Re: [Xen-devel] [PATCH 1/2] libxl: turn NUMA placement misconfigs into warnings

[Dario Faggioli <dario.faggioli@xxxxxxxxxx>]

On Fri, 2015-07-03 at 12:09 +0100, Ian Jackson wrote:
> Ian Campbell writes ("Re: [PATCH 1/2] libxl: turn NUMA placement misconfigs 
> into warnings"):
> > On Wed, 2015-07-01 at 15:38 +0100, Ian Jackson wrote:

> > > But, be aware that you are making a semantic change to the API which
> > > cannot be reverted for backwards-compatibility reasons.
> > 
> > Is any further action required due to this?
> 
> Not if it doesn't cause you or Dario to change your mind about the
> patch :-).
> 
It doesn't. This is the right thing to do, and should have been like
this from the beginning!

Just FTR, the reason why we were using the numa_autoplacement for sanity
checking was that, a while back, we were always passing a (potentially)
valid cpumap (for vcpu affinity) as part of the domain config. Hence we
needed something to distinguish the case were such a cpumap was full (or
empty, I can't recall) because of an explicit user choice, or as a mean
to ask libxl to perform the automatic placement.

I can't remember either why we did not do it, in the first place, the
way it is now (i.e., cpumaps *not allocated* --> do the placement), but
I never liked the fact that, if one[*] sets the affinity explicitly, but
forget to set numa_autoplacement to false, the domain does not build.

So, in summary, no, I don't think I will easily change my mind about
this needing to be a warning, rather than a fatal error.

Thanks and Regards,
Dario

[*] of course I'm talking about downstream toolstacks, like xl or
libvirt, not actual users, as in xl we do this consistently!
-- 
<<This happens because I choose it to happen!>> (Raistlin Majere)
-----------------------------------------------------------------
Dario Faggioli, Ph.D, http://about.me/dario.faggioli
Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK)

Attachment: signature.asc
Description: This is a digitally signed message part

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