lists.openwall.net   lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  linux-cve-announce  PHC 
Open Source and information security mailing list archives
 
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <aa5aeebe-6a0e-9c55-dc4b-6f8f02308373@molgen.mpg.de>
Date:   Tue, 20 Dec 2022 13:50:55 +0100
From:   Paul Menzel <pmenzel@...gen.mpg.de>
To:     Jim Liu <jim.t90615@...il.com>, Jim Liu <jjliu0@...oton.com>
Cc:     Joseph Liu <kwliu@...oton.com>, linus.walleij@...aro.org,
        brgl@...ev.pl, robh+dt@...nel.org,
        krzysztof.kozlowski+dt@...aro.org, linux-gpio@...r.kernel.org,
        openbmc@...ts.ozlabs.org, linux-kernel@...r.kernel.org,
        devicetree@...r.kernel.org
Subject: Re: [PATCH v3 1/3] gpio: nuvoton: Add Nuvoton NPCM7xx sgpio driver

Dear Jim,


Thank you for the patch.

Am 20.12.22 um 09:01 schrieb Jim Liu:
> Add Nuvoton BMC NPCM7xx sgpio driver support.

Please mention the datasheet, and how this is tested.

> Signed-off-by: Jim Liu <JJLIU0@...oton.com>

The email address does not match the address the patch was sent from.

 From the manual page git-send-email(1):

>        --from=<address>
>            Specify the sender of the emails. If not specified on the command
>            line, the value of the sendemail.from configuration option is used.
>            If neither the command-line option nor sendemail.from are set, then
>            the user will be prompted for the value. The default for the prompt
>            will be the value of GIT_AUTHOR_IDENT, or GIT_COMMITTER_IDENT if
>            that is not set, as returned by "git var -l".

(To avoid having to press the shift key, lowercase local part would be 
nice: jjliu0@...oton.com. Depends on the server configuration though, if 
it’s the same address.)

