On Fri, Aug 03, 2012 at 06:40:58PM +0200, Daniel Schwierzeck wrote:
Hi Tyler,
2012/8/3 Tyler Olmstead tyler.j.olmstead@gmail.com:
Yes, the #ifndef works perfectly for me. However, I also agree with your sentiment regarding build magic, which is why I wonder if removing the $GEN_UBOOT linker magic from the SPL makefile wouldn't be the best approach. If this was done, then my U-Boot command wouldn't have been linked into SPL in the first place, it wouldn't require any cluttering of #ifdef's, and would eliminate the potential of others encountering this same problem. This seems reasonable given that SPL shouldn't contain any command support. Thoughts?
Most of the spl/Makefile code was copied and adapted from the top Makefile. Unfortunately, we have not optimized the GEN_UBOOT macro so it still has the UNDEF_SYM magic which is obviously unnecessary.
I would suggest to remove the UNDEF_SYM magic.
So, this is another one of the problems with relying on the linker to discard stuff that's not needed. Current omap3_beagle: Configuring for omap3_beagle board... text data bss dec hex filename 326264 8460 266916 601640 92e28 omap3_beagle/u-boot 42856 1812 198020 242688 3b400 omap3_beagle/spl/u-boot-spl
Remove UNDEF_SYM, remove guards around the nandecc command in arch/arm/cpu/armv7/omap3/board.c: Configuring for omap3_beagle board... text data bss dec hex filename 326274 8460 266904 601638 92e26 omap3_beagle/u-boot 43014 1812 198020 242846 3b49e omap3_beagle/spl/u-boot-spl
So we don't discard the command and SPL grows slightly (confirmed by still removing UNDEF_SYM but putting the guards back on nandecc).
I don't know if we need to be more aggressive in linker commands or what but I know there's lots of other code not being discarded too, from when I poked at NAND before.