| Message ID | 1462782566-1791-1-git-send-email-jgross@suse.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On 09/05/2016 09:29, Juergen Gross wrote: > Domain-0 is always member of Pool-0 (or, to be precise: of the cpuppol > with cpupool-id 0). Document this in the xl man page. > > Signed-off-by: Juergen Gross <jgross@suse.com> Out of interest, why? Is this a limitation of the current implementation, or something which couldn't be made to work? Given that it is all just vcpus and linked lists in the end, I can't see why this should necessarily be the case. ~Andrew
On 09/05/16 10:36, Andrew Cooper wrote: > On 09/05/2016 09:29, Juergen Gross wrote: >> Domain-0 is always member of Pool-0 (or, to be precise: of the cpuppol >> with cpupool-id 0). Document this in the xl man page. >> >> Signed-off-by: Juergen Gross <jgross@suse.com> > > Out of interest, why? > > Is this a limitation of the current implementation, or something which > couldn't be made to work? > > Given that it is all just vcpus and linked lists in the end, I can't see > why this should necessarily be the case. Main reason was to keep cpu hotplugging simple. New cpus are always added to Pool-0 and should be available to dom0 immediately without having to move them to another cpupool. It could be changed, of course. But I don't see any advantage in allowing to move dom0 to another cpupool, while the code would be more complicated. Juergen
On Mon, May 09, 2016 at 10:53:30AM +0200, Juergen Gross wrote: > On 09/05/16 10:36, Andrew Cooper wrote: > > On 09/05/2016 09:29, Juergen Gross wrote: > >> Domain-0 is always member of Pool-0 (or, to be precise: of the cpuppol > >> with cpupool-id 0). Document this in the xl man page. > >> > >> Signed-off-by: Juergen Gross <jgross@suse.com> > > > > Out of interest, why? > > > > Is this a limitation of the current implementation, or something which > > couldn't be made to work? > > > > Given that it is all just vcpus and linked lists in the end, I can't see > > why this should necessarily be the case. > > Main reason was to keep cpu hotplugging simple. New cpus are always > added to Pool-0 and should be available to dom0 immediately without > having to move them to another cpupool. > > It could be changed, of course. But I don't see any advantage in > allowing to move dom0 to another cpupool, while the code would be > more complicated. > > I think this patch is valid in that it documents the current situation. Whether migration of Dom0 should be supported or not is a separate topic. Acked-by: Wei Liu <wei.liu2@citrix.com> Release-acked-by: Wei Liu <wei.liu2@citrix.com> I've queued this up for committing. Wei. > Juergen >
diff --git a/docs/man/xl.pod.1 b/docs/man/xl.pod.1 index e2bd32d..9887f1b 100644 --- a/docs/man/xl.pod.1 +++ b/docs/man/xl.pod.1 @@ -1263,6 +1263,7 @@ exact same syntax as in B<cpupool-cpu-add> above. =item B<cpupool-migrate> I<domain> I<cpu-pool> Moves a domain specified by domain-id or domain-name into a cpu-pool. +Domain-0 can't be moved to another cpu-pool. =item B<cpupool-numa-split>
Domain-0 is always member of Pool-0 (or, to be precise: of the cpuppol with cpupool-id 0). Document this in the xl man page. Signed-off-by: Juergen Gross <jgross@suse.com> --- docs/man/xl.pod.1 | 1 + 1 file changed, 1 insertion(+)