[mirror_ubuntu-bionic-kernel.git] / fs / xfs / xfs_attr_list.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_format.h"
#include "xfs_log_format.h"
#include "xfs_trans_resv.h"
#include "xfs_bit.h"
#include "xfs_mount.h"
#include "xfs_da_format.h"
#include "xfs_da_btree.h"
#include "xfs_inode.h"
#include "xfs_trans.h"
#include "xfs_inode_item.h"
#include "xfs_bmap.h"
#include "xfs_attr.h"
#include "xfs_attr_sf.h"
#include "xfs_attr_remote.h"
#include "xfs_attr_leaf.h"
#include "xfs_error.h"
#include "xfs_trace.h"
#include "xfs_buf_item.h"
#include "xfs_cksum.h"
#include "xfs_dir2.h"

STATIC int
xfs_attr_shortform_compare(const void *a, const void *b)
{
	xfs_attr_sf_sort_t *sa, *sb;

	sa = (xfs_attr_sf_sort_t *)a;
	sb = (xfs_attr_sf_sort_t *)b;
	if (sa->hash < sb->hash) {
		return -1;
	} else if (sa->hash > sb->hash) {
		return 1;
	} else {
		return sa->entno - sb->entno;
	}
}

#define XFS_ISRESET_CURSOR(cursor) \
	(!((cursor)->initted) && !((cursor)->hashval) && \
	 !((cursor)->blkno) && !((cursor)->offset))
/*
 * Copy out entries of shortform attribute lists for attr_list().
 * Shortform attribute lists are not stored in hashval sorted order.
 * If the output buffer is not large enough to hold them all, then we
 * we have to calculate each entries' hashvalue and sort them before
 * we can begin returning them to the user.
 */
static int
xfs_attr_shortform_list(xfs_attr_list_context_t *context)
{
	attrlist_cursor_kern_t *cursor;
	xfs_attr_sf_sort_t *sbuf, *sbp;
	xfs_attr_shortform_t *sf;
	xfs_attr_sf_entry_t *sfe;
	xfs_inode_t *dp;
	int sbsize, nsbuf, count, i;

	ASSERT(context != NULL);
	dp = context->dp;
	ASSERT(dp != NULL);
	ASSERT(dp->i_afp != NULL);
	sf = (xfs_attr_shortform_t *)dp->i_afp->if_u1.if_data;
	ASSERT(sf != NULL);
	if (!sf->hdr.count)
		return 0;
	cursor = context->cursor;
	ASSERT(cursor != NULL);

	trace_xfs_attr_list_sf(context);

	/*
	 * If the buffer is large enough and the cursor is at the start,
	 * do not bother with sorting since we will return everything in
	 * one buffer and another call using the cursor won't need to be
	 * made.
	 * Note the generous fudge factor of 16 overhead bytes per entry.
	 * If bufsize is zero then put_listent must be a search function
	 * and can just scan through what we have.
	 */
	if (context->bufsize == 0 ||
	    (XFS_ISRESET_CURSOR(cursor) &&
             (dp->i_afp->if_bytes + sf->hdr.count * 16) < context->bufsize)) {
		for (i = 0, sfe = &sf->list[0]; i < sf->hdr.count; i++) {
			context->put_listent(context,
					     sfe->flags,
					     sfe->nameval,
					     (int)sfe->namelen,
					     (int)sfe->valuelen);
			/*
			 * Either search callback finished early or
			 * didn't fit it all in the buffer after all.
			 */
			if (context->seen_enough)
				break;
			sfe = XFS_ATTR_SF_NEXTENTRY(sfe);
		}
		trace_xfs_attr_list_sf_all(context);
		return 0;
	}

	/* do no more for a search callback */
	if (context->bufsize == 0)
		return 0;

	/*
	 * It didn't all fit, so we have to sort everything on hashval.
	 */
	sbsize = sf->hdr.count * sizeof(*sbuf);
	sbp = sbuf = kmem_alloc(sbsize, KM_SLEEP | KM_NOFS);

	/*
	 * Scan the attribute list for the rest of the entries, storing
	 * the relevant info from only those that match into a buffer.
	 */
	nsbuf = 0;
	for (i = 0, sfe = &sf->list[0]; i < sf->hdr.count; i++) {
		if (unlikely(
		    ((char *)sfe < (char *)sf) ||
		    ((char *)sfe >= ((char *)sf + dp->i_afp->if_bytes)))) {
			XFS_CORRUPTION_ERROR("xfs_attr_shortform_list",
					     XFS_ERRLEVEL_LOW,
					     context->dp->i_mount, sfe);
			kmem_free(sbuf);
			return -EFSCORRUPTED;
		}

		sbp->entno = i;
		sbp->hash = xfs_da_hashname(sfe->nameval, sfe->namelen);
		sbp->name = sfe->nameval;
		sbp->namelen = sfe->namelen;
		/* These are bytes, and both on-disk, don't endian-flip */
		sbp->valuelen = sfe->valuelen;
		sbp->flags = sfe->flags;
		sfe = XFS_ATTR_SF_NEXTENTRY(sfe);
		sbp++;
		nsbuf++;
	}

	/*
	 * Sort the entries on hash then entno.
	 */
	xfs_sort(sbuf, nsbuf, sizeof(*sbuf), xfs_attr_shortform_compare);

	/*
	 * Re-find our place IN THE SORTED LIST.
	 */
	count = 0;
	cursor->initted = 1;
	cursor->blkno = 0;
	for (sbp = sbuf, i = 0; i < nsbuf; i++, sbp++) {
		if (sbp->hash == cursor->hashval) {
			if (cursor->offset == count) {
				break;
			}
			count++;
		} else if (sbp->hash > cursor->hashval) {
			break;
		}
	}
	if (i == nsbuf) {
		kmem_free(sbuf);
		return 0;
	}

