Tom <tom....@windriver.com> wrote on 26/11/2009 20:18:22: > > Joakim Tjernlund wrote: > >> From: Alessandro Rubini <rub...@unipv.it> > >> > >> While it's a matter of personal taste, I prefer to avoid ifdef when > >> possible. For example, I don't like to add BOARD_LATE_INIT in the > >> config file just to add a board_late_init() function. > >> Also, I think the file is more readable without the ifdef stuff. > >> This uses two trivial weak functions to provide defaults for all > >> functions that were ifdeffed. > >> > >> This patch was initially rejected in favor of a initcall mechanism > >> but that approach is not a work in progress any ore. > >> --- > >> > >> One complaint I got about this is the runtime overhead. Actually, > >> normal_nop is two instructions (plus the call to it) and void_nop > >> is one instruction (plus the call), similar to the overhead in > >> led management for platforms with no leds. > > > > I think (I have proposed this before) that you should get rid of > > the init_fnc_t *init_sequence[] array all together. It only > > adds relocation overhead(lots of it). Just make the array a function > > that calls the init functions directly. > > > > Please resend the RFC patch. > This would also go the arm/testing-arm_init branch.
I just sent a complete patch for ppc(as that is my platform). You choose if you want to do a similar one for ARM. There probably wont be any space savings for ARM as you don't use relocation yet. Jocke _______________________________________________ U-Boot mailing list U-Boot@lists.denx.de http://lists.denx.de/mailman/listinfo/u-boot