From mboxrd@z Thu Jan  1 00:00:00 1970
From: Radim =?utf-8?B?S3LEjW3DocWZ?= <rkrcmar@redhat.com>
Subject: Re: [CFT PATCH v2 2/2] KVM: x86: support XSAVES usage in the host
Date: Wed, 26 Nov 2014 14:53:22 +0100
Message-ID: <20141126135322.GA4887@potion.brq.redhat.com>
References: <1416847414-22253-1-git-send-email-pbonzini@redhat.com>
 <1416847414-22253-3-git-send-email-pbonzini@redhat.com>
 <20141126120753.GA31982@potion.redhat.com>
 <5475D20A.90505@redhat.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: QUOTED-PRINTABLE
Cc: linux-kernel@vger.kernel.org, kvm@vger.kernel.org,
	wanpeng.li@linux.intel.com, namit@cs.technion.ac.il,
	hpa@linux.intel.com, Fenghua Yu <fenghua.yu@intel.com>
To: Paolo Bonzini <pbonzini@redhat.com>
Return-path: <linux-kernel-owner@vger.kernel.org>
Content-Disposition: inline
In-Reply-To: <5475D20A.90505@redhat.com>
Sender: linux-kernel-owner@vger.kernel.org
List-Id: kvm.vger.kernel.org

2014-11-26 14:13+0100, Paolo Bonzini:
> On 26/11/2014 13:07, Radim Kr=C4=8Dm=C3=A1=C5=99 wrote:
> > 2014-11-24 17:43+0100, Paolo Bonzini:
> >> Userspace is expecting non-compacted format for KVM_GET_XSAVE, but
> >> struct xsave_struct might be using the compacted format.  Convert
> >> in order to preserve userspace ABI.
> >>
> >> Likewise, userspace is passing non-compacted format for KVM_SET_XS=
AVE
> >> but the kernel will pass it to XRSTORS, and we need to convert bac=
k.
> >=20
> > Future instructions might force us to calling xsave/xrstor directly=
, so
> > we could do that even now and save the explicit conversion ...
> >=20
> > What I mean is:  we could be using the native xsave.*/xrstor.* whil=
e in
> > kernel and use xsave/xrstor for communication with userspace.
> > Hardware would take care of everything in the conversion.
> >=20
> > get_xsave =3D native_xrstor(guest_xsave);  xsave(aligned_userspace_=
buffer)
> > set_xsave =3D xrstor(aligned_userspace_buffer);  native_xsave(guest=
_xsave)
> >=20
> > Could that work?
>=20
> It could, though it is more like
>=20
>    get_fpu()
>    native_xrstor(guest_xsave)
>    xsave(buffer)
>    put_fpu()
>=20
> and vice versa.  Also, the userspace buffer is mos likely not aligned=
,
> so you need some kind of bounce buffer.  It can be done if the CPUID
> turns out to be a bottleneck, apart from that it'd most likely be slo=
wer.

Yeah, it was mostly making this code more future-proof ... it is easier
to convince xsave.h to export its structures if CPUID is the problem.
(I still see some hope for Linux, so performance isn't my primary goal.=
)

I'm quite interested in CPUID now though, so I'll try to benchmark it,
someday.