diff mbox

[v2,2/8] irqchip: Supply new driver for STi based devices

Message ID 1416932705-16880-3-git-send-email-lee.jones@linaro.org
State New
Headers show

Commit Message

Lee Jones Nov. 25, 2014, 4:24 p.m. UTC
This driver is used to enable System Configuration Register controlled
External, CTI (Core Sight), PMU (Performance Management), and PL310 L2
Cache IRQs prior to use.

Signed-off-by: Lee Jones <lee.jones@linaro.org>
---
 drivers/irqchip/Kconfig  |   7 ++
 drivers/irqchip/Makefile |   1 +
 drivers/irqchip/irq-st.c | 208 +++++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 216 insertions(+)
 create mode 100644 drivers/irqchip/irq-st.c

Comments

Arnd Bergmann Nov. 25, 2014, 8:57 p.m. UTC | #1
On Tuesday 25 November 2014 16:24:59 Lee Jones wrote:

> diff --git a/drivers/irqchip/Kconfig b/drivers/irqchip/Kconfig
> index b21f12f..e502f15 100644
> --- a/drivers/irqchip/Kconfig
> +++ b/drivers/irqchip/Kconfig
> @@ -93,6 +93,13 @@ config RENESAS_IRQC
>  	bool
>  	select IRQ_DOMAIN
>  
> +config ST_IRQCHIP
> +	bool
> +	select REGMAP
> +	select MFD_SYSCON
> +	help
> +	  Enables SysCfg Controlled IRQs on STi based platforms.
> +

I'm confused by the purpose of this code. It's apparently a driver
in drivers/irqchip, the Kconfig symbol contains the string IRQCHIP,
yet it doesn't actually register an irq_chip.

Also, the name is a bit too generic, ST has lots of different irqchips,
and this apparently isn't even one of them ;-)

	Arnd
Lee Jones Nov. 27, 2014, 9:02 a.m. UTC | #2
On Tue, 25 Nov 2014, Arnd Bergmann wrote:

> On Tuesday 25 November 2014 16:24:59 Lee Jones wrote:
> 
> > diff --git a/drivers/irqchip/Kconfig b/drivers/irqchip/Kconfig
> > index b21f12f..e502f15 100644
> > --- a/drivers/irqchip/Kconfig
> > +++ b/drivers/irqchip/Kconfig
> > @@ -93,6 +93,13 @@ config RENESAS_IRQC
> >  	bool
> >  	select IRQ_DOMAIN
> >  
> > +config ST_IRQCHIP
> > +	bool
> > +	select REGMAP
> > +	select MFD_SYSCON
> > +	help
> > +	  Enables SysCfg Controlled IRQs on STi based platforms.
> > +
> 
> I'm confused by the purpose of this code. It's apparently a driver
> in drivers/irqchip, the Kconfig symbol contains the string IRQCHIP,
> yet it doesn't actually register an irq_chip.
> 
> Also, the name is a bit too generic, ST has lots of different irqchips,
> and this apparently isn't even one of them ;-)

Hmm... now you're going to ask me to remember who I had the
conversation with that alluded to this as the best location for this
driver.  Unfortunately, I cannot.  Can you think of a better place to
put it then?
Arnd Bergmann Nov. 27, 2014, 9:15 a.m. UTC | #3
On Thursday 27 November 2014 09:02:55 Lee Jones wrote:
> On Tue, 25 Nov 2014, Arnd Bergmann wrote:
> 
> > On Tuesday 25 November 2014 16:24:59 Lee Jones wrote:
> > 
> > > diff --git a/drivers/irqchip/Kconfig b/drivers/irqchip/Kconfig
> > > index b21f12f..e502f15 100644
> > > --- a/drivers/irqchip/Kconfig
> > > +++ b/drivers/irqchip/Kconfig
> > > @@ -93,6 +93,13 @@ config RENESAS_IRQC
> > >     bool
> > >     select IRQ_DOMAIN
> > >  
> > > +config ST_IRQCHIP
> > > +   bool
> > > +   select REGMAP
> > > +   select MFD_SYSCON
> > > +   help
> > > +     Enables SysCfg Controlled IRQs on STi based platforms.
> > > +
> > 
> > I'm confused by the purpose of this code. It's apparently a driver
> > in drivers/irqchip, the Kconfig symbol contains the string IRQCHIP,
> > yet it doesn't actually register an irq_chip.
> > 
> > Also, the name is a bit too generic, ST has lots of different irqchips,
> > and this apparently isn't even one of them 
> 
> Hmm... now you're going to ask me to remember who I had the
> conversation with that alluded to this as the best location for this
> driver.  Unfortunately, I cannot.  Can you think of a better place to
> put it then?

