Re: [PATCH 2/3] xfs: replace do_mod with native operations

["Darrick J. Wong" <darrick.wong@oracle.com>]

On Thu, Jun 07, 2018 at 07:42:05AM -0400, Brian Foster wrote:
> On Thu, Jun 07, 2018 at 03:27:50PM +1000, Dave Chinner wrote:
> > From: Dave Chinner <dchinner@redhat.com>
> > 
> > do_mod() is a hold-over from when we have different sizes for file
> > offsets and and other internal values for 40 bit XFS filesystems.
> > Hence depending on build flags variables passed to do_mod() could
> > change size. We no longer support those small format filesystems and
> > hence everything is of fixed size theses days, even on 32 bit
> > platforms.
> > 
> > As such, we can convert all the do_mod() callers to platform
> > optimised modulus operations as defined by linux/math64.h.
> > Individual conversions depend on the types of variables being used.
> > 
> > Signed-Off-By: Dave Chinner <dchinner@redhat.com>
> > ---
> >  fs/xfs/libxfs/xfs_bmap.c | 37 +++++++++++++++++++++++--------------
> >  fs/xfs/xfs_bmap_util.c   | 14 +++++++++-----
> >  fs/xfs/xfs_inode.c       |  2 +-
> >  fs/xfs/xfs_iomap.h       |  4 ++--
> >  fs/xfs/xfs_linux.h       | 19 -------------------
> >  fs/xfs/xfs_log_recover.c | 32 +++++++++++++++++++++++++-------
> >  fs/xfs/xfs_rtalloc.c     | 15 +++++++++++----
> >  7 files changed, 71 insertions(+), 52 deletions(-)
> > 
> ...
> > diff --git a/fs/xfs/xfs_log_recover.c b/fs/xfs/xfs_log_recover.c
> > index 7d897c58b0c8..4405ff21f9a9 100644
> > --- a/fs/xfs/xfs_log_recover.c
> > +++ b/fs/xfs/xfs_log_recover.c
> > @@ -1235,6 +1235,25 @@ xlog_verify_head(
> >  				be32_to_cpu((*rhead)->h_size));
> >  }
> >  
> > +/*
> > + * We need to make sure we handle log wrapping properly, so we can't use teh
> > + * calculated logbno directly. Make sure it wraps to teh correct bno inside teh
> > + * log.
> > + *
> 
> s/teh/the/
> 
> > + * The log is limited to 32 bit sizes, so we use the appropriate modulus
> > + * operation here and cast it back to a 64 bit daddr on return.
> > + */
> > +static inline xfs_daddr_t
> > +xlog_wrap_logbno(
> > +	struct xlog		*log,
> > +	xfs_daddr_t		bno)
> > +{
> > +	int			mod;
> > +
> > +	div_s64_rem(bno, log->l_logBBsize, &mod);
> > +	return mod;
> > +}
> > +
> >  /*
> >   * Check whether the head of the log points to an unmount record. In other
> >   * words, determine whether the log is clean. If so, update the in-core state
> ...
> > diff --git a/fs/xfs/xfs_rtalloc.c b/fs/xfs/xfs_rtalloc.c
> > index 80bbfe604ce0..776502a5dcb7 100644
> > --- a/fs/xfs/xfs_rtalloc.c
> > +++ b/fs/xfs/xfs_rtalloc.c
> ...
> > @@ -1262,8 +1266,11 @@ xfs_rtpick_extent(
> >  		resid = seq - (1ULL << log2);
> >  		b = (mp->m_sb.sb_rextents * ((resid << 1) + 1ULL)) >>
> >  		    (log2 + 1);
> > -		if (b >= mp->m_sb.sb_rextents)
> > -			b = do_mod(b, mp->m_sb.sb_rextents);
> > +		if (b >= mp->m_sb.sb_rextents) {
> > +			xfs_rtblock_t mod;
> > +			div64_u64_rem(b, mp->m_sb.sb_rextents, &mod);
> > +			b = mod;
> > +		}
> 
> Shouldn't we be able to do 'div64_u64_rem(b, mp->m_sb.sb_rextents, &b)'
> here? Otherwise looks fine:

I think Dave is trying to avoid aliasing the arguments in case
div64_u64_rem ever gets turned into a horrible macro (or over-optimized
by gcc)

--D

> 
> Reviewed-by: Brian Foster <bfoster@redhat.com>
> 
> >  		if (b + len > mp->m_sb.sb_rextents)
> >  			b = mp->m_sb.sb_rextents - len;
> >  	}
> > -- 
> > 2.17.0
> > 
> > --
> > 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
> --
> 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