changes from v4:
- patch 01 - no changes from v4
- patch 02:
    * fixed the size in memcpy
    * created a new macro to store the size of the vcpu
associativity array, VCPU_ASSOC_SIZE
    * simplified the logic using the new macro, dropping
the size argument from the helper
- patch 03:
    * added a comment in the loop where R4-R9 is set

The code turned out way cleaner after adding the the macro
Greg suggested. Patches were rebased on top of David's
ppc-for-5.2. 

Also, something that I forgot to mention: this series has
been tested with a guest with 4 NVlink GPUs and hotplug
memory enabled, and I'm comparing the FDTs created with
and without the series using dumpdtb and doing lsprop
manually. The idea is that, after all this code, we're still
in the same place as far as the guest can see. Guest
visible changes will be coming up after this.

Daniel Henrique Barboza (3):
  spapr: move h_home_node_associativity to spapr_numa.c
  spapr_numa: create a vcpu associativity helper
  spapr_numa: use spapr_numa_get_vcpu_assoc() in home_node hcall

 hw/ppc/spapr_hcall.c   |  40 ----------------
 hw/ppc/spapr_numa.c    | 103 +++++++++++++++++++++++++++++++++++------
 include/hw/ppc/spapr.h |   7 ++-
 3 files changed, 96 insertions(+), 54 deletions(-)

-- 
2.26.2