On 12/19/25 7:27 AM, Anton Johansson wrote:
On 17/12/25, Pierrick Bouvier wrote:
On 12/16/25 3:51 PM, Anton Johansson wrote:
TYPE_RISCV_CPU_BASE is used only to initialize the correct default
machine for 3 machines. Replace it with a runtime check.
Signed-off-by: Anton Johansson <[email protected]>
---
include/hw/riscv/virt.h | 11 +++++++++++
target/riscv/cpu.h | 6 ------
hw/riscv/microblaze-v-generic.c | 3 ++-
hw/riscv/spike.c | 3 ++-
hw/riscv/virt.c | 2 +-
5 files changed, 16 insertions(+), 9 deletions(-)
Using directly class->default_cpu_type instead of callback
class->get_default_cpu_type() works here compared to hw/arm/virt because
default cpu type does not depend on accelerator used, and we already will
know what would be the runtime target at this point.
Ah I see, appreciate the context!:)
It's subtle enough, and when I saw your patch, I forgot the initial
reason why we needed the callback with arm virt. Reading again through
it made me found the difference, so it was worth mentioning.
