linux-kernel - Re: [PATCH 10/11] KVM: Disallow read-only memory for x86 TDX

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: <YY6b4n8xPaKspoNI@google.com>
Date:   Fri, 12 Nov 2021 16:52:50 +0000
From:   Sean Christopherson <seanjc@...gle.com>
To:     Xiaoyao Li <xiaoyao.li@...el.com>
Cc:     Paolo Bonzini <pbonzini@...hat.com>,
        Vitaly Kuznetsov <vkuznets@...hat.com>,
        Wanpeng Li <wanpengli@...cent.com>,
        Jim Mattson <jmattson@...gle.com>,
        Joerg Roedel <joro@...tes.org>, erdemaktas@...gle.com,
        Connor Kuehl <ckuehl@...hat.com>, x86@...nel.org,
        linux-kernel@...r.kernel.org, kvm@...r.kernel.org,
        isaku.yamahata@...el.com, Kai Huang <kai.huang@...el.com>
Subject: Re: [PATCH 10/11] KVM: Disallow read-only memory for x86 TDX

On Fri, Nov 12, 2021, Xiaoyao Li wrote:
> From: Isaku Yamahata <isaku.yamahata@...el.com>
> 
> TDX doesn't expose permission bits to the VMM in the SEPT tables, i.e.,
> doesn't support read-only private memory.
> 
> Introduce kvm_arch_support_readonly_mem(), which returns true except for
> x86. x86 has its own implementation based on vm_type that returns faluse
> for TDX VM.
> 
> Propagate it to KVM_CAP_READONLY_MEM to allow reporting on a per-VM
> basis.

Assuming KVM gains support for private memslots (or memslots that _may_ be mapped
private), this is incorrect, the restriction on read-only memory only applies to
private memory.  Userspace should still be allowed to create read-only shared memory.
Ditto for dirty-logging in the next patch.

When this patch was originally created, it was "correct" because there was no
(proposed) concept of a private memslot or of a memslot that can be mapped private.

So these two patches at least need to wait until KVM has a defind ABI for managing
guest private memory.