linux-kernel - Re: [PATCH v2 24/30] KVM: x86/mmu: Allow yielding when zapping GFNs for defunct TDP MMU root

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] [day] [month] [year] [list]

Message-ID: <YdYOg6UnRPJsznjC@google.com>
Date:   Wed, 5 Jan 2022 21:32:51 +0000
From:   Sean Christopherson <seanjc@...gle.com>
To:     David Matlack <dmatlack@...gle.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>, kvm@...r.kernel.org,
        linux-kernel@...r.kernel.org, Ben Gardon <bgardon@...gle.com>,
        Mingwei Zhang <mizhang@...gle.com>
Subject: Re: [PATCH v2 24/30] KVM: x86/mmu: Allow yielding when zapping GFNs
 for defunct TDP MMU root

On Wed, Jan 05, 2022, David Matlack wrote:
> On Thu, Dec 23, 2021 at 10:23:12PM +0000, Sean Christopherson wrote:
> > +	/*
> > +	 * Invalidate the root to prevent it from being reused by a vCPU, and
> > +	 * mark it defunct so that kvm_tdp_mmu_zap_invalidated_roots() doesn't
> > +	 * try to put a reference it didn't acquire.
> > +	 */
> > +	root->role.invalid = true;
> > +	root->tdp_mmu_defunct_root = true;
> 
> Ah ok so tdp_mmu_defunct_root indicates the root became invalid due to
> losing all its references while it was valid. This is in contrast to
> kvm_tdp_mmu_invalidate_all_roots() which marks roots invalid while they
> still have valid references.
> 
> But I wonder if tdp_mmu_defunct_root is necessary? It's only used to
> skip a put in zap_invalidated_roots. Could we instead unconditionally
> grab a reference in invalidate_all_roots and then unconditionally drop
> it?

Hmm, it's probably not necessary.  I added tdp_mmu_defunct_root before realizing
that that kvm_tdp_mmu_invalidate_all_roots() was wrong about this:

 * Roots which have a zero refcount should be skipped as
 * they're already being torn down.

IIRC, I added the second flag because I was trying to honor that (incorrect) logic,
and never reconsidered the need for a second flag once I got everything working.
The only downside would be keeping the memory for defunct roots around a wee bit
longer, and that's not a big deal.

I'll yank it out for the next version, assuming I didn't forget some detail...