> ---
> Changes for v3:
>     - remove return in bank_reg function
> Changes for v2:
>     - add prefix
>     - write the enum values in all capitals
>     - remove _init in npcm_sgpio_probe
> ---
>   drivers/gpio/Kconfig           |   8 +
>   drivers/gpio/Makefile          |   1 +
>   drivers/gpio/gpio-npcm-sgpio.c | 639 +++++++++++++++++++++++++++++++++
>   3 files changed, 648 insertions(+)
>   create mode 100644 drivers/gpio/gpio-npcm-sgpio.c
> 
> diff --git a/drivers/gpio/Kconfig b/drivers/gpio/Kconfig
> index ec7cfd4f52b1..c23fff243461 100644
> --- a/drivers/gpio/Kconfig
> +++ b/drivers/gpio/Kconfig
> @@ -470,6 +470,14 @@ config GPIO_MXS
>   	select GPIO_GENERIC
>   	select GENERIC_IRQ_CHIP
>   
> +config GPIO_NPCM_SGPIO
> +	bool "Nuvoton SGPIO support"
> +	depends on (ARCH_NPCM || COMPILE_TEST) && OF_GPIO
> +	select GPIO_GENERIC
> +	select GPIOLIB_IRQCHIP
> +	help
> +	  Say Y here to support Nuvoton NPCM7XX/NPCM8XX SGPIO functionality.
> +
>   config GPIO_OCTEON
>   	tristate "Cavium OCTEON GPIO"
>   	depends on CAVIUM_OCTEON_SOC
> diff --git a/drivers/gpio/Makefile b/drivers/gpio/Makefile
> index 010587025fc8..106146f383f3 100644
> --- a/drivers/gpio/Makefile
> +++ b/drivers/gpio/Makefile
> @@ -111,6 +111,7 @@ obj-$(CONFIG_GPIO_MT7621)		+= gpio-mt7621.o
>   obj-$(CONFIG_GPIO_MVEBU)		+= gpio-mvebu.o
>   obj-$(CONFIG_GPIO_MXC)			+= gpio-mxc.o
>   obj-$(CONFIG_GPIO_MXS)			+= gpio-mxs.o
> +obj-$(CONFIG_GPIO_NPCM_SGPIO)           += gpio-npcm-sgpio.o
>   obj-$(CONFIG_GPIO_OCTEON)		+= gpio-octeon.o
>   obj-$(CONFIG_GPIO_OMAP)			+= gpio-omap.o
>   obj-$(CONFIG_GPIO_PALMAS)		+= gpio-palmas.o
> diff --git a/drivers/gpio/gpio-npcm-sgpio.c b/drivers/gpio/gpio-npcm-sgpio.c
> new file mode 100644
> index 000000000000..9c3896995f8f
> --- /dev/null
> +++ b/drivers/gpio/gpio-npcm-sgpio.c
> @@ -0,0 +1,639 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/*
> + * Nuvoton NPCM Serial GPIO Driver
> + *
> + * Copyright (C) 2021 Nuvoton Technologies
> + */
> +
> +#include <linux/bitfield.h>
> +#include <linux/clk.h>
> +#include <linux/gpio/driver.h>
> +#include <linux/hashtable.h>
> +#include <linux/init.h>
> +#include <linux/io.h>
> +#include <linux/kernel.h>
> +#include <linux/module.h>
> +#include <linux/platform_device.h>
> +#include <linux/spinlock.h>
> +#include <linux/string.h>
> +
> +#define MAX_NR_HW_SGPIO			64
> +
> +#define  NPCM_IOXCFG1				0x2A
> +#define  NPCM_IOXCFG1_SFT_CLK		GENMASK(3, 0)
> +#define  NPCM_IOXCFG1_SCLK_POL		BIT(4)
> +#define  NPCM_IOXCFG1_LDSH_POL		BIT(5)
> +
> +#define  NPCM_IOXCTS 0x28
> +#define  NPCM_IOXCTS_IOXIF_EN BIT(7)
> +#define  NPCM_IOXCTS_RD_MODE GENMASK(2, 1)
> +#define  NPCM_IOXCTS_RD_MODE_PERIODIC BIT(2)
> +#define  NPCM_IOXCTS_RD_MODE_CONTINUOUS GENMASK(2, 1)
> +
> +#define  NPCM_IOXCFG2 0x2B
> +#define  NPCM_IXOEVCFG_MASK 0x3
> +#define  NPCM_IXOEVCFG_BOTH 0x3
> +#define  NPCM_IXOEVCFG_FALLING 0x2
> +#define  NPCM_IXOEVCFG_RISING 0x1
> +
> +#define GPIO_BANK(x)    ((x) / 8)
> +#define GPIO_BIT(x)     ((x) % 8)
> +
> +/*
> + * Slect the freqency of shift clock.

