[PATCH v2 0/2] irqchip: Support Aspeed IRQ controller

[PATCH v2 0/2] irqchip: Support Aspeed IRQ controller

[Joel Stanley]

Hello All,

This adds support for the IRQ controller found in the Aspeed ast2400 and
ast2500 BMC SoCs.

It's been tested on the ast2400 in an OpenPower Palmetto system, and on the
ast2500-edk developer kit.

V2 has a fix for the compatible string in the device tree bindings.

Cheers,

Joel

Benjamin Herrenschmidt (1):
  irqchip: Add irq controller for Aspeed

Joel Stanley (1):
  doc/devicetree: Add Aspeed VIC bindings

 .../interrupt-controller/aspeed,ast2400-vic.txt    |  24 +++
 drivers/irqchip/Makefile                           |   1 +
 drivers/irqchip/irq-aspeed-vic.c                   | 238 +++++++++++++++++++++
 3 files changed, 263 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/interrupt-controller/aspeed,ast2400-vic.txt
 create mode 100644 drivers/irqchip/irq-aspeed-vic.c

-- 
2.8.1

[PATCH v2 1/2] doc/devicetree: Add Aspeed VIC bindings

[Joel Stanley]

Signed-off-by: Joel Stanley <joel@jms.id.au>
---
v2:
  Fix device tree binding example, thanks Baruch and Arnd for pointing this out

 .../interrupt-controller/aspeed,ast2400-vic.txt    | 24 ++++++++++++++++++++++
 1 file changed, 24 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/interrupt-controller/aspeed,ast2400-vic.txt

diff --git a/Documentation/devicetree/bindings/interrupt-controller/aspeed,ast2400-vic.txt b/Documentation/devicetree/bindings/interrupt-controller/aspeed,ast2400-vic.txt
new file mode 100644
index 000000000000..86dede1755a2
--- /dev/null
+++ b/Documentation/devicetree/bindings/interrupt-controller/aspeed,ast2400-vic.txt
@@ -0,0 +1,24 @@
+Aspeed Vectored Interrupt Controller
+
+These bindings are for the Aspeed AST2400 interrupt controller register layout.
+The SoC has an legacy register layout, but this driver does not support that
+mode of operation.
+
+Required properties:
+
+- compatible : should be "aspeed,ast2400-vic".
+
+- interrupt-controller : Identifies the node as an interrupt controller
+- #interrupt-cells : Specifies the number of cells needed to encode an
+  interrupt source. The value shall be 1.
+- valid-sources : bitmask of valid irq sources
+
+Example:
+
+ vic: interrupt-controller {
+      compatible = "aspeed,ast2400-vic";
+      interrupt-controller;
+      #interrupt-cells = <1>;
+      valid-sources = < 0xffffffff 0x0007ffff>;
+      reg = <0x1e6c0080 0x80>;
+ };
-- 
2.8.1

[PATCH v2 2/2] irqchip: Add irq controller for Aspeed

[Joel Stanley]

From: Benjamin Herrenschmidt <benh-XVmvHMARGAS8U2dJNN8I7kB+6BGkLq7r@public.gmane.org>

Signed-off-by: Benjamin Herrenschmidt <benh-XVmvHMARGAS8U2dJNN8I7kB+6BGkLq7r@public.gmane.org>
Signed-off-by: Joel Stanley <joel-U3u1mxZcP9KHXe+LvDLADg@public.gmane.org>
---
 drivers/irqchip/Makefile         |   1 +
 drivers/irqchip/irq-aspeed-vic.c | 238 +++++++++++++++++++++++++++++++++++++++
 2 files changed, 239 insertions(+)
 create mode 100644 drivers/irqchip/irq-aspeed-vic.c

