
29 Jan
2008
29 Jan
'08
10:11 a.m.
Le lundi 28 janvier 2008 à 18:22 +0100, Ulf Samuelsson a écrit :
Of course, u-boot does only need a (small) subset of it. If you prefer, I can simplify it.
==> Which will make it harder to maintain.
Note however this is the path the Linux kernel developers chose too.
These files are automatically generated from the chip database, and most of the stuff can conceivably be used by U-Boot. -- Even if they are not used today.
Smells like over-engineering at this point.
If a bug is detected in the file, then the place to update is inside the database. Obviously ansy script generating the data, should be modified to generate something which can be directly used by U-Boot without any further modification.
The question is whether it can be possible to make such a script to generate header files in accordance with the coding styles :)
It may make sense to split up the file as well, as Haavard proposes, but preferably, the split should be done at the source. I will forward this request to the AT91 & CAP teams
Sure, if they do this it would be perfect. Meanwhile, let's start with the one I've submitted (30 kb is better than 400 kb anyway). If/when Atmel cooks up clean and splitted header files then we will import those ones.
--
Stelian Pop stelian@popies.net