Selectª.
2.  frequency

> + * The shift clock is a division of the APB clock.
> + */
> +struct npcm_clk_cfg {
> +	const int *SFT_CLK;
> +	const u8 *CLK_SEL;
> +	const u8 cfg_opt;
> +};
> +
> +struct npcm_sgpio {
> +	struct gpio_chip chip;
> +	struct clk *pclk;
> +	struct irq_chip intc;
> +	spinlock_t lock; /*protect event config*/
> +	void __iomem *base;
> +	int irq;
> +	u8 nin_sgpio;
> +	u8 nout_sgpio;
> +	u8 in_port;
> +	u8 out_port;
> +	u8 int_type[MAX_NR_HW_SGPIO];
> +};
> +
> +struct npcm_sgpio_bank {
> +	u8 rdata_reg;
> +	u8 wdata_reg;
> +	u8 event_config;
> +	u8 event_status;
> +};
> +
> +enum npcm_sgpio_reg {
> +	READ_DATA,
> +	WRITE_DATA,
> +	EVENT_CFG,
> +	EVENT_STS,
> +};
> +
> +static const struct npcm_sgpio_bank npcm_sgpio_banks[] = {
> +	{
> +		.wdata_reg = 0x00,
> +		.rdata_reg = 0x08,
> +		.event_config = 0x10,
> +		.event_status = 0x20,
> +	},
> +	{
> +		.wdata_reg = 0x01,
> +		.rdata_reg = 0x09,
> +		.event_config = 0x12,
> +		.event_status = 0x21,
> +	},
> +	{
> +		.wdata_reg = 0x02,
> +		.rdata_reg = 0x0a,
> +		.event_config = 0x14,
> +		.event_status = 0x22,
> +	},
> +	{
> +		.wdata_reg = 0x03,
> +		.rdata_reg = 0x0b,
> +		.event_config = 0x16,
> +		.event_status = 0x23,
> +	},
> +	{
> +		.wdata_reg = 0x04,
> +		.rdata_reg = 0x0c,
> +		.event_config = 0x18,
> +		.event_status = 0x24,
> +	},
> +	{
> +		.wdata_reg = 0x05,
> +		.rdata_reg = 0x0d,
> +		.event_config = 0x1a,
> +		.event_status = 0x25,
> +	},
> +	{
> +		.wdata_reg = 0x06,
> +		.rdata_reg = 0x0e,
> +		.event_config = 0x1c,
> +		.event_status = 0x26,
> +	},
> +	{
> +		.wdata_reg = 0x07,
> +		.rdata_reg = 0x0f,
> +		.event_config = 0x1e,
> +		.event_status = 0x27,
> +	},
> +
> +};
> +
> +static void __iomem *bank_reg(struct npcm_sgpio *gpio,
> +			      const struct npcm_sgpio_bank *bank,
> +				const enum npcm_sgpio_reg reg)
> +{
> +	switch (reg) {
> +	case READ_DATA:
> +		return gpio->base + bank->rdata_reg;
> +	case WRITE_DATA:
> +		return gpio->base + bank->wdata_reg;
> +	case EVENT_CFG:
> +		return gpio->base + bank->event_config;
> +	case EVENT_STS:
> +		return gpio->base + bank->event_status;
> +	default:
> +		/* acturally if code runs to here, it's an error case */
> +		WARN(1, "NPCM SGPIO REG SET failed!!\n");
> +	}
> +}
> +
> +static const struct npcm_sgpio_bank *to_bank(unsigned int offset)
> +{
> +	unsigned int bank = GPIO_BANK(offset);
> +
> +	return &npcm_sgpio_banks[bank];
> +}
> +
> +static void irqd_to_npcm_sgpio_data(struct irq_data *d,
> +				    struct npcm_sgpio **gpio,
> +				    const struct npcm_sgpio_bank **bank,
> +				    u8 *bit, int *offset)
> +{
> +	struct npcm_sgpio *internal;
> +
> +	*offset = irqd_to_hwirq(d);
> +	internal = irq_data_get_irq_chip_data(d);
> +	WARN_ON(!internal);
> +
> +	*gpio = internal;
> +	*offset -= internal->nout_sgpio;
> +	*bank = to_bank(*offset);
> +	*bit = GPIO_BIT(*offset);
> +}
> +
> +static int npcm_sgpio_init_valid_mask(struct gpio_chip *gc,
> +				      unsigned long *valid_mask, unsigned int ngpios)
> +{
> +	struct npcm_sgpio *gpio = gpiochip_get_data(gc);
> +	u8 in_port, out_port, set_port;
> +
> +	in_port = gpio->nin_sgpio / 8;
> +	if (gpio->nin_sgpio % 8 > 0)
> +		in_port += 1;
> +
> +	out_port = gpio->nout_sgpio / 8;
> +	if (gpio->nout_sgpio % 8 > 0)
> +		out_port += 1;
> +
> +	gpio->in_port = in_port;
> +	gpio->out_port = out_port;
> +	set_port = ((out_port & 0xf) << 4) | (in_port & 0xf);
> +	iowrite8(set_port, gpio->base + NPCM_IOXCFG2);
> +
> +	return 0;
> +}
> +
> +static int npcm_sgpio_dir_in(struct gpio_chip *gc, unsigned int offset)
> +{
> +	struct npcm_sgpio *gpio = gpiochip_get_data(gc);
> +
> +	if (offset < gpio->nout_sgpio)
> +		return -EINVAL;
> +
> +	return 0;

