diff mbox series

[v4,6/6] gpio: uniphier: add UniPhier GPIO controller driver

Message ID 1504784522-26841-7-git-send-email-yamada.masahiro@socionext.com
State New
Headers show
Series irqdomain, gpio: expand irq_domain_push_irq() for DT use and use it for GPIO | expand

Commit Message

Masahiro Yamada Sept. 7, 2017, 11:42 a.m. UTC
This GPIO controller device is used on UniPhier SoCs.

It also serves as an interrupt controller, but interrupt signals are
just delivered to the parent irqchip without any latching or OR'ing.
This is implemented by using hierarchy IRQ domain.

Implementation note:
Unfortunately, the IRQ mapping from this controller to the parent is
random. (48, 49, ..., 63, 154, 155, ...)
If "interrupts" property is used, IRQ resources may be statically
allocated when platform devices are populated from DT.  This can be
a problem for the hierarchy IRQ domain because IRQ allocation must
happen from the outer-most domain up to the root domain in order to
build up the stacked IRQ.  (https://lkml.org/lkml/2017/7/6/758)
Solutions to work around it could be to hard-code parent hwirqs or
to invent a driver-specific DT property.

Here, the new API irq_domain_push_irq() was merged by v4.14-rc1.
It allows to add irq_data to the existing hierarchy.  It will help
to make this driver work whether the parent has already initialized
the hierarchy or not.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
---

Changes in v4:
  - Add COMPILE_TEST and select IRQ_DOMAIN_HIERARCHY
  - Reimplement irqchip part by using irq_domain_push_irq()

Changes in v3:
  - Add .irq_set_affinity() hook
  - Use irq_domain_create_hierarchy() instead of legacy
    irq_domain_add_hierarchy()

Changes in v2:
  - Remove +32 offset for parent interrupts to follow the GIC
    binding convention
  - Let uniphier_gpio_irq_alloc() fail if nr_irqs != 1
  - Allocate gpio_chip statically because just one instance is
    supported
  - Fix suspend and resume hooks

 .../devicetree/bindings/gpio/gpio-uniphier.txt     |  43 ++
 MAINTAINERS                                        |   1 +
 drivers/gpio/Kconfig                               |   8 +
 drivers/gpio/Makefile                              |   1 +
 drivers/gpio/gpio-uniphier.c                       | 486 +++++++++++++++++++++
 include/dt-bindings/gpio/uniphier-gpio.h           |  18 +
 6 files changed, 557 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/gpio/gpio-uniphier.txt
 create mode 100644 drivers/gpio/gpio-uniphier.c
 create mode 100644 include/dt-bindings/gpio/uniphier-gpio.h

Comments

Rob Herring Sept. 7, 2017, 7:41 p.m. UTC | #1
On Thu, Sep 7, 2017 at 6:42 AM, Masahiro Yamada
<yamada.masahiro@socionext.com> wrote:
> This GPIO controller device is used on UniPhier SoCs.
>
> It also serves as an interrupt controller, but interrupt signals are
> just delivered to the parent irqchip without any latching or OR'ing.
> This is implemented by using hierarchy IRQ domain.
>
> Implementation note:
> Unfortunately, the IRQ mapping from this controller to the parent is
> random. (48, 49, ..., 63, 154, 155, ...)
> If "interrupts" property is used, IRQ resources may be statically
> allocated when platform devices are populated from DT.  This can be
> a problem for the hierarchy IRQ domain because IRQ allocation must
> happen from the outer-most domain up to the root domain in order to
> build up the stacked IRQ.  (https://lkml.org/lkml/2017/7/6/758)
> Solutions to work around it could be to hard-code parent hwirqs or
> to invent a driver-specific DT property.
>
> Here, the new API irq_domain_push_irq() was merged by v4.14-rc1.
> It allows to add irq_data to the existing hierarchy.  It will help
> to make this driver work whether the parent has already initialized
> the hierarchy or not.
>
> Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
> ---
>
> Changes in v4:
>   - Add COMPILE_TEST and select IRQ_DOMAIN_HIERARCHY
>   - Reimplement irqchip part by using irq_domain_push_irq()
>
> Changes in v3:
>   - Add .irq_set_affinity() hook
>   - Use irq_domain_create_hierarchy() instead of legacy
>     irq_domain_add_hierarchy()
>
> Changes in v2:
>   - Remove +32 offset for parent interrupts to follow the GIC
>     binding convention
>   - Let uniphier_gpio_irq_alloc() fail if nr_irqs != 1
>   - Allocate gpio_chip statically because just one instance is
>     supported
>   - Fix suspend and resume hooks
>
>  .../devicetree/bindings/gpio/gpio-uniphier.txt     |  43 ++

What happened to my ack? One line here more that before, but I'm not
going to diff your patches for you.

BTW, it is preferred to split bindings to a separate patch.

