On Mon, 30 Jul 2012, Christoph Hellwig wrote:
> Date: Mon, 30 Jul 2012 22:33:36 -0400
> From: Christoph Hellwig <hch@xxxxxxxxxxxxx>
> To: Dave Chinner <david@xxxxxxxxxxxxx>
> Cc: Tomas Racek <tracek@xxxxxxxxxx>, lczerner@xxxxxxxxxx, xfs@xxxxxxxxxxx
> Subject: Re: [PATCH] xfstests: Change fstrim behaviour to be consistent with
> upstream version
> On Tue, Jul 31, 2012 at 08:04:13AM +1000, Dave Chinner wrote:
> > If we have duplicate code (i.e. a copy of the upstream utility) or
> > the local tool can be completely replaced by the upstream tool,
> > then we should use upstream and remove the local copy completely.
> > Distros have been shipping fstrim for long enough now that most
> > people running testing on upstream kernels will have it installed...
> > Adding a _require_fstrim() function that checks for the upstream
> > version of fstrim to be installed for each test that requires it
> > would go along with this.
> I would also vote for just using the upstream util-linux fstrim. Not
> quite sure what the history was here, but it might have been that the
> xfstests one actually was the earlier version. Lukas, any opinions?
The local xfstests version was indeed the earlier version and it was
not even in the util-linux back then. So now, when we already have
fstrim in util-linux and most distributions already ship it, I do
not see any reason for maintaining the local copy anymore.
I agree that we should be using upstream fstrim and remove the local