[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] Preface working plan for altp2m during freeze exception



>From: Jan Beulich [mailto:JBeulich@xxxxxxxx]
>Sent: Monday, July 20, 2015 12:12 AM
>
>>>> On 17.07.15 at 21:43, <ravi.sahita@xxxxxxxxx> wrote:
>> We are working on addressing review comments in this order (and you
>> will see this pattern in our review responses):
>>
>> * Category 1 - Address review comments that affect ABI - these are of
>> course required and will be addressed first.
>>
>> * Category 2 - Address review comments that do not affect ABI - we
>> will try to address ones that we think we can realistically meet
>> within the time bounds - we ask you for some flexibility on these. If
>> these cannot be addressed within the allotted exception time-frame,
>> hopefully these wont be blockers for 4.6 since they can be addressed by
>follow-on patches.
>
>Not sure - we've had bad experience with allowing code to go in with the
>promise for later adjustments (which then never happened)...
>

For some of the remaining items (not addressed by our v6) we have some 
tentative patches that we could share with you post 4.6, we just think we don't 
have time to get to Category 2 things (and probably not all of Category 4) to 
do a good job on them - please tell us based on your read of the v6 series if 
that is close to an acceptable "final - 1" series - with any minor i's to be 
dotted and t's to be crossed in the final - since realistically, we need to get 
a final patch series to you guys by Wednesday evening PDT (Thursday AM for you) 
- correct?

>> * Category 3 - Address review comments that are really design
>> questions - These we will try to address by short descriptions in
>> review replies that attempt to give a gist of the design we followed,
>> but of course design changes obviously cannot be done at this late
>> stage - hopefully that is expected.
>
>If you really just mean questions on the design (rather than questions possibly
>resulting int the requirement to change the design), then that'd be fine of
>course. I think you understand that we shouldn't be deferring issues that
>require design adjustments. Otoh I don't even recall what design questions
>there were.
>
>> * Category 4 - Address trivial changes as we naturally update patches,
>> however if we run out of time, some may remain un-addressed (to be
>> taken care of post 4.6).
>
>See above (point 2).
>
>> Can we please get a "yes - makes sense" sort of acknowledgement of
>> this plan from the Maintainers?
>
>Considering the limitations above, this is only a "maybe" from me.
>

Could you please review the v6 patches and see if your "maybe" can change to a 
"good for 4.6, pending changes post 4.6" - thanks.
Also please see my responses to your other open comments - I've tried to 
address all of them.

>> Y            N       [PATCH v3 05/15] x86/altp2m: basic data structures
>and support routines
>> Status if not acked: Category 3: we will write a short description of some
>> design questions in review replies
>>                      Category 2: moving altp2m struct to be dynamically
>allocated - this
>> has minor benefit and big downside so will be lower priority, also
>> some error handling fixes
>
>Big downside? You're not referring to the mechanical adjustments this
>implies, are you?
>

Just time to turn around and test the changes with our tests - like I said, we 
have a patch for this now - but time is short, so would request for this to not 
be the blocker.


>> Y            N       [PATCH v3 07/15] VMX: add VMFUNC leaf 0 (EPTP
>switching) to emulator
>> Status if not acked: Category 2/3 - changes staged after Jan's feedback on
>v5 -
>> ack with those addressed in v6?
>
>Let's see what v6 looks like.

Thanks.

>
>> Y            N       [PATCH v3 08/15] x86/altp2m: add control of
>suppress_ve
>> Status if not acked: Now has r-b both Jan and George -  need maintainers
>ack on
>> this one please
>
>Who do you refer to by "maintainer" here? I think the trivial adjustment to
>xen/arch/x86/mm/mem_sharing.c can in the worst case go in without Andres'
>ack. And everything else is covered by George's authorship and review.

Ok thanks,

Ravi

>
>Jan


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


 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.