I suspect it's in the right place but should actually be an irqchip
driver. I'm having trouble understanding what this code actually does,
can you you explain the functionality in more detail so we can figure
out what to do with it?

	Arnd
Lee Jones Nov. 27, 2014, 9:29 a.m. UTC | #4
On Thu, 27 Nov 2014, Arnd Bergmann wrote:

> On Thursday 27 November 2014 09:02:55 Lee Jones wrote:
> > On Tue, 25 Nov 2014, Arnd Bergmann wrote:
> > 
> > > On Tuesday 25 November 2014 16:24:59 Lee Jones wrote:
> > > 
> > > > diff --git a/drivers/irqchip/Kconfig b/drivers/irqchip/Kconfig
> > > > index b21f12f..e502f15 100644
> > > > --- a/drivers/irqchip/Kconfig
> > > > +++ b/drivers/irqchip/Kconfig
> > > > @@ -93,6 +93,13 @@ config RENESAS_IRQC
> > > >     bool
> > > >     select IRQ_DOMAIN
> > > >  
> > > > +config ST_IRQCHIP
> > > > +   bool
> > > > +   select REGMAP
> > > > +   select MFD_SYSCON
> > > > +   help
> > > > +     Enables SysCfg Controlled IRQs on STi based platforms.
> > > > +
> > > 
> > > I'm confused by the purpose of this code. It's apparently a driver
> > > in drivers/irqchip, the Kconfig symbol contains the string IRQCHIP,
> > > yet it doesn't actually register an irq_chip.
> > > 
> > > Also, the name is a bit too generic, ST has lots of different irqchips,
> > > and this apparently isn't even one of them 
> > 
> > Hmm... now you're going to ask me to remember who I had the
> > conversation with that alluded to this as the best location for this
> > driver.  Unfortunately, I cannot.  Can you think of a better place to
> > put it then?
> 
> I suspect it's in the right place but should actually be an irqchip
> driver. I'm having trouble understanding what this code actually does,
> can you you explain the functionality in more detail so we can figure
> out what to do with it?

In the simplest terms it's an IRQ unmasker.  A9 Core IRQs are disabled
on boot; PMU, CTI (CoreSight), PL310_L2 and EXT.  In order for you to
make use of them they need to be unmasked in SYSCFG.
Arnd Bergmann Nov. 27, 2014, 9:47 a.m. UTC | #5
On Thursday 27 November 2014 09:29:01 Lee Jones wrote:
> On Thu, 27 Nov 2014, Arnd Bergmann wrote:
> > On Thursday 27 November 2014 09:02:55 Lee Jones wrote:
> > > On Tue, 25 Nov 2014, Arnd Bergmann wrote:
> > > 
> > > > On Tuesday 25 November 2014 16:24:59 Lee Jones wrote:
> > > > 
> > > > > diff --git a/drivers/irqchip/Kconfig b/drivers/irqchip/Kconfig
> > > > > index b21f12f..e502f15 100644
> > > > > --- a/drivers/irqchip/Kconfig
> > > > > +++ b/drivers/irqchip/Kconfig
> > > > > @@ -93,6 +93,13 @@ config RENESAS_IRQC
> > > > >     bool
> > > > >     select IRQ_DOMAIN
> > > > >  
> > > > > +config ST_IRQCHIP
> > > > > +   bool
> > > > > +   select REGMAP
> > > > > +   select MFD_SYSCON
> > > > > +   help
> > > > > +     Enables SysCfg Controlled IRQs on STi based platforms.
> > > > > +
> > > > 
> > > > I'm confused by the purpose of this code. It's apparently a driver
> > > > in drivers/irqchip, the Kconfig symbol contains the string IRQCHIP,
> > > > yet it doesn't actually register an irq_chip.
> > > > 
> > > > Also, the name is a bit too generic, ST has lots of different irqchips,
> > > > and this apparently isn't even one of them 
> > > 
> > > Hmm... now you're going to ask me to remember who I had the
> > > conversation with that alluded to this as the best location for this
> > > driver.  Unfortunately, I cannot.  Can you think of a better place to
> > > put it then?
> > 
> > I suspect it's in the right place but should actually be an irqchip
> > driver. I'm having trouble understanding what this code actually does,
> > can you you explain the functionality in more detail so we can figure
> > out what to do with it?
> 
> In the simplest terms it's an IRQ unmasker.  A9 Core IRQs are disabled
> on boot; PMU, CTI (CoreSight), PL310_L2 and EXT.  In order for you to
> make use of them they need to be unmasked in SYSCFG.

