[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] [PATCH] xen: arm: arm64: Adding VFP save/restore support.



On Fri, 2014-02-07 at 15:28 +0000, George Dunlap wrote:
> On Thu, Feb 6, 2014 at 1:11 PM, Ian Campbell <Ian.Campbell@xxxxxxxxxx> wrote:
> > On Thu, 2014-02-06 at 13:08 +0000, Julien Grall wrote:
> >>
> >> On 06/02/14 12:57, Ian Campbell wrote:
> >> > On Thu, 2014-02-06 at 12:44 +0000, Julien Grall wrote:
> >> >>> +                 :: "r" ((char *)(&v->arch.vfp.fpregs)): "memory");
> >> >>
> >> >> I remember we had a discussion when I have implemented vfp context
> >> >> switch for arm32 for the memory constraints
> >> >> (http://lists.xen.org/archives/html/xen-devel/2013-06/msg00110.html).
> >> >>
> >> >> I think you should use "=Q" also here to avoid cloberring the whole 
> >> >> memory.
> >> >
> >> > Yes, I forgot to say: I think getting something in now is the priority,
> >> > which is why I committed it, but this should be tightened up, probably
> >> > for 4.5 unless the difference is benchmarkable.
> >>
> >> The fix is very simple (a matter of 2 lines changes). I would prefer to
> >> delay this patch for a couple of days and having a correct
> >> implementation from the beginning, so we will not forgot to change the
> >> code for Xen 4.5.
> >
> > And I would rather close this rather large hole right now and not wait
> > for two more days when we are looking at doing what might be the final
> > rc soon.
> >
> > I had already applied before you said anything, so the point is moot.
> 
> But just for future reference: Be on your guard against the "hurry and
> get in" mentality; it's likely to be counterproductive.  I'd rather
> have *more* care taken with the patches at this point than normally.

This was not a case of a careless rush to get something in. It was my
opinion that the minor shortcoming wasn't worth waiting for compared
with the benefit of getting that patch in sooner rather than later.

Note that the patch was not in any way wrong.

> If that means making a case to delay the release, that's what it will
> take.  This would obviously have been a complete blocker -- there's no
> way we could in good release without FPU support for guests. :-)
> 
>  -George



_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel


 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.