If you like, you could write:

     return offset < gpio->nout_sgpio ? -EINVAL : 0;

> +}
> +
> +static int npcm_sgpio_dir_out(struct gpio_chip *gc, unsigned int offset, int val)
> +{
> +	struct npcm_sgpio *gpio = gpiochip_get_data(gc);
> +
> +	if (offset < gpio->nout_sgpio) {
> +		gc->set(gc, offset, val);
> +		return 0;
> +	}
> +	return -EINVAL;

Above is a blank line before the `return` statement.

> +}
> +
> +static int npcm_sgpio_get_direction(struct gpio_chip *gc, unsigned int offset)
> +{
> +	struct npcm_sgpio *gpio = gpiochip_get_data(gc);
> +
> +	if (offset < gpio->nout_sgpio)
> +		return 0;
> +	return 1;
> +}
> +
> +static void npcm_sgpio_set(struct gpio_chip *gc, unsigned int offset, int val)
> +{
> +	struct npcm_sgpio *gpio = gpiochip_get_data(gc);
> +	const struct  npcm_sgpio_bank *bank = to_bank(offset);
> +	void __iomem *addr;
> +	u8 reg = 0;
> +
> +	addr = bank_reg(gpio, bank, WRITE_DATA);
> +	reg = ioread8(addr);
> +
> +	if (val) {
> +		reg |= (val << GPIO_BIT(offset));
> +		iowrite8(reg, addr);
> +	} else {
> +		reg &= ~(1 << GPIO_BIT(offset));
> +		iowrite8(reg, addr);
> +	}
> +}
> +
> +static int npcm_sgpio_get(struct gpio_chip *gc, unsigned int offset)
> +{
> +	struct npcm_sgpio *gpio = gpiochip_get_data(gc);
> +	const struct  npcm_sgpio_bank *bank;
> +	void __iomem *addr;
> +	u8 dir, reg;
> +
> +	dir = npcm_sgpio_get_direction(gc, offset);

Is the variable `dir` needed? If yes, please use `int` as that is the 
type of the return value, and is more performant than fixed width type [1].

