I would like to be Cc'ed on the matters. On Fri, Nov 29, 2019 at 9:13 PM Miquel Raynal <miquel.raynal@xxxxxxxxxxx> wrote: > > The MAX7313 chip is fully compatible with the PCA9535 on its basic > functions but can also manage the intensity on each of its ports with > PWM. Each output is independent and may be tuned with 16 values (4 > bits per output). The period is always 32kHz, only the duty-cycle may > be changed. One can use any output as GPIO or PWM. Thanks for an update! Still I think it's wrong approach. What should be done is: - adding a pin configuration type of PWM (when, for example, argument defines duty cycle and period) - conversion to pin control of this driver - enabling pin configuration PWM for it. For now it looks like a custom way of doing it. If GPIO maintainers are okay with it, I'll not object, just want to have than something like TODO updated for the matter. Taking above into consideration, I also provide my comments to the patch ... > #include <linux/bits.h> > +#include <linux/bitmap.h> It seems you need to take gpio/fixes branch as a base. ... > #define PCA_INT BIT(8) > #define PCA_PCAL BIT(9) > +#define MAX_PWM BIT(10) Use same prefix. ... > +#define PWM_MAX_COUNT 16 > +#define PWM_PER_REG 2 > +#define PWM_BITS_PER_REG (8 / PWM_PER_REG) Can we simple put 4 here? ... > +#define PWM_INTENSITY_MASK GENMASK(PWM_BITS_PER_REG - 1, 0) Please use plain numbers for the GENMASK() arguments. ... > +struct max7313_pwm_data { > + struct gpio_desc *desc; > +}; Are you plan to extend this? Can we directly use struct gpio_desc pointer? ... > + if (PCA_CHIP_TYPE(chip->driver_data) == PCA953X_TYPE && > + chip->driver_data & MAX_PWM) { Can't we simple check only for a flag for now? > + if (reg >= MAX7313_MASTER && > + reg < (MAX7313_INTENSITY + bank_sz)) > + return true; > + } ... > + if (PCA_CHIP_TYPE(chip->driver_data) == PCA953X_TYPE && > + chip->driver_data & MAX_PWM) { > + if (reg >= MAX7313_MASTER && > + reg < (MAX7313_INTENSITY + bank_sz)) > + return true; > + } This is a duplicate from above. Need a helper? ... > +/* > + * Max7313 PWM specific methods > + * > + * Limitations: > + * - Does not support a disabled state > + * - Period fixed to 31.25ms > + * - Only supports normal polarity > + * - Some glitches cannot be prevented > + */ Can we have below in a separate file and attach it to the gpio-pca953x code iff CONFIG_PWM != n? ... > + mutex_lock(&pca_chip->i2c_lock); > + regmap_read(pca_chip->regmap, reg, &val); No error check? > + mutex_unlock(&pca_chip->i2c_lock); ... > + if (shift) Redundant. > + val >>= shift; ... > + mutex_lock(&pca_chip->i2c_lock); > + regmap_read(pca_chip->regmap, reg, &output); > + mutex_unlock(&pca_chip->i2c_lock); No error check? ... > + mutex_lock(&pca_chip->i2c_lock); > + regmap_read(pca_chip->regmap, reg, &output); > + mutex_unlock(&pca_chip->i2c_lock); No error check? ... > +static int max7313_pwm_request(struct pwm_chip *chip, > + struct pwm_device *pwm) > +{ > + struct max7313_pwm *max_pwm = to_max7313_pwm(chip); > + struct pca953x_chip *pca_chip = to_pca953x(max_pwm); > + struct max7313_pwm_data *data; > + struct gpio_desc *desc; > + > + desc = gpiochip_request_own_desc(&pca_chip->gpio_chip, pwm->hwpwm, > + "max7313-pwm", GPIO_ACTIVE_HIGH, 0); > + if (IS_ERR(desc)) { > + dev_err(&pca_chip->client->dev, Can't we get to struct device easily? If it's possible maybe we could move next line to this one? > + "pin already in use (probably as GPIO): %ld\n", > + PTR_ERR(desc)); > + return PTR_ERR(desc); > + } > + return 0; > +} ... > + if (intensity) > + set_bit(pwm->hwpwm, max_pwm->active_pwm); > + else > + clear_bit(pwm->hwpwm, max_pwm->active_pwm); assign_bit() By the way, do you really need it to be atomic? Perhaps __asign_bit()? ... > + active = bitmap_weight(max_pwm->active_pwm, PWM_MAX_COUNT); > + if (!active) In this case more readable will be active == 0 since you compare this to the exact value. > + ret = max7313_pwm_set_master_intensity(pca_chip, 0); > + else if (active == 1) > + ret = max7313_pwm_set_master_intensity(pca_chip, > + PWM_INTENSITY_MASK); ... > + if (IS_ENABLED(CONFIG_PWM)) { I'm not sure it eliminates all PWM related callbacks. > + ret = max7313_pwm_probe(&client->dev, chip); > + if (ret) { > + dev_err(&client->dev, "pwm probe failed, %d\n", ret); > + return ret; > + } > + } -- With Best Regards, Andy Shevchenko