diff --git a/drivers/irqchip/Makefile b/drivers/irqchip/Makefile
index b03cfcbbac6b..ec82e6e15a38 100644
--- a/drivers/irqchip/Makefile
+++ b/drivers/irqchip/Makefile
@@ -65,3 +65,4 @@ obj-$(CONFIG_INGENIC_IRQ)		+= irq-ingenic.o
 obj-$(CONFIG_IMX_GPCV2)			+= irq-imx-gpcv2.o
 obj-$(CONFIG_PIC32_EVIC)		+= irq-pic32-evic.o
 obj-$(CONFIG_MVEBU_ODMI)		+= irq-mvebu-odmi.o
+obj-$(CONFIG_ARCH_ASPEED)		+= irq-aspeed-vic.o
diff --git a/drivers/irqchip/irq-aspeed-vic.c b/drivers/irqchip/irq-aspeed-vic.c
new file mode 100644
index 000000000000..db15fc3e831a
--- /dev/null
+++ b/drivers/irqchip/irq-aspeed-vic.c
@@ -0,0 +1,238 @@
+/*
+ *  Copyright (C) 2015 - Ben Herrenschmidt, IBM Corp.
+ *
+ *  Driver for Aspeed "new" VIC as found in SoC generation 3 and later
+ *
+ *  Based on irq-vic.c:
+ *
+ *  Copyright (C) 1999 - 2003 ARM Limited
+ *  Copyright (C) 2000 Deep Blue Solutions Ltd
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * 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/export.h>
+#include <linux/init.h>
+#include <linux/list.h>
+#include <linux/io.h>
+#include <linux/irq.h>
+#include <linux/irqchip.h>
+#include <linux/irqchip/chained_irq.h>
+#include <linux/irqdomain.h>
+#include <linux/of.h>
+#include <linux/of_address.h>
+#include <linux/of_irq.h>
+#include <linux/syscore_ops.h>
+#include <linux/device.h>
+#include <linux/slab.h>
+
+#include <asm/exception.h>
+#include <asm/irq.h>
+
+/* These definitions correspond to the "new mapping" of the
+ * register set that interleaves "high" and "low". The offsets
+ * below are for the "low" register, add 4 to get to the high one
+ */
+#define AVIC_IRQ_STATUS		0x00
+#define AVIC_FIQ_STATUS		0x08
+#define AVIC_RAW_STATUS		0x10
+#define AVIC_INT_SELECT		0x18
+#define AVIC_INT_ENABLE		0x20
+#define AVIC_INT_ENABLE_CLR	0x28
+#define AVIC_INT_TRIGGER	0x30
+#define AVIC_INT_TRIGGER_CLR	0x38
+#define AVIC_INT_SENSE		0x40
+#define AVIC_INT_DUAL_EDGE	0x48
+#define AVIC_INT_EVENT		0x50
+#define AVIC_EDGE_CLR		0x58
+#define AVIC_EDGE_STATUS	0x60
+
+struct aspeed_vic {
+	void __iomem		*base;
+	u32			valid_sources[2];
+	u32			edge_sources[2];
+	struct irq_domain	*dom;
+};
+static struct aspeed_vic *system_avic;
+
+static void vic_init_hw(struct aspeed_vic *vic)
+{
+	u32 sense;
+
+	/* Disable all interrupts */
+	writel(0xffffffff, vic->base + AVIC_INT_ENABLE_CLR);
+	writel(0xffffffff, vic->base + AVIC_INT_ENABLE_CLR + 4);
+
+	/* Make sure no soft trigger is on */
+	writel(0xffffffff, vic->base + AVIC_INT_TRIGGER_CLR);
+	writel(0xffffffff, vic->base + AVIC_INT_TRIGGER_CLR + 4);
+
+	/* Set everything to be IRQ */
+	writel(0, vic->base + AVIC_INT_SELECT);
+	writel(0, vic->base + AVIC_INT_SELECT + 4);
+
+	/* Some interrupts have a programable high/low level trigger
+	 * (4 GPIO direct inputs), for now we assume this was configured
+	 * by firmware. We read which ones are edge now.
+	 */
+	sense = readl(vic->base + AVIC_INT_SENSE);
+	vic->edge_sources[0] = ~sense;
+	sense = readl(vic->base + AVIC_INT_SENSE + 4);
+	vic->edge_sources[1] = ~sense;
+
+	/* Clear edge detection latches */
+	writel(0xffffffff, vic->base + AVIC_EDGE_CLR);
+	writel(0xffffffff, vic->base + AVIC_EDGE_CLR + 4);
+}
+
+static void __exception_irq_entry avic_handle_irq(struct pt_regs *regs)
+{
+	struct aspeed_vic *vic = system_avic;
+	u32 stat, irq;
+
+	for (;;) {
+		irq = 0;
+		stat = readl_relaxed(vic->base + AVIC_IRQ_STATUS);
+		if (!stat) {
+			stat = readl_relaxed(vic->base + AVIC_IRQ_STATUS + 4);
+			irq = 32;
+		}
