Re: [PATCH v3 1/3] x86/vmx: implement VMExit based guest Bus Lock detection

[Roger Pau Monné <roger.pau@xxxxxxxxxx>]

On Tue, Jul 19, 2022 at 07:26:08AM +0000, Tian, Kevin wrote:
> > From: Roger Pau Monne <roger.pau@xxxxxxxxxx>
> > Sent: Friday, July 1, 2022 9:17 PM
> > 
> > @@ -4065,6 +4065,11 @@ void vmx_vmexit_handler(struct cpu_user_regs
> > *regs)
> > 
> >      if ( unlikely(exit_reason & VMX_EXIT_REASONS_FAILED_VMENTRY) )
> >          return vmx_failed_vmentry(exit_reason, regs);
> 
> Add a blank line.
> 
> > +    if ( unlikely(exit_reason & VMX_EXIT_REASONS_BUS_LOCK) )
> > +    {
> > +        perfc_incr(buslock);
> > +        exit_reason &= ~VMX_EXIT_REASONS_BUS_LOCK;
> > +    }
> > 
> >      if ( v->arch.hvm.vmx.vmx_realmode )
> >      {
> > @@ -4561,6 +4566,15 @@ void vmx_vmexit_handler(struct cpu_user_regs
> > *regs)
> >          vmx_handle_descriptor_access(exit_reason);
> >          break;
> > 
> > +    case EXIT_REASON_BUS_LOCK:
> > +        /*
> > +         * Nothing to do: just taking a vmexit should be enough of a pause 
> > to
> > +         * prevent a VM from crippling the host with bus locks.  Note
> > +         * EXIT_REASON_BUS_LOCK will always have bit 26 set in exit_reason,
> > and
> > +         * hence the perf counter is already increased.
> > +         */
> > +        break;
> > +
> 
> Would it be helpful from diagnostic angle by throwing out a warning,
> once per the culprit domain?

Hm, not sure.  I've assumed that increasing the counter would be
enough, but that's not tied to a domain.

I will leave as-is unless someone else expresses interest in this (and
can also be added later if desired).

Thanks, Roger.