>  MAINTAINERS                                        |   1 +
>  drivers/gpio/Kconfig                               |   8 +
>  drivers/gpio/Makefile                              |   1 +
>  drivers/gpio/gpio-uniphier.c                       | 486 +++++++++++++++++++++
>  include/dt-bindings/gpio/uniphier-gpio.h           |  18 +
>  6 files changed, 557 insertions(+)
>  create mode 100644 Documentation/devicetree/bindings/gpio/gpio-uniphier.txt
>  create mode 100644 drivers/gpio/gpio-uniphier.c
>  create mode 100644 include/dt-bindings/gpio/uniphier-gpio.h
--
To unsubscribe from this list: send the line "unsubscribe linux-gpio" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Masahiro Yamada Sept. 8, 2017, 3:14 p.m. UTC | #2
Hi Rob,


2017-09-08 4:41 GMT+09:00 Rob Herring <robh+dt@kernel.org>:
> On Thu, Sep 7, 2017 at 6:42 AM, Masahiro Yamada
> <yamada.masahiro@socionext.com> wrote:
>> This GPIO controller device is used on UniPhier SoCs.
>>
>> It also serves as an interrupt controller, but interrupt signals are
>> just delivered to the parent irqchip without any latching or OR'ing.
>> This is implemented by using hierarchy IRQ domain.
>>
>> Implementation note:
>> Unfortunately, the IRQ mapping from this controller to the parent is
>> random. (48, 49, ..., 63, 154, 155, ...)
>> If "interrupts" property is used, IRQ resources may be statically
>> allocated when platform devices are populated from DT.  This can be
>> a problem for the hierarchy IRQ domain because IRQ allocation must
>> happen from the outer-most domain up to the root domain in order to
>> build up the stacked IRQ.  (https://lkml.org/lkml/2017/7/6/758)
>> Solutions to work around it could be to hard-code parent hwirqs or
>> to invent a driver-specific DT property.
>>
>> Here, the new API irq_domain_push_irq() was merged by v4.14-rc1.
>> It allows to add irq_data to the existing hierarchy.  It will help
>> to make this driver work whether the parent has already initialized
>> the hierarchy or not.
>>
>> Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
>> ---
>>
>> Changes in v4:
>>   - Add COMPILE_TEST and select IRQ_DOMAIN_HIERARCHY
>>   - Reimplement irqchip part by using irq_domain_push_irq()
>>
>> Changes in v3:
>>   - Add .irq_set_affinity() hook
>>   - Use irq_domain_create_hierarchy() instead of legacy
>>     irq_domain_add_hierarchy()
>>
>> Changes in v2:
>>   - Remove +32 offset for parent interrupts to follow the GIC
>>     binding convention
>>   - Let uniphier_gpio_irq_alloc() fail if nr_irqs != 1
>>   - Allocate gpio_chip statically because just one instance is
>>     supported
>>   - Fix suspend and resume hooks
>>
>>  .../devicetree/bindings/gpio/gpio-uniphier.txt     |  43 ++
>
> What happened to my ack? One line here more that before, but I'm not
> going to diff your patches for you.

I changed the binding for this version.
It includes a new one you have not acked yet.

Maybe I was too worried about it.


> BTW, it is preferred to split bindings to a separate patch.

I will do so next time.
kernel test robot Sept. 10, 2017, 12:13 p.m. UTC | #3
Hi Masahiro,

[auto build test ERROR on tip/irq/core]
[also build test ERROR on next-20170908]
[cannot apply to v4.13]
[if your patch is applied to the wrong git tree, please drop us a note to help improve the system]

url:    https://github.com/0day-ci/linux/commits/Masahiro-Yamada/irqdomain-gpio-expand-irq_domain_push_irq-for-DT-use-and-use-it-for-GPIO/20170910-160507
config: i386-allmodconfig (attached as .config)
compiler: gcc-6 (Debian 6.2.0-3) 6.2.0 20160901
reproduce:
        # save the attached .config to linux build tree
        make ARCH=i386 

All errors (new ones prefixed by >>):

>> ERROR: "of_irq_count" [drivers/gpio/gpio-uniphier.ko] undefined!

---
0-DAY kernel test infrastructure                Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all                   Intel Corporation
Rob Herring Sept. 11, 2017, 8:15 p.m. UTC | #4
On Sat, Sep 09, 2017 at 12:14:45AM +0900, Masahiro Yamada wrote:
> Hi Rob,
> 
> 
> 2017-09-08 4:41 GMT+09:00 Rob Herring <robh+dt@kernel.org>:
> > On Thu, Sep 7, 2017 at 6:42 AM, Masahiro Yamada
> > <yamada.masahiro@socionext.com> wrote:
> >> This GPIO controller device is used on UniPhier SoCs.
> >>
> >> It also serves as an interrupt controller, but interrupt signals are
> >> just delivered to the parent irqchip without any latching or OR'ing.
> >> This is implemented by using hierarchy IRQ domain.
> >>
> >> Implementation note:
> >> Unfortunately, the IRQ mapping from this controller to the parent is
> >> random. (48, 49, ..., 63, 154, 155, ...)
> >> If "interrupts" property is used, IRQ resources may be statically
> >> allocated when platform devices are populated from DT.  This can be
> >> a problem for the hierarchy IRQ domain because IRQ allocation must
> >> happen from the outer-most domain up to the root domain in order to
> >> build up the stacked IRQ.  (https://lkml.org/lkml/2017/7/6/758)
> >> Solutions to work around it could be to hard-code parent hwirqs or
> >> to invent a driver-specific DT property.
> >>
> >> Here, the new API irq_domain_push_irq() was merged by v4.14-rc1.
> >> It allows to add irq_data to the existing hierarchy.  It will help
> >> to make this driver work whether the parent has already initialized
> >> the hierarchy or not.
> >>
> >> Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
> >> ---
> >>
> >> Changes in v4:
> >>   - Add COMPILE_TEST and select IRQ_DOMAIN_HIERARCHY
> >>   - Reimplement irqchip part by using irq_domain_push_irq()
> >>
> >> Changes in v3:
> >>   - Add .irq_set_affinity() hook
> >>   - Use irq_domain_create_hierarchy() instead of legacy
> >>     irq_domain_add_hierarchy()
> >>
> >> Changes in v2:
> >>   - Remove +32 offset for parent interrupts to follow the GIC
> >>     binding convention
> >>   - Let uniphier_gpio_irq_alloc() fail if nr_irqs != 1
> >>   - Allocate gpio_chip statically because just one instance is
> >>     supported
> >>   - Fix suspend and resume hooks
> >>
> >>  .../devicetree/bindings/gpio/gpio-uniphier.txt     |  43 ++
> >
> > What happened to my ack? One line here more that before, but I'm not
> > going to diff your patches for you.
> 
> I changed the binding for this version.
> It includes a new one you have not acked yet.
> 
> Maybe I was too worried about it.

