2016-01-11 8:31 GMT+09:00 <richard.dorsch@xxxxxxxxx>:
> From: Richard Vidal-Dorsch <richard.dorsch@xxxxxxxxx>
>
> This patch adds Advantech iManager Embedded Controller MFD core driver.
> This mfd core dirver provides an interface to GPIO, I2C, HWmon,
> Watchdog, and Backlight/Brightness control.
>
> Signed-off-by: Richard Vidal-Dorsch <richard.dorsch@xxxxxxxxx>
> ---
> Documentation/devicetree/bindings/mfd/imanager.txt | 33 +
> MAINTAINERS | 13 +
> drivers/mfd/Kconfig | 20 +
> drivers/mfd/Makefile | 2 +
> drivers/mfd/imanager-core.c | 287 +++++
> drivers/mfd/imanager-ec.c | 1345 ++++++++++++++++++++
> include/linux/mfd/imanager/core.h | 31 +
> include/linux/mfd/imanager/ec.h | 210 +++
> 8 files changed, 1941 insertions(+)
> create mode 100644 Documentation/devicetree/bindings/mfd/imanager.txt
> create mode 100644 drivers/mfd/imanager-core.c
> create mode 100644 drivers/mfd/imanager-ec.c
> create mode 100644 include/linux/mfd/imanager/core.h
> create mode 100644 include/linux/mfd/imanager/ec.h
>
> diff --git a/Documentation/devicetree/bindings/mfd/imanager.txt b/Documentation/devicetree/bindings/mfd/imanager.txt
> new file mode 100644
> index 0000000..bf58a96
> --- /dev/null
> +++ b/Documentation/devicetree/bindings/mfd/imanager.txt
> @@ -0,0 +1,33 @@
> +Note
> +====
> +
> +This is a set of platform drivers which provide support for multiple
> +embedded features such as GPIO, I2C/SMBus, Hardware Monitoring, Watchdog,
> +and Backlight/Brightness control. Those features are available on Advantech
> +Embedded boards such as SOM, MIO, AIMB, and PCM.
> +Datasheets of each product line can be downloaded from www.advantech.com
> +
> +Author:
> + Richard Vidal-Dorsch <richard.dorsch@xxxxxxxxxxxxx>
> +
> +
> +Kernel driver imanager
> +======================
> +
> +This driver provides a communication layer to the Advantech iManager EC
> +firmware which is factory loaded onto ITE IT8518/28 chips. The type of
> +communication is message based. Clients (gpio, i2c, hwmon etc. drivers)
> +request data from Advantech iManager (polling, not interrupt driven). If
> +a response is been received within a time frame, the data is been extracted
> +from the message and then passed to the caller (clients).
> +
> + Supported chips:
> + * Advantech EC based on ITE IT8518
> + Prefix: imanager
> + Addresses: 0x029e/0x029f
> + Datasheet: Available from ITE upon request
> + * Advantech EC based on ITE IT8528
> + Prefix: imanager
> + Addresses: 0x0299/0x29a
> + Datasheet: Available from ITE upon request
Hi,
This does not describe the bindings. It is not the place for it. If
you have bindings please put the them as separate patch (first one).
> +
> diff --git a/MAINTAINERS b/MAINTAINERS
> index 233f834..8d25fdc 100644
> --- a/MAINTAINERS
> +++ b/MAINTAINERS
> @@ -5391,6 +5391,19 @@ M: Stanislaw Gruszka <stf_xl@xxxxx>
> S: Maintained
> F: drivers/usb/atm/ueagle-atm.c
>
> +IMANAGER ADVANTECH EC DRIVER
> +M: Richard Vidal-Dorsch <richard.dorsch@xxxxxxxxxxxxx>
> +S: Maintained
> +F: Documentation/devicetree/bindings/mfd/imanager.txt
> +F: Documentation/hwmon/imanager
> +F: Documentation/i2c/busses/i2c-imanager
> +F: drivers/mfd/imanager-core.c
> +F: drivers/gpio/imanager-gpio.c
> +F: drivers/hwmon/imanager-hwmon.c
> +F: drivers/i2c/busses/imanager-i2c.c
> +F: drivers/video/backlight/imanager-bl.c
> +F: drivers/watchdog/imanager-wdt.c
> +
> INA209 HARDWARE MONITOR DRIVER
> M: Guenter Roeck <linux@xxxxxxxxxxxx>
> L: lm-sensors@xxxxxxxxxxxxxx
> diff --git a/drivers/mfd/Kconfig b/drivers/mfd/Kconfig
> index 4d92df6..4dc7c13 100644
> --- a/drivers/mfd/Kconfig
> +++ b/drivers/mfd/Kconfig
> @@ -322,6 +322,26 @@ config MFD_INTEL_QUARK_I2C_GPIO
> their respective IO driver.
> The GPIO exports a total amount of 8 interrupt-capable GPIOs.
>
> +config MFD_IMANAGER
> + tristate "Advantech iManager Embedded Controller"
> + depends on PCI
> + depends on X86
> + select MFD_CORE
> + help
> + This is the core driver for Advantech iManager EC as found on some
> + Advantech SOM, MIO, AIMB, and PCM modules/boards. The EC may provide
> + functions like GPIO, I2C interface, HW monitoring, Watchdog, and
> + backlight/brightness control.
> +
> + The following Advantech boards are supported:
> + * All SOM modules newer than SOM-5788
> + * MIO-5250/5251/5270/5271 and newer
> + * PCM-9388
> + * AIMB-274/231
> +
> + This driver can also be built as a module. If so, the module
> + will be called imanager.
> +
> config LPC_ICH
> tristate "Intel ICH LPC"
> depends on PCI
> diff --git a/drivers/mfd/Makefile b/drivers/mfd/Makefile
> index a8b76b8..c9c63d9 100644
> --- a/drivers/mfd/Makefile
> +++ b/drivers/mfd/Makefile
> @@ -142,6 +142,8 @@ obj-$(CONFIG_MFD_DB8500_PRCMU) += db8500-prcmu.o
> obj-$(CONFIG_AB8500_CORE) += ab8500-core.o ab8500-sysctrl.o
> obj-$(CONFIG_MFD_TIMBERDALE) += timberdale.o
> obj-$(CONFIG_PMIC_ADP5520) += adp5520.o
> +imanager-objs := imanager-core.o imanager-ec.o
> +obj-$(CONFIG_MFD_IMANAGER) += imanager.o
> obj-$(CONFIG_MFD_KEMPLD) += kempld-core.o
> obj-$(CONFIG_MFD_INTEL_QUARK_I2C_GPIO) += intel_quark_i2c_gpio.o
> obj-$(CONFIG_LPC_SCH) += lpc_sch.o
> diff --git a/drivers/mfd/imanager-core.c b/drivers/mfd/imanager-core.c
> new file mode 100644
> index 0000000..d573ecb
> --- /dev/null
> +++ b/drivers/mfd/imanager-core.c
> @@ -0,0 +1,287 @@
> +/*
> + * Advantech iManager MFD core driver
> + *
> + * Copyright (C) 2016 Advantech Co., Ltd., Irvine, CA, USA
> + * Author: Richard Vidal-Dorsch <richard.dorsch@xxxxxxxxxxxxx>
> + *
> + * This program is free software; you can redistribute it and/or modify it
> + * under the terms of the GNU General Public License as published by the
> + * Free Software Foundation; either version 2 of the License, or (at your
> + * option) any later version.
> + */
> +
> +#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
> +
> +#include <linux/module.h>
> +#include <linux/init.h>
> +#include <linux/platform_device.h>
> +#include <linux/ioport.h>
Do you need ioport?
> +#include <linux/io.h>
> +#include <linux/mfd/core.h>
> +#include <linux/mfd/imanager/core.h>
> +#include <linux/mfd/imanager/ec.h>
> +
> +static struct platform_device *pdev;
Wait, why? Why you create a platform device?
> +
> +enum imanager_cells {
> + IMANAGER_BL,
> + IMANAGER_GPIO,
> + IMANAGER_HWMON,
> + IMANAGER_I2C,
> + IMANAGER_WDT,
> +};
Is the enum uses beside of the index in mfd_cell array? If not, remove it.
> +
> +static const char * const chip_names[] = {
> + "it8516",
> + "it8518",
> + "it8528",
> + NULL
> +};
> +
> +static struct resource imanager_ioresource = {
> + .start = IT8516_DAT_PORT,
> + .end = IT8518_DAT_PORT,
> + .flags = IORESOURCE_IO,
> +};
> +
> +/*
> + * Devices which are part of the iManager and are available via firmware.
