On Fri, Oct 19, 2018 at 8:24 PM Dan O'Donovan <dan@xxxxxxxxxx> wrote: > > From: Javier Arteaga <javier@xxxxxxxxxx> > > The UP2 board features a Raspberry Pi compatible pin header (HAT) and a > board-specific expansion connector (EXHAT). Both expose assorted > functions from either the SoC (such as GPIO, I2C, SPI, UART...) or other > on-board devices (ADC, FPGA IP blocks...). > > These lines are routed through an on-board FPGA. The platform controller > in its stock firmware provides register fields to change: > > - Line enable (FPGA pins enabled / high impedance) > - Line direction (SoC driven / FPGA driven) > > To enable using SoC GPIOs on the pin header, this arrangement requires > both configuring the platform controller, and updating the SoC pad > registers in sync. > > Add a frontend pinctrl/GPIO driver that registers a new set of GPIO > lines for the header pins. When these are requested, the driver > propagates this request to the backend SoC pinctrl/GPIO driver by > grabbing a GPIO descriptor for the matching SoC GPIO line. The needed > mapping for this is retrieved via ACPI properties. To Linus: please, don't consider this as anyhow part of Intel pin control infrastructure. Thus, if you are okay with the driver (personally I don't see any major issues with the code, though it might be required some clarification on design level, e.g. ACPI relationship) I have no objection. > +#define UPBOARD_BIT_TO_PIN(r, bit) \ > + ((r) * UPBOARD_REGISTER_SIZE + (bit)) One line? > +static int upboard_get_functions_count(struct pinctrl_dev *pctldev) > +{ > + return 0; > +} > + > +static int upboard_get_function_groups(struct pinctrl_dev *pctldev, > + unsigned int selector, > + const char * const **groups, > + unsigned int *num_groups) > +{ > + *groups = NULL; > + *num_groups = 0; > + return 0; > +} > + > +static const char *upboard_get_function_name(struct pinctrl_dev *pctldev, > + unsigned int selector) > +{ > + return NULL; > +} > + > +static int upboard_set_mux(struct pinctrl_dev *pctldev, unsigned int function, > + unsigned int group) > +{ > + return 0; > +} Hmm... Do you need those stubs? Same Q for other stubs in the file. > +static int upboard_gpio_request_enable(struct pinctrl_dev *pctldev, > + struct pinctrl_gpio_range *range, > + unsigned int pin) > +{ > + const struct pin_desc * const pd = pin_desc_get(pctldev, pin); > + const struct upboard_pin *p; > + int ret; > + > + if (!pd) > + return -EINVAL; When it possible to happen? Same Q for the rest same excerpts. > + p = pd->drv_data; > + > + /* if this pin has an associated function bit, disable it first */ > + if (p->func_en) { > + ret = regmap_field_write(p->func_en, 0); > + if (ret) > + return ret; > + } > + > + if (p->gpio_en) { > + ret = regmap_field_write(p->gpio_en, 1); > + if (ret) > + return ret; > + } > + > + return 0; > +} > +static struct gpio_desc *upboard_offset_to_soc_gpio(struct gpio_chip *gc, > + unsigned int offset) > +{ > + struct upboard_pinctrl *pctrl = > + container_of(gc, struct upboard_pinctrl, chip); One line? > + > + if (offset + 1 > pctrl->nsoc_gpios || !pctrl->soc_gpios[offset]) > + return ERR_PTR(-ENODEV); offset >= ? Is it even possible? > + > + return pctrl->soc_gpios[offset]; > +} > + > +static int upboard_gpio_request(struct gpio_chip *gc, unsigned int offset) > +{ > + struct upboard_pinctrl *pctrl = > + container_of(gc, struct upboard_pinctrl, chip); One line? > +} > + > +static void upboard_gpio_free(struct gpio_chip *gc, unsigned int offset) > +{ > + struct upboard_pinctrl *pctrl = > + container_of(gc, struct upboard_pinctrl, chip); Ditto. > + if (offset + 1 > pctrl->nsoc_gpios || !pctrl->soc_gpios[offset]) > + return; offset >= ? Is it even possible? > +} > +static int upboard_pinctrl_probe(struct platform_device *pdev) > +{ > + struct device *dev = &pdev->dev; > + struct pinctrl_desc *pctldesc; > + struct upboard_pinctrl *pctrl; > + struct upboard_pin *pins; > + struct acpi_device *adev; > + struct regmap *regmap; > + unsigned int i; > + int ret; > + adev = ACPI_COMPANION(dev); > + if (!adev || strcmp(acpi_device_hid(adev), "AANT0F01")) > + return -ENODEV; Same Q as per LED driver. > + for (i = 0; i < pctldesc->npins; i++) { > + struct upboard_pin *pin = &pins[i]; > + const struct pinctrl_pin_desc *pd = &pctldesc->pins[i]; > + pin->func_en = NULL; Useless. > + if (pd->drv_data) { > + struct reg_field *field = pd->drv_data; > + > + pin->func_en = devm_regmap_field_alloc(dev, regmap, > + *field); > + if (IS_ERR(pin->func_en)) > + return PTR_ERR(pin->func_en); > + } > + > + pin->gpio_en = upboard_field_alloc(dev, regmap, > + UPBOARD_REG_GPIO_EN0, i); > + if (IS_ERR(pin->gpio_en)) > + return PTR_ERR(pin->gpio_en); > + > + pin->gpio_dir = upboard_field_alloc(dev, regmap, > + UPBOARD_REG_GPIO_DIR0, i); > + if (IS_ERR(pin->gpio_dir)) > + return PTR_ERR(pin->gpio_dir); > + > + ((struct pinctrl_pin_desc *)pd)->drv_data = pin; I'm not sure I understand the purpose of this casting. > + } > + pctrl->soc_gpios = devm_kzalloc(dev, > + pctrl->nsoc_gpios * sizeof(*pctrl->soc_gpios), > + GFP_KERNEL); > + if (!pctrl->soc_gpios) > + return -ENOMEM; kzalloc -> kcalloc > +} -- With Best Regards, Andy Shevchenko