linux-kernel - Re: [PATCH] x86/mce: Check for hypervisor before enabling additional error logging

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: <7bd98718-f800-02ef-037a-4dfc5a7d1a54@redhat.com>
Date:   Tue, 10 Nov 2020 21:37:34 +0100
From:   Paolo Bonzini <pbonzini@...hat.com>
To:     "Luck, Tony" <tony.luck@...el.com>, Borislav Petkov <bp@...en8.de>
Cc:     Jim Mattson <jmattson@...gle.com>, Qian Cai <cai@...hat.com>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "linux-tip-commits@...r.kernel.org" 
        <linux-tip-commits@...r.kernel.org>, x86 <x86@...nel.org>,
        "kvm@...r.kernel.org" <kvm@...r.kernel.org>
Subject: Re: [PATCH] x86/mce: Check for hypervisor before enabling additional
 error logging

On 10/11/20 18:52, Luck, Tony wrote:
> Look at what it is trying to do ... change the behavior of the platform w.r.t. logging
> of memory errors.  How does that make any sense for a guest ...

Logging of memory errors certainly makes sense for a guest, KVM already 
does MCE forwarding as you probably know.

The exact set of information that MSR_ERROR_CONTROL[1] adds may not make 
much sense in the case of KVM, but it may make sense for other 
hypervisors that do nothing but partition the host.  (Difficult for me 
to say since the relevant part of the SDM might as well be written in 
Klingon :)).

In any case, checking HYPERVISOR is not enough because having it clear 
is a valid configuration.  So you would still have to switch to 
{rd,wr}msrl_safe, and then checking HYPERVISOR is pointless.

Paolo

> that doesn't even
> know what memory is present on the platform. Or have guarantees that what it sees
> as memory address 0x12345678 maps to the same set of cells in a DRAM from one
> second to the next?