RE: about __BITS_PER_LONG

["Guan Xuetao" <gxt@mprc.pku.edu.cn>]

> -----Original Message-----
> From: Arnd Bergmann [mailto:arnd@arndb.de]
> Sent: Monday, April 11, 2011 11:25 PM
> To: Guan Xuetao
> Cc: linux-kernel@vger.kernel.org; linux-arch@vger.kernel.org
> Subject: Re: about __BITS_PER_LONG
> 
> On Monday 11 April 2011, Guan Xuetao wrote:
> > --- a/include/asm-generic/bitsperlong.h
> > +++ b/include/asm-generic/bitsperlong.h
> > @@ -9,7 +9,11 @@
> >   * to decide it, but rather check a compiler provided macro.
> >   */
> >  #ifndef __BITS_PER_LONG
> > -#define __BITS_PER_LONG 32
> > +#   ifdef CONFIG_64BIT
> > +#      define __BITS_PER_LONG 64
> > +#   else
> > +#      define __BITS_PER_LONG 32
> > +#   endif /* CONFIG_64BIT */
> >  #endif
> >
> 
> Unfortunately, this does not work, because the __BITS_PER_LONG definition is meant
> for user space. You have to make this depend on a preprocessor macro that
> is provided by the compiler based on the command line switches (e.g. -m64) that
> a use could pass to the compiler.
However,  there is no conflict between the definition depending on a preprocessor macro and 
the definition of asm-generic header.

> 
> In user space, the CONFIG_* symbols are meaningless.
Yes, that's the problem. I have not notice it before.
So, only one value can be reserved in such circumstances.

> 
> 	Arnd

Thanks & Regards.

Guan Xuetao