Re: [RFC PATCH 00/56] Dynamic mitigations

[Alexander Graf]

On 14.10.25 20:06, Kaplan, David wrote:
> [AMD Official Use Only - AMD Internal Distribution Only]
>
> > -----Original Message-----
> > From: Josh Poimboeuf <jpoimboe@xxxxxxxxxx>
> > Sent: Tuesday, October 14, 2025 11:29 AM
> > To: Kaplan, David <David.Kaplan@xxxxxxx>
> > Cc: Thomas Gleixner <tglx@xxxxxxxxxxxxx>; Borislav Petkov <bp@xxxxxxxxx>; Peter
> > Zijlstra <peterz@xxxxxxxxxxxxx>; Pawan Gupta
> > <pawan.kumar.gupta@xxxxxxxxxxxxxxx>; Ingo Molnar <mingo@xxxxxxxxxx>; Dave
> > Hansen <dave.hansen@xxxxxxxxxxxxxxx>; x86@xxxxxxxxxx; H . Peter Anvin
> > <hpa@xxxxxxxxx>; Alexander Graf <graf@xxxxxxxxxx>; Boris Ostrovsky
> > <boris.ostrovsky@xxxxxxxxxx>; linux-kernel@xxxxxxxxxxxxxxx
> > Subject: Re: [RFC PATCH 00/56] Dynamic mitigations
> >
> > Caution: This message originated from an External Source. Use proper caution
> > when opening attachments, clicking links, or responding.
> >
> >
> > On Mon, Oct 13, 2025 at 09:33:48AM -0500, David Kaplan wrote:
> > > Dynamic mitigations enables changing the kernel CPU security mitigations at
> > > runtime without a reboot/kexec.
> > >
> > > Previously, mitigation choices had to be made on the kernel cmdline.  With
> > > this feature an administrator can select new mitigation choices by writing
> > > a sysfs file, after which the kernel will re-patch itself based on the new
> > > mitigations.
> > >
> > > As the performance cost of CPU mitigations can be significant, selecting
> > > the right set of mitigations is important to achieve the correct balance of
> > > performance/security.
> > >
> > > Use
> > > ---
> > > As described in the supplied documentation file, new mitigations are
> > > selected by writing cmdline options to a new sysfs file.  Only cmdline
> > > options related to mitigations are recognized via this interface.  All
> > > previous mitigation-related cmdline options are ignored and selections are
> > > done based on the new options.
> > >
> > > Examples:
> > >     echo "mitigations=off" > /sys/devices/system/cpu/mitigations
> > >     echo "spectre_v2=retpoline tsa=off" > /sys/devices/system/cpu/mitigations
> > >
> > >
> > > There are several use cases that will benefit from dynamic mitigations:
> > >
> > > Use Cases
> > > ---------
> > > 1. Runtime Policy
> > >
> > > Some workflows rely on booting a generic kernel before customizing the system.
> > > cloud-init is a popular example of this where a VM is started typically with
> > > default settings and then is customized based on a customer-provided
> > > configuration file.
> > I'm not really a fan of this.  It adds complexity to some areas that are
> > already struggling with too much complexity.
> >
> > IMO this would need some REALLY strong justification, more than just
> > "hey, this makes things more convenient."
> >
> > The mitigations should be a "set it and forget it" thing.  I don't see
> > anything here which justifies the considerable maintenance burden this
> > would add for all existing and future mitigations.
> The problem is there are environments like the one outlined where you can't just 'set it and forget it' because the kernel needs it set at boot-time, but in these environments you don't know how to configure the system until much later in boot.  So you end up running with the default settings all the time, even if you don't need them.  And the default settings can have significant performance impacts in many cases.
>
> The cloud guys on this thread may be able to offer some additional color here since I believe that's where you're most likely to have this situation.


The crux of the problem here is that the kernel command line is 
difficult to influence in most cloud environments.

In the cloud, you typically start from a generic base image which then 
boots, talks to a configuration mechanism (IMDS in EC2) which then 
contains all of the actual customization. For most customization, that 
is perfectly fine: You can install packages, run scripts, launch 
services, etc. But there is no simple way to modify the kernel command 
line. The story gets even worse when you try to abstract the cloud 
environment itself by using configuration layers on top like puppet, 
ansible, salt, etc. because you could not do a pre-boot environment hack 
even if you wanted to.

Users could in theory have a bootup script which checks the current 
command line, modifies the boot loader configuration, regenerates boot 
loader config files (If they even can. Signed UKIs make that difficult), 
and then reboot/kexec into the new environment. So we would punt a *lot* 
of complexity onto users and still degrade their experience by 
prolonging the launch phase.

I'm all ears for alternatives, but runtime setting seems like the most 
natural way to allow bootup / configuration scripts to actually instill 
policy.


Alex




Amazon Web Services Development Center Germany GmbH
Tamara-Danz-Str. 13
10243 Berlin
Geschaeftsfuehrung: Christian Schlaeger
Eingetragen am Amtsgericht Charlottenburg unter HRB 257764 B
Sitz: Berlin
Ust-ID: DE 365 538 597