Re: [PATCH v2 1/4] Input: Add driver for Cypress Generation 5 touchscreen

[Andy Shevchenko <andy.shevchenko@gmail.com>]

On Wed, Nov 3, 2021 at 1:49 PM Alistair Francis <alistair@alistair23.me> wrote:

...

> Message-Id: <20180703094309.18514-2-mylene.josserand@bootlin.com>

What is this?

...

> +config TOUCHSCREEN_CYTTSP5
> +       tristate "Cypress TrueTouch Gen5 Touchscreen Driver"

> +       depends on OF && I2C

Is it really OF dependent?

> +       select REGMAP_I2C
> +       select CRC_ITU_T
> +       help
> +         Driver for Parade TrueTouch Standard Product
> +         Generation 5 touchscreen controllers.
> +         I2C bus interface support only.
> +         Say Y here if you have a Cypress Gen5 touchscreen.
> +         If unsure, say N.
> +         To compile this driver as a module, choose M here: the
> +         module will be called cyttsp5.

Utilize lines better, i.e. increase density of the words on them.

...

> +/*
> + * Parade TrueTouch(TM) Standard Product V5 Module.
> + *
> + * Copyright (C) 2015 Parade Technologies
> + * Copyright (C) 2012-2015 Cypress Semiconductor
> + * Copyright (C) 2018 Bootlin
> + *
> + * Authors: Mylène Josserand <mylene.josserand@bootlin.com>
> + *                Alistair Francis <alistair@alistair23.me>

> + *

Redundant.

> + */

...

> +#include <asm/unaligned.h>

Can you move it after linux/* ones?

> +#include <linux/crc-itu-t.h>
> +#include <linux/delay.h>
> +#include <linux/device.h>

> +#include <linux/gpio.h>

This is wrong. New code shouldn't include this header.

> +#include <linux/input/mt.h>
> +#include <linux/input/touchscreen.h>
> +#include <linux/interrupt.h>
> +#include <linux/i2c.h>
> +#include <linux/module.h>
> +#include <linux/of_device.h>
> +#include <linux/regmap.h>

...

> +#define CY_BITS_PER_BTN                        1
> +#define CY_NUM_BTN_EVENT_ID                    ((1 << CY_BITS_PER_BTN) - 1)

Seems like a mask depending on the amount of bits.
Perhaps GENMASK(1, 0)?

...

> +#define HID_OUTPUT_RESPONSE_CMD_MASK           0x7F

GENMASK()

...

> +#define HID_SYSINFO_BTN_MASK                   0xFF

Ditto.

...

> +/*  Timeout in ms */

Drop this comment and use _MS suffix in the definitions.

> +#define CY_HID_OUTPUT_TIMEOUT                  200
> +#define CY_HID_OUTPUT_GET_SYSINFO_TIMEOUT      3000
> +#define CY_HID_GET_HID_DESCRIPTOR_TIMEOUT      4000

...

> +/* helpers */
> +#define HI_BYTE(x)                             ((u8)(((x) >> 8) & 0xFF))
> +#define LOW_BYTE(x)                            ((u8)((x) & 0xFF))

Why are these needed? Just use them directly.

...

> +enum cyttsp5_tch_abs { /* for ordering within the extracted touch data array */
> +       CY_TCH_X,       /* X */
> +       CY_TCH_Y,       /* Y */
> +       CY_TCH_P,       /* P (Z) */
> +       CY_TCH_T,       /* TOUCH ID */
> +       CY_TCH_MAJ,     /* TOUCH_MAJOR */
> +       CY_TCH_MIN,     /* TOUCH_MINOR */

> +       CY_TCH_NUM_ABS,

If it is a terminator, drop the comma.

> +};

...

> +/*
> + * For what understood in the datasheet, the register does not

is understood

> + * matter. For consistency, used the Input Register address

use

> + * but it does mean anything to the device. The important data
> + * to send is the I2C address
> + */


> +       /* The hardware wants to receive a frame with the address register

Be consistent with multi-line comments. Here you need to add a leading
empty line.

> +        * contains in the first two bytes. As the regmap_write function

contained

> +        * add the register adresse in the frame, we use the LOW_BYTE as
> +        * first frame byte for the address register and the first
> +        * data byte is the high register + left of the cmd to send
> +        */

...

> +       for (nbyte = 0, *axis = 0; nbyte < size; nbyte++)

> +               *axis = *axis + ((xy_data[nbyte] >> bofs) << (nbyte * 8));

+=

...

> +#ifdef CONFIG_OF

Why? You may easily combine the two.

> +static int cyttsp5_parse_dt_key_code(struct device *dev)
> +{
> +       struct cyttsp5 *ts = dev_get_drvdata(dev);
> +       struct cyttsp5_sysinfo *si = &ts->sysinfo;
> +       struct device_node *np;
> +       int i;

> +       np = dev->of_node;
> +       if (!np)
> +               return -EINVAL;

Redundant. The of_property_*() will do it anyway.

> +       if (!si->num_btns)
> +               return 0;
> +
> +       /* Initialize the button to RESERVED */

> +       for (i = 0; i < si->num_btns; i++)
> +               si->key_code[i] = KEY_RESERVED;

memset32() ?

> +       return of_property_read_u32_array(np, "linux,keycodes",
> +                                  si->key_code, si->num_btns);
> +}
> +#else
> +static int cyttsp5_parse_dt_key_code(struct device *dev)
> +{
> +       struct cyttsp5 *ts = dev_get_drvdata(dev);
> +       struct cyttsp5_sysinfo *si = &ts->sysinfo;
> +       int i;
> +
> +       if (!si->num_btns)
> +               return 0;
> +
> +       /* Initialize the button to RESERVED */
> +       for (i = 0; i < si->num_btns; i++)
> +               si->key_code[i] = KEY_RESERVED;
> +
> +       return 0;
> +}

Combine with the above and get rid of ugly ifdeffery.

