Re: [RFC PATCH] x86/msr: Filter MSR writes

[Sean Christopherson]

On Fri, Jun 12, 2020 at 06:46:02PM +0200, Borislav Petkov wrote:
> On Fri, Jun 12, 2020 at 09:34:06AM -0700, Sean Christopherson wrote:
> > The kernel should be tainted if the WRMSR is attempted, regardless of
> > whether it succeeds, and it should happen before the WRMSR.  E.g. pointing
> > MSR_IA32_DS_AREA at a bad address will likely cause an OOPS on the #PF
> 
> If the MSR write fails, MSR_IA32_DS_AREA won't have the bad address. If
> the writes fail, nothing has been changed.

DS_AREA takes a virtual (linear) address, i.e. the address can be legal from
the CPUs perspective but still lead to a #PF due to the address not being
mapped in the page tables.

> > This can be 0600, or maybe 0644, i.e. allow the user to enable/disable
> > writes after the module has been loaded.
> 
> What for?

So users don't have to unload and reload the module just to enable or
disable writes.  I don't think it changes the protections in any way, a
priveleged user still needs to explicitly toggle the control.

> crw------- 1 root root 202, 0 Jun 10 19:54 /dev/cpu/0/msr
> 
> You need root to write to the chrdev.
> 
> Also, the intent is *not* to open it more but to close it so that the
> incentive to design proper interfaces is there.
> 
> -- 
> Regards/Gruss,
>     Boris.
> 
> https://people.kernel.org/tglx/notes-about-netiquette