Re: [PATCH] x86/bigmem: eliminate struct domain address width restriction

[Ian Campbell <ian.campbell@citrix.com>]

On Tue, 2015-09-01 at 05:58 -0600, Jan Beulich wrote:
> > 
> > > > On 01.09.15 at 12:57, <ian.campbell@citrix.com> wrote:
> > On Wed, 2015-08-26 at 01:41 -0600, Jan Beulich wrote:
> > > @@ -225,10 +226,12 @@ static unsigned int __init noinline _dom
> > >  
> > >      return bits;
> > >  }
> > > +#endif
> > >  
> > >  struct domain *alloc_domain_struct(void)
> > >  {
> > >      struct domain *d;
> > > +#ifndef CONFIG_BIGMEM
> > >      /*
> > >       * We pack the PDX of the domain structure into a 32-bit field 
> > > within
> > >       * the page_info structure. Hence the MEMF_bits() restriction.
> > > @@ -237,12 +240,16 @@ struct domain *alloc_domain_struct(void)
> > >  
> > >      if ( unlikely(!bits) )
> > >           bits = _domain_struct_bits();
> > > +#else
> > > +# define bits 0
> > 
> > const unsigned int bits = 0;
> > 
> > Seems nicer than the trailing #undef?
> 
> Hmm, yes, in cases like this I agree (assuming that all compiler
> versions we care about will be intelligent enough to eliminate the
> variable).

Even if it weren't I don't think it would be noticeable in
alloc_domain_struct().

Ian.