Re: [PATCH 18/22] xfs: scrub extended attributes

From: Allison Henderson <allison.henderson@oracle.com>
To: "Darrick J. Wong" <darrick.wong@oracle.com>
Cc: linux-xfs@vger.kernel.org
Subject: Re: [PATCH 18/22] xfs: scrub extended attributes
Date: Sun, 23 Jul 2017 10:57:40 -0700	[thread overview]
Message-ID: <91fdeb68-b5ae-4829-8bc0-dfe0f9d276f0@oracle.com> (raw)
In-Reply-To: <150061202593.14732.2028384029545105215.stgit@magnolia>

Reviewed by: Allison Henderson <allison.henderson@oracle.com>

On 7/20/2017 9:40 PM, Darrick J. Wong wrote:
> From: Darrick J. Wong <darrick.wong@oracle.com>
>
> 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>
> ---
>  fs/xfs/Makefile        |    1
>  fs/xfs/libxfs/xfs_fs.h |    3 -
>  fs/xfs/scrub/attr.c    |  212 ++++++++++++++++++++++++++++++++++++++++++++++++
>  fs/xfs/scrub/common.c  |    8 ++
>  fs/xfs/scrub/common.h  |    3 +
>  fs/xfs/xfs_trace.h     |    3 -
>  6 files changed, 228 insertions(+), 2 deletions(-)
>  create mode 100644 fs/xfs/scrub/attr.c
>
>
> diff --git a/fs/xfs/Makefile b/fs/xfs/Makefile
> index c568d0d..da64bef 100644
> --- a/fs/xfs/Makefile
> +++ b/fs/xfs/Makefile
> @@ -142,6 +142,7 @@ ifeq ($(CONFIG_XFS_ONLINE_SCRUB),y)
>  xfs-y				+= $(addprefix scrub/, \
>  				   agheader.o \
>  				   alloc.o \
> +				   attr.o \
>  				   bmap.o \
>  				   btree.o \
>  				   common.o \
> diff --git a/fs/xfs/libxfs/xfs_fs.h b/fs/xfs/libxfs/xfs_fs.h
> index e646b5f..2f553ed 100644
> --- a/fs/xfs/libxfs/xfs_fs.h
> +++ b/fs/xfs/libxfs/xfs_fs.h
> @@ -498,7 +498,8 @@ struct xfs_scrub_metadata {
>  #define XFS_SCRUB_TYPE_BMBTA	14	/* attr fork block mapping */
>  #define XFS_SCRUB_TYPE_BMBTC	15	/* CoW fork block mapping */
>  #define XFS_SCRUB_TYPE_DIR	16	/* directory */
> -#define XFS_SCRUB_TYPE_MAX	16
> +#define XFS_SCRUB_TYPE_XATTR	17	/* extended attribute */
> +#define XFS_SCRUB_TYPE_MAX	17
>
>  /* i: repair this metadata */
>  #define XFS_SCRUB_FLAG_REPAIR		(1 << 0)
> diff --git a/fs/xfs/scrub/attr.c b/fs/xfs/scrub/attr.c
> new file mode 100644
> index 0000000..f6a4b59
> --- /dev/null
> +++ b/fs/xfs/scrub/attr.c
> @@ -0,0 +1,212 @@
> +/*
> + * 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_trace.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/common.h"
> +#include "scrub/dabtree.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;
> +};
> +
> +#define XFS_SCRUB_ATTR_CHECK(fs_ok) \
> +	XFS_SCRUB_DATA_CHECK(sx->sc, XFS_ATTR_FORK, args.blkno, "attr", fs_ok)
> +#define XFS_SCRUB_ATTR_OP_ERROR_GOTO(label) \
> +	XFS_SCRUB_FILE_OP_ERROR_GOTO(sx->sc, XFS_ATTR_FORK, args.blkno, "attr", &error, label)
> +/* Check that an extended attribute key can be looked up by hash. */
> +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);
> +
> +	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;
> +	XFS_SCRUB_ATTR_OP_ERROR_GOTO(fail_xref);
> +	XFS_SCRUB_ATTR_CHECK(args.valuelen == valuelen);
> +
> +fail_xref:
> +	return;
> +}
> +#undef XFS_SCRUB_ATTR_OP_ERROR_GOTO
> +#undef XFS_SCRUB_ATTR_CHECK
> +
> +/* 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);
> +	XFS_SCRUB_DA_GOTO(ds, nameidx >= hdrsize, out);
> +	XFS_SCRUB_DA_GOTO(ds, nameidx < mp->m_attr_geo->blksize, out);
> +
> +	/* 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);
> +	XFS_SCRUB_DA_CHECK(ds, (ent->flags & badflags) == 0);
