Re: [V2 PATCH 0/8]: PVH dom0....

["Roger Pau Monné" <roger.pau@citrix.com>]

On 25/11/13 11:49, George Dunlap wrote:
> On 11/25/2013 09:06 AM, Jan Beulich wrote:
>>>>> On 23.11.13 at 01:03, Mukesh Rathor <mukesh.rathor@oracle.com> wrote:
>>> These patches implement PVH dom0. Please note there is 1 fixme in
>>> this entire series that is being addressed under a different patch
>>> series. PVH dom0 creation is disabled until then.
>>>
>>> Patches 1 thru 4 implement changes in and around construct_dom0.
>>> Patches 5 thru 7 are to support tool stack on PVH dom0, and have
>>> been mostly looked at in the past. Finally, patch 8 adds option to
>>> boot a dom0 in PVH mode.
>>>
>>> These patches are based on c/s: b18e2d9 with the addition of fixes
>>> in epte_get_entry_emt and Roger's AP bringup/cleanup patches.
>> George,
>>
>> assuming we can get the remaining issues sorted out within
>> reasonable time, I'm inclined to recommend taking these despite
>> the code freeze. What's your opinion in this regard?
> 
> I hadn't even looked at them, assuming that "PVH for dom0" had missed
> the feature freeze.
> 
> Within our decision framework, we would accept this series if we
> considered PVH dom0 to be a "blocker" for the release -- a feature that
> is strategically important enough to risk slipping the release
> significantly.  That's the basis on which the ARMv8 stuff got in.
> 
> I assume (without having looked at all) that this series introduces new
> interfaces for PVH.  So in addition to the normal risk that a series
> like this may break existing functionality, there is the additional risk
> that by rushing the feature in at the last minute, we may not have had
> enough time for the interface to be reviewed, and we may end up having
> to support an interface that wasn't well thought-out.
> 
> So the important questions are:
> 
> * Is there a good reason that this feature can't wait until 4.5?
> 
> * What is the risk of these changes breaking existing functionality?

I agree with everything above...

> * What is the risk that the new interfaces will turn out to be a bad fit
> or difficult to support going forward?

But I don't think this applies to PVH Dom0, the DomU interface has
already been marked as experimental, and subject to change. The same
applies here, Dom0 PVH interface would be considered experimental.