Re: [PATCH] agpgart: Allow drm-populated agp memory types

[Arjan van de Ven <arjan@infradead.org>]

On Tue, 2007-01-09 at 09:23 +0100, thomas@tungstengraphics.com wrote:
> +
> +void agp_vkmalloc(size_t size, unsigned long **addr, u8 *vmalloc_flag)
> +{
> +	void *tmp = NULL;
> +
> +	*vmalloc_flag = 0;
> +
> +	if (size <= 2*PAGE_SIZE) {
> +		tmp = kmalloc(size, GFP_KERNEL);
> +	}
> +	if (tmp == NULL) {
> +		tmp = vmalloc(size);
> +		*vmalloc_flag = 1;
> +	}
> +
> +	*addr = tmp;
> +}
> +EXPORT_SYMBOL(agp_vkmalloc);


if you don't do this "fallback" thing the caller can just know it is
vmalloc due to the size... (and a 2 page kmalloc isn't going to fail on
you with GFP_KERNEL; also if you really want to deal with this failure
also tell the VM you're ok with failure (__GFP_NORETRY and such)...
but unless you do, things get a lot simpler by not doing this "fall
back" thing.

-- 
if you want to mail me at work (you don't), use arjan (at) linux.intel.com
Test the interaction between Linux and your BIOS via http://www.linuxfirmwarekit.org