Re: [Xen-devel] [PATCH 18/19] xen/mce: add support of vLMCE injection to XEN_MC_inject_v2

["Jan Beulich" <JBeulich@xxxxxxxx>]

>>> On 23.02.17 at 10:14, <haozhong.zhang@xxxxxxxxx> wrote:
> On 02/23/17 01:26 -0700, Jan Beulich wrote:
>> >>> On 23.02.17 at 06:14, <haozhong.zhang@xxxxxxxxx> wrote:
>> > On 02/22/17 08:59 -0700, Jan Beulich wrote:
>> >> >>> On 17.02.17 at 07:39, <haozhong.zhang@xxxxxxxxx> wrote:
>> >> > @@ -1552,6 +1553,21 @@ long do_mca(XEN_GUEST_HANDLE_PARAM(xen_mc_t) 
> u_xen_mc)
>> >> >                  send_IPI_mask(cpumap, cmci_apic_vector);
>> >> >              }
>> >> >              break;
>> >> > +        case XEN_MC_INJECT_TYPE_LMCE:
>> >> > +            if ( !lmce_support )
>> >> > +            {
>> >> > +                ret = x86_mcerr("No LMCE support in platform", 
>> >> > -EINVAL);
>> >> > +                break;
>> >> > +            }
>> >> > +            /* ensure at most one CPU is specified */
>> >> 
>> >> And what use is none at all? Also - comment style (should start with
>> >> a capital).
>> >>
>> > 
>> > Do you mean the check of empty cpumap? It's checked at the beginning of 
>> > case 
>> > XEN_MC_inject_v2.
>> 
>> To be honest, I don't see any such check. But looking at that code
>> makes me notice you should also forbid the combination of
>> XEN_MC_INJECT_CPU_BROADCAST and XEN_MC_INJECT_TYPE_LMCE.
>>
> 
> I mean the following check. Of course, the additional check you
> suggested must go before it.
>         if ( op->u.mc_inject_v2.flags & XEN_MC_INJECT_CPU_BROADCAST )
>             cpumap = &cpu_online_map;
>         else
>         {
>             ret = xenctl_bitmap_to_cpumask(&cmv, &op->u.mc_inject_v2.cpumap);
>             if ( ret )
>                 break;
>             cpumap = cmv;
>             if ( !cpumask_intersects(cpumap, &cpu_online_map) )  <=== I mean 
> this one exactly

I've been blind, so I'm sorry for the noise.

Jan


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
https://lists.xen.org/xen-devel