[GitHub/mt8127/android_kernel_alcatel_ttab.git] / fs / xfs / xfs_attr_remote.c

/*
 * Copyright (c) 2000-2005 Silicon Graphics, Inc.
 * Copyright (c) 2013 Red Hat, Inc.
 * All Rights Reserved.
 *
 * 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.
 *
 * 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_types.h"
#include "xfs_bit.h"
#include "xfs_log.h"
#include "xfs_trans.h"
#include "xfs_sb.h"
#include "xfs_ag.h"
#include "xfs_mount.h"
#include "xfs_error.h"
#include "xfs_da_btree.h"
#include "xfs_bmap_btree.h"
#include "xfs_dinode.h"
#include "xfs_inode.h"
#include "xfs_alloc.h"
#include "xfs_inode_item.h"
#include "xfs_bmap.h"
#include "xfs_attr.h"
#include "xfs_attr_leaf.h"
#include "xfs_attr_remote.h"
#include "xfs_trans_space.h"
#include "xfs_trace.h"
#include "xfs_cksum.h"
#include "xfs_buf_item.h"

#define ATTR_RMTVALUE_MAPSIZE	1	/* # of map entries at once */

/*
 * Each contiguous block has a header, so it is not just a simple attribute
 * length to FSB conversion.
 */
int
xfs_attr3_rmt_blocks(
	struct xfs_mount *mp,
	int		attrlen)
{
	if (xfs_sb_version_hascrc(&mp->m_sb)) {
		int buflen = XFS_ATTR3_RMT_BUF_SPACE(mp, mp->m_sb.sb_blocksize);
		return (attrlen + buflen - 1) / buflen;
	}
	return XFS_B_TO_FSB(mp, attrlen);
}

/*
 * Checking of the remote attribute header is split into two parts. The verifier
 * does CRC, location and bounds checking, the unpacking function checks the
 * attribute parameters and owner.
 */
static bool
xfs_attr3_rmt_hdr_ok(
	struct xfs_mount	*mp,
	void			*ptr,
	xfs_ino_t		ino,
	uint32_t		offset,
	uint32_t		size,
	xfs_daddr_t		bno)
{
	struct xfs_attr3_rmt_hdr *rmt = ptr;

	if (bno != be64_to_cpu(rmt->rm_blkno))
		return false;
	if (offset != be32_to_cpu(rmt->rm_offset))
		return false;
	if (size != be32_to_cpu(rmt->rm_bytes))
		return false;
	if (ino != be64_to_cpu(rmt->rm_owner))
		return false;

	/* ok */
	return true;
}

static bool
xfs_attr3_rmt_verify(
	struct xfs_mount	*mp,
	void			*ptr,
	int			fsbsize,
	xfs_daddr_t		bno)
{
	struct xfs_attr3_rmt_hdr *rmt = ptr;

	if (!xfs_sb_version_hascrc(&mp->m_sb))
		return false;
	if (rmt->rm_magic != cpu_to_be32(XFS_ATTR3_RMT_MAGIC))
		return false;
	if (!uuid_equal(&rmt->rm_uuid, &mp->m_sb.sb_uuid))
		return false;
	if (be64_to_cpu(rmt->rm_blkno) != bno)
		return false;
	if (be32_to_cpu(rmt->rm_bytes) > fsbsize - sizeof(*rmt))
		return false;
	if (be32_to_cpu(rmt->rm_offset) +
				be32_to_cpu(rmt->rm_bytes) >= XATTR_SIZE_MAX)
		return false;
	if (rmt->rm_owner == 0)
		return false;

	return true;
}

static void
xfs_attr3_rmt_read_verify(
	struct xfs_buf	*bp)
{
	struct xfs_mount *mp = bp->b_target->bt_mount;
	char		*ptr;
	int		len;
	bool		corrupt = false;
	xfs_daddr_t	bno;

	/* no verification of non-crc buffers */
	if (!xfs_sb_version_hascrc(&mp->m_sb))
		return;

	ptr = bp->b_addr;
	bno = bp->b_bn;
	len = BBTOB(bp->b_length);
	ASSERT(len >= XFS_LBSIZE(mp));

	while (len > 0) {
		if (!xfs_verify_cksum(ptr, XFS_LBSIZE(mp),
				      XFS_ATTR3_RMT_CRC_OFF)) {
			corrupt = true;
			break;
		}
		if (!xfs_attr3_rmt_verify(mp, ptr, XFS_LBSIZE(mp), bno)) {
			corrupt = true;
			break;
		}
		len -= XFS_LBSIZE(mp);
		ptr += XFS_LBSIZE(mp);
		bno += mp->m_bsize;
	}

	if (corrupt) {
		XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, bp->b_addr);
		xfs_buf_ioerror(bp, EFSCORRUPTED);
	} else
		ASSERT(len == 0);
}

static void
xfs_attr3_rmt_write_verify(
	struct xfs_buf	*bp)
{
	struct xfs_mount *mp = bp->b_target->bt_mount;
	struct xfs_buf_log_item	*bip = bp->b_fspriv;
	char		*ptr;
	int		len;
	xfs_daddr_t	bno;

	/* no verification of non-crc buffers */
	if (!xfs_sb_version_hascrc(&mp->m_sb))
		return;

	ptr = bp->b_addr;
	bno = bp->b_bn;
	len = BBTOB(bp->b_length);
	ASSERT(len >= XFS_LBSIZE(mp));

	while (len > 0) {
		if (!xfs_attr3_rmt_verify(mp, ptr, XFS_LBSIZE(mp), bno)) {
			XFS_CORRUPTION_ERROR(__func__,
					    XFS_ERRLEVEL_LOW, mp, bp->b_addr);
			xfs_buf_ioerror(bp, EFSCORRUPTED);
			return;
		}
		if (bip) {
			struct xfs_attr3_rmt_hdr *rmt;

			rmt = (struct xfs_attr3_rmt_hdr *)ptr;
			rmt->rm_lsn = cpu_to_be64(bip->bli_item.li_lsn);
		}
		xfs_update_cksum(ptr, XFS_LBSIZE(mp), XFS_ATTR3_RMT_CRC_OFF);

		len -= XFS_LBSIZE(mp);
		ptr += XFS_LBSIZE(mp);
		bno += mp->m_bsize;
	}
	ASSERT(len == 0);
}

const struct xfs_buf_ops xfs_attr3_rmt_buf_ops = {
	.verify_read = xfs_attr3_rmt_read_verify,
	.verify_write = xfs_attr3_rmt_write_verify,
};