> + */
> +static struct mfd_cell imanager_devs[] = {
> + [IMANAGER_BL] = {
> + .name = "imanager_backlight",
> + },
> + [IMANAGER_GPIO] = {
> + .name = "imanager_gpio",
> + },
> + [IMANAGER_HWMON] = {
> + .name = "imanager_hwmon",
> + },
> + [IMANAGER_I2C] = {
> + .name = "imanager_i2c",
> + },
> + [IMANAGER_WDT] = {
> + .name = "imanager_wdt",
> + },
> +};
> +
> +const char *project_type_to_str(int type)
static
> +{
> + const char *version_type;
> +
> + switch (type) {
> + case 'V':
> + version_type = "Release";
> + break;
> + case 'X':
> + version_type = "Engineering Sample";
> + break;
> + case 'A' ... 'U':
> + version_type = "Custom";
> + break;
> + default:
> + version_type = "Unknown";
> + break;
> + }
> +
> + return version_type;
> +}
> +
> +static ssize_t imanager_name_show(struct device *dev,
> + struct device_attribute *attr, char *buf)
> +{
> + struct imanager_platform_data *pdata = dev_get_platdata(dev);
> + const struct ec_info *info = &pdata->dev->info;
> +
> + return scnprintf(buf, PAGE_SIZE, "%s\n", info->pcb_name);
> +}
> +
> +static ssize_t imanager_kversion_show(struct device *dev,
> + struct device_attribute *attr, char *buf)
> +{
> + struct imanager_platform_data *pdata = dev_get_platdata(dev);
> + const struct ec_info *info = &pdata->dev->info;
> +
> + return scnprintf(buf, PAGE_SIZE, "%d.%d\n",
> + info->version.kernel_major,
> + info->version.kernel_minor);
> +}
> +
> +static ssize_t imanager_fwversion_show(struct device *dev,
> + struct device_attribute *attr, char *buf)
> +{
> + struct imanager_platform_data *pdata = dev_get_platdata(dev);
> + const struct ec_info *info = &pdata->dev->info;
> +
> + return scnprintf(buf, PAGE_SIZE, "%d.%d\n",
> + info->version.firmware_major,
> + info->version.firmware_minor);
> +}
> +
> +static ssize_t imanager_type_show(struct device *dev,
> + struct device_attribute *attr, char *buf)
> +{
> + struct imanager_platform_data *pdata = dev_get_platdata(dev);
> + const struct ec_info *info = &pdata->dev->info;
> +
> + return scnprintf(buf, PAGE_SIZE, "%s\n",
> + project_type_to_str(info->version.type));
> +}
> +
> +static ssize_t imanager_chip_name_show(struct device *dev,
> + struct device_attribute *attr, char *buf)
> +{
> + struct imanager_platform_data *pdata = dev_get_platdata(dev);
> +
> + return scnprintf(buf, PAGE_SIZE, "%s\n", pdata->chip_name);
> +}
> +
> +static DEVICE_ATTR(imanager_name, S_IRUGO, imanager_name_show, NULL);
> +static DEVICE_ATTR(imanager_kversion, S_IRUGO, imanager_kversion_show, NULL);
> +static DEVICE_ATTR(imanager_fwversion, S_IRUGO, imanager_fwversion_show, NULL);
> +static DEVICE_ATTR(imanager_type, S_IRUGO, imanager_type_show, NULL);
> +static DEVICE_ATTR(imanager_chip_name, S_IRUGO, imanager_chip_name_show, NULL);
You made so many of these just to print the name/version... What is
the purpose of exposing this to user-space? Who will use it and for
what purpose? What need does it fulfill?
For debugging purposes usually dev_info() is sufficient. If not (why
not?) I would expect at most one sysfs entry.
Anyway you are making this as part of ABI so it requires
documentation: Documentation/ABI.
> +
> +static struct attribute *imanager_core_attributes[] = {
> + &dev_attr_imanager_name.attr,
> + &dev_attr_imanager_kversion.attr,
> + &dev_attr_imanager_fwversion.attr,
> + &dev_attr_imanager_type.attr,
> + &dev_attr_imanager_chip_name.attr,
> + NULL
> +};
> +
> +static const struct attribute_group imanager_core_attr_group = {
> + .attrs = imanager_core_attributes,
> +};
> +
> +static int imanager_platform_create(void)
> +{
> + struct device *dev;
> + struct imanager_platform_data platdata;
> + int err;
> +
> + pdev = platform_device_alloc("imanager-core", -1);
> + if (!pdev)
> + return -ENOMEM;
> +
> + dev = &pdev->dev;
> +
> + err = platform_device_add_data(pdev, &platdata, sizeof(platdata));
> + if (err)
> + goto exit_device_put;
> +
> + err = platform_device_add_resources(pdev, &imanager_ioresource, 1);
> + if (err)
> + goto exit_device_put;
> +
> + err = platform_device_add(pdev);
> + if (err)
> + goto exit_device_put;
> +
> + err = mfd_add_devices(dev, pdev->id, imanager_devs,
> + ARRAY_SIZE(imanager_devs), NULL, -1, NULL);
This is part of probe, not __init. Beside that I don't have a clue why
entire imanager_platform_create() exists. :)
> + if (err)
> + goto exit_device_unregister;
> +
> + return 0;
> +
> +exit_device_unregister:
> + platform_device_unregister(pdev);
> +exit_device_put:
> + platform_device_put(pdev);
> +
> + return err;
> +}
> +
> +static int imanager_probe(struct platform_device *pdev)
> +{
> + struct device *dev = &pdev->dev;
> + struct imanager_platform_data *pdata = dev_get_platdata(dev);
> + struct imanager_device_data *imanager;
> + int ret;
> +
> + if (!pdev)
> + return -EINVAL;
Wait? Which pdev? Local or file-scope? Device cannot be probed for a
NULL device!
> +
> + imanager = devm_kzalloc(dev, sizeof(*imanager), GFP_KERNEL);
> + if (!imanager)
> + return -ENOMEM;
> +
> + imanager->dev = dev;
> + mutex_init(&imanager->lock);
> +
> + platform_set_drvdata(pdev, imanager);
> +
> + pdata->dev = imanager_get_ec_device();
> + pdata->chip_name = chip_names[pdata->dev->id];
> +
> + dev_info(dev, "Found Advantech iManager %s - %s %d.%d/%d.%d (%s)\n",
> + pdata->chip_name,
> + pdata->dev->info.pcb_name,
> + pdata->dev->info.version.kernel_major,
> + pdata->dev->info.version.kernel_minor,
> + pdata->dev->info.version.firmware_major,
> + pdata->dev->info.version.firmware_minor,
> + project_type_to_str(pdata->dev->info.version.type));
> +
> + ret = sysfs_create_group(&dev->kobj, &imanager_core_attr_group);
> + if (ret)
> + return ret;
> +
> + return 0;
> +}
> +
> +static int imanager_remove(struct platform_device *pdev)
> +{
> + struct device *dev = &pdev->dev;
> +
> + sysfs_remove_group(&dev->kobj, &imanager_core_attr_group);
> +
> + mfd_remove_devices(dev);
If devices were created in __init then they must be removed in __exit,
not in remove()... so this is also a good hint that you shouldn't put
mfd_add_devices() in __init.
> +
> + return 0;
> +}
> +
> +static struct platform_driver imanager_driver = {
> + .driver = {
> + .name = "imanager-core",
> + },
> + .probe = imanager_probe,
> + .remove = imanager_remove,
> +};
> +
> +static int __init imanager_init(void)
> +{
> + int ret;
> +
> + ret = imanager_ec_probe();
> + if (ret < 0)
> + return ret;
This confuses me even more: probing before register?
> +
> + ret = imanager_platform_create();
> + if (ret)
> + return ret;
> +
> + ret = platform_driver_register(&imanager_driver);
> + if (ret)
> + return ret;
> +
> + return 0;
> +}
> +
> +static void __exit imanager_exit(void)
> +{
> + if (pdev)
> + platform_device_unregister(pdev);
> +
> + platform_driver_unregister(&imanager_driver);
> +}
> +
> +module_init(imanager_init);
> +module_exit(imanager_exit);
The driver does not support DT... what is the purpose of DT bindings document?
> +
> +MODULE_DESCRIPTION("Advantech iManager Core Driver");
> +MODULE_AUTHOR("Richard Vidal-Dorsch <richard.dorsch at advantech.com>");
> +MODULE_LICENSE("GPL");
> +MODULE_ALIAS("platform:imanager-core");
> diff --git a/drivers/mfd/imanager-ec.c b/drivers/mfd/imanager-ec.c
> new file mode 100644
> index 0000000..be554ba
> --- /dev/null
> +++ b/drivers/mfd/imanager-ec.c
> @@ -0,0 +1,1345 @@
> +/*
> + * Advantech iManager Core - Firmware Interface
> + *
> + * Copyright (C) 2016 Advantech Co., Ltd., Irvine, CA, USA
> + * Author: Richard Vidal-Dorsch <richard.dorsch@xxxxxxxxxxxxx>
> + *
> + * This program is free software; you can redistribute it and/or modify it
> + * under the terms of the GNU General Public License as published by the
> + * Free Software Foundation; either version 2 of the License, or (at your
> + * option) any later version.