+		if (stat == 0)
+			break;
+		irq += ffs(stat) - 1;
+		handle_domain_irq(vic->dom, irq, regs);
+	}
+}
+
+static void avic_ack_irq(struct irq_data *d)
+{
+	struct aspeed_vic *vic = irq_data_get_irq_chip_data(d);
+	unsigned int sidx = d->hwirq >> 5;
+	unsigned int sbit = 1u << (d->hwirq & 0x1f);
+
+	/* Clear edge latch for edge interrupts, nop for level */
+	if (vic->edge_sources[sidx] & sbit)
+		writel(sbit, vic->base + AVIC_EDGE_CLR + sidx * 4);
+}
+
+static void avic_mask_irq(struct irq_data *d)
+{
+	struct aspeed_vic *vic = irq_data_get_irq_chip_data(d);
+	unsigned int sidx = d->hwirq >> 5;
+	unsigned int sbit = 1u << (d->hwirq & 0x1f);
+
+	writel(sbit, vic->base + AVIC_INT_ENABLE_CLR + sidx * 4);
+}
+
+static void avic_unmask_irq(struct irq_data *d)
+{
+	struct aspeed_vic *vic = irq_data_get_irq_chip_data(d);
+	unsigned int sidx = d->hwirq >> 5;
+	unsigned int sbit = 1u << (d->hwirq & 0x1f);
+
+	writel(sbit, vic->base + AVIC_INT_ENABLE + sidx * 4);
+}
+
+/* For level irq, faster than going through a nop "ack" and mask */
+static void avic_mask_ack_irq(struct irq_data *d)
+{
+	struct aspeed_vic *vic = irq_data_get_irq_chip_data(d);
+	unsigned int sidx = d->hwirq >> 5;
+	unsigned int sbit = 1u << (d->hwirq & 0x1f);
+
+	/* First mask */
+	writel(sbit, vic->base + AVIC_INT_ENABLE_CLR + sidx * 4);
+
+	/* Then clear edge latch for edge interrupts */
+	if (vic->edge_sources[sidx] & sbit)
+		writel(sbit, vic->base + AVIC_EDGE_CLR + sidx * 4);
+}
+
+static struct irq_chip avic_chip = {
+	.name		= "AVIC",
+	.irq_ack	= avic_ack_irq,
+	.irq_mask	= avic_mask_irq,
+	.irq_unmask	= avic_unmask_irq,
+	.irq_mask_ack	= avic_mask_ack_irq,
+};
+
+static int avic_map(struct irq_domain *d, unsigned int irq,
+		    irq_hw_number_t hwirq)
+{
+	struct aspeed_vic *vic = d->host_data;
+	unsigned int sidx = hwirq >> 5;
+	unsigned int sbit = 1u << (hwirq & 0x1f);
+
+	/* Check if interrupt exists */
+	if (sidx > 1 || !(vic->valid_sources[sidx] & sbit))
+		return -EPERM;
+
+	if (vic->edge_sources[sidx] & sbit)
+		irq_set_chip_and_handler(irq, &avic_chip, handle_edge_irq);
+	else
+		irq_set_chip_and_handler(irq, &avic_chip, handle_level_irq);
+	irq_set_chip_data(irq, vic);
+	irq_set_probe(irq);
+	return 0;
+}
+
+static struct irq_domain_ops avic_dom_ops = {
+	.map = avic_map,
+	.xlate = irq_domain_xlate_onetwocell,
+};
+
+static int __init avic_of_init(struct device_node *node,
+			       struct device_node *parent)
+{
+	void __iomem *regs;
+	struct aspeed_vic *vic;
+	int nirqs;
+
+	if (WARN(parent, "non-root Aspeed VIC not supported"))
+		return -EINVAL;
+	if (WARN(system_avic, "duplicate Aspeed VIC not supported"))
+		return -EINVAL;
+
+	regs = of_iomap(node, 0);
+	if (WARN_ON(!regs))
+		return -EIO;
+
+	vic = kzalloc(sizeof(struct aspeed_vic), GFP_KERNEL);
+	if (WARN_ON(!vic)) {
+		iounmap(regs);
+		return -ENOMEM;
+	}
+	vic->base = regs;
+
+	of_property_read_u32_index(node, "valid-sources", 0,
+				   &vic->valid_sources[0]);
+	of_property_read_u32_index(node, "valid-sources", 1,
+				   &vic->valid_sources[1]);
+
+	nirqs = hweight32(vic->valid_sources[0]) +
+		hweight32(vic->valid_sources[1]);
+
+	/* Initialize soures, all masked */
+	vic_init_hw(vic);
+
+	/* Ready to receive interrupts */
+	system_avic = vic;
+	set_handle_irq(avic_handle_irq);
+
+	/* Register our domain */
+	vic->dom = irq_domain_add_simple(node, nirqs, 0,
+					 &avic_dom_ops, vic);
+
+	return 0;
+}
+
+IRQCHIP_DECLARE(aspeed_new_vic, "aspeed,ast2400-vic", avic_of_init);
-- 
2.8.1

