Date:   Sat, 23 Jan 2021 09:00:11 -0800
From:   Guenter Roeck <linux@...ck-us.net>
To:     Campion Kang <campion.kang@...antech.com.tw>
Cc:     Lee Jones <lee.jones@...aro.org>, Rob Herring <robh+dt@...nel.org>,
        linux-kernel@...r.kernel.org, devicetree@...r.kernel.org,
        Jean Delvare <jdelvare@...e.com>,
        Wim Van Sebroeck <wim@...ux-watchdog.org>,
        linux-hwmon@...r.kernel.org, linux-watchdog@...r.kernel.org,
        AceLan Kao <chia-lin.kao@...onical.com>
Subject: Re: [PATCH v6 6/6] watchdog: ahc1ec0-wdt: Add sub-device watchdog
 for Advantech embedded controller

On Mon, Jan 18, 2021 at 08:37:49PM +0800, Campion Kang wrote:
> This is one of sub-device driver for Advantech embedded controller
> AHC1EC0. This driver provide watchdog functionality for Advantech
> related applications to restart the system.
> 
> Changed since V5:
> 	- remove unnecessary header files
> 	- bug fixed: reboot halt if watchdog enabled
> 	- Kconfig: add "AHC1EC0" string to clearly define the EC name
> 
> Signed-off-by: Campion Kang <campion.kang@...antech.com.tw>
> ---
>  drivers/watchdog/Kconfig       |  11 ++
>  drivers/watchdog/Makefile      |   1 +
>  drivers/watchdog/ahc1ec0-wdt.c | 261 +++++++++++++++++++++++++++++++++
>  3 files changed, 273 insertions(+)
>  create mode 100644 drivers/watchdog/ahc1ec0-wdt.c
> 
> diff --git a/drivers/watchdog/Kconfig b/drivers/watchdog/Kconfig
> index 7ff941e71b79..1a27836883ac 100644
> --- a/drivers/watchdog/Kconfig
> +++ b/drivers/watchdog/Kconfig
> @@ -1636,6 +1636,17 @@ config NIC7018_WDT
>  	  To compile this driver as a module, choose M here: the module will be
>  	  called nic7018_wdt.
>  
> +config AHC1EC0_WDT
> +	tristate "Advantech AHC1EC0 Watchdog Function"
> +	depends on MFD_AHC1EC0
> +	help
> +	  This is sub-device for Advantech AHC1EC0 embedded controller.
> +
> +	  This driver provide watchdog functionality for Advantech related
> +	  applications to restart the system.
> +	  To compile thie driver as a module, choose M here: the module will be
> +	  called ahc1ec0-wdt.
> +
>  # M68K Architecture
>  
>  config M54xx_WATCHDOG
> diff --git a/drivers/watchdog/Makefile b/drivers/watchdog/Makefile
> index 5c74ee19d441..7190811b1e50 100644
> --- a/drivers/watchdog/Makefile
> +++ b/drivers/watchdog/Makefile
> @@ -145,6 +145,7 @@ obj-$(CONFIG_INTEL_MID_WATCHDOG) += intel-mid_wdt.o
>  obj-$(CONFIG_INTEL_MEI_WDT) += mei_wdt.o
>  obj-$(CONFIG_NI903X_WDT) += ni903x_wdt.o
>  obj-$(CONFIG_NIC7018_WDT) += nic7018_wdt.o
> +obj-$(CONFIG_AHC1EC0_WDT) += ahc1ec0-wdt.o
>  obj-$(CONFIG_MLX_WDT) += mlx_wdt.o
>  
>  # M68K Architecture
> diff --git a/drivers/watchdog/ahc1ec0-wdt.c b/drivers/watchdog/ahc1ec0-wdt.c
> new file mode 100644
> index 000000000000..4497b6106b24
> --- /dev/null
> +++ b/drivers/watchdog/ahc1ec0-wdt.c
> @@ -0,0 +1,261 @@
> +// SPDX-License-Identifier: GPL-2.0-only
> +/*
> + * Watchdog Driver for Advantech Embedded Controller chip AHC1EC0
> + *
> + * Copyright 2020, Advantech IIoT Group
> + *
> + */
> +
> +#include <linux/errno.h>
> +#include <linux/kernel.h>
> +#include <linux/mfd/ahc1ec0.h>
> +#include <linux/module.h>
> +#include <linux/notifier.h>
> +#include <linux/platform_device.h>
> +#include <linux/reboot.h>
> +#include <linux/types.h>
> +#include <linux/watchdog.h>
> +
> +#define DRV_NAME      "ahc1ec0-wdt"
> +
> +struct ec_wdt_data {
> +	struct watchdog_device wdtdev;
> +	struct adv_ec_platform_data *adv_ec_data;
> +	struct notifier_block reboot_nb;
> +	struct mutex lock_ioctl;
> +	int is_enable;
> +	int current_timeout;

