-----Original Message-----
From: Stephen Boyd [mailto:sboyd@kernel.org]
Quoting Aisheng Dong (2018-12-10 18:36:29)

quoted

quoted

-----Original Message-----
From: Stephen Boyd [mailto:sboyd@kernel.org]
Sent: Tuesday, December 11, 2018 4:55 AM

[...]

quoted

Subject: Re: [PATCH V9 1/7] clk: imx: add configuration option for
mmio clks

Quoting Aisheng DONG (2018-12-04 06:39:22)

quoted

diff --git a/arch/arm/mach-imx/Kconfig b/arch/arm/mach-imx/Kconfig
index c12a05c..303082c 100644
--- a/arch/arm/mach-imx/Kconfig
+++ b/arch/arm/mach-imx/Kconfig

@@ -570,6 +580,7 @@ config SOC_IMX7ULP  config SOC_VF610
        bool "Vybrid Family VF610 support"
        select ARM_GIC if ARCH_MULTI_V7
+       select MXC_CLK
        select PINCTRL_VF610

Instead of select can we break this dependency on the arch Kconfig
and
have:

      config MXC_CLK
              def_bool ARCH_MXC && !ARM64

Thanks for the suggestion.
One little problem is that we also have LS1021 supported under
ARCH_MXC which does not use MXC_CLK.

Ok. So then your change is also limiting the compilation of drivers/clk/imx/ to
only the platforms that select it now, instead of all ARCH_MXC like it was done
before.

We can also have def_bool <big list of SoC platforms> if that helps for the
optimization that this patch is making.

arm-soc has generally pushed against having so many different arch level
Kconfig options because they make for unwieldy Kconfig fragments that may
become conflicting. Instead, the options for the different SoCs are removed
and we just have one for the platform and rely on defconfigs to pick the right
drivers.

Thanks for telling the detailed background. Very clear now.

Why can't we do that here?

quoted

quoted

quoted

diff --git a/drivers/clk/imx/Makefile b/drivers/clk/imx/Makefile

index
5c0b11e..f850424 100644

--- a/drivers/clk/imx/Makefile
+++ b/drivers/clk/imx/Makefile

@@ -1,6 +1,6 @@
 # SPDX-License-Identifier: GPL-2.0

-obj-y += \
+obj-$(CONFIG_MXC_CLK) += \

Because this changes to obj-$(CONFIG) we should change the
drivers/clk/Makefile to have obj-y for this Makefile instead of
depending on ARCH_MXC.

We also use ARCH_MXC for ARMv8 SoC clocks.

Yes, and? I'm suggesting drivers/clk/Makefile have an obj-y += drivers/clk/imx/
and then decide to compile or not compile the MXC_CLK "core" bits based on
CONFIG_MXC_CLK config option.

Sorry for the missunderstanding before.
I guess you point is have an ojb-y += drivrs/clk/imx/
Then in drivers/clk/imx/Kconfig
config MXC_CLK
		def_bool ARCH_MXC && !ARM64
		select MXC_clk bits
config MXC_CLK_SCU
		def_bool ARCH_MXC && ARM64
		select SCU clock bits

If wrong please let me know.

Will update the patch series and re-send.
Thanks for the suggestion.

Regards
Dong Aisheng
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel