Hi,

In this version I included a rework in riscv_load_fdt() to separate the
fdt address calculation from the fdt load process. Having both in the
same function doesn't give us much and can lead to confusion due to how
other archs handle their respective load_fdt() functions.

Patches are based on riscv-to-apply.next.

Changes from v1:
- former patches 1-6: already applied to riscv-to-apply.next
- former patch 7: removed
- patch 1 (new):
  - fix a potential issue with fdt_pack() called after fdt_totalsize()
- patch 2 (new):  
  - split fdt address compute from fdt load logic
- patch 3 (new):
  - simplify the new riscv_compute_fdt_addr() by using MachineState
- patches 4,5,6:
  - added Phil's r-b
v1 link: https://lists.gnu.org/archive/html/qemu-devel/2023-01/msg02246.html


Daniel Henrique Barboza (6):
  hw/riscv/boot.c: calculate fdt size after fdt_pack()
  hw/riscv: split fdt address calculation from fdt load
  hw/riscv: simplify riscv_compute_fdt_addr()
  hw/riscv/virt.c: calculate socket count once in create_fdt_imsic()
  hw/riscv/virt.c: rename MachineState 'mc' pointers to 'ms'
  hw/riscv/spike.c: rename MachineState 'mc' pointers to' ms'

 hw/riscv/boot.c            |  33 ++-
 hw/riscv/microchip_pfsoc.c |   6 +-
 hw/riscv/sifive_u.c        |   7 +-
 hw/riscv/spike.c           |  24 +-
 hw/riscv/virt.c            | 468 +++++++++++++++++++------------------
 include/hw/riscv/boot.h    |   3 +-
 6 files changed, 281 insertions(+), 260 deletions(-)

-- 
2.39.0