[Openchrome-devel] Early TTM patch

[James Simmons]

> > This patch is against the drm-next branch. Give it a try with the KMS xorg 
> > branch and let me know how it works. While you test that I'm going to put 
> > some gem test together to exercise the driver.
> > 
> The VRAM probing for the KM400/KM400A/P4M800 (chipset id 0x7205) will
> fail when on the P4M800 variant because it's using a different host
> bridge and then the VRAM needs to be read from another register. The
> chipset to host bridge mapping seems to be 1:1 in all but this case.
> The VRAM probing is broken in this case too for UMS, but it might worths
> fixing it right from the start for KMS.
> There's a bug filed for this issue :
> http://www.openchrome.org/trac/ticket/177
> 
> See
> http://cgit.freedesktop.org/~libv/xf86-video-unichrome/tree/src/via_host.c
> for a host bridge probing implementation.

Just like the SiS family and some embedded Nvidia chipsets. The current 
code in my patch should work since it does talk to the north bridge. I'm 
going to need to access the north bridge anyways to get the memory type 
that the card uses.

P.S
	I did post a second version of the TTM patch at the same place. 

http://www.infradead.org/~jsimmons/via-ttm.diff

The big difference is that besides the old api of allocation a chunk of 
memory at a offset in a particular region you can now allocate the 
size you want from any memory pool, which is the expected behavior of 
TTM/GEM. I did  test that using a special libkms with openchrome support 
that I put together. Still some issues tho with freeing the memory.