From mboxrd@z Thu Jan  1 00:00:00 1970
From: Paolo Bonzini <pbonzini@redhat.com>
Subject: Re: [PATCH 2/2] KVM: x86: map/unmap private slots in
 __x86_set_memory_region
Date: Tue, 13 Oct 2015 18:28:22 +0200
Message-ID: <561D3126.8060402@redhat.com>
References: <1444651758-6926-1-git-send-email-pbonzini@redhat.com>
 <1444651758-6926-3-git-send-email-pbonzini@redhat.com>
 <20151013153957.GA9987@potion.brq.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,
	aderumier@odiso.com, stable@vger.kernel.org
To: =?UTF-8?B?UmFkaW0gS3LEjW3DocWZ?= <rkrcmar@redhat.com>
Return-path: <stable-owner@vger.kernel.org>
In-Reply-To: <20151013153957.GA9987@potion.brq.redhat.com>
Sender: stable-owner@vger.kernel.org
List-Id: kvm.vger.kernel.org



On 13/10/2015 17:39, Radim Kr=C4=8Dm=C3=A1=C5=99 wrote:
> 2015-10-12 14:09+0200, Paolo Bonzini:
>> Otherwise, two copies (one of them never used and thus bogus) are
>> allocated for the regular and SMM address spaces.  This breaks
>> SMM with EPT but without unrestricted guest support, because the
>> SMM copy of the identity page map is all zeros.
>=20
> (Have you found out why EPT+unrestricted didn't use the alternative S=
MM
>  mapping as well?)

Yes, that I already knew; EPT+unrestricted uses CR0.PG=3D0 directly so
it doesn't use the identity page at all.  (CR0.PG=3D0 w/o unrestricted
instead runs with CR0.PG=3D1.  CR3 load and store exits are enabled,
and the guest CR3 always points to the identity page map while the
guest runs).

>> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
>> @@ -7717,23 +7717,53 @@ void kvm_arch_sync_events(struct kvm *kvm)
>>  int __x86_set_memory_region(struct kvm *kvm, int id, gpa_t gpa, u32=
 size)
>>  {
>>  	int i, r;
>> +	u64 hva;
>> +	struct kvm_memslots *slots =3D kvm_memslots(kvm);
>> +	struct kvm_memory_slot *slot, old;
> | [...]
>> +	slot =3D &slots->memslots[slots->id_to_index[id]];
>=20
> This seems better written as
>=20
>    	slot =3D id_to_memslot(slots, id);

Gah, I could not recall the right API!  I'll fix it.

> (Made me remember that I want to refactor the memslot API ...)
>=20
> | [...]
>> +	} else {
>> +		if (!slot->npages)
>> +			return 0;
>> +
>> +		hva =3D 0;
>> +	}
>> +
>> +	old =3D *slot;
>=20
> (Assignment could be in the 'else' =3D=3D !size branch, GCC would hav=
e fun.)

It would have fun _and_ warn, which is why it's not in the else branch.=
 :)

Paolo