xfs: scrub extended attributes

Scrub the hash tree, keys, and values in an extended attribute structure.
Refactor the attribute code to use the transaction if the caller supplied
one to avoid buffer deadocks.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>

fs/xfs/Makefile

@@ -145,6 +145,7 @@ xfs-y				+= $(addprefix scrub/, \
 				   trace.o \
 				   agheader.o \
 				   alloc.o \
+				   attr.o \
 				   bmap.o \
 				   btree.o \
 				   common.o \

fs/xfs/libxfs/xfs_fs.h

@@ -499,9 +499,10 @@ struct xfs_scrub_metadata {
 #define XFS_SCRUB_TYPE_BMBTA	13	/* attr fork block mapping */
 #define XFS_SCRUB_TYPE_BMBTC	14	/* CoW fork block mapping */
 #define XFS_SCRUB_TYPE_DIR	15	/* directory */
+#define XFS_SCRUB_TYPE_XATTR	16	/* extended attribute */
 
 /* Number of scrub subcommands. */
-#define XFS_SCRUB_TYPE_NR	16
+#define XFS_SCRUB_TYPE_NR	17
 
 /* i: Repair this metadata. */
 #define XFS_SCRUB_IFLAG_REPAIR		(1 << 0)

fs/xfs/scrub/attr.c

@@ -0,0 +1,260 @@
+/*
+ * Copyright (C) 2017 Oracle.  All Rights Reserved.
+ *
+ * Author: Darrick J. Wong <darrick.wong@oracle.com>
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; either version 2
+ * of the License, or (at your option) any later version.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301, USA.
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_mount.h"
+#include "xfs_defer.h"
+#include "xfs_btree.h"
+#include "xfs_bit.h"
+#include "xfs_log_format.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_inode.h"
+#include "xfs_da_format.h"
+#include "xfs_da_btree.h"
+#include "xfs_dir2.h"
+#include "xfs_attr.h"
+#include "xfs_attr_leaf.h"
+#include "scrub/xfs_scrub.h"
+#include "scrub/scrub.h"
+#include "scrub/common.h"
+#include "scrub/dabtree.h"
+#include "scrub/trace.h"
+
+#include <linux/posix_acl_xattr.h>
+#include <linux/xattr.h>
+
+/* Set us up to scrub an inode's extended attributes. */
+int
+xfs_scrub_setup_xattr(
+	struct xfs_scrub_context	*sc,
+	struct xfs_inode		*ip)
+{
+	/* Allocate the buffer without the inode lock held. */
+	sc->buf = kmem_zalloc_large(XATTR_SIZE_MAX, KM_SLEEP);
+	if (!sc->buf)
+		return -ENOMEM;
+
+	return xfs_scrub_setup_inode_contents(sc, ip, 0);
+}
+
+/* Extended Attributes */
+
+struct xfs_scrub_xattr {
+	struct xfs_attr_list_context	context;
+	struct xfs_scrub_context	*sc;
+};
+
+/*
+ * Check that an extended attribute key can be looked up by hash.
+ *
+ * We use the XFS attribute list iterator (i.e. xfs_attr_list_int_ilocked)
+ * to call this function for every attribute key in an inode.  Once
+ * we're here, we load the attribute value to see if any errors happen,
+ * or if we get more or less data than we expected.
+ */
+static void
+xfs_scrub_xattr_listent(
+	struct xfs_attr_list_context	*context,
+	int				flags,
+	unsigned char			*name,
+	int				namelen,
+	int				valuelen)
+{
+	struct xfs_scrub_xattr		*sx;
+	struct xfs_da_args		args = {0};
+	int				error = 0;
+
+	sx = container_of(context, struct xfs_scrub_xattr, context);
+
+	if (flags & XFS_ATTR_INCOMPLETE) {
+		/* Incomplete attr key, just mark the inode for preening. */
+		xfs_scrub_ino_set_preen(sx->sc, NULL);
+		return;
+	}
+
+	args.flags = ATTR_KERNOTIME;
+	if (flags & XFS_ATTR_ROOT)
+		args.flags |= ATTR_ROOT;
+	else if (flags & XFS_ATTR_SECURE)
+		args.flags |= ATTR_SECURE;
+	args.geo = context->dp->i_mount->m_attr_geo;
+	args.whichfork = XFS_ATTR_FORK;
+	args.dp = context->dp;
+	args.name = name;
+	args.namelen = namelen;
+	args.hashval = xfs_da_hashname(args.name, args.namelen);
+	args.trans = context->tp;
+	args.value = sx->sc->buf;
+	args.valuelen = XATTR_SIZE_MAX;
+
+	error = xfs_attr_get_ilocked(context->dp, &args);
+	if (error == -EEXIST)
+		error = 0;
+	if (!xfs_scrub_fblock_process_error(sx->sc, XFS_ATTR_FORK, args.blkno,
+			&error))
+		goto fail_xref;
+	if (args.valuelen != valuelen)
+		xfs_scrub_fblock_set_corrupt(sx->sc, XFS_ATTR_FORK,
+					     args.blkno);
+
+fail_xref:
+	return;
+}
+
+/* Scrub a attribute btree record. */