So you just apply static configuration once, or are there reasons why you
would mask the interrupts again later?

If it's all static, why doesn't the boot loader unmask all interrupts before
starting the kernel?

	Arnd
Lee Jones Nov. 27, 2014, 9:55 a.m. UTC | #6
On Thu, 27 Nov 2014, Arnd Bergmann wrote:

> On Thursday 27 November 2014 09:29:01 Lee Jones wrote:
> > On Thu, 27 Nov 2014, Arnd Bergmann wrote:
> > > On Thursday 27 November 2014 09:02:55 Lee Jones wrote:
> > > > On Tue, 25 Nov 2014, Arnd Bergmann wrote:
> > > > 
> > > > > On Tuesday 25 November 2014 16:24:59 Lee Jones wrote:
> > > > > 
> > > > > > diff --git a/drivers/irqchip/Kconfig b/drivers/irqchip/Kconfig
> > > > > > index b21f12f..e502f15 100644
> > > > > > --- a/drivers/irqchip/Kconfig
> > > > > > +++ b/drivers/irqchip/Kconfig
> > > > > > @@ -93,6 +93,13 @@ config RENESAS_IRQC
> > > > > >     bool
> > > > > >     select IRQ_DOMAIN
> > > > > >  
> > > > > > +config ST_IRQCHIP
> > > > > > +   bool
> > > > > > +   select REGMAP
> > > > > > +   select MFD_SYSCON
> > > > > > +   help
> > > > > > +     Enables SysCfg Controlled IRQs on STi based platforms.
> > > > > > +
> > > > > 
> > > > > I'm confused by the purpose of this code. It's apparently a driver
> > > > > in drivers/irqchip, the Kconfig symbol contains the string IRQCHIP,
> > > > > yet it doesn't actually register an irq_chip.
> > > > > 
> > > > > Also, the name is a bit too generic, ST has lots of different irqchips,
> > > > > and this apparently isn't even one of them 
> > > > 
> > > > Hmm... now you're going to ask me to remember who I had the
> > > > conversation with that alluded to this as the best location for this
> > > > driver.  Unfortunately, I cannot.  Can you think of a better place to
> > > > put it then?
> > > 
> > > I suspect it's in the right place but should actually be an irqchip
> > > driver. I'm having trouble understanding what this code actually does,
> > > can you you explain the functionality in more detail so we can figure
> > > out what to do with it?
> > 
> > In the simplest terms it's an IRQ unmasker.  A9 Core IRQs are disabled
> > on boot; PMU, CTI (CoreSight), PL310_L2 and EXT.  In order for you to
> > make use of them they need to be unmasked in SYSCFG.
> 
> So you just apply static configuration once, or are there reasons why you
> would mask the interrupts again later?
> 
> If it's all static, why doesn't the boot loader unmask all interrupts before
> starting the kernel?

We'd like to configure them at boot-time, via DT.
diff mbox

Patch

diff --git a/drivers/irqchip/Kconfig b/drivers/irqchip/Kconfig
index b21f12f..e502f15 100644
--- a/drivers/irqchip/Kconfig
+++ b/drivers/irqchip/Kconfig
@@ -93,6 +93,13 @@  config RENESAS_IRQC
 	bool
 	select IRQ_DOMAIN
 
