[Top] [All Lists]

Re: [PATCH 5/5] xfs: fold xfs_change_file_space into xfs_ioc_space

To: Christoph Hellwig <hch@xxxxxxxxxxxxx>
Subject: Re: [PATCH 5/5] xfs: fold xfs_change_file_space into xfs_ioc_space
From: Dave Chinner <david@xxxxxxxxxxxxx>
Date: Wed, 16 Oct 2013 08:47:33 +1100
Cc: xfs@xxxxxxxxxxx
Delivered-to: xfs@xxxxxxxxxxx
In-reply-to: <20131015153143.GA1612@xxxxxxxxxxxxx>
References: <20131012075503.370510641@xxxxxxxxxxxxxxxxxxxxxx> <20131012075640.637625757@xxxxxxxxxxxxxxxxxxxxxx> <20131014050807.GJ4446@dastard> <20131015153143.GA1612@xxxxxxxxxxxxx>
User-agent: Mutt/1.5.21 (2010-09-15)
On Tue, Oct 15, 2013 at 08:31:43AM -0700, Christoph Hellwig wrote:
> On Mon, Oct 14, 2013 at 04:08:07PM +1100, Dave Chinner wrote:
> > One question, though:
> > 
> > > + case XFS_IOC_ALLOCSP:
> > > + case XFS_IOC_ALLOCSP64:
> > > + case XFS_IOC_FREESP:
> > > + case XFS_IOC_FREESP64:
> > 
> > Should we, at this point, mark these ioctls as deprecated and
> > schedule then for removal given that we've recommended against using
> > them for the past 10 years and we have fallocate() now?
> I don't see any reason to remove them given that it's only about 15
> extra lines of code.  But if you care enough to get rid of them we
> probably need multiple years of actuall warnings emmited when used
> before actually removing them.  I would be very surprised if there
> aren't same users that wouldn't argue very vocally against their
> removal.

I'll put a significant quantity of beer on the table if anyone other
than xfstests is actually using these ioctls. In all my years of
working with XFS, I've never seen a single user of them, even on

The one person I know who was considering using XFS_IOC_ALLOCSP
convinced me (quite easily) to implement XFS_IOC_ZERO_RANGE for them
because writing all those zeros to re-initialise pre-allocated VM
images was going to be prohibitively expensive...

Anyway, it was just a thought.


Dave Chinner

<Prev in Thread] Current Thread [Next in Thread>