STATIC int
xfs_attr3_rmt_hdr_set(
	struct xfs_mount	*mp,
	void			*ptr,
	xfs_ino_t		ino,
	uint32_t		offset,
	uint32_t		size,
	xfs_daddr_t		bno)
{
	struct xfs_attr3_rmt_hdr *rmt = ptr;

	if (!xfs_sb_version_hascrc(&mp->m_sb))
		return 0;

	rmt->rm_magic = cpu_to_be32(XFS_ATTR3_RMT_MAGIC);
	rmt->rm_offset = cpu_to_be32(offset);
	rmt->rm_bytes = cpu_to_be32(size);
	uuid_copy(&rmt->rm_uuid, &mp->m_sb.sb_uuid);
	rmt->rm_owner = cpu_to_be64(ino);
	rmt->rm_blkno = cpu_to_be64(bno);

	return sizeof(struct xfs_attr3_rmt_hdr);
}

/*
 * Helper functions to copy attribute data in and out of the one disk extents
 */
STATIC int
xfs_attr_rmtval_copyout(
	struct xfs_mount *mp,
	struct xfs_buf	*bp,
	xfs_ino_t	ino,
	int		*offset,
	int		*valuelen,
	char		**dst)
{
	char		*src = bp->b_addr;
	xfs_daddr_t	bno = bp->b_bn;
	int		len = BBTOB(bp->b_length);

	ASSERT(len >= XFS_LBSIZE(mp));

	while (len > 0 && *valuelen > 0) {
		int hdr_size = 0;
		int byte_cnt = XFS_ATTR3_RMT_BUF_SPACE(mp, XFS_LBSIZE(mp));

		byte_cnt = min_t(int, *valuelen, byte_cnt);

		if (xfs_sb_version_hascrc(&mp->m_sb)) {
			if (!xfs_attr3_rmt_hdr_ok(mp, src, ino, *offset,
						  byte_cnt, bno)) {
				xfs_alert(mp,
"remote attribute header mismatch bno/off/len/owner (0x%llx/0x%x/Ox%x/0x%llx)",
					bno, *offset, byte_cnt, ino);
				return EFSCORRUPTED;
			}
			hdr_size = sizeof(struct xfs_attr3_rmt_hdr);
		}

		memcpy(*dst, src + hdr_size, byte_cnt);

		/* roll buffer forwards */
		len -= XFS_LBSIZE(mp);
		src += XFS_LBSIZE(mp);
		bno += mp->m_bsize;

		/* roll attribute data forwards */
		*valuelen -= byte_cnt;
		*dst += byte_cnt;
		*offset += byte_cnt;
	}
	return 0;
}

STATIC void
xfs_attr_rmtval_copyin(
	struct xfs_mount *mp,
	struct xfs_buf	*bp,
	xfs_ino_t	ino,
	int		*offset,
	int		*valuelen,
	char		**src)
{
	char		*dst = bp->b_addr;
	xfs_daddr_t	bno = bp->b_bn;
	int		len = BBTOB(bp->b_length);

	ASSERT(len >= XFS_LBSIZE(mp));

	while (len > 0 && *valuelen > 0) {
		int hdr_size;
		int byte_cnt = XFS_ATTR3_RMT_BUF_SPACE(mp, XFS_LBSIZE(mp));

		byte_cnt = min(*valuelen, byte_cnt);
		hdr_size = xfs_attr3_rmt_hdr_set(mp, dst, ino, *offset,
						 byte_cnt, bno);

		memcpy(dst + hdr_size, *src, byte_cnt);

		/*
		 * If this is the last block, zero the remainder of it.
		 * Check that we are actually the last block, too.
		 */
		if (byte_cnt + hdr_size < XFS_LBSIZE(mp)) {
			ASSERT(*valuelen - byte_cnt == 0);
			ASSERT(len == XFS_LBSIZE(mp));
			memset(dst + hdr_size + byte_cnt, 0,
					XFS_LBSIZE(mp) - hdr_size - byte_cnt);
		}

		/* roll buffer forwards */
		len -= XFS_LBSIZE(mp);
		dst += XFS_LBSIZE(mp);
		bno += mp->m_bsize;

		/* roll attribute data forwards */
		*valuelen -= byte_cnt;
		*src += byte_cnt;
		*offset += byte_cnt;
	}
}

/*
 * Read the value associated with an attribute from the out-of-line buffer
 * that we stored it in.
 */
int
xfs_attr_rmtval_get(
	struct xfs_da_args	*args)
{
	struct xfs_bmbt_irec	map[ATTR_RMTVALUE_MAPSIZE];
	struct xfs_mount	*mp = args->dp->i_mount;
	struct xfs_buf		*bp;
	xfs_dablk_t		lblkno = args->rmtblkno;
	char			*dst = args->value;
	int			valuelen = args->valuelen;
	int			nmap;
	int			error;
	int			blkcnt = args->rmtblkcnt;
	int			i;
	int			offset = 0;

	trace_xfs_attr_rmtval_get(args);

	ASSERT(!(args->flags & ATTR_KERNOVAL));

	while (valuelen > 0) {
		nmap = ATTR_RMTVALUE_MAPSIZE;
		error = xfs_bmapi_read(args->dp, (xfs_fileoff_t)lblkno,
				       blkcnt, map, &nmap,
				       XFS_BMAPI_ATTRFORK);
		if (error)
			return error;
		ASSERT(nmap >= 1);

		for (i = 0; (i < nmap) && (valuelen > 0); i++) {
			xfs_daddr_t	dblkno;
			int		dblkcnt;