is_enable and current_timeout are only written to and never read, and thus pointless.

> +};
> +
> +#define EC_WDT_MIN_TIMEOUT 1	/* The watchdog devices minimum timeout value (in seconds). */
> +#define EC_WDT_MAX_TIMEOUT 600  /* The watchdog devices maximum timeout value (in seconds) */
> +#define EC_WDT_DEFAULT_TIMEOUT 45
> +
> +static int set_delay(struct adv_ec_platform_data *adv_ec_data, unsigned short delay_timeout_in_ms)
> +{
> +	if (write_hw_ram(adv_ec_data, EC_RESET_DELAY_TIME_L, delay_timeout_in_ms & 0x00FF)) {

write_hw_ram() is too generic for a device specific function. Please add a
prefix.

> +		pr_err("Failed to set Watchdog Retset Time Low byte.");

Reset ? 

Either case, please drop those error messages. If they happen they would likely
be persistent and end up clogging the kernel log.

> +		return -EINVAL;
> +	}
> +
> +	if (write_hw_ram(adv_ec_data, EC_RESET_DELAY_TIME_H, (delay_timeout_in_ms & 0xFF00) >> 8)) {
> +		pr_err("Failed to set Watchdog Retset Time Hight byte.");
> +		return -EINVAL;
> +	}
> +
> +	return 0;
> +}
> +
> +static int advwdt_set_heartbeat(unsigned long t)
> +{
> +	if (t < 1 || t > 6553) {
> +		pr_err("%s: the input timeout is out of range.",  __func__);
> +		pr_err("Please choose valid data between 1 ~ 6553.");
> +		return -EINVAL;
> +	}

Unecessary check. The watchdog core performs limit checks.
Repeating them does not add value.

> +
> +	return (t * 10);

Given that, this helper function is quite pointless (and misleading, since
it doesn't set anything).

> +}
> +
> +/* Notifier for system down */
> +static int advwdt_notify_sys(struct notifier_block *nb, unsigned long code, void *data)
> +{
> +	if (code == SYS_DOWN || code == SYS_HALT) {
> +		struct ec_wdt_data *ec_wdt_data;
> +
> +		ec_wdt_data = container_of(nb, struct ec_wdt_data, reboot_nb);
> +		if (!ec_wdt_data)
> +			return NOTIFY_BAD;
> +
> +		/* Turn the WDT off */
> +		if (write_hwram_command(ec_wdt_data->adv_ec_data, EC_WDT_STOP)) {
> +			pr_err("Failed to set Watchdog stop.");
> +			return -EINVAL;
> +		}
> +		ec_wdt_data->is_enable = 0;
> +		pr_info("%s: notify sys shutdown", __func__);
> +	}
> +
> +	return NOTIFY_DONE;
> +}

Please use the watchdog subsystem instead of declaring yor own notifier.

> +
> +static int ec_wdt_start(struct watchdog_device *wdd)
> +{
> +	int ret;
> +	int timeout, timeout_in_ms;
> +	struct ec_wdt_data *ec_wdt_data = watchdog_get_drvdata(wdd);
> +	struct adv_ec_platform_data *adv_ec_data;
> +
> +	dev_dbg(wdd->parent, "%s\n", __func__);
> +
> +	adv_ec_data = ec_wdt_data->adv_ec_data;
> +	timeout = wdd->timeout; /* The watchdog devices timeout value (in seconds). */

'timeout' is used just once. Just use wdd->timeout directly below.