	/*
	 * Loop putting entries into the user buffer.
	 */
	for ( ; i < nsbuf; i++, sbp++) {
		if (cursor->hashval != sbp->hash) {
			cursor->hashval = sbp->hash;
			cursor->offset = 0;
		}
		context->put_listent(context,
				     sbp->flags,
				     sbp->name,
				     sbp->namelen,
				     sbp->valuelen);
		if (context->seen_enough)
			break;
		cursor->offset++;
	}

	kmem_free(sbuf);
	return 0;
}

/*
 * We didn't find the block & hash mentioned in the cursor state, so
 * walk down the attr btree looking for the hash.
 */
STATIC int
xfs_attr_node_list_lookup(
	struct xfs_attr_list_context	*context,
	struct attrlist_cursor_kern	*cursor,
	struct xfs_buf			**pbp)
{
	struct xfs_da3_icnode_hdr	nodehdr;
	struct xfs_da_intnode		*node;
	struct xfs_da_node_entry	*btree;
	struct xfs_inode		*dp = context->dp;
	struct xfs_mount		*mp = dp->i_mount;
	struct xfs_trans		*tp = context->tp;
	struct xfs_buf			*bp;
	int				i;
	int				error = 0;
	unsigned int			expected_level = 0;
	uint16_t			magic;

	ASSERT(*pbp == NULL);
	cursor->blkno = 0;
	for (;;) {
		error = xfs_da3_node_read(tp, dp, cursor->blkno, -1, &bp,
				XFS_ATTR_FORK);
		if (error)
			return error;
		node = bp->b_addr;
		magic = be16_to_cpu(node->hdr.info.magic);
		if (magic == XFS_ATTR_LEAF_MAGIC ||
		    magic == XFS_ATTR3_LEAF_MAGIC)
			break;
		if (magic != XFS_DA_NODE_MAGIC &&
		    magic != XFS_DA3_NODE_MAGIC) {
			XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp,
					node);
			goto out_corruptbuf;
		}

		dp->d_ops->node_hdr_from_disk(&nodehdr, node);

		/* Tree taller than we can handle; bail out! */
		if (nodehdr.level >= XFS_DA_NODE_MAXDEPTH)
			goto out_corruptbuf;

		/* Check the level from the root node. */
		if (cursor->blkno == 0)
			expected_level = nodehdr.level - 1;
		else if (expected_level != nodehdr.level)
			goto out_corruptbuf;
		else
			expected_level--;

		btree = dp->d_ops->node_tree_p(node);
		for (i = 0; i < nodehdr.count; btree++, i++) {
			if (cursor->hashval <= be32_to_cpu(btree->hashval)) {
				cursor->blkno = be32_to_cpu(btree->before);
				trace_xfs_attr_list_node_descend(context,
						btree);
				break;
			}
		}
		xfs_trans_brelse(tp, bp);

		if (i == nodehdr.count)
			return 0;

		/* We can't point back to the root. */
		if (cursor->blkno == 0)
			return -EFSCORRUPTED;
	}

	if (expected_level != 0)
		goto out_corruptbuf;

	*pbp = bp;
	return 0;

out_corruptbuf:
	xfs_trans_brelse(tp, bp);
	return -EFSCORRUPTED;
}

STATIC int
xfs_attr_node_list(
	struct xfs_attr_list_context	*context)
{
	struct xfs_attr3_icleaf_hdr	leafhdr;
	struct attrlist_cursor_kern	*cursor;
	struct xfs_attr_leafblock	*leaf;
	struct xfs_da_intnode		*node;
	struct xfs_buf			*bp;
	struct xfs_inode		*dp = context->dp;
	struct xfs_mount		*mp = dp->i_mount;
	int				error;

	trace_xfs_attr_node_list(context);

	cursor = context->cursor;
	cursor->initted = 1;

	/*
	 * Do all sorts of validation on the passed-in cursor structure.
	 * If anything is amiss, ignore the cursor and look up the hashval
	 * starting from the btree root.
	 */
	bp = NULL;
	if (cursor->blkno > 0) {
		error = xfs_da3_node_read(context->tp, dp, cursor->blkno, -1,
					      &bp, XFS_ATTR_FORK);
		if ((error != 0) && (error != -EFSCORRUPTED))
			return error;
		if (bp) {
			struct xfs_attr_leaf_entry *entries;

			node = bp->b_addr;
			switch (be16_to_cpu(node->hdr.info.magic)) {
			case XFS_DA_NODE_MAGIC:
			case XFS_DA3_NODE_MAGIC:
				trace_xfs_attr_list_wrong_blk(context);
				xfs_trans_brelse(context->tp, bp);
				bp = NULL;
				break;
			case XFS_ATTR_LEAF_MAGIC:
			case XFS_ATTR3_LEAF_MAGIC:
				leaf = bp->b_addr;
				xfs_attr3_leaf_hdr_from_disk(mp->m_attr_geo,
							     &leafhdr, leaf);
				entries = xfs_attr3_leaf_entryp(leaf);
				if (cursor->hashval > be32_to_cpu(
						entries[leafhdr.count - 1].hashval)) {
					trace_xfs_attr_list_wrong_blk(context);
					xfs_trans_brelse(context->tp, bp);
					bp = NULL;
				} else if (cursor->hashval <= be32_to_cpu(
						entries[0].hashval)) {
					trace_xfs_attr_list_wrong_blk(context);
					xfs_trans_brelse(context->tp, bp);
					bp = NULL;
				}
				break;
			default:
				trace_xfs_attr_list_wrong_blk(context);
				xfs_trans_brelse(context->tp, bp);
				bp = NULL;
			}
		}
	}

	/*
	 * We did not find what we expected given the cursor's contents,
	 * so we start from the top and work down based on the hash value.
	 * Note that start of node block is same as start of leaf block.
	 */
	if (bp == NULL) {
		error = xfs_attr_node_list_lookup(context, cursor, &bp);
		if (error || !bp)
			return error;
	}
	ASSERT(bp != NULL);

	/*
	 * Roll upward through the blocks, processing each leaf block in
	 * order.  As long as there is space in the result buffer, keep
	 * adding the information.
	 */
	for (;;) {
		leaf = bp->b_addr;
		xfs_attr3_leaf_list_int(bp, context);
		xfs_attr3_leaf_hdr_from_disk(mp->m_attr_geo, &leafhdr, leaf);
		if (context->seen_enough || leafhdr.forw == 0)
			break;
		cursor->blkno = leafhdr.forw;
		xfs_trans_brelse(context->tp, bp);
		error = xfs_attr3_leaf_read(context->tp, dp, cursor->blkno, -1, &bp);
		if (error)
			return error;
	}
	xfs_trans_brelse(context->tp, bp);
	return 0;
}

