linux-kernel - Re: [PATCH v3] KVM: SEV: Mark nested locking of vcpu->lock

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

Message-ID: <b63a373a-be17-33ca-11f6-2dd36162844b@redhat.com>
Date:   Fri, 29 Apr 2022 15:44:42 +0200
From:   Paolo Bonzini <pbonzini@...hat.com>
To:     Hillf Danton <hdanton@...a.com>
Cc:     Peter Gonda <pgonda@...gle.com>,
        Sean Christopherson <seanjc@...gle.com>,
        LKML <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH v3] KVM: SEV: Mark nested locking of vcpu->lock

On 4/29/22 13:40, Hillf Danton wrote:
> To avoid acquiring more than one mutexes at the same time, add a completion
> in paralle to the mutex in question and ask mutex locker for non-migration
> purpose to take a nap on completion instead of mutex.

Acquiring more than one mutex at a time is perfectly fine.  It also 
cannot get into philosophers-problem-like deadlocks because 1) it's got 
a well-defined iteration order 2) it's protected by sev_lock_two_vms.

Just, lockdep that has to be convinced that it is fine.  I don't shy 
away from complex locking schemes if they're necessary, but here it 
absolutely isn't.

Paolo