> +
> +	mutex_lock(&ec_wdt_data->lock_ioctl);

The watchdog core ensures that there will be no more than one caller.
What is this lock supposed to protect ?

> +
> +	timeout_in_ms = advwdt_set_heartbeat(timeout);

timeout_in_ms is the timeout in 10th of seconds. "_ms" is misleading.

> +	if (timeout_in_ms < 0) {
> +		mutex_unlock(&ec_wdt_data->lock_ioctl);
> +		return timeout_in_ms;
> +	}
> +
> +	ret = set_delay(adv_ec_data, (unsigned short)(timeout_in_ms-1));

Unnecessary type cast.

> +	if (ret) {
> +		dev_err(wdd->parent, "Failed to set Watchdog delay (ret=%x).\n", ret);

This driver is extremely noisy when it comes to error messages.

> +		mutex_unlock(&ec_wdt_data->lock_ioctl);
> +		return ret;
> +	}
> +	ret = write_hwram_command(adv_ec_data, EC_WDT_STOP);

Errors from this call are ignored. If that is on purpose, the assignment
to 'ret' is unnecessary.

> +	ret = write_hwram_command(adv_ec_data, EC_WDT_START);
> +	if (ret) {
> +		dev_err(wdd->parent, "Failed to set Watchdog start (ret=%x).\n", ret);
> +		mutex_unlock(&ec_wdt_data->lock_ioctl);
> +		return ret;

This should be  handled with a goto to the end of the function.
See error handling in coding style.

> +	}
> +	ec_wdt_data->is_enable = 1;
> +	ec_wdt_data->current_timeout = timeout_in_ms/10;

The conversion from timeout -> timeout_in_ms -> timeout
seems to be quite pointless.

> +
> +	mutex_unlock(&ec_wdt_data->lock_ioctl);
> +	return 0;
> +}
> +
> +static int ec_wdt_stop(struct watchdog_device *wdd)
> +{
> +	int ret;
> +	struct ec_wdt_data *ec_wdt_data = watchdog_get_drvdata(wdd);
> +	struct adv_ec_platform_data *adv_ec_data;
> +
> +	dev_dbg(wdd->parent, "%s\n", __func__);
> +
> +	adv_ec_data = ec_wdt_data->adv_ec_data;
> +
> +	mutex_lock(&ec_wdt_data->lock_ioctl);
> +	ret = write_hwram_command(adv_ec_data, EC_WDT_STOP);
> +	mutex_unlock(&ec_wdt_data->lock_ioctl);
> +	if (ret)
> +		pr_err("Failed to set Watchdog stop.");
> +	else
> +		ec_wdt_data->is_enable = 0;
> +
> +	return ret;
> +}
> +
> +static int ec_wdt_ping(struct watchdog_device *wdd)
> +{
> +	int ret;
> +	struct ec_wdt_data *ec_wdt_data = watchdog_get_drvdata(wdd);
> +	struct adv_ec_platform_data *adv_ec_data;
> +
> +	dev_dbg(wdd->parent, "%s\n", __func__);
> +
> +	adv_ec_data = ec_wdt_data->adv_ec_data;
> +
> +	mutex_lock(&ec_wdt_data->lock_ioctl);
> +	ret = write_hwram_command(adv_ec_data, EC_WDT_RESET);
> +	mutex_unlock(&ec_wdt_data->lock_ioctl);
> +	if (ret) {
> +		pr_err("Failed to set Watchdog reset.");
> +		return -EINVAL;
> +	}
> +
> +	return 0;
> +}
> +
> +static int ec_wdt_set_timeout(struct watchdog_device *wdd,
> +				unsigned int timeout)
> +{
> +	dev_dbg(wdd->parent, "%s, timeout=%d\n", __func__, timeout);
> +
> +	wdd->timeout = timeout;
> +
> +	if (watchdog_active(wdd))
> +		return ec_wdt_start(wdd);
> +
> +	return 0;
> +}
> +
> +static const struct watchdog_info ec_watchdog_info = {
> +	.options = WDIOF_SETTIMEOUT | WDIOF_KEEPALIVEPING | WDIOF_MAGICCLOSE,
> +	.identity = "AHC1EC0 Watchdog",
> +};
> +
> +static const struct watchdog_ops ec_watchdog_ops = {
> +	.owner = THIS_MODULE,
> +	.start = ec_wdt_start,
> +	.stop = ec_wdt_stop,
> +	.ping = ec_wdt_ping,
> +	.set_timeout = ec_wdt_set_timeout,
> +};
> +
> +static int adv_ec_wdt_probe(struct platform_device *pdev)
> +{
> +	int ret;
> +	struct device *dev = &pdev->dev;
> +	struct adv_ec_platform_data *adv_ec_data;
> +	struct ec_wdt_data *ec_wdt_data;
> +	struct watchdog_device *wdd;
> +
> +	dev_info(dev, "watchdog probe start\n");
> +
> +	adv_ec_data = dev_get_drvdata(dev->parent);
> +	if (!adv_ec_data)
> +		return -EINVAL;
> +
> +	ec_wdt_data = devm_kzalloc(dev, sizeof(struct ec_wdt_data), GFP_KERNEL);
> +	if (!ec_wdt_data)
> +		return -ENOMEM;
> +
> +	mutex_init(&ec_wdt_data->lock_ioctl);
> +
> +	ec_wdt_data->adv_ec_data = adv_ec_data;
> +	wdd = &ec_wdt_data->wdtdev;
> +
> +	watchdog_init_timeout(&ec_wdt_data->wdtdev, 0, dev);
> +
> +	//watchdog_set_nowayout(&ec_wdt_data->wdtdev, WATCHDOG_NOWAYOUT);

