Skip to contentSkip to conversation

mfd: intel_quark_i2c_gpio: Add Intel Quark X1000 I2C-GPIO MFD Driver

PATCH Superseded Raymond Tan 4 participants 14 messages Dec 11, 2014 – Jan 26, 2015 Multifunction devices (MFD)

v3 of 4 Superseded Latest is v4 →

lore
Share & tools
lore
#0 lore raw

v3 cover mfd: intel_quark_i2c_gpio: Add Intel Quark X1000 I2C-GPIO MFD Driver

From: "Tan, Raymond" <[email protected]>

Hi,
This patch is for enabling support of Intel Quark X1000 I2C controller and GPIO controller. In Quark X1000, the platform exports a single PCI device with both I2C and GPIO functions.

This MFD driver will split the 2 devices for their respective drivers.

---
Changes in v3:
* Simplify the call path for both gpio and i2c controllers initialization
* Move the global variables clk and clk_lookup into a struct

Changes in v2:
* Fix possible memory leak

Raymond Tan (1):
mfd: intel_quark_i2c_gpio: Add Intel Quark X1000 I2C-GPIO MFD Driver

 drivers/mfd/Kconfig                |   12 ++
 drivers/mfd/Makefile               |    1 +
 drivers/mfd/intel_quark_i2c_gpio.c |  279 ++++++++++++++++++++++++++++++++++++
 3 files changed, 292 insertions(+)
 create mode 100644 drivers/mfd/intel_quark_i2c_gpio.c
Signature · 1 line
1.7.9.5
#1 lore raw

PATCH v3 1/1 mfd: intel_quark_i2c_gpio: Add Intel Quark X1000 I2C-GPIO MFD Driver

In Quark X1000, there's a single PCI device that provides both an I2C controller and a GPIO controller. This MFD driver will split the 2 devices for their respective drivers.

This patch is based on Josef Ahmad's initial work for Quark enabling.

---
 drivers/mfd/Kconfig                |   12 ++
 drivers/mfd/Makefile               |    1 +
 drivers/mfd/intel_quark_i2c_gpio.c |  279 ++++++++++++++++++++++++++++++++++++
 3 files changed, 292 insertions(+)
 create mode 100644 drivers/mfd/intel_quark_i2c_gpio.c
drivers/mfd/Kconfig +12 -0
index b7c74a7..2f2f62d 100644
@@ -219,6 +219,18 @@ config HTC_I2CPLD
 	  This device provides input and output GPIOs through an I2C
 	  interface to one or more sub-chips.
 
+config MFD_INTEL_QUARK_I2C_GPIO
+	tristate "Intel Quark MFD I2C GPIO"
+	depends on PCI
+	depends on X86
+	depends on COMMON_CLK
+	select MFD_CORE
+	help
+	  This MFD provides support for I2C and GPIO that exist only
+	  in a single PCI device. It splits the 2 IO devices to
+	  their respective IO driver.
+	  The GPIO exports a total amount of 8 interrupt-capable GPIOs.
+
 config LPC_ICH
 	tristate "Intel ICH LPC"
 	depends on PCI
drivers/mfd/Makefile +1 -0
index 8a28dc9..d42652d 100644
@@ -133,6 +133,7 @@ obj-$(CONFIG_AB8500_CORE)	+= ab8500-core.o ab8500-sysctrl.o
 obj-$(CONFIG_MFD_TIMBERDALE)    += timberdale.o
 obj-$(CONFIG_PMIC_ADP5520)	+= adp5520.o
 obj-$(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
 obj-$(CONFIG_LPC_ICH)		+= lpc_ich.o
 obj-$(CONFIG_MFD_RDC321X)	+= rdc321x-southbridge.o
drivers/mfd/intel_quark_i2c_gpio.c new file +279 -0

279 lines not on this page · raw message

Signature · 1 line
1.7.9.5
#2 lore raw
Quoted 5 lines · Raymond Tan

In Quark X1000, there's a single PCI device that provides both an I2C controller and a GPIO controller. This MFD driver will split the 2 devices for their respective drivers.

This patch is based on Josef Ahmad's initial work for Quark enabling.

See my comments below.

Quoted 8 lines
---
 drivers/mfd/Kconfig                |   12 ++
 drivers/mfd/Makefile               |    1 +
 drivers/mfd/intel_quark_i2c_gpio.c |  279 ++++++++++++++++++++++++++++++++++++

So, there is a proposal to bring the order into the source tree a bit for Intel related MFD drivers.

What about to move this module to
drivers/mfd/intel/quark/i2c_gpio.c ?

At least what I can tell now, it will come more Intel stuff and we have few drivers already there.

I can see the following:
drivers/mfd/intel/
-> quark/
-> mid/ (Intel MID platforms, existing drivers)
-> some new platform/ (whatever is the one of the new Intel SoCs)

Lee, what's your opinion?

I could change a layout for existing in-tree drivers.

drivers/mfd/Kconfig config HTC_I2CPLD · line 222

Quoted 12 lines
 3 files changed, 292 insertions(+)
 create mode 100644 drivers/mfd/intel_quark_i2c_gpio.c
drivers/mfd/Kconfig +1 -0
index b7c74a7..2f2f62d 100644
@@ -219,6 +219,18 @@ config HTC_I2CPLD
 	  This device provides input and output GPIOs through an I2C
 	  interface to one or more sub-chips.
 
+config MFD_INTEL_QUARK_I2C_GPIO

Depending on the further discussion about layout.

Quoted 135 lines
drivers/mfd/Kconfig +11 -0
+	tristate "Intel Quark MFD I2C GPIO"
+	depends on PCI
+	depends on X86
+	depends on COMMON_CLK
+	select MFD_CORE
+	help
+	  This MFD provides support for I2C and GPIO that exist only
+	  in a single PCI device. It splits the 2 IO devices to
+	  their respective IO driver.
+	  The GPIO exports a total amount of 8 interrupt-capable GPIOs.
+
 config LPC_ICH
 	tristate "Intel ICH LPC"
 	depends on PCI
drivers/mfd/Makefile +1 -0
index 8a28dc9..d42652d 100644
@@ -133,6 +133,7 @@ obj-$(CONFIG_AB8500_CORE)	+= ab8500-core.o ab8500-sysctrl.o
 obj-$(CONFIG_MFD_TIMBERDALE)    += timberdale.o
 obj-$(CONFIG_PMIC_ADP5520)	+= adp5520.o
 obj-$(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
 obj-$(CONFIG_LPC_ICH)		+= lpc_ich.o
 obj-$(CONFIG_MFD_RDC321X)	+= rdc321x-southbridge.o
drivers/mfd/intel_quark_i2c_gpio.c new file +103 -0
new file mode 100644
index 0000000..a3a7043
@@ -0,0 +1,279 @@
+/*
+ * Intel Quark MFD PCI driver for I2C & GPIO
+ *
+ * Copyright(c) 2014 Intel Corporation.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * Intel Quark PCI device for I2C and GPIO controller sharing the same
+ * PCI function. This PCI driver will split the 2 devices into their
+ * respective drivers.
+ */
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/pci.h>
+#include <linux/mfd/core.h>
+#include <linux/clkdev.h>
+#include <linux/clk-provider.h>
+#include <linux/dmi.h>
+#include <linux/platform_data/gpio-dwapb.h>
+#include <linux/platform_data/i2c-designware.h>
+
+/* PCI BAR for register base address */
+#define MFD_I2C_BAR		0
+#define MFD_GPIO_BAR		1
+
+/* The base GPIO number under GPIOLIB framework */
+#define INTEL_QUARK_MFD_GPIO_BASE	8
+
+/* The default number of South-Cluster GPIO on Quark. */
+#define INTEL_QUARK_MFD_NGPIO		8
+
+/* The DesignWare GPIO ports on Quark. */
+#define INTEL_QUARK_GPIO_NPORTS	1
+
+#define INTEL_QUARK_IORES_MEM	0
+#define INTEL_QUARK_IORES_IRQ	1
+
+#define INTEL_QUARK_I2C_CONTROLLER_CLK "i2c_designware.0"
+
+/* The Quark I2C controller source clock */
+#define INTEL_QUARK_I2C_CLK_HZ	33000000
+
+#define INTEL_QUARK_I2C_NCLK	1
+
+struct intel_quark_mfd {
+	struct pci_dev		*pdev;
+	struct clk			*i2c_clk;
+	struct clk_lookup	*i2c_clk_lookup;
+};
+
+struct i2c_mode_info {
+	const char *name;
+	unsigned int i2c_scl_freq;
+};
+
+static const struct i2c_mode_info platform_i2c_mode_info[] = {
+	{
+		.name = "Galileo",
+		.i2c_scl_freq = 100000,
+	},
+	{
+		.name = "GalileoGen2",
+		.i2c_scl_freq = 400000,
+	},
+};
+
+static struct resource intel_quark_i2c_res[] = {
+	[INTEL_QUARK_IORES_MEM] = {
+		.flags = IORESOURCE_MEM,
+	},
+	[INTEL_QUARK_IORES_IRQ] = {
+		.flags = IORESOURCE_IRQ,
+	},
+};
+
+static struct resource intel_quark_gpio_res[] = {
+	[INTEL_QUARK_IORES_MEM] = {
+		.flags = IORESOURCE_MEM,
+	},
+};
+
+static struct mfd_cell intel_quark_mfd_cells[] = {
+	{
+		.id = MFD_I2C_BAR,
+		.name = "i2c_designware",
+		.num_resources = ARRAY_SIZE(intel_quark_i2c_res),
+		.resources = intel_quark_i2c_res,
+		.ignore_resource_conflicts = true,
+	},
+	{
+		.id = MFD_GPIO_BAR,
+		.name = "gpio-dwapb",
+		.num_resources = ARRAY_SIZE(intel_quark_gpio_res),
+		.resources = intel_quark_gpio_res,
+		.ignore_resource_conflicts = true,

drivers/mfd/intel_quark_i2c_gpio.c line 113

Quoted 10 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +10 -0
+	},
+};
+
+static const struct pci_device_id intel_quark_mfd_ids[] = {
+	{ PCI_VDEVICE(INTEL, 0x0934), },
+	{ 0,}
+};
+MODULE_DEVICE_TABLE(pci, intel_quark_mfd_ids);
+
+static int intel_quark_register_i2c_clk(struct intel_quark_mfd *quark_mfd)

What about quark_mfd -> mfd ? This prefix is kinda useless inside the functions in this module.

