[coreboot] patch: add mcp55 pcie device
c-d.hailfinger.devel.2006 at gmx.net
Tue Aug 12 03:50:12 CEST 2008
On 11.08.2008 18:29, ron minnich wrote:
> Committed revision 743.
> On Mon, Aug 11, 2008 at 3:53 AM, Carl-Daniel Hailfinger
>> How about multi-level dts instead of the current two-level model? An
>> intermediate level could aggregate device-level dts. The mainboard dts
>> would include intermediate-level dts unless low-level stuff was needed.
> Horrors. Let's not go there right now. This has already been the
> subject of multi-month discussions. Keep it simple and regular.
I missed those discussions. It seems I should be happy I missed them.
>>> Note that it *is* clearer from the point of view of seeing exactly what
>>> is being built in in one place: the dts. In v2, the magic of linker scripts
>>> added considerable confusion, and we got complaints about that, so
>>> we are trying to avoid such magic in v3.
>> Point taken.
> Let's never forget: we need to think about it from the point of view
> of non-coreboot-experts, i.e. people who are using the tool, and might
> have to change it, but don't wnat to make a living at it.
If anything, they are most likely to use automated tools (to be written)
for dts creation. But the important thing is that people should not be
forced to understand magic multilevel include stuff.
> That should be our target: people who don't do this for a living and
> don't want to -- they just want a quick build and an easy set of
> changes. One issue that came up on OLPC was that a number of people
> found v2 hard to figure out, and they did not have time to plumb its
> depths. I don't blame them one bit.
> I think u-boot does a great job that way. You can walk into that code
> and find your way around. It's not nearly as capable as v2 but i can
> tell you -- there are some big PPC machines running u-boot, not
> coreboot, and the choice was made by people who looked at both
I have not looked at U-Boot code in depth, but coreboot v3 code was
pretty readable and the structure was intelligible some weeks ago. I
strive to keep it that way.
> I think v2 veered off the path to some degree, in recent years, as
> more and more complexity was added in via cpp tricks, linker sets and
> complex code paths. In some sense, I have been trying to reduce
> trickiness and increase transparency with v3. I hope it works out.
I'll keep an eye on it as well.
>> Could you dig up the revision? I'd like to take a look. Thanks.
> I don't think I can :-) I did that in the very early going at FOSDEM
> 2007 and we killed it about a year later IIRC. It was documented in
> early versions of the newboot.lyx document.
I'll read through that once I have time.
>> With the comments below addressed:
> renamed per your suggestions.
More information about the coreboot