			ASSERT((map[i].br_startblock != DELAYSTARTBLOCK) &&
			       (map[i].br_startblock != HOLESTARTBLOCK));
			dblkno = XFS_FSB_TO_DADDR(mp, map[i].br_startblock);
			dblkcnt = XFS_FSB_TO_BB(mp, map[i].br_blockcount);
			error = xfs_trans_read_buf(mp, NULL, mp->m_ddev_targp,
						   dblkno, dblkcnt, 0, &bp,
						   &xfs_attr3_rmt_buf_ops);
			if (error)
				return error;

			error = xfs_attr_rmtval_copyout(mp, bp, args->dp->i_ino,
							&offset, &valuelen,
							&dst);
			xfs_buf_relse(bp);
			if (error)
				return error;

			/* roll attribute extent map forwards */
			lblkno += map[i].br_blockcount;
			blkcnt -= map[i].br_blockcount;
		}
	}
	ASSERT(valuelen == 0);
	return 0;
}

/*
 * Write the value associated with an attribute into the out-of-line buffer
 * that we have defined for it.
 */
int
xfs_attr_rmtval_set(
	struct xfs_da_args	*args)
{
	struct xfs_inode	*dp = args->dp;
	struct xfs_mount	*mp = dp->i_mount;
	struct xfs_bmbt_irec	map;
	xfs_dablk_t		lblkno;
	xfs_fileoff_t		lfileoff = 0;
	char			*src = args->value;
	int			blkcnt;
	int			valuelen;
	int			nmap;
	int			error;
	int			offset = 0;

	trace_xfs_attr_rmtval_set(args);

	/*
	 * Find a "hole" in the attribute address space large enough for
	 * us to drop the new attribute's value into. Because CRC enable
	 * attributes have headers, we can't just do a straight byte to FSB
	 * conversion and have to take the header space into account.
	 */
	blkcnt = xfs_attr3_rmt_blocks(mp, args->valuelen);
	error = xfs_bmap_first_unused(args->trans, args->dp, blkcnt, &lfileoff,
						   XFS_ATTR_FORK);
	if (error)
		return error;

	args->rmtblkno = lblkno = (xfs_dablk_t)lfileoff;
	args->rmtblkcnt = blkcnt;

	/*
	 * Roll through the "value", allocating blocks on disk as required.
	 */
	while (blkcnt > 0) {
		int	committed;

		/*
		 * Allocate a single extent, up to the size of the value.
		 */
		xfs_bmap_init(args->flist, args->firstblock);
		nmap = 1;
		error = xfs_bmapi_write(args->trans, dp, (xfs_fileoff_t)lblkno,
				  blkcnt,
				  XFS_BMAPI_ATTRFORK | XFS_BMAPI_METADATA,
				  args->firstblock, args->total, &map, &nmap,
				  args->flist);
		if (!error) {
			error = xfs_bmap_finish(&args->trans, args->flist,
						&committed);
		}
		if (error) {
			ASSERT(committed);
			args->trans = NULL;
			xfs_bmap_cancel(args->flist);
			return(error);
		}

		/*
		 * bmap_finish() may have committed the last trans and started
		 * a new one.  We need the inode to be in all transactions.
		 */
		if (committed)
			xfs_trans_ijoin(args->trans, dp, 0);

		ASSERT(nmap == 1);
		ASSERT((map.br_startblock != DELAYSTARTBLOCK) &&
		       (map.br_startblock != HOLESTARTBLOCK));
		lblkno += map.br_blockcount;
		blkcnt -= map.br_blockcount;

		/*
		 * Start the next trans in the chain.
		 */
		error = xfs_trans_roll(&args->trans, dp);
		if (error)
			return (error);
	}

	/*
	 * Roll through the "value", copying the attribute value to the
	 * already-allocated blocks.  Blocks are written synchronously
	 * so that we can know they are all on disk before we turn off
	 * the INCOMPLETE flag.
	 */
	lblkno = args->rmtblkno;
	blkcnt = args->rmtblkcnt;
	valuelen = args->valuelen;
	while (valuelen > 0) {
		struct xfs_buf	*bp;
		xfs_daddr_t	dblkno;
		int		dblkcnt;

		ASSERT(blkcnt > 0);

		xfs_bmap_init(args->flist, args->firstblock);
		nmap = 1;
		error = xfs_bmapi_read(dp, (xfs_fileoff_t)lblkno,
				       blkcnt, &map, &nmap,
				       XFS_BMAPI_ATTRFORK);
		if (error)
			return(error);
		ASSERT(nmap == 1);
		ASSERT((map.br_startblock != DELAYSTARTBLOCK) &&
		       (map.br_startblock != HOLESTARTBLOCK));

		dblkno = XFS_FSB_TO_DADDR(mp, map.br_startblock),
		dblkcnt = XFS_FSB_TO_BB(mp, map.br_blockcount);

		bp = xfs_buf_get(mp->m_ddev_targp, dblkno, dblkcnt, 0);
		if (!bp)
			return ENOMEM;
		bp->b_ops = &xfs_attr3_rmt_buf_ops;

		xfs_attr_rmtval_copyin(mp, bp, args->dp->i_ino, &offset,
				       &valuelen, &src);

		error = xfs_bwrite(bp);	/* GROT: NOTE: synchronous write */
		xfs_buf_relse(bp);
		if (error)
			return error;


		/* roll attribute extent map forwards */
		lblkno += map.br_blockcount;
		blkcnt -= map.br_blockcount;
	}
	ASSERT(valuelen == 0);
	return 0;
}

/*
 * Remove the value associated with an attribute by deleting the
 * out-of-line buffer that it is stored on.
 */
int
xfs_attr_rmtval_remove(
	struct xfs_da_args	*args)
{
	struct xfs_mount	*mp = args->dp->i_mount;
	xfs_dablk_t		lblkno;
	int			blkcnt;
	int			error;
	int			done;

	trace_xfs_attr_rmtval_remove(args);