Quoted 23 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +23 -0
+{
+	struct pci_dev *pdev = quark_mfd->pdev;
+	struct clk_lookup *i2c_clk_lookup;
+	struct clk *i2c_clk;
+	int retval;
+
+	i2c_clk_lookup = devm_kcalloc(
+		&pdev->dev, INTEL_QUARK_I2C_NCLK,
+		sizeof(*i2c_clk_lookup), GFP_KERNEL);
+
+	if (!i2c_clk_lookup)
+		return -ENOMEM;
+
+	i2c_clk_lookup[0].dev_id = INTEL_QUARK_I2C_CONTROLLER_CLK;
+
+	i2c_clk = clk_register_fixed_rate(
+		&pdev->dev, INTEL_QUARK_I2C_CONTROLLER_CLK, NULL,
+		CLK_IS_ROOT, INTEL_QUARK_I2C_CLK_HZ);
+
+	quark_mfd->i2c_clk_lookup = i2c_clk_lookup;
+	quark_mfd->i2c_clk = i2c_clk;
+
+	retval = clk_register_clkdevs(i2c_clk, i2c_clk_lookup,

drivers/mfd/intel_quark_i2c_gpio.c line 146

Quoted 10 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +10 -0
+				      INTEL_QUARK_I2C_NCLK);
+	if (retval)
+		dev_err(&pdev->dev, "Fixed clk register failed: %d\n", retval);
+
+	return retval;
+}
+
+static void intel_quark_unregister_i2c_clk(struct pci_dev *pdev)
+{
+	struct intel_quark_mfd *quark_mfd = dev_get_drvdata(&pdev->dev);

Ditto.

drivers/mfd/intel_quark_i2c_gpio.c line 159

Quoted 13 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +13 -0
+
+	if (!quark_mfd->i2c_clk || !quark_mfd->i2c_clk_lookup)
+		return;
+
+	clkdev_drop(quark_mfd->i2c_clk_lookup);
+	clk_unregister(quark_mfd->i2c_clk);
+}
+
+static int intel_quark_i2c_setup(struct pci_dev *pdev, struct mfd_cell *cell)
+{
+	const char *board_name = dmi_get_system_info(DMI_BOARD_NAME);
+	struct dw_i2c_platform_data *pdata;
+	struct resource *res = (struct resource *)cell->resources;

Here you know what the exact resource you is going to change.

What if you update the global variable first, and assign it here to the cell?

Quoted 23 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +23 -0
+	struct device *dev = &pdev->dev;
+	unsigned int i;
+
+	res[INTEL_QUARK_IORES_MEM].start =
+		pci_resource_start(pdev, MFD_I2C_BAR);
+	res[INTEL_QUARK_IORES_MEM].end =
+		pci_resource_end(pdev, MFD_I2C_BAR);
+
+	res[INTEL_QUARK_IORES_IRQ].start = pdev->irq;
+	res[INTEL_QUARK_IORES_IRQ].end = pdev->irq;
+
+	pdata = devm_kzalloc(dev, sizeof(*pdata), GFP_KERNEL);
+	if (!pdata)
+		return -ENOMEM;
+
+	/* Fast mode by default */
+	pdata->i2c_scl_freq = 400000;
+
+	for (i = 0; i < ARRAY_SIZE(platform_i2c_mode_info); i++)
+		if (!strcmp(board_name, platform_i2c_mode_info[i].name))
+			pdata->i2c_scl_freq
+				= platform_i2c_mode_info[i].i2c_scl_freq;
+

drivers/mfd/intel_quark_i2c_gpio.c line 192

Quoted 10 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +10 -0
+	cell->platform_data = pdata;
+	cell->pdata_size = sizeof(*pdata);
+
+	return 0;
+}
+
+static int intel_quark_gpio_setup(struct pci_dev *pdev, struct mfd_cell *cell)
+{
+	struct dwapb_platform_data *pdata;
+	struct resource *res = (struct resource *)cell->resources;

Same idea here.

Quoted 87 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +87 -0
+	struct device *dev = &pdev->dev;
+
+	res[INTEL_QUARK_IORES_MEM].start =
+		pci_resource_start(pdev, MFD_GPIO_BAR);
+	res[INTEL_QUARK_IORES_MEM].end =
+		pci_resource_end(pdev, MFD_GPIO_BAR);
+
+	pdata = devm_kzalloc(dev, sizeof(*pdata), GFP_KERNEL);
+	if (!pdata)
+		return -ENOMEM;
+
+	/* For intel quark x1000, it has only one port: portA */
+	pdata->nports = INTEL_QUARK_GPIO_NPORTS;
+	pdata->properties = devm_kcalloc(dev, pdata->nports,
+					 sizeof(*pdata->properties),
+					 GFP_KERNEL);
+	if (!pdata->properties)
+		return -ENOMEM;
+
+	/* Set the properties for portA */
+	pdata->properties->node	= NULL;
+	pdata->properties->name	= "intel-quark-x1000-gpio-portA";
+	pdata->properties->idx	= 0;
+	pdata->properties->ngpio	= INTEL_QUARK_MFD_NGPIO;
+	pdata->properties->gpio_base	= INTEL_QUARK_MFD_GPIO_BASE;
+	pdata->properties->irq	= pdev->irq;
+	pdata->properties->irq_shared	= true;
+
+	cell->platform_data = pdata;
+	cell->pdata_size = sizeof(*pdata);
+
+	return 0;
+}
+
+static int intel_quark_mfd_probe(struct pci_dev *pdev,
+				 const struct pci_device_id *id)
+{
+	struct intel_quark_mfd *quark_mfd;
+	int retval;
+
+	retval = pcim_enable_device(pdev);
+	if (retval)
+		return retval;
+
+	quark_mfd = devm_kzalloc(&pdev->dev, sizeof(*quark_mfd), GFP_KERNEL);
+	if (!quark_mfd)
+		return -ENOMEM;
+	quark_mfd->pdev = pdev;
+
+	retval = intel_quark_register_i2c_clk(quark_mfd);
+	if (retval)
+		return retval;
+
+	dev_set_drvdata(&pdev->dev, quark_mfd);
+
+	retval = intel_quark_i2c_setup(pdev,
+				       &intel_quark_mfd_cells[MFD_I2C_BAR]);
+	if (retval)
+		return retval;
+
+	retval = intel_quark_gpio_setup(pdev,
+					&intel_quark_mfd_cells[MFD_GPIO_BAR]);
+	if (retval)
+		return retval;
+
+	return mfd_add_devices(&pdev->dev, 0, intel_quark_mfd_cells,
+		ARRAY_SIZE(intel_quark_mfd_cells), NULL, 0, NULL);
+}
+
+static void intel_quark_mfd_remove(struct pci_dev *pdev)
+{
+	intel_quark_unregister_i2c_clk(pdev);
+	mfd_remove_devices(&pdev->dev);
+}
+
+static struct pci_driver intel_quark_mfd_driver = {
+	.name		= "intel_quark_mfd_i2c_gpio",
+	.id_table	= intel_quark_mfd_ids,
+	.probe		= intel_quark_mfd_probe,
+	.remove		= intel_quark_mfd_remove,
+};
+
+module_pci_driver(intel_quark_mfd_driver);
+
+MODULE_AUTHOR("Raymond Tan <[email protected]>");
+MODULE_DESCRIPTION("Intel Quark MFD PCI driver for I2C & GPIO");
+MODULE_LICENSE("GPL v2");
Signature · 13 lines
Andy Shevchenko <[email protected]>
Intel Finland Oy
---------------------------------------------------------------------
Intel Finland Oy
Registered Address: PL 281, 00181 Helsinki 
Business Identity Code: 0357606 - 4 
Domiciled in Helsinki 

This e-mail and any attachments may contain confidential material for
the sole use of the intended recipient(s). Any review or distribution
by others is strictly prohibited. If you are not the intended
recipient, please contact the sender and delete all copies.
���{.n�+�������+%�����ݶ��w��{.n�+����{��G�����{ay�ʇڙ�,j��f���h��������z_�(�階�ݢj"���m������G������?����&���~��iO��z��v�^�m����������?�I�
#3 lore raw

Hi Lee Jones,

Happy New Year.

Regarding Andriy's suggestion to split and reorganize the MFD drivers for IA platform in the previous email, any thoughts from you? Should I proceed to work in splitting it or keep the structure as-is for now? Thanks.

Warm Regards,

 Raymond Tan

Quoted 397 lines

-----Original Message-----
From: Shevchenko, Andriy
Sent: Thursday, December 11, 2014 6:27 PM
To: Tan, Raymond
Cc: [email protected]; [email protected]; linux- [email protected]; Chen, Alvin; [email protected]
Subject: Re: [PATCH v3 1/1] mfd: intel_quark_i2c_gpio: Add Intel Quark X1000 I2C-GPIO MFD Driver

Quoted 5 lines · Raymond Tan

In Quark X1000, there's a single PCI device that provides both an I2C controller and a GPIO controller. This MFD driver will split the 2 devices for their respective drivers.

This patch is based on Josef Ahmad's initial work for Quark enabling.

See my comments below.

Quoted 9 lines
---
 drivers/mfd/Kconfig                |   12 ++
 drivers/mfd/Makefile               |    1 +
 drivers/mfd/intel_quark_i2c_gpio.c |  279

++++++++++++++++++++++++++++++++++++

So, there is a proposal to bring the order into the source tree a bit for Intel related MFD drivers.

What about to move this module to
drivers/mfd/intel/quark/i2c_gpio.c ?

At least what I can tell now, it will come more Intel stuff and we have few drivers already there.

I can see the following:
drivers/mfd/intel/
-> quark/
-> mid/ (Intel MID platforms, existing drivers) -> some new platform/
(whatever is the one of the new Intel SoCs)

Lee, what's your opinion?

I could change a layout for existing in-tree drivers.

drivers/mfd/Kconfig config HTC_I2CPLD · line 222

Quoted 12 lines
 3 files changed, 292 insertions(+)
 create mode 100644 drivers/mfd/intel_quark_i2c_gpio.c
drivers/mfd/Kconfig index renamed from drivers/mfd/Kconfig +0 -0

b7c74a7..2f2f62d 100644

drivers/mfd/Kconfig +1 -0
@@ -219,6 +219,18 @@ config HTC_I2CPLD
 	  This device provides input and output GPIOs through an I2C
 	  interface to one or more sub-chips.
 