Probably so, but I still don't know what you changed.

Rob
--
To unsubscribe from this list: send the line "unsubscribe linux-gpio" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Linus Walleij Sept. 12, 2017, 2:03 p.m. UTC | #5
On Thu, Sep 7, 2017 at 1:42 PM, Masahiro Yamada
<yamada.masahiro@socionext.com> wrote:

> This GPIO controller device is used on UniPhier SoCs.
>
> It also serves as an interrupt controller, but interrupt signals are
> just delivered to the parent irqchip without any latching or OR'ing.
> This is implemented by using hierarchy IRQ domain.
>
> Implementation note:
> Unfortunately, the IRQ mapping from this controller to the parent is
> random. (48, 49, ..., 63, 154, 155, ...)
> If "interrupts" property is used, IRQ resources may be statically
> allocated when platform devices are populated from DT.  This can be
> a problem for the hierarchy IRQ domain because IRQ allocation must
> happen from the outer-most domain up to the root domain in order to
> build up the stacked IRQ.  (https://lkml.org/lkml/2017/7/6/758)
> Solutions to work around it could be to hard-code parent hwirqs or
> to invent a driver-specific DT property.
>
> Here, the new API irq_domain_push_irq() was merged by v4.14-rc1.
> It allows to add irq_data to the existing hierarchy.  It will help
> to make this driver work whether the parent has already initialized
> the hierarchy or not.
>
> Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
> ---
>
> Changes in v4:
>   - Add COMPILE_TEST and select IRQ_DOMAIN_HIERARCHY
>   - Reimplement irqchip part by using irq_domain_push_irq()

Awesome improvement. There was a build error and I also
would like David Daney to have a look at this so we know we
use things the right way, but overall I am happy after this
so I hope I will be able to apply next version.

Yours,
Linus Walleij
--
To unsubscribe from this list: send the line "unsubscribe linux-gpio" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
David Daney Sept. 12, 2017, 3:44 p.m. UTC | #6
On 09/12/2017 07:03 AM, Linus Walleij wrote:
> On Thu, Sep 7, 2017 at 1:42 PM, Masahiro Yamada
> <yamada.masahiro@socionext.com> wrote:
> 
>> This GPIO controller device is used on UniPhier SoCs.
>>
>> It also serves as an interrupt controller, but interrupt signals are
>> just delivered to the parent irqchip without any latching or OR'ing.
>> This is implemented by using hierarchy IRQ domain.
>>
>> Implementation note:
>> Unfortunately, the IRQ mapping from this controller to the parent is
>> random. (48, 49, ..., 63, 154, 155, ...)
>> If "interrupts" property is used, IRQ resources may be statically
>> allocated when platform devices are populated from DT.  This can be
>> a problem for the hierarchy IRQ domain because IRQ allocation must
>> happen from the outer-most domain up to the root domain in order to
>> build up the stacked IRQ.  (https://lkml.org/lkml/2017/7/6/758)
>> Solutions to work around it could be to hard-code parent hwirqs or
>> to invent a driver-specific DT property.
>>
>> Here, the new API irq_domain_push_irq() was merged by v4.14-rc1.
>> It allows to add irq_data to the existing hierarchy.  It will help
>> to make this driver work whether the parent has already initialized
>> the hierarchy or not.
>>
>> Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
>> ---
>>
>> Changes in v4:
>>    - Add COMPILE_TEST and select IRQ_DOMAIN_HIERARCHY
>>    - Reimplement irqchip part by using irq_domain_push_irq()
> 
> Awesome improvement. There was a build error and I also
> would like David Daney to have a look at this so we know we
> use things the right way,

It looks correct to me.

I haven't verified it, but I think the OF device-tree probing code for 
the platform devices will automatically xlat-and-map all those irqs, so 
that the  irq_domain_push_irq() is required to get the domain hierarchy 
properly configured.  It would be similar to the PCI case where we 
configure all the MSI-X and then do the irq_domain_push_irq() in the 
Cavium ThunderX driver.

If interrupt handling has been verified to work with this driver, I 
would say that we are probably using things "the right way".

David.



> but overall I am happy after this
> so I hope I will be able to apply next version.
> 
> Yours,
> Linus Walleij
> 

--
To unsubscribe from this list: send the line "unsubscribe linux-gpio" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Masahiro Yamada Sept. 13, 2017, 8:31 a.m. UTC | #7
Hi.

