Hi,
You mentioned that /spl can not be used for source files. Isn't there a way to workaround this problem?
Why should we have source files in a SPL directory? I would prefer to have spl specific sources right where the rest ist - maybe marked with something like _spl or excluded by some #define-test. If we have a SPL specific directory we have to copy most of the tree (arch/cpu etc.) which in my eyes is totally unnecessary if we don't do the symlinking stuff...
Also, I agree with Scott's opinion that re-compiling some files while re-using the binary of some other files won't be a good idea. In this case, CONFIG_PRELOADER will be honored in some files but not in other files. That will be a source of confusion for developers.
I also see this as the biggest problem with reusing the object-files. It will add more complexity than a simple re-run with different flags like suggested by Daniel.
BTW, John Rigby had sent out a series sometime back for OMAP3 NAND SPL. That can be integrated with my work and we will get an SPL that supports both MMC and NAND. I guess Simon Schwarz is also doing some work lately on OMAP3.
I am working on OMAP3 (on devkit8000). If this discussion comes to a conclusion soon I would prefer sending the patches with the new SPL format.
Regards Simon