+config MFD_INTEL_QUARK_I2C_GPIO

Depending on the further discussion about layout.

drivers/mfd/Makefile obj-$(CONFIG_AB8500_CORE) += ab8500-

Quoted 19 lines
drivers/mfd/Kconfig +11 -0
+	tristate "Intel Quark MFD I2C GPIO"
+	depends on PCI
+	depends on X86
+	depends on COMMON_CLK
+	select MFD_CORE
+	help
+	  This MFD provides support for I2C and GPIO that exist only
+	  in a single PCI device. It splits the 2 IO devices to
+	  their respective IO driver.
+	  The GPIO exports a total amount of 8 interrupt-capable GPIOs.
+
 config LPC_ICH
 	tristate "Intel ICH LPC"
 	depends on PCI
drivers/mfd/Makefile index renamed from drivers/mfd/Makefile +0 -0

8a28dc9..d42652d 100644

drivers/mfd/Makefile +0 -0
@@ -133,6 +133,7 @@ obj-$(CONFIG_AB8500_CORE)	+= ab8500-

core.o ab8500-sysctrl.o

drivers/mfd/Makefile obj-$(CONFIG_AB8500_CORE) += ab8500-core.o ab8500-sysctrl.o

Quoted 4 lines
drivers/mfd/Makefile +1 -0
 obj-$(CONFIG_MFD_TIMBERDALE)    += timberdale.o
 obj-$(CONFIG_PMIC_ADP5520)	+= adp5520.o
 obj-$(CONFIG_MFD_KEMPLD)	+= kempld-core.o
+obj-$(CONFIG_MFD_INTEL_QUARK_I2C_GPIO)	+=

intel_quark_i2c_gpio.o

Quoted 13 lines
drivers/mfd/Makefile +0 -0
 obj-$(CONFIG_LPC_SCH)		+= lpc_sch.o
 obj-$(CONFIG_LPC_ICH)		+= lpc_ich.o
 obj-$(CONFIG_MFD_RDC321X)	+= rdc321x-southbridge.o
drivers/mfd/intel_quark_i2c_gpio.c new file +3 -0
new file mode 100644
index 0000000..a3a7043
@@ -0,0 +1,279 @@
+/*
+ * Intel Quark MFD PCI driver for I2C & GPIO
+ *

drivers/mfd/intel_quark_i2c_gpio.c line 13

Quoted 10 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +10 -0
+ * Copyright(c) 2014 Intel Corporation.
+ *
+ * This program is free software; you can redistribute it and/or
+modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but
+WITHOUT
+ * ANY WARRANTY; without even the implied warranty of

MERCHANTABILITY

Quoted 96 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +96 -0
+or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public
+License for
+ * more details.
+ *
+ * Intel Quark PCI device for I2C and GPIO controller sharing the
+same
+ * PCI function. This PCI driver will split the 2 devices into their
+ * respective drivers.
+ */
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/pci.h>
+#include <linux/mfd/core.h>
+#include <linux/clkdev.h>
+#include <linux/clk-provider.h>
+#include <linux/dmi.h>
+#include <linux/platform_data/gpio-dwapb.h>
+#include <linux/platform_data/i2c-designware.h>
+
+/* PCI BAR for register base address */
+#define MFD_I2C_BAR		0
+#define MFD_GPIO_BAR		1
+
+/* The base GPIO number under GPIOLIB framework */
+#define INTEL_QUARK_MFD_GPIO_BASE	8
+
+/* The default number of South-Cluster GPIO on Quark. */
+#define INTEL_QUARK_MFD_NGPIO		8
+
+/* The DesignWare GPIO ports on Quark. */
+#define INTEL_QUARK_GPIO_NPORTS	1
+
+#define INTEL_QUARK_IORES_MEM	0
+#define INTEL_QUARK_IORES_IRQ	1
+
+#define INTEL_QUARK_I2C_CONTROLLER_CLK "i2c_designware.0"
+
+/* The Quark I2C controller source clock */
+#define INTEL_QUARK_I2C_CLK_HZ	33000000
+
+#define INTEL_QUARK_I2C_NCLK	1
+
+struct intel_quark_mfd {
+	struct pci_dev		*pdev;
+	struct clk			*i2c_clk;
+	struct clk_lookup	*i2c_clk_lookup;
+};
+
+struct i2c_mode_info {
+	const char *name;
+	unsigned int i2c_scl_freq;
+};
+
+static const struct i2c_mode_info platform_i2c_mode_info[] = {
+	{
+		.name = "Galileo",
+		.i2c_scl_freq = 100000,
+	},
+	{
+		.name = "GalileoGen2",
+		.i2c_scl_freq = 400000,
+	},
+};
+
+static struct resource intel_quark_i2c_res[] = {
+	[INTEL_QUARK_IORES_MEM] = {
+		.flags = IORESOURCE_MEM,
+	},
+	[INTEL_QUARK_IORES_IRQ] = {
+		.flags = IORESOURCE_IRQ,
+	},
+};
+
+static struct resource intel_quark_gpio_res[] = {
+	[INTEL_QUARK_IORES_MEM] = {
+		.flags = IORESOURCE_MEM,
+	},
+};
+
+static struct mfd_cell intel_quark_mfd_cells[] = {
+	{
+		.id = MFD_I2C_BAR,
+		.name = "i2c_designware",
+		.num_resources = ARRAY_SIZE(intel_quark_i2c_res),
+		.resources = intel_quark_i2c_res,
+		.ignore_resource_conflicts = true,
+	},
+	{
+		.id = MFD_GPIO_BAR,
+		.name = "gpio-dwapb",
+		.num_resources = ARRAY_SIZE(intel_quark_gpio_res),
+		.resources = intel_quark_gpio_res,
+		.ignore_resource_conflicts = true,
+	},

drivers/mfd/intel_quark_i2c_gpio.c

Quoted 10 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +10 -0
+};
+
+static const struct pci_device_id intel_quark_mfd_ids[] = {
+	{ PCI_VDEVICE(INTEL, 0x0934), },
+	{ 0,}
+};
+MODULE_DEVICE_TABLE(pci, intel_quark_mfd_ids);
+
+static int intel_quark_register_i2c_clk(struct intel_quark_mfd
+*quark_mfd)

What about quark_mfd -> mfd ? This prefix is kinda useless inside the functions in this module.

Quoted 16 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +16 -0
+{
+	struct pci_dev *pdev = quark_mfd->pdev;
+	struct clk_lookup *i2c_clk_lookup;
+	struct clk *i2c_clk;
+	int retval;
+
+	i2c_clk_lookup = devm_kcalloc(
+		&pdev->dev, INTEL_QUARK_I2C_NCLK,
+		sizeof(*i2c_clk_lookup), GFP_KERNEL);
+
+	if (!i2c_clk_lookup)
+		return -ENOMEM;
+
+	i2c_clk_lookup[0].dev_id = INTEL_QUARK_I2C_CONTROLLER_CLK;
+
+	i2c_clk = clk_register_fixed_rate(

drivers/mfd/intel_quark_i2c_gpio.c

Quoted 10 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +10 -0
+		&pdev->dev, INTEL_QUARK_I2C_CONTROLLER_CLK, NULL,
+		CLK_IS_ROOT, INTEL_QUARK_I2C_CLK_HZ);
+
+	quark_mfd->i2c_clk_lookup = i2c_clk_lookup;
+	quark_mfd->i2c_clk = i2c_clk;
+
+	retval = clk_register_clkdevs(i2c_clk, i2c_clk_lookup,
+				      INTEL_QUARK_I2C_NCLK);
+	if (retval)
+		dev_err(&pdev->dev, "Fixed clk register failed: %d\n",

retval);

Quoted 7 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +6 -0
+
+	return retval;
+}
+
+static void intel_quark_unregister_i2c_clk(struct pci_dev *pdev) {
+	struct intel_quark_mfd *quark_mfd = dev_get_drvdata(&pdev-

dev);

Ditto.

drivers/mfd/intel_quark_i2c_gpio.c

Quoted 11 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +11 -0
+
+	if (!quark_mfd->i2c_clk || !quark_mfd->i2c_clk_lookup)
+		return;
+
+	clkdev_drop(quark_mfd->i2c_clk_lookup);
+	clk_unregister(quark_mfd->i2c_clk);
+}
+
+static int intel_quark_i2c_setup(struct pci_dev *pdev, struct
+mfd_cell *cell) {
+	const char *board_name =

dmi_get_system_info(DMI_BOARD_NAME);

drivers/mfd/intel_quark_i2c_gpio.c

Quoted 2 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +2 -0
+	struct dw_i2c_platform_data *pdata;
+	struct resource *res = (struct resource *)cell->resources;

Here you know what the exact resource you is going to change.

What if you update the global variable first, and assign it here to the cell?

Quoted 23 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +23 -0
+	struct device *dev = &pdev->dev;
+	unsigned int i;
+
+	res[INTEL_QUARK_IORES_MEM].start =
+		pci_resource_start(pdev, MFD_I2C_BAR);
+	res[INTEL_QUARK_IORES_MEM].end =
+		pci_resource_end(pdev, MFD_I2C_BAR);
+
+	res[INTEL_QUARK_IORES_IRQ].start = pdev->irq;
+	res[INTEL_QUARK_IORES_IRQ].end = pdev->irq;
+
+	pdata = devm_kzalloc(dev, sizeof(*pdata), GFP_KERNEL);
+	if (!pdata)
+		return -ENOMEM;
+
+	/* Fast mode by default */
+	pdata->i2c_scl_freq = 400000;
+
+	for (i = 0; i < ARRAY_SIZE(platform_i2c_mode_info); i++)
+		if (!strcmp(board_name, platform_i2c_mode_info[i].name))
+			pdata->i2c_scl_freq
+				= platform_i2c_mode_info[i].i2c_scl_freq;
+

drivers/mfd/intel_quark_i2c_gpio.c line 192

Quoted 10 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +10 -0
+	cell->platform_data = pdata;
+	cell->pdata_size = sizeof(*pdata);
+
+	return 0;
+}
+
+static int intel_quark_gpio_setup(struct pci_dev *pdev, struct
+mfd_cell *cell) {
+	struct dwapb_platform_data *pdata;
+	struct resource *res = (struct resource *)cell->resources;

Same idea here.

Quoted 35 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +35 -0
+	struct device *dev = &pdev->dev;
+
+	res[INTEL_QUARK_IORES_MEM].start =
+		pci_resource_start(pdev, MFD_GPIO_BAR);
+	res[INTEL_QUARK_IORES_MEM].end =
+		pci_resource_end(pdev, MFD_GPIO_BAR);
+
+	pdata = devm_kzalloc(dev, sizeof(*pdata), GFP_KERNEL);
+	if (!pdata)
+		return -ENOMEM;
+
+	/* For intel quark x1000, it has only one port: portA */
+	pdata->nports = INTEL_QUARK_GPIO_NPORTS;
+	pdata->properties = devm_kcalloc(dev, pdata->nports,
+					 sizeof(*pdata->properties),
+					 GFP_KERNEL);
+	if (!pdata->properties)
+		return -ENOMEM;
+
+	/* Set the properties for portA */
+	pdata->properties->node	= NULL;
+	pdata->properties->name	= "intel-quark-x1000-gpio-portA";
+	pdata->properties->idx	= 0;
+	pdata->properties->ngpio	= INTEL_QUARK_MFD_NGPIO;
+	pdata->properties->gpio_base	= INTEL_QUARK_MFD_GPIO_BASE;
+	pdata->properties->irq	= pdev->irq;
+	pdata->properties->irq_shared	= true;
+
+	cell->platform_data = pdata;
+	cell->pdata_size = sizeof(*pdata);
+
+	return 0;
+}
+
+static int intel_quark_mfd_probe(struct pci_dev *pdev,

drivers/mfd/intel_quark_i2c_gpio.c

Quoted 10 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +10 -0
+				 const struct pci_device_id *id)
+{
+	struct intel_quark_mfd *quark_mfd;
+	int retval;
+
+	retval = pcim_enable_device(pdev);
+	if (retval)
+		return retval;
+
+	quark_mfd = devm_kzalloc(&pdev->dev, sizeof(*quark_mfd),

GFP_KERNEL);

Quoted 7 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +7 -0
+	if (!quark_mfd)
+		return -ENOMEM;
+	quark_mfd->pdev = pdev;
+
+	retval = intel_quark_register_i2c_clk(quark_mfd);
+	if (retval)
+		return retval;

drivers/mfd/intel_quark_i2c_gpio.c line 257

Quoted 10 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +10 -0
+
+	dev_set_drvdata(&pdev->dev, quark_mfd);
+
+	retval = intel_quark_i2c_setup(pdev,
+				       &intel_quark_mfd_cells[MFD_I2C_BAR]);
+	if (retval)
+		return retval;
+
+	retval = intel_quark_gpio_setup(pdev,
+
	&intel_quark_mfd_cells[MFD_GPIO_BAR]);
Quoted 19 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +19 -0
+	if (retval)
+		return retval;
+
+	return mfd_add_devices(&pdev->dev, 0, intel_quark_mfd_cells,
+		ARRAY_SIZE(intel_quark_mfd_cells), NULL, 0, NULL); }
+
+static void intel_quark_mfd_remove(struct pci_dev *pdev) {
+	intel_quark_unregister_i2c_clk(pdev);
+	mfd_remove_devices(&pdev->dev);
+}
+
+static struct pci_driver intel_quark_mfd_driver = {
+	.name		= "intel_quark_mfd_i2c_gpio",
+	.id_table	= intel_quark_mfd_ids,
+	.probe		= intel_quark_mfd_probe,
+	.remove		= intel_quark_mfd_remove,
+};
+
+module_pci_driver(intel_quark_mfd_driver);
Quoted 8 lines

drivers/mfd/intel_quark_i2c_gpio.c line 279

Quoted 4 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +4 -0
+
+MODULE_AUTHOR("Raymond Tan <[email protected]>");
+MODULE_DESCRIPTION("Intel Quark MFD PCI driver for I2C & GPIO");
+MODULE_LICENSE("GPL v2");

--
Andy Shevchenko <[email protected]> Intel Finland Oy

���{.n�+�������+%�����ݶ��w��{.n�+����{��G�����{ay�ʇڙ�,j��f���h��������z_�(�階�ݢj"���m������G������?����&���~��iO��z��v�^�m���� ������?�I�

#4 lore raw
Quoted 1 line

Happy New Year.

