Re: [PATCH v2 2/2] input: misc: Add Spreadtrum vibrator driver

[Dmitry Torokhov <dmitry.torokhov@gmail.com>]

Hi Xiaotong,

On Mon, Apr 23, 2018 at 10:33:36AM +0800, Baolin Wang wrote:
> From: Xiaotong Lu <xiaotong.lu@spreadtrum.com>
> 
> This patch adds the Spreadtrum vibrator driver, which embedded in the
> Spreadtrum SC27xx series PMICs.
> 
> Signed-off-by: Xiaotong Lu <xiaotong.lu@spreadtrum.com>
> Signed-off-by: Baolin Wang <baolin.wang@linaro.org>
> ---
> Changes since v1:
>  - Remove input_ff_destroy() and input_unregister_device()
> ---
>  drivers/input/misc/Kconfig        |   10 +++
>  drivers/input/misc/Makefile       |    1 +
>  drivers/input/misc/sc27xx-vibra.c |  156 +++++++++++++++++++++++++++++++++++++
>  3 files changed, 167 insertions(+)
>  create mode 100644 drivers/input/misc/sc27xx-vibra.c
> 
> diff --git a/drivers/input/misc/Kconfig b/drivers/input/misc/Kconfig
> index 572b15f..c761c0c 100644
> --- a/drivers/input/misc/Kconfig
> +++ b/drivers/input/misc/Kconfig
> @@ -841,4 +841,14 @@ config INPUT_RAVE_SP_PWRBUTTON
>  	  To compile this driver as a module, choose M here: the
>  	  module will be called rave-sp-pwrbutton.
>  
> +config INPUT_SC27XX_VIBRA
> +	tristate "Spreadtrum sc27xx vibrator support"
> +	depends on MFD_SC27XX_PMIC || COMPILE_TEST
> +	select INPUT_FF_MEMLESS
> +	help
> +	  This option enables support for Spreadtrum sc27xx vibrator driver.
> +
> +	  To compile this driver as a module, choose M here. The module will
> +	  be called sc27xx_vibra.
> +
>  endif
> diff --git a/drivers/input/misc/Makefile b/drivers/input/misc/Makefile
> index 72cde28..9d0f9d1 100644
> --- a/drivers/input/misc/Makefile
> +++ b/drivers/input/misc/Makefile
> @@ -66,6 +66,7 @@ obj-$(CONFIG_INPUT_RETU_PWRBUTTON)	+= retu-pwrbutton.o
>  obj-$(CONFIG_INPUT_AXP20X_PEK)		+= axp20x-pek.o
>  obj-$(CONFIG_INPUT_GPIO_ROTARY_ENCODER)	+= rotary_encoder.o
>  obj-$(CONFIG_INPUT_RK805_PWRKEY)	+= rk805-pwrkey.o
> +obj-$(CONFIG_INPUT_SC27XX_VIBRA)	+= sc27xx-vibra.o
>  obj-$(CONFIG_INPUT_SGI_BTNS)		+= sgi_btns.o
>  obj-$(CONFIG_INPUT_SIRFSOC_ONKEY)	+= sirfsoc-onkey.o
>  obj-$(CONFIG_INPUT_SOC_BUTTON_ARRAY)	+= soc_button_array.o
> diff --git a/drivers/input/misc/sc27xx-vibra.c b/drivers/input/misc/sc27xx-vibra.c
> new file mode 100644
> index 0000000..f78e70f
> --- /dev/null
> +++ b/drivers/input/misc/sc27xx-vibra.c
> @@ -0,0 +1,156 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/*
> + * Copyright (C) 2018 Spreadtrum Communications Inc.
> + */
> +
> +#include <linux/module.h>
> +#include <linux/of_address.h>
> +#include <linux/platform_device.h>
> +#include <linux/regmap.h>
> +#include <linux/input.h>
> +#include <linux/workqueue.h>
> +
> +#define CUR_DRV_CAL_SEL		GENMASK(13, 12)
> +#define SLP_LDOVIBR_PD_EN	BIT(9)
> +#define LDO_VIBR_PD		BIT(8)
> +
> +struct vibra_info {
> +	struct input_dev	*input_dev;
> +	struct work_struct	play_work;
> +	struct regmap		*regmap;
> +	u32			base;
> +	u32			strength;
> +	bool			enabled;
> +};
> +
> +static void sc27xx_vibra_set(struct vibra_info *info, bool on)
> +{
> +	if (on) {
> +		regmap_update_bits(info->regmap, info->base, LDO_VIBR_PD, 0);
> +		regmap_update_bits(info->regmap, info->base,
> +				   SLP_LDOVIBR_PD_EN, 0);
> +		info->enabled = true;
> +	} else {
> +		regmap_update_bits(info->regmap, info->base, LDO_VIBR_PD,
> +				   LDO_VIBR_PD);
> +		regmap_update_bits(info->regmap, info->base,
> +				   SLP_LDOVIBR_PD_EN, SLP_LDOVIBR_PD_EN);
> +		info->enabled = false;
> +	}
> +}
> +
> +static int sc27xx_vibra_hw_init(struct vibra_info *info)
> +{
> +	return regmap_update_bits(info->regmap, info->base, CUR_DRV_CAL_SEL, 0);
> +}
> +
> +static void sc27xx_vibra_play_work(struct work_struct *work)
> +{
> +	struct vibra_info *info = container_of(work, struct vibra_info,
> +					       play_work);
> +
> +	if (info->strength && !info->enabled)
> +		sc27xx_vibra_set(info, true);
> +	else if (info->enabled)
> +		sc27xx_vibra_set(info, false);

I do not think this is correct. If you issue 2 play requests with
info->strength that is not 0 then you'll end up disabling the vibrator.