/*
 * Copy out attribute list entries for attr_list(), for leaf attribute lists.
 */
void
xfs_attr3_leaf_list_int(
	struct xfs_buf			*bp,
	struct xfs_attr_list_context	*context)
{
	struct attrlist_cursor_kern	*cursor;
	struct xfs_attr_leafblock	*leaf;
	struct xfs_attr3_icleaf_hdr	ichdr;
	struct xfs_attr_leaf_entry	*entries;
	struct xfs_attr_leaf_entry	*entry;
	int				i;
	struct xfs_mount		*mp = context->dp->i_mount;

	trace_xfs_attr_list_leaf(context);

	leaf = bp->b_addr;
	xfs_attr3_leaf_hdr_from_disk(mp->m_attr_geo, &ichdr, leaf);
	entries = xfs_attr3_leaf_entryp(leaf);

	cursor = context->cursor;
	cursor->initted = 1;

	/*
	 * Re-find our place in the leaf block if this is a new syscall.
	 */
	if (context->resynch) {
		entry = &entries[0];
		for (i = 0; i < ichdr.count; entry++, i++) {
			if (be32_to_cpu(entry->hashval) == cursor->hashval) {
				if (cursor->offset == context->dupcnt) {
					context->dupcnt = 0;
					break;
				}
				context->dupcnt++;
			} else if (be32_to_cpu(entry->hashval) >
					cursor->hashval) {
				context->dupcnt = 0;
				break;
			}
		}
		if (i == ichdr.count) {
			trace_xfs_attr_list_notfound(context);
			return;
		}
	} else {
		entry = &entries[0];
		i = 0;
	}
	context->resynch = 0;

	/*
	 * We have found our place, start copying out the new attributes.
	 */
	for (; i < ichdr.count; entry++, i++) {
		char *name;
		int namelen, valuelen;

		if (be32_to_cpu(entry->hashval) != cursor->hashval) {
			cursor->hashval = be32_to_cpu(entry->hashval);
			cursor->offset = 0;
		}

		if ((entry->flags & XFS_ATTR_INCOMPLETE) &&
		    !(context->flags & ATTR_INCOMPLETE))
			continue;		/* skip incomplete entries */

		if (entry->flags & XFS_ATTR_LOCAL) {
			xfs_attr_leaf_name_local_t *name_loc;

			name_loc = xfs_attr3_leaf_name_local(leaf, i);
			name = name_loc->nameval;
			namelen = name_loc->namelen;
			valuelen = be16_to_cpu(name_loc->valuelen);
		} else {
			xfs_attr_leaf_name_remote_t *name_rmt;

			name_rmt = xfs_attr3_leaf_name_remote(leaf, i);
			name = name_rmt->name;
			namelen = name_rmt->namelen;
			valuelen = be32_to_cpu(name_rmt->valuelen);
		}

		context->put_listent(context, entry->flags,
					      name, namelen, valuelen);
		if (context->seen_enough)
			break;
		cursor->offset++;
	}
	trace_xfs_attr_list_leaf_end(context);
	return;
}

/*
 * Copy out attribute entries for attr_list(), for leaf attribute lists.
 */
STATIC int
xfs_attr_leaf_list(xfs_attr_list_context_t *context)
{
	int error;
	struct xfs_buf *bp;

	trace_xfs_attr_leaf_list(context);

	context->cursor->blkno = 0;
	error = xfs_attr3_leaf_read(context->tp, context->dp, 0, -1, &bp);
	if (error)
		return error;

	xfs_attr3_leaf_list_int(bp, context);
	xfs_trans_brelse(context->tp, bp);
	return 0;
}

int
xfs_attr_list_int_ilocked(
	struct xfs_attr_list_context	*context)
{
	struct xfs_inode		*dp = context->dp;

	ASSERT(xfs_isilocked(dp, XFS_ILOCK_SHARED | XFS_ILOCK_EXCL));

	/*
	 * Decide on what work routines to call based on the inode size.
	 */
	if (!xfs_inode_hasattr(dp))
		return 0;
	else if (dp->i_d.di_aformat == XFS_DINODE_FMT_LOCAL)
		return xfs_attr_shortform_list(context);
	else if (xfs_bmap_one_block(dp, XFS_ATTR_FORK))
		return xfs_attr_leaf_list(context);
	return xfs_attr_node_list(context);
}

int
xfs_attr_list_int(
	xfs_attr_list_context_t *context)
{
	int error;
	xfs_inode_t *dp = context->dp;
	uint		lock_mode;

	XFS_STATS_INC(dp->i_mount, xs_attr_list);

	if (XFS_FORCED_SHUTDOWN(dp->i_mount))
		return -EIO;

	lock_mode = xfs_ilock_attr_map_shared(dp);
	error = xfs_attr_list_int_ilocked(context);
	xfs_iunlock(dp, lock_mode);
	return error;
}

#define	ATTR_ENTBASESIZE		/* minimum bytes used by an attr */ \
	(((struct attrlist_ent *) 0)->a_name - (char *) 0)
#define	ATTR_ENTSIZE(namelen)		/* actual bytes used by an attr */ \
	((ATTR_ENTBASESIZE + (namelen) + 1 + sizeof(uint32_t)-1) \
	 & ~(sizeof(uint32_t)-1))

/*
 * Format an attribute and copy it out to the user's buffer.
 * Take care to check values and protect against them changing later,
 * we may be reading them directly out of a user buffer.
 */
STATIC void
xfs_attr_put_listent(
	xfs_attr_list_context_t *context,
	int		flags,
	unsigned char	*name,
	int		namelen,
	int		valuelen)
{
	struct attrlist *alist = (struct attrlist *)context->alist;
	attrlist_ent_t *aep;
	int arraytop;

