On Sunday, October 03, 2010 14:27:13 Wolfgang Denk wrote:
> Mike Frysinger wrote:
> > Many people like the current nand_init() behavior where it is always
> > initialized during boot and the flash size shown, but there are cases
> > where we are willing to forgo this niceness for speed/functionality.
> > So rather than change the default, introduce a delayed config option
> > people may enable. This way the nand is only poked when someone tries
> > to actually use it.
> >
> > extern void nand_init(void);
> >
> > +#ifdef CONFIG_SYS_NAND_DELAYED_INIT
> > +# define nand_delayed_init() nand_init()
> > +#else
> > +# define nand_delayed_init() do { } while (0)
> > +#endif
>
> Would it not be esier to rename your nand_delayed_init() into
> nand_init(), and add a "#ifndef CONFIG_SYS_NAND_DELAYED_INIT" around
> the current call to nand_init()?nand_init() cant handle being called multiple times. and i need to add more nand_init() points that only apply to when things are delayed. so when delayed init is not enabled (the default), there is no change in compiled code size. > Question: is there a risk of problems with boards that have the > environment in NAND? that's why my patch adds delayed init points to the major nand env entry points. my understanding is that these must be called before the env read/write funcs may be called. -mike
signature.asc
Description: This is a digitally signed message part.
_______________________________________________ U-Boot mailing list U-Boot@lists.denx.de http://lists.denx.de/mailman/listinfo/u-boot
