[linux.git] / fs / xfs / xfs_bmap_item.c

// SPDX-License-Identifier: GPL-2.0+
/*
 * Copyright (C) 2016 Oracle.  All Rights Reserved.
 * Author: Darrick J. Wong <[email protected]>
 */
#include "xfs.h"
#include "xfs_fs.h"
#include "xfs_format.h"
#include "xfs_log_format.h"
#include "xfs_trans_resv.h"
#include "xfs_bit.h"
#include "xfs_mount.h"
#include "xfs_defer.h"
#include "xfs_inode.h"
#include "xfs_trans.h"
#include "xfs_trans_priv.h"
#include "xfs_buf_item.h"
#include "xfs_bmap_item.h"
#include "xfs_log.h"
#include "xfs_bmap.h"
#include "xfs_icache.h"
#include "xfs_trace.h"
#include "xfs_bmap_btree.h"
#include "xfs_trans_space.h"


kmem_zone_t	*xfs_bui_zone;
kmem_zone_t	*xfs_bud_zone;

static inline struct xfs_bui_log_item *BUI_ITEM(struct xfs_log_item *lip)
{
	return container_of(lip, struct xfs_bui_log_item, bui_item);
}

void
xfs_bui_item_free(
	struct xfs_bui_log_item	*buip)
{
	kmem_zone_free(xfs_bui_zone, buip);
}

/*
 * Freeing the BUI requires that we remove it from the AIL if it has already
 * been placed there. However, the BUI may not yet have been placed in the AIL
 * when called by xfs_bui_release() from BUD processing due to the ordering of
 * committed vs unpin operations in bulk insert operations. Hence the reference
 * count to ensure only the last caller frees the BUI.
 */
void
xfs_bui_release(
	struct xfs_bui_log_item	*buip)
{
	ASSERT(atomic_read(&buip->bui_refcount) > 0);
	if (atomic_dec_and_test(&buip->bui_refcount)) {
		xfs_trans_ail_remove(&buip->bui_item, SHUTDOWN_LOG_IO_ERROR);
		xfs_bui_item_free(buip);
	}
}


STATIC void
xfs_bui_item_size(
	struct xfs_log_item	*lip,
	int			*nvecs,
	int			*nbytes)
{
	struct xfs_bui_log_item	*buip = BUI_ITEM(lip);

	*nvecs += 1;
	*nbytes += xfs_bui_log_format_sizeof(buip->bui_format.bui_nextents);
}

/*
 * This is called to fill in the vector of log iovecs for the
 * given bui log item. We use only 1 iovec, and we point that
 * at the bui_log_format structure embedded in the bui item.
 * It is at this point that we assert that all of the extent
 * slots in the bui item have been filled.
 */
STATIC void
xfs_bui_item_format(
	struct xfs_log_item	*lip,
	struct xfs_log_vec	*lv)
{
	struct xfs_bui_log_item	*buip = BUI_ITEM(lip);
	struct xfs_log_iovec	*vecp = NULL;

	ASSERT(atomic_read(&buip->bui_next_extent) ==
			buip->bui_format.bui_nextents);

	buip->bui_format.bui_type = XFS_LI_BUI;
	buip->bui_format.bui_size = 1;

	xlog_copy_iovec(lv, &vecp, XLOG_REG_TYPE_BUI_FORMAT, &buip->bui_format,
			xfs_bui_log_format_sizeof(buip->bui_format.bui_nextents));
}

/*
 * Pinning has no meaning for an bui item, so just return.
 */
STATIC void
xfs_bui_item_pin(
	struct xfs_log_item	*lip)
{
}

/*
 * The unpin operation is the last place an BUI is manipulated in the log. It is
 * either inserted in the AIL or aborted in the event of a log I/O error. In
 * either case, the BUI transaction has been successfully committed to make it
 * this far. Therefore, we expect whoever committed the BUI to either construct
 * and commit the BUD or drop the BUD's reference in the event of error. Simply
 * drop the log's BUI reference now that the log is done with it.
 */
STATIC void
xfs_bui_item_unpin(
	struct xfs_log_item	*lip,
	int			remove)
{
	struct xfs_bui_log_item	*buip = BUI_ITEM(lip);

	xfs_bui_release(buip);
}

/*
 * BUI items have no locking or pushing.  However, since BUIs are pulled from
 * the AIL when their corresponding BUDs are committed to disk, their situation
 * is very similar to being pinned.  Return XFS_ITEM_PINNED so that the caller
 * will eventually flush the log.  This should help in getting the BUI out of
 * the AIL.
 */
STATIC uint
xfs_bui_item_push(
	struct xfs_log_item	*lip,
	struct list_head	*buffer_list)
{
	return XFS_ITEM_PINNED;
}

