Re: [PATCH 2/2] make new alloc_pages_exact()

[Dave Hansen <dave@linux.vnet.ibm.com>]

On Thu, 2011-04-07 at 15:03 -0700, David Rientjes wrote:
> On Thu, 7 Apr 2011, Dave Hansen wrote:
> > What I really wanted in the end was a highmem-capable alloc_pages_exact(),
> > so here it is.
> 
> Perhaps expand upon how the new alloc_pages_exact() works and what it will 
> be used for instead of only talking about how it's named?

Will do.

> > +/* 'struct page' version */
> > +struct page *__alloc_pages_exact(gfp_t, size_t);
> > +void __free_pages_exact(struct page *, size_t);
> 
> They're not required, but these should have the names of the arguments 
> like the other prototypes in this file.

Fair enough.

> > -	addr = __get_free_pages(gfp_mask, order);
> > -	if (addr) {
> > -		unsigned long alloc_end = addr + (PAGE_SIZE << order);
> > -		unsigned long used = addr + PAGE_ALIGN(size);
> > +	page = alloc_pages(gfp_mask, order);
> > +	if (page) {
> > +		struct page *alloc_end = page + (1 << order);
> > +		struct page *used = page + PAGE_ALIGN(size)/PAGE_SIZE;
> 
> Wouldn't it better to call this "unused" rather than "used" since it 
> represents a cursor over pages that we want to free?

Yeah, agreed.  I think I screwed this up once when coding this because I
misread it.  I'll change it.

-- Dave