From:   "Nakajima, Jun" <jun.nakajima@intel.com>
To:     Thomas Gleixner <tglx@linutronix.de>
CC:     "Liu, Jing2" <jing2.liu@intel.com>,
        Paolo Bonzini <pbonzini@redhat.com>,
        LKML <linux-kernel@vger.kernel.org>,
        "x86@kernel.org" <x86@kernel.org>,
        "kvm@vger.kernel.org" <kvm@vger.kernel.org>,
        Dave Hansen <dave.hansen@linux.intel.com>,
        Arjan van de Ven <arjan@linux.intel.com>,
        Jing Liu <jing2.liu@linux.intel.com>,
        "seanjc@google.com" <seanjc@google.com>,
        "Cooper, Andrew" <andrew.cooper3@citrix.com>,
        "Bae, Chang Seok" <chang.seok.bae@intel.com>
Subject: Re: Thoughts of AMX KVM support based on latest kernel
Thread-Topic: Thoughts of AMX KVM support based on latest kernel
Thread-Index: AdfWMJGMz5/jeSLQRn+nYCX+7Qj8nwE7nDEAABP7RQA=
Date:   Wed, 17 Nov 2021 04:52:35 +0000
Message-ID: <16BF8BE6-B7B1-4F3E-B972-9D82CD2F23C8@intel.com>
References: <BYAPR11MB325685AB8E3DFD245846F854A9939@BYAPR11MB3256.namprd11.prod.outlook.com>
 <878rxn6h6t.ffs@tglx>
In-Reply-To: <878rxn6h6t.ffs@tglx>
Accept-Language: en-US
Content-Language: en-US
x-ms-exchange-antispam-messagedata-chunkcount: 1
x-ms-exchange-antispam-messagedata-0: =?us-ascii?Q?3qJVlhdb5Py/Vjkfp7AstBHDqom/00ItiFhctTLYVv/rCdCO5WvoxPZIx4c2?=
 =?us-ascii?Q?B2KEcVB3SWVZIpqPmYGphJfJ9wYkYQs4Xk8H8yiXbc+a0rVoCFaBKbTQ8WvM?=
 =?us-ascii?Q?wFA5X5vU/zUANVKVaQCzgMGBjbO8TTkAHRzbwF0zxX7dYxaPuCGdbYgwj1Nx?=
 =?us-ascii?Q?wlTRaGgmiJbvAW061hVAticoLMnQyJWuW3+e1BDHCV4JeuQwr5Ws4V/M/PG9?=
 =?us-ascii?Q?dX/SSQrBSZXknOCxMXcZWWgUqW2/uwsSt/ghNCrO0/PJx+ozj6vgoQWmkNVv?=
 =?us-ascii?Q?qJjn42zDPy+RFeulRZNJ/BQpWVtxrg9vB3uaX94BCV8pyxnqav7JyyiVmvVy?=
 =?us-ascii?Q?fIpmcUnnvD0K2Li837nWmg++9g6hegIFEw4xm0BIX/ildV/RweCGZTpg3LKn?=
 =?us-ascii?Q?B+NC8eJ+i9cFBnBwkaGlzrx77eANiUNTIfog2KRAtbc1i5nVMvTRey1tkx36?=
 =?us-ascii?Q?3XKkm37I85j5rd/CJW+0IViK9qTLWStiV3c/v6gY5TBh5qhHYcyBxwIfkAcZ?=
 =?us-ascii?Q?OC4H3/ajx3vwnjkA/An7rcICP/KSOMknOI/ZzSfUZ5BnYKGpPU/C4ZZwkehE?=
 =?us-ascii?Q?vj78g2UqS+UP+Ko06VIqI876VCKKPBsiHbeenDD6UAGjKU72gG7+xIMI2xQ6?=
 =?us-ascii?Q?J2n0bH8fDP1Vq++fw07x1dApUa/OO/jUC0jQU1F2nxGXqbfHGP/QyuWviM8E?=
 =?us-ascii?Q?+NmbCAOxSTWlpNc5xdSjVaiU0QiTT4BEtX2nFW/VbqyaaSHlzhVWV9n8ImB3?=
 =?us-ascii?Q?OTpHts9kvu2eOwnduNYqZwJbpNl77kDtYG6CpgRUWytzFUJY6og+FMsGAZHf?=
 =?us-ascii?Q?Su7VaLokN3rayB39N91pfvNmUZurIKQ6epCOg3XXombeQJjJznlvRq0X0z8T?=
 =?us-ascii?Q?00vsjflgaJDmJHtKU3xmhoN0YyD0NGTDDaNpzprrhjVE7jqXEBvGNmlW8+Oh?=
 =?us-ascii?Q?YOEUl+dNqY303q2sr+Mlwi5wIqqZ7J46fiH4ZO0wp7xMrnnYUAkWtI8/vPOT?=
 =?us-ascii?Q?t81anc/sKWS9FnkFXixa0rUtPcbI5+E8jwZz6DZv/WerN4koY9/+Bg4xtv5v?=
 =?us-ascii?Q?rO/U8PlsDFf+hUHaMhbQwkbAp2GX3Ae919JCYp3f8SxuIuJpAHWmBKM1MEtu?=
 =?us-ascii?Q?y/l/PxkTEruZ7GlmQIDrbkHaUDkl1mElw1RKp9/ihoAefHTetzihUsBNWVNZ?=
 =?us-ascii?Q?1cCKoE5tLlmON4ZPolITKwL9jBku1mudPLDouvFfZTQ8O34MEsIYo2/VRNBN?=
 =?us-ascii?Q?AQffC+oRS1qlF7SSjFHoBUD4ost5Tzra27vb4FsZ4FnIBrzvQyie82SobvZD?=
 =?us-ascii?Q?GD8EH43J91SV8dX4xDy5L4Qq/teiw/gPVszB1OKF+990cxV2009Ls+pql5ov?=
 =?us-ascii?Q?UFHkusu/0Tmm9N650Xzht676VkAc0mRf0msOo2dpMv1LaKT7WI3XxHjA0Zgt?=
 =?us-ascii?Q?nIbsojKSoAWZfVtBHqRIdsqjsO5RwcUnR7DcB26OCfIaN7cvsRNOkU1tosla?=
 =?us-ascii?Q?Qq0rtGRD/nTxI6XbCH+be2702XPB8mHS7KBReFo1FRdXtIj04bmJWg/oQrlF?=
 =?us-ascii?Q?QMKacYIIDOxyMgA4kIF/Zh+mgj/2YzBeY36icN1o4liEyieXPoJtZiEA3pQZ?=
 =?us-ascii?Q?tPrQlimtbS78k2ixOABlT4E=3D?=
Content-Type: text/plain; charset="us-ascii"
Content-ID: <CAB01BE743E22444946D4A12DC08887A@namprd11.prod.outlook.com>
Content-Transfer-Encoding: quoted-printable
MIME-Version: 1.0
X-MS-Exchange-CrossTenant-AuthAs: Internal
X-MS-Exchange-CrossTenant-AuthSource: BY5PR11MB4435.namprd11.prod.outlook.com
X-MS-Exchange-CrossTenant-Network-Message-Id: 396722d9-d8d3-4d0c-1151-08d9a9861311
X-MS-Exchange-CrossTenant-originalarrivaltime: 17 Nov 2021 04:52:35.8874
 (UTC)
X-MS-Exchange-CrossTenant-fromentityheader: Hosted
X-MS-Exchange-CrossTenant-id: 46c98d88-e344-4ed4-8496-4ed7712e255d
X-MS-Exchange-CrossTenant-mailboxtype: HOSTED
X-MS-Exchange-CrossTenant-userprincipalname: 04W6i89PFBpHJqiKoKYEJ1+NrPHt5CfSSpsz7OI1HEVRVjmgv37OdLGa/NFUV7OGQJZ6SDZsGNWPOmn5MHDZsA==
X-MS-Exchange-Transport-CrossTenantHeadersStamped: BYAPR11MB3669
X-OriginatorOrg: intel.com
Precedence: bulk
List-ID: <linux-kernel.vger.kernel.org>
X-Mailing-List: linux-kernel@vger.kernel.org