--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH v2 1/2] doc/devicetree: Add Aspeed VIC bindings

[Rob Herring]

On Mon, May 09, 2016 at 11:28:36PM +0930, Joel Stanley wrote:
> Signed-off-by: Joel Stanley <joel-U3u1mxZcP9KHXe+LvDLADg@public.gmane.org>
> ---
> v2:
>   Fix device tree binding example, thanks Baruch and Arnd for pointing this out
> 
>  .../interrupt-controller/aspeed,ast2400-vic.txt    | 24 ++++++++++++++++++++++
>  1 file changed, 24 insertions(+)
>  create mode 100644 Documentation/devicetree/bindings/interrupt-controller/aspeed,ast2400-vic.txt
> 
> diff --git a/Documentation/devicetree/bindings/interrupt-controller/aspeed,ast2400-vic.txt b/Documentation/devicetree/bindings/interrupt-controller/aspeed,ast2400-vic.txt
> new file mode 100644
> index 000000000000..86dede1755a2
> --- /dev/null
> +++ b/Documentation/devicetree/bindings/interrupt-controller/aspeed,ast2400-vic.txt
> @@ -0,0 +1,24 @@
> +Aspeed Vectored Interrupt Controller
> +
> +These bindings are for the Aspeed AST2400 interrupt controller register layout.
> +The SoC has an legacy register layout, but this driver does not support that
> +mode of operation.
> +
> +Required properties:
> +
> +- compatible : should be "aspeed,ast2400-vic".
> +
> +- interrupt-controller : Identifies the node as an interrupt controller
> +- #interrupt-cells : Specifies the number of cells needed to encode an
> +  interrupt source. The value shall be 1.

No need for level vs. edge flags?

> +- valid-sources : bitmask of valid irq sources

Drop this. Either all interrupt controllers need this or none of 
them do. The valid sources are the ones described in the DT.

> +
> +Example:
> +
> + vic: interrupt-controller {

Needs a unit address.

> +      compatible = "aspeed,ast2400-vic";
> +      interrupt-controller;
> +      #interrupt-cells = <1>;
> +      valid-sources = < 0xffffffff 0x0007ffff>;
> +      reg = <0x1e6c0080 0x80>;
> + };
> -- 
> 2.8.1
> 
> --
> To unsubscribe from this list: send the line "unsubscribe devicetree" in
> the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH v2 1/2] doc/devicetree: Add Aspeed VIC bindings

[Benjamin Herrenschmidt]

On Wed, 2016-05-11 at 09:33 -0500, Rob Herring wrote:
> > +- interrupt-controller : Identifies the node as an interrupt controller
> > +- #interrupt-cells : Specifies the number of cells needed to encode an
> > +  interrupt source. The value shall be 1.
> 
> No need for level vs. edge flags?

That's an open question. Most interrupts are fixed. A handful of GPIOs
can be configured either way. For now I am relying on uboot setting up
the right config for them and I read it back at boot time, but we could
make it part of the binding I suppose.

> > +- valid-sources : bitmask of valid irq sources
> 
> Drop this. Either all interrupt controllers need this or none of 
> them do. The valid sources are the ones described in the DT.

Looking at the code (I wrote that ages ago), this is only used for:

  - Failing map on an unsupported source, so we could drop it

  - Counting the number of sources in order to optimize the
    revmap size allocation. We could unconditionally allocate
    64, I don't see a big deal here.

So yes Joel, feel free to just ditch this.

> > +
> > +Example:
> > +
> > + vic: interrupt-controller {
> 
> Needs a unit address.
> 
> > +      compatible = "aspeed,ast2400-vic";
> > +      interrupt-controller;
> > +      #interrupt-cells = <1>;
> > +      valid-sources = < 0xffffffff 0x0007ffff>;
> > +      reg = <0x1e6c0080 0x80>;
> > + };
> > -- 
> > 2.8.1
> > 
> > --
> > To unsubscribe from this list: send the line "unsubscribe devicetree" in
> > the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
> > More majordomo info at  http://vger.kernel.org/majordomo-info.html
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH v3 0/2] irqchip: Support Aspeed IRQ controller

[Joel Stanley]

Hello All,

This adds support for the IRQ controller found in the Aspeed ast2400 and
ast2500 BMC SoCs.

It's been tested on the ast2400 in an OpenPower Palmetto system, and on the
ast2500-edk developer kit.

V2 has a fix for the compatible string in the device tree bindings.

V3 drops the valid-sources bitmasks and adds a unit address to the example in
the bindings

Cheers,

Joel

Benjamin Herrenschmidt (1):
  irqchip: Add irq controller for Aspeed

Joel Stanley (1):
  doc/devicetree: Add Aspeed VIC bindings

 .../interrupt-controller/aspeed,ast2400-vic.txt    |  22 ++
 drivers/irqchip/Makefile                           |   1 +
 drivers/irqchip/irq-aspeed-vic.c                   | 230 +++++++++++++++++++++
 3 files changed, 253 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/interrupt-controller/aspeed,ast2400-vic.txt
 create mode 100644 drivers/irqchip/irq-aspeed-vic.c

-- 
2.8.1

--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH v3 1/2] doc/devicetree: Add Aspeed VIC bindings

[Joel Stanley]

