Bug 1662

(In reply to comment #0)
> Hi. I'm using the 2.6.9-rc4-love1 kernel (based on the mm kernel) and while the 
> Radeon driver compiles okay, it failes to load because of an unknown symbol 
> called remap_page_range. I did a search for this and apparently Andrew Morton 
> has changed the name of the function and the arguments it takes. This page only 
> mentions the name change but I tried simply changing the name in the source and 
> that didn't work.
> 
> http://www.ussg.iu.edu/hypermail/linux/kernel/0409.3/1106.html
> 
> This is fairly bleeding edge stuff so maybe you don't want to address the issue 
> yet but I thought I'd make you aware of it in case this change makes it back to 
> the mainstream kernel.

Yeah changing the name alone wont work
The problem is that the two functions remap_page_range and remap_pfn_range count
sizes in different units.

As a work around try this.

In drm_vm.h (or someother header file add this function definition)
 
---------------8<----------------8<-------------------------------------------
static inline __deprecated /* since 25 Sept 2004 -- wli */
int remap_page_range(struct vm_area_struct *vma, unsigned long uvaddr,
                     unsigned long paddr, unsigned long size, pgprot_t prot)
{
  return remap_pfn_range(vma, uvaddr, paddr >> PAGE_SHIFT, size, prot);
}
---------------8<-----------------8<------------------------------------------

This is only a temporary work around and might not compile in older kernels that
already have the definition of remap_page_range.  I am putting it here because
it worked for me (savage on linux 2.6.10-rc1-mm2)

A cleaner solution would be to eliminate usage of remap_page_range but I am not
sure of backward compat.

Best of luck

--
Rajsekar
IIT Madras
raj--cutme--sekar@cse.iDELTHISitm.ernet.in

I'm using the Gentoo kernel now which isn't effected by the problem but thanks 
anyway. It may help someone else out.

This is fixed in DRM CVS.