[v12,1/2] cpufreq: mediatek-hw: Add support for CPUFREQ HW

From: "Hector.Yuan" <hector.yuan@mediatek.com>

Add cpufreq HW support.

Signed-off-by: Hector.Yuan <hector.yuan@mediatek.com>
---
 drivers/cpufreq/Kconfig.arm           |   12 ++
 drivers/cpufreq/Makefile              |    1 +
 drivers/cpufreq/mediatek-cpufreq-hw.c |  370 +++++++++++++++++++++++++++++++++
 3 files changed, 383 insertions(+)
 create mode 100644 drivers/cpufreq/mediatek-cpufreq-hw.c

On 30-05-21, 00:52, Hector Yuan wrote:
> From: "Hector.Yuan" <hector.yuan@mediatek.com>
> 
> Add cpufreq HW support.
> 
> Signed-off-by: Hector.Yuan <hector.yuan@mediatek.com>
> ---
>  drivers/cpufreq/Kconfig.arm           |   12 ++
>  drivers/cpufreq/Makefile              |    1 +
>  drivers/cpufreq/mediatek-cpufreq-hw.c |  370 +++++++++++++++++++++++++++++++++
>  3 files changed, 383 insertions(+)
>  create mode 100644 drivers/cpufreq/mediatek-cpufreq-hw.c
> 
> diff --git a/drivers/cpufreq/Kconfig.arm b/drivers/cpufreq/Kconfig.arm
> index a5c5f70..954749a 100644
> --- a/drivers/cpufreq/Kconfig.arm
> +++ b/drivers/cpufreq/Kconfig.arm
> @@ -133,6 +133,18 @@ config ARM_MEDIATEK_CPUFREQ
>  	help
>  	  This adds the CPUFreq driver support for MediaTek SoCs.
>  
> +config ARM_MEDIATEK_CPUFREQ_HW
> +	tristate "MediaTek CPUFreq HW driver"
> +	depends on ARCH_MEDIATEK || COMPILE_TEST
> +	default m
> +	help
> +	  Support for the CPUFreq HW driver.
> +	  Some MediaTek chipsets have a HW engine to offload the steps
> +	  necessary for changing the frequency of the CPUs. Firmware loaded
> +	  in this engine exposes a programming interface to the OS.
> +	  The driver implements the cpufreq interface for this HW engine.
> +	  Say Y if you want to support CPUFreq HW.
> +
>  config ARM_OMAP2PLUS_CPUFREQ
>  	bool "TI OMAP2+"
>  	depends on ARCH_OMAP2PLUS
> diff --git a/drivers/cpufreq/Makefile b/drivers/cpufreq/Makefile
> index 27d3bd7..48ee585 100644
> --- a/drivers/cpufreq/Makefile
> +++ b/drivers/cpufreq/Makefile
> @@ -56,6 +56,7 @@ obj-$(CONFIG_ARM_IMX6Q_CPUFREQ)		+= imx6q-cpufreq.o
>  obj-$(CONFIG_ARM_IMX_CPUFREQ_DT)	+= imx-cpufreq-dt.o
>  obj-$(CONFIG_ARM_KIRKWOOD_CPUFREQ)	+= kirkwood-cpufreq.o
>  obj-$(CONFIG_ARM_MEDIATEK_CPUFREQ)	+= mediatek-cpufreq.o
> +obj-$(CONFIG_ARM_MEDIATEK_CPUFREQ_HW)	+= mediatek-cpufreq-hw.o
>  obj-$(CONFIG_MACH_MVEBU_V7)		+= mvebu-cpufreq.o
>  obj-$(CONFIG_ARM_OMAP2PLUS_CPUFREQ)	+= omap-cpufreq.o
>  obj-$(CONFIG_ARM_PXA2xx_CPUFREQ)	+= pxa2xx-cpufreq.o
> diff --git a/drivers/cpufreq/mediatek-cpufreq-hw.c b/drivers/cpufreq/mediatek-cpufreq-hw.c
> new file mode 100644
> index 0000000..6f3a461
> --- /dev/null
> +++ b/drivers/cpufreq/mediatek-cpufreq-hw.c
> @@ -0,0 +1,370 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/*
> + * Copyright (c) 2020 MediaTek Inc.
> + */
> +
> +#include <linux/bitfield.h>
> +#include <linux/cpufreq.h>
> +#include <linux/energy_model.h>
> +#include <linux/init.h>
> +#include <linux/iopoll.h>
