Re: [PATCH v2 16/30] x86/cpu: Move set_cpumask() calls into c_early_init()

["Jan Beulich" <JBeulich@suse.com>]

>>> On 17.02.16 at 11:45, <andrew.cooper3@citrix.com> wrote:
> On 16/02/16 14:10, Jan Beulich wrote:
>>>>> On 05.02.16 at 14:42, <andrew.cooper3@citrix.com> wrote:
>>> Before c/s 44e24f8567 "x86: don't call generic_identify() redundantly", the
>>> commandline-provided masks would take effect in Xen's view of the features.
>>>
>>> As the masks got applied after the query for features, the redundant call to
>>> generic_identify() would clobber the pre-masking feature information with the
>>> post-masking information.
>>>
>>> Move the set_cpumask() calls into c_early_init() so their effects take place
>>> before the main query for features in generic_identify().
>>>
>>> The cpuid_mask_* command line parameters now limit the entire system, a
>>> feature XenServer was relying on for testing purposes.
>> And I continue to view this as a step backwards, and hence can't
>> really approve of this change.
> 
> It is not a step backwards.  Being able to disable features in Xen is
> critical for testing.  You accidentally broke that with a patch which
> was supposed to be no functional change.

Views differ: I would say I unknowingly fixed this with that patch
(as to me it was always clear that this masking should not apply to
Xen itself).

If that behavior is critical for testing, add a command line option
to enable it.

> This series means that these command line options are not required for
> levelling.

By the end of the series, agreed.

Jan