And to you.

Quoted 1 line

Regarding Andriy's suggestion to split and reorganize the MFD drivers for IA platform in the previous email, any thoughts from you? Should I proceed to work in splitting it or keep the structure as-is for now? Thanks.

I will review it as it is for now. I need to take a closer look at the proposal when I get down to it. I am currently only half way through my post-vacation backlog.

Quoted 403 lines
Quoted 403 lines

-----Original Message-----
From: Shevchenko, Andriy
Sent: Thursday, December 11, 2014 6:27 PM
To: Tan, Raymond
Cc: [email protected]; [email protected]; linux- [email protected]; Chen, Alvin; [email protected]
Subject: Re: [PATCH v3 1/1] mfd: intel_quark_i2c_gpio: Add Intel Quark X1000 I2C-GPIO MFD Driver

Quoted 5 lines · Raymond Tan

In Quark X1000, there's a single PCI device that provides both an I2C controller and a GPIO controller. This MFD driver will split the 2 devices for their respective drivers.

This patch is based on Josef Ahmad's initial work for Quark enabling.

See my comments below.

Quoted 9 lines
---
 drivers/mfd/Kconfig                |   12 ++
 drivers/mfd/Makefile               |    1 +
 drivers/mfd/intel_quark_i2c_gpio.c |  279

++++++++++++++++++++++++++++++++++++

So, there is a proposal to bring the order into the source tree a bit for Intel related MFD drivers.

What about to move this module to
drivers/mfd/intel/quark/i2c_gpio.c ?

At least what I can tell now, it will come more Intel stuff and we have few drivers already there.

I can see the following:
drivers/mfd/intel/
-> quark/
-> mid/ (Intel MID platforms, existing drivers) -> some new platform/
(whatever is the one of the new Intel SoCs)

Lee, what's your opinion?

I could change a layout for existing in-tree drivers.

drivers/mfd/Kconfig config HTC_I2CPLD · line 222

Quoted 12 lines
 3 files changed, 292 insertions(+)
 create mode 100644 drivers/mfd/intel_quark_i2c_gpio.c
drivers/mfd/Kconfig index renamed from drivers/mfd/Kconfig +0 -0

b7c74a7..2f2f62d 100644

drivers/mfd/Kconfig +1 -0
@@ -219,6 +219,18 @@ config HTC_I2CPLD
 	  This device provides input and output GPIOs through an I2C
 	  interface to one or more sub-chips.
 
+config MFD_INTEL_QUARK_I2C_GPIO

Depending on the further discussion about layout.

drivers/mfd/Makefile obj-$(CONFIG_AB8500_CORE) += ab8500-

Quoted 19 lines
drivers/mfd/Kconfig +11 -0
+	tristate "Intel Quark MFD I2C GPIO"
+	depends on PCI
+	depends on X86
+	depends on COMMON_CLK
+	select MFD_CORE
+	help
+	  This MFD provides support for I2C and GPIO that exist only
+	  in a single PCI device. It splits the 2 IO devices to
+	  their respective IO driver.
+	  The GPIO exports a total amount of 8 interrupt-capable GPIOs.
+
 config LPC_ICH
 	tristate "Intel ICH LPC"
 	depends on PCI
drivers/mfd/Makefile index renamed from drivers/mfd/Makefile +0 -0

8a28dc9..d42652d 100644

drivers/mfd/Makefile +0 -0
@@ -133,6 +133,7 @@ obj-$(CONFIG_AB8500_CORE)	+= ab8500-

core.o ab8500-sysctrl.o

drivers/mfd/Makefile obj-$(CONFIG_AB8500_CORE) += ab8500-core.o ab8500-sysctrl.o

Quoted 4 lines
drivers/mfd/Makefile +1 -0
 obj-$(CONFIG_MFD_TIMBERDALE)    += timberdale.o
 obj-$(CONFIG_PMIC_ADP5520)	+= adp5520.o
 obj-$(CONFIG_MFD_KEMPLD)	+= kempld-core.o
+obj-$(CONFIG_MFD_INTEL_QUARK_I2C_GPIO)	+=

intel_quark_i2c_gpio.o

Quoted 13 lines
drivers/mfd/Makefile +0 -0
 obj-$(CONFIG_LPC_SCH)		+= lpc_sch.o
 obj-$(CONFIG_LPC_ICH)		+= lpc_ich.o
 obj-$(CONFIG_MFD_RDC321X)	+= rdc321x-southbridge.o
drivers/mfd/intel_quark_i2c_gpio.c new file +3 -0
new file mode 100644
index 0000000..a3a7043
@@ -0,0 +1,279 @@
+/*
+ * Intel Quark MFD PCI driver for I2C & GPIO
+ *

drivers/mfd/intel_quark_i2c_gpio.c line 13

Quoted 10 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +10 -0
+ * Copyright(c) 2014 Intel Corporation.
+ *
+ * This program is free software; you can redistribute it and/or
+modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but
+WITHOUT
+ * ANY WARRANTY; without even the implied warranty of

MERCHANTABILITY

Quoted 96 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +96 -0
+or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public
+License for
+ * more details.
+ *
+ * Intel Quark PCI device for I2C and GPIO controller sharing the
+same
+ * PCI function. This PCI driver will split the 2 devices into their
+ * respective drivers.
+ */
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/pci.h>
+#include <linux/mfd/core.h>
+#include <linux/clkdev.h>
+#include <linux/clk-provider.h>
+#include <linux/dmi.h>
+#include <linux/platform_data/gpio-dwapb.h>
+#include <linux/platform_data/i2c-designware.h>
+
+/* PCI BAR for register base address */
+#define MFD_I2C_BAR		0
+#define MFD_GPIO_BAR		1
+
+/* The base GPIO number under GPIOLIB framework */
+#define INTEL_QUARK_MFD_GPIO_BASE	8
+
+/* The default number of South-Cluster GPIO on Quark. */
+#define INTEL_QUARK_MFD_NGPIO		8
+
+/* The DesignWare GPIO ports on Quark. */
+#define INTEL_QUARK_GPIO_NPORTS	1
+
+#define INTEL_QUARK_IORES_MEM	0
+#define INTEL_QUARK_IORES_IRQ	1
+
+#define INTEL_QUARK_I2C_CONTROLLER_CLK "i2c_designware.0"
+
+/* The Quark I2C controller source clock */
+#define INTEL_QUARK_I2C_CLK_HZ	33000000
+
+#define INTEL_QUARK_I2C_NCLK	1
+
+struct intel_quark_mfd {
+	struct pci_dev		*pdev;
+	struct clk			*i2c_clk;
+	struct clk_lookup	*i2c_clk_lookup;
+};
+
+struct i2c_mode_info {
+	const char *name;
+	unsigned int i2c_scl_freq;
+};
+
+static const struct i2c_mode_info platform_i2c_mode_info[] = {
+	{
+		.name = "Galileo",
+		.i2c_scl_freq = 100000,
+	},
+	{
+		.name = "GalileoGen2",
+		.i2c_scl_freq = 400000,
+	},
+};
+
+static struct resource intel_quark_i2c_res[] = {
+	[INTEL_QUARK_IORES_MEM] = {
+		.flags = IORESOURCE_MEM,
+	},
+	[INTEL_QUARK_IORES_IRQ] = {
+		.flags = IORESOURCE_IRQ,
+	},
+};
+
+static struct resource intel_quark_gpio_res[] = {
+	[INTEL_QUARK_IORES_MEM] = {
+		.flags = IORESOURCE_MEM,
+	},
+};
+
+static struct mfd_cell intel_quark_mfd_cells[] = {
+	{
+		.id = MFD_I2C_BAR,
+		.name = "i2c_designware",
+		.num_resources = ARRAY_SIZE(intel_quark_i2c_res),
+		.resources = intel_quark_i2c_res,
+		.ignore_resource_conflicts = true,
+	},
+	{
+		.id = MFD_GPIO_BAR,
+		.name = "gpio-dwapb",
+		.num_resources = ARRAY_SIZE(intel_quark_gpio_res),
+		.resources = intel_quark_gpio_res,
+		.ignore_resource_conflicts = true,
+	},

drivers/mfd/intel_quark_i2c_gpio.c

Quoted 10 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +10 -0
+};
+
+static const struct pci_device_id intel_quark_mfd_ids[] = {
+	{ PCI_VDEVICE(INTEL, 0x0934), },
+	{ 0,}
+};
+MODULE_DEVICE_TABLE(pci, intel_quark_mfd_ids);
+
+static int intel_quark_register_i2c_clk(struct intel_quark_mfd
+*quark_mfd)

What about quark_mfd -> mfd ? This prefix is kinda useless inside the functions in this module.

