On Sun, Aug 02, 2015 at 06:13:50PM -0600, Simon Glass wrote:
This reverts commit 5b34436035fc862b5e8d0d2c3eab74ba36f1a7f4.
This function has a few problems. It calls fdt_parent_offset() which as mentioned in code review is very slow.
https://patchwork.ozlabs.org/patch/499482/ https://patchwork.ozlabs.org/patch/452604/
It also happens to break SPI flash on Minnowboard max which is how I noticed that this was applied. I can send a patch to tidy that up, but in any case I think we should consider a revert until the function is better implemented.
I suspect that what breaks is when you try to parse the "memory-map" property from this (taken from arch/x86/dts/minnowmax.dts):
spi { #address-cells = <1>; #size-cells = <0>; compatible = "intel,ich-spi"; spi-flash@0 { reg = <0>; compatible = "stmicro,n25q064a", "spi-flash"; memory-map = <0xff800000 0x00800000>; }; };
That's obviously not going to work because as Stephen and I have pointed out it specifies an address range in a different address space than that of the SPI flash. The spi parent node defines #address-cells and #size-cells, both of which define the layout of the "reg" property (and some others like "ranges"). I don't see a way around this other than to read two individual cells from the "memory-map" property. I don't know of a concept in DT that would allow you to translate from memory-map in the spi-flash@0 node to the address space in the parent node of the spi node (the root node).
Thierry