> +#endif

...

> +       size = get_unaligned_le16(&ts->response_buf[0]);

> +

Redundant blank line.

> +       if (!size)
> +               return 0;

...

> +               crc = cyttsp5_compute_crc(&ts->response_buf[4], size - 7);
> +               if (ts->response_buf[size - 3] != LOW_BYTE(crc) ||
> +                   ts->response_buf[size - 2] != HI_BYTE(crc)) {

What's wrong with a direct comparison with _le16 value?

> +                       dev_err(ts->dev, "HID output response, wrong CRC 0x%X\n",
> +                               crc);
> +                       return -EPROTO;
> +               }

...

> +       si->num_btns = 0;
> +       for (i = 0; i < HID_SYSINFO_MAX_BTN; i++) {
> +               if (btns & BIT(i))
> +                       si->num_btns++;
> +       }

NIH for_each_set_bit(), but do one more step to see that this is
actually a Hamming weight. We have functions. Hence, make it one line
with one simple call.

...

> +       struct cyttsp5_sensing_conf_data_dev *scd_dev =

> +               (struct cyttsp5_sensing_conf_data_dev *)

Why casting?

> +               &ts->response_buf[HID_SYSINFO_SENSING_OFFSET];

...

> +       si->xy_data = devm_kzalloc(ts->dev, scd->max_tch * TOUCH_REPORT_SIZE,
> +                                  GFP_KERNEL);

Use _kcalloc().

> +       if (!si->xy_data)
> +               return -ENOMEM;

...

> +       cmd[0] = LOW_BYTE(HID_OUTPUT_GET_SYSINFO_SIZE);
> +       cmd[1] = HI_BYTE(HID_OUTPUT_GET_SYSINFO_SIZE);

Use put_unaligned_le16() instead.

...

> +       rc = wait_event_timeout(ts->wait_q, (ts->hid_cmd_state == HID_CMD_DONE),
> +                               msecs_to_jiffies(CY_HID_OUTPUT_GET_SYSINFO_TIMEOUT));
> +       if (!rc) {
> +               dev_err(ts->dev, "HID output cmd execution timed out\n");

> +               rc = -ETIME;

What is this supposed to mean?

> +               goto error;
> +       }

...

> +       cmd[0] = LOW_BYTE(HID_OUTPUT_BL_LAUNCH_APP_SIZE);
> +       cmd[1] = HI_BYTE(HID_OUTPUT_BL_LAUNCH_APP_SIZE);

put_unaligned_le16()

...

> +       cmd[6] = 0x0; /* Low bytes of data */
> +       cmd[7] = 0x0; /* Hi bytes of data */

Ditto.

...

> +       crc = cyttsp5_compute_crc(&cmd[4], 4);
> +       cmd[8] = LOW_BYTE(crc);
> +       cmd[9] = HI_BYTE(crc);

Ditto.

...

> +       rc = wait_event_timeout(ts->wait_q, (ts->hid_cmd_state == HID_CMD_DONE),
> +                               msecs_to_jiffies(CY_HID_OUTPUT_TIMEOUT));
> +       if (!rc) {
> +               dev_err(ts->dev, "HID output cmd execution timed out\n");

> +               rc = -ETIME;

???

> +               goto error;
> +       }

...

> +       memcpy(desc, ts->response_buf, sizeof(struct cyttsp5_hid_desc));

sizeof(*desc) ?

...

> +       rc = regmap_bulk_read(ts->regmap, HID_INPUT_REG, buf, 2);

sizeof(buf)

> +       if (rc < 0)
> +               return rc;

...

> +#ifdef CONFIG_OF

Drop this. It usually brings more pain than helps.

> +static const struct of_device_id cyttsp5_of_match[] = {
> +       { .compatible = "cypress,tt21000", },
> +       { }
> +};
> +MODULE_DEVICE_TABLE(of, cyttsp5_of_match);
> +#endif
> +
> +static const struct i2c_device_id cyttsp5_i2c_id[] = {
> +       { CYTTSP5_NAME, 0, },
> +       { }
> +};
> +MODULE_DEVICE_TABLE(i2c, cyttsp5_i2c_id);

Move both closer to their users (below).

...

> +               dev_set_drvdata(dev, NULL);

No need to repeat what device core does for everybody for the last 10+ years.

...

> +               dev_set_drvdata(dev, NULL);

Ditto.

...

> +       /* Reset the gpio to be in a reset state */
> +       ts->reset_gpio = devm_gpiod_get_optional(dev, "reset", GPIOD_OUT_LOW);
> +       if (IS_ERR(ts->reset_gpio)) {
> +               rc = PTR_ERR(ts->reset_gpio);
> +               dev_err(dev, "Failed to request reset gpio, error %d\n", rc);
> +               return rc;
> +       }

No minimum reset timeout?

> +       gpiod_set_value(ts->reset_gpio, 1);

Or what does this all mean? Shouldn't you simply use _OUT_HIGH above?

> +       /* Need a delay to have device up */
> +       msleep(20);

...

> +static struct i2c_driver cyttsp5_i2c_driver = {
> +       .driver = {
> +               .name = CYTTSP5_NAME,

> +               .of_match_table = of_match_ptr(cyttsp5_of_match),

Drop of_match_ptr() as well.

> +       },
> +       .probe = cyttsp5_i2c_probe,
> +       .remove = cyttsp5_i2c_remove,
> +       .id_table = cyttsp5_i2c_id,
> +};

> +

Redundant blank line.

> +module_i2c_driver(cyttsp5_i2c_driver);

--
With Best Regards,
Andy Shevchenko