> + */
> +
> +#include <linux/types.h>
> +#include <linux/errno.h>
> +#include <linux/bug.h>
> +#include <linux/io.h>
> +#include <linux/delay.h>
> +#include <linux/string.h>
> +#include <linux/byteorder/generic.h>
> +#include <linux/module.h>
> +#include <linux/swab.h>
> +#include <linux/mfd/imanager/ec.h>
> +
> +/**
> + * This is the delay time between two EC transactions.
> + * Values lower than 200us are not encouraged and may
> + * cause I/O errors
> + */
> +#define EC_MICRO_DELAY 200
> +#define EC_MAX_RETRY 1000
> +
> +#define EC_MSG_OFFSET_CMD 0UL
> +#define EC_MSG_OFFSET_STATUS 1UL
> +#define EC_MSG_OFFSET_PARAM 2UL
> +#define EC_MSG_OFFSET_DATA(N) (3UL + N)
> +#define EC_MSG_OFFSET_PAYLOAD(N) (7UL + N)
> +
> +/* The Device ID registers - 16 bit */
> +#define DEVID_REG_MSB 0x20
> +#define DEVID_REG_LSB 0x21
> +
> +/*
> + * IT8528 based firmware require a read/write command offset.
> + */
> +#define EC_CMD_OFFSET_READ 0xA0UL
> +#define EC_CMD_OFFSET_WRITE 0x50UL
> +
> +#define EC_STATUS_SUCCESS BIT(0)
> +#define EC_STATUS_CMD_COMPLETE BIT(7)
> +
> +#define PCB_NAME_MAX_SIZE 8UL
> +#define EC_I2C_BLOCK_SIZE 32
> +#define EC_MAX_DID 32UL
> +
> +#define DID_LABEL_SIZE 24UL
> +#define DID_DESC_SIZE 32UL
> +
> +#define EC_KERNEL_MINOR(x) (LOBYTE16(x))
> +#define EC_KERNEL_MAJOR(x) ({ \
> + typeof(x) __x = (HIBYTE16(x)); \
> + ((__x >> 4) * 10 + (__x & 0x000f)); })
> +#define EC_FIRMWARE_MINOR(x) (LOBYTE16(x))
> +#define EC_FIRMWARE_MAJOR(x) EC_KERNEL_MAJOR(x)
> +#define EC_PROJECT_CODE(x) ((char)(LOBYTE16(x)))
> +
> +enum ec_device_type {
> + ADC = 1,
> + DAC,
> + GPIO,
> + IRQ,
> + PWM,
> + SMB,
> + TACH
> +};
> +
> +enum ec_device_id {
> + /* GPIO */
> + ALTGPIO0 = 0x10,
> + ALTGPIO1,
> + ALTGPIO2,
> + ALTGPIO3,
> + ALTGPIO4,
> + ALTGPIO5,
> + ALTGPIO6,
> + ALTGPIO7,
> + /* Button (GPIO) */
> + BUTTON0,
> + BUTTON1,
> + BUTTON2,
> + BUTTON3,
> + BUTTON4,
> + BUTTON5,
> + BUTTON6,
> + BUTTON7,
> + /* FAN */
> + CPUFAN_2P,
> + CPUFAN_4P,
> + SYSFAN1_2P,
> + SYSFAN1_4P,
> + SYSFAN2_2P,
> + SYSFAN2_4P,
> + /* Brightness Control */
> + BRIGHTNESS,
> + /* System Speaker */
> + PCBEEP,
> + /* SMBus */
> + SMBOEM0,
> + SMBOEM1,
> + SMBOEM2,
> + SMBEEPROM,
> + SMBTHERMAL0,
> + SMBTHERMAL1,
> + SMBSECEEP,
> + I2COEM,
> + /* Speaker */
> + SPEAKER = 0x30,
> + /* SMBus */
> + SMBEEP2K = 0x38,
> + OEMEEP,
> + OEMEEP2K,
> + PECI,
> + SMBOEM3,
> + SMLINK,
> + SMBSLV,
> + /* LED */
> + POWERLED = 0x40,
> + BATLEDG,
> + OEMLED0,
> + OEMLED1,
> + OEMLED2,
> + BATLEDR,
> + /* Smart Battery */
> + SMARTBAT1 = 0x48,
> + SMARTBAT2,
> + /* ADC */
> + CMOSBAT = 0x50,
> + CMOSBATx2,
> + CMOSBATx10,
> + LIBAT,
> + LIBATx2,
> + LIBATx10,
> + ADC5VS0,
> + ADC5VS0x2,
> + ADC5VS0x10,
> + ADC5VS5,
> + ADC5VS5x2,
> + ADC5VS5x10,
> + ADC33VS0,
> + ADC33VS0x2,
> + ADC33VS0x10,
> + ADC33VS5,
> + ADC33VS5x2, /* 0x60 */
> + ADC33VS5x10,
> + ADC12VS0,
> + ADC12VS0x2,
> + ADC12VS0x10,
> + VCOREA,
> + VCOREAx2,
> + VCOREAx10,
> + VCOREB,
> + VCOREBx2,
> + VCOREBx10,
> + ADCDC,
> + ADCDCx2,
> + ADCDCx10,
> + VSTBY,
> + VSTBYx2,
> + VSTBYx10, /* 0x70 */
> + VAUX,
> + VAUXx2,
> + VAUXx10,
> + CURRENT,
> + /* Watchdog */
> + WDIRQ = 0x78,
> + WDNMI,
> + /* FAN Tacho */
> + TACHO0 = 0x80,
> + TACHO1,
> + TACHO2,
> + /* Brightness/Backlight Control */
> + BRIGHTNESS2 = 0x88,
> + BACKLIGHT1,
> + BACKLIGHT2
> +};
> +
> +enum ec_dynamic_table_type {
> + EC_DYN_DID,
> + EC_DYN_HWP,
> + EC_DYN_POL
> +};
> +
> +struct ec_devtbl {
> + int id;
> + int type;
> + int scale;
> + const char label[DID_LABEL_SIZE];
> + const char description[DID_DESC_SIZE];
> +};
> +
> +struct ec_dyn_devtbl {
> + int did; /* Device ID */
> + int hwp; /* Hardware Pin */
> + int pol; /* Polarity */
> + const struct ec_devtbl *devtbl; /* Device table Entry */
> +};
> +
> +struct imanager_data {
> + int (*read)(int cmd);
> + int (*write)(int cmd, int value);
> +
> + struct ec_dyn_devtbl dyn[EC_MAX_DID];
> +
> + struct imanager_ec_device dev;
> + struct imanager_hwmon_device sensors;
> + struct imanager_gpio_device gpio;
> + struct imanager_i2c_device i2c;
> + struct imanager_watchdog_device wdt;
> + struct imanager_backlight_device blc;
> +};
> +
> +struct ec_version_raw {
> + u16 kernel,
> + chipid,
> + project_code,
> + firmware;
> +};
> +
> +enum ec_ram_type {
> + EC_RAM_ACPI = 1,
> + EC_RAM_HW,
> + EC_RAM_EXT
> +};
> +
> +static struct imanager_data ec;
Why this is file-scope, not stored in device state?
> +
> +static const struct ec_devtbl devtbl[] = {
I didn't quite get the purpose of this array.