	/*
	 * Roll through the "value", invalidating the attribute value's blocks.
	 * Note that args->rmtblkcnt is the minimum number of data blocks we'll
	 * see for a CRC enabled remote attribute. Each extent will have a
	 * header, and so we may have more blocks than we realise here.  If we
	 * fail to map the blocks correctly, we'll have problems with the buffer
	 * lookups.
	 */
	lblkno = args->rmtblkno;
	blkcnt = args->rmtblkcnt;
	while (blkcnt > 0) {
		struct xfs_bmbt_irec	map;
		struct xfs_buf		*bp;
		xfs_daddr_t		dblkno;
		int			dblkcnt;
		int			nmap;

		/*
		 * Try to remember where we decided to put the value.
		 */
		nmap = 1;
		error = xfs_bmapi_read(args->dp, (xfs_fileoff_t)lblkno,
				       blkcnt, &map, &nmap, XFS_BMAPI_ATTRFORK);
		if (error)
			return(error);
		ASSERT(nmap == 1);
		ASSERT((map.br_startblock != DELAYSTARTBLOCK) &&
		       (map.br_startblock != HOLESTARTBLOCK));

		dblkno = XFS_FSB_TO_DADDR(mp, map.br_startblock),
		dblkcnt = XFS_FSB_TO_BB(mp, map.br_blockcount);

		/*
		 * If the "remote" value is in the cache, remove it.
		 */
		bp = xfs_incore(mp->m_ddev_targp, dblkno, dblkcnt, XBF_TRYLOCK);
		if (bp) {
			xfs_buf_stale(bp);
			xfs_buf_relse(bp);
			bp = NULL;
		}

		lblkno += map.br_blockcount;
		blkcnt -= map.br_blockcount;
	}

	/*
	 * Keep de-allocating extents until the remote-value region is gone.
	 */
	lblkno = args->rmtblkno;
	blkcnt = args->rmtblkcnt;
	done = 0;
	while (!done) {
		int committed;

		xfs_bmap_init(args->flist, args->firstblock);
		error = xfs_bunmapi(args->trans, args->dp, lblkno, blkcnt,
				    XFS_BMAPI_ATTRFORK | XFS_BMAPI_METADATA,
				    1, args->firstblock, args->flist,
				    &done);
		if (!error) {
			error = xfs_bmap_finish(&args->trans, args->flist,
						&committed);
		}
		if (error) {
			ASSERT(committed);
			args->trans = NULL;
			xfs_bmap_cancel(args->flist);
			return error;
		}

		/*
		 * bmap_finish() may have committed the last trans and started
		 * a new one.  We need the inode to be in all transactions.
		 */
		if (committed)
			xfs_trans_ijoin(args->trans, args->dp, 0);