Quoted 16 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +16 -0
+{
+	struct pci_dev *pdev = quark_mfd->pdev;
+	struct clk_lookup *i2c_clk_lookup;
+	struct clk *i2c_clk;
+	int retval;
+
+	i2c_clk_lookup = devm_kcalloc(
+		&pdev->dev, INTEL_QUARK_I2C_NCLK,
+		sizeof(*i2c_clk_lookup), GFP_KERNEL);
+
+	if (!i2c_clk_lookup)
+		return -ENOMEM;
+
+	i2c_clk_lookup[0].dev_id = INTEL_QUARK_I2C_CONTROLLER_CLK;
+
+	i2c_clk = clk_register_fixed_rate(

drivers/mfd/intel_quark_i2c_gpio.c

Quoted 10 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +10 -0
+		&pdev->dev, INTEL_QUARK_I2C_CONTROLLER_CLK, NULL,
+		CLK_IS_ROOT, INTEL_QUARK_I2C_CLK_HZ);
+
+	quark_mfd->i2c_clk_lookup = i2c_clk_lookup;
+	quark_mfd->i2c_clk = i2c_clk;
+
+	retval = clk_register_clkdevs(i2c_clk, i2c_clk_lookup,
+				      INTEL_QUARK_I2C_NCLK);
+	if (retval)
+		dev_err(&pdev->dev, "Fixed clk register failed: %d\n",

retval);

Quoted 7 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +6 -0
+
+	return retval;
+}
+
+static void intel_quark_unregister_i2c_clk(struct pci_dev *pdev) {
+	struct intel_quark_mfd *quark_mfd = dev_get_drvdata(&pdev-

dev);

Ditto.

drivers/mfd/intel_quark_i2c_gpio.c

Quoted 11 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +11 -0
+
+	if (!quark_mfd->i2c_clk || !quark_mfd->i2c_clk_lookup)
+		return;
+
+	clkdev_drop(quark_mfd->i2c_clk_lookup);
+	clk_unregister(quark_mfd->i2c_clk);
+}
+
+static int intel_quark_i2c_setup(struct pci_dev *pdev, struct
+mfd_cell *cell) {
+	const char *board_name =

dmi_get_system_info(DMI_BOARD_NAME);

drivers/mfd/intel_quark_i2c_gpio.c

Quoted 2 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +2 -0
+	struct dw_i2c_platform_data *pdata;
+	struct resource *res = (struct resource *)cell->resources;

Here you know what the exact resource you is going to change.

What if you update the global variable first, and assign it here to the cell?

Quoted 23 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +23 -0
+	struct device *dev = &pdev->dev;
+	unsigned int i;
+
+	res[INTEL_QUARK_IORES_MEM].start =
+		pci_resource_start(pdev, MFD_I2C_BAR);
+	res[INTEL_QUARK_IORES_MEM].end =
+		pci_resource_end(pdev, MFD_I2C_BAR);
+
+	res[INTEL_QUARK_IORES_IRQ].start = pdev->irq;
+	res[INTEL_QUARK_IORES_IRQ].end = pdev->irq;
+
+	pdata = devm_kzalloc(dev, sizeof(*pdata), GFP_KERNEL);
+	if (!pdata)
+		return -ENOMEM;
+
+	/* Fast mode by default */
+	pdata->i2c_scl_freq = 400000;
+
+	for (i = 0; i < ARRAY_SIZE(platform_i2c_mode_info); i++)
+		if (!strcmp(board_name, platform_i2c_mode_info[i].name))
+			pdata->i2c_scl_freq
+				= platform_i2c_mode_info[i].i2c_scl_freq;
+

drivers/mfd/intel_quark_i2c_gpio.c line 192

Quoted 10 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +10 -0
+	cell->platform_data = pdata;
+	cell->pdata_size = sizeof(*pdata);
+
+	return 0;
+}
+
+static int intel_quark_gpio_setup(struct pci_dev *pdev, struct
+mfd_cell *cell) {
+	struct dwapb_platform_data *pdata;
+	struct resource *res = (struct resource *)cell->resources;

Same idea here.

Quoted 35 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +35 -0
+	struct device *dev = &pdev->dev;
+
+	res[INTEL_QUARK_IORES_MEM].start =
+		pci_resource_start(pdev, MFD_GPIO_BAR);
+	res[INTEL_QUARK_IORES_MEM].end =
+		pci_resource_end(pdev, MFD_GPIO_BAR);
+
+	pdata = devm_kzalloc(dev, sizeof(*pdata), GFP_KERNEL);
+	if (!pdata)
+		return -ENOMEM;
+
+	/* For intel quark x1000, it has only one port: portA */
+	pdata->nports = INTEL_QUARK_GPIO_NPORTS;
+	pdata->properties = devm_kcalloc(dev, pdata->nports,
+					 sizeof(*pdata->properties),
+					 GFP_KERNEL);
+	if (!pdata->properties)
+		return -ENOMEM;
+
+	/* Set the properties for portA */
+	pdata->properties->node	= NULL;
+	pdata->properties->name	= "intel-quark-x1000-gpio-portA";
+	pdata->properties->idx	= 0;
+	pdata->properties->ngpio	= INTEL_QUARK_MFD_NGPIO;
+	pdata->properties->gpio_base	= INTEL_QUARK_MFD_GPIO_BASE;
+	pdata->properties->irq	= pdev->irq;
+	pdata->properties->irq_shared	= true;
+
+	cell->platform_data = pdata;
+	cell->pdata_size = sizeof(*pdata);
+
+	return 0;
+}
+
+static int intel_quark_mfd_probe(struct pci_dev *pdev,

drivers/mfd/intel_quark_i2c_gpio.c

Quoted 10 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +10 -0
+				 const struct pci_device_id *id)
+{
+	struct intel_quark_mfd *quark_mfd;
+	int retval;
+
+	retval = pcim_enable_device(pdev);
+	if (retval)
+		return retval;
+
+	quark_mfd = devm_kzalloc(&pdev->dev, sizeof(*quark_mfd),

GFP_KERNEL);

Quoted 7 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +7 -0
+	if (!quark_mfd)
+		return -ENOMEM;
+	quark_mfd->pdev = pdev;
+
+	retval = intel_quark_register_i2c_clk(quark_mfd);
+	if (retval)
+		return retval;

drivers/mfd/intel_quark_i2c_gpio.c line 257

Quoted 10 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +10 -0
+
+	dev_set_drvdata(&pdev->dev, quark_mfd);
+
+	retval = intel_quark_i2c_setup(pdev,
+				       &intel_quark_mfd_cells[MFD_I2C_BAR]);
+	if (retval)
+		return retval;
+
+	retval = intel_quark_gpio_setup(pdev,
+
	&intel_quark_mfd_cells[MFD_GPIO_BAR]);
Quoted 23 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +23 -0
+	if (retval)
+		return retval;
+
+	return mfd_add_devices(&pdev->dev, 0, intel_quark_mfd_cells,
+		ARRAY_SIZE(intel_quark_mfd_cells), NULL, 0, NULL); }
+
+static void intel_quark_mfd_remove(struct pci_dev *pdev) {
+	intel_quark_unregister_i2c_clk(pdev);
+	mfd_remove_devices(&pdev->dev);
+}
+
+static struct pci_driver intel_quark_mfd_driver = {
+	.name		= "intel_quark_mfd_i2c_gpio",
+	.id_table	= intel_quark_mfd_ids,
+	.probe		= intel_quark_mfd_probe,
+	.remove		= intel_quark_mfd_remove,
+};
+
+module_pci_driver(intel_quark_mfd_driver);
+
+MODULE_AUTHOR("Raymond Tan <[email protected]>");
+MODULE_DESCRIPTION("Intel Quark MFD PCI driver for I2C & GPIO");
+MODULE_LICENSE("GPL v2");
Signature · 4 lines
Lee Jones
Linaro STMicroelectronics Landing Team Lead
Linaro.org │ Open source software for ARM SoCs
Follow Linaro: Facebook | Twitter | Blog
#5 lore raw
In reply to Raymond Tan

Quoting Raymond Tan (2014-12-11 01:38:30)

Quoted 15 lines

In Quark X1000, there's a single PCI device that provides both an I2C controller and a GPIO controller. This MFD driver will split the 2 devices for their respective drivers.

This patch is based on Josef Ahmad's initial work for Quark enabling.

---
 drivers/mfd/Kconfig                |   12 ++
 drivers/mfd/Makefile               |    1 +
 drivers/mfd/intel_quark_i2c_gpio.c |  279 ++++++++++++++++++++++++++++++++++++
 3 files changed, 292 insertions(+)
 create mode 100644 drivers/mfd/intel_quark_i2c_gpio.c

<snip>

Quoted 15 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +15 -0
+static int intel_quark_register_i2c_clk(struct intel_quark_mfd *quark_mfd)
+{
+       struct pci_dev *pdev = quark_mfd->pdev;
+       struct clk_lookup *i2c_clk_lookup;
+       struct clk *i2c_clk;
+       int retval;
+
+       i2c_clk_lookup = devm_kcalloc(
+               &pdev->dev, INTEL_QUARK_I2C_NCLK,
+               sizeof(*i2c_clk_lookup), GFP_KERNEL);
+
+       if (!i2c_clk_lookup)
+               return -ENOMEM;
+
+       i2c_clk_lookup[0].dev_id = INTEL_QUARK_I2C_CONTROLLER_CLK;

drivers/mfd/intel_quark_i2c_gpio.c line 137

Quoted 10 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +10 -0
+
+       i2c_clk = clk_register_fixed_rate(
+               &pdev->dev, INTEL_QUARK_I2C_CONTROLLER_CLK, NULL,
+               CLK_IS_ROOT, INTEL_QUARK_I2C_CLK_HZ);
+
+       quark_mfd->i2c_clk_lookup = i2c_clk_lookup;
+       quark_mfd->i2c_clk = i2c_clk;
+
+       retval = clk_register_clkdevs(i2c_clk, i2c_clk_lookup,
+                                     INTEL_QUARK_I2C_NCLK);

Lee asked about this in V2, so I'll follow up here in V3. It is OK for a driver to use the clock provider api to register clocks with the clk framework if that device truly is the provider of that clock signal. A good example can be found here:

drivers/media/platform/omap3isp/isp.c

The OMAP3 ISP receives a clock signal as a input. Within the image signal processor IP block it also has some basic clock controls of it's own which it feeds to downstream IP blocks. As such it is both a clock consumer and a provider and this is a common pattern amongst SoC designs.

So my question for this driver is if i2c_clk is provided by whatever the hell this mfd device is supposed to be, or if it's just a convenient place to call the code?

Another concern is that fact that this is a fixed clock. For architectures that use device tree to desribe board topology (ARM, MIPS, PPC) it is common to simply put the fixed-rate clocks there and not directly into the drive code. This prevents having to hack a lot of conditionals into your driver when rev 2.0 of your hardware comes out with a faster fixed rate clock, but you still need to support 1.0 hardware users at the slower rate. I don't know if x86 has a similar way of describing board topology but it might something to look into.

Regards,
Mike

#6 lore raw

Hi Mike,

Thanks for your reply. I've answered the questions as below.

Warm Regards,

Raymond Tan

Quoted 60 lines

-----Original Message-----
From: Mike Turquette [mailto:[email protected]]
Sent: Friday, December 12, 2014 6:26 AM
To: Tan, Raymond; Lee Jones; Samuel Ortiz
Cc: [email protected]; Chen, Alvin; Shevchenko, Andriy; Tan, Raymond
Subject: Re: [PATCH v3 1/1] mfd: intel_quark_i2c_gpio: Add Intel Quark X1000 I2C-GPIO MFD Driver

Quoting Raymond Tan (2014-12-11 01:38:30)

Quoted 16 lines

In Quark X1000, there's a single PCI device that provides both an I2C controller and a GPIO controller. This MFD driver will split the 2 devices for their respective drivers.

This patch is based on Josef Ahmad's initial work for Quark enabling.