/*
 * The BUI has been either committed or aborted if the transaction has been
 * cancelled. If the transaction was cancelled, an BUD isn't going to be
 * constructed and thus we free the BUI here directly.
 */
STATIC void
xfs_bui_item_unlock(
	struct xfs_log_item	*lip)
{
	if (test_bit(XFS_LI_ABORTED, &lip->li_flags))
		xfs_bui_release(BUI_ITEM(lip));
}

/*
 * The BUI is logged only once and cannot be moved in the log, so simply return
 * the lsn at which it's been logged.
 */
STATIC xfs_lsn_t
xfs_bui_item_committed(
	struct xfs_log_item	*lip,
	xfs_lsn_t		lsn)
{
	return lsn;
}

/*
 * The BUI dependency tracking op doesn't do squat.  It can't because
 * it doesn't know where the free extent is coming from.  The dependency
 * tracking has to be handled by the "enclosing" metadata object.  For
 * example, for inodes, the inode is locked throughout the extent freeing
 * so the dependency should be recorded there.
 */
STATIC void
xfs_bui_item_committing(
	struct xfs_log_item	*lip,
	xfs_lsn_t		lsn)
{
}

/*
 * This is the ops vector shared by all bui log items.
 */
static const struct xfs_item_ops xfs_bui_item_ops = {
	.iop_size	= xfs_bui_item_size,
	.iop_format	= xfs_bui_item_format,
	.iop_pin	= xfs_bui_item_pin,
	.iop_unpin	= xfs_bui_item_unpin,
	.iop_unlock	= xfs_bui_item_unlock,
	.iop_committed	= xfs_bui_item_committed,
	.iop_push	= xfs_bui_item_push,
	.iop_committing = xfs_bui_item_committing,
};

/*
 * Allocate and initialize an bui item with the given number of extents.
 */
struct xfs_bui_log_item *
xfs_bui_init(
	struct xfs_mount		*mp)

{
	struct xfs_bui_log_item		*buip;

	buip = kmem_zone_zalloc(xfs_bui_zone, KM_SLEEP);

	xfs_log_item_init(mp, &buip->bui_item, XFS_LI_BUI, &xfs_bui_item_ops);
	buip->bui_format.bui_nextents = XFS_BUI_MAX_FAST_EXTENTS;
	buip->bui_format.bui_id = (uintptr_t)(void *)buip;
	atomic_set(&buip->bui_next_extent, 0);
	atomic_set(&buip->bui_refcount, 2);

	return buip;
}

static inline struct xfs_bud_log_item *BUD_ITEM(struct xfs_log_item *lip)
{
	return container_of(lip, struct xfs_bud_log_item, bud_item);
}

STATIC void
xfs_bud_item_size(
	struct xfs_log_item	*lip,
	int			*nvecs,
	int			*nbytes)
{
	*nvecs += 1;
	*nbytes += sizeof(struct xfs_bud_log_format);
}

/*
 * This is called to fill in the vector of log iovecs for the
 * given bud log item. We use only 1 iovec, and we point that
 * at the bud_log_format structure embedded in the bud item.
 * It is at this point that we assert that all of the extent
 * slots in the bud item have been filled.
 */
STATIC void
xfs_bud_item_format(
	struct xfs_log_item	*lip,
	struct xfs_log_vec	*lv)
{
	struct xfs_bud_log_item	*budp = BUD_ITEM(lip);
	struct xfs_log_iovec	*vecp = NULL;

	budp->bud_format.bud_type = XFS_LI_BUD;
	budp->bud_format.bud_size = 1;

	xlog_copy_iovec(lv, &vecp, XLOG_REG_TYPE_BUD_FORMAT, &budp->bud_format,
			sizeof(struct xfs_bud_log_format));
}

/*
 * Pinning has no meaning for an bud item, so just return.
 */
STATIC void
xfs_bud_item_pin(
	struct xfs_log_item	*lip)
{
}

/*
 * Since pinning has no meaning for an bud item, unpinning does
 * not either.
 */
STATIC void
xfs_bud_item_unpin(
	struct xfs_log_item	*lip,
	int			remove)
{
}

/*
 * There isn't much you can do to push on an bud item.  It is simply stuck
 * waiting for the log to be flushed to disk.
 */
STATIC uint
xfs_bud_item_push(
	struct xfs_log_item	*lip,
	struct list_head	*buffer_list)
{
	return XFS_ITEM_PINNED;
}

/*
 * The BUD is either committed or aborted if the transaction is cancelled. If
 * the transaction is cancelled, drop our reference to the BUI and free the
 * BUD.
 */
STATIC void
xfs_bud_item_unlock(
	struct xfs_log_item	*lip)
{
	struct xfs_bud_log_item	*budp = BUD_ITEM(lip);

	if (test_bit(XFS_LI_ABORTED, &lip->li_flags)) {
		xfs_bui_release(budp->bud_buip);
		kmem_zone_free(xfs_bud_zone, budp);
	}
}

