Re: [PATCH]MTRR suspend/resume cleanup

[Shaohua Li <shaohua.li@intel.com>]

On Tue, 2005-06-28 at 13:24 +0800, Andrew Morton wrote:
> > @@ -332,6 +332,7 @@ int mtrr_add_page(unsigned long base, un
> > 
> >       error = -EINVAL;
> > 
> > +     lock_cpu_hotplug();
> >       /*  Search for existing MTRR  */
> >       down(&main_lock);
> >       for (i = 0; i < num_var_ranges; ++i) {
>        
> Is this new locking?  If so, please describe it.
> > +
> > +static int __init mtrr_init_finialize(void)
> > +{
> > +     if (!mtrr_if)
> > +             return 0;
> > +     if (use_intel())
> > +             mtrr_state_warn();
> > +     else
> > +     /* The CPUs haven't MTRR and seemes not support SMP. They have
> specific
> > +      * drivers, we use a tricky method to support suspend/resume
> for them
> > +      * TBD: is there any system with such CPU which supports
> suspend/resume?
> > +      * if no, we should remove the code.
> > +      */
> > +             sysdev_driver_register(&cpu_sysdev_class,
> > +                     &mtrr_sysdev_driver);
> > +     return 0;
> > +}
> 
> That's pretty hard to read.  Braces will improve it:
> 
>         if (use_intel()) {
>                 mtrr_state_warn();
>         } else {
>                 /*
>                  * The CPUs haven't MTRR and seemes not support SMP.
> They have
>                  * specific drivers, we use a tricky method to support
>                  * suspend/resume for them.
>                  * TBD: is there any system with such CPU which
> supports
>                  * suspend/resume?  If no, we should remove the code.
>                  */
>                 sysdev_driver_register(&cpu_sysdev_class,
>                         &mtrr_sysdev_driver);
>         }
> 
> > +#ifdef CONFIG_MTRR
> > +extern int mtrr_ap_init(void);
> > +extern int mtrr_bp_init(void);
> > +#else
> > +#define mtrr_ap_init()
> > +#define mtrr_bp_init()
> > +#endif
> 
> If this works, it means we're always ignoring the return values and
> these
> things should return void.
> 
> Otherwise the !CONFIG_MTRR stubs should be returning an integer.  So
> we
> need either:
> 
> #ifdef CONFIG_MTRR
> extern void mtrr_ap_init(void);
> extern void mtrr_bp_init(void);
> #else
> #define mtrr_ap_init()  do {} while (0)         (or static inlines)
> #define mtrr_bp_init()  do {} while (0)
> #endif
> 
> or
> 
> #ifdef CONFIG_MTRR
> extern int mtrr_ap_init(void);
> extern int mtrr_bp_init(void);
> #else
> static inline int mtrr_ap_init(void) { return 0; }
> static inline int mtrr_bp_init(void) { return 0; }
> #endif
> > +#ifdef CONFIG_MTRR
> > +extern int mtrr_ap_init(void);
> > +extern int mtrr_bp_init(void);
> > +#else
> > +#define mtrr_ap_init()
> > +#define mtrr_bp_init();
> > +#endif
> Ditto.
Thanks for your feedback. I will update the patch when I collect all
comments.

Thanks,
Shaohua