From: Nikita Shubin <n.shubin@yadro.com>

Currently it is not possible to overload instance of RISCVCPU, 
i.e. something like this:

static const TypeInfo riscv_cpu_type_infos[] = {
     {
        .name = TYPE_ANOTHER_RISCV_CPU,
        .parent = TYPE_RISCV_CPU,
        .instance_size = sizeof(MyCPUState),
        .instance_init = riscv_my_cpu_init,
    }
};

Because we have RISCVHartArrayState.harts with exactly 
the size of RISCVCPU.

Using own instances can be used to store some internal hart state.

Cc: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Link: https://patchwork.kernel.org/project/qemu-devel/patch/20230727080545.7908-1-nikita.shubin@maquefel.me/

Nikita Shubin (2):
  hw/riscv: hart: replace array access with qemu_get_cpu()
  hw/riscv: hart: allow other cpu instance

 hw/riscv/boot.c               |  6 ++++--
 hw/riscv/riscv_hart.c         | 20 ++++++++++++--------
 hw/riscv/sifive_u.c           |  7 +++++--
 hw/riscv/spike.c              | 17 ++++++++++-------
 hw/riscv/virt-acpi-build.c    |  2 +-
 hw/riscv/virt.c               | 17 +++++++++--------
 include/hw/riscv/riscv_hart.h |  2 +-
 7 files changed, 42 insertions(+), 29 deletions(-)

-- 
2.39.2