	ASSERT(!(context->flags & ATTR_KERNOVAL));
	ASSERT(context->count >= 0);
	ASSERT(context->count < (ATTR_MAX_VALUELEN/8));
	ASSERT(context->firstu >= sizeof(*alist));
	ASSERT(context->firstu <= context->bufsize);

	/*
	 * Only list entries in the right namespace.
	 */
	if (((context->flags & ATTR_SECURE) == 0) !=
	    ((flags & XFS_ATTR_SECURE) == 0))
		return;
	if (((context->flags & ATTR_ROOT) == 0) !=
	    ((flags & XFS_ATTR_ROOT) == 0))
		return;

	arraytop = sizeof(*alist) +
			context->count * sizeof(alist->al_offset[0]);
	context->firstu -= ATTR_ENTSIZE(namelen);
	if (context->firstu < arraytop) {
		trace_xfs_attr_list_full(context);
		alist->al_more = 1;
		context->seen_enough = 1;
		return;
	}

	aep = (attrlist_ent_t *)&context->alist[context->firstu];
	aep->a_valuelen = valuelen;
	memcpy(aep->a_name, name, namelen);
	aep->a_name[namelen] = 0;
	alist->al_offset[context->count++] = context->firstu;
	alist->al_count = context->count;
	trace_xfs_attr_list_add(context);
	return;
}

/*
 * Generate a list of extended attribute names and optionally
 * also value lengths.  Positive return value follows the XFS
 * convention of being an error, zero or negative return code
 * is the length of the buffer returned (negated), indicating
 * success.
 */
int
xfs_attr_list(
	xfs_inode_t	*dp,
	char		*buffer,
	int		bufsize,
	int		flags,
	attrlist_cursor_kern_t *cursor)
{
	xfs_attr_list_context_t context;
	struct attrlist *alist;
	int error;

	/*
	 * Validate the cursor.
	 */
	if (cursor->pad1 || cursor->pad2)
		return -EINVAL;
	if ((cursor->initted == 0) &&
	    (cursor->hashval || cursor->blkno || cursor->offset))
		return -EINVAL;

	/* Only internal consumers can retrieve incomplete attrs. */
	if (flags & ATTR_INCOMPLETE)
		return -EINVAL;

	/*
	 * Check for a properly aligned buffer.
	 */
	if (((long)buffer) & (sizeof(int)-1))
		return -EFAULT;
	if (flags & ATTR_KERNOVAL)
		bufsize = 0;

	/*
	 * Initialize the output buffer.
	 */
	memset(&context, 0, sizeof(context));
	context.dp = dp;
	context.cursor = cursor;
	context.resynch = 1;
	context.flags = flags;
	context.alist = buffer;
	context.bufsize = (bufsize & ~(sizeof(int)-1));  /* align */
	context.firstu = context.bufsize;
	context.put_listent = xfs_attr_put_listent;

	alist = (struct attrlist *)context.alist;
	alist->al_count = 0;
	alist->al_more = 0;
	alist->al_offset[0] = context.bufsize;

