2005-11-16 15:22:56

by Mika Penttilä

Subject: DMA32 zone unusable

The new DMA32 zone (which at least x86-64 has) is quite "interesting" :

#define __GFP_DMA32 ((__force gfp_t)0x04) <-----!!!!!

#define GFP_ZONEMASK 0x03 <------!!!!!

#define gfp_zone(mask) ((__force int)((mask) & (__force gfp_t)GFP_ZONEMASK))

static inline struct page *alloc_pages_node(int nid, gfp_t gfp_mask,
unsigned int order)
{
if (unlikely(order >= MAX_ORDER))
return NULL;

return __alloc_pages(gfp_mask, order,
NODE_DATA(nid)->node_zonelists + gfp_zone(gfp_mask));
}


So with GFP_DMA32 you never get those pages (but DMA instead).

--Mika

2005-11-16 15:53:17

by Andi Kleen

Subject: Re: DMA32 zone unusable

On Wednesday 16 November 2005 16:22, Mika Penttil? wrote:
> The new DMA32 zone (which at least x86-64 has) is quite "interesting" :
>
> #define __GFP_DMA32 ((__force gfp_t)0x04) <-----!!!!!
>
> #define GFP_ZONEMASK 0x03 <------!!!!!
>
> #define gfp_zone(mask) ((__force int)((mask) & (__force gfp_t)GFP_ZONEMASK))

Yes that was a last minute change that was wrong. I will submit
an update. Thanks for reviewing.

-Andi

> static inline struct page *alloc_pages_node(int nid, gfp_t gfp_mask,
> unsigned int order)
> {
> if (unlikely(order >= MAX_ORDER))
> return NULL;
>
> return __alloc_pages(gfp_mask, order,
> NODE_DATA(nid)->node_zonelists + gfp_zone(gfp_mask));
> }
>
>
> So with GFP_DMA32 you never get those pages (but DMA instead).
>
> --Mika
>
> --
> To unsubscribe, send a message with 'unsubscribe linux-mm' in
> the body to [email protected]. For more info on Linux MM,
> see: http://www.linux-mm.org/ .
> Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
>