Signed-off-by: Joel Stanley <joel-U3u1mxZcP9KHXe+LvDLADg@public.gmane.org>
---
 .../interrupt-controller/aspeed,ast2400-vic.txt    | 22 ++++++++++++++++++++++
 1 file changed, 22 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/interrupt-controller/aspeed,ast2400-vic.txt

diff --git a/Documentation/devicetree/bindings/interrupt-controller/aspeed,ast2400-vic.txt b/Documentation/devicetree/bindings/interrupt-controller/aspeed,ast2400-vic.txt
new file mode 100644
index 000000000000..6c6e85324b9d
--- /dev/null
+++ b/Documentation/devicetree/bindings/interrupt-controller/aspeed,ast2400-vic.txt
@@ -0,0 +1,22 @@
+Aspeed Vectored Interrupt Controller
+
+These bindings are for the Aspeed AST2400 interrupt controller register layout.
+The SoC has an legacy register layout, but this driver does not support that
+mode of operation.
+
+Required properties:
+
+- compatible : should be "aspeed,ast2400-vic".
+
+- interrupt-controller : Identifies the node as an interrupt controller
+- #interrupt-cells : Specifies the number of cells needed to encode an
+  interrupt source. The value shall be 1.
+
+Example:
+
+ vic: interrupt-controller@1e6c0080 {
+      compatible = "aspeed,ast2400-vic";
+      interrupt-controller;
+      #interrupt-cells = <1>;
+      reg = <0x1e6c0080 0x80>;
+ };
-- 
2.8.1

--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH v3 2/2] irqchip: Add irq controller for Aspeed

[Joel Stanley]

From: Benjamin Herrenschmidt <benh-XVmvHMARGAS8U2dJNN8I7kB+6BGkLq7r@public.gmane.org>

Signed-off-by: Benjamin Herrenschmidt <benh-XVmvHMARGAS8U2dJNN8I7kB+6BGkLq7r@public.gmane.org>
Signed-off-by: Joel Stanley <joel-U3u1mxZcP9KHXe+LvDLADg@public.gmane.org>
---
 drivers/irqchip/Makefile         |   1 +
 drivers/irqchip/irq-aspeed-vic.c | 230 +++++++++++++++++++++++++++++++++++++++
 2 files changed, 231 insertions(+)
 create mode 100644 drivers/irqchip/irq-aspeed-vic.c