> +#include <linux/kernel.h>
> +#include <linux/module.h>
> +#include <linux/of_address.h>
> +#include <linux/of_platform.h>
> +#include <linux/pm_qos.h>
> +#include <linux/slab.h>
> +
> +#define LUT_MAX_ENTRIES			32U
> +#define LUT_FREQ			GENMASK(11, 0)
> +#define LUT_ROW_SIZE			0x4
> +#define CPUFREQ_HW_STATUS		BIT(0)
> +#define SVS_HW_STATUS			BIT(1)
> +#define POLL_USEC			1000
> +#define TIMEOUT_USEC			300000
> +
> +enum {
> +	REG_FREQ_LUT_TABLE,
> +	REG_FREQ_ENABLE,
> +	REG_FREQ_PERF_STATE,
> +	REG_FREQ_HW_STATE,
> +	REG_EM_POWER_TBL,
> +	REG_FREQ_LATENCY,
> +
> +	REG_ARRAY_SIZE,
> +};
> +
> +struct cpufreq_mtk {
> +	struct cpufreq_frequency_table *table;
> +	void __iomem *reg_bases[REG_ARRAY_SIZE];
> +	int nr_opp;
> +	cpumask_t related_cpus;
> +};
> +
> +static const u16 cpufreq_mtk_offsets[REG_ARRAY_SIZE] = {
> +	[REG_FREQ_LUT_TABLE]	= 0x0,
> +	[REG_FREQ_ENABLE]	= 0x84,
> +	[REG_FREQ_PERF_STATE]	= 0x88,
> +	[REG_FREQ_HW_STATE]	= 0x8c,
> +	[REG_EM_POWER_TBL]	= 0x90,
> +	[REG_FREQ_LATENCY]	= 0x110,
> +};
> +
> +static struct cpufreq_mtk *mtk_freq_domain_map[NR_CPUS];
> +
> +static int __maybe_unused
> +mtk_cpufreq_get_cpu_power(unsigned long *mW,
> +			  unsigned long *KHz, struct device *cpu_dev)
> +{
> +	struct cpufreq_mtk *c = mtk_freq_domain_map[cpu_dev->id];
> +	int i;
> +
> +	for (i = 0; i < c->nr_opp; i++) {
> +		if (c->table[i].frequency < *KHz)
> +			break;
> +	}
> +	i--;
> +
> +	*KHz = c->table[i].frequency;
> +	*mW = readl_relaxed(c->reg_bases[REG_EM_POWER_TBL] +
> +			    i * LUT_ROW_SIZE) / 1000;
> +
> +	return 0;
> +}
> +
> +static int mtk_cpufreq_hw_target_index(struct cpufreq_policy *policy,
> +				       unsigned int index)
> +{
> +	struct cpufreq_mtk *c = policy->driver_data;
> +
> +	writel_relaxed(index, c->reg_bases[REG_FREQ_PERF_STATE]);
> +
> +	return 0;
> +}
> +
> +static unsigned int mtk_cpufreq_hw_get(unsigned int cpu)
> +{
> +	struct cpufreq_mtk *c;
> +	unsigned int index;
> +
> +	c = mtk_freq_domain_map[cpu];
> +
> +	index = readl_relaxed(c->reg_bases[REG_FREQ_PERF_STATE]);
> +	index = min(index, LUT_MAX_ENTRIES - 1);
> +
> +	return c->table[index].frequency;
> +}
> +
> +static unsigned int mtk_cpufreq_hw_fast_switch(struct cpufreq_policy *policy,
> +					       unsigned int target_freq)
> +{
> +	struct cpufreq_mtk *c = policy->driver_data;
> +	unsigned int index;
> +
> +	index = cpufreq_table_find_index_dl(policy, target_freq);
> +
> +	writel_relaxed(index, c->reg_bases[REG_FREQ_PERF_STATE]);
> +
> +	return policy->freq_table[index].frequency;
> +}
> +
> +static int mtk_cpufreq_hw_cpu_init(struct cpufreq_policy *policy)
> +{
> +	struct cpufreq_mtk *c;
> +	struct device *cpu_dev;
> +	struct em_data_callback em_cb = EM_DATA_CB(mtk_cpufreq_get_cpu_power);
> +	struct pm_qos_request *qos_request;
> +	int sig, pwr_hw = CPUFREQ_HW_STATUS | SVS_HW_STATUS;
> +	unsigned int latency;
> +
> +	qos_request = kzalloc(sizeof(*qos_request), GFP_KERNEL);
> +	if (!qos_request)
> +		return -ENOMEM;

