On 10 April 2018 at 09:51, Neil Armstrong <narmstr...@baylibre.com> wrote:
> On 08/04/2018 15:50, Simon Glass wrote:
>> On 28 March 2018 at 05:54, Neil Armstrong <narmstr...@baylibre.com> wrote:
>>> The Amlogic SoCs have a registers containing the die revision
>>> and packaging type to determine the SoC family and package marketing
>>> name like S905X for the GXL SoC Family.
>>> This code is taken for the Linux meson-gx-socinfo driver and adapted
>>> to U-Boot printing.
>>> Signed-off-by: Neil Armstrong <narmstr...@baylibre.com>
>>> arch/arm/include/asm/arch-meson/gx.h | 1 +
>>> arch/arm/mach-meson/Makefile | 2 +-
>>> arch/arm/mach-meson/cpu_info.c | 105
>>> configs/khadas-vim_defconfig | 2 +-
>>> configs/libretech-cc_defconfig | 2 +-
>>> configs/odroid-c2_defconfig | 2 +-
>>> configs/odroid_defconfig | 1 +
>>> configs/p212_defconfig | 2 +-
>>> 8 files changed, 112 insertions(+), 5 deletions(-)
>>> create mode 100644 arch/arm/mach-meson/cpu_info.c
>> Please can you use driver model's CPU interface for this?
> Hi Simon,
> The CPU uclass is designed for the /cpu/* nodes, here we fetch the SoC
> in some /soc/ subnodes which cannot be used with the current CPU uclass and
> won't be probed before relocation.
> Either I push is as a MISC driver (with the pre-reloc issue) or we leave this
> in mach-meson.
> What do you think ?
Do you think it could search both /cpu and /soc ? They seem to be
fairly equivalent. Is that the only problem?
U-Boot mailing list