/*
 * When the bud item is committed to disk, all we need to do is delete our
 * reference to our partner bui item and then free ourselves. Since we're
 * freeing ourselves we must return -1 to keep the transaction code from
 * further referencing this item.
 */
STATIC xfs_lsn_t
xfs_bud_item_committed(
	struct xfs_log_item	*lip,
	xfs_lsn_t		lsn)
{
	struct xfs_bud_log_item	*budp = BUD_ITEM(lip);

	/*
	 * Drop the BUI reference regardless of whether the BUD has been
	 * aborted. Once the BUD transaction is constructed, it is the sole
	 * responsibility of the BUD to release the BUI (even if the BUI is
	 * aborted due to log I/O error).
	 */
	xfs_bui_release(budp->bud_buip);
	kmem_zone_free(xfs_bud_zone, budp);

	return (xfs_lsn_t)-1;
}

/*
 * The BUD dependency tracking op doesn't do squat.  It can't because
 * it doesn't know where the free extent is coming from.  The dependency
 * tracking has to be handled by the "enclosing" metadata object.  For
 * example, for inodes, the inode is locked throughout the extent freeing
 * so the dependency should be recorded there.
 */
STATIC void
xfs_bud_item_committing(
	struct xfs_log_item	*lip,
	xfs_lsn_t		lsn)
{
}

/*
 * This is the ops vector shared by all bud log items.
 */
static const struct xfs_item_ops xfs_bud_item_ops = {
	.iop_size	= xfs_bud_item_size,
	.iop_format	= xfs_bud_item_format,
	.iop_pin	= xfs_bud_item_pin,
	.iop_unpin	= xfs_bud_item_unpin,
	.iop_unlock	= xfs_bud_item_unlock,
	.iop_committed	= xfs_bud_item_committed,
	.iop_push	= xfs_bud_item_push,
	.iop_committing = xfs_bud_item_committing,
};

/*
 * Allocate and initialize an bud item with the given number of extents.
 */
struct xfs_bud_log_item *
xfs_bud_init(
	struct xfs_mount		*mp,
	struct xfs_bui_log_item		*buip)

{
	struct xfs_bud_log_item	*budp;

	budp = kmem_zone_zalloc(xfs_bud_zone, KM_SLEEP);
	xfs_log_item_init(mp, &budp->bud_item, XFS_LI_BUD, &xfs_bud_item_ops);
	budp->bud_buip = buip;
	budp->bud_format.bud_bui_id = buip->bui_format.bui_id;

	return budp;
}

/*
 * Process a bmap update intent item that was recovered from the log.
 * We need to update some inode's bmbt.
 */
int
xfs_bui_recover(
	struct xfs_trans		*parent_tp,
	struct xfs_bui_log_item		*buip)
{
	int				error = 0;
	unsigned int			bui_type;
	struct xfs_map_extent		*bmap;
	xfs_fsblock_t			startblock_fsb;
	xfs_fsblock_t			inode_fsb;
	xfs_filblks_t			count;
	bool				op_ok;
	struct xfs_bud_log_item		*budp;
	enum xfs_bmap_intent_type	type;
	int				whichfork;
	xfs_exntst_t			state;
	struct xfs_trans		*tp;
	struct xfs_inode		*ip = NULL;
	struct xfs_bmbt_irec		irec;
	struct xfs_mount		*mp = parent_tp->t_mountp;

	ASSERT(!test_bit(XFS_BUI_RECOVERED, &buip->bui_flags));

	/* Only one mapping operation per BUI... */
	if (buip->bui_format.bui_nextents != XFS_BUI_MAX_FAST_EXTENTS) {
		set_bit(XFS_BUI_RECOVERED, &buip->bui_flags);
		xfs_bui_release(buip);
		return -EIO;
	}

	/*
	 * First check the validity of the extent described by the
	 * BUI.  If anything is bad, then toss the BUI.
	 */
	bmap = &buip->bui_format.bui_extents[0];
	startblock_fsb = XFS_BB_TO_FSB(mp,
			   XFS_FSB_TO_DADDR(mp, bmap->me_startblock));
	inode_fsb = XFS_BB_TO_FSB(mp, XFS_FSB_TO_DADDR(mp,
			XFS_INO_TO_FSB(mp, bmap->me_owner)));
	switch (bmap->me_flags & XFS_BMAP_EXTENT_TYPE_MASK) {
	case XFS_BMAP_MAP:
	case XFS_BMAP_UNMAP:
		op_ok = true;
		break;
	default:
		op_ok = false;
		break;
	}
	if (!op_ok || startblock_fsb == 0 ||
	    bmap->me_len == 0 ||
	    inode_fsb == 0 ||
	    startblock_fsb >= mp->m_sb.sb_dblocks ||
	    bmap->me_len >= mp->m_sb.sb_agblocks ||
	    inode_fsb >= mp->m_sb.sb_dblocks ||
	    (bmap->me_flags & ~XFS_BMAP_EXTENT_FLAGS)) {
		/*
		 * This will pull the BUI from the AIL and
		 * free the memory associated with it.
		 */
		set_bit(XFS_BUI_RECOVERED, &buip->bui_flags);
		xfs_bui_release(buip);
		return -EIO;
	}

	error = xfs_trans_alloc(mp, &M_RES(mp)->tr_itruncate,
			XFS_EXTENTADD_SPACE_RES(mp, XFS_DATA_FORK), 0, 0, &tp);
	if (error)
		return error;
	/*
	 * Recovery stashes all deferred ops during intent processing and
	 * finishes them on completion. Transfer current dfops state to this
	 * transaction and transfer the result back before we return.
	 */
	xfs_defer_move(tp, parent_tp);
	budp = xfs_trans_get_bud(tp, buip);