Since this is temporarily required, please allocate it while you are
trying to use it.

> +
> +	cpu_dev = get_cpu_device(policy->cpu);
> +	if (!cpu_dev) {
> +		pr_err("failed to get cpu%d device\n", policy->cpu);
> +		return -ENODEV;
> +	}
> +
> +	c = mtk_freq_domain_map[policy->cpu];
> +	if (!c) {
> +		pr_err("No scaling support for CPU%d\n", policy->cpu);
> +		return -ENODEV;
> +	}
> +
> +	cpumask_copy(policy->cpus, &c->related_cpus);
> +
> +	policy->freq_table = c->table;
> +	policy->driver_data = c;
> +
> +	latency = readl_relaxed(c->reg_bases[REG_FREQ_LATENCY]);
> +	if (!latency)
> +		latency = CPUFREQ_ETERNAL;
> +
> +	/* us convert to ns */
> +	policy->cpuinfo.transition_latency = latency * 1000;
> +
> +	policy->fast_switch_possible = true;
> +
> +	/* Let CPUs leave idle-off state for SVS CPU initializing */
> +	cpu_latency_qos_add_request(qos_request, 0);

I am not sure I understand what's going on here.

> +	/* HW should be in enabled state to proceed now */
> +	writel_relaxed(0x1, c->reg_bases[REG_FREQ_ENABLE]);
> +
> +	if (readl_poll_timeout(c->reg_bases[REG_FREQ_HW_STATE], sig,
> +			       (sig & pwr_hw) == pwr_hw, POLL_USEC,
> +			       TIMEOUT_USEC)) {
> +		if (!(sig & CPUFREQ_HW_STATUS)) {
> +			pr_info("cpufreq hardware of CPU%d is not enabled\n",
> +				policy->cpu);
> +			return -ENODEV;
> +		}
> +
> +		pr_info("SVS of CPU%d is not enabled\n", policy->cpu);
> +	}
> +
> +	em_dev_register_perf_domain(cpu_dev, c->nr_opp, &em_cb, policy->cpus, true);
> +
> +	cpu_latency_qos_remove_request(qos_request);

Can this be done before em_dev_register_perf_domain() ?

> +	kfree(qos_request);
> +
> +	return 0;
> +}
> +
> +static int mtk_cpufreq_hw_cpu_exit(struct cpufreq_policy *policy)
> +{
> +	struct cpufreq_mtk *c;
> +
> +	c = mtk_freq_domain_map[policy->cpu];
> +	if (!c) {

This can't be NULL here.