+STATIC int
+xfs_scrub_xattr_rec(
+	struct xfs_scrub_da_btree	*ds,
+	int				level,
+	void				*rec)
+{
+	struct xfs_mount		*mp = ds->state->mp;
+	struct xfs_attr_leaf_entry	*ent = rec;
+	struct xfs_da_state_blk		*blk;
+	struct xfs_attr_leaf_name_local	*lentry;
+	struct xfs_attr_leaf_name_remote	*rentry;
+	struct xfs_buf			*bp;
+	xfs_dahash_t			calc_hash;
+	xfs_dahash_t			hash;
+	int				nameidx;
+	int				hdrsize;
+	unsigned int			badflags;
+	int				error;
+
+	blk = &ds->state->path.blk[level];
+
+	/* Check the hash of the entry. */
+	error = xfs_scrub_da_btree_hash(ds, level, &ent->hashval);
+	if (error)
+		goto out;
+
+	/* Find the attr entry's location. */
+	bp = blk->bp;
+	hdrsize = xfs_attr3_leaf_hdr_size(bp->b_addr);
+	nameidx = be16_to_cpu(ent->nameidx);
+	if (nameidx < hdrsize || nameidx >= mp->m_attr_geo->blksize) {
+		xfs_scrub_da_set_corrupt(ds, level);
+		goto out;
+	}
+
+	/* Check all the padding. */
+	if (xfs_sb_version_hascrc(&ds->sc->mp->m_sb)) {
+		struct xfs_attr3_leafblock	*leaf = bp->b_addr;
+
+		if (leaf->hdr.pad1 != 0 ||
+		    leaf->hdr.pad2 != cpu_to_be32(0) ||
+		    leaf->hdr.info.hdr.pad != cpu_to_be16(0))
+			xfs_scrub_da_set_corrupt(ds, level);
+	} else {
+		struct xfs_attr_leafblock	*leaf = bp->b_addr;
+
+		if (leaf->hdr.pad1 != 0 ||
+		    leaf->hdr.info.pad != cpu_to_be16(0))
+			xfs_scrub_da_set_corrupt(ds, level);
+	}
+	if (ent->pad2 != 0)
+		xfs_scrub_da_set_corrupt(ds, level);
+
+	/* Retrieve the entry and check it. */
+	hash = be32_to_cpu(ent->hashval);
+	badflags = ~(XFS_ATTR_LOCAL | XFS_ATTR_ROOT | XFS_ATTR_SECURE |
+			XFS_ATTR_INCOMPLETE);
+	if ((ent->flags & badflags) != 0)
+		xfs_scrub_da_set_corrupt(ds, level);
+	if (ent->flags & XFS_ATTR_LOCAL) {
+		lentry = (struct xfs_attr_leaf_name_local *)
+				(((char *)bp->b_addr) + nameidx);
+		if (lentry->namelen <= 0) {
+			xfs_scrub_da_set_corrupt(ds, level);
+			goto out;
+		}
+		calc_hash = xfs_da_hashname(lentry->nameval, lentry->namelen);
+	} else {
+		rentry = (struct xfs_attr_leaf_name_remote *)
+				(((char *)bp->b_addr) + nameidx);
+		if (rentry->namelen <= 0) {
+			xfs_scrub_da_set_corrupt(ds, level);
+			goto out;
+		}
+		calc_hash = xfs_da_hashname(rentry->name, rentry->namelen);
+	}
+	if (calc_hash != hash)
+		xfs_scrub_da_set_corrupt(ds, level);
+
+out:
+	return error;
+}
+
+/* Scrub the extended attribute metadata. */
+int
+xfs_scrub_xattr(
+	struct xfs_scrub_context	*sc)
+{
+	struct xfs_scrub_xattr		sx = { 0 };
+	struct attrlist_cursor_kern	cursor = { 0 };
+	int				error = 0;
+
+	if (!xfs_inode_hasattr(sc->ip))
+		return -ENOENT;
+
+	memset(&sx, 0, sizeof(sx));
+	/* Check attribute tree structure */
+	error = xfs_scrub_da_btree(sc, XFS_ATTR_FORK, xfs_scrub_xattr_rec);
+	if (error)
+		goto out;
+
+	if (sc->sm->sm_flags & XFS_SCRUB_OFLAG_CORRUPT)
+		goto out;
+
+	/* Check that every attr key can also be looked up by hash. */
+	sx.context.dp = sc->ip;
+	sx.context.cursor = &cursor;
+	sx.context.resynch = 1;
+	sx.context.put_listent = xfs_scrub_xattr_listent;
+	sx.context.tp = sc->tp;
+	sx.context.flags = ATTR_INCOMPLETE;
+	sx.sc = sc;
+
+	/*
+	 * Look up every xattr in this file by name.
+	 *
+	 * Use the backend implementation of xfs_attr_list to call
+	 * xfs_scrub_xattr_listent on every attribute key in this inode.
+	 * In other words, we use the same iterator/callback mechanism
+	 * that listattr uses to scrub extended attributes, though in our
+	 * _listent function, we check the value of the attribute.
+	 *
+	 * The VFS only locks i_rwsem when modifying attrs, so keep all
+	 * three locks held because that's the only way to ensure we're
+	 * the only thread poking into the da btree.  We traverse the da
+	 * btree while holding a leaf buffer locked for the xattr name
+	 * iteration, which doesn't really follow the usual buffer
+	 * locking order.
+	 */
+	error = xfs_attr_list_int_ilocked(&sx.context);
+	if (!xfs_scrub_fblock_process_error(sc, XFS_ATTR_FORK, 0, &error))
+		goto out;
+out:
+	return error;
+}

