Hi Philippe,
On Fri, Mar 15, 2024 at 02:08:53PM +0100, Philippe Mathieu-Daudé wrote:
> Date: Fri, 15 Mar 2024 14:08:53 +0100
> From: Philippe Mathieu-Daudé
> Subject: [PATCH-for-9.1 05/21] cpus: Open code OBJECT_DECLARE_TYPE() in
> OBJECT_DECLARE_CPU_TYPE()
> X-Mailer: git-send-email 2.41.0
>
> Since the OBJECT_DECLARE_CPU_TYPE() macro uses the abstract ArchCPU
> type, when declaring multiple CPUs of the same ArchCPU type we get
> an error related to the indirect G_DEFINE_AUTOPTR_CLEANUP_FUNC()
> use within OBJECT_DECLARE_TYPE():
>
> target/mips/cpu-qom.h:31:1: error: redefinition of
> 'glib_autoptr_clear_ArchCPU'
> OBJECT_DECLARE_CPU_TYPE(MIPS64CPU, MIPSCPUClass, MIPS64_CPU)
> ^
> include/hw/core/cpu.h:82:5: note: expanded from macro
> 'OBJECT_DECLARE_CPU_TYPE'
> OBJECT_DECLARE_TYPE(ArchCPU, CpuClassType, CPU_MODULE_OBJ_NAME);
> ^
> include/qom/object.h:237:5: note: expanded from macro 'OBJECT_DECLARE_TYPE'
> G_DEFINE_AUTOPTR_CLEANUP_FUNC(InstanceType, object_unref) \
> ^
> /usr/include/glib-2.0/glib/gmacros.h:1371:3: note: expanded from macro
> 'G_DEFINE_AUTOPTR_CLEANUP_FUNC'
> _GLIB_DEFINE_AUTOPTR_CLEANUP_FUNCS(TypeName, TypeName, func)
> ^
> /usr/include/glib-2.0/glib/gmacros.h:1354:36: note: expanded from macro
> '_GLIB_DEFINE_AUTOPTR_CLEANUP_FUNCS'
> static G_GNUC_UNUSED inline void _GLIB_AUTOPTR_CLEAR_FUNC_NAME(TypeName)
> (TypeName *_ptr) \
> ^
> /usr/include/glib-2.0/glib/gmacros.h:1338:49: note: expanded from macro
> '_GLIB_AUTOPTR_CLEAR_FUNC_NAME'
> #define _GLIB_AUTOPTR_CLEAR_FUNC_NAME(TypeName)
> glib_autoptr_clear_##TypeName
> ^
> :54:1: note: expanded from here
> glib_autoptr_clear_ArchCPU
> ^
> target/mips/cpu-qom.h:30:1: note: previous definition is here
> OBJECT_DECLARE_CPU_TYPE(MIPS32CPU, MIPSCPUClass, MIPS32_CPU)
> ^
>
> Avoid that problem by expanding the OBJECT_DECLARE_TYPE() macro
> within OBJECT_DECLARE_CPU_TYPE().
>
> Signed-off-by: Philippe Mathieu-Daudé
> Acked-by: Richard Henderson
> ---
> TODO: check rth comment:
> What about adding an OBJECT_DECLARE_CPU_SUBTYPE that omits half the stuff
> instead?
> We don't need another object typedef at all, for instance.
> ---
> include/hw/core/cpu.h | 7 ++-
> 1 file changed, 6 insertions(+), 1 deletion(-)
>
> diff --git a/include/hw/core/cpu.h b/include/hw/core/cpu.h
> index ec14f74ce5..4c2e5095bf 100644
> --- a/include/hw/core/cpu.h
> +++ b/include/hw/core/cpu.h
> @@ -78,7 +78,12 @@ DECLARE_CLASS_CHECKERS(CPUClass, CPU,
> */
> #define OBJECT_DECLARE_CPU_TYPE(CpuInstanceType, CpuClassType,
> CPU_MODULE_OBJ_NAME) \
> typedef struct ArchCPU CpuInstanceType; \
> -OBJECT_DECLARE_TYPE(ArchCPU, CpuClassType, CPU_MODULE_OBJ_NAME);
> +typedef struct CpuClassType CpuClassType; \
> +\
> +G_DEFINE_AUTOPTR_CLEANUP_FUNC(CpuInstanceType, object_unref) \
> +\
> +DECLARE_OBJ_CHECKERS(CpuInstanceType, CpuClassType, \
> + CPU_MODULE_OBJ_NAME, TYPE_##CPU_MODULE_OBJ_NAME)
>
The OBJECT_DECLARE_TYPE is expaneded as the following:
#define OBJECT_DECLARE_TYPE(InstanceType, ClassType, MODULE_OBJ_NAME) \
typedef struct InstanceType InstanceType; \
typedef struct ClassType ClassType; \
\
G_DEFINE_AUTOPTR_CLEANUP_FUNC(InstanceType, object_unref) \
\
DECLARE_OBJ_CHECKERS(InstanceType, ClassType, \
MODULE_OBJ_NAME, TYPE_##MODULE_OBJ_NAME)
So the above code change deletes a typedef:
typedef struct ArchCPU ArchCPU;
Will this deletion break the direct uses of ArchCPU? e.g., in
accel/tcg/translator.c:
static void set_can_do_io(DisasContextBase *db, bool val)
{
if (db->saved_can_do_io != val) {
...
tcg_gen_st8_i32(tcg_constant_i32(val), tcg_env,
offsetof(ArchCPU, parent_obj.neg.can_do_io) -
offsetof(ArchCPU, env));
}
}
Thanks,
Zhao