linux-kernel - Re: [RFC: timer passthrough 0/9] Support timer passthrough for VM

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: <X+pqH77bs9nyhK8w@google.com>
Date:   Mon, 28 Dec 2020 15:28:31 -0800
From:   Sean Christopherson <seanjc@...gle.com>
To:     Zhimin Feng <fengzhimin@...edance.com>
Cc:     zhouyibo@...edance.com, zhanghaozhong@...edance.com,
        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>,
        Thomas Gleixner <tglx@...utronix.de>,
        Ingo Molnar <mingo@...hat.com>, Borislav Petkov <bp@...en8.de>,
        x86@...nel.org, "H. Peter Anvin" <hpa@...or.com>,
        Frederic Weisbecker <fweisbec@...il.com>, kvm@...r.kernel.org,
        linux-kernel@...r.kernel.org
Subject: Re: [RFC: timer passthrough 0/9] Support timer passthrough for VM

On Mon, Dec 28, 2020, Zhimin Feng wrote:
> The main motivation for this patch is to improve the performance of VM.

Do you have numbers that show when this improves performance, and by how much?

This adds hundreds of cycles of overhead (VMWRITEs, WRMSRs, RDMSRs, etc...) to
_every_ VM-Exit roundtrip, and the timer IRQ still incurs a VM-Exit.  It's not
obvious that this is a net win, and it adds a fair amount of complexity and
subtlety, e.g. there are multiple corner cases this gets wrong.  I suspect
you'll have a tough time getting this reviewed, let alone merged, without hard
numbers to justify the complexity and review effort.