+config ST_IRQCHIP
+	bool
+	select REGMAP
+	select MFD_SYSCON
+	help
+	  Enables SysCfg Controlled IRQs on STi based platforms.
+
 config TB10X_IRQC
 	bool
 	select IRQ_DOMAIN
diff --git a/drivers/irqchip/Makefile b/drivers/irqchip/Makefile
index 173bb5f..293b68d 100644
--- a/drivers/irqchip/Makefile
+++ b/drivers/irqchip/Makefile
@@ -31,6 +31,7 @@  obj-$(CONFIG_RENESAS_IRQC)		+= irq-renesas-irqc.o
 obj-$(CONFIG_VERSATILE_FPGA_IRQ)	+= irq-versatile-fpga.o
 obj-$(CONFIG_ARCH_NSPIRE)		+= irq-zevio.o
 obj-$(CONFIG_ARCH_VT8500)		+= irq-vt8500.o
+obj-$(CONFIG_ST_IRQCHIP)		+= irq-st.o
 obj-$(CONFIG_TB10X_IRQC)		+= irq-tb10x.o
 obj-$(CONFIG_XTENSA)			+= irq-xtensa-pic.o
 obj-$(CONFIG_XTENSA_MX)			+= irq-xtensa-mx.o
diff --git a/drivers/irqchip/irq-st.c b/drivers/irqchip/irq-st.c
new file mode 100644
index 0000000..181dad1
--- /dev/null
+++ b/drivers/irqchip/irq-st.c
@@ -0,0 +1,208 @@ 
+/*
+ *  drivers/irqchip/irq-st.c
+ *
+ *  Copyright (C) 2014 STMicroelectronics – All Rights Reserved
+ *
+ *  Author: Lee Jones <lee.jones@linaro.org>
+ *
+ *  This is a re-write of Christophe Kerello's PMU driver.
+ *
+ * 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.
+ */
+
+#include <dt-bindings/interrupt-controller/irq-st.h>
+#include <linux/err.h>
+#include <linux/mfd/syscon.h>
+#include <linux/of_device.h>
+#include <linux/platform_device.h>
+#include <linux/regmap.h>
+#include <linux/slab.h>
+
+#define STIH415_SYSCFG_642		0x0a8
+#define STIH416_SYSCFG_7543		0x87c
+#define STIH407_SYSCFG_5102		0x198
+#define STID127_SYSCFG_734		0x088
+
+#define ST_A9_IRQ_MASK			0x001FFFFF
+#define ST_A9_IRQ_MAX_CHANS		2
+
+#define ST_A9_IRQ_EN_CTI_0		BIT(0)
+#define ST_A9_IRQ_EN_CTI_1		BIT(1)
+#define ST_A9_IRQ_EN_PMU_0		BIT(2)
+#define ST_A9_IRQ_EN_PMU_1		BIT(3)
+#define ST_A9_IRQ_EN_PL310_L2		BIT(4)
+#define ST_A9_IRQ_EN_EXT_0		BIT(5)
+#define ST_A9_IRQ_EN_EXT_1		BIT(6)
+#define ST_A9_IRQ_EN_EXT_2		BIT(7)
+
+#define ST_A9_FIQ_N_SEL(dev, chan)	(dev << (8  + (chan * 3)))
+#define ST_A9_IRQ_N_SEL(dev, chan)	(dev << (14 + (chan * 3)))
+#define ST_A9_EXTIRQ_INV_SEL(dev)	(dev << 20)
+
+struct st_irq_syscfg {
+	struct regmap *regmap;
+	unsigned int syscfg;
+	unsigned int config;
+	bool ext_inverted;
+};
+
+static const struct of_device_id st_irq_syscfg_match[] = {
+	{
+		.compatible = "st,stih415-irq-syscfg",
+		.data = (void *)STIH415_SYSCFG_642,
+	},
+	{
+		.compatible = "st,stih416-irq-syscfg",
+		.data = (void *)STIH416_SYSCFG_7543,
+	},
+	{
+		.compatible = "st,stih407-irq-syscfg",
+		.data = (void *)STIH407_SYSCFG_5102,
+	},
+	{
+		.compatible = "st,stid127-irq-syscfg",
+		.data = (void *)STID127_SYSCFG_734,
+	},
+	{}
+};
+
+static int st_irq_xlate(struct platform_device *pdev,
+			int device, int channel, bool irq)
+{
+	struct st_irq_syscfg *ddata = dev_get_drvdata(&pdev->dev);
+
+	/* Set the device enable bit. */
+	switch (device) {
+	case ST_IRQ_SYSCFG_EXT_0 :
+		ddata->config |= ST_A9_IRQ_EN_EXT_0;
+		break;
+	case ST_IRQ_SYSCFG_EXT_1 :
+		ddata->config |= ST_A9_IRQ_EN_EXT_1;
+		break;
+	case ST_IRQ_SYSCFG_EXT_2 :
+		ddata->config |= ST_A9_IRQ_EN_EXT_2;
+		break;
+	case ST_IRQ_SYSCFG_CTI_0 :
+		ddata->config |= ST_A9_IRQ_EN_CTI_0;
+		break;