> + { ALTGPIO0, GPIO, -1, "gpio0" },
> + { ALTGPIO1, GPIO, -1, "gpio1" },
> + { ALTGPIO2, GPIO, -1, "gpio2" },
> + { ALTGPIO3, GPIO, -1, "gpio3" },
> + { ALTGPIO4, GPIO, -1, "gpio4" },
> + { ALTGPIO5, GPIO, -1, "gpio5" },
> + { ALTGPIO6, GPIO, -1, "gpio6" },
> + { ALTGPIO7, GPIO, -1, "gpio7" },
> + { BUTTON0, GPIO, -1, "button0" },
> + { BUTTON1, GPIO, -1, "button1" },
> + { BUTTON2, GPIO, -1, "button2" },
> + { BUTTON3, GPIO, -1, "button3" },
> + { BUTTON4, GPIO, -1, "button4" },
> + { BUTTON5, GPIO, -1, "button4" },
> + { BUTTON6, GPIO, -1, "button4" },
> + { BUTTON7, GPIO, -1, "button4" },
> + { CPUFAN_2P, PWM, 2, "FAN CPU" },
> + { CPUFAN_4P, PWM, 4, "FAN CPU" },
> + { SYSFAN1_2P, PWM, 2, "FAN SYS1" },
> + { SYSFAN1_4P, PWM, 4, "FAN SYS1" },
> + { SYSFAN2_2P, PWM, 2, "FAN SYS2" },
> + { SYSFAN2_4P, PWM, 4, "FAN SYS2" },
> + { BRIGHTNESS, PWM, -1, "Brightness1" },
> + { PCBEEP, PWM, -1, "Beep" },
> + { SMBOEM0, SMB, -1, "SMB1" },
> + { SMBOEM1, SMB, -1, "SMB2" },
> + { SMBOEM2, SMB, -1, "SMB3" },
> + { SMBEEPROM, SMB, -1, "SMBEEP" },
> + { SMBTHERMAL0, SMB, -1, "SMBTHERM0" },
> + { SMBTHERMAL1, SMB, -1, "SMBTHERM1" },
> + { SMBSECEEP, SMB, -1, "SMBSECEEP" },
> + { I2COEM, SMB, -1, "I2COEM" },
> + { SPEAKER, DAC, -1, "Speaker" },
> + { SMBEEP2K, SMB, -1, "SMBEEP2K" },
> + { OEMEEP, SMB, -1, "OEMEEP" },
> + { OEMEEP2K, SMB, -1, "OEMEEP2K" },
> + { PECI, SMB, -1, "SMB_PECI" },
> + { SMBOEM3, SMB, -1, "SMBOEM3" },
> + { SMLINK, SMB, -1, "SMLINK" },
> + { SMBSLV, SMB, -1, "SMBSLV" },
> + { POWERLED, GPIO, -1, "Power LED" },
> + { BATLEDG, GPIO, -1, "BATLEDG" },
> + { OEMLED0, GPIO, -1, "OEMLED0" },
> + { OEMLED1, GPIO, -1, "OEMLED1" },
> + { OEMLED2, GPIO, -1, "OEMLED2" },
> + { BATLEDR, GPIO, -1, "OEMLEDR" },
> + { SMARTBAT1, SMB, -1, "SmartBat1" },
> + { SMARTBAT2, SMB, -1, "SmartBat2" },
> + { CMOSBAT, ADC, 1, "VBat" },
> + { CMOSBATx2, ADC, 2, "VBat" },
> + { CMOSBATx10, ADC, 10, "VBat" },
> + { LIBAT, ADC, 1, "VBat2" },
> + { LIBATx2, ADC, 2, "VBat2" },
> + { LIBATx10, ADC, 10, "VBat2" },
> + { ADC5VS0, ADC, 1, "+5V" },
> + { ADC5VS0x2, ADC, 2, "+5V" },
> + { ADC5VS0x10, ADC, 10, "+5V" },
> + { ADC5VS5, ADC, 1, "+5V" },
> + { ADC5VS5x2, ADC, 2, "+5V" },
> + { ADC5VS5x10, ADC, 10, "+5V" },
> + { ADC33VS0, ADC, 1, "+3.3V" },
> + { ADC33VS0x2, ADC, 2, "+3.3V" },
> + { ADC33VS0x10, ADC, 10, "+3.3V" },
> + { ADC33VS5, ADC, 1, "+3.3V" },
> + { ADC33VS5x2, ADC, 2, "+3.3V" },
> + { ADC33VS5x10, ADC, 10, "+3.3V" },
> + { ADC12VS0, ADC, 1, "+12V" },
> + { ADC12VS0x2, ADC, 2, "+12V" },
> + { ADC12VS0x10, ADC, 10, "+12V" },
> + { VCOREA, ADC, 1, "VCore" },
> + { VCOREAx2, ADC, 2, "VCore" },
> + { VCOREAx10, ADC, 10, "VCore" },
> + { VCOREB, ADC, 1, "VCore2" },
> + { VCOREBx2, ADC, 2, "VCore2" },
> + { VCOREBx10, ADC, 10, "VCore2" },
> + { ADCDC, ADC, 1, "ADCDC" },
> + { ADCDCx2, ADC, 2, "ADCDCx2" },
> + { ADCDCx10, ADC, 10, "ADCDCx10" },
> + { VSTBY, ADC, 1, "Vsb" },
> + { VSTBYx2, ADC, 2, "Vsb" },
> + { VSTBYx10, ADC, 10, "Vsb" },
> + { VAUX, ADC, 1, "VAUX" },
> + { VAUXx2, ADC, 2, "VAUX" },
> + { VAUXx10, ADC, 10, "VAUX" },
> + { CURRENT, ADC, 1, "Imon" },
> + { WDIRQ, IRQ, -1, "WDIRQ" },
> + { WDNMI, GPIO, -1, "WDNMI" },
> + { TACHO0, TACH, -1, "Tacho1" },
> + { TACHO1, TACH, -1, "Tacho2" },
> + { TACHO2, TACH, -1, "Tacho3" },
> + { BRIGHTNESS2, PWM, -1, "Brightness2" },
> + { BACKLIGHT1, GPIO, -1, "Backlight1" },
> + { BACKLIGHT2, GPIO, -1, "Backlight2" },
> + { 0, 0, 0, "" }
> +};
> +
> +/**
> + * EC I/O
> + */
> +
> +static inline void imanager_delay(void)
> +{
> + udelay(EC_MICRO_DELAY);
> +}
Factoring it to inline does not help. Just use udelay() where you need to.
> +
> +static int wait_ibf_cleared(void)
> +{
> + int i = 0;
> +
> + do {
> + if (!(inb(IT8516_CMD_PORT) & BIT(1)))
Please define a developer-friendly name for BIT(1)
> + return 0;
> + imanager_delay();
> + } while (i++ < EC_MAX_RETRY);
> +
> + return -ETIME;
> +}
> +
> +static int wait_obf_set(void)
> +{
> + int i = 0;
> +
> + do {
> + if (inb(IT8516_CMD_PORT) & BIT(0))
ditto
> + return 0;
> + imanager_delay();
> + } while (i++ < EC_MAX_RETRY);
> +
> + return -ETIME;
> +}
> +
> +static inline int ec_inb(int addr, int reg)
> +{
> + outb(reg, addr);
> + return inb(addr + 1);
> +}
> +
> +static inline void ec_outb(int addr, int reg, int val)
> +{
> + outb(reg, addr);
> + outb(val, addr + 1);
> +}
> +
> +static inline int ec_io28_inb(int addr, int reg)
> +{
> + int ret;
> +
> + ret = wait_ibf_cleared();
> + if (ret)
> + return ret;
> +
> + /* clear data to prevent lock */
> + inb(addr - 1);
> +
> + outb(reg, addr);
> +
> + ret = wait_obf_set();
> + if (ret)
> + return ret;
> +
> + return inb(addr - 1);
> +}
> +
> +static inline int ec_io28_outb(int addr, int reg, int val)
> +{
> + int ret;
> +
> + ret = wait_ibf_cleared();
> + if (ret)
> + return ret;
> +
> + outb(reg, addr);
> +
> + ret = wait_ibf_cleared();
> + if (ret)
> + return ret;
> +
> + outb(val, addr - 1);
> +
> + return 0;
> +}
> +
> +static int ec_io18_read(int cmd)
> +{
> + return ec_inb(IT8518_CMD_PORT, cmd);
> +}
> +
> +static int ec_io18_write(int cmd, int value)
> +{
> + ec_outb(IT8518_CMD_PORT, cmd, value);
> +
> + return 0;
> +}
> +
> +static int ec_io28_read(int cmd)
> +{
> + return ec_io28_inb(IT8516_CMD_PORT, cmd + EC_CMD_OFFSET_READ);
> +}
> +
> +static int ec_io28_write(int cmd, int value)
> +{
> + return ec_io28_outb(IT8516_CMD_PORT, cmd + EC_CMD_OFFSET_WRITE, value);
> +}
> +
> +/* Prevent FW lock */
> +static void ec_clear_ports(void)
> +{
> + inb(IT8516_DAT_PORT);
> + inb(IT8518_DAT_PORT);
> +}
> +
> +static inline u16 ec_read_chipid(u16 addr)
> +{
> + return (ec_inb(addr, DEVID_REG_MSB) << 8 |
> + ec_inb(addr, DEVID_REG_LSB));
> +}
> +
> +static int ec_wait_cmd_clear(void)
> +{
> + int i = 0;
> +
> + do {
> + if (!ec.read(0))
> + return 0;
> + imanager_delay();
> + } while (i++ < EC_MAX_RETRY);
> +
> + pr_err("No respons from EC (timeout)\n");
> +
> + return -ETIME;
> +}
> +
> +static int ec_read_ram(u8 bank, u8 offset, u8 len, u8 *buf, u8 bufsz)
> +{
> + int i;
> + int ret;
> +
> + if (WARN_ON(!buf))
> + return -EINVAL;
> +
> + ret = ec_wait_cmd_clear();
> + if (ret)
> + return ret;
> +
> + ec.write(EC_MSG_OFFSET_PARAM, bank);
> + ec.write(EC_MSG_OFFSET_DATA(0), offset);
> + ec.write(EC_MSG_OFFSET_DATA(0x2C), len);
> + ec.write(EC_MSG_OFFSET_CMD, EC_CMD_RAM_RD);
> +
> + ret = ec_wait_cmd_clear();
> + if (ret)
> + return ret;
> +
