Re: [Qemu-devel] [RFC PATCH v1 2/4] numa, pc-dimm: Store pc-dimm memory information in numa_info

[Bharata B Rao <bharata@linux.vnet.ibm.com>]

On Mon, Jun 15, 2015 at 11:17:59AM +0200, Igor Mammedov wrote:
> On Fri, 12 Jun 2015 14:30:26 +0530
> Bharata B Rao <bharata@linux.vnet.ibm.com> wrote:
> 
> > Start storing the (start_addr, size, nodeid) of the pc-dimm memory
> > in numa_info so that this information can be used to lookup
> > node by address.
> > 
> > Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
> > ---
> >  hw/mem/pc-dimm.c      |  4 ++++
> >  include/sysemu/numa.h | 10 ++++++++++
> >  numa.c                | 26 ++++++++++++++++++++++++++
> >  3 files changed, 40 insertions(+)
> > 
> > diff --git a/hw/mem/pc-dimm.c b/hw/mem/pc-dimm.c
> > index 98971b7..bb04862 100644
> > --- a/hw/mem/pc-dimm.c
> > +++ b/hw/mem/pc-dimm.c
> > @@ -97,6 +97,7 @@ void pc_dimm_memory_plug(DeviceState *dev, MemoryHotplugState *hpms,
> >  
> >      memory_region_add_subregion(&hpms->mr, addr - hpms->base, mr);
> >      vmstate_register_ram(mr, dev);
> > +    numa_set_mem_node_id(addr, memory_region_size(mr), dimm->node);
> >  
> >  out:
> >      error_propagate(errp, local_err);
> > @@ -105,6 +106,9 @@ out:
> >  void pc_dimm_memory_unplug(DeviceState *dev, MemoryHotplugState *hpms,
> >                             MemoryRegion *mr)
> >  {
> > +    PCDIMMDevice *dimm = PC_DIMM(dev);
> > +
> > +    numa_unset_mem_node_id(dimm->addr, memory_region_size(mr), dimm->node);
> Wouldn't that cause pc-dimm range appear in SRAT table?

I don't think so.

numa_set_mem_node_id() and numa_unset_mem_node_id() APIs store/remove
address range and node id information of realized pc-dimm device in/from a
linked list in numa_info structure so that we can lookup the node id for a
given address from numa.c in a self-contained manner.

So unless I am missing something, I don't see this affecting ACPI/SRAT
table in any way.

> Before this pc-dimm-s are only added as ACPI devices but don't
> advertised in SRAT ACPI table.
> 
> Perhaps make it up to target to decide if it want's to
> report dimms with numa_unset_mem_node_id() and not in generic code.

Regards,
Bharata.