+	case ST_IRQ_SYSCFG_CTI_1 :
+		ddata->config |= ST_A9_IRQ_EN_CTI_1;
+		break;
+	case ST_IRQ_SYSCFG_PMU_0 :
+		ddata->config |= ST_A9_IRQ_EN_PMU_0;
+		break;
+	case ST_IRQ_SYSCFG_PMU_1 :
+		ddata->config |= ST_A9_IRQ_EN_PMU_1;
+		break;
+	case ST_IRQ_SYSCFG_pl310_L2 :
+		ddata->config |= ST_A9_IRQ_EN_PL310_L2;
+		break;
+	case ST_IRQ_SYSCFG_DISABLED :
+		return 0;
+	default :
+		dev_err(&pdev->dev, "Unrecognised device %d\n", device);
+		return -EINVAL;
+	}
+
+	/* Select IRQ/FIQ channel for device. */
+	ddata->config |= irq ?
+		ST_A9_IRQ_N_SEL(device, channel) :
+		ST_A9_FIQ_N_SEL(device, channel);
+
+	return 0;
+}
+
+static int st_irq_syscfg_enable(struct platform_device *pdev)
+{
+	struct device_node *np = pdev->dev.of_node;
+	struct st_irq_syscfg *ddata = dev_get_drvdata(&pdev->dev);
+	int channels, ret, i;
+	u32 device, invert;
+
+	channels = of_property_count_u32_elems(np, "st,irq-device");
+	if (channels != ST_A9_IRQ_MAX_CHANS) {
+		dev_err(&pdev->dev, "st,enable-irq-device must have 2 elems\n");
+		return -EINVAL;
+	}
+
+	channels = of_property_count_u32_elems(np, "st,fiq-device");
+	if (channels != ST_A9_IRQ_MAX_CHANS) {
+		dev_err(&pdev->dev, "st,enable-fiq-device must have 2 elems\n");
+		return -EINVAL;
+	}
+
+	for (i = 0; i < ST_A9_IRQ_MAX_CHANS; i++) {
+		of_property_read_u32_index(np, "st,irq-device", i, &device);
+
+		ret = st_irq_xlate(pdev, device, i, true);
+		if (ret)
+			return ret;
+
+		of_property_read_u32_index(np, "st,fiq-device", i, &device);
+
+		ret = st_irq_xlate(pdev, device, i, false);
+		if (ret)
+			return ret;
+	}
+
+	/* External IRQs may be inverted. */
+	of_property_read_u32(np, "st,invert-ext", &invert);
+	ddata->config |= ST_A9_EXTIRQ_INV_SEL(invert);
+
+	return regmap_update_bits(ddata->regmap, ddata->syscfg,
+				  ST_A9_IRQ_MASK, ddata->config);
+}
+
+static int st_irq_syscfg_probe(struct platform_device *pdev)
+{
+	struct device_node *np = pdev->dev.of_node;
+	const struct of_device_id *match;
+	struct st_irq_syscfg *ddata;
+
+	ddata = devm_kzalloc(&pdev->dev, sizeof(*ddata), GFP_KERNEL);
+	if (!ddata)
+		return -ENOMEM;
+
+	match = of_match_device(st_irq_syscfg_match, &pdev->dev);
+	if (!match)
+		return -ENODEV;
+
+	ddata->syscfg = (unsigned int)match->data;
+
+	ddata->regmap = syscon_regmap_lookup_by_phandle(np, "st,syscfg");
+	if (IS_ERR(ddata->regmap)) {
+		dev_err(&pdev->dev, "syscfg phandle missing\n");
+		return PTR_ERR(ddata->regmap);
+	}
+
+	dev_set_drvdata(&pdev->dev, ddata);
+
+	return st_irq_syscfg_enable(pdev);
+}
+
+static int st_irq_syscfg_resume(struct device *dev)
+{
+	struct st_irq_syscfg *ddata = dev_get_drvdata(dev);
+
+	return regmap_update_bits(ddata->regmap, ddata->syscfg,
+				  ST_A9_IRQ_MASK, ddata->config);
+}
+
+static SIMPLE_DEV_PM_OPS(st_irq_syscfg_pm_ops, NULL, st_irq_syscfg_resume);
+
+static struct platform_driver st_irq_syscfg_driver = {
+	.driver = {
+		.name = "st_irq_syscfg",
+		.pm = &st_irq_syscfg_pm_ops,
+		.of_match_table = st_irq_syscfg_match,
+	},
+	.probe = st_irq_syscfg_probe,
+};
+
+static int __init st_irq_syscfg_init(void)
+{
+	return platform_driver_register(&st_irq_syscfg_driver);
+}
+core_initcall(st_irq_syscfg_init);