linux-kernel - Re: [PATCH v2 00/11] KVM: Support guest MAXPHYADDR

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <53dc2f58-1373-b0ef-9888-713712c6a85a@amd.com>
Date:   Mon, 22 Jun 2020 11:35:39 -0500
From:   Tom Lendacky <thomas.lendacky@....com>
To:     Paolo Bonzini <pbonzini@...hat.com>,
        Mohammed Gamal <mgamal@...hat.com>, kvm@...r.kernel.org
Cc:     linux-kernel@...r.kernel.org, vkuznets@...hat.com,
        sean.j.christopherson@...el.com, wanpengli@...cent.com,
        jmattson@...gle.com, joro@...tes.org, babu.moger@....com
Subject: Re: [PATCH v2 00/11] KVM: Support guest MAXPHYADDR < host MAXPHYADDR

On 6/22/20 10:23 AM, Paolo Bonzini wrote:
> On 22/06/20 17:08, Mohammed Gamal wrote:
>>> Also, something to consider. On AMD, when memory encryption is 
>>> enabled (via the SYS_CFG MSR), a guest can actually have a larger
>>> MAXPHYADDR than the host. How do these patches all play into that?
> 
> As long as the NPT page tables handle the guest MAXPHYADDR just fine,
> there's no need to do anything.  I think that's the case?

Yes, it does.

Thanks,
Tom

> 
> Paolo
> 
>> Well the patches definitely don't address that case. It's assumed a
>> guest VM's MAXPHYADDR <= host MAXPHYADDR, and hence we handle the case
>> where a guests's physical address space is smaller and try to trap
>> faults that may go unnoticed by the host.
>>
>> My question is in the case of guest MAXPHYADDR > host MAXPHYADDR, do we
>> expect somehow that there might be guest physical addresses that
>> contain what the host could see as reserved bits? And how'd the host
>> handle that?
>