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

From: Paolo Bonzini <pbonzini@redhat.com>
To: Borislav Petkov <bp@alien8.de>
Cc: "Luck, Tony" <tony.luck@intel.com>,
	Jim Mattson <jmattson@google.com>, Qian Cai <cai@redhat.com>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
	"linux-tip-commits@vger.kernel.org" 
	<linux-tip-commits@vger.kernel.org>, x86 <x86@kernel.org>,
	"kvm@vger.kernel.org" <kvm@vger.kernel.org>
Subject: Re: [PATCH] x86/mce: Check for hypervisor before enabling additional error logging
Date: Tue, 10 Nov 2020 17:08:31 +0100	[thread overview]
Message-ID: <1b587b45-a5a8-2147-ae53-06d1b284ea11@redhat.com> (raw)
In-Reply-To: <20201110155013.GE9857@nazgul.tnic>

On 10/11/20 16:50, Borislav Petkov wrote:
> I was thinking of
> having a mapping between f/m/s and a list of MSRs which those models
> have - even non-architectural ones - but that's a waste of energy. Why?
> Because using the *msr_safe() variants will give you the same thing

Yes, pretty much.

>> If it makes sense to emulate certain non-architectural MSRs we can add them.
> See above - probably not worth the effort.

When we do, certain Microsoft OSes are usually involved. :)

> I'll take a look at how ugly it would become to make the majority of MSR
> accesses safe.

I think most of them already are, especially the non-architectural ones, 
because I remember going through a similar discussion a few years ago 
and Andy said basically "screw it, let's just use *_safe anywhere" as 
well.  I don't see any need to do anything but add it to your review 
checklist if you have it (and do it now for MSR_ERROR_CONTROL).

Paolo