On Wed, 27 Aug 2025 at 06:39, Richard Henderson
<richard.hender...@linaro.org> wrote:
>
> During reset.hold, the cpu is in an inconsistent state,
> where the leaf class has not had a chance to initialize
> state at all.
>
> This is visible as a SIGSEGV in "qemu-system-sparc64 -d cpu_reset".
>
> Move the dump to the exit phase, where all initialization
> is certain to be complete.
>
> Reported-by: Henk van der Laak <h...@laaksoft.nl>
> Signed-off-by: Richard Henderson <richard.hender...@linaro.org>
> ---
> hw/core/cpu-common.c | 16 +++++++++++-----
> 1 file changed, 11 insertions(+), 5 deletions(-)
>
> diff --git a/hw/core/cpu-common.c b/hw/core/cpu-common.c
> index 39e674aca2..26321be785 100644
> --- a/hw/core/cpu-common.c
> +++ b/hw/core/cpu-common.c
> @@ -119,11 +119,6 @@ static void cpu_common_reset_hold(Object *obj, ResetType
> type)
> {
> CPUState *cpu = CPU(obj);
>
> - if (qemu_loglevel_mask(CPU_LOG_RESET)) {
> - qemu_log("CPU Reset (CPU %d)\n", cpu->cpu_index);
> - log_cpu_state(cpu, cpu->cc->reset_dump_flags);
> - }
> -
> cpu->interrupt_request = 0;
> cpu->halted = cpu->start_powered_off;
> cpu->mem_io_pc = 0;
> @@ -137,6 +132,16 @@ static void cpu_common_reset_hold(Object *obj, ResetType
> type)
> cpu_exec_reset_hold(cpu);
> }
>
> +static void cpu_common_reset_exit(Object *obj, ResetType type)
> +{
> + if (qemu_loglevel_mask(CPU_LOG_RESET)) {
> + CPUState *cpu = CPU(obj);
> +
> + qemu_log("CPU Reset (CPU %d)\n", cpu->cpu_index);
> + log_cpu_state(cpu, cpu->cc->reset_dump_flags);
> + }
> +}
> +
> ObjectClass *cpu_class_by_name(const char *typename, const char *cpu_model)
> {
> ObjectClass *oc;
> @@ -380,6 +385,7 @@ static void cpu_common_class_init(ObjectClass *klass,
> const void *data)
> dc->realize = cpu_common_realizefn;
> dc->unrealize = cpu_common_unrealizefn;
> rc->phases.hold = cpu_common_reset_hold;
> + rc->phases.exit = cpu_common_reset_exit;
> cpu_class_init_props(dc);
> /*
> * Reason: CPUs still need special care by board code: wiring up
If we ever have CPUs that actually update their state in
the reset exit phase (e.g. if we manage to complete the refactoring
that would let us implement M-profile "load starting PC and SP
from memory" in reset-exit after rom blob loading rather than
having a hack to do it in reset-hold), this won't capture that.
But it's clearly better than trying to do it in the common
reset-hold method...
Reviewed-by: Peter Maydell <peter.mayd...@linaro.org>
thanks
-- PMM
