linux-kernel - Re: RFC: Getting rid of LTR in VMX

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]

Date:   Sat, 18 Feb 2017 16:19:17 +0100
From:   Paolo Bonzini <pbonzini@...hat.com>
To:     Andy Lutomirski <luto@...capital.net>,
        kvm list <kvm@...r.kernel.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        X86 ML <x86@...nel.org>, "H. Peter Anvin" <hpa@...or.com>,
        Borislav Petkov <bpetkov@...e.de>
Subject: Re: RFC: Getting rid of LTR in VMX



On 18/02/2017 04:29, Andy Lutomirski wrote:
> There's no code here because the patch is trivial, but I want to run
> the idea by you all first to see if there are any issues.
> 
> VMX is silly and forces the TSS limit to the minimum on VM exits.  KVM
> wastes lots of cycles bumping it back up to accomodate the io bitmap.
> 
> I propose that we rework this.  Add a percpu variable that indicates
> whether the TSS limit needs to be refreshed.  On task switch, if the
> new task has TIF_IO_BITMAP set, then check that flag and, if set,
> refresh TR and clear the flag.  On VMX exit, set the flag.

That's nice.  An alternative is to just use the user return notifier,
which does not need any global change.  This is even simpler and can be
used to benchmark.

> The TSS limit is (phew!) invisible to userspace, so we don't have ABI
> issues to worry about here.  We also shouldn't have security issues
> because a too-low TSS limit just results in unprivileged IO operations
> generating #GP, which is exactly what we want.
> 
> What do you all think?  I expect a speedup of a couple hundred cycles
> on each VM exit.
>