Re: [Qemu-devel] [RFC v2 3/5] linux-user: prepare for switching to cpu_create() API

[Igor Mammedov]

On Thu, 18 Jan 2018 18:33:37 -0200
Eduardo Habkost <address@hidden> wrote:

> On Thu, Jan 18, 2018 at 06:34:00PM +0100, Igor Mammedov wrote:
> > temporarily add #ifdef CPU_RESOLVING_TYPE in linux-user/main.c
> > so that  each target could gradually switch to cpu_create() and
> > not converted yet could continue to use cpu_init().
> > 
> > Once all targets are converted
> >     temporary ifdefs and cpu_init() API
> > will be removed
> > 
> > Signed-off-by: Igor Mammedov <address@hidden>
> > ---
> >  linux-user/main.c | 13 ++++++++++++-
> >  1 file changed, 12 insertions(+), 1 deletion(-)
> > 
> > diff --git a/linux-user/main.c b/linux-user/main.c
> > index 450eb3c..d481e8c 100644
> > --- a/linux-user/main.c
> > +++ b/linux-user/main.c
> > @@ -44,6 +44,7 @@ static const char *argv0;
> >  static int gdbstub_port;
> >  static envlist_t *envlist;
> >  static const char *cpu_model;
> > +static const char *cpu_type;
> >  unsigned long mmap_min_addr;
> >  unsigned long guest_base;
> >  int have_guest_base;
> > @@ -3847,7 +3848,7 @@ void init_task_state(TaskState *ts)
> >  CPUArchState *cpu_copy(CPUArchState *env)
> >  {
> >      CPUState *cpu = ENV_GET_CPU(env);
> > -    CPUState *new_cpu = cpu_init(cpu_model);
> > +    CPUState *new_cpu = cpu_create(cpu_type);
> >      CPUArchState *new_env = new_cpu->env_ptr;
> >      CPUBreakpoint *bp;
> >      CPUWatchpoint *wp;
> > @@ -4362,7 +4363,17 @@ int main(int argc, char **argv, char **envp)
> >      tcg_exec_init(0);
> >      /* NOTE: we need to init the CPU at this stage to get
> >         qemu_host_page_size */
> > +
> > +#ifdef CPU_RESOLVING_TYPE
> > +    cpu_type = cpu_parse_cpu_model(CPU_RESOLVING_TYPE, cpu_model);
> > +    if (!cpu_type) {
> > +        exit(1);
> > +    }
> > +    cpu_create(cpu_type);
> > +#else
> >      cpu = cpu_init(cpu_model);
> > +    cpu_type = object_class_get_name(object_get_class(OBJECT(cpu)));  
> 
> Will the final version of this series require the temporary
> #ifdefs?
> 
> I'd prefer to first apply patches adding the CPU_RESOLVING_TYPE
> defines and then patches adding the code without any #ifdef at
> all.
sure, it's possible to do without temporary #ifdef

> > +#endif
> >      env = cpu->env_ptr;
> >      cpu_reset(cpu);
> >  
> > -- 
> > 2.7.4
> >   
>