2017-09-13 0:44 GMT+09:00 David Daney <ddaney@caviumnetworks.com>:
> On 09/12/2017 07:03 AM, Linus Walleij wrote:
>>
>> On Thu, Sep 7, 2017 at 1:42 PM, Masahiro Yamada
>> <yamada.masahiro@socionext.com> wrote:
>>
>>> This GPIO controller device is used on UniPhier SoCs.
>>>
>>> It also serves as an interrupt controller, but interrupt signals are
>>> just delivered to the parent irqchip without any latching or OR'ing.
>>> This is implemented by using hierarchy IRQ domain.
>>>
>>> Implementation note:
>>> Unfortunately, the IRQ mapping from this controller to the parent is
>>> random. (48, 49, ..., 63, 154, 155, ...)
>>> If "interrupts" property is used, IRQ resources may be statically
>>> allocated when platform devices are populated from DT.  This can be
>>> a problem for the hierarchy IRQ domain because IRQ allocation must
>>> happen from the outer-most domain up to the root domain in order to
>>> build up the stacked IRQ.  (https://lkml.org/lkml/2017/7/6/758)
>>> Solutions to work around it could be to hard-code parent hwirqs or
>>> to invent a driver-specific DT property.
>>>
>>> Here, the new API irq_domain_push_irq() was merged by v4.14-rc1.
>>> It allows to add irq_data to the existing hierarchy.  It will help
>>> to make this driver work whether the parent has already initialized
>>> the hierarchy or not.
>>>
>>> Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
>>> ---
>>>
>>> Changes in v4:
>>>    - Add COMPILE_TEST and select IRQ_DOMAIN_HIERARCHY
>>>    - Reimplement irqchip part by using irq_domain_push_irq()
>>
>>
>> Awesome improvement. There was a build error and I also
>> would like David Daney to have a look at this so we know we
>> use things the right way,
>
>
> It looks correct to me.
>
> I haven't verified it, but I think the OF device-tree probing code for the
> platform devices will automatically xlat-and-map all those irqs, so that the
> irq_domain_push_irq() is required to get the domain hierarchy properly
> configured.  It would be similar to the PCI case where we configure all the
> MSI-X and then do the irq_domain_push_irq() in the Cavium ThunderX driver.
>
> If interrupt handling has been verified to work with this driver, I would
> say that we are probably using things "the right way".
>
> David.
>

V4 depends on 5 patches that got negative feedback in irqdomain subsystem.
One more problem for this approach is to virtual IRQs are statically
allocated during the driver probe.
This looks a step backward to me.
Recently, gpio_irqchip migrated to on-the-fly allocation in case some
controllers may have lots of
GPIO lines.


Finally, I came up with another (I think, better) solution.
I think v5 is less controversial,
and works very well in on-the-fly manner.

I am sending it shortly...
diff mbox series

Patch

diff --git a/Documentation/devicetree/bindings/gpio/gpio-uniphier.txt b/Documentation/devicetree/bindings/gpio/gpio-uniphier.txt
new file mode 100644
index 0000000..10326df
--- /dev/null
+++ b/Documentation/devicetree/bindings/gpio/gpio-uniphier.txt
@@ -0,0 +1,43 @@ 
+UniPhier GPIO controller
+
+Required properties:
+- compatible: Should be "socionext,uniphier-gpio".
+- reg: Specifies offset and length of the register set for the device.
+- gpio-controller: Marks the device node as a GPIO controller.
+- #gpio-cells: Should be 2.  The first cell is the pin number and the second
+  cell is used to specify optional parameters.
+- interrupt-parent: Specifies the parent interrupt controller.
+- interrupts: Specifies interrupts connected to the parent interrupt controller.
+- interrupt-controller: Marks the device node as an interrupt controller.
+- #interrupt-cells: Should be 2.  The first cell defines the interrupt number.
+  The second cell bits[3:0] is used to specify trigger type as follows:
+    1 = low-to-high edge triggered
+    2 = high-to-low edge triggered
+    4 = active high level-sensitive
+    8 = active low level-sensitive
+  Valid combinations are 1, 2, 3, 4, 8.
+- ngpios: Specifies the number of GPIO lines.
+- gpio-ranges: Mapping to pin controller pins (as described in gpio.txt)
+
+Optional properties:
+- gpio-ranges-group-names: Used for named gpio ranges (as described in gpio.txt)
+
+Example:
+	gpio: gpio@55000000 {
+		compatible = "socionext,uniphier-gpio";
+		reg = <0x55000000 0x200>;
+		interrupt-parent = <&aidet>;
+		interrupts = <48 4>, <49 4>, <50 4>, <51 4>,
+			     <52 4>, <53 4>, <54 4>, <55 4>,
+			     <56 4>, <57 4>, <58 4>, <59 4>,
+			     <60 4>, <61 4>, <62 4>, <63 4>,
+			     <154 4>, <155 4>, <156 4>, <157 4>,
+			     <158 4>;
+		interrupt-controller;
+		#interrupt-cells = <2>;
+		gpio-controller;
+		#gpio-cells = <2>;
+		gpio-ranges = <&pinctrl 0 0 0>;
+		gpio-ranges-group-names = "gpio_range";
+		ngpios = <248>;
+	};
diff --git a/MAINTAINERS b/MAINTAINERS
index 11dde28..df67191 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -2007,6 +2007,7 @@  F:	arch/arm/mm/cache-uniphier.c
 F:	arch/arm64/boot/dts/socionext/
 F:	drivers/bus/uniphier-system-bus.c
 F:	drivers/clk/uniphier/
+F:	drivers/gpio/gpio-uniphier.c
 F:	drivers/i2c/busses/i2c-uniphier*
 F:	drivers/irqchip/irq-uniphier-aidet.c
 F:	drivers/pinctrl/uniphier/
diff --git a/drivers/gpio/Kconfig b/drivers/gpio/Kconfig
index 3f80f16..25c0f308 100644
--- a/drivers/gpio/Kconfig
+++ b/drivers/gpio/Kconfig
@@ -475,6 +475,14 @@  config GPIO_TZ1090_PDC
 	help
 	  Say yes here to support Toumaz Xenif TZ1090 PDC GPIOs.
 