> +		pr_err("No scaling support for CPU%d\n", policy->cpu);
> +		return -ENODEV;
> +	}
> +
> +	/* HW should be in paused state now */
> +	writel_relaxed(0x0, c->reg_bases[REG_FREQ_ENABLE]);
> +
> +	return 0;
> +}
> +
> +static struct cpufreq_driver cpufreq_mtk_hw_driver = {
> +	.flags		= CPUFREQ_NEED_INITIAL_FREQ_CHECK |
> +			  CPUFREQ_HAVE_GOVERNOR_PER_POLICY |
> +			  CPUFREQ_IS_COOLING_DEV,
> +	.verify		= cpufreq_generic_frequency_table_verify,
> +	.target_index	= mtk_cpufreq_hw_target_index,
> +	.get		= mtk_cpufreq_hw_get,
> +	.init		= mtk_cpufreq_hw_cpu_init,
> +	.exit		= mtk_cpufreq_hw_cpu_exit,
> +	.fast_switch	= mtk_cpufreq_hw_fast_switch,
> +	.name		= "mtk-cpufreq-hw",
> +	.attr		= cpufreq_generic_attr,
> +};
> +
> +static int mtk_cpu_create_freq_table(struct platform_device *pdev,
> +				     struct cpufreq_mtk *c)
> +{
> +	struct device *dev = &pdev->dev;
> +	void __iomem *base_table;
> +	u32 data, i, freq, prev_freq = 0;
> +
> +	c->table = devm_kcalloc(dev, LUT_MAX_ENTRIES + 1,
> +				sizeof(*c->table), GFP_KERNEL);
> +	if (!c->table)
> +		return -ENOMEM;
> +
> +	base_table = c->reg_bases[REG_FREQ_LUT_TABLE];
> +
> +	for (i = 0; i < LUT_MAX_ENTRIES; i++) {
> +		data = readl_relaxed(base_table + (i * LUT_ROW_SIZE));
> +		freq = FIELD_GET(LUT_FREQ, data) * 1000;
> +
> +		if (freq == prev_freq)
> +			break;
> +
> +		c->table[i].frequency = freq;
> +
> +		dev_dbg(dev, "index=%d freq=%d\n",
> +			i, c->table[i].frequency);
> +
> +		prev_freq = freq;
> +	}
> +
> +	c->table[i].frequency = CPUFREQ_TABLE_END;
> +	c->nr_opp = i;
> +
> +	return 0;
> +}
> +
> +static int mtk_get_related_cpus(int index, struct cpufreq_mtk *c)
> +{
> +	struct device_node *cpu_np;
> +	struct of_phandle_args args;
> +	int cpu, ret;
> +
> +	for_each_possible_cpu(cpu) {
> +		cpu_np = of_cpu_device_node_get(cpu);
> +		if (!cpu_np)
> +			continue;
> +
> +		ret = of_parse_phandle_with_args(cpu_np, "performance-domains",
> +						 "#performance-domain-cells", 0,
> +						 &args);
> +		of_node_put(cpu_np);
> +		if (ret < 0)
> +			continue;
> +
> +		if (index == args.args[0]) {
> +			cpumask_set_cpu(cpu, &c->related_cpus);
> +			mtk_freq_domain_map[cpu] = c;
> +		}
> +	}
> +
> +	return 0;
> +}

I really hope this can be moved to a common place as more than one
drier should be required to parse this thing.

