Re: [PATCH v6 05/10] NFSD first draft of async copy

["J. Bruce Fields" <bfields@redhat.com>]

On Fri, Feb 02, 2018 at 02:50:01PM -0500, Olga Kornievskaia wrote:
> On Fri, Jan 26, 2018 at 4:34 PM, J. Bruce Fields <bfields@fieldses.org> wrote:
> > If I understand correctly (I may be wrong), once this patch is applied a
> > COPY may fail that previously worked--because we're switching over to
> > the asynchronous copy implementation before it's actually complete.
> 
> I will have to double check this with testing but I think after this
> patch the asynchronous copy is functional but doesn't comply with the
> spec (eg., doesn't generate the unique stateid).
> 
> > Of course, that's fixed by the end of this series.  But we try to avoid
> > that situation, where functionality is temporarily broken in the middle
> > of a patch series and then fixed later.
> >
> > Options might be to squash this patch together with some of the later
> > patches.  Or go ahead and add this code but don't actually enable it
> > till later.  (E.g. arrange that the "if (!copy->cp_synchronous)" case
> > won't be taken till the last patch.  Maybe it already works that way, I
> > can't tell.)  Or maybe there's some slicker way that I don't see right
> > now.
> 
> I could do if (!copy->cp_synchronous && 0) and then add a patch that removes 0.

OK.  I still don't see the "slick" solution, so let's go with that.

--b.