+config GPIO_UNIPHIER
+	tristate "UniPhier GPIO support"
+	depends on ARCH_UNIPHIER || COMPILE_TEST
+	depends on OF_GPIO
+	select IRQ_DOMAIN_HIERARCHY
+	help
+	  Say yes here to support UniPhier GPIOs.
+
 config GPIO_VF610
 	def_bool y
 	depends on ARCH_MXC && SOC_VF610
diff --git a/drivers/gpio/Makefile b/drivers/gpio/Makefile
index aeb70e9d..472f675 100644
--- a/drivers/gpio/Makefile
+++ b/drivers/gpio/Makefile
@@ -131,6 +131,7 @@  obj-$(CONFIG_GPIO_TWL6040)	+= gpio-twl6040.o
 obj-$(CONFIG_GPIO_TZ1090)	+= gpio-tz1090.o
 obj-$(CONFIG_GPIO_TZ1090_PDC)	+= gpio-tz1090-pdc.o
 obj-$(CONFIG_GPIO_UCB1400)	+= gpio-ucb1400.o
+obj-$(CONFIG_GPIO_UNIPHIER)	+= gpio-uniphier.o
 obj-$(CONFIG_GPIO_VF610)	+= gpio-vf610.o
 obj-$(CONFIG_GPIO_VIPERBOARD)	+= gpio-viperboard.o
 obj-$(CONFIG_GPIO_VR41XX)	+= gpio-vr41xx.o
diff --git a/drivers/gpio/gpio-uniphier.c b/drivers/gpio/gpio-uniphier.c
new file mode 100644
index 0000000..135c91a
--- /dev/null
+++ b/drivers/gpio/gpio-uniphier.c
@@ -0,0 +1,486 @@ 
+/*
+ * Copyright (C) 2017 Socionext Inc.
+ *   Author: Masahiro Yamada <yamada.masahiro@socionext.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that 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.
+ */
+
+#include <linux/bitops.h>
+#include <linux/gpio/driver.h>
+#include <linux/irq.h>
+#include <linux/irqdomain.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/of_device.h>
+#include <linux/of_irq.h>
+#include <linux/platform_device.h>
+#include <linux/spinlock.h>
+#include <dt-bindings/gpio/uniphier-gpio.h>
+
+#define UNIPHIER_GPIO_BANK_MASK		\
+				GENMASK((UNIPHIER_GPIO_LINES_PER_BANK) - 1, 0)
+
+#define UNIPHIER_GPIO_IRQ_MAX_NUM	32
+
+#define UNIPHIER_GPIO_PORT_DATA		0x0	/* data */
+#define UNIPHIER_GPIO_PORT_DIR		0x4	/* direction (1:in, 0:out) */
+#define UNIPHIER_GPIO_IRQ_EN		0x90	/* irq enable */
+#define UNIPHIER_GPIO_IRQ_MODE		0x94	/* irq mode (1: both edge) */
+#define UNIPHIER_GPIO_IRQ_FLT_EN	0x98	/* noise filter enable */
+#define UNIPHIER_GPIO_IRQ_FLT_CYC	0x9c	/* noise filter clock cycle */
+
+struct uniphier_gpio_priv {
+	struct gpio_chip chip;
+	struct irq_chip irq_chip;
+	struct irq_domain *domain;
+	void __iomem *regs;
+	spinlock_t lock;
+	unsigned int nirqs;
+	u32 saved_vals[0];
+};
+
+static unsigned int uniphier_gpio_bank_to_reg(unsigned int bank)
+{
+	unsigned int reg;
+
+	reg = (bank + 1) * 8;
+
+	/*
+	 * Unfortunately, the GPIO port registers are not contiguous because
+	 * offset 0x90-0x9f is used for IRQ.  Add 0x10 when crossing the region.
+	 */
+	if (reg >= UNIPHIER_GPIO_IRQ_EN)
+		reg += 0x10;
+
+	return reg;
+}
+
+static void uniphier_gpio_get_bank_and_mask(unsigned int offset,
+					    unsigned int *bank, u32 *mask)
+{
+	*bank = offset / UNIPHIER_GPIO_LINES_PER_BANK;
+	*mask = BIT(offset % UNIPHIER_GPIO_LINES_PER_BANK);
+}
+
+static void uniphier_gpio_reg_update(struct uniphier_gpio_priv *priv,
+				     unsigned int reg, u32 mask, u32 val)
+{
+	unsigned long flags;
+	u32 tmp;
+
+	spin_lock_irqsave(&priv->lock, flags);
+	tmp = readl(priv->regs + reg);
+	tmp &= ~mask;
+	tmp |= mask & val;
+	writel(tmp, priv->regs + reg);
+	spin_unlock_irqrestore(&priv->lock, flags);
+}
+
+static void uniphier_gpio_bank_write(struct gpio_chip *chip, unsigned int bank,
+				     unsigned int reg, u32 mask, u32 val)
+{
+	struct uniphier_gpio_priv *priv = gpiochip_get_data(chip);
+
+	if (!mask)
+		return;
+
+	uniphier_gpio_reg_update(priv, uniphier_gpio_bank_to_reg(bank) + reg,
+				 mask, val);
+}
+
+static void uniphier_gpio_offset_write(struct gpio_chip *chip,
+				       unsigned int offset, unsigned int reg,
+				       int val)
+{
+	unsigned int bank;
+	u32 mask;
+
+	uniphier_gpio_get_bank_and_mask(offset, &bank, &mask);
+
+	uniphier_gpio_bank_write(chip, bank, reg, mask, val ? mask : 0);
+}
+
+static int uniphier_gpio_offset_read(struct gpio_chip *chip,