		/*
		 * Close out trans and start the next one in the chain.
		 */
		error = xfs_trans_roll(&args->trans, args->dp);
		if (error)
			return (error);
	}
	return(0);
}
Commit	Line	Data
95920cd6 DC	1	/*
95920cd6 DC	2	* Copyright (c) 2000-2005 Silicon Graphics, Inc.
d2e448d5	3	* Copyright (c) 2013 Red Hat, Inc.
95920cd6 DC	4	* All Rights Reserved.
	5	*
	6	* This program is free software; you can redistribute it and/or
	7	* modify it under the terms of the GNU General Public License as
	8	* published by the Free Software Foundation.
	9	*
	10	* This program is distributed in the hope that it would be useful,
	11	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	12	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	13	* GNU General Public License for more details.
	14	*
	15	* You should have received a copy of the GNU General Public License
	16	* along with this program; if not, write the Free Software Foundation,
	17	* Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
	18	*/
	19	#include "xfs.h"
	20	#include "xfs_fs.h"
	21	#include "xfs_types.h"
	22	#include "xfs_bit.h"
	23	#include "xfs_log.h"
	24	#include "xfs_trans.h"
	25	#include "xfs_sb.h"
	26	#include "xfs_ag.h"
	27	#include "xfs_mount.h"
	28	#include "xfs_error.h"
	29	#include "xfs_da_btree.h"
	30	#include "xfs_bmap_btree.h"
	31	#include "xfs_dinode.h"
	32	#include "xfs_inode.h"
	33	#include "xfs_alloc.h"
	34	#include "xfs_inode_item.h"
	35	#include "xfs_bmap.h"
	36	#include "xfs_attr.h"
	37	#include "xfs_attr_leaf.h"
	38	#include "xfs_attr_remote.h"
	39	#include "xfs_trans_space.h"
	40	#include "xfs_trace.h"
d2e448d5 DC	41	#include "xfs_cksum.h"
d2e448d5 DC	42	#include "xfs_buf_item.h"
95920cd6 DC	43
	44	#define ATTR_RMTVALUE_MAPSIZE 1 /* # of map entries at once */
	45
d2e448d5 DC	46	/*
	47	* Each contiguous block has a header, so it is not just a simple attribute
	48	* length to FSB conversion.
	49	*/
7bc0dc27	50	int
d2e448d5 DC	51	xfs_attr3_rmt_blocks(
	52	struct xfs_mount *mp,
	53	int attrlen)
	54	{
551b382f DC	55	if (xfs_sb_version_hascrc(&mp->m_sb)) {
	56	int buflen = XFS_ATTR3_RMT_BUF_SPACE(mp, mp->m_sb.sb_blocksize);
	57	return (attrlen + buflen - 1) / buflen;
	58	}
	59	return XFS_B_TO_FSB(mp, attrlen);
d2e448d5 DC	60	}
d2e448d5 DC	61
7bc0dc27 DC	62	/*
	63	* Checking of the remote attribute header is split into two parts. The verifier
	64	* does CRC, location and bounds checking, the unpacking function checks the
	65	* attribute parameters and owner.
	66	*/
	67	static bool
	68	xfs_attr3_rmt_hdr_ok(
	69	struct xfs_mount *mp,
	70	void *ptr,
	71	xfs_ino_t ino,
	72	uint32_t offset,
	73	uint32_t size,
	74	xfs_daddr_t bno)
	75	{
	76	struct xfs_attr3_rmt_hdr *rmt = ptr;
	77
	78	if (bno != be64_to_cpu(rmt->rm_blkno))
	79	return false;
	80	if (offset != be32_to_cpu(rmt->rm_offset))
	81	return false;
	82	if (size != be32_to_cpu(rmt->rm_bytes))
	83	return false;
	84	if (ino != be64_to_cpu(rmt->rm_owner))
	85	return false;
	86
	87	/* ok */
	88	return true;
	89	}
	90
d2e448d5 DC	91	static bool
d2e448d5 DC	92	xfs_attr3_rmt_verify(
7bc0dc27 DC	93	struct xfs_mount *mp,
	94	void *ptr,
	95	int fsbsize,
	96	xfs_daddr_t bno)
d2e448d5	97	{
7bc0dc27	98	struct xfs_attr3_rmt_hdr *rmt = ptr;
d2e448d5 DC	99
	100	if (!xfs_sb_version_hascrc(&mp->m_sb))
	101	return false;
	102	if (rmt->rm_magic != cpu_to_be32(XFS_ATTR3_RMT_MAGIC))
	103	return false;
	104	if (!uuid_equal(&rmt->rm_uuid, &mp->m_sb.sb_uuid))
	105	return false;
7bc0dc27 DC	106	if (be64_to_cpu(rmt->rm_blkno) != bno)
	107	return false;
	108	if (be32_to_cpu(rmt->rm_bytes) > fsbsize - sizeof(*rmt))
d2e448d5 DC	109	return false;
d2e448d5 DC	110	if (be32_to_cpu(rmt->rm_offset) +
946217ba	111	be32_to_cpu(rmt->rm_bytes) >= XATTR_SIZE_MAX)
d2e448d5 DC	112	return false;
	113	if (rmt->rm_owner == 0)
	114	return false;
	115
	116	return true;
	117	}
	118
	119	static void
	120	xfs_attr3_rmt_read_verify(
	121	struct xfs_buf *bp)
	122	{
	123	struct xfs_mount *mp = bp->b_target->bt_mount;
7bc0dc27 DC	124	char *ptr;
	125	int len;
	126	bool corrupt = false;
	127	xfs_daddr_t bno;
d2e448d5 DC	128
	129	/* no verification of non-crc buffers */
	130	if (!xfs_sb_version_hascrc(&mp->m_sb))
	131	return;
	132
7bc0dc27 DC	133	ptr = bp->b_addr;
	134	bno = bp->b_bn;
	135	len = BBTOB(bp->b_length);
	136	ASSERT(len >= XFS_LBSIZE(mp));
	137
	138	while (len > 0) {
	139	if (!xfs_verify_cksum(ptr, XFS_LBSIZE(mp),
	140	XFS_ATTR3_RMT_CRC_OFF)) {
	141	corrupt = true;
	142	break;
	143	}
	144	if (!xfs_attr3_rmt_verify(mp, ptr, XFS_LBSIZE(mp), bno)) {
	145	corrupt = true;
	146	break;
	147	}
	148	len -= XFS_LBSIZE(mp);
	149	ptr += XFS_LBSIZE(mp);
	150	bno += mp->m_bsize;
	151	}
	152
	153	if (corrupt) {
d2e448d5 DC	154	XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, bp->b_addr);
d2e448d5 DC	155	xfs_buf_ioerror(bp, EFSCORRUPTED);