	error = xfs_attr_list_int(&context);
	ASSERT(error <= 0);
	return error;
}
Commit	Line	Data
abec5f2b DC	1	/*
	2	* Copyright (c) 2000-2005 Silicon Graphics, Inc.
	3	* Copyright (c) 2013 Red Hat, Inc.
	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"
a4fbe6ab	21	#include "xfs_format.h"
239880ef DC	22	#include "xfs_log_format.h"
239880ef DC	23	#include "xfs_trans_resv.h"
abec5f2b	24	#include "xfs_bit.h"
abec5f2b	25	#include "xfs_mount.h"
57062787	26	#include "xfs_da_format.h"
abec5f2b	27	#include "xfs_da_btree.h"
abec5f2b	28	#include "xfs_inode.h"
239880ef	29	#include "xfs_trans.h"
abec5f2b DC	30	#include "xfs_inode_item.h"
	31	#include "xfs_bmap.h"
	32	#include "xfs_attr.h"
a4fbe6ab DC	33	#include "xfs_attr_sf.h"
a4fbe6ab DC	34	#include "xfs_attr_remote.h"
abec5f2b DC	35	#include "xfs_attr_leaf.h"
	36	#include "xfs_error.h"
	37	#include "xfs_trace.h"
	38	#include "xfs_buf_item.h"
	39	#include "xfs_cksum.h"
4bceb18f	40	#include "xfs_dir2.h"
abec5f2b DC	41
	42	STATIC int
	43	xfs_attr_shortform_compare(const void a, const void b)
	44	{
	45	xfs_attr_sf_sort_t sa, sb;
	46
	47	sa = (xfs_attr_sf_sort_t *)a;
	48	sb = (xfs_attr_sf_sort_t *)b;
	49	if (sa->hash < sb->hash) {
d99831ff	50	return -1;
abec5f2b	51	} else if (sa->hash > sb->hash) {
d99831ff	52	return 1;
abec5f2b	53	} else {
d99831ff	54	return sa->entno - sb->entno;
abec5f2b DC	55	}
	56	}
	57
	58	#define XFS_ISRESET_CURSOR(cursor) \
	59	(!((cursor)->initted) && !((cursor)->hashval) && \
	60	!((cursor)->blkno) && !((cursor)->offset))
	61	/*
	62	* Copy out entries of shortform attribute lists for attr_list().
	63	* Shortform attribute lists are not stored in hashval sorted order.
	64	* If the output buffer is not large enough to hold them all, then we
	65	* we have to calculate each entries' hashvalue and sort them before
	66	* we can begin returning them to the user.
	67	*/
0d5a75e9	68	static int
abec5f2b DC	69	xfs_attr_shortform_list(xfs_attr_list_context_t *context)
	70	{
	71	attrlist_cursor_kern_t *cursor;
	72	xfs_attr_sf_sort_t sbuf, sbp;
	73	xfs_attr_shortform_t *sf;
	74	xfs_attr_sf_entry_t *sfe;
	75	xfs_inode_t *dp;
	76	int sbsize, nsbuf, count, i;
abec5f2b DC	77
	78	ASSERT(context != NULL);
	79	dp = context->dp;
	80	ASSERT(dp != NULL);
	81	ASSERT(dp->i_afp != NULL);
	82	sf = (xfs_attr_shortform_t *)dp->i_afp->if_u1.if_data;
	83	ASSERT(sf != NULL);
	84	if (!sf->hdr.count)
d99831ff	85	return 0;
abec5f2b DC	86	cursor = context->cursor;
	87	ASSERT(cursor != NULL);
	88
	89	trace_xfs_attr_list_sf(context);
	90
	91	/*
	92	* If the buffer is large enough and the cursor is at the start,
	93	* do not bother with sorting since we will return everything in
	94	* one buffer and another call using the cursor won't need to be
	95	* made.
	96	* Note the generous fudge factor of 16 overhead bytes per entry.
	97	* If bufsize is zero then put_listent must be a search function
	98	* and can just scan through what we have.
	99	*/
	100	if (context->bufsize == 0 \|\|
	101	(XFS_ISRESET_CURSOR(cursor) &&
	102	(dp->i_afp->if_bytes + sf->hdr.count * 16) < context->bufsize)) {
	103	for (i = 0, sfe = &sf->list[0]; i < sf->hdr.count; i++) {
f7a136ae ES	104	context->put_listent(context,
	105	sfe->flags,
	106	sfe->nameval,
	107	(int)sfe->namelen,
	108	(int)sfe->valuelen);
abec5f2b DC	109	/*
	110	* Either search callback finished early or
	111	* didn't fit it all in the buffer after all.
	112	*/
	113	if (context->seen_enough)
	114	break;
abec5f2b DC	115	sfe = XFS_ATTR_SF_NEXTENTRY(sfe);
	116	}
	117	trace_xfs_attr_list_sf_all(context);
d99831ff	118	return 0;
abec5f2b DC	119	}
	120
	121	/* do no more for a search callback */
	122	if (context->bufsize == 0)
	123	return 0;
	124
	125	/*
	126	* It didn't all fit, so we have to sort everything on hashval.
	127	*/
	128	sbsize = sf->hdr.count * sizeof(*sbuf);
	129	sbp = sbuf = kmem_alloc(sbsize, KM_SLEEP \| KM_NOFS);
	130
	131	/*
	132	* Scan the attribute list for the rest of the entries, storing
	133	* the relevant info from only those that match into a buffer.
	134	*/
	135	nsbuf = 0;
	136	for (i = 0, sfe = &sf->list[0]; i < sf->hdr.count; i++) {
	137	if (unlikely(
	138	((char )sfe < (char )sf) \|\|
	139	((char )sfe >= ((char )sf + dp->i_afp->if_bytes)))) {
	140	XFS_CORRUPTION_ERROR("xfs_attr_shortform_list",
	141	XFS_ERRLEVEL_LOW,
	142	context->dp->i_mount, sfe);
	143	kmem_free(sbuf);
2451337d	144	return -EFSCORRUPTED;
abec5f2b DC	145	}
	146
	147	sbp->entno = i;
	148	sbp->hash = xfs_da_hashname(sfe->nameval, sfe->namelen);
	149	sbp->name = sfe->nameval;
	150	sbp->namelen = sfe->namelen;
	151	/* These are bytes, and both on-disk, don't endian-flip */
	152	sbp->valuelen = sfe->valuelen;
	153	sbp->flags = sfe->flags;
	154	sfe = XFS_ATTR_SF_NEXTENTRY(sfe);
	155	sbp++;
	156	nsbuf++;
	157	}
	158
	159	/*
	160	* Sort the entries on hash then entno.
	161	*/
	162	xfs_sort(sbuf, nsbuf, sizeof(*sbuf), xfs_attr_shortform_compare);
	163
	164	/*
	165	* Re-find our place IN THE SORTED LIST.
	166	*/
	167	count = 0;
	168	cursor->initted = 1;
	169	cursor->blkno = 0;
