Re: [PATCH v4 2/3] xfs: introduce xfs_validate_stripe_factors()

[Gao Xiang <hsiangkao@redhat.com>]

On Wed, Oct 07, 2020 at 03:29:42PM -0700, Darrick J. Wong wrote:
> On Wed, Oct 07, 2020 at 10:04:01PM +0800, Gao Xiang wrote:
> > From: Gao Xiang <hsiangkao@redhat.com>
> > 
> > Introduce a common helper to consolidate
> > stripe validation process. Also make kernel
> > code xfs_validate_sb_common() use it first.
> 
> Please use all 72(?) columns here.

will fix.

> 
> > 
> > Signed-off-by: Gao Xiang <hsiangkao@redhat.com>
> > ---
> >  libxfs/xfs_sb.c | 54 +++++++++++++++++++++++++++++++++++++++----------
> >  libxfs/xfs_sb.h |  3 +++
> 
> These libxfs changes will have to go through the kernel first.

will send another patch together with the next version.

> 
> >  2 files changed, 46 insertions(+), 11 deletions(-)
> > 
> > diff --git a/libxfs/xfs_sb.c b/libxfs/xfs_sb.c
> > index d37d60b39a52..bd65828c844e 100644
> > --- a/libxfs/xfs_sb.c
> > +++ b/libxfs/xfs_sb.c
> > @@ -357,21 +357,13 @@ xfs_validate_sb_common(
> >  		}
> >  	}
> >  
> > -	if (sbp->sb_unit) {
> > -		if (!xfs_sb_version_hasdalign(sbp) ||
> > -		    sbp->sb_unit > sbp->sb_width ||
> > -		    (sbp->sb_width % sbp->sb_unit) != 0) {
> > -			xfs_notice(mp, "SB stripe unit sanity check failed");
> > -			return -EFSCORRUPTED;
> > -		}
> > -	} else if (xfs_sb_version_hasdalign(sbp)) {
> > +	if (!sbp->sb_unit ^ !xfs_sb_version_hasdalign(sbp)) {
> 
> Urgh, this logic makes my brain hurt.
> 
> "If the zeroness of sb_unit differs from the unsetness of the dalign
> feature"?  This might need some kind of comment, such as:
> 
> 	/*
> 	 * Either sb_unit and hasdalign are both set, or they are zero
> 	 * and not set, respectively.
> 	 */
> 	if (!sbp->sb_unit ^ !xfs_sb_version_hasdalign(sbp)) {
> 

Ok, yet I think the comment might describe failure condition (which causes
-EFSCORRUPTED) instead directly, like,

	/*
	 * Either (sb_unit and !hasdalign) or (!sb_unit and hasdalign)
	 * would imply the image is corrupted.
	 */
 	if (!sbp->sb_unit ^ !xfs_sb_version_hasdalign(sbp)) {

>
> >  		xfs_notice(mp, "SB stripe alignment sanity check failed");
> >  		return -EFSCORRUPTED;
> > -	} else if (sbp->sb_width) {
> > -		xfs_notice(mp, "SB stripe width sanity check failed");
> > -		return -EFSCORRUPTED;
> >  	}
> >  
> > +	if (!xfs_validate_stripe_factors(mp, sbp->sb_unit, sbp->sb_width, 0))
> > +		return -EFSCORRUPTED;
> >  
> >  	if (xfs_sb_version_hascrc(&mp->m_sb) &&
> >  	    sbp->sb_blocksize < XFS_MIN_CRC_BLOCKSIZE) {
> > @@ -1208,3 +1200,43 @@ xfs_sb_get_secondary(
> >  	*bpp = bp;
> >  	return 0;
> >  }
> > +
> > +/*
> > + * If sectorsize is specified, sunit / swidth must be in bytes;
> > + * or both can be in any kind of units (e.g. 512B sector or blocksize).
> > + */
> > +bool
> > +xfs_validate_stripe_factors(
> > +	struct xfs_mount	*mp,
> > +	int			sunit,
> > +	int			swidth,
> > +	int			sectorsize)
> > +{
> > +	if (sectorsize && sunit % sectorsize) {
> > +		xfs_notice(mp,
> > +"stripe unit (%d) must be a multiple of the sector size (%d)",
> > +			   sunit, sectorsize);
> > +		return false;
> > +	}
> > +
> > +	if ((sunit && !swidth) || (!sunit && swidth)) {
> > +		xfs_notice(mp,
> > +"stripe unit (%d) and width (%d) are partially valid", sunit, swidth);
> 
> I would break these into separate checks and messages.

Ok, will update in the next version.

> 
> > +		return false;
> > +	}
> > +
> > +	if (sunit > swidth) {
> > +		xfs_notice(mp,
> > +"stripe unit (%d) is too large of the stripe width (%d)", sunit, swidth);
> 
> "stripe unit (%d) is larger than the stripe width..."

Will update too.

Thanks,
Gao Xiang

> 
> --D