From mboxrd@z Thu Jan  1 00:00:00 1970
From: Avi Kivity <avi@redhat.com>
Subject: Re: [PATCH 08/20] target-ppc: Add hooks for handling tcg and kvm
 limitations
Date: Mon, 16 Apr 2012 15:31:43 +0300
Message-ID: <4F8C112F.1070502@redhat.com>
References: <1334515144-26485-1-git-send-email-afaerber@suse.de> <1334515144-26485-9-git-send-email-afaerber@suse.de> <4F8BF0D0.2090109@redhat.com> <4F8BF242.2080807@suse.de>
Mime-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: QUOTED-PRINTABLE
Cc: qemu-devel@nongnu.org, qemu-ppc@nongnu.org, blauwirbel@gmail.com,
	agraf@suse.de, David Gibson <david@gibson.dropbear.id.au>,
	"Marcelo Tosatti (supporter:Overall)" <mtosatti@redhat.com>,
	open@suse.de, "list@suse.de:Overall" <kvm@vger.kernel.org>
To: =?UTF-8?B?QW5kcmVhcyBGw6RyYmVy?= <afaerber@suse.de>
Return-path: <kvm-owner@vger.kernel.org>
Received: from mx1.redhat.com ([209.132.183.28]:45367 "EHLO mx1.redhat.com"
	rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP
	id S1753627Ab2DPMb5 (ORCPT <rfc822;kvm@vger.kernel.org>);
	Mon, 16 Apr 2012 08:31:57 -0400
In-Reply-To: <4F8BF242.2080807@suse.de>
Sender: kvm-owner@vger.kernel.org
List-ID: <kvm.vger.kernel.org>

On 04/16/2012 01:19 PM, Andreas F=C3=A4rber wrote:
> Am 16.04.2012 12:13, schrieb Avi Kivity:
> > On 04/15/2012 09:38 PM, Andreas F=C3=A4rber wrote:
> >> From: David Gibson <david@gibson.dropbear.id.au>
> >>
> >> On target-ppc, our table of CPU types and features encodes the fea=
tures as
> >> found on the hardware, regardless of whether these features are ac=
tually
> >> usable under TCG or KVM.  We already have cases where the informat=
ion from
> >> the cpu table must be fixed up to account for limitations in the e=
mulation
> >> method we're using.  e.g. TCG does not support the DFP and VSX ins=
tructions
> >> and KVM needs different numbering of the CPUs in order to tell it =
the
> >> correct thread to core mappings.
> >>
> >> This patch cleans up these hacks to handle emulation limitations b=
y
> >> consolidating them into a pair of functions specifically for the p=
urpose.
> >>
> >>      env->check_pow =3D def->check_pow;
> >> +
> >> +    if (kvm_enabled()) {
> >> +        if (kvmppc_fixup_cpu(env) !=3D 0) {
> >> +            fprintf(stderr, "Unable to virtualize selected CPU wi=
th KVM\n");
> >> +            exit(1);
> >> +        }
> >> +    } else {
> >> +        if (ppc_fixup_cpu(env) !=3D 0) {
> >> +            fprintf(stderr, "Unable to emulate selected CPU with =
TCG\n");
> >> +            exit(1);
> >> +        }
> >> +    }
> >> +
> >>
> >=20
> > I wish we had an error like that for x86.  Right now we silently ad=
d and
> > drop cpu features.
>
> x86 has an ,+enforce option for that, doesn't it?
>

So it does.  And one day it will be the default.

--=20
error compiling committee.c: too many arguments to function