diff --git a/drivers/irqchip/Makefile b/drivers/irqchip/Makefile
index b03cfcbbac6b..ec82e6e15a38 100644
--- a/drivers/irqchip/Makefile
+++ b/drivers/irqchip/Makefile
@@ -65,3 +65,4 @@ obj-$(CONFIG_INGENIC_IRQ)		+= irq-ingenic.o
 obj-$(CONFIG_IMX_GPCV2)			+= irq-imx-gpcv2.o
 obj-$(CONFIG_PIC32_EVIC)		+= irq-pic32-evic.o
 obj-$(CONFIG_MVEBU_ODMI)		+= irq-mvebu-odmi.o
+obj-$(CONFIG_ARCH_ASPEED)		+= irq-aspeed-vic.o
diff --git a/drivers/irqchip/irq-aspeed-vic.c b/drivers/irqchip/irq-aspeed-vic.c
new file mode 100644
index 000000000000..d24451d5bf8a
--- /dev/null
+++ b/drivers/irqchip/irq-aspeed-vic.c
@@ -0,0 +1,230 @@
+/*
+ *  Copyright (C) 2015 - Ben Herrenschmidt, IBM Corp.
+ *
+ *  Driver for Aspeed "new" VIC as found in SoC generation 3 and later
+ *
+ *  Based on irq-vic.c:
+ *
+ *  Copyright (C) 1999 - 2003 ARM Limited
+ *  Copyright (C) 2000 Deep Blue Solutions Ltd
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * 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/export.h>
+#include <linux/init.h>
+#include <linux/list.h>
+#include <linux/io.h>
+#include <linux/irq.h>
+#include <linux/irqchip.h>
+#include <linux/irqchip/chained_irq.h>
+#include <linux/irqdomain.h>
+#include <linux/of.h>
+#include <linux/of_address.h>
+#include <linux/of_irq.h>
+#include <linux/syscore_ops.h>
+#include <linux/device.h>
+#include <linux/slab.h>
+
+#include <asm/exception.h>
+#include <asm/irq.h>
+
+/* These definitions correspond to the "new mapping" of the
+ * register set that interleaves "high" and "low". The offsets
+ * below are for the "low" register, add 4 to get to the high one
+ */
+#define AVIC_IRQ_STATUS		0x00
+#define AVIC_FIQ_STATUS		0x08
+#define AVIC_RAW_STATUS		0x10
+#define AVIC_INT_SELECT		0x18
+#define AVIC_INT_ENABLE		0x20
+#define AVIC_INT_ENABLE_CLR	0x28
+#define AVIC_INT_TRIGGER	0x30
+#define AVIC_INT_TRIGGER_CLR	0x38
+#define AVIC_INT_SENSE		0x40
+#define AVIC_INT_DUAL_EDGE	0x48
+#define AVIC_INT_EVENT		0x50
+#define AVIC_EDGE_CLR		0x58
+#define AVIC_EDGE_STATUS	0x60
+
+#define NUM_IRQS		64
+
+struct aspeed_vic {
+	void __iomem		*base;
+	u32			edge_sources[2];
+	struct irq_domain	*dom;
+};
+static struct aspeed_vic *system_avic;
+
+static void vic_init_hw(struct aspeed_vic *vic)
+{
+	u32 sense;
+
+	/* Disable all interrupts */
+	writel(0xffffffff, vic->base + AVIC_INT_ENABLE_CLR);
+	writel(0xffffffff, vic->base + AVIC_INT_ENABLE_CLR + 4);
+
+	/* Make sure no soft trigger is on */
+	writel(0xffffffff, vic->base + AVIC_INT_TRIGGER_CLR);
+	writel(0xffffffff, vic->base + AVIC_INT_TRIGGER_CLR + 4);
+
+	/* Set everything to be IRQ */
+	writel(0, vic->base + AVIC_INT_SELECT);
+	writel(0, vic->base + AVIC_INT_SELECT + 4);
+
+	/* Some interrupts have a programable high/low level trigger
+	 * (4 GPIO direct inputs), for now we assume this was configured
+	 * by firmware. We read which ones are edge now.
+	 */
+	sense = readl(vic->base + AVIC_INT_SENSE);
+	vic->edge_sources[0] = ~sense;
+	sense = readl(vic->base + AVIC_INT_SENSE + 4);
+	vic->edge_sources[1] = ~sense;
+
+	/* Clear edge detection latches */
+	writel(0xffffffff, vic->base + AVIC_EDGE_CLR);
+	writel(0xffffffff, vic->base + AVIC_EDGE_CLR + 4);
+}
+
+static void __exception_irq_entry avic_handle_irq(struct pt_regs *regs)
+{
+	struct aspeed_vic *vic = system_avic;
+	u32 stat, irq;
+
+	for (;;) {
+		irq = 0;
+		stat = readl_relaxed(vic->base + AVIC_IRQ_STATUS);
+		if (!stat) {
+			stat = readl_relaxed(vic->base + AVIC_IRQ_STATUS + 4);
+			irq = 32;
+		}
+		if (stat == 0)
+			break;
+		irq += ffs(stat) - 1;
+		handle_domain_irq(vic->dom, irq, regs);
+	}
+}
+
+static void avic_ack_irq(struct irq_data *d)
+{
+	struct aspeed_vic *vic = irq_data_get_irq_chip_data(d);
+	unsigned int sidx = d->hwirq >> 5;
+	unsigned int sbit = 1u << (d->hwirq & 0x1f);
+
+	/* Clear edge latch for edge interrupts, nop for level */
+	if (vic->edge_sources[sidx] & sbit)
+		writel(sbit, vic->base + AVIC_EDGE_CLR + sidx * 4);
+}
+
+static void avic_mask_irq(struct irq_data *d)
+{
+	struct aspeed_vic *vic = irq_data_get_irq_chip_data(d);
+	unsigned int sidx = d->hwirq >> 5;
+	unsigned int sbit = 1u << (d->hwirq & 0x1f);
+
+	writel(sbit, vic->base + AVIC_INT_ENABLE_CLR + sidx * 4);
+}
+
+static void avic_unmask_irq(struct irq_data *d)
+{
+	struct aspeed_vic *vic = irq_data_get_irq_chip_data(d);
+	unsigned int sidx = d->hwirq >> 5;
+	unsigned int sbit = 1u << (d->hwirq & 0x1f);
+
+	writel(sbit, vic->base + AVIC_INT_ENABLE + sidx * 4);
+}
+
+/* For level irq, faster than going through a nop "ack" and mask */
+static void avic_mask_ack_irq(struct irq_data *d)
+{
+	struct aspeed_vic *vic = irq_data_get_irq_chip_data(d);
+	unsigned int sidx = d->hwirq >> 5;
+	unsigned int sbit = 1u << (d->hwirq & 0x1f);
+
+	/* First mask */
+	writel(sbit, vic->base + AVIC_INT_ENABLE_CLR + sidx * 4);
+
+	/* Then clear edge latch for edge interrupts */
+	if (vic->edge_sources[sidx] & sbit)
+		writel(sbit, vic->base + AVIC_EDGE_CLR + sidx * 4);
+}
+
+static struct irq_chip avic_chip = {
+	.name		= "AVIC",
+	.irq_ack	= avic_ack_irq,
+	.irq_mask	= avic_mask_irq,
+	.irq_unmask	= avic_unmask_irq,
+	.irq_mask_ack	= avic_mask_ack_irq,
+};
+
+static int avic_map(struct irq_domain *d, unsigned int irq,
+		    irq_hw_number_t hwirq)
+{
+	struct aspeed_vic *vic = d->host_data;
+	unsigned int sidx = hwirq >> 5;
+	unsigned int sbit = 1u << (hwirq & 0x1f);
+
+	/* Check if interrupt exists */
+	if (sidx > 1)
+		return -EPERM;
+
+	if (vic->edge_sources[sidx] & sbit)
+		irq_set_chip_and_handler(irq, &avic_chip, handle_edge_irq);
+	else
+		irq_set_chip_and_handler(irq, &avic_chip, handle_level_irq);
+	irq_set_chip_data(irq, vic);
+	irq_set_probe(irq);
+	return 0;
+}
+
+static struct irq_domain_ops avic_dom_ops = {
+	.map = avic_map,
+	.xlate = irq_domain_xlate_onetwocell,
+};
+
+static int __init avic_of_init(struct device_node *node,
+			       struct device_node *parent)
+{
+	void __iomem *regs;
+	struct aspeed_vic *vic;
+
+	if (WARN(parent, "non-root Aspeed VIC not supported"))
+		return -EINVAL;
+	if (WARN(system_avic, "duplicate Aspeed VIC not supported"))
+		return -EINVAL;
+
+	regs = of_iomap(node, 0);
+	if (WARN_ON(!regs))
+		return -EIO;
+
+	vic = kzalloc(sizeof(struct aspeed_vic), GFP_KERNEL);
+	if (WARN_ON(!vic)) {
+		iounmap(regs);
+		return -ENOMEM;
+	}
+	vic->base = regs;
+
+	/* Initialize soures, all masked */
+	vic_init_hw(vic);
+
+	/* Ready to receive interrupts */
+	system_avic = vic;
+	set_handle_irq(avic_handle_irq);
+
+	/* Register our domain */
+	vic->dom = irq_domain_add_simple(node, NUM_IRQS, 0,
+					 &avic_dom_ops, vic);
+
+	return 0;
+}
+
+IRQCHIP_DECLARE(aspeed_new_vic, "aspeed,ast2400-vic", avic_of_init);
-- 
2.8.1