> +static int mtk_cpu_resources_init(struct platform_device *pdev,
> +				  unsigned int cpu, int index,
> +				  const u16 *offsets)
> +{
> +	struct cpufreq_mtk *c;
> +	struct device *dev = &pdev->dev;
> +	int ret, i;
> +	void __iomem *base;
> +
> +	if (mtk_freq_domain_map[cpu])
> +		return 0;
> +
> +	c = devm_kzalloc(dev, sizeof(*c), GFP_KERNEL);
> +	if (!c)
> +		return -ENOMEM;
> +
> +	base = devm_platform_ioremap_resource(pdev, index);
> +	if (IS_ERR(base))
> +		return PTR_ERR(base);
> +
> +	for (i = REG_FREQ_LUT_TABLE; i < REG_ARRAY_SIZE; i++)
> +		c->reg_bases[i] = base + offsets[i];
> +
> +	ret = mtk_get_related_cpus(index, c);
> +	if (ret) {
> +		dev_err(dev, "Domain-%d failed to get related CPUs\n", index);
> +		return ret;
> +	}
> +
> +	ret = mtk_cpu_create_freq_table(pdev, c);
> +	if (ret) {
> +		dev_err(dev, "Domain-%d failed to create freq table\n", index);
> +		return ret;
> +	}
> +
> +	return 0;
> +}
> +
> +static int mtk_cpufreq_hw_driver_probe(struct platform_device *pdev)
> +{
> +	struct device_node *cpu_np;
> +	struct of_phandle_args args;
> +	const u16 *offsets;
> +	unsigned int cpu;
> +	int ret;
> +
> +	offsets = of_device_get_match_data(&pdev->dev);
> +	if (!offsets)
> +		return -EINVAL;
> +
> +	for_each_possible_cpu(cpu) {
> +		cpu_np = of_cpu_device_node_get(cpu);
> +		if (!cpu_np) {
> +			dev_err(&pdev->dev, "Failed to get cpu %d device\n",
> +				cpu);
> +			return -ENODEV;
> +		}
> +
> +		ret = of_parse_phandle_with_args(cpu_np, "performance-domains",
> +						 "#performance-domain-cells", 0,
> +						 &args);
> +		if (ret < 0)
> +			return ret;
> +
> +		/* Get the bases of cpufreq for domains */
> +		ret = mtk_cpu_resources_init(pdev, cpu, args.args[0], offsets);
> +		if (ret) {
> +			dev_err(&pdev->dev, "CPUFreq resource init failed\n");
> +			return ret;
> +		}
> +	}

Why don't you do above stuff in driver's ->init() ? As this really
isn't required to be done per CPU, but per policy.

> +
> +	ret = cpufreq_register_driver(&cpufreq_mtk_hw_driver);
> +	if (ret) {
> +		dev_err(&pdev->dev, "CPUFreq HW driver failed to register\n");
> +		return ret;
> +	}
> +
> +	return 0;
> +}
> +
> +static int mtk_cpufreq_hw_driver_remove(struct platform_device *pdev)
> +{
> +	return cpufreq_unregister_driver(&cpufreq_mtk_hw_driver);
> +}
> +
> +static const struct of_device_id mtk_cpufreq_hw_match[] = {
> +	{ .compatible = "mediatek,cpufreq-hw", .data = &cpufreq_mtk_offsets },
> +	{}
> +};
> +
> +static struct platform_driver mtk_cpufreq_hw_driver = {
> +	.probe = mtk_cpufreq_hw_driver_probe,
> +	.remove = mtk_cpufreq_hw_driver_remove,
> +	.driver = {
> +		.name = "mtk-cpufreq-hw",
> +		.of_match_table = mtk_cpufreq_hw_match,
> +	},
> +};
> +module_platform_driver(mtk_cpufreq_hw_driver);
> +
> +MODULE_DESCRIPTION("Mediatek cpufreq-hw driver");
> +MODULE_LICENSE("GPL v2");
> -- 
> 1.7.9.5

On 27-06-21, 16:17, Hector Yuan wrote:
> On Mon, 2021-06-14 at 16:10 +0530, Viresh Kumar wrote:

> > On 30-05-21, 00:52, Hector Yuan wrote:

> > > +static int mtk_get_related_cpus(int index, struct cpufreq_mtk *c)

> > > +{

> > > +	struct device_node *cpu_np;

> > > +	struct of_phandle_args args;

> > > +	int cpu, ret;

> > > +

> > > +	for_each_possible_cpu(cpu) {

> > > +		cpu_np = of_cpu_device_node_get(cpu);