> + ret = ec.read(EC_MSG_OFFSET_STATUS);
> + if (ret != EC_STATUS_SUCCESS)
> + return -EIO;
> +
> + for (i = 0; (i < len) && (len < EC_MSG_SIZE) && (len <= bufsz); i++)
> + buf[i] = ec.read(EC_MSG_OFFSET_DATA(i + 1));
> +
> + return 0;
> +}
> +
> +static int ec_write_ram(u8 bank, u8 offset, u8 len, u8 *buf)
> +{
> + int i;
> + int ret;
> +
> + if (WARN_ON(!buf))
> + return -EINVAL;
> +
> + ret = ec_wait_cmd_clear();
> + if (ret)
> + return ret;
> +
> + ec.write(EC_MSG_OFFSET_PARAM, bank);
> + ec.write(EC_MSG_OFFSET_DATA(0), offset);
> + ec.write(EC_MSG_OFFSET_DATA(0x2C), len);
> +
> + for (i = 0; (i < len) && (len < EC_MSG_SIZE); i++)
> + ec.write(EC_MSG_OFFSET_DATA(i + 1), buf[i]);
> +
> + ec.write(EC_MSG_OFFSET_CMD, EC_CMD_RAM_WR);
> +
> + ret = ec_wait_cmd_clear();
> + if (ret)
> + return ret;
> +
> + ret = ec.read(EC_MSG_OFFSET_STATUS);
> + if (ret != EC_STATUS_SUCCESS)
> + return -EIO;
> +
> + return 0;
> +}
> +
> +static int ec_read_dynamic_devtbl(struct imanager_data *ec)
> +{
> + u32 i, j;
> + int ret;
> + struct ec_message did = {
> + .rlen = EC_MAX_DID,
> + .wlen = 0,
> + };
> + struct ec_message hwp = {
> + .rlen = EC_MAX_DID,
> + .wlen = 0,
> + };
> + struct ec_message pol = {
> + .rlen = EC_MAX_DID,
> + .wlen = 0,
> + };
> + struct ec_dyn_devtbl *dyn;
> +
> + memset(ec->dyn, 0, sizeof(ec->dyn));
> +
> + ret = imanager_msg_read(EC_CMD_DYN_TBL_RD, EC_DYN_DID, &did);
> + if (ret)
> + return -EIO;
> +
> + ret = imanager_msg_read(EC_CMD_DYN_TBL_RD, EC_DYN_HWP, &hwp);
> + if (ret)
> + return -EIO;
> +
> + ret = imanager_msg_read(EC_CMD_DYN_TBL_RD, EC_DYN_POL, &pol);
> + if (ret)
> + return -EIO;
> +
> + for (i = 0; (i < EC_MAX_DID) && did.u.data[i]; i++) {
> + dyn = &ec->dyn[i];
> + for (j = 0; j < ARRAY_SIZE(devtbl); j++) {
> + if (devtbl[j].id == did.u.data[i]) {
> + dyn->did = did.u.data[i];
> + dyn->hwp = hwp.u.data[i];
> + dyn->pol = pol.u.data[i];
> + dyn->devtbl = &devtbl[j];
> + break;
> + }
> + }
> + }
> +
> + return 0;
> +}
> +
> +static int ec_read_buffer(u8 *data, int rlen)
> +{
> + int ret, i, j;
> + int pages = rlen % EC_I2C_BLOCK_SIZE;
> + int rem = rlen / EC_I2C_BLOCK_SIZE;
> +
> + /* pre-condition: rlen <= 256 */
> +
> + ret = ec_wait_cmd_clear();
> + if (ret)
> + return ret;
> +
> + for (i = 0; i < pages; i++) {
> + ec.write(EC_MSG_OFFSET_PARAM, i);
> + ec.write(EC_MSG_OFFSET_CMD, EC_CMD_BUF_RD);
> +
> + ret = ec_wait_cmd_clear();
> + if (ret)
> + return ret;
> +
> + ret = ec.read(EC_MSG_OFFSET_STATUS);
> + if (ret != EC_STATUS_SUCCESS)
> + return -EIO;
> +
> + for (j = 0; j < EC_I2C_BLOCK_SIZE; j++)
> + data[i * EC_I2C_BLOCK_SIZE + j] =
> + ec.read(EC_MSG_OFFSET_DATA(j));
> + }
> +
> + if (rem) {
> + ec.write(EC_MSG_OFFSET_PARAM, pages);
> + ec.write(EC_MSG_OFFSET_CMD, EC_CMD_BUF_RD);
> +
> + ret = ec_wait_cmd_clear();
> + if (ret)
> + return ret;
> +
> + ret = ec.read(EC_MSG_OFFSET_STATUS);
> + if (ret != EC_STATUS_SUCCESS)
> + return -EIO;
> +
> + for (j = 0; j < rem; j++)
> + data[pages * EC_I2C_BLOCK_SIZE + j] =
> + ec.read(EC_MSG_OFFSET_DATA(j));
> + }
> +
> + return 0;
> +}
> +
> +static int
> +imanager_msg_trans(u8 cmd, u8 param, struct ec_message *msg, bool payload)
> +{
> + int ret, i, len;
> + u32 offset;
> +
> + ret = ec_wait_cmd_clear();
> + if (ret)
> + return ret;
> +
> + ec.write(EC_MSG_OFFSET_PARAM, param);
> +
> + if (msg && msg->wlen) {
> + if (!msg->data) {
> + for (i = 0; i < msg->wlen; i++)
> + ec.write(EC_MSG_OFFSET_DATA(i),
> + msg->u.data[i]);
> + } else {
> + for (i = 0; i < msg->wlen; i++)
> + ec.write(EC_MSG_OFFSET_DATA(i), msg->data[i]);
> + ec.write(EC_MSG_OFFSET_DATA(0x2c), msg->wlen);
> + }
> + }
> +
> + ec.write(EC_MSG_OFFSET_CMD, cmd);
> + ret = ec_wait_cmd_clear();
> + if (ret)
> + return ret;
> +
> + /* GPIO and I2C have different success return values */
> + ret = ec.read(EC_MSG_OFFSET_STATUS);
> + if ((ret != EC_STATUS_SUCCESS) && !(ret & EC_STATUS_CMD_COMPLETE))
> + return -EIO;
> + /*
> + * EC I2C may return an error code which we need to hand-off
> + * to the caller
> + */
> + else if (ret & 0x07e)
> + return ret;
> +
> + if (msg && msg->data) {
> + ret = ec_read_buffer(msg->data, msg->rlen);
> + if (ret < 0)
> + return ret;
> + } else if (msg && msg->rlen) {
> + if (msg->rlen == 0xff)
> + /* Use alternate message body for hwmon */
> + len = ec.read(EC_MSG_OFFSET_DATA(0x2C));
> + else
> + len = (msg->rlen > EC_MSG_SIZE ? EC_MSG_SIZE :
> + msg->rlen);
> + offset = payload ? EC_MSG_OFFSET_PAYLOAD(0) :
> + EC_MSG_OFFSET_DATA(0);
> + for (i = 0; i < len; i++)
> + msg->u.data[i] = ec.read(offset + i);
> + }
> +
> + return 0;
> +}
> +
> +int imanager_msg_read(u8 cmd, u8 param, struct ec_message *msg)
> +{
> + return imanager_msg_trans(cmd, param, msg, false);
> +}
> +EXPORT_SYMBOL_GPL(imanager_msg_read);
> +
> +int imanager_msg_write(u8 cmd, u8 param, struct ec_message *msg)
> +{
> + return imanager_msg_trans(cmd, param, msg, true);
> +}
> +EXPORT_SYMBOL_GPL(imanager_msg_write);
> +
> +int imanager_read_byte(u8 cmd, u8 param)
> +{
> + int ret;
> + struct ec_message msg = {
> + .rlen = 1,
> + .wlen = 0,
> + };
> +
> + ret = imanager_msg_read(cmd, param, &msg);
> + if (ret)
> + return ret;
> +
> + return msg.u.data[0];
> +}
> +EXPORT_SYMBOL_GPL(imanager_read_byte);
> +
> +int imanager_read_word(u8 cmd, u8 param)
> +{
> + int ret;
> + struct ec_message msg = {
> + .rlen = 2,
> + .wlen = 0,
> + };
> +
> + ret = imanager_msg_read(cmd, param, &msg);
> + if (ret)
> + return ret;
> +
> + return (msg.u.data[0] << 8 | msg.u.data[1]);
> +}
> +EXPORT_SYMBOL_GPL(imanager_read_word);
> +
> +int imanager_write_byte(u8 cmd, u8 param, u8 byte)
> +{
> + struct ec_message msg = {
> + .rlen = 0,
> + .wlen = 1,
> + .u = {
> + .data = { byte, 0 },
> + },
> + };
> +
> + return imanager_msg_write(cmd, param, &msg);
> +}
> +EXPORT_SYMBOL_GPL(imanager_write_byte);
> +
> +int imanager_write_word(u8 cmd, u8 param, u16 word)
> +{
> + struct ec_message msg = {
> + .rlen = 0,
> + .wlen = 2,
> + .u = {
> + .data = { HIBYTE16(word), LOBYTE16(word), 0 },
> + },
> + };
> +
> + return imanager_msg_write(cmd, param, &msg);
> +}
> +EXPORT_SYMBOL_GPL(imanager_write_word);
> +
> +static int ec_hwram_read_byte(u8 offset)
> +{
> + int ret;
> + u8 val;
> +
> + ret = ec_read_ram(EC_RAM_HW, offset, sizeof(val), &val, sizeof(val));
> + if (ret < 0) {
> + pr_err("Failed to read from HWRAM @ 0x%02X\n", offset);
Here and in other places - this should be dev_err.