I think you want the 2nd condition to be:

	else if (info->strength == 0 && info->enabled)
		sc27xx_vibra_set(info, false);


> +}
> +
> +static int sc27xx_vibra_play(struct input_dev *input, void *data,
> +			     struct ff_effect *effect)
> +{
> +	struct vibra_info *info = input_get_drvdata(input);
> +
> +	info->strength = effect->u.rumble.weak_magnitude;
> +	schedule_work(&info->play_work);
> +
> +	return 0;
> +}
> +
> +static void sc27xx_vibra_close(struct input_dev *input)
> +{
> +	struct vibra_info *info = input_get_drvdata(input);
> +
> +	cancel_work_sync(&info->play_work);
> +	if (info->enabled)
> +		sc27xx_vibra_set(info, false);
> +}
> +
> +static int sc27xx_vibra_probe(struct platform_device *pdev)
> +{
> +	struct device_node *node = pdev->dev.of_node;
> +	struct vibra_info *info;
> +	int ret;

Can we please call this variable "error"?

> +
> +	info = devm_kzalloc(&pdev->dev, sizeof(*info), GFP_KERNEL);
> +	if (!info)
> +		return -ENOMEM;
> +
> +	info->regmap = dev_get_regmap(pdev->dev.parent, NULL);
> +	if (!info->regmap) {
> +		dev_err(&pdev->dev, "failed to get vibrator regmap.\n");
> +		return -ENODEV;
> +	}
> +
> +	ret = of_property_read_u32(node, "reg", &info->base);

I am fan of generic device properties, please change to
device_property_read_u32().

> +	if (ret) {
> +		dev_err(&pdev->dev, "failed to get vibrator base address.\n");
> +		return ret;
> +	}
> +
> +	info->input_dev = devm_input_allocate_device(&pdev->dev);
> +	if (!info->input_dev) {
> +		dev_err(&pdev->dev, "failed to allocate input device.\n");
> +		return -ENOMEM;
> +	}
> +
> +	info->input_dev->name = "sc27xx:vibrator";
> +	info->input_dev->id.version = 0;
> +	info->input_dev->dev.parent = pdev->dev.parent;

Why is device reparented to the parent of platform device? This breaks
devm implementation for the input device. If you really need this,
you'll have to switch to unmanaged API.

> +	info->input_dev->close = sc27xx_vibra_close;
> +
> +	input_set_drvdata(info->input_dev, info);
> +	input_set_capability(info->input_dev, EV_FF, FF_RUMBLE);
> +	INIT_WORK(&info->play_work, sc27xx_vibra_play_work);
> +	info->enabled = false;
> +
> +	ret = input_ff_create_memless(info->input_dev, NULL, sc27xx_vibra_play);
> +	if (ret) {
> +		dev_err(&pdev->dev, "failed to register vibrator to FF.\n");
> +		return ret;
> +	}
> +
> +	ret = input_register_device(info->input_dev);
> +	if (ret) {
> +		dev_err(&pdev->dev, "failed to register input device.\n");
> +		return ret;
> +	}
> +
> +	ret = sc27xx_vibra_hw_init(info);
> +	if (ret) {
> +		dev_err(&pdev->dev, "failed to initialize the vibrator.\n");
> +		return ret;
> +	}

It is too late to initialize hardware after registering the input
device, as once registered it should be fully functional. I'd recommend
calling this before calling input_register_device(), or (maybe even
better) doing this as part of open() method.

> +
> +	platform_set_drvdata(pdev, info);

I do not think you are using platform device's driver data anywhere.

> +	return 0;
> +}
> +
> +static const struct of_device_id sc27xx_vibra_of_match[] = {
> +	{ .compatible = "sprd,sc27xx-vibrator", },
> +	{}
> +};
> +MODULE_DEVICE_TABLE(of, sc27xx_vibra_of_match);
> +
> +static struct platform_driver sc27xx_vibra_driver = {
> +	.driver = {
> +		.name = "sc27xx-vibrator",
> +		.of_match_table = sc27xx_vibra_of_match,

Do you need suspend support by chance? To shut off the vibrator when
system transitions to sleep?

> +	},
> +	.probe = sc27xx_vibra_probe,
> +};
> +
> +module_platform_driver(sc27xx_vibra_driver);
> +
> +MODULE_DESCRIPTION("Spreadtrum SC27xx Vibrator Driver");
> +MODULE_LICENSE("GPL v2");
> +MODULE_AUTHOR("Xiaotong Lu <xiaotong.lu@spreadtrum.com>");
> -- 
> 1.7.9.5
> 

Thanks.

-- 
Dmitry