	/* Grab the inode. */
	error = xfs_iget(mp, tp, bmap->me_owner, 0, XFS_ILOCK_EXCL, &ip);
	if (error)
		goto err_inode;

	if (VFS_I(ip)->i_nlink == 0)
		xfs_iflags_set(ip, XFS_IRECOVERY);

	/* Process deferred bmap item. */
	state = (bmap->me_flags & XFS_BMAP_EXTENT_UNWRITTEN) ?
			XFS_EXT_UNWRITTEN : XFS_EXT_NORM;
	whichfork = (bmap->me_flags & XFS_BMAP_EXTENT_ATTR_FORK) ?
			XFS_ATTR_FORK : XFS_DATA_FORK;
	bui_type = bmap->me_flags & XFS_BMAP_EXTENT_TYPE_MASK;
	switch (bui_type) {
	case XFS_BMAP_MAP:
	case XFS_BMAP_UNMAP:
		type = bui_type;
		break;
	default:
		error = -EFSCORRUPTED;
		goto err_inode;
	}
	xfs_trans_ijoin(tp, ip, 0);

	count = bmap->me_len;
	error = xfs_trans_log_finish_bmap_update(tp, budp, type, ip, whichfork,
			bmap->me_startoff, bmap->me_startblock, &count, state);
	if (error)
		goto err_inode;

	if (count > 0) {
		ASSERT(type == XFS_BMAP_UNMAP);
		irec.br_startblock = bmap->me_startblock;
		irec.br_blockcount = count;
		irec.br_startoff = bmap->me_startoff;
		irec.br_state = state;
		error = xfs_bmap_unmap_extent(tp, ip, &irec);
		if (error)
			goto err_inode;
	}

	set_bit(XFS_BUI_RECOVERED, &buip->bui_flags);
	xfs_defer_move(parent_tp, tp);
	error = xfs_trans_commit(tp);
	xfs_iunlock(ip, XFS_ILOCK_EXCL);
	xfs_irele(ip);

	return error;