--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH v2 1/2] doc/devicetree: Add Aspeed VIC bindings

[Rob Herring]

On Wed, May 11, 2016 at 8:01 PM, Benjamin Herrenschmidt
<benh-XVmvHMARGAS8U2dJNN8I7kB+6BGkLq7r@public.gmane.org> wrote:
> On Wed, 2016-05-11 at 09:33 -0500, Rob Herring wrote:
>> > +- interrupt-controller : Identifies the node as an interrupt controller
>> > +- #interrupt-cells : Specifies the number of cells needed to encode an
>> > +  interrupt source. The value shall be 1.
>>
>> No need for level vs. edge flags?
>
> That's an open question. Most interrupts are fixed. A handful of GPIOs
> can be configured either way. For now I am relying on uboot setting up
> the right config for them and I read it back at boot time, but we could
> make it part of the binding I suppose.

It is almost standard to just use 2 cells here even if reserved for
future use. Especially since the IP block seems to have registers to
control that.

>
>> > +- valid-sources : bitmask of valid irq sources
>>
>> Drop this. Either all interrupt controllers need this or none of
>> them do. The valid sources are the ones described in the DT.
>
> Looking at the code (I wrote that ages ago), this is only used for:

I'm guessing it came from the VIC binding. I never really liked the
property, but not enough to worry about it.

Rob
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH v2 1/2] doc/devicetree: Add Aspeed VIC bindings

[Joel Stanley]

On Fri, May 13, 2016 at 2:50 AM, Rob Herring <robh-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org> wrote:
> On Wed, May 11, 2016 at 8:01 PM, Benjamin Herrenschmidt
> <benh-XVmvHMARGAS8U2dJNN8I7kB+6BGkLq7r@public.gmane.org> wrote:
>> On Wed, 2016-05-11 at 09:33 -0500, Rob Herring wrote:
>>> > +- interrupt-controller : Identifies the node as an interrupt controller
>>> > +- #interrupt-cells : Specifies the number of cells needed to encode an
>>> > +  interrupt source. The value shall be 1.
>>>
>>> No need for level vs. edge flags?
>>
>> That's an open question. Most interrupts are fixed. A handful of GPIOs
>> can be configured either way. For now I am relying on uboot setting up
>> the right config for them and I read it back at boot time, but we could
>> make it part of the binding I suppose.
>
> It is almost standard to just use 2 cells here even if reserved for
> future use. Especially since the IP block seems to have registers to
> control that.

Yep, makes sense. I've added this to the bindings document.

I was trying to use the second cell in our driver:

  static struct irq_domain_ops avic_dom_ops = {
          .map = avic_map,
         .xlate = irq_domain_xlate_twocell,
  };

 So we have irq_domain_xlate_twocell to parse the device tree and
grabs the type property from the second cell.

