linux-kernel - Re: [RFC][PATCH] KVM: Remove HIGH_RES

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: <ZtgNqv1r7S738osp@freefall.freebsd.org>
Date: Wed, 4 Sep 2024 07:35:06 +0000
From: Suleiman Souhlal <ssouhlal@...ebsd.org>
To: Sean Christopherson <seanjc@...gle.com>
Cc: Steven Rostedt <rostedt@...dmis.org>,
	LKML <linux-kernel@...r.kernel.org>, kvm@...r.kernel.org,
	Paolo Bonzini <pbonzini@...hat.com>,
	Thomas Gleixner <tglx@...utronix.de>,
	"x86@...nel.org" <x86@...nel.org>,
	Joel Fernandes <joel@...lfernandes.org>,
	Vineeth Pillai <vineeth@...byteword.org>,
	Borislav Petkov <bp@...en8.de>,
	Anna-Maria Behnsen <anna-maria@...utronix.de>,
	Peter Zijlstra <peterz@...radead.org>,
	Viresh Kumar <viresh.kumar@...aro.org>,
	Frederic Weisbecker <fweisbec@...il.com>, suleiman@...gle.com
Subject: Re: [RFC][PATCH] KVM: Remove HIGH_RES_TIMERS dependency

On Wed, Aug 28, 2024 at 12:34:26PM -0700, Sean Christopherson wrote:
> On Wed, Aug 21, 2024, Steven Rostedt wrote:
> > From: Steven Rostedt <rostedt@...dmis.org>
> > 
> > Commit 92b5265d38f6a ("KVM: Depend on HIGH_RES_TIMERS") added a dependency
> > to high resolution timers with the comment:
> > 
> >     KVM lapic timer and tsc deadline timer based on hrtimer,
> >     setting a leftmost node to rb tree and then do hrtimer reprogram.
> >     If hrtimer not configured as high resolution, hrtimer_enqueue_reprogram
> >     do nothing and then make kvm lapic timer and tsc deadline timer fail.
> > 
> > That was back in 2012, where hrtimer_start_range_ns() would do the
> > reprogramming with hrtimer_enqueue_reprogram(). But as that was a nop with
> > high resolution timers disabled, this did not work. But a lot has changed
> > in the last 12 years.
> > 
> > For example, commit 49a2a07514a3a ("hrtimer: Kick lowres dynticks targets on
> > timer enqueue") modifies __hrtimer_start_range_ns() to work with low res
> > timers. There's been lots of other changes that make low res work.
> > 
> > I added this change to my main server that runs all my VMs (my mail
> > server, my web server, my ssh server) and disabled HIGH_RES_TIMERS and the
> > system has been running just fine for over a month.
> > 
> > ChromeOS has tested this before as well, and it hasn't seen any issues with
> > running KVM with high res timers disabled.
> 
> Can you provide some background on why this is desirable, and what the effective
> tradeoffs are?  Mostly so that future users have some chance of making an
> informed decision.  Realistically, anyone running with HIGH_RES_TIMERS=n is likely
> already aware of the tradeoffs, but it'd be nice to capture the info here.

We have found that disabling HR timers saves power without degrading
the user experience too much.

> 
> > Signed-off-by: Steven Rostedt (Google) <rostedt@...dmis.org>
> > ---
> >  arch/x86/kvm/Kconfig | 1 -
> >  1 file changed, 1 deletion(-)
> > 
> > diff --git a/arch/x86/kvm/Kconfig b/arch/x86/kvm/Kconfig
> > index 472a1537b7a9..c65127e796a9 100644
> > --- a/arch/x86/kvm/Kconfig
> > +++ b/arch/x86/kvm/Kconfig
> > @@ -19,7 +19,6 @@ if VIRTUALIZATION
> >  
> >  config KVM
> >  	tristate "Kernel-based Virtual Machine (KVM) support"
> > -	depends on HIGH_RES_TIMERS
> 
> I did some very basic testing and nothing exploded on me either.  So long as
> nothing in the host catches fire, I don't see a good reason to make high resolution
> timers a hard requirement.
> 
> My only concern is that this could, at least in theory, result in people
> unintentionally breaking their setups, but that seems quite unlikely.
> 
> One thought would be to require the user to enable EXPERT in order to break the
> HIGH_RES_TIMERS dependency.  In practice, I doubt that will be much of a deterrent
> since (IIRC) many distros ship with EXPERT=y.  But it would at least document that
> using KVM x86 without HIGH_RES_TIMERS may come with caveats.  E.g.
> 
> 	depends on HIGH_RES_TIMERS || EXPERT

This sounds like a good compromise.

-- Suleiman