Re: KVM: x86: update masterclock when kvmclock_offset is calculated

[Paolo Bonzini <pbonzini@redhat.com>]

Il 03/09/2013 15:56, Marcelo Tosatti ha scritto:
>>> The point of REQ_MCLOCK_INPROGRESS request is to guarantee that the
>>> following is not possible:
>>>
>>> - 2 vcpus in guest mode with per-vcpu kvmclock areas with
>>> different {system_timestamp, tsc_offset} values.
>>
>> Understood.
>>
>>> To achieve that:
>>>
>>> - Kick all vcpus out of guest mode (via a request bit that can't be
>>>   cleared).
>>> - Update the {system_timestamp, tsc_offset} values.
>>> - Clear the request bit.
>>
>> After make_all_cpus_request, all VCPUs will be out of guest mode, and
>> the request bit will not be cleared until pvclock_gtod_sync_lock is
>> released.
> 
> It seems more obfuscated to rely on an implicit pvclock_gtod_sync_lock
> blocking than an explicit request bit that can't be cleared, but perhaps
> thats personal opinion.

Yes, this gets dangerously close to personal opinion territory. :)

> OTOH, you can also argue that the request bit abuses the vcpu->requests
> request mechanism, because there is not a request in fact (it only
> blocks guest entry).

The busy-wait is a bit ugly, but otherwise it's fine.

> If you have reasons to update, and you are sure its safe, feel
> free to modify it.

No reason yet apart from removing a few lines of code, but thanks for
discussing this.  If you want to do the seqlock change for
pvclock_gtod_sync_lock, I'll be glad to review the patch.

Paolo