Re: [Xen-devel] [PATCH v2] x86: generic MSRs save/restore

["Liu, Jinsong" <jinsong.liu@xxxxxxxxx>]

Jan Beulich wrote:
>>>> On 16.12.13 at 11:05, "Liu, Jinsong" <jinsong.liu@xxxxxxxxx> wrote:
>> Jan Beulich wrote:
>>>>>> On 16.12.13 at 10:53, "Liu, Jinsong" <jinsong.liu@xxxxxxxxx>
>>>>>> wrote: 
>>>> @@ -467,6 +487,7 @@ int main(int argc, char **argv)
>>>>          case HVM_SAVE_CODE(VIRIDIAN_VCPU): dump_viridian_vcpu();
>>>>          break; case HVM_SAVE_CODE(VMCE_VCPU): dump_vmce_vcpu();
>>>>          break; case HVM_SAVE_CODE(TSC_ADJUST): dump_tsc_adjust();
>>>> break; +        case HVM_SAVE_CODE(CPU_MSR): dump_cpu_msr(); break;
>>>>          case HVM_SAVE_CODE(END): break;
>>>>          default:
>>>>              printf(" ** Don't understand type %u: skipping\n",
>>>> ======================= 
>>>> 
>>>> w/o the declaration, case HVM_SAVE_CODE(CPU_MSR) compiling fail.
>>> 
>>> Right, because you ought to use CPU_MSR_CODE here, which the
>>> public header does define.
>> 
>> Isn't it ugly? not unified style w/ other typecodes.
>> Why not simply add declaration at Xen side?
> 
> We absolutely should not declare things that aren't correct.
> 
> Jan

You have had the declaration in the public header with variable size data 
structure:
struct hvm_msr {
    uint32_t count;
    struct hvm_one_msr {
        uint32_t index;
        uint32_t _rsvd;
        uint64_t val;
    } msr[1 /* variable size */];
};
Why can't declare DECLARE_HVM_SAVE_TYPE(CPU_MSR, CPU_MSR_CODE, struct hvm_msr)? 
They are both variable size.

Thanks,
Jinsong
_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel