cpufreq: frequency scaling spec in DT node

[Viresh Kumar]

On 11-07-17, 15:36, Mason wrote:
> On 11/07/2017 11:27, Mason wrote:
> 
> > I'll experiment with the other solution of creating the OPP table
> > at init.
> 
> For the sake of discussion, based on your suggestion to use
> dev_pm_opp_add(), I have tested the patch below.
> 
> There is one remaining error:
> [    2.071139] of: dev_pm_opp_of_cpumask_add_table: couldn't find opp table for cpu:0, -19
> 
> Reached through:
> [    1.878461] [<c0359a0c>] (dev_pm_opp_of_add_table) from [<c0359ff0>] (dev_pm_opp_of_cpumask_add_table+0x40/0xc8)
> [    1.888695] [<c0359ff0>] (dev_pm_opp_of_cpumask_add_table) from [<c03d66d8>] (cpufreq_init+0x23c/0x2dc)
> [    1.898139] [<c03d66d8>] (cpufreq_init) from [<c03d3d94>] (cpufreq_online+0xb8/0x660)
> [    1.906009] [<c03d3d94>] (cpufreq_online) from [<c03d43f8>] (cpufreq_add_dev+0xbc/0xcc)
> [    1.914058] [<c03d43f8>] (cpufreq_add_dev) from [<c0350560>] (subsys_interface_register+0x90/0xcc)
> [    1.923065] [<c0350560>] (subsys_interface_register) from [<c03d32f4>] (cpufreq_register_driver+0x15c/0x1e0)
> [    1.932945] [<c03d32f4>] (cpufreq_register_driver) from [<c03d67f0>] (dt_cpufreq_probe+0x78/0xec)
> [    1.941866] [<c03d67f0>] (dt_cpufreq_probe) from [<c035327c>] (platform_drv_probe+0x34/0x6c)

I have sent a patch to fix that.

> The cpufreq-dt driver seems to recover despite the error:
> 
> /sys/devices/system/cpu/cpu0/cpufreq$ cat scaling_available_frequencies
> 135000 243000 405000 607500 1215000 

Yeah, it worked as expected.

>  arch/arm/boot/dts/tango4-smp8758.dtsi |  1 -
>  drivers/cpufreq/Makefile              |  1 +
>  drivers/cpufreq/cpufreq-dt-platdev.c  |  2 --
>  drivers/cpufreq/tango-cpufreq.c       | 45 +++++++++++++++++++++++++++++++++++
>  4 files changed, 46 insertions(+), 3 deletions(-)
> 
> diff --git a/arch/arm/boot/dts/tango4-smp8758.dtsi b/arch/arm/boot/dts/tango4-smp8758.dtsi
> index d2e65c46bcc7..eca33d568690 100644
> --- a/arch/arm/boot/dts/tango4-smp8758.dtsi
> +++ b/arch/arm/boot/dts/tango4-smp8758.dtsi
> @@ -13,7 +13,6 @@
>  			reg = <0>;
>  			clocks = <&clkgen CPU_CLK>;
>  			clock-latency = <1>;
> -			operating-points = <1215000 0 607500 0 405000 0 243000 0 135000 0>;
>  		};
>  
>  		cpu1: cpu at 1 {
> diff --git a/drivers/cpufreq/Makefile b/drivers/cpufreq/Makefile
> index 0a9b6a093646..5ae156dc0d35 100644
> --- a/drivers/cpufreq/Makefile
> +++ b/drivers/cpufreq/Makefile
> @@ -75,6 +75,7 @@ obj-$(CONFIG_ARM_SA1110_CPUFREQ)	+= sa1110-cpufreq.o
>  obj-$(CONFIG_ARM_SCPI_CPUFREQ)		+= scpi-cpufreq.o
>  obj-$(CONFIG_ARM_SPEAR_CPUFREQ)		+= spear-cpufreq.o
>  obj-$(CONFIG_ARM_STI_CPUFREQ)		+= sti-cpufreq.o
> +obj-$(CONFIG_ARCH_TANGO)		+= tango-cpufreq.o
>  obj-$(CONFIG_ARM_TEGRA20_CPUFREQ)	+= tegra20-cpufreq.o
>  obj-$(CONFIG_ARM_TEGRA124_CPUFREQ)	+= tegra124-cpufreq.o
>  obj-$(CONFIG_ARM_VEXPRESS_SPC_CPUFREQ)	+= vexpress-spc-cpufreq.o
> diff --git a/drivers/cpufreq/cpufreq-dt-platdev.c b/drivers/cpufreq/cpufreq-dt-platdev.c
> index 71267626456b..1490e4ce3fbc 100644
> --- a/drivers/cpufreq/cpufreq-dt-platdev.c
> +++ b/drivers/cpufreq/cpufreq-dt-platdev.c
> @@ -70,8 +70,6 @@
>  	{ .compatible = "rockchip,rk3368", },
>  	{ .compatible = "rockchip,rk3399", },
>  
> -	{ .compatible = "sigma,tango4" },
> -
>  	{ .compatible = "ti,am33xx", },
>  	{ .compatible = "ti,dra7", },
>  	{ .compatible = "ti,omap2", },
> diff --git a/drivers/cpufreq/tango-cpufreq.c b/drivers/cpufreq/tango-cpufreq.c
> new file mode 100644
> index 000000000000..270c8261ebfa
> --- /dev/null
> +++ b/drivers/cpufreq/tango-cpufreq.c
> @@ -0,0 +1,45 @@
> +#include <linux/of.h>
> +#include <linux/cpu.h>
> +#include <linux/clk.h>
> +#include <linux/pm_opp.h>
> +#include <linux/platform_device.h>
> +
> +static const struct of_device_id machines[] __initconst = {
> +	{ .compatible = "sigma,tango4" },
> +	{ .compatible = "sigma,tango5" },
> +	{ /* sentinel */ }
> +};
> +
> +static void __init build_opp_table(struct device *cpu_dev, struct clk *cpu_clk)
> +{
> +	unsigned long max_freq = clk_get_rate(cpu_clk);

I kind of get the idea on how is the bootloader fixing your max. So you take the
frequency at which the CPU booted to be Max, right ?

I think that is part of the problem here. IMHO, the kernel really isn't required
to do that. It can have a MAX value of its own. Not sure if we can get it via DT
somehow today, otherwise statically should be fine as well.

And if kernel can take care of max and not depend on the value set by
bootloader, you wouldn't be required to have this patch and use DT based OPP.

> +	dev_pm_opp_add(cpu_dev, max_freq / 1, 0);
> +	dev_pm_opp_add(cpu_dev, max_freq / 2, 0);
> +	dev_pm_opp_add(cpu_dev, max_freq / 3, 0);
> +	dev_pm_opp_add(cpu_dev, max_freq / 5, 0);
> +	dev_pm_opp_add(cpu_dev, max_freq / 9, 0);
> +}

-- 
viresh