Re: [Qemu-devel] [PATCH 03/22] target-i386: split out CPU creation and features parsing into cpu_x86_create()

[Eduardo Habkost]

On Fri, Apr 05, 2013 at 04:36:55PM +0200, Igor Mammedov wrote:
> Move CPU creation and features parsing into a separate cpu_x86_create()
> function, so that board would be able to set board specific CPU
> properties before CPU is realized.
> 
> Keep cpu_x86_init() for compatibility with the code that uses cpu_init()
> and doesn't need to modify CPU properties.
> 
> Signed-off-by: Igor Mammedov <address@hidden>
> ---
>  target-i386/cpu.c | 28 +++++++++++++++++++---------
>  target-i386/cpu.h |  1 +
>  2 files changed, 20 insertions(+), 9 deletions(-)
> 
> diff --git a/target-i386/cpu.c b/target-i386/cpu.c
> index 41f0f47..269a681 100644
> --- a/target-i386/cpu.c
> +++ b/target-i386/cpu.c
> @@ -1563,17 +1563,16 @@ static void cpu_x86_register(X86CPU *cpu, const char 
> *name, Error **errp)
>      object_property_set_str(OBJECT(cpu), def->model_id, "model-id", errp);
>  }
>  
> -X86CPU *cpu_x86_init(const char *cpu_model)
> +X86CPU *cpu_x86_create(const char *cpu_model, Error **errp)
>  {
>      X86CPU *cpu = NULL;
>      CPUX86State *env;
>      gchar **model_pieces;
>      char *name, *features;
> -    Error *error = NULL;
>  
>      model_pieces = g_strsplit(cpu_model, ",", 2);
>      if (!model_pieces[0]) {
> -        error_setg(&error, "Invalid/empty CPU model name");
> +        error_setg(errp, "Invalid/empty CPU model name");
>          goto out;
>      }
>      name = model_pieces[0];
> @@ -1583,23 +1582,34 @@ X86CPU *cpu_x86_init(const char *cpu_model)
>      env = &cpu->env;
>      env->cpu_model_str = cpu_model;
>  
> -    cpu_x86_register(cpu, name, &error);
> -    if (error) {
> +    cpu_x86_register(cpu, name, errp);
> +    if (error_is_set(errp)) {

So the function now does error checking properly if and only if errp is
not NULL. Do we really want to do that?

>          goto out;
>      }
>  
> -    cpu_x86_parse_featurestr(cpu, features, &error);
> -    if (error) {
> +    cpu_x86_parse_featurestr(cpu, features, errp);
> +    if (error_is_set(errp)) {
>          goto out;
>      }
>  
> -    object_property_set_bool(OBJECT(cpu), true, "realized", &error);
> +out:
> +    g_strfreev(model_pieces);

Any specific reason you didn't choose to keep 'Error *error = NULL'
inside cpu_x86_create() as well, and use error_propagate() here? I
believe it would make the patch simpler and easier to review, and at the
same time make cpu_x86_init() check for errors properly even if errp is
NULL. This is the opposite of what you did on x86_cpu_realizefn() at
patch 01/22.

I am not against it if you want to keep this style of error-checking,
but I believe an error_propagate()-based version would be simpler and
safer.


> +    return cpu;
> +}
> +
> +X86CPU *cpu_x86_init(const char *cpu_model)
> +{
> +    Error *error = NULL;
> +    X86CPU *cpu;
> +
> +    cpu = cpu_x86_create(cpu_model, &error);
>      if (error) {
>          goto out;
>      }
>  
> +    object_property_set_bool(OBJECT(cpu), true, "realized", &error);
> +
>  out:
> -    g_strfreev(model_pieces);
>      if (error) {
>          fprintf(stderr, "%s\n", error_get_pretty(error));
>          error_free(error);
> diff --git a/target-i386/cpu.h b/target-i386/cpu.h
> index 069a2e2..b98efd2 100644
> --- a/target-i386/cpu.h
> +++ b/target-i386/cpu.h
> @@ -896,6 +896,7 @@ typedef struct CPUX86State {
>  #include "cpu-qom.h"
>  
>  X86CPU *cpu_x86_init(const char *cpu_model);
> +X86CPU *cpu_x86_create(const char *cpu_model, Error **errp);
>  int cpu_x86_exec(CPUX86State *s);
>  void x86_cpu_list(FILE *f, fprintf_function cpu_fprintf);
>  void x86_cpudef_setup(void);
> -- 
> 1.8.1.4
> 
> 

-- 
Eduardo