Re: [PATCH v4 07/10] ARM: rzn1: basic support for Renesas RZ/N1 SoC

[Ralph Siemsen <ralph.siemsen@linaro.org>]

On Mon, Apr 17, 2023 at 07:15:07PM +0200, Marek Vasut wrote:
>On 3/8/23 21:26, Ralph Siemsen wrote:
>
>[...]
>
>>+++ b/arch/arm/Kconfig
>>@@ -1031,6 +1031,21 @@ config ARCH_RMOBILE
>>  	imply SYS_THUMB_BUILD
>>  	imply ARCH_MISC_INIT if DISPLAY_CPUINFO
>>+config ARCH_RZN1
>>+	bool "Reneasa RZ/N1 SoC"
>>+	select CLK
>>+	select CLK_RENESAS
>>+	select CLK_R9A06G032
>
>Does 'select CLK_R9A06G032' automatically activate the dependencies 
>like 'CLK_RENESAS' too ?

Do you mean things like CLK_RCAR_CPG_LIB for example?

The new clock driver has no dependencies (except perhaps ARCH_RZN1). 
Since this clock driver is essential to boot, I figured it was best to 
select it in Kconfig, rather than require each board defconfig to 
specify the same options.

If there is a better/preferred approach, I will implement it.

>
>>+	select DM
>>+	select DM_ETH
>>+	select DM_SERIAL
>>+	select PINCTRL
>>+	select PINCONF
>>+	select REGMAP
>>+	select SYSRESET
>>+	select SYSRESET_SYSCON
>>+	imply CMD_DM
>>+
>>  config ARCH_SNAPDRAGON
>>  	bool "Qualcomm Snapdragon SoCs"
>>  	select ARM64
>>@@ -2207,6 +2222,8 @@ source "arch/arm/mach-owl/Kconfig"
>>  source "arch/arm/mach-rmobile/Kconfig"
>>+source "arch/arm/mach-rzn1/Kconfig"
>
>Should this be in mach-rmobile (which, maybe, should be renamed to 
>mach-renesas) ?

I vaguely recall that I discussed this with someone (possibly you), 
likely on IRC. And I think the conclusion was there was a significant 
enough difference (possibly Cortex-A7 versus A9?).

I don't have a strong preference, if moving to mach-rmobile makes sense, 
let me know and I can give it a go.

>>+#if !CONFIG_IS_ENABLED(SYS_DCACHE_OFF)
>>+void enable_caches(void)
>>+{
>
>Why not enable icache with icache_enable() too ?

The I-cache is enabled quite early in arch/arm/cpu/armv7/start.S

Ralph