Re: [Qemu-devel] [PATCH v2 10/24] numa: mirror cpu to node mapping in MachineState::possible_cpus

[Igor Mammedov <imammedo@redhat.com>]

On Fri, 5 May 2017 14:04:15 -0300
Eduardo Habkost <ehabkost@redhat.com> wrote:

> On Fri, May 05, 2017 at 02:16:48PM +0200, Igor Mammedov wrote:
> > On Wed, 3 May 2017 12:20:22 -0300
> > Eduardo Habkost <ehabkost@redhat.com> wrote:
> > 
> > > On Wed, May 03, 2017 at 02:57:04PM +0200, Igor Mammedov wrote:
> > > > Introduce machine_set_cpu_numa_node() helper that stores
> > > > node mapping for CPU in MachineState::possible_cpus.
> > > > CPU and node it belongs to is specified by 'props' argument.
> > > > 
> > > > Patch doesn't remove old way of storing mapping in
> > > > numa_info[X].node_cpu as removing it at the same time
> > > > makes patch rather big. Instead it just mirrors mapping
> > > > in possible_cpus and follow up per target patches will
> > > > switch to possible_cpus and numa_info[X].node_cpu will
> > > > be removed once there isn't any users left.
> > > > 
> > > > Signed-off-by: Igor Mammedov <imammedo@redhat.com>
> > > > Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
> > > > ---
> > > >  include/hw/boards.h |  2 ++
> > > >  hw/core/machine.c   | 68 +++++++++++++++++++++++++++++++++++++++++++++++++++++
> > > >  numa.c              |  8 +++++++
> > > >  3 files changed, 78 insertions(+)
> > > > 
> > > > diff --git a/include/hw/boards.h b/include/hw/boards.h
> > > > index 5d6af21..1f518a1 100644
> > > > --- a/include/hw/boards.h
> > > > +++ b/include/hw/boards.h
> > > > @@ -42,6 +42,8 @@ bool machine_dump_guest_core(MachineState *machine);
> > > >  bool machine_mem_merge(MachineState *machine);
> > > >  void machine_register_compat_props(MachineState *machine);
> > > >  HotpluggableCPUList *machine_query_hotpluggable_cpus(MachineState *machine);
> > > > +void machine_set_cpu_numa_node(MachineState *machine,
> > > > +                               CpuInstanceProperties *props, Error **errp);
> > > >  
> > > >  /**
> > > >   * CPUArchId:
> > > > diff --git a/hw/core/machine.c b/hw/core/machine.c
> > > > index ada9eea..a63f17b 100644
> > > > --- a/hw/core/machine.c
> > > > +++ b/hw/core/machine.c
> > > > @@ -388,6 +388,74 @@ HotpluggableCPUList *machine_query_hotpluggable_cpus(MachineState *machine)
> > > >      return head;
> > > >  }
> > > >  
> > > > +void machine_set_cpu_numa_node(MachineState *machine,
> > > > +                               CpuInstanceProperties *props, Error **errp)  
> > > 
> > > As the semantics of this function aren't trivial, it would be
> > > nice to have a comment explaining what exactly this function do.
> > > 
> > > e.g.:
> > > * make it clear that it could affect multiple CPU slots;
> > > * make it clear what does it mean to have props->has_node_id=false as
> > >   argument (is it really valid?);
> > > * make it clear that it will refuse to change an existing mapping.
> > Will be following comment sufficient?
> > 
> > +/**
> > + * machine_set_cpu_numa_node:
> > + * @machine: machine object to modify
> > + * @props: specifies which cpu objects to assign to
> > + *         numa node specified by @props.node_id
> > + * @errp: if an error occurs, a pointer to an area to store the error
> > + *
> > + * Associate NUMA node specified by @props.node_id with cpu slots that
> > + * match socket/core/thread-ids specified by @props. It's recommended to use
> > + * query-hotpluggable-cpus.props values to specify affected cpu slots,
> > + * which would lead to exact 1:1 mapping of cpu slots to NUMA node.
> > + *
> > + * However for CLI convenience it's possible to pass in subset of properties,
> > + * which would affect all cpu slots that match it.
> > + * Ex for pc machine:
> > + *    -smp 4,cores=2,sockets=2 -numa node,nodeid=0 -numa node,nodeid=1 \
> > + *    -numa cpu,node-id=0,socket_id=0 \
> > + *    -numa cpu,node-id=1,socket_id=1
> > + * will assign all child cores of socket 0 to node 0 and
> > + * of socket 1 to node 1.
> > + *
> > + * Empty subset is disallowed and function will return with error in this case.
> > + */
> >  void machine_set_cpu_numa_node(MachineState *machine,
> >                                 CpuInstanceProperties *props, Error **errp)
> 
> Sounds good to me.
> 
> While at it, we could make 'props' const, as it's not going to be
> touched by the function.
sure