public inbox for linux-xfs@vger.kernel.org
 help / color / mirror / Atom feed
From: Brian Foster <bfoster@redhat.com>
To: Eric Sandeen <sandeen@redhat.com>
Cc: xfs-oss <xfs@oss.sgi.com>
Subject: Re: [PATCH] xfs_metadump: obfuscate attrs on CRC fs
Date: Fri, 19 Jun 2015 09:19:39 -0400	[thread overview]
Message-ID: <20150619131938.GA12833@bfoster.bfoster> (raw)
In-Reply-To: <557B6222.4030903@redhat.com>

On Fri, Jun 12, 2015 at 05:50:10PM -0500, Eric Sandeen wrote:
> Lots of issues in xfs_metadump obfuscation of extended
> attributes with CRC filesystems; this fixes it up.
> 
> The main issues are that the headers differ, and the
> space in the remote blocks differ.
> 
> Tested with a script I'm using to do other metadump
> work; I still owe an xfstest for this and other bits.
> 
> Signed-off-by: Eric Sandeen <sandeen@redhat.com>
> ---
> 
> Brief/hacky testcase; populate w/ many attr types with:
> 
> 	MNTPT=/mount/test
> 
>         # FMT_LOCAL
>         touch $MNTPT/S_IFREG.ATTR.FMT_LOCAL
>         setfattr -n user.localattrname -v localattrvalue $MNTPT/S_IFREG.ATTR.FMT_LOCAL
> 
>         # FMT_EXTENTS
>         touch $MNTPT/S_IFREG.ATTR.FMT_EXTENTS
>         for I in `seq 1 50`; do
>                 setfattr -n user.extentattrname$I -v extentattrvalue $MNTPT/S_IFREG.ATTR.FMT_EXTENTS
>         done
> 
>         # FMT_EXTENTS with a single remote 3k value, fill with "C"
>         touch $MNTPT/S_IFREG.ATTR.FMT_EXTENTS_REMOTE3K
>         xfs_io -f -c "pwrite -S 0x43 0 3k" attrvalfile &>/dev/null
>         attr -q -s user.remotebtreeattrname $MNTPT/S_IFREG.ATTR.FMT_EXTENTS_REMOTE3K < attrvalfile
> 
>         # FMT_EXTENTS with a single remote 4k value, fill with "D"
>         touch $MNTPT/S_IFREG.ATTR.FMT_EXTENTS_REMOTE4K
>         xfs_io -f -c "pwrite -S 0x44 0 4k" attrvalfile &>/dev/null
>         attr -q -s user.remotebtreeattrname $MNTPT/S_IFREG.ATTR.FMT_EXTENTS_REMOTE4K < attrvalfile
> 
>         # FMT_BTREE
>         touch $MNTPT/S_IFREG.ATTR.FMT_BTREE
>         for I in `seq 1 1000`; do
>                 setfattr -n user.btreeattrname$I -v btreelongerattrvalue $MNTPT/S_IFREG.ATTR.FMT_BTREE
>         done
> 
> on a crc filesystem, an obfuscated metadump will show all these attrs in the clear
> with getfattr -dR $MNTPT
> 
> diff --git a/db/metadump.c b/db/metadump.c
> index 94f92bc..793f39e 100644
> --- a/db/metadump.c
> +++ b/db/metadump.c
> @@ -1268,39 +1268,54 @@ add_remote_vals(
>  	}
>  }
>  
> +/* Handle remote and leaf attributes */
>  static void
>  obfuscate_attr_block(
> -	char			*block,
> -	xfs_fileoff_t		offset)
> +	char				*block,
> +	xfs_fileoff_t			offset)
>  {
> -	xfs_attr_leafblock_t	*leaf;
> -	int			i;
> -	int			nentries;
> -	xfs_attr_leaf_entry_t 	*entry;
> -	xfs_attr_leaf_name_local_t *local;
> -	xfs_attr_leaf_name_remote_t *remote;
> +	struct xfs_attr_leafblock	*leaf;
> +	struct xfs_attr3_icleaf_hdr	hdr;
> +	int				i;
> +	int				nentries;
> +	xfs_attr_leaf_entry_t 		*entry;
> +	xfs_attr_leaf_name_local_t 	*local;
> +	xfs_attr_leaf_name_remote_t 	*remote;
> +	__uint32_t			bs = mp->m_sb.sb_blocksize;
> +
>  
>  	leaf = (xfs_attr_leafblock_t *)block;
>  
> -	if (be16_to_cpu(leaf->hdr.info.magic) != XFS_ATTR_LEAF_MAGIC) {
> +	/* Get header; accounts for crc & non-crc */
> +	xfs_attr3_leaf_hdr_from_disk(mp->m_attr_geo, &hdr, leaf);
> +

The above conversion helper assert fails on a non-leaf magic. It seems
like effectively the same check below with this change, but it might be
more appropriate to defer the leaf header conversion to after the hunk
below..?

Brian

> +	/* Remote attributes */
> +	if ((hdr.magic != XFS_ATTR_LEAF_MAGIC) &&
> +	    (hdr.magic != XFS_ATTR3_LEAF_MAGIC)) {
>  		for (i = 0; i < attr_data.remote_val_count; i++) {
> -			/* XXX: need to handle CRC headers */
>  			if (attr_data.remote_vals[i] == offset)
> -				memset(block, 0, mp->m_sb.sb_blocksize);
> +				/* magic to handle attr and attr3 */
> +				memset(block +
> +					(bs - XFS_ATTR3_RMT_BUF_SPACE(mp, bs)),
> +				      0, XFS_ATTR3_RMT_BUF_SPACE(mp, bs));
>  		}
>  		return;
>  	}
>  
> -	nentries = be16_to_cpu(leaf->hdr.count);
> +	nentries = hdr.count;
> +
>  	if (nentries * sizeof(xfs_attr_leaf_entry_t) +
> -			sizeof(xfs_attr_leaf_hdr_t) > mp->m_sb.sb_blocksize) {
> +			xfs_attr3_leaf_hdr_size(leaf) >
> +				XFS_ATTR3_RMT_BUF_SPACE(mp, bs)) {
>  		if (show_warnings)
>  			print_warning("invalid attr count in inode %llu",
>  					(long long)cur_ino);
>  		return;
>  	}
>  
> -	for (i = 0, entry = &leaf->entries[0]; i < nentries; i++, entry++) {
> +	entry = xfs_attr3_leaf_entryp(leaf);
> +
> +	for (i = 0; i < nentries; i++, entry++) {
>  		if (be16_to_cpu(entry->nameidx) > mp->m_sb.sb_blocksize) {
>  			if (show_warnings)
>  				print_warning(
> 
> _______________________________________________
> xfs mailing list
> xfs@oss.sgi.com
> http://oss.sgi.com/mailman/listinfo/xfs

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

  reply	other threads:[~2015-06-19 13:19 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-06-12 22:50 [PATCH] xfs_metadump: obfuscate attrs on CRC fs Eric Sandeen
2015-06-19 13:19 ` Brian Foster [this message]
2015-06-25 20:57 ` Eric Sandeen

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20150619131938.GA12833@bfoster.bfoster \
    --to=bfoster@redhat.com \
    --cc=sandeen@redhat.com \
    --cc=xfs@oss.sgi.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox