Re: [PATCH 2/6] soc: ti: omap-prm: Add basic power domain support

[Tero Kristo <t-kristo@ti.com>]

On 12/05/2020 23:38, Tony Lindgren wrote:
> The PRM controller has currently only support for resets while the power
> domains are still handled in the platform code.
> 
> Let's add basic power domain support to enable and disable a PRM
> controlled power domain if configured in the devicetree. This can be
> used for various hardware accelerators, and interconnect instances.
> 
> Further support can be added later on as needed for runtime configuration
> based on domain-idle-states.
> 
> Signed-off-by: Tony Lindgren <tony@atomide.com>
> ---
>   arch/arm/mach-omap2/Kconfig |   1 +
>   drivers/soc/ti/omap_prm.c   | 281 +++++++++++++++++++++++++++++++++++-
>   2 files changed, 281 insertions(+), 1 deletion(-)
> 
> diff --git a/arch/arm/mach-omap2/Kconfig b/arch/arm/mach-omap2/Kconfig
> --- a/arch/arm/mach-omap2/Kconfig
> +++ b/arch/arm/mach-omap2/Kconfig
> @@ -7,6 +7,7 @@ config ARCH_OMAP2
>   	depends on ARCH_MULTI_V6
>   	select ARCH_OMAP2PLUS
>   	select CPU_V6
> +	select PM_GENERIC_DOMAINS if PM
>   	select SOC_HAS_OMAP2_SDRC
>   
>   config ARCH_OMAP3
> diff --git a/drivers/soc/ti/omap_prm.c b/drivers/soc/ti/omap_prm.c
> --- a/drivers/soc/ti/omap_prm.c
> +++ b/drivers/soc/ti/omap_prm.c
> @@ -6,18 +6,50 @@
>    *	Tero Kristo <t-kristo@ti.com>
>    */
>   
> +#include <linux/clk.h>
>   #include <linux/kernel.h>
>   #include <linux/device.h>
>   #include <linux/io.h>
>   #include <linux/iopoll.h>
> +#include <linux/module.h>
>   #include <linux/of.h>
>   #include <linux/of_device.h>
>   #include <linux/platform_device.h>
> +#include <linux/pm_domain.h>
>   #include <linux/reset-controller.h>
>   #include <linux/delay.h>
>   
>   #include <linux/platform_data/ti-prm.h>
>   
> +enum omap_prm_clocks {
> +	OMAP_PRM_FCK,
> +	OMAP_PRM_ICK,
> +	OMAP_PRM_NR_CLOCKS,
> +};
> +
> +enum omap_prm_domain_mode {
> +	OMAP_PRMD_OFF,
> +	OMAP_PRMD_RETENTION,
> +	OMAP_PRMD_ON_INACTIVE,
> +	OMAP_PRMD_ON_ACTIVE,
> +};
> +
> +struct omap_prm_domain_map {
> +	unsigned int usable_modes;	/* Mask of hardware supported modes */
> +	unsigned long statechange:1;	/* Optional low-power state change */
> +	unsigned long logicretstate:1;	/* Optional logic off mode */
> +};
> +
> +struct omap_prm_domain {
> +	struct device *dev;
> +	struct omap_prm *prm;
> +	struct generic_pm_domain pd;
> +	void __iomem *pwrstctrl;
> +	void __iomem *pwrstst;

I think the pwrstst is not really used as of now, it is just part of 
couple of dev_dbg prints.

> +	const struct omap_prm_domain_map *cap;
> +	u32 pwrstctrl_saved;
> +};
> +
>   struct omap_rst_map {
>   	s8 rst;
>   	s8 st;
> @@ -27,6 +59,9 @@ struct omap_prm_data {
>   	u32 base;
>   	const char *name;
>   	const char *clkdm_name;
> +	u16 pwrstctrl;
> +	u16 pwrstst;
> +	const struct omap_prm_domain_map *dmap;
>   	u16 rstctrl;
>   	u16 rstst;
>   	const struct omap_rst_map *rstmap;
> @@ -36,6 +71,8 @@ struct omap_prm_data {
>   struct omap_prm {
>   	const struct omap_prm_data *data;
>   	void __iomem *base;
> +	struct clk *clocks[OMAP_PRM_NR_CLOCKS];
> +	struct omap_prm_domain *prmd;
>   };
>   
>   struct omap_reset_data {
> @@ -47,6 +84,7 @@ struct omap_reset_data {
>   	struct device *dev;
>   };
>   
> +#define genpd_to_prm_domain(gpd) container_of(gpd, struct omap_prm_domain, pd)
>   #define to_omap_reset_data(p) container_of((p), struct omap_reset_data, rcdev)
>   
>   #define OMAP_MAX_RESETS		8
> @@ -58,6 +96,40 @@ struct omap_reset_data {
>   
>   #define OMAP_PRM_HAS_RESETS	(OMAP_PRM_HAS_RSTCTRL | OMAP_PRM_HAS_RSTST)
>   
> +#define PRM_LOGICRETSTATE	BIT(2)
> +#define PRM_LOWPOWERSTATECHANGE	BIT(4)
> +#define PRM_POWERSTATE_MASK	OMAP_PRMD_ON_ACTIVE
> +
> +static const struct __maybe_unused
> +omap_prm_domain_map omap_prm_all = {
> +	.usable_modes = BIT(OMAP_PRMD_ON_ACTIVE) | BIT(OMAP_PRMD_ON_INACTIVE) |
> +			BIT(OMAP_PRMD_RETENTION) | BIT(OMAP_PRMD_OFF),
> +	.statechange = 1,
> +	.logicretstate = 1,
> +};
> +
> +static const struct __maybe_unused
> +omap_prm_domain_map omap_prm_noinact = {
> +	.usable_modes = BIT(OMAP_PRMD_ON_ACTIVE) | BIT(OMAP_PRMD_RETENTION) |
> +			BIT(OMAP_PRMD_OFF),
> +	.statechange = 1,
> +	.logicretstate = 1,
> +};
> +
> +static const struct __maybe_unused
> +omap_prm_domain_map omap_prm_nooff = {
> +	.usable_modes = BIT(OMAP_PRMD_ON_ACTIVE) | BIT(OMAP_PRMD_ON_INACTIVE) |
> +			BIT(OMAP_PRMD_RETENTION),
> +	.statechange = 1,
> +	.logicretstate = 1,
> +};
> +
> +static const struct __maybe_unused
> +omap_prm_domain_map omap_prm_onoff_noauto = {
> +	.usable_modes = BIT(OMAP_PRMD_ON_ACTIVE) | BIT(OMAP_PRMD_OFF),
> +	.statechange = 1,
> +};
> +
>   static const struct omap_rst_map rst_map_0[] = {
>   	{ .rst = 0, .st = 0 },
>   	{ .rst = -1 },
> @@ -151,6 +223,152 @@ static const struct of_device_id omap_prm_id_table[] = {
>   	{ },
>   };
>   
> +static int omap_prm_domain_power_on(struct generic_pm_domain *domain)
> +{
> +	struct omap_prm_domain *prmd;
> +	u32 v;
> +
> +	prmd = genpd_to_prm_domain(domain);
> +	if (!prmd->cap)
> +		return 0;
> +
> +	dev_dbg(prmd->dev, "%s: %s: old state: pwrstctrl: %08x pwrstst: %08x\n",
> +		__func__, prmd->pd.name, readl_relaxed(prmd->pwrstctrl),
> +		readl_relaxed(prmd->pwrstst));
> +
> +	if (prmd->pwrstctrl_saved)
> +		v = prmd->pwrstctrl_saved;
> +	else
> +		v = readl_relaxed(prmd->pwrstctrl);
> +
> +	writel_relaxed(v | OMAP_PRMD_ON_ACTIVE, prmd->pwrstctrl);
> +	dev_dbg(prmd->dev, "%s: %s: new state pwrstctrl: %08x\n",
> +		__func__, prmd->pd.name, readl_relaxed(prmd->pwrstctrl));

Should we wait for the transition to complete here?

> +
> +	return 0;
> +}
> +
> +/* No need to check for holes in the mask for the lowest mode */
> +static int omap_prm_domain_find_lowest(struct omap_prm_domain *prmd)
> +{
> +	return __ffs(prmd->cap->usable_modes);
> +}
> +
> +static int omap_prm_domain_power_off(struct generic_pm_domain *domain)
> +{
> +	struct omap_prm_domain *prmd;
> +	u32 v;
> +
> +	prmd = genpd_to_prm_domain(domain);
> +	if (!prmd->cap)
> +		return 0;
> +
> +	v = readl_relaxed(prmd->pwrstctrl);
> +	prmd->pwrstctrl_saved = v;
> +
> +	dev_dbg(prmd->dev, "%s: %s: old state: pwrstctrl: %08x pwrstst: %08x\n",
> +		__func__, prmd->pd.name, v, readl_relaxed(prmd->pwrstst));
> +
> +	v &= ~PRM_POWERSTATE_MASK;
> +	v |= omap_prm_domain_find_lowest(prmd);
> +
> +	if (prmd->cap->statechange)
> +		v |= PRM_LOWPOWERSTATECHANGE;
> +	if (prmd->cap->logicretstate)
> +		v &= ~PRM_LOGICRETSTATE;
> +	else
> +		v |= PRM_LOGICRETSTATE;
> +
> +	writel_relaxed(v, prmd->pwrstctrl);

Should we wait for the transition to complete here?

> +
> +	dev_dbg(prmd->dev, "%s: %s new state pwrstctrl: %08x\n",
> +		__func__, prmd->pd.name, readl_relaxed(prmd->pwrstctrl));
> +
> +	return 0;
> +}
> +
> +static int omap_prm_domain_attach_dev(struct generic_pm_domain *domain,
> +				      struct device *dev)
> +{
> +	struct generic_pm_domain_data *genpd_data;
> +	struct of_phandle_args pd_args;
> +	struct omap_prm_domain *prmd;
> +	struct device_node *np;
> +	int ret;
> +
> +	prmd = genpd_to_prm_domain(domain);
> +	np = dev->of_node;
> +
> +	ret = of_parse_phandle_with_args(np, "power-domains",
> +					 "#power-domain-cells", 0, &pd_args);
> +	if (ret < 0)
> +		return ret;
> +
> +	if (pd_args.args_count != 0)
> +		dev_warn(dev, "%s: unusupported #power-domain-cells: %i\n",
> +			 prmd->pd.name, pd_args.args_count);
> +
> +	genpd_data = dev_gpd_data(dev);
> +	genpd_data->data = NULL;
> +
> +	return 0;
> +}
> +
> +static void omap_prm_domain_detach_dev(struct generic_pm_domain *domain,
> +				       struct device *dev)
> +{
> +	struct generic_pm_domain_data *genpd_data;
> +	struct omap_prm_domain *prmd;
> +
> +	prmd = genpd_to_prm_domain(domain);
> +
> +	genpd_data = dev_gpd_data(dev);
> +	genpd_data->data = NULL;
> +}
> +
> +static int omap_prm_domain_init(struct device *dev, struct omap_prm *prm)
> +{
> +	struct omap_prm_domain *prmd;
> +	struct device_node *np = dev->of_node;
> +	const struct omap_prm_data *data;
> +	const char *name;
> +	int error;
> +
> +	if (!of_find_property(dev->of_node, "#power-domain-cells", NULL))
> +		return 0;
> +
> +	of_node_put(dev->of_node);
> +
> +	prmd = devm_kzalloc(dev, sizeof(*prmd), GFP_KERNEL);
> +	if (!prmd)
> +		return -ENOMEM;
> +
> +	data = prm->data;
> +	name = devm_kasprintf(dev, GFP_KERNEL, "prm_%s",
> +			      data->name);
> +
> +	prmd->dev = dev;
> +	prmd->prm = prm;
> +	prmd->cap = prmd->prm->data->dmap;
> +	prmd->pwrstctrl = prmd->prm->base + prmd->prm->data->pwrstctrl;
> +	prmd->pwrstst = prmd->prm->base + prmd->prm->data->pwrstst;
> +
> +	prmd->pd.name = name;
> +	prmd->pd.power_on = omap_prm_domain_power_on;
> +	prmd->pd.power_off = omap_prm_domain_power_off;
> +	prmd->pd.attach_dev = omap_prm_domain_attach_dev;
> +	prmd->pd.detach_dev = omap_prm_domain_detach_dev;
> +
> +	pm_genpd_init(&prmd->pd, NULL, true);
> +	error = of_genpd_add_provider_simple(np, &prmd->pd);
> +	if (error)
> +		pm_genpd_remove(&prmd->pd);
> +	else
> +		prm->prmd = prmd;
> +
> +	return error;
> +}
> +
>   static bool _is_valid_reset(struct omap_reset_data *reset, unsigned long id)
>   {
>   	if (reset->mask & BIT(id))
> @@ -345,12 +563,48 @@ static int omap_prm_reset_init(struct platform_device *pdev,
>   	return devm_reset_controller_register(&pdev->dev, &reset->rcdev);
>   }
>   