7bc0dc27 DC	156	} else
7bc0dc27 DC	157	ASSERT(len == 0);
d2e448d5 DC	158	}
	159
	160	static void
	161	xfs_attr3_rmt_write_verify(
	162	struct xfs_buf *bp)
	163	{
	164	struct xfs_mount *mp = bp->b_target->bt_mount;
	165	struct xfs_buf_log_item *bip = bp->b_fspriv;
7bc0dc27 DC	166	char *ptr;
	167	int len;
	168	xfs_daddr_t bno;
d2e448d5 DC	169
	170	/* no verification of non-crc buffers */
	171	if (!xfs_sb_version_hascrc(&mp->m_sb))
	172	return;
	173
7bc0dc27 DC	174	ptr = bp->b_addr;
	175	bno = bp->b_bn;
	176	len = BBTOB(bp->b_length);
	177	ASSERT(len >= XFS_LBSIZE(mp));
	178
	179	while (len > 0) {
	180	if (!xfs_attr3_rmt_verify(mp, ptr, XFS_LBSIZE(mp), bno)) {
	181	XFS_CORRUPTION_ERROR(__func__,
	182	XFS_ERRLEVEL_LOW, mp, bp->b_addr);
	183	xfs_buf_ioerror(bp, EFSCORRUPTED);
	184	return;
	185	}
	186	if (bip) {
	187	struct xfs_attr3_rmt_hdr *rmt;
d2e448d5	188
7bc0dc27 DC	189	rmt = (struct xfs_attr3_rmt_hdr *)ptr;
	190	rmt->rm_lsn = cpu_to_be64(bip->bli_item.li_lsn);
	191	}
	192	xfs_update_cksum(ptr, XFS_LBSIZE(mp), XFS_ATTR3_RMT_CRC_OFF);
	193
	194	len -= XFS_LBSIZE(mp);
	195	ptr += XFS_LBSIZE(mp);
	196	bno += mp->m_bsize;
d2e448d5	197	}
7bc0dc27	198	ASSERT(len == 0);
d2e448d5 DC	199	}
	200
	201	const struct xfs_buf_ops xfs_attr3_rmt_buf_ops = {
	202	.verify_read = xfs_attr3_rmt_read_verify,
	203	.verify_write = xfs_attr3_rmt_write_verify,
	204	};
	205
7bc0dc27	206	STATIC int
d2e448d5 DC	207	xfs_attr3_rmt_hdr_set(
d2e448d5 DC	208	struct xfs_mount *mp,
7bc0dc27	209	void *ptr,
d2e448d5 DC	210	xfs_ino_t ino,
	211	uint32_t offset,
	212	uint32_t size,
7bc0dc27	213	xfs_daddr_t bno)
d2e448d5	214	{
7bc0dc27	215	struct xfs_attr3_rmt_hdr *rmt = ptr;
d2e448d5 DC	216
	217	if (!xfs_sb_version_hascrc(&mp->m_sb))
	218	return 0;
	219
	220	rmt->rm_magic = cpu_to_be32(XFS_ATTR3_RMT_MAGIC);
	221	rmt->rm_offset = cpu_to_be32(offset);
	222	rmt->rm_bytes = cpu_to_be32(size);
	223	uuid_copy(&rmt->rm_uuid, &mp->m_sb.sb_uuid);
	224	rmt->rm_owner = cpu_to_be64(ino);
7bc0dc27	225	rmt->rm_blkno = cpu_to_be64(bno);
d2e448d5 DC	226
	227	return sizeof(struct xfs_attr3_rmt_hdr);
	228	}
	229
	230	/*
7bc0dc27	231	* Helper functions to copy attribute data in and out of the one disk extents
d2e448d5	232	*/
7bc0dc27 DC	233	STATIC int
	234	xfs_attr_rmtval_copyout(
	235	struct xfs_mount *mp,
	236	struct xfs_buf *bp,
	237	xfs_ino_t ino,
	238	int *offset,
	239	int *valuelen,
	240	char **dst)
d2e448d5	241	{
7bc0dc27 DC	242	char *src = bp->b_addr;
	243	xfs_daddr_t bno = bp->b_bn;
	244	int len = BBTOB(bp->b_length);
d2e448d5	245
7bc0dc27	246	ASSERT(len >= XFS_LBSIZE(mp));
d2e448d5	247
7bc0dc27 DC	248	while (len > 0 && *valuelen > 0) {
	249	int hdr_size = 0;
	250	int byte_cnt = XFS_ATTR3_RMT_BUF_SPACE(mp, XFS_LBSIZE(mp));
	251
	252	byte_cnt = min_t(int, *valuelen, byte_cnt);
	253
	254	if (xfs_sb_version_hascrc(&mp->m_sb)) {
	255	if (!xfs_attr3_rmt_hdr_ok(mp, src, ino, *offset,
	256	byte_cnt, bno)) {
	257	xfs_alert(mp,
	258	"remote attribute header mismatch bno/off/len/owner (0x%llx/0x%x/Ox%x/0x%llx)",
	259	bno, *offset, byte_cnt, ino);
	260	return EFSCORRUPTED;
	261	}
	262	hdr_size = sizeof(struct xfs_attr3_rmt_hdr);
	263	}
	264
	265	memcpy(*dst, src + hdr_size, byte_cnt);
	266
	267	/* roll buffer forwards */
	268	len -= XFS_LBSIZE(mp);
	269	src += XFS_LBSIZE(mp);
	270	bno += mp->m_bsize;
	271
	272	/* roll attribute data forwards */
	273	*valuelen -= byte_cnt;
	274	*dst += byte_cnt;
	275	*offset += byte_cnt;
	276	}
	277	return 0;
	278	}
	279
	280	STATIC void
	281	xfs_attr_rmtval_copyin(
	282	struct xfs_mount *mp,
	283	struct xfs_buf *bp,
	284	xfs_ino_t ino,
	285	int *offset,
	286	int *valuelen,
	287	char **src)
	288	{
	289	char *dst = bp->b_addr;
	290	xfs_daddr_t bno = bp->b_bn;
	291	int len = BBTOB(bp->b_length);
	292
	293	ASSERT(len >= XFS_LBSIZE(mp));
	294
	295	while (len > 0 && *valuelen > 0) {
	296	int hdr_size;
	297	int byte_cnt = XFS_ATTR3_RMT_BUF_SPACE(mp, XFS_LBSIZE(mp));
	298
	299	byte_cnt = min(*valuelen, byte_cnt);
	300	hdr_size = xfs_attr3_rmt_hdr_set(mp, dst, ino, *offset,
	301	byte_cnt, bno);
	302
	303	memcpy(dst + hdr_size, *src, byte_cnt);
	304
	305	/*
	306	* If this is the last block, zero the remainder of it.
	307	* Check that we are actually the last block, too.
	308	*/
	309	if (byte_cnt + hdr_size < XFS_LBSIZE(mp)) {
	310	ASSERT(*valuelen - byte_cnt == 0);
	311	ASSERT(len == XFS_LBSIZE(mp));
312	memset(dst + hdr_size + byte_cnt, 0,
313	XFS_LBSIZE(mp) - hdr_size - byte_cnt);
314	}
315
316	/* roll buffer forwards */
317	len -= XFS_LBSIZE(mp);
318	dst += XFS_LBSIZE(mp);
319	bno += mp->m_bsize;
320
321	/* roll attribute data forwards */
322	*valuelen -= byte_cnt;
323	*src += byte_cnt;
324	*offset += byte_cnt;
325	}
d2e448d5 DC	326	}
d2e448d5 DC	327
95920cd6 DC	328	/*