> + return ret;
> + }
> +
> + return val;
> +}
> +
> +int imanager_acpiram_read_byte(u8 offset)
> +{
> + int ret;
> + u8 value;
> +
> + ret = ec_read_ram(EC_RAM_ACPI, offset, sizeof(value), (u8 *)&value,
> + sizeof(value));
> + if (ret < 0) {
> + pr_err("Failed to read from ACPI RAM @ 0x%02X\n", offset);
> + return ret;
> + }
> +
> + return value;
> +}
> +EXPORT_SYMBOL_GPL(imanager_acpiram_read_byte);
Here and in rest of file: I am not sure whether mfd tree is suitable
for this. I assume you want to share it between all of the drivers as
a part of common transport layer?
> +
> +int imanager_acpiram_write_byte(u8 offset, u8 value)
> +{
> + int ret;
> +
> + ret = ec_write_ram(EC_RAM_ACPI, offset, sizeof(value), &value);
> + if (ret) {
> + pr_err("Failed to write to ACPI RAM @ 0x%02X\n", offset);
> + return ret;
> + }
> +
> + return 0;
> +}
> +EXPORT_SYMBOL_GPL(imanager_acpiram_write_byte);
> +
> +int imanager_acpiram_read_block(u8 offset, u8 *buf, u8 len)
> +{
> + int ret;
> +
> + ret = ec_read_ram(EC_RAM_ACPI, offset, len, buf, len);
> + if (ret < 0) {
> + pr_err("Failed to read from ACPI RAM @ 0x%02X\n", offset);
> + return ret;
> + }
> +
> + return 0;
> +}
> +EXPORT_SYMBOL_GPL(imanager_acpiram_read_block);
> +
> +int imanager_acpiram_write_block(u8 offset, u8 *buf, u8 len)
> +{
> + int ret;
> +
> + ret = ec_write_ram(EC_RAM_ACPI, offset, len, buf);
> + if (ret) {
> + pr_err("Failed to write to ACPI RAM @ 0x%02X\n", offset);
> + return ret;
> + }
> +
> + return 0;
> +}
> +EXPORT_SYMBOL_GPL(imanager_acpiram_write_block);
> +
> +int imanager_wait_proc_complete(u8 offset, int cond)
> +{
> + int ret, i;
> +
> + for (i = 0; i < EC_MAX_RETRY; i++) {
> + ret = ec_hwram_read_byte(offset);
> + if (ret < 0)
> + return ret;
> +
> + if (ret == cond)
> + return 0;
> +
> + imanager_delay();
> + }
> +
> + return -EIO;
> +}
> +EXPORT_SYMBOL_GPL(imanager_wait_proc_complete);
> +
> +static inline void ec_get_dev_attr(struct ec_dev_attr *attr,
> + const struct ec_dyn_devtbl *tbl)
> +{
> + attr->did = tbl->did;
> + attr->hwp = tbl->hwp;
> + attr->pol = tbl->pol;
> + attr->scale = tbl->devtbl->scale;
> + attr->label = tbl->devtbl->label;
> +}
> +
> +static int ec_get_dev_gpio(struct imanager_data *ec)
> +{
> + size_t i;
> + struct ec_dyn_devtbl *dyn;
> + struct imanager_gpio_device *gpio = &ec->gpio;
> +
> + for (i = 0; i < ARRAY_SIZE(ec->dyn); i++) {
> + dyn = &ec->dyn[i];
> + if (dyn->did && (dyn->devtbl->type == GPIO)) {
> + switch (dyn->did) {
> + case ALTGPIO0:
> + ec_get_dev_attr(&gpio->attr[0], dyn);
> + gpio->num++;
> + break;
> + case ALTGPIO1:
> + ec_get_dev_attr(&gpio->attr[1], dyn);
> + gpio->num++;
> + break;
> + case ALTGPIO2:
> + ec_get_dev_attr(&gpio->attr[2], dyn);
> + gpio->num++;
> + break;
> + case ALTGPIO3:
> + ec_get_dev_attr(&gpio->attr[3], dyn);
> + gpio->num++;
> + break;
> + case ALTGPIO4:
> + ec_get_dev_attr(&gpio->attr[4], dyn);
> + gpio->num++;
> + break;
> + case ALTGPIO5:
> + ec_get_dev_attr(&gpio->attr[5], dyn);
> + gpio->num++;
> + break;
> + case ALTGPIO6:
> + ec_get_dev_attr(&gpio->attr[6], dyn);
> + gpio->num++;
> + break;
> + case ALTGPIO7:
> + ec_get_dev_attr(&gpio->attr[7], dyn);
> + gpio->num++;
> + break;
> + case BUTTON0:
> + case BUTTON1:
> + case BUTTON2:
> + case BUTTON3:
> + case BUTTON4:
> + case BUTTON5:
> + case BUTTON6:
> + case BUTTON7:
> + case POWERLED:
> + case BATLEDG:
> + case OEMLED0:
> + case OEMLED1:
> + case OEMLED2:
> + case BATLEDR:
> + case WDNMI:
> + case BACKLIGHT1:
> + case BACKLIGHT2:
> + break;
> + default:
> + pr_err("DID 0x%02X not handled\n", dyn->did);
> + return -EINVAL;
> + }
> + }
> + }
> +
> + gpio->info = &ec->dev.info;
> +
> + return 0;
> +}
> +
> +static int ec_get_dev_adc(struct imanager_data *ec)
Naming of this and other similar functions is confusing to me. It
suggests that some device will be GET and returned. If not returned
directly then put under argument (like **). But it behaves completely
differently.
> +{
> + size_t i;
> + struct ec_dyn_devtbl *dyn;
> + struct dev_adc *adc = &ec->sensors.adc;
> +
> + for (i = 0; i < ARRAY_SIZE(ec->dyn); i++) {
> + dyn = &ec->dyn[i];
> + if (dyn->did && (dyn->devtbl->type == ADC)) {
> + switch (dyn->did) {
> + case ADC12VS0:
> + case ADC12VS0x2:
> + case ADC12VS0x10:
> + ec_get_dev_attr(&adc->attr[0], dyn);
> + adc->num++;
> + break;
> + case ADC5VS5:
> + case ADC5VS5x2:
> + case ADC5VS5x10:
> + ec_get_dev_attr(&adc->attr[1], dyn);
> + adc->num++;
> + break;
> + case CMOSBAT:
> + case CMOSBATx2:
> + case CMOSBATx10:
> + ec_get_dev_attr(&adc->attr[2], dyn);
> + adc->num++;
> + break;
> + case VCOREA:
> + case ADC5VS0:
> + case ADC5VS0x2:
> + case ADC5VS0x10:
> + ec_get_dev_attr(&adc->attr[3], dyn);
> + adc->num++;
> + break;
> + case CURRENT:
> + case ADC33VS0:
> + case ADC33VS0x2:
> + case ADC33VS0x10:
> + ec_get_dev_attr(&adc->attr[4], dyn);
> + adc->num++;
> + break;
> + default:
> + pr_err("DID 0x%02X not handled\n", dyn->did);
dev_err()... but is it really an error? Critical error? Maybe dev_info
or dev_warn?