err_inode:
	xfs_defer_move(parent_tp, tp);
	xfs_trans_cancel(tp);
	if (ip) {
		xfs_iunlock(ip, XFS_ILOCK_EXCL);
		xfs_irele(ip);
	}
	return error;
}
Commit	Line	Data
0b61f8a4	1	// SPDX-License-Identifier: GPL-2.0+
6413a014 DW	2	/*
6413a014 DW	3	* Copyright (C) 2016 Oracle. All Rights Reserved.
6413a014	4	* Author: Darrick J. Wong <[email protected]>
6413a014 DW	5	*/
	6	#include "xfs.h"
	7	#include "xfs_fs.h"
	8	#include "xfs_format.h"
	9	#include "xfs_log_format.h"
	10	#include "xfs_trans_resv.h"
77d61fe4	11	#include "xfs_bit.h"
6413a014	12	#include "xfs_mount.h"
77d61fe4 DW	13	#include "xfs_defer.h"
77d61fe4 DW	14	#include "xfs_inode.h"
6413a014 DW	15	#include "xfs_trans.h"
	16	#include "xfs_trans_priv.h"
	17	#include "xfs_buf_item.h"
	18	#include "xfs_bmap_item.h"
	19	#include "xfs_log.h"
77d61fe4 DW	20	#include "xfs_bmap.h"
	21	#include "xfs_icache.h"
	22	#include "xfs_trace.h"
fe0be23e DW	23	#include "xfs_bmap_btree.h"
fe0be23e DW	24	#include "xfs_trans_space.h"
6413a014 DW	25
	26
	27	kmem_zone_t *xfs_bui_zone;
	28	kmem_zone_t *xfs_bud_zone;
	29
	30	static inline struct xfs_bui_log_item BUI_ITEM(struct xfs_log_item lip)
	31	{
	32	return container_of(lip, struct xfs_bui_log_item, bui_item);
	33	}
	34
	35	void
	36	xfs_bui_item_free(
	37	struct xfs_bui_log_item *buip)
	38	{
	39	kmem_zone_free(xfs_bui_zone, buip);
	40	}
	41
0612d116 DC	42	/*
	43	* Freeing the BUI requires that we remove it from the AIL if it has already
	44	* been placed there. However, the BUI may not yet have been placed in the AIL
	45	* when called by xfs_bui_release() from BUD processing due to the ordering of
	46	* committed vs unpin operations in bulk insert operations. Hence the reference
	47	* count to ensure only the last caller frees the BUI.
	48	*/
	49	void
	50	xfs_bui_release(
	51	struct xfs_bui_log_item *buip)
	52	{
	53	ASSERT(atomic_read(&buip->bui_refcount) > 0);
	54	if (atomic_dec_and_test(&buip->bui_refcount)) {
	55	xfs_trans_ail_remove(&buip->bui_item, SHUTDOWN_LOG_IO_ERROR);
	56	xfs_bui_item_free(buip);
	57	}
	58	}
	59
	60
6413a014 DW	61	STATIC void
	62	xfs_bui_item_size(
	63	struct xfs_log_item *lip,
	64	int *nvecs,
	65	int *nbytes)
	66	{
	67	struct xfs_bui_log_item *buip = BUI_ITEM(lip);
	68
	69	*nvecs += 1;
	70	*nbytes += xfs_bui_log_format_sizeof(buip->bui_format.bui_nextents);
	71	}
	72
	73	/*
	74	* This is called to fill in the vector of log iovecs for the
	75	* given bui log item. We use only 1 iovec, and we point that
	76	* at the bui_log_format structure embedded in the bui item.
	77	* It is at this point that we assert that all of the extent
	78	* slots in the bui item have been filled.
	79	*/
	80	STATIC void
	81	xfs_bui_item_format(
	82	struct xfs_log_item *lip,
	83	struct xfs_log_vec *lv)
	84	{
	85	struct xfs_bui_log_item *buip = BUI_ITEM(lip);
	86	struct xfs_log_iovec *vecp = NULL;
	87
	88	ASSERT(atomic_read(&buip->bui_next_extent) ==
	89	buip->bui_format.bui_nextents);
	90
	91	buip->bui_format.bui_type = XFS_LI_BUI;
	92	buip->bui_format.bui_size = 1;
	93
	94	xlog_copy_iovec(lv, &vecp, XLOG_REG_TYPE_BUI_FORMAT, &buip->bui_format,
	95	xfs_bui_log_format_sizeof(buip->bui_format.bui_nextents));
	96	}
	97
	98	/*
	99	* Pinning has no meaning for an bui item, so just return.
	100	*/
	101	STATIC void
	102	xfs_bui_item_pin(
	103	struct xfs_log_item *lip)
	104	{
	105	}
	106
	107	/*
	108	* The unpin operation is the last place an BUI is manipulated in the log. It is
	109	* either inserted in the AIL or aborted in the event of a log I/O error. In
	110	* either case, the BUI transaction has been successfully committed to make it
	111	* this far. Therefore, we expect whoever committed the BUI to either construct
	112	* and commit the BUD or drop the BUD's reference in the event of error. Simply
	113	* drop the log's BUI reference now that the log is done with it.
	114	*/
	115	STATIC void
	116	xfs_bui_item_unpin(
	117	struct xfs_log_item *lip,
	118	int remove)
	119	{
	120	struct xfs_bui_log_item *buip = BUI_ITEM(lip);
	121
	122	xfs_bui_release(buip);