> > > +		if (!cpu_np)

> > > +			continue;

> > > +

> > > +		ret = of_parse_phandle_with_args(cpu_np, "performance-domains",

> > > +						 "#performance-domain-cells", 0,

> > > +						 &args);

> > > +		of_node_put(cpu_np);

> > > +		if (ret < 0)

> > > +			continue;

> > > +

> > > +		if (index == args.args[0]) {

> > > +			cpumask_set_cpu(cpu, &c->related_cpus);

> > > +			mtk_freq_domain_map[cpu] = c;

> > > +		}

> > > +	}

> > > +

> > > +	return 0;

> > > +}

> > 

> > I really hope this can be moved to a common place as more than one

> > drier should be required to parse this thing.

> > 

> 

> Yes, this can be a common part for all performance domain users. But may

> I know whats your suggestion? Put this API in another file or? Thanks


Rob, Sudeep: You guys have a suggestion on where can we keep a routine for this
?

-- 
viresh

On 28-06-21, 10:09, Sudeep Holla wrote:
> Probably in driver/cpufreq or some related headers if it needs to access
> related_cpus and is more cpufreq related in that way ?

It just needs to set a mask, so doesn't really depend on cpufreq. I
was wondering if drivers/opp/of.c may be used for this, and I am not
sure.

> Orthogonal to that, I prefer to make the generic function take list_name
> and cells_name as generic. I see we can reuse that qcom-cpufreq-hw.c
> with "qcom,freq-domain" and "#freq-domain-cells".

Yes.

On Tue, Jun 29, 2021 at 08:17:19AM +0530, Viresh Kumar wrote:
> On 28-06-21, 10:09, Sudeep Holla wrote:
> > Probably in driver/cpufreq or some related headers if it needs to access
> > related_cpus and is more cpufreq related in that way ?
>
> It just needs to set a mask, so doesn't really depend on cpufreq. I
> was wondering if drivers/opp/of.c may be used for this, and I am not
> sure.
>

Sounds good to me.

--
Regards,
Sudeep

On 29-06-21, 10:02, Sudeep Holla wrote:
> On Tue, Jun 29, 2021 at 08:17:19AM +0530, Viresh Kumar wrote:

> > On 28-06-21, 10:09, Sudeep Holla wrote:

> > > Probably in driver/cpufreq or some related headers if it needs to access

> > > related_cpus and is more cpufreq related in that way ?

> >

> > It just needs to set a mask, so doesn't really depend on cpufreq. I

> > was wondering if drivers/opp/of.c may be used for this, and I am not

> > sure.

> >

> 

> Sounds good to me.


Okay.

Hector, add the routine to of.c in OPP core.

-- 
viresh

On 05-07-21, 01:01, Hector Yuan wrote:
> Hi, Viresh:
> 
> OK, Thank you.
> 
> I'll add this routine in of.c.
> May I know the below prototype is okay to you?
> 
> static int _of_get_related_cpus(const char *list_name, const char
> *cell_name, int index, struct cpumask *m)

of_perf_domain_get_sharing_cpumask(int cpu, const char *list_name,
        const char *cell_name, struct cpumask *cpumask);

And actually add this in cpufreq.h, this doesn't have anything to do with OPP
and I can assume that a platform not enabling the OPP functionality to end up
using it later.

Rafael, will that be fine ? Or we add it in a completely new header ? The users
are going to be cpufreq drivers only I believe.

> And maybe I have to find somewhere to update my
> mtk_freq_domain_map[cpu]?
> 
> static int mtk_get_related_cpus(int index, struct cpufreq_mtk *c)
> {
> 	for_each_possible_cpu(cpu) {
> 		cpumask_set_cpu(cpu, &c->related_cpus);//related_cpu can copy policy
> cpus
> 		mtk_freq_domain_map[cpu] = c;//need to find somewhere to assign
> 
> 	}
> }

I am not sure I understood this part well.