In avic_map we set the irq handler:

         if (vic->edge_sources[sidx] & sbit) {
                 /* TODO: Check aginst type from dt and warn if not edge */
                 irq_set_chip_and_handler(irq, &avic_chip, handle_edge_irq);
         } else {
                 /* TODO: Check aginst type from dt and warn if not level */
                 irq_set_chip_and_handler(irq, &avic_chip, handle_level_irq);
         }

However, we don't have the type here, so we can't use it to check that
we're setting the correct edge/level callback (or use it to set the
register in the future if we want to use the device tree to override
the bootloader settings).

I had a look at some other drivers and they don't seem to use the dt
property when mapping the irq. What am I missing here?

Cheers,

Joel
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH v2 1/2] doc/devicetree: Add Aspeed VIC bindings

[Rob Herring]

On Wed, May 18, 2016 at 8:50 AM, Joel Stanley <joel-U3u1mxZcP9KHXe+LvDLADg@public.gmane.org> wrote:
> On Fri, May 13, 2016 at 2:50 AM, Rob Herring <robh-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org> wrote:
>> On Wed, May 11, 2016 at 8:01 PM, Benjamin Herrenschmidt
>> <benh-XVmvHMARGAS8U2dJNN8I7kB+6BGkLq7r@public.gmane.org> wrote:
>>> On Wed, 2016-05-11 at 09:33 -0500, Rob Herring wrote:
>>>> > +- interrupt-controller : Identifies the node as an interrupt controller
>>>> > +- #interrupt-cells : Specifies the number of cells needed to encode an
>>>> > +  interrupt source. The value shall be 1.
>>>>
>>>> No need for level vs. edge flags?
>>>
>>> That's an open question. Most interrupts are fixed. A handful of GPIOs
>>> can be configured either way. For now I am relying on uboot setting up
>>> the right config for them and I read it back at boot time, but we could
>>> make it part of the binding I suppose.
>>
>> It is almost standard to just use 2 cells here even if reserved for
>> future use. Especially since the IP block seems to have registers to
>> control that.
>
> Yep, makes sense. I've added this to the bindings document.
>
> I was trying to use the second cell in our driver:
>
>   static struct irq_domain_ops avic_dom_ops = {
>           .map = avic_map,
>          .xlate = irq_domain_xlate_twocell,
>   };
>
>  So we have irq_domain_xlate_twocell to parse the device tree and
> grabs the type property from the second cell.
>
> In avic_map we set the irq handler:
>
>          if (vic->edge_sources[sidx] & sbit) {
>                  /* TODO: Check aginst type from dt and warn if not edge */
>                  irq_set_chip_and_handler(irq, &avic_chip, handle_edge_irq);
>          } else {
>                  /* TODO: Check aginst type from dt and warn if not level */
>                  irq_set_chip_and_handler(irq, &avic_chip, handle_level_irq);
>          }
>
> However, we don't have the type here, so we can't use it to check that
> we're setting the correct edge/level callback (or use it to set the
> register in the future if we want to use the device tree to override
> the bootloader settings).
>
> I had a look at some other drivers and they don't seem to use the dt
> property when mapping the irq. What am I missing here?

The type will be set by the irqdomain core when the mapping is created
and this should result in irq_set_type() being called on your irqchip.
The map function doesn't need to deal with type.

Rob
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH v3 0/2] irqchip: Support Aspeed IRQ controller

[Jason Cooper]

Hey Joel, Ben,

On Fri, May 13, 2016 at 12:13:11AM +0930, Joel Stanley wrote:
> Hello All,
> 
> This adds support for the IRQ controller found in the Aspeed ast2400 and
> ast2500 BMC SoCs.
> 
> It's been tested on the ast2400 in an OpenPower Palmetto system, and on the
> ast2500-edk developer kit.
> 
> V2 has a fix for the compatible string in the device tree bindings.
> 
> V3 drops the valid-sources bitmasks and adds a unit address to the example in
> the bindings
> 
> Cheers,
> 
> Joel
> 
> Benjamin Herrenschmidt (1):
>   irqchip: Add irq controller for Aspeed
> 
> Joel Stanley (1):
>   doc/devicetree: Add Aspeed VIC bindings
> 
>  .../interrupt-controller/aspeed,ast2400-vic.txt    |  22 ++
>  drivers/irqchip/Makefile                           |   1 +
>  drivers/irqchip/irq-aspeed-vic.c                   | 230 +++++++++++++++++++++
>  3 files changed, 253 insertions(+)
>  create mode 100644 Documentation/devicetree/bindings/interrupt-controller/aspeed,ast2400-vic.txt
>  create mode 100644 drivers/irqchip/irq-aspeed-vic.c

Series applied to irqchip/aspeed.  I'll only rebase it if you tell me
to. :-)

Should be in -next the next time it gets built.

thx,

Jason.