cpufreq: scmi: add COMMON_CLK dependency

From: Arnd Bergmann <[email protected]>

Wtihout CONFIG_COMMON_CLK, the scmi driver fails to link:

arm-linux-gnueabi-ld: drivers/cpufreq/scmi-cpufreq.o: in function `scmi_cpufreq_probe':
scmi-cpufreq.c:(.text+0x20c): undefined reference to `devm_of_clk_add_hw_provider'
arm-linux-gnueabi-ld: scmi-cpufreq.c:(.text+0x22c): undefined reference to `of_clk_hw_simple_get'

Add a Kconfig dependency for it.

---
 drivers/cpufreq/Kconfig.arm | 1 +
 1 file changed, 1 insertion(+)

index 1f73fa75b1a0..434ef03d2762 100644
@@ -264,6 +264,7 @@ config ARM_SA1110_CPUFREQ
 config ARM_SCMI_CPUFREQ
 	tristate "SCMI based CPUfreq driver"
 	depends on ARM_SCMI_PROTOCOL || COMPILE_TEST
+	depends on COMMON_CLK
 	select PM_OPP
 	help
 	  This adds the CPUfreq driver support for ARM platforms using SCMI

2.27.0

There is a fix already upstream in later -rc(rc6 IIRC), I assume you are seeing this prior to that.

Commit f943849f7206 ("cpufreq: scmi: Fix build for !CONFIG_COMMON_CLK")

Since the only dependency on CONFIG_COMMON_CLK is to satisfy OPP adding dummy clock provider, I avoided adding dependency on CLK for this driver as this works fine for !CONFIG_COMMON_CLK.

--
Regards,
Sudeep

P.S: There are 2 copies of this patch, I chose to reply on this, other one is @[1]

[1] https://lore.kernel.org/lkml/[email protected]

Ok, I missed that during my rebase.

Right. I don't think it makes much difference either way, though I usually prefer code checks to be written as

        /* dummy clock provider as needed by OPP if clocks property is used */
        if (IS_ENABLED(CONFIG_COMMON_CLK) &&
           of_find_property(dev->of_node, "#clock-cells", NULL))
                devm_of_clk_add_hw_provider(dev, of_clk_hw_simple_get, NULL);

which helps readability as well as compile coverage. Doesn't matter here either, since there are only two lines inside.

Arnd