---
 drivers/mfd/Kconfig                |   12 ++
 drivers/mfd/Makefile               |    1 +
 drivers/mfd/intel_quark_i2c_gpio.c |  279
++++++++++++++++++++++++++++++++++++
 3 files changed, 292 insertions(+)
 create mode 100644 drivers/mfd/intel_quark_i2c_gpio.c

<snip>

Quoted 15 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +15 -0
+static int intel_quark_register_i2c_clk(struct intel_quark_mfd
+*quark_mfd) {
+       struct pci_dev *pdev = quark_mfd->pdev;
+       struct clk_lookup *i2c_clk_lookup;
+       struct clk *i2c_clk;
+       int retval;
+
+       i2c_clk_lookup = devm_kcalloc(
+               &pdev->dev, INTEL_QUARK_I2C_NCLK,
+               sizeof(*i2c_clk_lookup), GFP_KERNEL);
+
+       if (!i2c_clk_lookup)
+               return -ENOMEM;
+
+       i2c_clk_lookup[0].dev_id = INTEL_QUARK_I2C_CONTROLLER_CLK;

drivers/mfd/intel_quark_i2c_gpio.c line 137

Quoted 10 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +10 -0
+
+       i2c_clk = clk_register_fixed_rate(
+               &pdev->dev, INTEL_QUARK_I2C_CONTROLLER_CLK, NULL,
+               CLK_IS_ROOT, INTEL_QUARK_I2C_CLK_HZ);
+
+       quark_mfd->i2c_clk_lookup = i2c_clk_lookup;
+       quark_mfd->i2c_clk = i2c_clk;
+
+       retval = clk_register_clkdevs(i2c_clk, i2c_clk_lookup,
+                                     INTEL_QUARK_I2C_NCLK);

Lee asked about this in V2, so I'll follow up here in V3. It is OK for a driver to use the clock provider api to register clocks with the clk framework if that device truly is the provider of that clock signal. A good example can be found here:

Quoted 6 lines

drivers/media/platform/omap3isp/isp.c

The OMAP3 ISP receives a clock signal as a input. Within the image signal processor IP block it also has some basic clock controls of it's own which it feeds to downstream IP blocks. As such it is both a clock consumer and a provider and this is a common pattern amongst SoC designs.

Thanks for the reference, however the mfd driver is purely a clk provider in this case.

Quoted 3 lines

So my question for this driver is if i2c_clk is provided by whatever the hell this mfd device is supposed to be, or if it's just a convenient place to call the code?

As you've noticed, this is a fixed clock which only consumed by the I2C controller. Following the structure of the designware i2c controller device driver, a clk is needed for it, and on this platform, it is a fixed clk.
I'm putting the clk functions in this mfd driver is due to the fact that, this mfd driver is splitting the function of the PCI device to 2 controllers downstream.

Quoted 9 lines

Another concern is that fact that this is a fixed clock. For architectures that use device tree to desribe board topology (ARM, MIPS,
PPC) it is common to simply put the fixed-rate clocks there and not directly into the drive code. This prevents having to hack a lot of conditionals into your driver when rev 2.0 of your hardware comes out with a faster fixed rate clock, but you still need to support 1.0 hardware users at the slower rate. I don't know if x86 has a similar way of describing board topology but it might something to look into.

I checked the kernel source for x86 arch, sadly there's no similar implementation of fixed clk being developed/written on the architectures code.
That being said, for this platform, we do have a separate platform board file for those onboard peripherals, do you think that it's better I put the clk function under the board file instead? My reasoning behind is if I were to introduce clk in general to x86 in this way, it's effect will be on x86 unless I introduce further checking during compilation / runtime.

Quoted 3 lines

Regards,
Mike

���{.n�+�������+%�����ݶ��w��{.n�+����{��G�����{ay�ʇڙ�,j��f���h��������z_�(�階�ݢj"���m������G������?����&���~��iO��z��v�^�m���� ������?�I�

#7 lore raw

Quoting Tan, Raymond (2014-12-21 18:33:42)

Quoted 97 lines

Hi Mike,

Thanks for your reply. I've answered the questions as below.

Warm Regards,

Raymond Tan

Quoted 60 lines

-----Original Message-----
From: Mike Turquette [mailto:[email protected]]
Sent: Friday, December 12, 2014 6:26 AM
To: Tan, Raymond; Lee Jones; Samuel Ortiz
Cc: [email protected]; Chen, Alvin; Shevchenko, Andriy; Tan, Raymond
Subject: Re: [PATCH v3 1/1] mfd: intel_quark_i2c_gpio: Add Intel Quark X1000 I2C-GPIO MFD Driver

Quoting Raymond Tan (2014-12-11 01:38:30)

Quoted 16 lines

In Quark X1000, there's a single PCI device that provides both an I2C controller and a GPIO controller. This MFD driver will split the 2 devices for their respective drivers.

This patch is based on Josef Ahmad's initial work for Quark enabling.

---
 drivers/mfd/Kconfig                |   12 ++
 drivers/mfd/Makefile               |    1 +
 drivers/mfd/intel_quark_i2c_gpio.c |  279
++++++++++++++++++++++++++++++++++++
 3 files changed, 292 insertions(+)
 create mode 100644 drivers/mfd/intel_quark_i2c_gpio.c

<snip>

Quoted 15 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +15 -0
+static int intel_quark_register_i2c_clk(struct intel_quark_mfd
+*quark_mfd) {
+       struct pci_dev *pdev = quark_mfd->pdev;
+       struct clk_lookup *i2c_clk_lookup;
+       struct clk *i2c_clk;
+       int retval;
+
+       i2c_clk_lookup = devm_kcalloc(
+               &pdev->dev, INTEL_QUARK_I2C_NCLK,
+               sizeof(*i2c_clk_lookup), GFP_KERNEL);
+
+       if (!i2c_clk_lookup)
+               return -ENOMEM;
+
+       i2c_clk_lookup[0].dev_id = INTEL_QUARK_I2C_CONTROLLER_CLK;

drivers/mfd/intel_quark_i2c_gpio.c line 137

Quoted 10 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +10 -0
+
+       i2c_clk = clk_register_fixed_rate(
+               &pdev->dev, INTEL_QUARK_I2C_CONTROLLER_CLK, NULL,
+               CLK_IS_ROOT, INTEL_QUARK_I2C_CLK_HZ);
+
+       quark_mfd->i2c_clk_lookup = i2c_clk_lookup;
+       quark_mfd->i2c_clk = i2c_clk;
+
+       retval = clk_register_clkdevs(i2c_clk, i2c_clk_lookup,
+                                     INTEL_QUARK_I2C_NCLK);

Lee asked about this in V2, so I'll follow up here in V3. It is OK for a driver to use the clock provider api to register clocks with the clk framework if that device truly is the provider of that clock signal. A good example can be found here:

Quoted 6 lines

drivers/media/platform/omap3isp/isp.c

The OMAP3 ISP receives a clock signal as a input. Within the image signal processor IP block it also has some basic clock controls of it's own which it feeds to downstream IP blocks. As such it is both a clock consumer and a provider and this is a common pattern amongst SoC designs.

Thanks for the reference, however the mfd driver is purely a clk provider in this case.

Quoted 3 lines

So my question for this driver is if i2c_clk is provided by whatever the hell this mfd device is supposed to be, or if it's just a convenient place to call the code?

As you've noticed, this is a fixed clock which only consumed by the I2C controller. Following the structure of the designware i2c controller device driver, a clk is needed for it, and on this platform, it is a fixed clk.
I'm putting the clk functions in this mfd driver is due to the fact that, this mfd driver is splitting the function of the PCI device to 2 controllers downstream.

Quoted 9 lines

Another concern is that fact that this is a fixed clock. For architectures that use device tree to desribe board topology (ARM, MIPS,
PPC) it is common to simply put the fixed-rate clocks there and not directly into the drive code. This prevents having to hack a lot of conditionals into your driver when rev 2.0 of your hardware comes out with a faster fixed rate clock, but you still need to support 1.0 hardware users at the slower rate. I don't know if x86 has a similar way of describing board topology but it might something to look into.

Quoted 7 lines

I checked the kernel source for x86 arch, sadly there's no similar implementation of fixed clk being developed/written on the architectures code.
That being said, for this platform, we do have a separate platform board file for those onboard peripherals, do you think that it's better I put the clk function under the board file instead? My reasoning behind is if I were to introduce clk in general to x86 in this way, it's effect will be on x86 unless I introduce further checking during compilation / runtime.

Thanks for the explanation. One final question, who consumes this clock?

The clk bits of the driver look good to me so please add my:

Thanks,
Mike

Quoted 4 lines
Quoted 4 lines

Regards,
Mike

#8 lore raw

Hi Mike,

Thanks for the acknowledgement. The clk will be consumed by the desginware i2c controller.

Warm Regards,

 Raymond Tan
Software Engineer
Malaysia IT Flex Services
INET: 8-253-0075
Flex Website: http://flexservices.intel.com

Quoted 126 lines

-----Original Message-----
From: Mike Turquette [mailto:[email protected]]
Sent: Wednesday, January 21, 2015 1:30 AM
To: Tan, Raymond; Lee Jones; Samuel Ortiz
Cc: [email protected]; Chen, Alvin; Shevchenko, Andriy; Tan, Raymond
Subject: RE: [PATCH v3 1/1] mfd: intel_quark_i2c_gpio: Add Intel Quark X1000 I2C-GPIO MFD Driver

Quoting Tan, Raymond (2014-12-21 18:33:42)

Quoted 76 lines

Hi Mike,

Thanks for your reply. I've answered the questions as below.

Warm Regards,

Raymond Tan

Quoted 68 lines

-----Original Message-----
From: Mike Turquette [mailto:[email protected]]
Sent: Friday, December 12, 2014 6:26 AM
To: Tan, Raymond; Lee Jones; Samuel Ortiz
Cc: [email protected]; Chen, Alvin; Shevchenko, Andriy; Tan, Raymond
Subject: Re: [PATCH v3 1/1] mfd: intel_quark_i2c_gpio: Add Intel Quark
X1000 I2C-GPIO MFD Driver

Quoting Raymond Tan (2014-12-11 01:38:30)

Quoted 16 lines

In Quark X1000, there's a single PCI device that provides both an I2C controller and a GPIO controller. This MFD driver will split the 2 devices for their respective drivers.

This patch is based on Josef Ahmad's initial work for Quark enabling.