	329	* Read the value associated with an attribute from the out-of-line buffer
	330	* that we stored it in.
	331	*/
	332	int
d2e448d5 DC	333	xfs_attr_rmtval_get(
d2e448d5 DC	334	struct xfs_da_args *args)
95920cd6	335	{
d2e448d5 DC	336	struct xfs_bmbt_irec map[ATTR_RMTVALUE_MAPSIZE];
	337	struct xfs_mount *mp = args->dp->i_mount;
	338	struct xfs_buf *bp;
d2e448d5	339	xfs_dablk_t lblkno = args->rmtblkno;
7bc0dc27	340	char *dst = args->value;
d2e448d5 DC	341	int valuelen = args->valuelen;
	342	int nmap;
	343	int error;
7bc0dc27	344	int blkcnt = args->rmtblkcnt;
d2e448d5 DC	345	int i;
d2e448d5 DC	346	int offset = 0;
95920cd6 DC	347
	348	trace_xfs_attr_rmtval_get(args);
	349
	350	ASSERT(!(args->flags & ATTR_KERNOVAL));
	351
95920cd6 DC	352	while (valuelen > 0) {
	353	nmap = ATTR_RMTVALUE_MAPSIZE;
	354	error = xfs_bmapi_read(args->dp, (xfs_fileoff_t)lblkno,
551b382f	355	blkcnt, map, &nmap,
95920cd6 DC	356	XFS_BMAPI_ATTRFORK);
95920cd6 DC	357	if (error)
d2e448d5	358	return error;
95920cd6 DC	359	ASSERT(nmap >= 1);
	360
	361	for (i = 0; (i < nmap) && (valuelen > 0); i++) {
7bc0dc27 DC	362	xfs_daddr_t dblkno;
7bc0dc27 DC	363	int dblkcnt;
d2e448d5	364
95920cd6 DC	365	ASSERT((map[i].br_startblock != DELAYSTARTBLOCK) &&
	366	(map[i].br_startblock != HOLESTARTBLOCK));
	367	dblkno = XFS_FSB_TO_DADDR(mp, map[i].br_startblock);
7bc0dc27	368	dblkcnt = XFS_FSB_TO_BB(mp, map[i].br_blockcount);
95920cd6	369	error = xfs_trans_read_buf(mp, NULL, mp->m_ddev_targp,
7bc0dc27	370	dblkno, dblkcnt, 0, &bp,
d2e448d5	371	&xfs_attr3_rmt_buf_ops);
95920cd6	372	if (error)
d2e448d5 DC	373	return error;
d2e448d5 DC	374
7bc0dc27 DC	375	error = xfs_attr_rmtval_copyout(mp, bp, args->dp->i_ino,
	376	&offset, &valuelen,
	377	&dst);
95920cd6	378	xfs_buf_relse(bp);
7bc0dc27 DC	379	if (error)
7bc0dc27 DC	380	return error;
d2e448d5	381
7bc0dc27	382	/* roll attribute extent map forwards */
95920cd6	383	lblkno += map[i].br_blockcount;
7bc0dc27	384	blkcnt -= map[i].br_blockcount;
95920cd6 DC	385	}
	386	}
	387	ASSERT(valuelen == 0);
d2e448d5	388	return 0;
95920cd6 DC	389	}
	390
	391	/*
	392	* Write the value associated with an attribute into the out-of-line buffer
	393	* that we have defined for it.
	394	*/
	395	int
d2e448d5 DC	396	xfs_attr_rmtval_set(
d2e448d5 DC	397	struct xfs_da_args *args)
95920cd6	398	{
d2e448d5 DC	399	struct xfs_inode *dp = args->dp;
	400	struct xfs_mount *mp = dp->i_mount;
	401	struct xfs_bmbt_irec map;
d2e448d5 DC	402	xfs_dablk_t lblkno;
d2e448d5 DC	403	xfs_fileoff_t lfileoff = 0;
7bc0dc27	404	char *src = args->value;
d2e448d5 DC	405	int blkcnt;
	406	int valuelen;
	407	int nmap;
	408	int error;
d2e448d5	409	int offset = 0;
95920cd6 DC	410
	411	trace_xfs_attr_rmtval_set(args);
	412
95920cd6 DC	413	/*
95920cd6 DC	414	* Find a "hole" in the attribute address space large enough for
d2e448d5 DC	415	* us to drop the new attribute's value into. Because CRC enable
d2e448d5 DC	416	* attributes have headers, we can't just do a straight byte to FSB
7bc0dc27	417	* conversion and have to take the header space into account.
95920cd6	418	*/
26f71445	419	blkcnt = xfs_attr3_rmt_blocks(mp, args->valuelen);
95920cd6 DC	420	error = xfs_bmap_first_unused(args->trans, args->dp, blkcnt, &lfileoff,
95920cd6 DC	421	XFS_ATTR_FORK);
d2e448d5 DC	422	if (error)
	423	return error;
	424
95920cd6 DC	425	args->rmtblkno = lblkno = (xfs_dablk_t)lfileoff;
	426	args->rmtblkcnt = blkcnt;
	427
	428	/*
	429	* Roll through the "value", allocating blocks on disk as required.
	430	*/
	431	while (blkcnt > 0) {
d2e448d5 DC	432	int committed;
d2e448d5 DC	433
95920cd6 DC	434	/*
	435	* Allocate a single extent, up to the size of the value.
	436	*/
	437	xfs_bmap_init(args->flist, args->firstblock);
	438	nmap = 1;
	439	error = xfs_bmapi_write(args->trans, dp, (xfs_fileoff_t)lblkno,
	440	blkcnt,
	441	XFS_BMAPI_ATTRFORK \| XFS_BMAPI_METADATA,
	442	args->firstblock, args->total, &map, &nmap,
	443	args->flist);
	444	if (!error) {
	445	error = xfs_bmap_finish(&args->trans, args->flist,
	446	&committed);
	447	}
	448	if (error) {
	449	ASSERT(committed);
	450	args->trans = NULL;
	451	xfs_bmap_cancel(args->flist);
	452	return(error);
	453	}
	454
	455	/*
	456	* bmap_finish() may have committed the last trans and started
	457	* a new one. We need the inode to be in all transactions.
	458	*/
	459	if (committed)
	460	xfs_trans_ijoin(args->trans, dp, 0);
	461
	462	ASSERT(nmap == 1);
	463	ASSERT((map.br_startblock != DELAYSTARTBLOCK) &&
	464	(map.br_startblock != HOLESTARTBLOCK));
	465	lblkno += map.br_blockcount;
	466	blkcnt -= map.br_blockcount;
	467
	468	/*
	469	* Start the next trans in the chain.
	470	*/
	471	error = xfs_trans_roll(&args->trans, dp);
	472	if (error)
	473	return (error);
	474	}
	475
	476	/*
	477	* Roll through the "value", copying the attribute value to the