+				     unsigned int offset, unsigned int reg)
+{
+	struct uniphier_gpio_priv *priv = gpiochip_get_data(chip);
+	unsigned int bank, reg_offset;
+	u32 mask;
+
+	uniphier_gpio_get_bank_and_mask(offset, &bank, &mask);
+	reg_offset = uniphier_gpio_bank_to_reg(bank) + reg;
+
+	return !!(readl(priv->regs + reg_offset) & mask);
+}
+
+static int uniphier_gpio_get_direction(struct gpio_chip *chip,
+				       unsigned int offset)
+{
+	return uniphier_gpio_offset_read(chip, offset, UNIPHIER_GPIO_PORT_DIR);
+}
+
+static int uniphier_gpio_direction_input(struct gpio_chip *chip,
+					 unsigned int offset)
+{
+	uniphier_gpio_offset_write(chip, offset, UNIPHIER_GPIO_PORT_DIR, 1);
+
+	return 0;
+}
+
+static int uniphier_gpio_direction_output(struct gpio_chip *chip,
+					  unsigned int offset, int val)
+{
+	uniphier_gpio_offset_write(chip, offset, UNIPHIER_GPIO_PORT_DATA, val);
+	uniphier_gpio_offset_write(chip, offset, UNIPHIER_GPIO_PORT_DIR, 0);
+
+	return 0;
+}
+
+static int uniphier_gpio_get(struct gpio_chip *chip, unsigned int offset)
+{
+	return uniphier_gpio_offset_read(chip, offset, UNIPHIER_GPIO_PORT_DATA);
+}
+
+static void uniphier_gpio_set(struct gpio_chip *chip,
+			      unsigned int offset, int val)
+{
+	uniphier_gpio_offset_write(chip, offset, UNIPHIER_GPIO_PORT_DATA, val);
+}
+
+static void uniphier_gpio_set_multiple(struct gpio_chip *chip,
+				       unsigned long *mask, unsigned long *bits)
+{
+	unsigned int bank, shift, bank_mask, bank_bits;
+	int i;
+
+	for (i = 0; i < chip->ngpio; i += UNIPHIER_GPIO_LINES_PER_BANK) {
+		bank = i / UNIPHIER_GPIO_LINES_PER_BANK;
+		shift = i % BITS_PER_LONG;
+		bank_mask = (mask[BIT_WORD(i)] >> shift) &
+						UNIPHIER_GPIO_BANK_MASK;
+		bank_bits = bits[BIT_WORD(i)] >> shift;
+
+		uniphier_gpio_bank_write(chip, bank, UNIPHIER_GPIO_PORT_DATA,
+					 bank_mask, bank_bits);
+	}
+}
+
+static int uniphier_gpio_to_irq(struct gpio_chip *chip, unsigned int offset)
+{
+	struct uniphier_gpio_priv *priv = gpiochip_get_data(chip);
+
+	if (offset < UNIPHIER_GPIO_IRQ_OFFSET)
+		return -ENXIO;
+
+	return irq_find_mapping(priv->domain,
+				offset - UNIPHIER_GPIO_IRQ_OFFSET);
+}
+
+static void uniphier_gpio_irq_mask(struct irq_data *data)
+{
+	struct uniphier_gpio_priv *priv = data->chip_data;
+	u32 mask = BIT(data->hwirq);
+
+	uniphier_gpio_reg_update(priv, UNIPHIER_GPIO_IRQ_EN, mask, 0);
+
+	return irq_chip_mask_parent(data);
+}
+
+static void uniphier_gpio_irq_unmask(struct irq_data *data)
+{
+	struct uniphier_gpio_priv *priv = data->chip_data;
+	u32 mask = BIT(data->hwirq);
+
+	uniphier_gpio_reg_update(priv, UNIPHIER_GPIO_IRQ_EN, mask, mask);
+
+	return irq_chip_unmask_parent(data);
+}
+
+static int uniphier_gpio_irq_set_type(struct irq_data *data, unsigned int type)
+{
+	struct uniphier_gpio_priv *priv = data->chip_data;
+	u32 mask = BIT(data->hwirq);
+	u32 val = 0;
+
+	if (type == IRQ_TYPE_EDGE_BOTH) {
+		val = mask;
+		type = IRQ_TYPE_EDGE_FALLING;
+	}
+
+	uniphier_gpio_reg_update(priv, UNIPHIER_GPIO_IRQ_MODE, mask, val);
+	/* To enable both edge detection, the noise filter must be enabled. */
+	uniphier_gpio_reg_update(priv, UNIPHIER_GPIO_IRQ_FLT_EN, mask, val);
+
+	return irq_chip_set_type_parent(data, type);
+}
+
+static int uniphier_gpio_irq_domain_alloc(struct irq_domain *domain,
+					  unsigned int virq,
+					  unsigned int nr_irqs, void *arg)
+{
+	struct uniphier_gpio_priv *priv = domain->host_data;
+	irq_hw_number_t hwirq = (long)arg;
+
+	if (WARN_ON(nr_irqs != 1))
+		return -EINVAL;
+
+	if (WARN_ON(hwirq >= priv->nirqs))
+		return -EINVAL;
+
+	return irq_domain_set_hwirq_and_chip(domain, virq, hwirq,
+					     &priv->irq_chip, priv);
+}
+
+static void uniphier_gpio_irq_domain_activate(struct irq_domain *domain,
+					      struct irq_data *data)
+{
+	struct uniphier_gpio_priv *priv = domain->host_data;
