From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from ipmail06.adl2.internode.on.net ([150.101.137.129]:12498 "EHLO ipmail06.adl2.internode.on.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750732AbeAOVc2 (ORCPT ); Mon, 15 Jan 2018 16:32:28 -0500 Date: Tue, 16 Jan 2018 08:31:02 +1100 From: Dave Chinner Subject: Re: [PATCH v2 1/5] xfs: check sb_agblocks and sb_agblklog when validating superblock Message-ID: <20180115213102.GZ16421@dastard> References: <151579463950.8694.320025813242574491.stgit@magnolia> <151579464601.8694.5478076755339350941.stgit@magnolia> <20180115200313.GC5602@magnolia> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20180115200313.GC5602@magnolia> Sender: linux-xfs-owner@vger.kernel.org List-ID: List-Id: xfs To: "Darrick J. Wong" Cc: linux-xfs@vger.kernel.org, Brian Foster On Mon, Jan 15, 2018 at 12:03:13PM -0800, Darrick J. Wong wrote: > From: Darrick J. Wong > > Currently, we don't check sb_agblocks or sb_agblklog when we validate > the superblock, which means that we can fuzz garbage values into those > values and the mount succeeds. This leads to all sorts of UBSAN > warnings in xfs/350 since we can then coerce other parts of xfs into > shifting by ridiculously large values. > > Signed-off-by: Darrick J. Wong > --- > v2: simplify ag min/max size definitions > --- > fs/xfs/libxfs/xfs_fs.h | 7 +++++++ > fs/xfs/libxfs/xfs_sb.c | 3 +++ > 2 files changed, 10 insertions(+) > > diff --git a/fs/xfs/libxfs/xfs_fs.h b/fs/xfs/libxfs/xfs_fs.h > index fc4386a..3ab1870 100644 > --- a/fs/xfs/libxfs/xfs_fs.h > +++ b/fs/xfs/libxfs/xfs_fs.h > @@ -233,6 +233,13 @@ typedef struct xfs_fsop_resblks { > #define XFS_MAX_LOG_BLOCKS (1024 * 1024ULL) > #define XFS_MIN_LOG_BYTES (10 * 1024 * 1024ULL) > > +/* > + * Limits on sb_agblocks/sb_agblklog -- mkfs won't format AGs smaller than > + * 16MB or larger than 1TB. > + */ > +#define XFS_AG_MIN_BYTES (1ULL << 24) /* 16 MB */ > +#define XFS_AG_MAX_BYTES (1ULL << 40) /* 1 TB */ Sure, but.... > + > /* keep the maximum size under 2^31 by a small amount */ > #define XFS_MAX_LOG_BYTES \ > ((2 * 1024 * 1024 * 1024ULL) - XFS_MIN_LOG_BYTES) > diff --git a/fs/xfs/libxfs/xfs_sb.c b/fs/xfs/libxfs/xfs_sb.c > index 08e44a0..bdb4f74 100644 > --- a/fs/xfs/libxfs/xfs_sb.c > +++ b/fs/xfs/libxfs/xfs_sb.c > @@ -253,6 +253,9 @@ xfs_mount_validate_sb( > sbp->sb_inodesize != (1 << sbp->sb_inodelog) || > sbp->sb_logsunit > XLOG_MAX_RECORD_BSIZE || > sbp->sb_inopblock != howmany(sbp->sb_blocksize,sbp->sb_inodesize) || > + XFS_FSB_TO_B(mp, sbp->sb_agblocks) < XFS_AG_MIN_BYTES || > + XFS_FSB_TO_B(mp, sbp->sb_agblocks) > XFS_AG_MAX_BYTES || .... this really needs to be checked against sb_dblocks / sb_agcount. i.e. check against current filesystem size and geometry, not against the theoretical maximum. Cheers, Dave. -- Dave Chinner david@fromorbit.com