> +	if (ent->flags & XFS_ATTR_LOCAL) {
> +		lentry = (struct xfs_attr_leaf_name_local *)
> +				(((char *)bp->b_addr) + nameidx);
> +		XFS_SCRUB_DA_GOTO(ds, lentry->namelen < MAXNAMELEN, out);
> +		calc_hash = xfs_da_hashname(lentry->nameval, lentry->namelen);
> +	} else {
> +		rentry = (struct xfs_attr_leaf_name_remote *)
> +				(((char *)bp->b_addr) + nameidx);
> +		XFS_SCRUB_DA_GOTO(ds, rentry->namelen < MAXNAMELEN, out);
> +		calc_hash = xfs_da_hashname(rentry->name, rentry->namelen);
> +	}
> +	XFS_SCRUB_DA_CHECK(ds, calc_hash == hash);
> +
> +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 };
> +	struct xfs_mount		*mp = sc->mp;
> +	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;
> +
> +	/* 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.sc = sc;
> +
> +	/*
> +	 * Look up every xattr in this file by name.
> +	 *
> +	 * 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);
> +	XFS_SCRUB_OP_ERROR_GOTO(sc,
> +			XFS_INO_TO_AGNO(mp, sc->ip->i_ino),
> +			XFS_INO_TO_AGBNO(mp, sc->ip->i_ino),
> +			"inode", &error, out);
> +out:
> +	return error;
> +}
> diff --git a/fs/xfs/scrub/common.c b/fs/xfs/scrub/common.c
> index 92627e9..a47c654 100644
> --- a/fs/xfs/scrub/common.c
> +++ b/fs/xfs/scrub/common.c
> @@ -686,6 +686,10 @@ xfs_scrub_teardown(
>  			IRELE(sc->ip);
>  		sc->ip = NULL;
>  	}
> +	if (sc->buf) {
> +		kmem_free(sc->buf);
> +		sc->buf = NULL;
> +	}
>  	return error;
>  }
>
> @@ -844,6 +848,10 @@ static const struct xfs_scrub_meta_fns meta_scrub_fns[] = {
>  		.setup	= xfs_scrub_setup_directory,
>  		.scrub	= xfs_scrub_directory,
>  	},
> +	{ /* extended attributes */
> +		.setup	= xfs_scrub_setup_xattr,
> +		.scrub	= xfs_scrub_xattr,
> +	},
>  };
>
>  /* Dispatch metadata scrubbing. */
> diff --git a/fs/xfs/scrub/common.h b/fs/xfs/scrub/common.h
> index 7baaa2d..1cfe0cc 100644
> --- a/fs/xfs/scrub/common.h
> +++ b/fs/xfs/scrub/common.h
> @@ -52,6 +52,7 @@ struct xfs_scrub_context {
>  	const struct xfs_scrub_meta_fns	*fns;
>  	struct xfs_trans		*tp;
>  	struct xfs_inode		*ip;
> +	void				*buf;
>  	uint				ilock_flags;
>  	bool				try_harder;
>
> @@ -221,6 +222,7 @@ SETUP_FN(xfs_scrub_setup_inode);
>  SETUP_FN(xfs_scrub_setup_inode_bmap_data);
>  SETUP_FN(xfs_scrub_setup_inode_bmap);
>  SETUP_FN(xfs_scrub_setup_directory);
> +SETUP_FN(xfs_scrub_setup_xattr);
>  #undef SETUP_FN
>
>  /* Metadata scrubbers */
> @@ -243,6 +245,7 @@ SCRUB_FN(xfs_scrub_bmap_data);
>  SCRUB_FN(xfs_scrub_bmap_attr);
>  SCRUB_FN(xfs_scrub_bmap_cow);
>  SCRUB_FN(xfs_scrub_directory);
> +SCRUB_FN(xfs_scrub_xattr);
>  #undef SCRUB_FN
>
>  #endif	/* __XFS_REPAIR_COMMON_H__ */
> diff --git a/fs/xfs/xfs_trace.h b/fs/xfs/xfs_trace.h
> index ccd27ec..fe4b313 100644
> --- a/fs/xfs/xfs_trace.h
> +++ b/fs/xfs/xfs_trace.h
> @@ -3328,7 +3328,8 @@ DEFINE_GETFSMAP_EVENT(xfs_getfsmap_mapping);
>  	{ XFS_SCRUB_TYPE_BMBTD,		"bmapbtd" }, \
>  	{ XFS_SCRUB_TYPE_BMBTA,		"bmapbta" }, \
>  	{ XFS_SCRUB_TYPE_BMBTC,		"bmapbtc" }, \
> -	{ XFS_SCRUB_TYPE_DIR,		"dir" }
> +	{ XFS_SCRUB_TYPE_DIR,		"dir" }, \
> +	{ XFS_SCRUB_TYPE_XATTR,		"xattr" }
>  DECLARE_EVENT_CLASS(xfs_scrub_class,
>  	TP_PROTO(struct xfs_inode *ip, struct xfs_scrub_metadata *sm,
>  		 int error),
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-xfs" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>