+	struct gpio_chip *chip = &priv->chip;
+
+	gpiochip_lock_as_irq(chip, data->hwirq + UNIPHIER_GPIO_IRQ_OFFSET);
+}
+
+static void uniphier_gpio_irq_domain_deactivate(struct irq_domain *domain,
+						struct irq_data *data)
+{
+	struct uniphier_gpio_priv *priv = domain->host_data;
+	struct gpio_chip *chip = &priv->chip;
+
+	gpiochip_unlock_as_irq(chip, data->hwirq + UNIPHIER_GPIO_IRQ_OFFSET);
+}
+
+static int uniphier_gpio_irq_domain_translate(struct irq_domain *domain,
+					      struct irq_fwspec *fwspec,
+					      unsigned long *out_hwirq,
+					      unsigned int *out_type)
+{
+	if (WARN_ON(fwspec->param_count < 2))
+		return -EINVAL;
+
+	*out_hwirq = fwspec->param[0];
+	*out_type = fwspec->param[1] & IRQ_TYPE_SENSE_MASK;
+
+	return 0;
+}
+
+static const struct irq_domain_ops uniphier_gpio_irq_domain_ops = {
+	.alloc = uniphier_gpio_irq_domain_alloc,
+	.free = irq_domain_free_irqs_common,
+	.activate = uniphier_gpio_irq_domain_activate,
+	.deactivate = uniphier_gpio_irq_domain_deactivate,
+	.translate = uniphier_gpio_irq_domain_translate,
+};
+
+static void uniphier_gpio_hw_init(struct uniphier_gpio_priv *priv)
+{
+	/*
+	 * Due to the hardware design, the noise filter must be enabled to
+	 * detect both edge interrupts.  This filter is intended to remove the
+	 * noise from the irq lines.  It does not work for GPIO input, so GPIO
+	 * debounce is not supported.  Unfortunately, the filter period is
+	 * shared among all irq lines.  Just choose a sensible period here.
+	 */
+	writel(0xff, priv->regs + UNIPHIER_GPIO_IRQ_FLT_CYC);
+}
+
+static unsigned int uniphier_gpio_get_nbanks(unsigned int ngpio)
+{
+	return DIV_ROUND_UP(ngpio, UNIPHIER_GPIO_LINES_PER_BANK);
+}
+
+static int uniphier_gpio_probe(struct platform_device *pdev)
+{
+	struct device *dev = &pdev->dev;
+	struct device_node *parent_np;
+	struct irq_domain *parent_domain;
+	struct uniphier_gpio_priv *priv;
+	struct gpio_chip *chip;
+	struct irq_chip *irq_chip;
+	struct resource *regs;
+	unsigned int nregs;
+	u32 ngpios;
+	int virq, ret, i;
+
+	parent_np = of_irq_find_parent(dev->of_node);
+	if (!parent_np)
+		return -ENXIO;
+
+	parent_domain = irq_find_host(parent_np);
+	of_node_put(parent_np);
+	if (!parent_domain)
+		return -EPROBE_DEFER;
+
+	ret = of_property_read_u32(dev->of_node, "ngpios", &ngpios);
+	if (ret)
+		return ret;
+
+	nregs = uniphier_gpio_get_nbanks(ngpios) * 2 + 3;
+	priv = devm_kzalloc(dev,
+			    sizeof(*priv) + sizeof(priv->saved_vals[0]) * nregs,
+			    GFP_KERNEL);
+	if (!priv)
+		return -ENOMEM;
+
+	priv->nirqs = of_irq_count(dev->of_node);
+	if (priv->nirqs > UNIPHIER_GPIO_IRQ_MAX_NUM)
+		return -EINVAL;
+
+	regs = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+	priv->regs = devm_ioremap_resource(dev, regs);
+	if (IS_ERR(priv->regs))
+		return PTR_ERR(priv->regs);
+
+	spin_lock_init(&priv->lock);
+
+	chip = &priv->chip;
+	chip->label = dev_name(dev);
+	chip->parent = dev;
+	chip->request = gpiochip_generic_request;
+	chip->free = gpiochip_generic_free;
+	chip->get_direction = uniphier_gpio_get_direction;
+	chip->direction_input = uniphier_gpio_direction_input;
+	chip->direction_output = uniphier_gpio_direction_output;
+	chip->get = uniphier_gpio_get;
+	chip->set = uniphier_gpio_set;
+	chip->set_multiple = uniphier_gpio_set_multiple;
+	chip->to_irq = uniphier_gpio_to_irq;
+	chip->base = -1;
+	chip->ngpio = ngpios;
+
+	irq_chip = &priv->irq_chip;
+	irq_chip->name = dev_name(dev);
+	irq_chip->irq_mask = uniphier_gpio_irq_mask;
+	irq_chip->irq_unmask = uniphier_gpio_irq_unmask;
+	irq_chip->irq_eoi = irq_chip_eoi_parent;
+	irq_chip->irq_set_affinity = irq_chip_set_affinity_parent;
+	irq_chip->irq_set_type = uniphier_gpio_irq_set_type;
+
+	uniphier_gpio_hw_init(priv);
+
+	ret = devm_gpiochip_add_data(dev, chip, priv);
+	if (ret)
+		return ret;
+
