Re: [PATCH v2 1/2] x86/xen: remove xen_have_vcpu_info_placement flag

[Juergen Gross <jgross@xxxxxxxx>]

On 22.09.21 23:43, Boris Ostrovsky wrote:
> On 9/22/21 6:31 AM, Juergen Gross wrote:
> >   
> > -	if (xen_have_vcpu_info_placement) {
> > -               vcpup = &per_cpu(xen_vcpu_info, cpu);
> > -               info.mfn = arbitrary_virt_to_mfn(vcpup);
> > -               info.offset = offset_in_page(vcpup);
> > +       vcpup = &per_cpu(xen_vcpu_info, cpu);
> > +       info.mfn = arbitrary_virt_to_mfn(vcpup);
> > +       info.offset = offset_in_page(vcpup);
> >   
> > -		/*
> > -                * Check to see if the hypervisor will put the vcpu_info
> > -                * structure where we want it, which allows direct access via
> > -                * a percpu-variable.
> > -                * N.B. This hypercall can _only_ be called once per CPU.
> > -                * Subsequent calls will error out with -EINVAL. This is due to
> > -                * the fact that hypervisor has no unregister variant and this
> > -                * hypercall does not allow to over-write info.mfn and
> > -                * info.offset.
> > -                */
> > -               err = HYPERVISOR_vcpu_op(VCPUOP_register_vcpu_info,
> > -                                        xen_vcpu_nr(cpu), &info);
> > -
> > -               if (err) {
> > -                       pr_warn_once("register_vcpu_info failed: cpu=%d 
> > err=%d\n",
> > -                                    cpu, err);
> > -                       xen_have_vcpu_info_placement = 0;
> > -               } else {
> > -                       /*
> > -                        * This cpu is using the registered vcpu info, even if
> > -                        * later ones fail to.
> > -                        */
> > -                       per_cpu(xen_vcpu, cpu) = vcpup;
> > -               }
> > -       }
> > -
> > -       if (!xen_have_vcpu_info_placement)
> > -               xen_vcpu_info_reset(cpu);
> > +       /*
> > +        * N.B. This hypercall can _only_ be called once per CPU.
> > +        * Subsequent calls will error out with -EINVAL. This is due to
> > +        * the fact that hypervisor has no unregister variant and this
> > +        * hypercall does not allow to over-write info.mfn and
> > +        * info.offset.
> > +        */
> > +       err = HYPERVISOR_vcpu_op(VCPUOP_register_vcpu_info, xen_vcpu_nr(cpu),
> > +                                &info);
> > +       if (err)
> > +               panic("register_vcpu_info failed: cpu=%d err=%d\n", cpu, err);
> >   
>
> This is change in behavior. Before if the hypercall failed we still try to boot. I 
> am not sure we need to worry about this (since it's not clear it actually 
> works)  but I'd at least mention this in the commit message.
Hmm, maybe I should have been more explicit saying that the hypercall
was introduced in Xen 3.4, and only reason of failure is either an
illegal vcpu, an invalid mapping specification, or a try to reissue the
hypercall for a vcpu. None of those should ever happen.


Juergen

Attachment: OpenPGP_0xB0DE9DD628BF132F.asc
Description: OpenPGP public key

Attachment: OpenPGP_signature
Description: OpenPGP digital signature