[U-Boot] [PATCH v4 12/17] board_f: Use static print_cpuinfo if CONFIG_CPU is active

[Stefan Roese]

Hi Bin,

On 12.10.2018 17:34, Bin Meng wrote:
> On Fri, Oct 12, 2018 at 10:00 PM Stefan Roese <sr at denx.de> wrote:
>>
>> Hi!
>>
>> I just tested the latest U-Boot version on one of my x86
>> platforms (Baytrail theadorable-x86-dfi-bt700_defconfig).
>> Unfortunately this patch break the board port, resulting in
>> this output:
>>
>> U-Boot 2018.11-rc1-00268-g1f7888b626 (Oct 12 2018 - 15:23:52 +0200)
>>
>> initcall sequence fff5aa40 failed at call fff129bc (err=-19)
>> ### ERROR ### Please RESET the board ###
>>
>> This is related to the fact that this platform does not support
>> a proper DM CPU driver. When I revert this patch, all is fine
>> again:
>>
>> U-Boot 2018.11-rc1-00269-g82e3eecc2b (Oct 12 2018 - 15:50:36 +0200)
>>
>> CPU: x86_64, vendor Intel, device 30679h
>> DRAM:  4 GiB
>> MMC:   pci_mmc: 0, pci_mmc: 1, pci_mmc: 2
>> Loading Environment from SPI Flash... SF: Detected w25q64cv with page size 256 Bytes, erase size 4 KiB, total 8 MiB
>> OK
>> Model: theadorable-x86-DFI-BT700
>> ...
>>
>> How should we proceed with this patch. Since multiple boards /
>> platforms could be affected (I remember this also hitting me
>> with my latest MIPS port), we should probably revert this patch
>> for this upcoming release.
>>
>> Thoughts?
>>
> 
> Thanks for testing and reporting this! I suspect the following patch
> can resolve this:
> 
> diff --git a/arch/x86/cpu/baytrail/cpu.c b/arch/x86/cpu/baytrail/cpu.c
> index 56e9813..2eb9172 100644
> --- a/arch/x86/cpu/baytrail/cpu.c
> +++ b/arch/x86/cpu/baytrail/cpu.c
> @@ -203,4 +203,5 @@ U_BOOT_DRIVER(cpu_x86_baytrail_drv) = {
>          .bind           = cpu_x86_bind,
>          .probe          = cpu_x86_baytrail_probe,
>          .ops            = &cpu_x86_baytrail_ops,
> +       .flags          = DM_FLAG_PRE_RELOC,
>   }
> 
> This patch needs to be applied on top of u-boot-x86/dm-fixes. Could
> you please have a try?

I've applied your patches from u-boot-x86/dm-fixes and added
the .flag as suggested above. And yes, this fixes this issue.

Many thanks for the quick help,
Stefan