---
 drivers/mfd/Kconfig                |   12 ++
 drivers/mfd/Makefile               |    1 +
 drivers/mfd/intel_quark_i2c_gpio.c |  279
++++++++++++++++++++++++++++++++++++
 3 files changed, 292 insertions(+)  create mode 100644

drivers/mfd/intel_quark_i2c_gpio.c

<snip>

Quoted 15 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +15 -0
+static int intel_quark_register_i2c_clk(struct intel_quark_mfd
+*quark_mfd) {
+       struct pci_dev *pdev = quark_mfd->pdev;
+       struct clk_lookup *i2c_clk_lookup;
+       struct clk *i2c_clk;
+       int retval;
+
+       i2c_clk_lookup = devm_kcalloc(
+               &pdev->dev, INTEL_QUARK_I2C_NCLK,
+               sizeof(*i2c_clk_lookup), GFP_KERNEL);
+
+       if (!i2c_clk_lookup)
+               return -ENOMEM;
+
+       i2c_clk_lookup[0].dev_id = INTEL_QUARK_I2C_CONTROLLER_CLK;

drivers/mfd/intel_quark_i2c_gpio.c line 137

Quoted 10 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +10 -0
+
+       i2c_clk = clk_register_fixed_rate(
+               &pdev->dev, INTEL_QUARK_I2C_CONTROLLER_CLK, NULL,
+               CLK_IS_ROOT, INTEL_QUARK_I2C_CLK_HZ);
+
+       quark_mfd->i2c_clk_lookup = i2c_clk_lookup;
+       quark_mfd->i2c_clk = i2c_clk;
+
+       retval = clk_register_clkdevs(i2c_clk, i2c_clk_lookup,
+                                     INTEL_QUARK_I2C_NCLK);

Lee asked about this in V2, so I'll follow up here in V3. It is OK for a driver to use the clock provider api to register clocks with the clk framework if that device truly is the provider of that clock signal. A good example can be found
here:

drivers/media/platform/omap3isp/isp.c

The OMAP3 ISP receives a clock signal as a input. Within the image signal processor IP block it also has some basic clock controls of it's own which it feeds to downstream IP blocks. As such it is both a clock consumer and a provider and this is a common pattern amongst

SoC designs.

Quoted 2 lines

Thanks for the reference, however the mfd driver is purely a clk provider in

this case.

Quoted 4 lines
Quoted 4 lines

So my question for this driver is if i2c_clk is provided by whatever the hell this mfd device is supposed to be, or if it's just a convenient place

to call the code?

Quoted 2 lines

As you've noticed, this is a fixed clock which only consumed by the I2C

controller.

Quoted 4 lines

Following the structure of the designware i2c controller device driver, a clk is needed for it, and on this platform, it is a fixed clk. I'm putting the clk functions in this mfd driver is due to the fact that, this mfd driver is splitting the function of the PCI device to 2

controllers downstream.

Quoted 11 lines
Quoted 11 lines

Another concern is that fact that this is a fixed clock. For architectures that use device tree to desribe board topology (ARM, MIPS,
PPC) it is common to simply put the fixed-rate clocks there and not directly into the drive code. This prevents having to hack a lot of conditionals into your driver when rev 2.0 of your hardware comes out with a faster fixed rate clock, but you still need to support 1.0 hardware users at the slower rate. I don't know if x86 has a similar way of describing board topology but it might something to look

into.

Quoted 3 lines

I checked the kernel source for x86 arch, sadly there's no similar implementation of fixed clk being developed/written on the architectures

code.

Quoted 5 lines

That being said, for this platform, we do have a separate platform board file for those onboard peripherals, do you think that it's better I put the clk function under the board file instead? My reasoning behind is if I were to introduce clk in general to x86 in this way, it's effect will be on x86 unless I introduce further checking during

compilation / runtime.

Quoted 13 lines

Thanks for the explanation. One final question, who consumes this clock?

The clk bits of the driver look good to me so please add my:

Thanks,
Mike

Quoted 4 lines
Quoted 4 lines

Regards,
Mike

���{.n�+�������+%�����ݶ��w��{.n�+����{��G�����{ay�ʇڙ�,j��f���h��������z_�(�階�ݢj"���m������G������?����&���~��iO��z��v�^�m���� ������?�I�

#9 lore raw
In reply to Raymond Tan
Quoted 120 lines · Raymond Tan

In Quark X1000, there's a single PCI device that provides both an I2C controller and a GPIO controller. This MFD driver will split the 2 devices for their respective drivers.

This patch is based on Josef Ahmad's initial work for Quark enabling.

---
 drivers/mfd/Kconfig                |   12 ++
 drivers/mfd/Makefile               |    1 +
 drivers/mfd/intel_quark_i2c_gpio.c |  279 ++++++++++++++++++++++++++++++++++++
 3 files changed, 292 insertions(+)
 create mode 100644 drivers/mfd/intel_quark_i2c_gpio.c
drivers/mfd/Kconfig +12 -0
index b7c74a7..2f2f62d 100644
@@ -219,6 +219,18 @@ config HTC_I2CPLD
 	  This device provides input and output GPIOs through an I2C
 	  interface to one or more sub-chips.
 
+config MFD_INTEL_QUARK_I2C_GPIO
+	tristate "Intel Quark MFD I2C GPIO"
+	depends on PCI
+	depends on X86
+	depends on COMMON_CLK
+	select MFD_CORE
+	help
+	  This MFD provides support for I2C and GPIO that exist only
+	  in a single PCI device. It splits the 2 IO devices to
+	  their respective IO driver.
+	  The GPIO exports a total amount of 8 interrupt-capable GPIOs.
+
 config LPC_ICH
 	tristate "Intel ICH LPC"
 	depends on PCI
drivers/mfd/Makefile +1 -0
index 8a28dc9..d42652d 100644
@@ -133,6 +133,7 @@ obj-$(CONFIG_AB8500_CORE)	+= ab8500-core.o ab8500-sysctrl.o
 obj-$(CONFIG_MFD_TIMBERDALE)    += timberdale.o
 obj-$(CONFIG_PMIC_ADP5520)	+= adp5520.o
 obj-$(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
 obj-$(CONFIG_LPC_ICH)		+= lpc_ich.o
 obj-$(CONFIG_MFD_RDC321X)	+= rdc321x-southbridge.o
drivers/mfd/intel_quark_i2c_gpio.c new file +63 -0
new file mode 100644
index 0000000..a3a7043
@@ -0,0 +1,279 @@
+/*
+ * Intel Quark MFD PCI driver for I2C & GPIO
+ *
+ * Copyright(c) 2014 Intel Corporation.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * Intel Quark PCI device for I2C and GPIO controller sharing the same
+ * PCI function. This PCI driver will split the 2 devices into their
+ * respective drivers.
+ */
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/pci.h>
+#include <linux/mfd/core.h>
+#include <linux/clkdev.h>
+#include <linux/clk-provider.h>
+#include <linux/dmi.h>
+#include <linux/platform_data/gpio-dwapb.h>
+#include <linux/platform_data/i2c-designware.h>
+
+/* PCI BAR for register base address */
+#define MFD_I2C_BAR		0
+#define MFD_GPIO_BAR		1
+
+/* The base GPIO number under GPIOLIB framework */
+#define INTEL_QUARK_MFD_GPIO_BASE	8
+
+/* The default number of South-Cluster GPIO on Quark. */
+#define INTEL_QUARK_MFD_NGPIO		8
+
+/* The DesignWare GPIO ports on Quark. */
+#define INTEL_QUARK_GPIO_NPORTS	1
+
+#define INTEL_QUARK_IORES_MEM	0
+#define INTEL_QUARK_IORES_IRQ	1
+
+#define INTEL_QUARK_I2C_CONTROLLER_CLK "i2c_designware.0"
+
+/* The Quark I2C controller source clock */
+#define INTEL_QUARK_I2C_CLK_HZ	33000000
+
+#define INTEL_QUARK_I2C_NCLK	1
+
+struct intel_quark_mfd {
+	struct pci_dev		*pdev;
+	struct clk			*i2c_clk;
+	struct clk_lookup	*i2c_clk_lookup;
+};
+
+struct i2c_mode_info {
+	const char *name;
+	unsigned int i2c_scl_freq;
+};
+

drivers/mfd/intel_quark_i2c_gpio.c line 73

Quoted 10 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +10 -0
+static const struct i2c_mode_info platform_i2c_mode_info[] = {
+	{
+		.name = "Galileo",
+		.i2c_scl_freq = 100000,
+	},
+	{
+		.name = "GalileoGen2",
+		.i2c_scl_freq = 400000,
+	},
+};

We normally do not capitalise device names.

Quoted 25 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +25 -0
+static struct resource intel_quark_i2c_res[] = {
+	[INTEL_QUARK_IORES_MEM] = {
+		.flags = IORESOURCE_MEM,
+	},
+	[INTEL_QUARK_IORES_IRQ] = {
+		.flags = IORESOURCE_IRQ,
+	},
+};
+
+static struct resource intel_quark_gpio_res[] = {
+	[INTEL_QUARK_IORES_MEM] = {
+		.flags = IORESOURCE_MEM,
+	},
+};
+
+static struct mfd_cell intel_quark_mfd_cells[] = {
+	{
+		.id = MFD_I2C_BAR,
+		.name = "i2c_designware",
+		.num_resources = ARRAY_SIZE(intel_quark_i2c_res),
+		.resources = intel_quark_i2c_res,
+		.ignore_resource_conflicts = true,
+	},
+	{
+		.id = MFD_GPIO_BAR,

drivers/mfd/intel_quark_i2c_gpio.c line 109

Quoted 10 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +10 -0
+		.name = "gpio-dwapb",
+		.num_resources = ARRAY_SIZE(intel_quark_gpio_res),
+		.resources = intel_quark_gpio_res,
+		.ignore_resource_conflicts = true,
+	},
+};
+
+static const struct pci_device_id intel_quark_mfd_ids[] = {
+	{ PCI_VDEVICE(INTEL, 0x0934), },
+	{ 0,}

No need to populate this; just { }, will do fine.

drivers/mfd/intel_quark_i2c_gpio.c line 121

Quoted 12 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +12 -0
+};
+MODULE_DEVICE_TABLE(pci, intel_quark_mfd_ids);
+
+static int intel_quark_register_i2c_clk(struct intel_quark_mfd *quark_mfd)
+{
+	struct pci_dev *pdev = quark_mfd->pdev;
+	struct clk_lookup *i2c_clk_lookup;
+	struct clk *i2c_clk;
+	int retval;
+
+	i2c_clk_lookup = devm_kcalloc(
+		&pdev->dev, INTEL_QUARK_I2C_NCLK,

Place this line back on the one above and line subsequent lines up against the '('.

drivers/mfd/intel_quark_i2c_gpio.c line 122

Quoted 2 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +2 -0
+		sizeof(*i2c_clk_lookup), GFP_KERNEL);
+

Rid this empty line.

drivers/mfd/intel_quark_i2c_gpio.c line 130

Quoted 7 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +7 -0
+	if (!i2c_clk_lookup)
+		return -ENOMEM;
+
+	i2c_clk_lookup[0].dev_id = INTEL_QUARK_I2C_CONTROLLER_CLK;
+
+	i2c_clk = clk_register_fixed_rate(
+		&pdev->dev, INTEL_QUARK_I2C_CONTROLLER_CLK, NULL,

Same here, put this back up on the line above.

drivers/mfd/intel_quark_i2c_gpio.c line 142

Quoted 12 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +12 -0
+		CLK_IS_ROOT, INTEL_QUARK_I2C_CLK_HZ);
+
+	quark_mfd->i2c_clk_lookup = i2c_clk_lookup;
+	quark_mfd->i2c_clk = i2c_clk;
+
+	retval = clk_register_clkdevs(i2c_clk, i2c_clk_lookup,
+				      INTEL_QUARK_I2C_NCLK);
+	if (retval)
+		dev_err(&pdev->dev, "Fixed clk register failed: %d\n", retval);
+
+	return retval;
+}

I wouldn't mind a clk Ack on this. Did you Cc Mike already?