fs/xfs/scrub/common.h

@@ -95,6 +95,8 @@ int xfs_scrub_setup_inode_bmap_data(struct xfs_scrub_context *sc,
 				    struct xfs_inode *ip);
 int xfs_scrub_setup_directory(struct xfs_scrub_context *sc,
 			      struct xfs_inode *ip);
+int xfs_scrub_setup_xattr(struct xfs_scrub_context *sc,
+			  struct xfs_inode *ip);
 
 void xfs_scrub_ag_free(struct xfs_scrub_context *sc, struct xfs_scrub_ag *sa);
 int xfs_scrub_ag_init(struct xfs_scrub_context *sc, xfs_agnumber_t agno,

fs/xfs/scrub/scrub.c

@@ -158,6 +158,10 @@ xfs_scrub_teardown(
 			iput(VFS_I(sc->ip));
 		sc->ip = NULL;
 	}
+	if (sc->buf) {
+		kmem_free(sc->buf);
+		sc->buf = NULL;
+	}
 	return error;
 }
 
@@ -231,6 +235,10 @@ static const struct xfs_scrub_meta_ops meta_scrub_ops[] = {
 		.setup	= xfs_scrub_setup_directory,
 		.scrub	= xfs_scrub_directory,
 	},
+	{ /* extended attributes */
+		.setup	= xfs_scrub_setup_xattr,
+		.scrub	= xfs_scrub_xattr,
+	},
 };
 
 /* This isn't a stable feature, warn once per day. */

fs/xfs/scrub/scrub.h

@@ -59,6 +59,7 @@ struct xfs_scrub_context {
 	const struct xfs_scrub_meta_ops	*ops;
 	struct xfs_trans		*tp;
 	struct xfs_inode		*ip;
+	void				*buf;
 	uint				ilock_flags;
 	bool				try_harder;
 
@@ -83,5 +84,6 @@ int xfs_scrub_bmap_data(struct xfs_scrub_context *sc);
 int xfs_scrub_bmap_attr(struct xfs_scrub_context *sc);
 int xfs_scrub_bmap_cow(struct xfs_scrub_context *sc);
 int xfs_scrub_directory(struct xfs_scrub_context *sc);
+int xfs_scrub_xattr(struct xfs_scrub_context *sc);
 
 #endif	/* __XFS_SCRUB_SCRUB_H__ */

fs/xfs/xfs_attr.h

@@ -48,6 +48,8 @@ struct xfs_attr_list_context;
 #define ATTR_KERNOTIME	0x1000	/* [kernel] don't update inode timestamps */
 #define ATTR_KERNOVAL	0x2000	/* [kernel] get attr size only, not value */
 
+#define ATTR_INCOMPLETE	0x4000	/* [kernel] return INCOMPLETE attr keys */
+
 #define XFS_ATTR_FLAGS \
 	{ ATTR_DONTFOLLOW, 	"DONTFOLLOW" }, \
 	{ ATTR_ROOT,		"ROOT" }, \
@@ -56,7 +58,8 @@ struct xfs_attr_list_context;
 	{ ATTR_CREATE,		"CREATE" }, \
 	{ ATTR_REPLACE,		"REPLACE" }, \
 	{ ATTR_KERNOTIME,	"KERNOTIME" }, \
-	{ ATTR_KERNOVAL,	"KERNOVAL" }
+	{ ATTR_KERNOVAL,	"KERNOVAL" }, \
+	{ ATTR_INCOMPLETE,	"INCOMPLETE" }
 
 /*
  * The maximum size (into the kernel or returned from the kernel) of an

fs/xfs/xfs_attr_list.c

@@ -407,7 +407,8 @@ xfs_attr3_leaf_list_int(
 			cursor->offset = 0;
 		}
 
-		if (entry->flags & XFS_ATTR_INCOMPLETE)
+		if ((entry->flags & XFS_ATTR_INCOMPLETE) &&
+		    !(context->flags & ATTR_INCOMPLETE))
 			continue;		/* skip incomplete entries */
 
 		if (entry->flags & XFS_ATTR_LOCAL) {
@@ -583,6 +584,10 @@ xfs_attr_list(
 	    (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.
 	 */