Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Mon, 2 Jan 2023 21:02:11 +0100
From:   Borislav Petkov <bp@alien8.de>
To:     "Nikunj A. Dadhania" <nikunj@amd.com>
Cc:     David Rientjes <rientjes@google.com>, linux-kernel@vger.kernel.org,
        x86@kernel.org, kvm@vger.kernel.org, mingo@redhat.com,
        tglx@linutronix.de, dave.hansen@linux.intel.com, seanjc@google.com,
        pbonzini@redhat.com, thomas.lendacky@amd.com, michael.roth@amd.com,
        stable@kernel.org
Subject: Re: [PATCH v3] x86/sev: Add SEV-SNP guest feature negotiation support
Message-ID: <Y7M4IzNYBtfEJe6Z@zn.tnic>
References: <20230102083810.71178-1-nikunj@amd.com>
 <3169b54b-d990-7707-5ec4-cde7261318fe@google.com>
 <45487a87-764a-7ff3-292b-4a55fe29f7ba@amd.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
In-Reply-To: <45487a87-764a-7ff3-292b-4a55fe29f7ba@amd.com>
Precedence: bulk

On Mon, Jan 02, 2023 at 08:50:23PM +0530, Nikunj A. Dadhania wrote:
> >> +		/*
> >> +		 * Terminate the boot if hypervisor has enabled any feature
> >> +		 * lacking guest side implementation.
> >> +		 */
> >> +		if (sev_status & SNP_FEATURES_IMPL_REQ & ~SNP_FEATURES_PRESENT)
> >> +			sev_es_terminate(SEV_TERM_SET_GEN, GHCB_SNP_FEAT_NOT_IMPLEMENTED);
> > 
> > We can't help out by specifying which feature(s)?
> 
> The purpose of SNP_FEATURES_PRESENT is just that, at present no features that need guest 
> implementation is part of the kernel. For e.g. I will be posting patches with SecureTSC 
> enabled, that will make the following change.

I think what David means is, can we have sev_es_terminate() say exactly which
feature wasn't implemented instead of users having to dig out which one exactly
wasn't by trying to find out what their SNP_FEATURES_IMPL_REQ and
SNP_FEATURES_PRESENT masks are.

Looking at the GHCB protocol, where GHCB_SNP_FEAT_NOT_IMPLEMENTED reason code
goes is GHCBData[23:16] which is not enough... And the VMSA has SEV_FEATURES but
that's guest-only.

I guess we need a way to communicate those masks in a more user-friendly way so
that it is exactly clear because of which missing feature(s) has the guest
terminated.

Hmm.

-- 
Regards/Gruss,
    Boris.

https://people.kernel.org/tglx/notes-about-netiquette