> + return -EINVAL;
> + }
> + }
> + }
> +
> + return 0;
> +}
> +
> +static int ec_get_dev_fan(struct imanager_data *ec)
> +{
> + size_t i;
> + struct ec_dyn_devtbl *dyn;
> + struct dev_fan *fan = &ec->sensors.fan;
> +
> + for (i = 0; i < ARRAY_SIZE(ec->dyn); i++) {
> + dyn = &ec->dyn[i];
> + if (dyn->did && ((dyn->devtbl->type == TACH) ||
> + (dyn->devtbl->type == PWM))) {
> + switch (dyn->did) {
> + case CPUFAN_2P:
> + case CPUFAN_4P:
> + ec_get_dev_attr(&fan->attr[0], dyn);
> + fan->num++;
> + fan->active |= 1 << 0;
> + break;
> + case SYSFAN1_2P:
> + case SYSFAN1_4P:
> + ec_get_dev_attr(&fan->attr[1], dyn);
> + fan->num++;
> + fan->active |= 1 << 1;
> + break;
> + case SYSFAN2_2P:
> + case SYSFAN2_4P:
> + ec_get_dev_attr(&fan->attr[2], dyn);
> + fan->num++;
> + fan->active |= 1 << 2;
> + break;
> + case TACHO0:
> + case TACHO1:
> + case TACHO2:
> + case BRIGHTNESS:
> + case BRIGHTNESS2:
> + case PCBEEP:
> + break;
> + default:
> + pr_err("DID 0x%02X not handled\n", dyn->did);
ditto
> + return -EINVAL;
> + }
> + }
> + }
> +
> + return 0;
> +}
> +
> +static int ec_get_dev_hwmon(struct imanager_data *ec)
> +{
> + int ret;
> +
> + ret = ec_get_dev_adc(ec);
> + if (ret < 0)
> + return ret;
> +
> + ret = ec_get_dev_fan(ec);
> + if (ret < 0)
> + return ret;
> +
> + ec->sensors.ecdev = &ec->dev;
> +
> + return 0;
> +}
> +
> +static int ec_get_dev_i2c(struct imanager_data *ec)
> +{
> + size_t i;
> + struct ec_dyn_devtbl *dyn;
> + struct imanager_i2c_device *i2c = &ec->i2c;
> +
> + for (i = 0; i < ARRAY_SIZE(ec->dyn); i++) {
> + dyn = &ec->dyn[i];
> + if (dyn->did && (dyn->devtbl->type == SMB)) {
> + switch (dyn->did) {
> + case SMBEEPROM:
> + ec_get_dev_attr(&i2c->attr[0], dyn);
> + i2c->eeprom = &i2c->attr[0];
> + i2c->num++;
> + break;
> + case I2COEM:
> + ec_get_dev_attr(&i2c->attr[1], dyn);
> + i2c->i2coem = &i2c->attr[1];
> + i2c->num++;
> + break;
> + case SMBOEM0:
> + case SMBOEM1:
> + case SMBOEM2:
> + case SMBTHERMAL0:
> + case SMBTHERMAL1:
> + case SMBSECEEP:
> + case SMBEEP2K:
> + case OEMEEP:
> + case OEMEEP2K:
> + case PECI:
> + case SMBOEM3:
> + case SMLINK:
> + case SMBSLV:
> + case SMARTBAT1:
> + case SMARTBAT2:
> + break;
> + default:
> + pr_err("DID 0x%02X not handled\n", dyn->did);
> + return -EINVAL;
> + }
> + }
> + }
> +
> + i2c->ecdev = &ec->dev;
> +
> + return 0;
> +}
> +
> +static int ec_get_dev_blc(struct imanager_data *ec)
> +{
> + size_t i;
> + struct ec_dyn_devtbl *dyn;
> + struct imanager_backlight_device *blc = &ec->blc;
> +
> + for (i = 0; i < ARRAY_SIZE(ec->dyn); i++) {
> + dyn = &ec->dyn[i];
> + if (dyn->did && (dyn->devtbl->type == PWM)) {
> + switch (dyn->did) {
> + case BRIGHTNESS:
> + ec_get_dev_attr(&blc->attr[0], dyn);
> + blc->brightness[0] = EC_ACPIRAM_BRIGHTNESS1;
> + blc->num++;
> + break;
> + case BRIGHTNESS2:
> + ec_get_dev_attr(&blc->attr[1], dyn);
> + blc->brightness[1] = EC_ACPIRAM_BRIGHTNESS2;
> + blc->num++;
> + break;
> + case CPUFAN_2P:
> + case CPUFAN_4P:
> + case SYSFAN1_2P:
> + case SYSFAN1_4P:
> + case SYSFAN2_2P:
> + case SYSFAN2_4P:
> + case PCBEEP:
> + case TACHO0:
> + case TACHO1:
> + case TACHO2:
> + break;
> + default:
> + pr_err("DID 0x%02X not handled\n", dyn->did);
> + return -EINVAL;
> + }
> + }
> + }
> +
> + blc->info = &ec->dev.info;
> +
> + return 0;
> +}
> +
> +static int ec_get_dev_wdt(struct imanager_data *ec)
> +{
> + size_t i;
> + struct ec_dyn_devtbl *dyn;
> + struct imanager_watchdog_device *wdt = &ec->wdt;
> +
> + for (i = 0; i < ARRAY_SIZE(ec->dyn); i++) {
> + dyn = &ec->dyn[i];
> + if (dyn->did && (dyn->devtbl->type == IRQ)) {
> + switch (dyn->did) {
> + case WDIRQ:
> + ec_get_dev_attr(&wdt->attr[0], dyn);
> + wdt->irq = &wdt->attr[0];
> + wdt->num++;
> + break;
> + case WDNMI:
> + ec_get_dev_attr(&wdt->attr[1], dyn);
> + wdt->nmi = &wdt->attr[1];
> + wdt->num++;
> + break;
> + default:
> + pr_err("DID 0x%02X not handled\n", dyn->did);
> + return -EINVAL;
> + }
> + }
> + }
> +
> + wdt->info = &ec->dev.info;
> +
> + return 0;
> +}
> +
> +const struct imanager_ec_device *imanager_get_ec_device(void)
> +{
> + return &ec.dev;
> +};
> +EXPORT_SYMBOL_GPL(imanager_get_ec_device);
> +
> +const struct imanager_hwmon_device *imanager_get_hwmon_device(void)
> +{
> + return &ec.sensors;
> +}
> +EXPORT_SYMBOL_GPL(imanager_get_hwmon_device);
> +
> +const struct imanager_gpio_device *imanager_get_gpio_device(void)
> +{
> + return &ec.gpio;
> +}
> +EXPORT_SYMBOL_GPL(imanager_get_gpio_device);
> +
> +const struct imanager_i2c_device *imanager_get_i2c_device(void)
> +{
> + return &ec.i2c;
> +}
> +EXPORT_SYMBOL_GPL(imanager_get_i2c_device);
> +
> +const struct imanager_backlight_device *imanager_get_backlight_device(void)
> +{
> + return &ec.blc;
> +}
> +EXPORT_SYMBOL_GPL(imanager_get_backlight_device);
> +
> +const struct imanager_watchdog_device *imanager_get_watchdog_device(void)
> +{
> + return &ec.wdt;
> +}
> +EXPORT_SYMBOL_GPL(imanager_get_watchdog_device);
1. All of these helpers should have rather a format of:
to_blablabla(pointer) {
container_of(...)
}
Don't use the globals, don't rely on them. If you have to rely, then why?
2. You are mixing the order of declarations - static, export, static.
Please be consistent. First static, last exported.
3. Do you really need all of these exported functions? I didn't check
rest of patches so I am just guessing - all of them are used?
> +
> +static int ec_get_version(struct ec_version *version)
> +{
> + int ret;
> + u16 raw;
> + struct ec_version_raw ver;
> +
> + if (WARN_ON(!version))
> + return -EINVAL;
> +
> + ret = ec_read_ram(EC_RAM_ACPI, EC_ACPIRAM_FW_RELEASE_RD, sizeof(ver),
> + (u8 *)&ver, sizeof(ver));
> + if (ret < 0)
> + return ret;
> +
> + raw = swab16(ver.kernel);
> + version->kernel_major = EC_KERNEL_MAJOR(raw);
> + version->kernel_minor = EC_KERNEL_MINOR(raw);
> +
> + raw = swab16(ver.firmware);
> + version->firmware_major = EC_FIRMWARE_MAJOR(raw);
> + version->firmware_minor = EC_FIRMWARE_MINOR(raw);
> +
> + raw = swab16(ver.project_code);
> + version->type = EC_PROJECT_CODE(raw);
> +
> + return 0;
> +}
> +
> +static int ec_get_pcb_name(struct ec_info *info)
> +{
> + int ret;
> + struct ec_message msg = {
> + .rlen = ARRAY_SIZE(info->pcb_name),
> + .wlen = 0,
> + };
> +
> + if (WARN_ON(!info))
> + return -EINVAL;
So this can be called with NULL? Or cannot? If cannot, then no warn,
just return... but this is a static function so it has only local
users which means that you control the users. How can the info be NULL
then?
> +
> + ret = imanager_msg_read(EC_CMD_FW_INFO_RD, 0, &msg);
> + if (ret)
> + return ret;
> +
> + /*
> + * Sadly, the string is not Null-terminated so we will need to read a
> + * fixed amount of chars. There is, apparently, no exact definition
> + * of board name (SOM6867 vs. MIO-5271).
> + */
> + memset(info->pcb_name, 0, ARRAY_SIZE(info->pcb_name));
Isn't info allocated with kzalloc()? Why you need memset?