	170	for (sbp = sbuf, i = 0; i < nsbuf; i++, sbp++) {
	171	if (sbp->hash == cursor->hashval) {
	172	if (cursor->offset == count) {
	173	break;
	174	}
	175	count++;
	176	} else if (sbp->hash > cursor->hashval) {
	177	break;
	178	}
	179	}
	180	if (i == nsbuf) {
	181	kmem_free(sbuf);
d99831ff	182	return 0;
abec5f2b DC	183	}
	184
	185	/*
	186	* Loop putting entries into the user buffer.
	187	*/
	188	for ( ; i < nsbuf; i++, sbp++) {
	189	if (cursor->hashval != sbp->hash) {
	190	cursor->hashval = sbp->hash;
	191	cursor->offset = 0;
	192	}
f7a136ae ES	193	context->put_listent(context,
	194	sbp->flags,
	195	sbp->name,
	196	sbp->namelen,
	197	sbp->valuelen);
abec5f2b DC	198	if (context->seen_enough)
	199	break;
	200	cursor->offset++;
	201	}
	202
	203	kmem_free(sbuf);
d99831ff	204	return 0;
abec5f2b DC	205	}
abec5f2b DC	206
bdaac93f DW	207	/*
	208	* We didn't find the block & hash mentioned in the cursor state, so
	209	* walk down the attr btree looking for the hash.
	210	*/
abec5f2b	211	STATIC int
bdaac93f DW	212	xfs_attr_node_list_lookup(
	213	struct xfs_attr_list_context *context,
	214	struct attrlist_cursor_kern *cursor,
	215	struct xfs_buf **pbp)
abec5f2b	216	{
bdaac93f DW	217	struct xfs_da3_icnode_hdr nodehdr;
	218	struct xfs_da_intnode *node;
	219	struct xfs_da_node_entry *btree;
	220	struct xfs_inode *dp = context->dp;
	221	struct xfs_mount *mp = dp->i_mount;
	222	struct xfs_trans *tp = context->tp;
	223	struct xfs_buf *bp;
	224	int i;
	225	int error = 0;
8210f4dd	226	unsigned int expected_level = 0;
bdaac93f DW	227	uint16_t magic;
	228
	229	ASSERT(*pbp == NULL);
	230	cursor->blkno = 0;
	231	for (;;) {
	232	error = xfs_da3_node_read(tp, dp, cursor->blkno, -1, &bp,
	233	XFS_ATTR_FORK);
	234	if (error)
	235	return error;
	236	node = bp->b_addr;
	237	magic = be16_to_cpu(node->hdr.info.magic);
	238	if (magic == XFS_ATTR_LEAF_MAGIC \|\|
	239	magic == XFS_ATTR3_LEAF_MAGIC)
	240	break;
	241	if (magic != XFS_DA_NODE_MAGIC &&
	242	magic != XFS_DA3_NODE_MAGIC) {
	243	XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp,
	244	node);
	245	goto out_corruptbuf;
	246	}
	247
	248	dp->d_ops->node_hdr_from_disk(&nodehdr, node);
	249
8210f4dd DW	250	/* Tree taller than we can handle; bail out! */
	251	if (nodehdr.level >= XFS_DA_NODE_MAXDEPTH)
	252	goto out_corruptbuf;
	253
	254	/* Check the level from the root node. */
	255	if (cursor->blkno == 0)
	256	expected_level = nodehdr.level - 1;
	257	else if (expected_level != nodehdr.level)
	258	goto out_corruptbuf;
	259	else
	260	expected_level--;
	261
bdaac93f DW	262	btree = dp->d_ops->node_tree_p(node);
	263	for (i = 0; i < nodehdr.count; btree++, i++) {
	264	if (cursor->hashval <= be32_to_cpu(btree->hashval)) {
	265	cursor->blkno = be32_to_cpu(btree->before);
	266	trace_xfs_attr_list_node_descend(context,
	267	btree);
	268	break;
	269	}
	270	}
	271	xfs_trans_brelse(tp, bp);
	272
	273	if (i == nodehdr.count)
	274	return 0;
8210f4dd DW	275
	276	/* We can't point back to the root. */
	277	if (cursor->blkno == 0)
	278	return -EFSCORRUPTED;
bdaac93f DW	279	}
bdaac93f DW	280
8210f4dd DW	281	if (expected_level != 0)
	282	goto out_corruptbuf;
	283
bdaac93f DW	284	*pbp = bp;
	285	return 0;
	286
	287	out_corruptbuf:
	288	xfs_trans_brelse(tp, bp);
	289	return -EFSCORRUPTED;
	290	}
	291
	292	STATIC int
	293	xfs_attr_node_list(
	294	struct xfs_attr_list_context *context)
	295	{
	296	struct xfs_attr3_icleaf_hdr leafhdr;
	297	struct attrlist_cursor_kern *cursor;
	298	struct xfs_attr_leafblock *leaf;
	299	struct xfs_da_intnode *node;
	300	struct xfs_buf *bp;
	301	struct xfs_inode *dp = context->dp;
	302	struct xfs_mount *mp = dp->i_mount;
	303	int error;
abec5f2b DC	304
	305	trace_xfs_attr_node_list(context);
	306
	307	cursor = context->cursor;
	308	cursor->initted = 1;
	309
	310	/*
	311	* Do all sorts of validation on the passed-in cursor structure.
	312	* If anything is amiss, ignore the cursor and look up the hashval
	313	* starting from the btree root.
	314	*/
	315	bp = NULL;
	316	if (cursor->blkno > 0) {
ad017f65	317	error = xfs_da3_node_read(context->tp, dp, cursor->blkno, -1,
abec5f2b	318	&bp, XFS_ATTR_FORK);
2451337d	319	if ((error != 0) && (error != -EFSCORRUPTED))
d99831ff	320	return error;
abec5f2b DC	321	if (bp) {
	322	struct xfs_attr_leaf_entry *entries;
	323
	324	node = bp->b_addr;
	325	switch (be16_to_cpu(node->hdr.info.magic)) {
	326	case XFS_DA_NODE_MAGIC:
	327	case XFS_DA3_NODE_MAGIC:
	328	trace_xfs_attr_list_wrong_blk(context);
ad017f65	329	xfs_trans_brelse(context->tp, bp);
abec5f2b DC	330	bp = NULL;
	331	break;
	332	case XFS_ATTR_LEAF_MAGIC:
	333	case XFS_ATTR3_LEAF_MAGIC:
	334	leaf = bp->b_addr;
2f661241 BF	335	xfs_attr3_leaf_hdr_from_disk(mp->m_attr_geo,
2f661241 BF	336	&leafhdr, leaf);
abec5f2b DC	337	entries = xfs_attr3_leaf_entryp(leaf);
	338	if (cursor->hashval > be32_to_cpu(
	339	entries[leafhdr.count - 1].hashval)) {
	340	trace_xfs_attr_list_wrong_blk(context);
ad017f65	341	xfs_trans_brelse(context->tp, bp);
abec5f2b DC	342	bp = NULL;
	343	} else if (cursor->hashval <= be32_to_cpu(
