linux-kernel - Re: [PATCH 1/3] KVM: x86: introduce kvm_mmu_invalidate

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 for Android: free password hash cracker in your pocket

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

Message-ID: <bb7b1075-a4fc-e0d3-d8fd-f516d107d5e2@redhat.com>
Date:   Tue, 31 Mar 2020 14:16:27 +0200
From:   Paolo Bonzini <pbonzini@...hat.com>
To:     Vitaly Kuznetsov <vkuznets@...hat.com>,
        Sean Christopherson <sean.j.christopherson@...el.com>
Cc:     linux-kernel@...r.kernel.org, kvm@...r.kernel.org,
        Junaid Shahid <junaids@...gle.com>
Subject: Re: [PATCH 1/3] KVM: x86: introduce kvm_mmu_invalidate_gva

On 31/03/20 12:33, Vitaly Kuznetsov wrote:
>> Works for me.  My vote is for anything other than guest_mmu :-)
>
> Oh come on guys, nobody protested when I called it this way :-)

Sure I take full responsibility for that. :)

> Peronally, I don't quite like 'shadow_tdp_mmu' because it doesn't have
> any particular reference to the fact that it is a nested/L2 related
> thing (maybe it's just a shadow MMU?)

Well, nested virt is the only case in which you shadow TDP.  Both
interpretations work:

* "shadow tdp_mmu": an MMU for two-dimensional page tables that employs
shadowing

* "shadow_tdp MMU": the MMU for two-dimensional page tables.

> Also, we already have a thing
> called 'nested_mmu'... Maybe let's be bold and rename all three things,
> like
>
> root_mmu -> l1_mmu
> guest_mmu -> l1_nested_mmu
> nested_mmu -> l2_mmu (l2_walk_mmu)

I am not particularly fond of using l1/l2 outside code that specifically
deals with nested virt.  Also, l1_nested_mmu is too confusing with
respect to the current nested_mmu (likewise for root_mmu I would rename
it to guest_mmu but it would be an awful source of mental confusion as
well as semantic source code conflicts).

That said, I wouldn't mind replacing nested_mmu to something else, for
example nested_walk_mmu.

Paolo