	123	}
	124
125	/*
126	* BUI items have no locking or pushing. However, since BUIs are pulled from
127	* the AIL when their corresponding BUDs are committed to disk, their situation
128	* is very similar to being pinned. Return XFS_ITEM_PINNED so that the caller
129	* will eventually flush the log. This should help in getting the BUI out of
130	* the AIL.
131	*/
132	STATIC uint
133	xfs_bui_item_push(
134	struct xfs_log_item *lip,
135	struct list_head *buffer_list)
136	{
137	return XFS_ITEM_PINNED;
138	}
139
140	/*
141	* The BUI has been either committed or aborted if the transaction has been
142	* cancelled. If the transaction was cancelled, an BUD isn't going to be
143	* constructed and thus we free the BUI here directly.
144	*/
145	STATIC void
146	xfs_bui_item_unlock(
147	struct xfs_log_item *lip)
148	{
22525c17	149	if (test_bit(XFS_LI_ABORTED, &lip->li_flags))
0612d116	150	xfs_bui_release(BUI_ITEM(lip));
6413a014 DW	151	}
	152
	153	/*
	154	* The BUI is logged only once and cannot be moved in the log, so simply return
	155	* the lsn at which it's been logged.
	156	*/
	157	STATIC xfs_lsn_t
	158	xfs_bui_item_committed(
	159	struct xfs_log_item *lip,
	160	xfs_lsn_t lsn)
	161	{
	162	return lsn;
	163	}
	164
	165	/*
	166	* The BUI dependency tracking op doesn't do squat. It can't because
	167	* it doesn't know where the free extent is coming from. The dependency
	168	* tracking has to be handled by the "enclosing" metadata object. For
	169	* example, for inodes, the inode is locked throughout the extent freeing
	170	* so the dependency should be recorded there.
	171	*/
	172	STATIC void
	173	xfs_bui_item_committing(
	174	struct xfs_log_item *lip,
	175	xfs_lsn_t lsn)
	176	{
	177	}
	178
	179	/*
	180	* This is the ops vector shared by all bui log items.
	181	*/
	182	static const struct xfs_item_ops xfs_bui_item_ops = {
	183	.iop_size = xfs_bui_item_size,
	184	.iop_format = xfs_bui_item_format,
	185	.iop_pin = xfs_bui_item_pin,
	186	.iop_unpin = xfs_bui_item_unpin,
	187	.iop_unlock = xfs_bui_item_unlock,
	188	.iop_committed = xfs_bui_item_committed,
	189	.iop_push = xfs_bui_item_push,
	190	.iop_committing = xfs_bui_item_committing,
	191	};
	192
	193	/*
	194	* Allocate and initialize an bui item with the given number of extents.
	195	*/
	196	struct xfs_bui_log_item *
	197	xfs_bui_init(
	198	struct xfs_mount *mp)
	199
	200	{
	201	struct xfs_bui_log_item *buip;
	202
	203	buip = kmem_zone_zalloc(xfs_bui_zone, KM_SLEEP);
	204
	205	xfs_log_item_init(mp, &buip->bui_item, XFS_LI_BUI, &xfs_bui_item_ops);
	206	buip->bui_format.bui_nextents = XFS_BUI_MAX_FAST_EXTENTS;
	207	buip->bui_format.bui_id = (uintptr_t)(void *)buip;
	208	atomic_set(&buip->bui_next_extent, 0);
	209	atomic_set(&buip->bui_refcount, 2);
	210
	211	return buip;
	212	}
	213
6413a014 DW	214	static inline struct xfs_bud_log_item BUD_ITEM(struct xfs_log_item lip)
	215	{
	216	return container_of(lip, struct xfs_bud_log_item, bud_item);
	217	}
	218
	219	STATIC void
	220	xfs_bud_item_size(
	221	struct xfs_log_item *lip,
	222	int *nvecs,
	223	int *nbytes)
	224	{
	225	*nvecs += 1;
	226	*nbytes += sizeof(struct xfs_bud_log_format);
	227	}
	228
	229	/*
	230	* This is called to fill in the vector of log iovecs for the
	231	* given bud log item. We use only 1 iovec, and we point that
	232	* at the bud_log_format structure embedded in the bud item.
	233	* It is at this point that we assert that all of the extent
	234	* slots in the bud item have been filled.
	235	*/
	236	STATIC void
	237	xfs_bud_item_format(
	238	struct xfs_log_item *lip,
	239	struct xfs_log_vec *lv)
	240	{
	241	struct xfs_bud_log_item *budp = BUD_ITEM(lip);
	242	struct xfs_log_iovec *vecp = NULL;
	243
	244	budp->bud_format.bud_type = XFS_LI_BUD;
	245	budp->bud_format.bud_size = 1;
	246
	247	xlog_copy_iovec(lv, &vecp, XLOG_REG_TYPE_BUD_FORMAT, &budp->bud_format,
	248	sizeof(struct xfs_bud_log_format));
	249	}
	250
	251	/*
	252	* Pinning has no meaning for an bud item, so just return.
	253	*/
	254	STATIC void