>=20
> On Nov 16, 2021, at 11:20 AM, Thomas Gleixner <tglx@linutronix.de> wrote:
>=20
> Jing,
>=20
> On Wed, Nov 10 2021 at 13:01, Liu, Jing2 wrote:
>=20
> more thoughts.
>=20
>> Once we start passthrough the XFD MSR, we need to save/restore
>> them at VM exit/entry time. If we immediately resume the guest
>> without enabling interrupts/preemptions (exit fast-path), we have no
>> issues. We don't need to save the MSR.
>=20
> Correct.
>=20
>> The question is how the host XFD MSR is restored while control is in
>> KVM.
>>=20
>> The XSAVE(S) instruction saves the (guest) state component[x] as 0 or
>> doesn't save when XFD[x] !=3D 0. Accordingly, XRSTOR(S) cannot restore
>> that (guest state). And it is possible that XFD !=3D 0 and the guest is =
using
>> extended feature at VM exit;
>=20
> You mean on creative guests which just keep AMX state alive and set
> XFD[AMX] =3D 1 to later restore it to XFD[AMX] =3D 0?


Typically a (usual) guest saves the AMX state for the previous process and =
sets XFD[AMX] =3D 1 for the next at context switch time, and a VM exit can =
happen anytime, e.g. right after XFD[AMX] =3D 1.=20
But this case is okay because the state is already saved by the guest.

If a (creative) guest wants to set XFD[AMX] =3D 1 for fun while keeping AMX=
 state alive without saving the AXM state, it may lose the state after VM e=
xit/entry. I think the right thing to do is to avoid such programming in th=
e first place. Let me find out if we can add such notes in the programming =
references.


---=20
Jun