Quoted 66 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +66 -0
+static void intel_quark_unregister_i2c_clk(struct pci_dev *pdev)
+{
+	struct intel_quark_mfd *quark_mfd = dev_get_drvdata(&pdev->dev);
+
+	if (!quark_mfd->i2c_clk || !quark_mfd->i2c_clk_lookup)
+		return;
+
+	clkdev_drop(quark_mfd->i2c_clk_lookup);
+	clk_unregister(quark_mfd->i2c_clk);
+}
+
+static int intel_quark_i2c_setup(struct pci_dev *pdev, struct mfd_cell *cell)
+{
+	const char *board_name = dmi_get_system_info(DMI_BOARD_NAME);
+	struct dw_i2c_platform_data *pdata;
+	struct resource *res = (struct resource *)cell->resources;
+	struct device *dev = &pdev->dev;
+	unsigned int i;
+
+	res[INTEL_QUARK_IORES_MEM].start =
+		pci_resource_start(pdev, MFD_I2C_BAR);
+	res[INTEL_QUARK_IORES_MEM].end =
+		pci_resource_end(pdev, MFD_I2C_BAR);
+
+	res[INTEL_QUARK_IORES_IRQ].start = pdev->irq;
+	res[INTEL_QUARK_IORES_IRQ].end = pdev->irq;
+
+	pdata = devm_kzalloc(dev, sizeof(*pdata), GFP_KERNEL);
+	if (!pdata)
+		return -ENOMEM;
+
+	/* Fast mode by default */
+	pdata->i2c_scl_freq = 400000;
+
+	for (i = 0; i < ARRAY_SIZE(platform_i2c_mode_info); i++)
+		if (!strcmp(board_name, platform_i2c_mode_info[i].name))
+			pdata->i2c_scl_freq
+				= platform_i2c_mode_info[i].i2c_scl_freq;
+
+	cell->platform_data = pdata;
+	cell->pdata_size = sizeof(*pdata);
+
+	return 0;
+}
+
+static int intel_quark_gpio_setup(struct pci_dev *pdev, struct mfd_cell *cell)
+{
+	struct dwapb_platform_data *pdata;
+	struct resource *res = (struct resource *)cell->resources;
+	struct device *dev = &pdev->dev;
+
+	res[INTEL_QUARK_IORES_MEM].start =
+		pci_resource_start(pdev, MFD_GPIO_BAR);
+	res[INTEL_QUARK_IORES_MEM].end =
+		pci_resource_end(pdev, MFD_GPIO_BAR);
+
+	pdata = devm_kzalloc(dev, sizeof(*pdata), GFP_KERNEL);
+	if (!pdata)
+		return -ENOMEM;
+
+	/* For intel quark x1000, it has only one port: portA */
+	pdata->nports = INTEL_QUARK_GPIO_NPORTS;
+	pdata->properties = devm_kcalloc(dev, pdata->nports,
+					 sizeof(*pdata->properties),
+					 GFP_KERNEL);
+	if (!pdata->properties)

drivers/mfd/intel_quark_i2c_gpio.c line 219

Quoted 10 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +10 -0
+		return -ENOMEM;
+
+	/* Set the properties for portA */
+	pdata->properties->node	= NULL;
+	pdata->properties->name	= "intel-quark-x1000-gpio-portA";
+	pdata->properties->idx	= 0;
+	pdata->properties->ngpio	= INTEL_QUARK_MFD_NGPIO;
+	pdata->properties->gpio_base	= INTEL_QUARK_MFD_GPIO_BASE;
+	pdata->properties->irq	= pdev->irq;
+	pdata->properties->irq_shared	= true;

Nit: This looks untidy. Either line up all of the '=' or don't do any aligning at all.

drivers/mfd/intel_quark_i2c_gpio.c line 231

Quoted 11 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +11 -0
+	cell->platform_data = pdata;
+	cell->pdata_size = sizeof(*pdata);
+
+	return 0;
+}
+
+static int intel_quark_mfd_probe(struct pci_dev *pdev,
+				 const struct pci_device_id *id)
+{
+	struct intel_quark_mfd *quark_mfd;
+	int retval;

I prefer the use of 'ret' or 'err', as it's more consistent with the rest of the kernel.

Quoted 17 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +17 -0
+	retval = pcim_enable_device(pdev);
+	if (retval)
+		return retval;
+
+	quark_mfd = devm_kzalloc(&pdev->dev, sizeof(*quark_mfd), GFP_KERNEL);
+	if (!quark_mfd)
+		return -ENOMEM;
+	quark_mfd->pdev = pdev;
+
+	retval = intel_quark_register_i2c_clk(quark_mfd);
+	if (retval)
+		return retval;
+
+	dev_set_drvdata(&pdev->dev, quark_mfd);
+
+	retval = intel_quark_i2c_setup(pdev,
+				       &intel_quark_mfd_cells[MFD_I2C_BAR]);

drivers/mfd/intel_quark_i2c_gpio.c line 259

Quoted 10 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +10 -0
+	if (retval)
+		return retval;
+
+	retval = intel_quark_gpio_setup(pdev,
+					&intel_quark_mfd_cells[MFD_GPIO_BAR]);
+	if (retval)
+		return retval;
+
+	return mfd_add_devices(&pdev->dev, 0, intel_quark_mfd_cells,
+		ARRAY_SIZE(intel_quark_mfd_cells), NULL, 0, NULL);

Odd alignment. Please use the '('.

Quoted 20 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +20 -0
+}
+
+static void intel_quark_mfd_remove(struct pci_dev *pdev)
+{
+	intel_quark_unregister_i2c_clk(pdev);
+	mfd_remove_devices(&pdev->dev);
+}
+
+static struct pci_driver intel_quark_mfd_driver = {
+	.name		= "intel_quark_mfd_i2c_gpio",
+	.id_table	= intel_quark_mfd_ids,
+	.probe		= intel_quark_mfd_probe,
+	.remove		= intel_quark_mfd_remove,
+};
+
+module_pci_driver(intel_quark_mfd_driver);
+
+MODULE_AUTHOR("Raymond Tan <[email protected]>");
+MODULE_DESCRIPTION("Intel Quark MFD PCI driver for I2C & GPIO");
+MODULE_LICENSE("GPL v2");
Signature · 4 lines
Lee Jones
Linaro STMicroelectronics Landing Team Lead
Linaro.org │ Open source software for ARM SoCs
Follow Linaro: Facebook | Twitter | Blog
#10 lore raw
Quoted 1 line · Lee Jones

On Thu, 11 Dec 2014, Raymond Tan wrote:

[]

Quoted 4 lines
Quoted 4 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +4 -0
+static const struct i2c_mode_info platform_i2c_mode_info[] = {
+	{
+		.name = "Galileo",
+		.i2c_scl_freq = 100000,
Quoted 8 lines

drivers/mfd/intel_quark_i2c_gpio.c line 73

Quoted 6 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +6 -0
+	},
+	{
+		.name = "GalileoGen2",
+		.i2c_scl_freq = 400000,
+	},
+};

We normally do not capitalise device names.

This comes from DMI, so, can't be altered I suppose.

Signature · 13 lines
Andy Shevchenko <[email protected]>
Intel Finland Oy
---------------------------------------------------------------------
Intel Finland Oy
Registered Address: PL 281, 00181 Helsinki 
Business Identity Code: 0357606 - 4 
Domiciled in Helsinki 

This e-mail and any attachments may contain confidential material for
the sole use of the intended recipient(s). Any review or distribution
by others is strictly prohibited. If you are not the intended
recipient, please contact the sender and delete all copies.
���{.n�+�������+%�����ݶ��w��{.n�+����{��G�����{ay�ʇڙ�,j��f���h��������z_�(�階�ݢj"���m������G������?����&���~��iO��z��v�^�m����������?�I�
#11 lore raw
Quoted 18 lines · Andriy
Quoted 1 line · Lee Jones

On Thu, 11 Dec 2014, Raymond Tan wrote:

[]

Quoted 4 lines
Quoted 4 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +4 -0
+static const struct i2c_mode_info platform_i2c_mode_info[] = {
+	{
+		.name = "Galileo",
+		.i2c_scl_freq = 100000,
Quoted 8 lines

drivers/mfd/intel_quark_i2c_gpio.c line 73

Quoted 6 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +6 -0
+	},
+	{
+		.name = "GalileoGen2",
+		.i2c_scl_freq = 400000,
+	},
+};

We normally do not capitalise device names.

Quoted 1 line

This comes from DMI, so, can't be altered I suppose.

I'm not sure what that means, but sounds reasonable.

Signature · 4 lines
Lee Jones
Linaro STMicroelectronics Landing Team Lead
Linaro.org │ Open source software for ARM SoCs
Follow Linaro: Facebook | Twitter | Blog
#12 lore raw
Quoted 20 lines · Lee Jones
Quoted 18 lines · Andriy
Quoted 1 line · Lee Jones

On Thu, 11 Dec 2014, Raymond Tan wrote:

[]

Quoted 12 lines

drivers/mfd/intel_quark_i2c_gpio.c line 73

Quoted 10 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +10 -0
+static const struct i2c_mode_info platform_i2c_mode_info[] = {
+	{
+		.name = "Galileo",
+		.i2c_scl_freq = 100000,
+	},
+	{
+		.name = "GalileoGen2",
+		.i2c_scl_freq = 400000,
+	},
+};

We normally do not capitalise device names.

Quoted 3 lines
Quoted 1 line

This comes from DMI, so, can't be altered I suppose.

I'm not sure what that means, but sounds reasonable.

http://en.wikipedia.org/wiki/Desktop_Management_Interface

In short, it comes from firmware that's why we have to use exact names here.

Signature · 13 lines
Andy Shevchenko <[email protected]>
Intel Finland Oy
---------------------------------------------------------------------
Intel Finland Oy
Registered Address: PL 281, 00181 Helsinki 
Business Identity Code: 0357606 - 4 
Domiciled in Helsinki 

This e-mail and any attachments may contain confidential material for
the sole use of the intended recipient(s). Any review or distribution
by others is strictly prohibited. If you are not the intended
recipient, please contact the sender and delete all copies.
���{.n�+�������+%�����ݶ��w��{.n�+����{��G�����{ay�ʇڙ�,j��f���h��������z_�(�階�ݢj"���m������G������?����&���~��iO��z��v�^�m����������?�I�
#13 lore raw
In reply to Lee Jones

Quoting Lee Jones (2015-01-20 04:47:46)

Quoted 7 lines · Raymond Tan
Quoted 6 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +6 -0
+             CLK_IS_ROOT, INTEL_QUARK_I2C_CLK_HZ);
+
+     quark_mfd->i2c_clk_lookup = i2c_clk_lookup;
+     quark_mfd->i2c_clk = i2c_clk;
+
+     retval = clk_register_clkdevs(i2c_clk, i2c_clk_lookup,
Quoted 8 lines

drivers/mfd/intel_quark_i2c_gpio.c line 142

Quoted 6 lines
drivers/mfd/intel_quark_i2c_gpio.c new file +6 -0
+                                   INTEL_QUARK_I2C_NCLK);
+     if (retval)
+             dev_err(&pdev->dev, "Fixed clk register failed: %d\n", retval);
+
+     return retval;
+}

I wouldn't mind a clk Ack on this. Did you Cc Mike already?

Yes he did and I just supplied it in a parallel thread.

Regards,
Mike