Re: "struct page" and allocating high memory

[Rene Herman <rene.herman@xxxxxxxxxxxx>]

On 28-04-08 22:15, Rene Herman wrote:
> On 27-04-08 18:44, Robert P. J. Day wrote:
>
> >   so what does that comment mean?  it seems to be exactly backwards
> > from what i'm used to believing.
>
> The comment seems confused yes. alloc_pages() is just the buddy 
> allocator and gets you contiguous pages. For vmalloc() purposes, an 
> array of struct page * is allocated, allocating one page at a time with 
> alloc_page() (in __vmalloc_area_node).

Err. That's more me who seems confused. The vmalloc() bit is correct and 
yes, alloc_pages() is "just the buddy allocator and gets you [physically] 
contiguous pages" but that in fact ofcourse means the comment is fine. The 
single struct page * is fine for a higher order alloc_pages, since its's 
adjacent struct page pointers in the mem_map. That is:

	struct page *page = alloc_pages(gfp_mask, order);

means your struct page structures are at *page to *(page + 2^order) in the 
node's mem_map, even if they are highmem pages.

Rene.

--
To unsubscribe from this list: send an email with
"unsubscribe kernelnewbies" to ecartis@xxxxxxxxxxxx
Please read the FAQ at http://kernelnewbies.org/FAQ