+	priv->domain = irq_domain_create_hierarchy(
+					parent_domain,
+					IRQ_DOMAIN_FLAG_NO_CREATE, priv->nirqs,
+					of_node_to_fwnode(dev->of_node),
+					&uniphier_gpio_irq_domain_ops, priv);
+	if (!priv->domain)
+		return -ENOMEM;
+
+	for (i = 0; i < priv->nirqs; i++) {
+		virq = platform_get_irq(pdev, i);
+		if (virq < 0)
+			continue;
+
+		ret = irq_domain_push_irq(priv->domain, virq, (void *)(long)i);
+		if (ret)
+			dev_warn(dev,
+				 "failed to push to irq hierarchy. IRQ%d is not usable.",
+				 i);
+	}
+
+	platform_set_drvdata(pdev, priv);
+
+	return 0;
+}
+
+static int uniphier_gpio_remove(struct platform_device *pdev)
+{
+	struct uniphier_gpio_priv *priv = platform_get_drvdata(pdev);
+	unsigned int virq;
+	int ret, i;
+
+	for (i = 0; i < priv->nirqs; i++) {
+		virq = irq_find_mapping(priv->domain, i);
+		if (!virq)
+			continue;
+		ret = irq_domain_pop_irq(priv->domain, virq);
+		if (ret)
+			return ret;
+	}
+
+	irq_domain_remove(priv->domain);
+
+	return 0;
+}
+
+static int __maybe_unused uniphier_gpio_suspend(struct device *dev)
+{
+	struct uniphier_gpio_priv *priv = dev_get_drvdata(dev);
+	unsigned int nbanks = uniphier_gpio_get_nbanks(priv->chip.ngpio);
+	u32 *val = priv->saved_vals;
+	unsigned int reg;
+	int i;
+
+	for (i = 0; i < nbanks; i++) {
+		reg = uniphier_gpio_bank_to_reg(i);
+
+		*val++ = readl(priv->regs + reg + UNIPHIER_GPIO_PORT_DATA);
+		*val++ = readl(priv->regs + reg + UNIPHIER_GPIO_PORT_DIR);
+	}
+
+	*val++ = readl(priv->regs + UNIPHIER_GPIO_IRQ_EN);
+	*val++ = readl(priv->regs + UNIPHIER_GPIO_IRQ_MODE);
+	*val++ = readl(priv->regs + UNIPHIER_GPIO_IRQ_FLT_EN);
+
+	return 0;
+}
+
+static int __maybe_unused uniphier_gpio_resume(struct device *dev)
+{
+	struct uniphier_gpio_priv *priv = dev_get_drvdata(dev);
+	unsigned int nbanks = uniphier_gpio_get_nbanks(priv->chip.ngpio);
+	const u32 *val = priv->saved_vals;
+	unsigned int reg;
+	int i;
+
+	for (i = 0; i < nbanks; i++) {
+		reg = uniphier_gpio_bank_to_reg(i);
+
+		writel(*val++, priv->regs + reg + UNIPHIER_GPIO_PORT_DATA);
+		writel(*val++, priv->regs + reg + UNIPHIER_GPIO_PORT_DIR);
+	}
+
+	writel(*val++, priv->regs + UNIPHIER_GPIO_IRQ_EN);
+	writel(*val++, priv->regs + UNIPHIER_GPIO_IRQ_MODE);
+	writel(*val++, priv->regs + UNIPHIER_GPIO_IRQ_FLT_EN);
+
+	uniphier_gpio_hw_init(priv);
+
+	return 0;
+}
+
+static const struct dev_pm_ops uniphier_gpio_pm_ops = {
+	SET_LATE_SYSTEM_SLEEP_PM_OPS(uniphier_gpio_suspend,
+				     uniphier_gpio_resume)
+};
+
+static const struct of_device_id uniphier_gpio_match[] = {
+	{ .compatible = "socionext,uniphier-gpio" },
+	{ /* sentinel */ }
+};
+MODULE_DEVICE_TABLE(of, uniphier_gpio_match);
+
+static struct platform_driver uniphier_gpio_driver = {
+	.probe = uniphier_gpio_probe,
+	.remove = uniphier_gpio_remove,
+	.driver = {
+		.name = "uniphier-gpio",
+		.of_match_table = uniphier_gpio_match,
+		.pm = &uniphier_gpio_pm_ops,
+	},
+};
+module_platform_driver(uniphier_gpio_driver);
+
+MODULE_AUTHOR("Masahiro Yamada <yamada.masahiro@socionext.com>");
+MODULE_DESCRIPTION("UniPhier GPIO driver");
+MODULE_LICENSE("GPL");
diff --git a/include/dt-bindings/gpio/uniphier-gpio.h b/include/dt-bindings/gpio/uniphier-gpio.h
new file mode 100644
index 0000000..9f0ad17
--- /dev/null
+++ b/include/dt-bindings/gpio/uniphier-gpio.h
@@ -0,0 +1,18 @@ 
+/*
+ * Copyright (C) 2017 Socionext Inc.
+ *   Author: Masahiro Yamada <yamada.masahiro@socionext.com>
+ */
+
+#ifndef _DT_BINDINGS_GPIO_UNIPHIER_H
+#define _DT_BINDINGS_GPIO_UNIPHIER_H
+
+#define UNIPHIER_GPIO_LINES_PER_BANK	8
+
+#define UNIPHIER_GPIO_IRQ_OFFSET	((UNIPHIER_GPIO_LINES_PER_BANK) * 15)
+
+#define UNIPHIER_GPIO_PORT(bank, line)	\
+			((UNIPHIER_GPIO_LINES_PER_BANK) * (bank) + (line))
+
+#define UNIPHIER_GPIO_IRQ(n)		((UNIPHIER_GPIO_IRQ_OFFSET) + (n))
+
+#endif /* _DT_BINDINGS_GPIO_UNIPHIER_H */