Hi Wolfgang et al., (replying both to 5/9 and 6/9 here)
Le 25/04/2011 01:55, Wolfgang Denk a écrit : > This is an inconsistency between architectures which I dislike. Then these two config options could be defined for other architectures as well. > Also, what in case we should want to skip only cpu_init_crit and not > lowlevel_init? If we need to handle these separately, then we need 2 > CONFIG options. Agreed this complete separation could be useful in case U-Boot is not the code executed on startup, and this startup code does the CPU initialization but leaves the rest to U-Boot for commodity. > How can the SDRAM be initialized when the CPU is not? Debugging sessions come to mind, where the debugger resets the target and initializes SDRAM to load the target code. I've also seen cases on multiple-core ICs where an ARM core starts with SDRAM access already working, but U-Boot was not involved. The idea is that during board bring-up, rather than have people try and neutralize some code in assembly language startup files, they just temporarily define the configuration options CONFIG_SKIP_* in their board's config header file (and, for those boards that have an IPL of some sort before U-Boot, they keep the relevant CONFIG_SKIP_* there once debugged). > Best regards, > > Wolfgang Denk Amicalement, -- Albert. _______________________________________________ U-Boot mailing list U-Boot@lists.denx.de http://lists.denx.de/mailman/listinfo/u-boot