	255	xfs_bud_item_pin(
	256	struct xfs_log_item *lip)
	257	{
	258	}
	259
	260	/*
	261	* Since pinning has no meaning for an bud item, unpinning does
	262	* not either.
	263	*/
	264	STATIC void
	265	xfs_bud_item_unpin(
	266	struct xfs_log_item *lip,
	267	int remove)
	268	{
	269	}
	270
	271	/*
	272	* There isn't much you can do to push on an bud item. It is simply stuck
	273	* waiting for the log to be flushed to disk.
	274	*/
	275	STATIC uint
	276	xfs_bud_item_push(
	277	struct xfs_log_item *lip,
278	struct list_head *buffer_list)
279	{
280	return XFS_ITEM_PINNED;
281	}
282
283	/*
284	* The BUD is either committed or aborted if the transaction is cancelled. If
285	* the transaction is cancelled, drop our reference to the BUI and free the
286	* BUD.
287	*/
288	STATIC void
289	xfs_bud_item_unlock(
290	struct xfs_log_item *lip)
291	{
292	struct xfs_bud_log_item *budp = BUD_ITEM(lip);
293
22525c17	294	if (test_bit(XFS_LI_ABORTED, &lip->li_flags)) {
6413a014 DW	295	xfs_bui_release(budp->bud_buip);
	296	kmem_zone_free(xfs_bud_zone, budp);
	297	}
	298	}
	299
	300	/*
	301	* When the bud item is committed to disk, all we need to do is delete our
	302	* reference to our partner bui item and then free ourselves. Since we're
	303	* freeing ourselves we must return -1 to keep the transaction code from
	304	* further referencing this item.
	305	*/
	306	STATIC xfs_lsn_t
	307	xfs_bud_item_committed(
	308	struct xfs_log_item *lip,
	309	xfs_lsn_t lsn)
	310	{
	311	struct xfs_bud_log_item *budp = BUD_ITEM(lip);
	312
	313	/*
	314	* Drop the BUI reference regardless of whether the BUD has been
	315	* aborted. Once the BUD transaction is constructed, it is the sole
	316	* responsibility of the BUD to release the BUI (even if the BUI is
	317	* aborted due to log I/O error).
	318	*/
	319	xfs_bui_release(budp->bud_buip);
	320	kmem_zone_free(xfs_bud_zone, budp);
	321
	322	return (xfs_lsn_t)-1;
	323	}
	324
	325	/*
	326	* The BUD dependency tracking op doesn't do squat. It can't because
	327	* it doesn't know where the free extent is coming from. The dependency
	328	* tracking has to be handled by the "enclosing" metadata object. For
	329	* example, for inodes, the inode is locked throughout the extent freeing
	330	* so the dependency should be recorded there.
	331	*/
	332	STATIC void
	333	xfs_bud_item_committing(
	334	struct xfs_log_item *lip,
	335	xfs_lsn_t lsn)
	336	{
	337	}
	338
	339	/*
	340	* This is the ops vector shared by all bud log items.
	341	*/
	342	static const struct xfs_item_ops xfs_bud_item_ops = {
	343	.iop_size = xfs_bud_item_size,
	344	.iop_format = xfs_bud_item_format,
	345	.iop_pin = xfs_bud_item_pin,
	346	.iop_unpin = xfs_bud_item_unpin,
	347	.iop_unlock = xfs_bud_item_unlock,
	348	.iop_committed = xfs_bud_item_committed,
	349	.iop_push = xfs_bud_item_push,
	350	.iop_committing = xfs_bud_item_committing,
	351	};
	352
	353	/*
	354	* Allocate and initialize an bud item with the given number of extents.
	355	*/
	356	struct xfs_bud_log_item *
	357	xfs_bud_init(
	358	struct xfs_mount *mp,
359	struct xfs_bui_log_item *buip)
360
361	{
362	struct xfs_bud_log_item *budp;
363
364	budp = kmem_zone_zalloc(xfs_bud_zone, KM_SLEEP);
365	xfs_log_item_init(mp, &budp->bud_item, XFS_LI_BUD, &xfs_bud_item_ops);
366	budp->bud_buip = buip;
367	budp->bud_format.bud_bui_id = buip->bui_format.bui_id;
368
369	return budp;
370	}
77d61fe4 DW	371
	372	/*
	373	* Process a bmap update intent item that was recovered from the log.
	374	* We need to update some inode's bmbt.
	375	*/
	376	int
	377	xfs_bui_recover(
fbfa977d BF	378	struct xfs_trans *parent_tp,
fbfa977d BF	379	struct xfs_bui_log_item *buip)
77d61fe4 DW	380	{
77d61fe4 DW	381	int error = 0;
9f3afb57	382	unsigned int bui_type;
77d61fe4 DW	383	struct xfs_map_extent *bmap;
	384	xfs_fsblock_t startblock_fsb;
	385	xfs_fsblock_t inode_fsb;
e1a4e37c	386	xfs_filblks_t count;
77d61fe4	387	bool op_ok;
9f3afb57 DW	388	struct xfs_bud_log_item *budp;
	389	enum xfs_bmap_intent_type type;
	390	int whichfork;
	391	xfs_exntst_t state;
