Re: [coreboot] Changes for removing IDs from dts, etc. etc.

[Carl-Daniel Hailfinger]

On 15.02.2008 19:23, ron minnich wrote:
> This turned into a pretty big cleanup. However, it reduced (by a small
> amount) the amount of code, and complexity, and I like it.
>
> Comments welcome. This one is signed off, though there is a bit of
> todo left as noted in the patch.
>
> boots on qemu and alix1c.

Very nice.
> TODO:
> final removal of support for/use of 'constructor' property, replace with
>
> 'device_operations' property

Your decision whether to do this now or later.
> change lpc back to superio

Please do so before commit if at all possible. I'm not sure the
"superio" keyword is the right one, but keeping that change out of the
commit helps reduce code churn.
> Change limitation in dtc that makes it hard to use hex in pci@ notation.

Can wait for a later patch.
> whatever else you tell me to do :-)

Two questions about superio/lpc specification in the dts, though.
- If a southbridge requires special setup before the superio is
accessible, where do we handle that?
- A superio is attached via LPC, but there is no way to be sure that
future superios will be addressed via I/O Ports. Do we want to change
the superio part of the dts to something like this:

===================================================================
--- mainboard/pcengines/alix1c/dts	(revision 600)
+++ mainboard/pcengines/alix1c/dts	(working copy)
@@ -54,7 +45,7 @@
    		 * See virtual PIC spec. */
    		enable_gpio_int_route = "0x0D0C0700";
    	};
-		superio {
+		ioport@46 {
    		/config/("superio/winbond/w83627hf/dts");
    		com1enable = "1";
    	};
> Signed-off-by: Ronald G. Minnich

With point 2 from the TODO fixed or an answer for the superio/lpc/ioport
problem, the patch is
Acked-by: Carl-Daniel Hailfinger

Thanks for doing that work! The code/dts structure is much cleaner and
more obvious now.

Regards,
Carl-Daniel