Re: [PATCH]agp for i820 chipset

Robert Love (rml@tech9.net)
05 Nov 2001 15:19:12 -0500


On Mon, 2001-11-05 at 13:03, Nicolas Aspert wrote:
> This would be great. This source _does_ need cleanup. Since I saw taht
> everyone did copy all the functions I just did the same :-)

Yah, it is currently a mess. I hope to have it done soon.

> I sent the patch to linux-kernel, but do you think I have to send it
> personnally to Alan & Linus ?

You will certainly need to send it to Linus, perhaps multiple times :)

Alan may pick it up, but if he doesn't it wouldn't hurt to cut a diff
against his most recent kernel and send it to him and the list.

> > I'm not too sure why you need this. I see other chipsets have their
> > device 0:01 defined but I can't reason why. When I add AGP drivers I
> > never add it. If you remove it, I think you will find everything still
> > works.
>
> Damn ! You're right :-). The first entry is needed, because the i810
> chipset uses the secondary device (at least this is what is written in
> the code. see the 'agp_find_supported_device' routine. I remember this
> is needed for on-board chips. Does that make any sense :-) ? ), but the
> i820 related one is useless (afaik).

Yah, you should be able to remove this no problem.

> > You can just use intel_generic_fetch_size or even one of the
> > i840-specific or whatever versions, here. Note you don't use anything
> > specific to the i820, so reduce the footprint and ditch it.
>
> The reason for rewriting this function is that, according to the Intel specs,
> the APSIZE register is 8bits long (at least for the i820)! The generic function reads
> 16 bits, so who knows what will be in the neighbouting register ? I guess it was
> working by accident, but if you have an argument for sticking to the generic 'fetch_size'
> I am all ready to replace my part by the generic one :-)

Ohh, I didn't notice this. It is odd the two sizes don't match. If
they really are different then I agree; don't use the generic one.

> I have to leave by now, I'll check my e-mail tonight, but it is
> likely that I won't make any patch before tomorrow :-(

No problem ... send it along when you do.

Robert Love

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/