	392	struct xfs_trans *tp;
	393	struct xfs_inode *ip = NULL;
e1a4e37c	394	struct xfs_bmbt_irec irec;
fbfa977d	395	struct xfs_mount *mp = parent_tp->t_mountp;
77d61fe4 DW	396
	397	ASSERT(!test_bit(XFS_BUI_RECOVERED, &buip->bui_flags));
	398
	399	/* Only one mapping operation per BUI... */
	400	if (buip->bui_format.bui_nextents != XFS_BUI_MAX_FAST_EXTENTS) {
	401	set_bit(XFS_BUI_RECOVERED, &buip->bui_flags);
	402	xfs_bui_release(buip);
	403	return -EIO;
	404	}
	405
	406	/*
	407	* First check the validity of the extent described by the
	408	* BUI. If anything is bad, then toss the BUI.
	409	*/
	410	bmap = &buip->bui_format.bui_extents[0];
	411	startblock_fsb = XFS_BB_TO_FSB(mp,
	412	XFS_FSB_TO_DADDR(mp, bmap->me_startblock));
	413	inode_fsb = XFS_BB_TO_FSB(mp, XFS_FSB_TO_DADDR(mp,
	414	XFS_INO_TO_FSB(mp, bmap->me_owner)));
	415	switch (bmap->me_flags & XFS_BMAP_EXTENT_TYPE_MASK) {
	416	case XFS_BMAP_MAP:
	417	case XFS_BMAP_UNMAP:
	418	op_ok = true;
	419	break;
	420	default:
	421	op_ok = false;
	422	break;
	423	}
	424	if (!op_ok \|\| startblock_fsb == 0 \|\|
	425	bmap->me_len == 0 \|\|
	426	inode_fsb == 0 \|\|
	427	startblock_fsb >= mp->m_sb.sb_dblocks \|\|
	428	bmap->me_len >= mp->m_sb.sb_agblocks \|\|
	429	inode_fsb >= mp->m_sb.sb_dblocks \|\|
	430	(bmap->me_flags & ~XFS_BMAP_EXTENT_FLAGS)) {
	431	/*
	432	* This will pull the BUI from the AIL and
	433	* free the memory associated with it.
	434	*/
	435	set_bit(XFS_BUI_RECOVERED, &buip->bui_flags);
	436	xfs_bui_release(buip);
	437	return -EIO;
	438	}
	439
fe0be23e DW	440	error = xfs_trans_alloc(mp, &M_RES(mp)->tr_itruncate,
fe0be23e DW	441	XFS_EXTENTADD_SPACE_RES(mp, XFS_DATA_FORK), 0, 0, &tp);
9f3afb57 DW	442	if (error)
9f3afb57 DW	443	return error;
91ef75b6 BF	444	/*
	445	* Recovery stashes all deferred ops during intent processing and
	446	* finishes them on completion. Transfer current dfops state to this
	447	* transaction and transfer the result back before we return.
	448	*/
ce356d64	449	xfs_defer_move(tp, parent_tp);
9f3afb57 DW	450	budp = xfs_trans_get_bud(tp, buip);
	451
	452	/* Grab the inode. */
	453	error = xfs_iget(mp, tp, bmap->me_owner, 0, XFS_ILOCK_EXCL, &ip);
	454	if (error)
	455	goto err_inode;
	456
17c12bcd DW	457	if (VFS_I(ip)->i_nlink == 0)
17c12bcd DW	458	xfs_iflags_set(ip, XFS_IRECOVERY);
9f3afb57 DW	459
	460	/* Process deferred bmap item. */
	461	state = (bmap->me_flags & XFS_BMAP_EXTENT_UNWRITTEN) ?
	462	XFS_EXT_UNWRITTEN : XFS_EXT_NORM;
	463	whichfork = (bmap->me_flags & XFS_BMAP_EXTENT_ATTR_FORK) ?
	464	XFS_ATTR_FORK : XFS_DATA_FORK;
	465	bui_type = bmap->me_flags & XFS_BMAP_EXTENT_TYPE_MASK;
	466	switch (bui_type) {
	467	case XFS_BMAP_MAP:
	468	case XFS_BMAP_UNMAP:
	469	type = bui_type;
	470	break;
	471	default:
	472	error = -EFSCORRUPTED;
50995582	473	goto err_inode;
9f3afb57 DW	474	}
	475	xfs_trans_ijoin(tp, ip, 0);
	476
e1a4e37c	477	count = bmap->me_len;
7dbddbac BF	478	error = xfs_trans_log_finish_bmap_update(tp, budp, type, ip, whichfork,
7dbddbac BF	479	bmap->me_startoff, bmap->me_startblock, &count, state);
9f3afb57	480	if (error)
50995582	481	goto err_inode;
9f3afb57	482
e1a4e37c DW	483	if (count > 0) {
	484	ASSERT(type == XFS_BMAP_UNMAP);
	485	irec.br_startblock = bmap->me_startblock;
	486	irec.br_blockcount = count;
	487	irec.br_startoff = bmap->me_startoff;
	488	irec.br_state = state;
0f37d178	489	error = xfs_bmap_unmap_extent(tp, ip, &irec);
e1a4e37c	490	if (error)
50995582	491	goto err_inode;
e1a4e37c DW	492	}
e1a4e37c DW	493
77d61fe4	494	set_bit(XFS_BUI_RECOVERED, &buip->bui_flags);
ce356d64	495	xfs_defer_move(parent_tp, tp);
9f3afb57 DW	496	error = xfs_trans_commit(tp);
9f3afb57 DW	497	xfs_iunlock(ip, XFS_ILOCK_EXCL);
44a8736b	498	xfs_irele(ip);
9f3afb57 DW	499
	500	return error;
	501
9f3afb57	502	err_inode:
ce356d64	503	xfs_defer_move(parent_tp, tp);
9f3afb57 DW	504	xfs_trans_cancel(tp);
	505	if (ip) {
	506	xfs_iunlock(ip, XFS_ILOCK_EXCL);
44a8736b	507	xfs_irele(ip);
9f3afb57	508	}
77d61fe4 DW	509	return error;
77d61fe4 DW	510	}