agf_bp, &sa->agfl_bp);
	if (error)
		return error;

	return xfs_scrub_ag_btcur_init(sc, sa);
}

/* Per-scrubber setup functions */

/* Set us up with a transaction and an empty context. */
int
xfs_scrub_setup_fs(
	struct xfs_scrub_context	*sc,
	struct xfs_inode		*ip)
{
	return xfs_scrub_trans_alloc(sc->sm, sc->mp, &sc->tp);
}

/* Set us up with AG headers and btree cursors. */
int
xfs_scrub_setup_ag_btree(
	struct xfs_scrub_context	*sc,
	struct xfs_inode		*ip,
	bool				force_log)
{
	struct xfs_mount		*mp = sc->mp;
	int				error;

	/*
	 * If the caller asks us to checkpont the log, do so.  This
	 * expensive operation should be performed infrequently and only
	 * as a last resort.  Any caller that sets force_log should
	 * document why they need to do so.
	 */
	if (force_log) {
		error = xfs_scrub_checkpoint_log(mp);
		if (error)
			return error;
	}

	error = xfs_scrub_setup_ag_header(sc, ip);
	if (error)
		return error;

	return xfs_scrub_ag_init(sc, sc->sm->sm_agno, &sc->sa);
}

/* Push everything out of the log onto disk. */
int
xfs_scrub_checkpoint_log(
	struct xfs_mount	*mp)
{
	int			error;

	error = _xfs_log_force(mp, XFS_LOG_SYNC, NULL);
	if (error)
		return error;
	xfs_ail_push_all_sync(mp->m_ail);
	return 0;
}

/*
 * Given an inode and the scrub control structure, grab either the
 * inode referenced in the control structure or the inode passed in.
 * The inode is not locked.
 */
int
xfs_scrub_get_inode(
	struct xfs_scrub_context	*sc,
	struct xfs_inode		*ip_in)
{
	struct xfs_mount		*mp = sc->mp;
	struct xfs_inode		*ip = NULL;
	int				error;

	/*
	 * If userspace passed us an AG number or a generation number
	 * without an inode number, they haven't got a clue so bail out
	 * immediately.
	 */
	if (sc->sm->sm_agno || (sc->sm->sm_gen && !sc->sm->sm_ino))
		return -EINVAL;

	/* We want to scan the inode we already had opened. */
	if (sc->sm->sm_ino == 0 || sc->sm->sm_ino == ip_in->i_ino) {
		sc->ip = ip_in;
		return 0;
	}

	/* Look up the inode, see if the generation number matches. */
	if (xfs_internal_inum(mp, sc->sm->sm_ino))
		return -ENOENT;
	error = xfs_iget(mp, NULL, sc->sm->sm_ino,
			XFS_IGET_UNTRUSTED | XFS_IGET_DONTCACHE, 0, &ip);
	if (error == -ENOENT || error == -EINVAL) {
		/* inode doesn't exist... */
		return -ENOENT;
	} else if (error) {
		trace_xfs_scrub_op_error(sc,
				XFS_INO_TO_AGNO(mp, sc->sm->sm_ino),
				XFS_INO_TO_AGBNO(mp, sc->sm->sm_ino),
				error, __return_address);
		return error;
	}
	if (VFS_I(ip)->i_generation != sc->sm->sm_gen) {
		iput(VFS_I(ip));
		return -ENOENT;
	}

	sc->ip = ip;
	return 0;
}