	344	entries[0].hashval)) {
	345	trace_xfs_attr_list_wrong_blk(context);
ad017f65	346	xfs_trans_brelse(context->tp, bp);
abec5f2b DC	347	bp = NULL;
	348	}
	349	break;
	350	default:
	351	trace_xfs_attr_list_wrong_blk(context);
ad017f65	352	xfs_trans_brelse(context->tp, bp);
abec5f2b DC	353	bp = NULL;
	354	}
	355	}
	356	}
	357
	358	/*
	359	* We did not find what we expected given the cursor's contents,
	360	* so we start from the top and work down based on the hash value.
	361	* Note that start of node block is same as start of leaf block.
	362	*/
	363	if (bp == NULL) {
bdaac93f DW	364	error = xfs_attr_node_list_lookup(context, cursor, &bp);
	365	if (error \|\| !bp)
	366	return error;
abec5f2b DC	367	}
	368	ASSERT(bp != NULL);
	369
	370	/*
	371	* Roll upward through the blocks, processing each leaf block in
	372	* order. As long as there is space in the result buffer, keep
	373	* adding the information.
	374	*/
	375	for (;;) {
	376	leaf = bp->b_addr;
f7a136ae	377	xfs_attr3_leaf_list_int(bp, context);
2f661241	378	xfs_attr3_leaf_hdr_from_disk(mp->m_attr_geo, &leafhdr, leaf);
abec5f2b DC	379	if (context->seen_enough \|\| leafhdr.forw == 0)
	380	break;
	381	cursor->blkno = leafhdr.forw;
ad017f65 DW	382	xfs_trans_brelse(context->tp, bp);
ad017f65 DW	383	error = xfs_attr3_leaf_read(context->tp, dp, cursor->blkno, -1, &bp);
abec5f2b DC	384	if (error)
	385	return error;
	386	}
ad017f65	387	xfs_trans_brelse(context->tp, bp);
abec5f2b DC	388	return 0;
	389	}
	390
	391	/*
	392	* Copy out attribute list entries for attr_list(), for leaf attribute lists.
	393	*/
f7a136ae	394	void
abec5f2b DC	395	xfs_attr3_leaf_list_int(
	396	struct xfs_buf *bp,
	397	struct xfs_attr_list_context *context)
	398	{
	399	struct attrlist_cursor_kern *cursor;
	400	struct xfs_attr_leafblock *leaf;
	401	struct xfs_attr3_icleaf_hdr ichdr;
	402	struct xfs_attr_leaf_entry *entries;
	403	struct xfs_attr_leaf_entry *entry;
abec5f2b	404	int i;
2f661241	405	struct xfs_mount *mp = context->dp->i_mount;
abec5f2b DC	406
	407	trace_xfs_attr_list_leaf(context);
	408
	409	leaf = bp->b_addr;
2f661241	410	xfs_attr3_leaf_hdr_from_disk(mp->m_attr_geo, &ichdr, leaf);
abec5f2b DC	411	entries = xfs_attr3_leaf_entryp(leaf);
	412
	413	cursor = context->cursor;
	414	cursor->initted = 1;
	415
	416	/*
	417	* Re-find our place in the leaf block if this is a new syscall.
	418	*/
	419	if (context->resynch) {
	420	entry = &entries[0];
	421	for (i = 0; i < ichdr.count; entry++, i++) {
	422	if (be32_to_cpu(entry->hashval) == cursor->hashval) {
	423	if (cursor->offset == context->dupcnt) {
	424	context->dupcnt = 0;
	425	break;
	426	}
	427	context->dupcnt++;
	428	} else if (be32_to_cpu(entry->hashval) >
	429	cursor->hashval) {
	430	context->dupcnt = 0;
	431	break;
	432	}
	433	}
	434	if (i == ichdr.count) {
	435	trace_xfs_attr_list_notfound(context);
f7a136ae	436	return;
abec5f2b DC	437	}
	438	} else {
	439	entry = &entries[0];
	440	i = 0;
	441	}
	442	context->resynch = 0;
	443
	444	/*
	445	* We have found our place, start copying out the new attributes.
	446	*/
abec5f2b	447	for (; i < ichdr.count; entry++, i++) {
3ab3ffca ES	448	char *name;
	449	int namelen, valuelen;
	450
abec5f2b DC	451	if (be32_to_cpu(entry->hashval) != cursor->hashval) {
	452	cursor->hashval = be32_to_cpu(entry->hashval);
	453	cursor->offset = 0;
	454	}
	455
eec0482e DW	456	if ((entry->flags & XFS_ATTR_INCOMPLETE) &&
eec0482e DW	457	!(context->flags & ATTR_INCOMPLETE))
abec5f2b DC	458	continue; /* skip incomplete entries */
	459
	460	if (entry->flags & XFS_ATTR_LOCAL) {
3ab3ffca	461	xfs_attr_leaf_name_local_t *name_loc;
abec5f2b	462
3ab3ffca ES	463	name_loc = xfs_attr3_leaf_name_local(leaf, i);
	464	name = name_loc->nameval;
	465	namelen = name_loc->namelen;
	466	valuelen = be16_to_cpu(name_loc->valuelen);
	467	} else {
	468	xfs_attr_leaf_name_remote_t *name_rmt;
abec5f2b	469
3ab3ffca ES	470	name_rmt = xfs_attr3_leaf_name_remote(leaf, i);
	471	name = name_rmt->name;
	472	namelen = name_rmt->namelen;
	473	valuelen = be32_to_cpu(name_rmt->valuelen);
abec5f2b	474	}
3ab3ffca	475
f7a136ae	476	context->put_listent(context, entry->flags,
3ab3ffca	477	name, namelen, valuelen);
abec5f2b DC	478	if (context->seen_enough)
	479	break;
	480	cursor->offset++;
	481	}
	482	trace_xfs_attr_list_leaf_end(context);
f7a136ae	483	return;
abec5f2b DC	484	}
	485
	486	/*
	487	* Copy out attribute entries for attr_list(), for leaf attribute lists.
	488	*/
	489	STATIC int
	490	xfs_attr_leaf_list(xfs_attr_list_context_t *context)
	491	{
	492	int error;
	493	struct xfs_buf *bp;
	494
	495	trace_xfs_attr_leaf_list(context);
	496
	497	context->cursor->blkno = 0;
ad017f65	498	error = xfs_attr3_leaf_read(context->tp, context->dp, 0, -1, &bp);
abec5f2b	499	if (error)
b474c7ae	500	return error;
abec5f2b	501
f7a136ae	502	xfs_attr3_leaf_list_int(bp, context);
ad017f65	503	xfs_trans_brelse(context->tp, bp);
f7a136ae	504	return 0;
abec5f2b DC	505	}
abec5f2b DC	506
ad017f65 DW	507	int
	508	xfs_attr_list_int_ilocked(
