Re: [RFC PATCH 2/7] lib: Add huge I/O map capability interfaces

[Andrew Morton <akpm@linux-foundation.org>]

On Mon, 26 Jan 2015 18:01:55 -0700 Toshi Kani <toshi.kani@hp.com> wrote:

> > >  static int ioremap_pte_range(pmd_t *pmd, unsigned long addr,
> > >  		unsigned long end, phys_addr_t phys_addr, pgprot_t prot)
> > >  {
> > > @@ -74,6 +112,12 @@ int ioremap_page_range(unsigned long addr,
> > >  	unsigned long start;
> > >  	unsigned long next;
> > >  	int err;
> > > +	static int ioremap_huge_init_done;
> > > +
> > > +	if (!ioremap_huge_init_done) {
> > > +		ioremap_huge_init_done = 1;
> > > +		ioremap_huge_init();
> > > +	}
> > 
> > Looks hacky.  Why can't we just get the startup ordering correct?  It
> > at least needs a comment which fully explains the situation.
> 
> How about calling it from mm_init() after vmalloc_init()?  
> 
> void __init mm_init(void)
> 		:
>         percpu_init_late();
>         pgtable_init();
>         vmalloc_init();
> +       ioremap_huge_init();
>  }

Sure, that would be better, assuming it can be made to work.  Don't
forget to mark ioremap_huge_init() as __init.