linux-kernel - Re: [PATCH v4 13/17] kvm: i8254: Deactivate APICv when using in-kernel PIT re-injection mode.

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: <20191105084024.GA10988@rkaganb.lan>
Date:   Tue, 5 Nov 2019 08:40:28 +0000
From:   Roman Kagan <rkagan@...tuozzo.com>
To:     "Graf (AWS), Alexander" <graf@...zon.de>
CC:     Paolo Bonzini <pbonzini@...hat.com>,
        "Suthikulpanit, Suravee" <Suravee.Suthikulpanit@....com>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "kvm@...r.kernel.org" <kvm@...r.kernel.org>,
        "rkrcmar@...hat.com" <rkrcmar@...hat.com>,
        "joro@...tes.org" <joro@...tes.org>,
        "vkuznets@...hat.com" <vkuznets@...hat.com>,
        "Schoenherr, Jan H." <jschoenh@...zon.de>,
        "Raslan, KarimAllah" <karahmed@...zon.de>,
        "Lukaszewicz, Rimas" <rimasluk@...zon.com>,
        "Grimm, Jon" <Jon.Grimm@....com>
Subject: Re: [PATCH v4 13/17] kvm: i8254: Deactivate APICv when using
 in-kernel PIT re-injection mode.

On Tue, Nov 05, 2019 at 07:05:57AM +0000, Graf (AWS), Alexander wrote:
> 
> 
> > Am 04.11.2019 um 22:50 schrieb Paolo Bonzini <pbonzini@...hat.com>:
> > 
> > On 04/11/19 19:54, Suthikulpanit, Suravee wrote:
> >> I see you point.
> >> 
> >>> We can work around it by adding a global mask of inhibit reasons that
> >>> apply to the vendor, and initializing it as soon as possible in vmx.c/svm.c.
> >>> 
> >>> Then kvm_request_apicv_update can ignore reasons that the vendor doesn't
> >>> care about.
> >> 
> >> What about we enhance the pre_update_apivc_exec_ctrl() to also return 
> >> success/fail. In here, the vendor specific code can decide to update 
> >> APICv state or not.
> > 
> > That works for me, too.  Something like return false for deactivate and
> > true for activate.
> 
> I'm trying to wrap my head around how that will work with live
> migration. Do we also need to save/restore the deactivate reasons?

Nope, this is all invisible to userland.  The target will deduce the
deactivation reasons on its own from the user-visible setup like PIT
configuration, Hyper-V SynIC, etc.

Roman.