	509	struct xfs_attr_list_context *context)
	510	{
	511	struct xfs_inode *dp = context->dp;
	512
5af7777e CH	513	ASSERT(xfs_isilocked(dp, XFS_ILOCK_SHARED \| XFS_ILOCK_EXCL));
5af7777e CH	514
ad017f65 DW	515	/*
	516	* Decide on what work routines to call based on the inode size.
	517	*/
	518	if (!xfs_inode_hasattr(dp))
	519	return 0;
	520	else if (dp->i_d.di_aformat == XFS_DINODE_FMT_LOCAL)
	521	return xfs_attr_shortform_list(context);
	522	else if (xfs_bmap_one_block(dp, XFS_ATTR_FORK))
	523	return xfs_attr_leaf_list(context);
	524	return xfs_attr_node_list(context);
	525	}
	526
abec5f2b DC	527	int
	528	xfs_attr_list_int(
	529	xfs_attr_list_context_t *context)
	530	{
	531	int error;
	532	xfs_inode_t *dp = context->dp;
568d994e	533	uint lock_mode;
abec5f2b	534
ff6d6af2	535	XFS_STATS_INC(dp->i_mount, xs_attr_list);
abec5f2b DC	536
abec5f2b DC	537	if (XFS_FORCED_SHUTDOWN(dp->i_mount))
2451337d	538	return -EIO;
abec5f2b	539
568d994e	540	lock_mode = xfs_ilock_attr_map_shared(dp);
ad017f65	541	error = xfs_attr_list_int_ilocked(context);
568d994e	542	xfs_iunlock(dp, lock_mode);
abec5f2b DC	543	return error;
	544	}
	545
	546	#define ATTR_ENTBASESIZE /* minimum bytes used by an attr */ \
	547	(((struct attrlist_ent ) 0)->a_name - (char ) 0)
	548	#define ATTR_ENTSIZE(namelen) /* actual bytes used by an attr */ \
65a7935d DW	549	((ATTR_ENTBASESIZE + (namelen) + 1 + sizeof(uint32_t)-1) \
65a7935d DW	550	& ~(sizeof(uint32_t)-1))
abec5f2b DC	551
	552	/*
	553	* Format an attribute and copy it out to the user's buffer.
	554	* Take care to check values and protect against them changing later,
	555	* we may be reading them directly out of a user buffer.
	556	*/
f7a136ae	557	STATIC void
abec5f2b DC	558	xfs_attr_put_listent(
	559	xfs_attr_list_context_t *context,
	560	int flags,
	561	unsigned char *name,
	562	int namelen,
e5bd12bf	563	int valuelen)
abec5f2b DC	564	{
	565	struct attrlist alist = (struct attrlist )context->alist;
	566	attrlist_ent_t *aep;
	567	int arraytop;
	568
	569	ASSERT(!(context->flags & ATTR_KERNOVAL));
	570	ASSERT(context->count >= 0);
	571	ASSERT(context->count < (ATTR_MAX_VALUELEN/8));
	572	ASSERT(context->firstu >= sizeof(*alist));
	573	ASSERT(context->firstu <= context->bufsize);
	574
	575	/*
	576	* Only list entries in the right namespace.
	577	*/
	578	if (((context->flags & ATTR_SECURE) == 0) !=
	579	((flags & XFS_ATTR_SECURE) == 0))
f7a136ae	580	return;
abec5f2b DC	581	if (((context->flags & ATTR_ROOT) == 0) !=
abec5f2b DC	582	((flags & XFS_ATTR_ROOT) == 0))
f7a136ae	583	return;
abec5f2b DC	584
	585	arraytop = sizeof(*alist) +
	586	context->count * sizeof(alist->al_offset[0]);
	587	context->firstu -= ATTR_ENTSIZE(namelen);
	588	if (context->firstu < arraytop) {
	589	trace_xfs_attr_list_full(context);
	590	alist->al_more = 1;
	591	context->seen_enough = 1;
f7a136ae	592	return;
abec5f2b DC	593	}
	594
	595	aep = (attrlist_ent_t *)&context->alist[context->firstu];
	596	aep->a_valuelen = valuelen;
	597	memcpy(aep->a_name, name, namelen);
	598	aep->a_name[namelen] = 0;
	599	alist->al_offset[context->count++] = context->firstu;
	600	alist->al_count = context->count;
	601	trace_xfs_attr_list_add(context);
f7a136ae	602	return;
abec5f2b DC	603	}
	604
	605	/*
	606	* Generate a list of extended attribute names and optionally
	607	* also value lengths. Positive return value follows the XFS
	608	* convention of being an error, zero or negative return code
	609	* is the length of the buffer returned (negated), indicating
	610	* success.
	611	*/
	612	int
	613	xfs_attr_list(
	614	xfs_inode_t *dp,
	615	char *buffer,
	616	int bufsize,
	617	int flags,
	618	attrlist_cursor_kern_t *cursor)
	619	{
	620	xfs_attr_list_context_t context;
	621	struct attrlist *alist;
	622	int error;
	623
	624	/*
	625	* Validate the cursor.
	626	*/
	627	if (cursor->pad1 \|\| cursor->pad2)
2451337d	628	return -EINVAL;
abec5f2b DC	629	if ((cursor->initted == 0) &&
abec5f2b DC	630	(cursor->hashval \|\| cursor->blkno \|\| cursor->offset))
2451337d	631	return -EINVAL;
abec5f2b	632
eec0482e DW	633	/* Only internal consumers can retrieve incomplete attrs. */
	634	if (flags & ATTR_INCOMPLETE)
	635	return -EINVAL;
	636
abec5f2b DC	637	/*
	638	* Check for a properly aligned buffer.
	639	*/
	640	if (((long)buffer) & (sizeof(int)-1))
2451337d	641	return -EFAULT;
abec5f2b DC	642	if (flags & ATTR_KERNOVAL)
	643	bufsize = 0;
	644
	645	/*
	646	* Initialize the output buffer.
	647	*/
	648	memset(&context, 0, sizeof(context));
	649	context.dp = dp;
	650	context.cursor = cursor;
	651	context.resynch = 1;
	652	context.flags = flags;
	653	context.alist = buffer;
	654	context.bufsize = (bufsize & ~(sizeof(int)-1)); /* align */
	655	context.firstu = context.bufsize;
	656	context.put_listent = xfs_attr_put_listent;
	657
	658	alist = (struct attrlist *)context.alist;
	659	alist->al_count = 0;
	660	alist->al_more = 0;
	661	alist->al_offset[0] = context.bufsize;
	662
	663	error = xfs_attr_list_int(&context);
2451337d	664	ASSERT(error <= 0);
abec5f2b DC	665	return error;
abec5f2b DC	666	}