Re: [PATCH 2/4] x86: Introduce MSR_UNHANDLED

[Jan Beulich <jbeulich@xxxxxxxx>]

On 08.01.2021 21:31, boris.ostrovsky@xxxxxxxxxx wrote:
> On 1/8/21 9:55 AM, Jan Beulich wrote:
>> On 07.01.2021 21:34, Boris Ostrovsky wrote:
>>> + *  0 = return #GP, warning emitted
>>> + *  1 = read as 0, writes are dropped, no warning
>>> + *  2 = read as 0, writes are dropped, warning emitted
>>> + */
>>> +#define MSR_UNHANDLED_NEVER     0
>>> +#define MSR_UNHANDLED_SILENT    1
>>> +#define MSR_UNHANDLED_VERBOSE   2
>>> +
>>> +/* MSR that is not explicitly processed by emulation */
>>> +#define MSR_UNHANDLED -1
>> Not sure about this choice: I'd consider an MSR index in the Xen
>> range more safe to use, not the least because this value
>> effectively becomes part of the migration ABI. Or if you use one
>> outside, then maybe a less prominent one than 0xffffffff (I
>> guess -1, irrespective of the missing parentheses, isn't
>> appropriate to use here).
> 
> 
> All MSRs in Xen range are currently handled (although most return
> an exception). I can reserve the last one (0x400002ff) if you feel
> it's more appropriate.

I do, yes, but I'd prefer to also have Andrew's general view here.
Difficulty is his email delivery issue, so I don't know how soon
we could hope for a reply.

Jan