[6/8] ARM: imx8m: Fix reset in SPL on NXP iMX8MP EVK

On Mon, May 4, 2020 at 12:21 PM Harald Seiler <hws at denx.de> wrote:

> With or without the revert?

When I change the defconfig like:


It prints a single SPL line with the revert and also without the revert.

Hello Fabio,

On Mon, 2020-05-04 at 12:28 -0300, Fabio Estevam wrote:
> On Mon, May 4, 2020 at 12:21 PM Harald Seiler <hws at denx.de> wrote:
> 
> > With or without the revert?
> 
> When I change the defconfig like:
> 
> --- a/configs/imx8mp_evk_defconfig
> +++ b/configs/imx8mp_evk_defconfig
> @@ -57,7 +57,9 @@ CONFIG_ENV_IS_IN_MMC=y
>  CONFIG_SYS_RELOC_GD_ENV_ADDR=y
>  CONFIG_ENV_VARS_UBOOT_RUNTIME_CONFIG=y
>  CONFIG_SPL_DM=y
> +CONFIG_SPL_CLK_COMPOSITE_CCF=y
>  CONFIG_CLK_COMPOSITE_CCF=y
> +CONFIG_SPL_CLK_IMX8MP=y
>  CONFIG_CLK_IMX8MP=y
>  CONFIG_MXC_GPIO=y
>  CONFIG_DM_PCA953X=y
> 
> It prints a single SPL line with the revert and also without the revert.

Ok, I guess this means the imx8mp clock driver in SPL is broken and we
can't easily fix the DM_WATCHDOG issue without it.  I don't really know
much about imx8 nor do I have any hardware which I could use to debug with
so I can't help much with this.

Maybe Peng, who wrote the clock driver, can comment?

Otherwise, the series which contains this patch also fixed the non-DM
reset in SPL so we could revert back to that if all else fails ...

On 5/5/20 10:54 AM, Harald Seiler wrote:
> Hello Fabio,

Hi

> On Mon, 2020-05-04 at 12:28 -0300, Fabio Estevam wrote:
>> On Mon, May 4, 2020 at 12:21 PM Harald Seiler <hws at denx.de> wrote:
>>
>>> With or without the revert?
>>
>> When I change the defconfig like:
>>
>> --- a/configs/imx8mp_evk_defconfig
>> +++ b/configs/imx8mp_evk_defconfig
>> @@ -57,7 +57,9 @@ CONFIG_ENV_IS_IN_MMC=y
>>  CONFIG_SYS_RELOC_GD_ENV_ADDR=y
>>  CONFIG_ENV_VARS_UBOOT_RUNTIME_CONFIG=y
>>  CONFIG_SPL_DM=y
>> +CONFIG_SPL_CLK_COMPOSITE_CCF=y
>>  CONFIG_CLK_COMPOSITE_CCF=y
>> +CONFIG_SPL_CLK_IMX8MP=y
>>  CONFIG_CLK_IMX8MP=y
>>  CONFIG_MXC_GPIO=y
>>  CONFIG_DM_PCA953X=y
>>
>> It prints a single SPL line with the revert and also without the revert.
> 
> Ok, I guess this means the imx8mp clock driver in SPL is broken and we
> can't easily fix the DM_WATCHDOG issue without it.  I don't really know
> much about imx8 nor do I have any hardware which I could use to debug with
> so I can't help much with this.
> 
> Maybe Peng, who wrote the clock driver, can comment?
> 
> Otherwise, the series which contains this patch also fixed the non-DM
> reset in SPL so we could revert back to that if all else fails ...

Fix the clock driver. Reverting a patch to work around bugs is bad.