Dear Tom Rini,
In message 1397047800-26221-1-git-send-email-trini@ti.com you wrote:
The fdt_fixup_memory function is only used on PowerPC where we only claim one memory bank in U-Boot (and then in the device tree) so we can call a function that just calls fdt_fixup_memory_banks(..., 1). Call this directly for consistency with other architectures.
I understand what you mean and what you want, but I'm not really happy about it.
First, the description is not correct. In my understanding a "bank" of memory is some memory device which, on the hardware level, is addressed using one specific chip select signal. On PPC, we usually have flexible memory controllers, so e can always map all existing memory banks such that they for a single, contiguous region.
It is causing me some creepes to introduce code that claims it is fixing memory for only one bank - I feel this is wrong.
And is dropping the (u64) not a problem? bd->bi_memstart is just an "unsigned long", but fdt_fixup_memory_banks() expects a u64 ?
Best regards,
Wolfgang Denk