> +	if (dir == 0) {
> +		bank = to_bank(offset);
> +
> +		addr = bank_reg(gpio, bank, WRITE_DATA);
> +		reg = ioread8(addr);
> +		reg = (reg >> GPIO_BIT(offset)) & 0x01;
> +	} else {
> +		offset -= gpio->nout_sgpio;
> +		bank = to_bank(offset);
> +
> +		addr = bank_reg(gpio, bank, READ_DATA);
> +		reg = ioread8(addr);
> +		reg = (reg >> GPIO_BIT(offset)) & 0x01;
> +	}
> +
> +	return reg;
> +}
> +
> +static void npcm_sgpio_setup_enable(struct npcm_sgpio *gpio, bool enable)
> +{
> +	u8 reg = 0;
> +
> +	reg = ioread8(gpio->base + NPCM_IOXCTS);
> +	reg = reg & ~NPCM_IOXCTS_RD_MODE;
> +	reg = reg | NPCM_IOXCTS_RD_MODE_PERIODIC;
> +
> +	if (enable) {
> +		reg |= NPCM_IOXCTS_IOXIF_EN;
> +		iowrite8(reg, gpio->base + NPCM_IOXCTS);
> +	} else {
> +		reg &= ~NPCM_IOXCTS_IOXIF_EN;
> +		iowrite8(reg, gpio->base + NPCM_IOXCTS);
> +	}
> +}
> +
> +static int npcm_sgpio_setup_clk(struct npcm_sgpio *gpio,
> +				const struct npcm_clk_cfg *clk_cfg, u32 sgpio_freq)
> +{
> +	unsigned long apb_freq;
> +	u32 sgpio_clk_div;
> +	u8 tmp;
> +	int i;
> +
> +	apb_freq = clk_get_rate(gpio->pclk);
> +	sgpio_clk_div = (apb_freq / sgpio_freq);
> +	if ((apb_freq % sgpio_freq) != 0)
> +		sgpio_clk_div += 1;
> +
> +	tmp = ioread8(gpio->base + NPCM_IOXCFG1) & ~NPCM_IOXCFG1_SFT_CLK;
> +
> +	for (i = 0; i < clk_cfg->cfg_opt; i++) {
> +		if (sgpio_clk_div >= clk_cfg->SFT_CLK[i]) {
> +			iowrite8(clk_cfg->CLK_SEL[i] | tmp, gpio->base + NPCM_IOXCFG1);
> +			return 0;
> +		}
> +	}
> +
> +	return -EINVAL;
> +}
> +
> +static void npcm_sgpio_irq_init_valid_mask(struct gpio_chip *gc,
> +					   unsigned long *valid_mask, unsigned int ngpios)
> +{
> +	struct npcm_sgpio *gpio = gpiochip_get_data(gc);
> +	int n = gpio->nin_sgpio;
> +
> +	/* input GPIOs in the high range */
> +	bitmap_set(valid_mask, gpio->nout_sgpio, n);
> +	bitmap_clear(valid_mask, 0, gpio->nout_sgpio);
> +}
> +
> +static void npcm_sgpio_irq_set_mask(struct irq_data *d, bool set)
> +{
> +	const struct npcm_sgpio_bank *bank;
> +	struct npcm_sgpio *gpio;
> +	unsigned long flags;
> +	void __iomem *addr;
> +	int offset;
> +	u16 reg, type;
> +	u8 bit;
> +
> +	irqd_to_npcm_sgpio_data(d, &gpio, &bank, &bit, &offset);
> +	addr = bank_reg(gpio, bank, EVENT_CFG);
> +
> +	spin_lock_irqsave(&gpio->lock, flags);
> +
> +	npcm_sgpio_setup_enable(gpio, false);
> +
> +	reg = ioread16(addr);
> +	if (set) {
> +		reg &= ~(NPCM_IXOEVCFG_MASK << (bit * 2));
> +	} else {
> +		type = gpio->int_type[offset];
> +		reg |= (type << (bit * 2));
> +	}
> +
> +	iowrite16(reg, addr);
> +
> +	npcm_sgpio_setup_enable(gpio, true);
> +
> +	addr = bank_reg(gpio, bank, EVENT_STS);
> +	reg = ioread8(addr);
> +	reg |= BIT(bit);
> +	iowrite8(reg, addr);
> +
> +	spin_unlock_irqrestore(&gpio->lock, flags);
> +}
> +
> +static void npcm_sgpio_irq_ack(struct irq_data *d)
> +{
> +	const struct npcm_sgpio_bank *bank;
> +	struct npcm_sgpio *gpio;
> +	unsigned long flags;
> +	void __iomem *status_addr;
> +	int offset;
> +	u8 bit;
> +
> +	irqd_to_npcm_sgpio_data(d, &gpio, &bank, &bit, &offset);
> +	status_addr = bank_reg(gpio, bank, EVENT_STS);
> +	spin_lock_irqsave(&gpio->lock, flags);
> +	iowrite8(BIT(bit), status_addr);
> +	spin_unlock_irqrestore(&gpio->lock, flags);
> +}
> +
> +static void npcm_sgpio_irq_mask(struct irq_data *d)
> +{
> +	npcm_sgpio_irq_set_mask(d, true);
> +}
> +
> +static void npcm_sgpio_irq_unmask(struct irq_data *d)
> +{
> +	npcm_sgpio_irq_set_mask(d, false);
> +}
> +
> +static int npcm_sgpio_set_type(struct irq_data *d, unsigned int type)
> +{
> +	const struct npcm_sgpio_bank *bank;
> +	irq_flow_handler_t handler;
> +	struct npcm_sgpio *gpio;
> +	unsigned long flags;
> +	void __iomem *addr;