> + strncpy(info->pcb_name, (const char *)msg.u.data, PCB_NAME_MAX_SIZE);
> +
> + if (strchr(info->pcb_name, '-') == NULL)
> + info->pcb_name[PCB_NAME_MAX_SIZE - 1] = '\0';
If there is no '-' and contents exceeded PCB_NAME_MAX_SIZE then no
need of NULL-terminating?
> +
> + return 0;
> +}
> +
> +static int ec_get_fw_info(struct ec_info *info)
> +{
> + int ret;
> +
> + if (WARN_ON(!info))
> + return -EINVAL;
The same weird pattern. Just no.
> +
> + ret = ec_get_version(&info->version);
> + if (ret)
> + return ret;
> +
> + return ec_get_pcb_name(info);
> +}
> +
> +static int ec_init(void)
init not marked as __init looks a little weird. It also does not
accept any arguments so again you depend on some magic
globals/file-scope variables.
> +{
> + int ret;
> +
> + ec_clear_ports();
> +
> + ret = ec_read_dynamic_devtbl(&ec);
> + if (ret)
> + return ret;
> +
> + ret = ec_get_fw_info(&ec.dev.info);
> + if (ret < 0)
> + return ret;
> +
> + ret = ec_get_dev_gpio(&ec);
> + if (ret < 0)
> + return ret;
> +
> + ret = ec_get_dev_hwmon(&ec);
> + if (ret < 0)
> + return ret;
> +
> + ret = ec_get_dev_i2c(&ec);
> + if (ret < 0)
> + return ret;
> +
> + ret = ec_get_dev_blc(&ec);
> + if (ret < 0)
> + return ret;
> +
> + ret = ec_get_dev_wdt(&ec);
> + if (ret < 0)
> + return ret;
> +
> + return 0;
> +}
> +
> +int imanager_ec_probe(void)
This is not a probe. Or it is seriously broken for a probe. :)
> +{
> + int chipid = ec_read_chipid(EC_BASE_ADDR);
> +
> + memset((void *)&ec, 0, sizeof(ec));
Why? The driver should dynamically allocate its state during a probe
(real probe) and use kzalloc so no memset needed.
> +
> + switch (chipid) {
> + case EC_DEVID_IT8516:
> + pr_err("EC IT8516 not supported\n");
dev_err.
The fact that you cannot refer to a device here is also interesting...
What is the context for this function?
> + ec.dev.id = IT8516;
> + return -ENODEV;
> + case EC_DEVID_IT8518:
> + ec.read = ec_io18_read;
> + ec.write = ec_io18_write;
> + ec.dev.id = IT8518;
> + break;
> + case EC_DEVID_IT8528:
> + ec.read = ec_io28_read;
> + ec.write = ec_io28_write;
> + ec.dev.id = IT8528;
> + break;
> + default:
> + return -ENODEV;
> + }
> +
> + ec.dev.addr = EC_BASE_ADDR;
> +
> + return ec_init();
> +}
> +
> diff --git a/include/linux/mfd/imanager/core.h b/include/linux/mfd/imanager/core.h
> new file mode 100644
> index 0000000..05c77fa
> --- /dev/null
> +++ b/include/linux/mfd/imanager/core.h
> @@ -0,0 +1,31 @@
> +/*
> + * Advantech iManager MFD core
> + *
> + * Copyright (C) 2016 Advantech Co., Ltd., Irvine, CA, USA
> + * Author: Richard Vidal-Dorsch <richard.dorsch@xxxxxxxxxxxxx>
> + *
> + * This program is free software; you can redistribute it and/or modify it
> + * under the terms of the GNU General Public License as published by the
> + * Free Software Foundation; either version 2 of the License, or (at your
> + * option) any later version.
> + */
> +
> +#ifndef __CORE_H__
> +#define __CORE_H__
Bigger protection - like __LINUX_MFD_IMANAGER_CORE_H__ (probably LINUX
is too much... but anyway...).
> +
> +#include <linux/mutex.h>
> +#include <linux/io.h>
> +#include <linux/types.h>
Why do you need io.h and types.h?
> +#include <linux/mfd/imanager/ec.h>
> +
> +struct imanager_platform_data {
> + const struct imanager_ec_device *dev;
> + const char *chip_name;
> +};
> +
> +struct imanager_device_data {
> + struct device *dev;
> + struct mutex lock;
> +};
> +
> +#endif
> diff --git a/include/linux/mfd/imanager/ec.h b/include/linux/mfd/imanager/ec.h
> new file mode 100644
> index 0000000..783f268
> --- /dev/null
> +++ b/include/linux/mfd/imanager/ec.h
> @@ -0,0 +1,210 @@
> +/*
> + * Advantech iManager core - firmware interface
> + *
> + * Copyright (C) 2016 Advantech Co., Ltd., Irvine, CA, USA
> + * Author: Richard Vidal-Dorsch <richard.dorsch@xxxxxxxxxxxxx>
> + *
> + * This program is free software; you can redistribute it and/or modify it
> + * under the terms of the GNU General Public License as published by the
> + * Free Software Foundation; either version 2 of the License, or (at your
> + * option) any later version.
> + */
> +
> +#ifndef __EC_H__
> +#define __EC_H__
> +
> +#include <linux/types.h>
> +
> +#define EC_DEVID_IT8516 0x8516
> +#define EC_DEVID_IT8518 0x8518
> +#define EC_DEVID_IT8528 0x8528
> +
> +#define EC_BASE_ADDR 0x029C
> +
> +#define IT8516_CMD_PORT 0x029A /* it8528 */
> +#define IT8516_DAT_PORT 0x0299
> +
> +#define IT8518_CMD_PORT 0x029E /* it8518 */
> +#define IT8518_DAT_PORT 0x029F
> +
> +#define EC_GPIO_MAX_NUM 8
> +#define EC_HWM_MAX_ADC 5
> +#define EC_HWM_MAX_FAN 3
> +#define EC_BLC_MAX_NUM 2
> +#define EC_SMB_MAX_NUM 4
> +#define EC_WDT_MAX_NUM 2
> +
> +#define PCB_NAME_SIZE 32
> +#define EC_PAYLOAD_SIZE 40
> +#define EC_MSG_SIZE sizeof(struct ec_smb_message)
> +
> +#define LOBYTE16(x) (x & 0x00FF)
> +#define HIBYTE16(x) (LOBYTE16(x >> 8))
> +#define LOADDR16(x) LOBYTE16(x)
> +#define HIADDR16(x) (x >= 0xF000 ? LOBYTE16(x >> 8) : 0)
> +
> +/*
> + * iManager commands
> + */
> +#define EC_CMD_HWP_RD 0x11UL
> +#define EC_CMD_HWP_WR 0x12UL
> +#define EC_CMD_GPIO_DIR_RD 0x30UL
> +#define EC_CMD_GPIO_DIR_WR 0x31UL
> +#define EC_CMD_PWM_FREQ_RD 0x36UL
> +#define EC_CMD_PWM_FREQ_WR 0x32UL
> +#define EC_CMD_PWM_POL_RD 0x37UL
> +#define EC_CMD_PWM_POL_WR 0x33UL
> +#define EC_CMD_SMB_FREQ_RD 0x34UL
> +#define EC_CMD_SMB_FREQ_WR 0x35UL
> +#define EC_CMD_FAN_CTL_RD 0x40UL
> +#define EC_CMD_FAN_CTL_WR 0x41UL
> +#define EC_CMD_THZ_RD 0x42UL
> +#define EC_CMD_DYN_TBL_RD 0x20UL
> +#define EC_CMD_FW_INFO_RD 0xF0UL
> +#define EC_CMD_BUF_CLR 0xC0UL
> +#define EC_CMD_BUF_RD 0xC1UL
> +#define EC_CMD_BUF_WR 0xC2UL
> +#define EC_CMD_RAM_RD 0x1EUL
> +#define EC_CMD_RAM_WR 0x1FUL
> +#define EC_CMD_I2C_RW 0x0EUL
> +#define EC_CMD_I2C_WR 0x0FUL
> +#define EC_CMD_WDT_CTRL 0x28UL
> +
> +/*
> + * ACPI and HW RAM offsets
> + */
> +#define EC_ACPIRAM_FAN_ALERT 0x6FUL
> +#define EC_ACPIRAM_FAN_SPEED_LIMIT 0x76UL
> +#define EC_ACPIRAM_BRIGHTNESS1 0x50UL
> +#define EC_ACPIRAM_BRIGHTNESS2 0x52UL
> +#define EC_ACPIRAM_BLC_CTRL 0x99UL
> +#define EC_ACPIRAM_FW_RELEASE_RD 0xF8UL
> +
> +enum chips { IT8516, IT8518, IT8528 };
Is the order anyhow related to order of chip_names array?
Best regards,
Krzysztof
--
To unsubscribe from this list: send the line "unsubscribe linux-i2c" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