Please no C++ comments in the code.

> +	watchdog_set_drvdata(&ec_wdt_data->wdtdev, ec_wdt_data);
> +	platform_set_drvdata(pdev, ec_wdt_data);
> +
> +	wdd->info = &ec_watchdog_info;
> +	wdd->ops = &ec_watchdog_ops;
> +	wdd->min_timeout = EC_WDT_MIN_TIMEOUT;
> +	wdd->max_timeout = EC_WDT_MAX_TIMEOUT;
> +	wdd->parent = dev;
> +
> +	ec_wdt_data->wdtdev.timeout = EC_WDT_DEFAULT_TIMEOUT;
> +	ec_wdt_data->is_enable = 0;
> +	ec_wdt_data->current_timeout = EC_WDT_DEFAULT_TIMEOUT;
> +
> +	watchdog_stop_on_unregister(wdd);
> +
> +	ec_wdt_data->reboot_nb.notifier_call = advwdt_notify_sys;
> +	ret = devm_register_reboot_notifier(dev, &ec_wdt_data->reboot_nb);
> +	if (ret) {
> +		dev_err(dev, "watchdog%d: Cannot register reboot notifier (%d)\n",
> +			wdd->id, ret);
> +		return ret;
> +	}

Use watchdog_stop_on_reboot().

> +
> +	ret = devm_watchdog_register_device(dev, wdd);
> +	if (ret != 0)
> +		dev_err(dev, "watchdog_register_device() failed: %d\n",
> +			ret);

And right afterwards "watchdog register success", and the error
is ignored.

> +
> +	dev_info(dev, "watchdog register success\n");
> +
> +	return 0;
> +}
> +
> +static struct platform_driver adv_wdt_drv = {
> +	.driver = {
> +		.name = DRV_NAME,
> +	},
> +	.probe = adv_ec_wdt_probe,
> +};
> +module_platform_driver(adv_wdt_drv);
> +
> +MODULE_LICENSE("GPL");
> +MODULE_ALIAS("platform:" DRV_NAME);
> +MODULE_DESCRIPTION("Advantech Embedded Controller Watchdog Driver.");
> +MODULE_AUTHOR("Campion Kang <campion.kang@...antech.com.tw>");
> +MODULE_VERSION("1.0");
> -- 
> 2.17.1
> 

Powered by blists - more mailing lists