> +	int offset;
> +	u16 reg, val;
> +	u8 bit;
> +
> +	irqd_to_npcm_sgpio_data(d, &gpio, &bank, &bit, &offset);
> +
> +	switch (type & IRQ_TYPE_SENSE_MASK) {
> +	case IRQ_TYPE_EDGE_BOTH:
> +		val = NPCM_IXOEVCFG_BOTH;
> +		handler = handle_edge_irq;
> +		break;
> +	case IRQ_TYPE_EDGE_RISING:
> +		val = NPCM_IXOEVCFG_RISING;
> +		handler = handle_edge_irq;
> +		break;
> +	case IRQ_TYPE_EDGE_FALLING:
> +		val = NPCM_IXOEVCFG_FALLING;
> +		handler = handle_edge_irq;
> +		break;
> +	case IRQ_TYPE_LEVEL_HIGH:
> +		val = NPCM_IXOEVCFG_RISING;
> +		handler = handle_level_irq;
> +		break;
> +	case IRQ_TYPE_LEVEL_LOW:
> +		val = NPCM_IXOEVCFG_FALLING;
> +		handler = handle_level_irq;
> +		break;
> +	default:
> +		return -EINVAL;
> +	}
> +
> +	gpio->int_type[offset] = val;
> +
> +	spin_lock_irqsave(&gpio->lock, flags);
> +	npcm_sgpio_setup_enable(gpio, false);
> +	addr = bank_reg(gpio, bank, EVENT_CFG);
> +	reg = ioread16(addr);
> +
> +	reg |= (val << (bit * 2));
> +
> +	iowrite16(reg, addr);
> +	npcm_sgpio_setup_enable(gpio, true);
> +	spin_unlock_irqrestore(&gpio->lock, flags);
> +
> +	irq_set_handler_locked(d, handler);
> +
> +	return 0;
> +}
> +
> +static void npcm_sgpio_irq_handler(struct irq_desc *desc)
> +{
> +	struct gpio_chip *gc = irq_desc_get_handler_data(desc);
> +	struct irq_chip *ic = irq_desc_get_chip(desc);
> +	struct npcm_sgpio *gpio = gpiochip_get_data(gc);
> +	unsigned int i, j, girq;
> +	unsigned long reg;
> +
> +	chained_irq_enter(ic, desc);
> +
> +	for (i = 0; i < ARRAY_SIZE(npcm_sgpio_banks); i++) {
> +		const struct npcm_sgpio_bank *bank = &npcm_sgpio_banks[i];
> +
> +		reg = ioread8(bank_reg(gpio, bank, EVENT_STS));
> +		for_each_set_bit(j, &reg, 8) {
> +			girq = irq_find_mapping(gc->irq.domain, i * 8 + gpio->nout_sgpio + j);
> +			generic_handle_irq(girq);
> +		}
> +	}
> +
> +	chained_irq_exit(ic, desc);
> +}
> +
> +static int npcm_sgpio_setup_irqs(struct npcm_sgpio *gpio,
> +				 struct platform_device *pdev)
> +{
> +	int rc, i;
> +	struct gpio_irq_chip *irq;
> +
> +	rc = platform_get_irq(pdev, 0);
> +	if (rc < 0)
> +		return rc;
> +
> +	gpio->irq = rc;
> +
> +	npcm_sgpio_setup_enable(gpio, false);
> +
> +	/* Disable IRQ and clear Interrupt status registers for all SGPIO Pins. */
> +	for (i = 0; i < ARRAY_SIZE(npcm_sgpio_banks); i++) {
> +		const struct npcm_sgpio_bank *bank = &npcm_sgpio_banks[i];
> +
> +		iowrite16(0x0000, bank_reg(gpio, bank, EVENT_CFG));
> +		iowrite8(0xff, bank_reg(gpio, bank, EVENT_STS));
> +	}
> +
> +	gpio->intc.name = dev_name(&pdev->dev);
> +	gpio->intc.irq_ack = npcm_sgpio_irq_ack;
> +	gpio->intc.irq_mask = npcm_sgpio_irq_mask;
> +	gpio->intc.irq_unmask = npcm_sgpio_irq_unmask;
> +	gpio->intc.irq_set_type = npcm_sgpio_set_type;
> +
> +	irq = &gpio->chip.irq;
> +	irq->chip = &gpio->intc;
> +	irq->init_valid_mask = npcm_sgpio_irq_init_valid_mask;
> +	irq->handler = handle_bad_irq;
> +	irq->default_type = IRQ_TYPE_NONE;
> +	irq->parent_handler = npcm_sgpio_irq_handler;
> +	irq->parent_handler_data = gpio;
> +	irq->parents = &gpio->irq;
> +	irq->num_parents = 1;
> +
> +	return 0;
> +}
> +
> +static const int npcm750_SFT_CLK[] = {
> +		1024, 32, 8, 4, 3, 2,
> +};
> +
> +static const u8 npcm750_CLK_SEL[] = {
> +		0x00, 0x05, 0x07, 0x0C, 0x0D, 0x0E,
> +};
> +
> +static const int npcm845_SFT_CLK[] = {
> +		1024, 32, 16, 8, 4,
> +};
> +
> +static const u8 npcm845_CLK_SEL[] = {
> +		0x00, 0x05, 0x06, 0x07, 0x0C,
> +};
> +
> +static const struct npcm_clk_cfg npcm750_sgpio_pdata = {
> +	.SFT_CLK = npcm750_SFT_CLK,
> +	.CLK_SEL = npcm750_CLK_SEL,
> +	.cfg_opt = 6,
> +};
> +
> +static const struct npcm_clk_cfg npcm845_sgpio_pdata = {