Is any of the following clock handling needed, and if yes, whats its 
purpose?

It looks like this is only used for ABE clkctrl handling on omap4/omap5 
(at least for now), but afaik, ABE clkctrl is read only so this code 
would effectively do nothing (and potentially just even fail.)

-Tero

> +static int omap_prm_init_clock(struct device *dev, struct omap_prm *prm,
> +			       const char *name, enum omap_prm_clocks index)
> +{
> +	struct clk *clock;
> +	int error;
> +
> +	clock = devm_clk_get(dev, name);
> +	if (IS_ERR(clock)) {
> +		if (PTR_ERR(clock) == -ENOENT)
> +			error = 0;
> +		else
> +			error = PTR_ERR(prm->clocks[index]);
> +
> +		goto out_done;
> +	}
> +
> +	error = clk_prepare_enable(clock);
> +
> +out_done:
> +	prm->clocks[index] = clock;
> +
> +	return error;
> +}
> +
> +static void omap_prm_disable_clocks(struct omap_prm *prm)
> +{
> +	int i;
> +
> +	for (i = 0; i < OMAP_PRM_NR_CLOCKS; i++) {
> +		if (IS_ERR(prm->clocks[i]))
> +			continue;
> +		clk_disable_unprepare(prm->clocks[i]);
> +	}
> +}
> +
>   static int omap_prm_probe(struct platform_device *pdev)
>   {
>   	struct resource *res;
>   	const struct omap_prm_data *data;
>   	struct omap_prm *prm;
>   	const struct of_device_id *match;
> +	int ret;
>   
>   	res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
>   	if (!res)
> @@ -378,7 +632,32 @@ static int omap_prm_probe(struct platform_device *pdev)
>   	if (IS_ERR(prm->base))
>   		return PTR_ERR(prm->base);
>   
> -	return omap_prm_reset_init(pdev, prm);
> +	ret = omap_prm_init_clock(&pdev->dev, prm, "fck", OMAP_PRM_FCK);
> +	if (ret)
> +		goto err_disable;
> +
> +	ret = omap_prm_init_clock(&pdev->dev, prm, "ick", OMAP_PRM_ICK);
> +	if (ret)
> +		goto err_disable;
> +
> +	ret = omap_prm_domain_init(&pdev->dev, prm);
> +	if (ret)
> +		goto err_disable;
> +
> +	ret = omap_prm_reset_init(pdev, prm);
> +	if (ret)
> +		goto err_domain;
> +
> +	return 0;
> +
> +err_domain:
> +	of_genpd_del_provider(pdev->dev.of_node);
> +	pm_genpd_remove(&prm->prmd->pd);
> +
> +err_disable:
> +	omap_prm_disable_clocks(prm);
> +
> +	return ret;
>   }
>   
>   static struct platform_driver omap_prm_driver = {
> 

--
Texas Instruments Finland Oy, Porkkalankatu 22, 00180 Helsinki. Y-tunnus/Business ID: 0615521-4. Kotipaikka/Domicile: Helsinki