linux-kernel - Re: [PATCH RFC 1/6] KVM: fix guest_mode optimization in kvm_make_all_cpus

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: <20170411193124.GA395@potion>
Date:   Tue, 11 Apr 2017 21:31:24 +0200
From:   Radim Krčmář <rkrcmar@...hat.com>
To:     James Hogan <james.hogan@...tec.com>
Cc:     Paolo Bonzini <pbonzini@...hat.com>, linux-kernel@...r.kernel.org,
        kvm@...r.kernel.org, Christoffer Dall <cdall@...aro.org>,
        Andrew Jones <drjones@...hat.com>,
        Marc Zyngier <marc.zyngier@....com>,
        Christian Borntraeger <borntraeger@...ibm.com>,
        Cornelia Huck <cornelia.huck@...ibm.com>,
        Paul Mackerras <paulus@...abs.org>
Subject: Re: [PATCH RFC 1/6] KVM: fix guest_mode optimization in
 kvm_make_all_cpus_request()

2017-04-11 10:37+0100, James Hogan:
> Hi Paolo,
> 
> On Tue, Apr 11, 2017 at 01:25:04PM +0800, Paolo Bonzini wrote:
>> On 07/04/2017 05:02, James Hogan wrote:
>> > This presumably changes the behaviour on x86, from != OUTSIDE_GUEST_MODE
>> > to == IN_GUEST_MODE. so:
>> > - you'll no longer get IPIs if its in READING_SHADOW_PAGE_TABLES (which
>> >   MIPS also now uses when accessing mappings outside of guest mode and
>> >   depends upon to wait until the old mappings are no longer in use).
>> 
>> This is wrong, the purpose of READING_SHADOW_PAGE_TABLES is "kvm_flush_remote_tlbs
>> should send me an IPI, because I want to stop kvm_flush_remote_tlbs until I'm done
>> reading the page tables".
> 
> That sounds equivalent to what I meant for MIPS, i.e.
> kvm_flush_remote_tlbs() does the waiting (not the thing accessing guest
> mappings).

I agree, thanks for noticing this.  It would be a huge mistake to drop
the synchronization.