You do not mention NPCM845 in the commit message.

> +	.SFT_CLK = npcm845_SFT_CLK,
> +	.CLK_SEL = npcm845_CLK_SEL,
> +	.cfg_opt = 5,
> +};
> +
> +static const struct of_device_id npcm_sgpio_of_table[] = {
> +	{ .compatible = "nuvoton,npcm750-sgpio", .data = &npcm750_sgpio_pdata, },
> +	{ .compatible = "nuvoton,npcm845-sgpio", .data = &npcm845_sgpio_pdata, },
> +	{}
> +};
> +
> +MODULE_DEVICE_TABLE(of, npcm_sgpio_of_table);
> +
> +static int npcm_sgpio_probe(struct platform_device *pdev)
> +{
> +	struct npcm_sgpio *gpio;
> +	const struct npcm_clk_cfg *clk_cfg;
> +	int rc;
> +	u32 nin_gpios, nout_gpios, sgpio_freq;
> +
> +	gpio = devm_kzalloc(&pdev->dev, sizeof(*gpio), GFP_KERNEL);
> +	if (!gpio)
> +		return -ENOMEM;
> +
> +	gpio->base = devm_platform_ioremap_resource(pdev, 0);
> +	if (IS_ERR(gpio->base))
> +		return PTR_ERR(gpio->base);
> +
> +	clk_cfg = device_get_match_data(&pdev->dev);
> +	if (!clk_cfg)
> +		return -EINVAL;
> +
> +	rc = device_property_read_u32(&pdev->dev, "nin_gpios", &nin_gpios);
> +	if (rc < 0) {
> +		dev_err(&pdev->dev, "Could not read ngpios property\n");

Mention nin_gpios?