	478	* already-allocated blocks. Blocks are written synchronously
	479	* so that we can know they are all on disk before we turn off
	480	* the INCOMPLETE flag.
	481	*/
	482	lblkno = args->rmtblkno;
26f71445	483	blkcnt = args->rmtblkcnt;
7bc0dc27	484	valuelen = args->valuelen;
95920cd6	485	while (valuelen > 0) {
7bc0dc27 DC	486	struct xfs_buf *bp;
	487	xfs_daddr_t dblkno;
	488	int dblkcnt;
	489
	490	ASSERT(blkcnt > 0);
95920cd6	491
95920cd6 DC	492	xfs_bmap_init(args->flist, args->firstblock);
	493	nmap = 1;
	494	error = xfs_bmapi_read(dp, (xfs_fileoff_t)lblkno,
26f71445	495	blkcnt, &map, &nmap,
95920cd6 DC	496	XFS_BMAPI_ATTRFORK);
	497	if (error)
	498	return(error);
	499	ASSERT(nmap == 1);
	500	ASSERT((map.br_startblock != DELAYSTARTBLOCK) &&
	501	(map.br_startblock != HOLESTARTBLOCK));
	502
	503	dblkno = XFS_FSB_TO_DADDR(mp, map.br_startblock),
26f71445	504	dblkcnt = XFS_FSB_TO_BB(mp, map.br_blockcount);
95920cd6	505
26f71445	506	bp = xfs_buf_get(mp->m_ddev_targp, dblkno, dblkcnt, 0);
95920cd6 DC	507	if (!bp)
95920cd6 DC	508	return ENOMEM;
d2e448d5	509	bp->b_ops = &xfs_attr3_rmt_buf_ops;
26f71445	510
7bc0dc27 DC	511	xfs_attr_rmtval_copyin(mp, bp, args->dp->i_ino, &offset,
7bc0dc27 DC	512	&valuelen, &src);
95920cd6 DC	513
	514	error = xfs_bwrite(bp); /* GROT: NOTE: synchronous write */
	515	xfs_buf_relse(bp);
	516	if (error)
	517	return error;
d2e448d5	518
95920cd6	519
7bc0dc27	520	/* roll attribute extent map forwards */
95920cd6	521	lblkno += map.br_blockcount;
26f71445	522	blkcnt -= map.br_blockcount;
95920cd6 DC	523	}
95920cd6 DC	524	ASSERT(valuelen == 0);
d2e448d5	525	return 0;
95920cd6 DC	526	}
	527
	528	/*
	529	* Remove the value associated with an attribute by deleting the
	530	* out-of-line buffer that it is stored on.
	531	*/
	532	int
7bc0dc27 DC	533	xfs_attr_rmtval_remove(
7bc0dc27 DC	534	struct xfs_da_args *args)
95920cd6	535	{
7bc0dc27 DC	536	struct xfs_mount *mp = args->dp->i_mount;
	537	xfs_dablk_t lblkno;
	538	int blkcnt;
	539	int error;
	540	int done;
95920cd6 DC	541
	542	trace_xfs_attr_rmtval_remove(args);
	543
95920cd6	544	/*
58a72281 DC	545	* Roll through the "value", invalidating the attribute value's blocks.
	546	* Note that args->rmtblkcnt is the minimum number of data blocks we'll
	547	* see for a CRC enabled remote attribute. Each extent will have a
	548	* header, and so we may have more blocks than we realise here. If we
	549	* fail to map the blocks correctly, we'll have problems with the buffer
	550	* lookups.
95920cd6 DC	551	*/
95920cd6 DC	552	lblkno = args->rmtblkno;
7bc0dc27 DC	553	blkcnt = args->rmtblkcnt;
	554	while (blkcnt > 0) {
	555	struct xfs_bmbt_irec map;
	556	struct xfs_buf *bp;
	557	xfs_daddr_t dblkno;
	558	int dblkcnt;
	559	int nmap;
58a72281	560
95920cd6 DC	561	/*
	562	* Try to remember where we decided to put the value.
	563	*/
	564	nmap = 1;
	565	error = xfs_bmapi_read(args->dp, (xfs_fileoff_t)lblkno,
58a72281	566	blkcnt, &map, &nmap, XFS_BMAPI_ATTRFORK);
95920cd6 DC	567	if (error)
	568	return(error);
	569	ASSERT(nmap == 1);
	570	ASSERT((map.br_startblock != DELAYSTARTBLOCK) &&
	571	(map.br_startblock != HOLESTARTBLOCK));
	572
	573	dblkno = XFS_FSB_TO_DADDR(mp, map.br_startblock),
58a72281	574	dblkcnt = XFS_FSB_TO_BB(mp, map.br_blockcount);
95920cd6 DC	575
	576	/*
	577	* If the "remote" value is in the cache, remove it.
	578	*/
58a72281	579	bp = xfs_incore(mp->m_ddev_targp, dblkno, dblkcnt, XBF_TRYLOCK);
95920cd6 DC	580	if (bp) {
	581	xfs_buf_stale(bp);
	582	xfs_buf_relse(bp);
	583	bp = NULL;
	584	}
	585
95920cd6	586	lblkno += map.br_blockcount;
58a72281	587	blkcnt -= map.br_blockcount;
95920cd6 DC	588	}
	589
	590	/*
	591	* Keep de-allocating extents until the remote-value region is gone.
	592	*/
	593	lblkno = args->rmtblkno;
7bc0dc27	594	blkcnt = args->rmtblkcnt;
95920cd6 DC	595	done = 0;
95920cd6 DC	596	while (!done) {
7bc0dc27 DC	597	int committed;
7bc0dc27 DC	598
95920cd6 DC	599	xfs_bmap_init(args->flist, args->firstblock);
	600	error = xfs_bunmapi(args->trans, args->dp, lblkno, blkcnt,
	601	XFS_BMAPI_ATTRFORK \| XFS_BMAPI_METADATA,
	602	1, args->firstblock, args->flist,
	603	&done);
	604	if (!error) {
	605	error = xfs_bmap_finish(&args->trans, args->flist,
	606	&committed);
	607	}
	608	if (error) {
	609	ASSERT(committed);
	610	args->trans = NULL;
	611	xfs_bmap_cancel(args->flist);
d2e448d5	612	return error;
95920cd6 DC	613	}
	614
	615	/*
	616	* bmap_finish() may have committed the last trans and started
	617	* a new one. We need the inode to be in all transactions.
	618	*/
	619	if (committed)
	620	xfs_trans_ijoin(args->trans, args->dp, 0);
	621
	622	/*
	623	* Close out trans and start the next one in the chain.
	624	*/
	625	error = xfs_trans_roll(&args->trans, args->dp);
	626	if (error)
	627	return (error);
	628	}
	629	return(0);
	630	}
	631