> +		return -EINVAL;
> +	}
> +	rc = device_property_read_u32(&pdev->dev, "nout_gpios", &nout_gpios);
> +	if (rc < 0) {
> +		dev_err(&pdev->dev, "Could not read ngpios property\n");
> +		return -EINVAL;
> +	}
> +
> +	gpio->nin_sgpio = nin_gpios;
> +	gpio->nout_sgpio = nout_gpios;
> +	if (gpio->nin_sgpio > MAX_NR_HW_SGPIO || gpio->nout_sgpio > MAX_NR_HW_SGPIO) {
> +		dev_err(&pdev->dev, "Number of GPIOs exceeds the maximum of %d: input: %d output: %d\n",
> +			MAX_NR_HW_SGPIO, nin_gpios, nout_gpios);
> +		return -EINVAL;
> +	}
> +
> +	rc = device_property_read_u32(&pdev->dev, "bus-frequency", &sgpio_freq);
> +	if (rc < 0) {
> +		dev_err(&pdev->dev, "Could not read bus-frequency property\n");
> +		return -EINVAL;
> +	}
> +
> +	gpio->pclk = devm_clk_get(&pdev->dev, NULL);
> +	if (IS_ERR(gpio->pclk)) {
> +		dev_err(&pdev->dev, "devm_clk_get failed\n");

For a user, reading such an error message, I’d prefer a more elaborate 
message, for example, what won’t work, and that the serial GPIO driver 
initialization is aborted.

> +		return PTR_ERR(gpio->pclk);
> +	}
> +
> +	rc = npcm_sgpio_setup_clk(gpio, clk_cfg, sgpio_freq);
> +	if (rc < 0) {
> +		dev_err(&pdev->dev, "Failed to setup clock\n");
> +		return -EINVAL;
> +	}
> +	spin_lock_init(&gpio->lock);
> +	gpio->chip.parent = &pdev->dev;
> +	gpio->chip.ngpio = gpio->nin_sgpio + gpio->nout_sgpio;
> +	gpio->chip.init_valid_mask = npcm_sgpio_init_valid_mask;
> +	gpio->chip.direction_input = npcm_sgpio_dir_in;
> +	gpio->chip.direction_output = npcm_sgpio_dir_out;
> +	gpio->chip.get_direction = npcm_sgpio_get_direction;
> +	gpio->chip.request = NULL;
> +	gpio->chip.free = NULL;
> +	gpio->chip.get = npcm_sgpio_get;
> +	gpio->chip.set = npcm_sgpio_set;
> +	gpio->chip.set_config = NULL;
> +	gpio->chip.label = dev_name(&pdev->dev);
> +	gpio->chip.base = -1;
> +
> +	rc = npcm_sgpio_setup_irqs(gpio, pdev);
> +	if (rc < 0)
> +		return rc;
> +
> +	rc = devm_gpiochip_add_data(&pdev->dev, &gpio->chip, gpio);
> +	if (rc < 0)
> +		return rc;
> +
> +	npcm_sgpio_setup_enable(gpio, true);
> +	return 0;
> +}
> +
> +static struct platform_driver npcm_sgpio_driver = {
> +	.driver = {
> +		.name = KBUILD_MODNAME,
> +		.of_match_table = npcm_sgpio_of_table,
> +	},
> +	.probe	= npcm_sgpio_probe,
> +};
> +
> +module_platform_driver(npcm_sgpio_driver);
> +
> +MODULE_AUTHOR("Jim Liu <jjliu0@...oton.com>");
> +MODULE_AUTHOR("Joseph Liu <kwliu@...oton.com>");
> +MODULE_DESCRIPTION("Nuvoton NPCM Serial GPIO Driver");
> +MODULE_LICENSE("GPL v2");


Kind regards,

Paul


[1]: https://notabs.org/coding/smallIntsBigPenalty.htm

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