Skip to contentSkip to conversation

Extend irqchip ocelot driver to support other SoCs

PATCH Superseded Gregory CLEMENT 4 participants 12 messages Nov 16, 2020 – Nov 18, 2020

Subsystems (2)

v3 of 5 Superseded Latest is v5 →

lore
Share & tools
lore
#0 lore raw

v3 cover Extend irqchip ocelot driver to support other SoCs

Hello,

Ocelot SoC belongs to a larger family of SoCs which use the same interrupt controller with a few variation.

This series of patches add support for Luton, Serval and Jaguar2, they are all MIPS based.

The first patches of the series also updates the binding documentation with the new compatible strings.

Gregory

Changelog:
v2 -> v3
- Fix new-line-at-end-of-file error in the yaml file

v1 -> v2:
- Convert the binding to yaml
- Squashed the patches adding new binding in a single one

Gregory CLEMENT (5):
dt-bindings: interrupt-controller: convert icpu intr bindings to
json-schema
dt-bindings: interrupt-controller: Add binding for few Microsemi
interrupt controllers
irqchip: ocelot: Add support for Luton platforms
irqchip: ocelot: Add support for Serval platforms
irqchip: ocelot: Add support for Jaguar2 platforms

 .../mscc,ocelot-icpu-intr.txt                 |  21 --
 .../mscc,ocelot-icpu-intr.yaml                |  63 ++++++
 drivers/irqchip/irq-mscc-ocelot.c             | 183 ++++++++++++++++--
 3 files changed, 225 insertions(+), 42 deletions(-)
 delete mode 100644 Documentation/devicetree/bindings/interrupt-controller/mscc,ocelot-icpu-intr.txt
 create mode 100644 Documentation/devicetree/bindings/interrupt-controller/mscc,ocelot-icpu-intr.yaml
Signature · 1 line
2.29.2
#1 lore raw

PATCH v3 1/5 dt-bindings: interrupt-controller: convert icpu intr bindings to json-schema

Convert device tree bindings for Microsemi Ocelot SoC ICPU Interrupt Controller to YAML format

---
 .../mscc,ocelot-icpu-intr.txt                 | 21 -------
 .../mscc,ocelot-icpu-intr.yaml                | 59 +++++++++++++++++++
 2 files changed, 59 insertions(+), 21 deletions(-)
 delete mode 100644 Documentation/devicetree/bindings/interrupt-controller/mscc,ocelot-icpu-intr.txt
 create mode 100644 Documentation/devicetree/bindings/interrupt-controller/mscc,ocelot-icpu-intr.yaml
Documentation/devicetree/bindings/interrupt-controller/mscc,ocelot-icpu-intr.txt deleted +0 -21
deleted file mode 100644
index f5baeccb689f..000000000000
@@ -1,21 +0,0 @@
-Microsemi Ocelot SoC ICPU Interrupt Controller
-
-Required properties:
-
-- compatible : should be "mscc,ocelot-icpu-intr"
-- reg : Specifies base physical address and size of the registers.
-- 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.
-- interrupts : Specifies the CPU interrupt the controller is connected to.
-
-Example:
-
-		intc: interrupt-controller@70000070 {
-			compatible = "mscc,ocelot-icpu-intr";
-			reg = <0x70000070 0x70>;
-			#interrupt-cells = <1>;
-			interrupt-controller;
-			interrupt-parent = <&cpuintc>;
-			interrupts = <2>;
-		};
Documentation/devicetree/bindings/interrupt-controller/mscc,ocelot-icpu-intr.yaml new file +59 -0

59 lines not on this page · raw message

Signature · 1 line
2.29.2
maintainer
#2 lore raw

Re: [PATCH v3 1/5] dt-bindings: interrupt-controller: convert icpu intr bindings to json-schema

Quoted 11 lines · Gregory CLEMENT

Convert device tree bindings for Microsemi Ocelot SoC ICPU Interrupt Controller to YAML format

---
 .../mscc,ocelot-icpu-intr.txt                 | 21 -------
 .../mscc,ocelot-icpu-intr.yaml                | 59 +++++++++++++++++++
 2 files changed, 59 insertions(+), 21 deletions(-)
 delete mode 100644 Documentation/devicetree/bindings/interrupt-controller/mscc,ocelot-icpu-intr.txt
 create mode 100644 Documentation/devicetree/bindings/interrupt-controller/mscc,ocelot-icpu-intr.yaml

My bot found errors running 'make dt_binding_check' on your patch:

yamllint warnings/errors:

dtschema/dtc warnings/errors:
/builds/robherring/linux-dt-review/Documentation/devicetree/bindings/interrupt-controller/mscc,ocelot-icpu-intr.example.dt.yaml: interrupt-controller@70000070: '#address-cells' is a required property
From schema: /builds/robherring/linux-dt-review/Documentation/devicetree/bindings/interrupt-controller/mscc,ocelot-icpu-intr.yaml

See https://patchwork.ozlabs.org/patch/1401029

The base for the patch is generally the last rc1. Any dependencies should be noted.

If you already ran 'make dt_binding_check' and didn't see the above error(s), then make sure 'yamllint' is installed and dt-schema is up to date:

pip3 install dtschema --upgrade

Please check and re-submit.

#3 lore raw

PATCH v3 2/5 dt-bindings: interrupt-controller: Add binding for few Microsemi interrupt controllers

In reply to Gregory CLEMENT

Add the Device Tree binding documentation for the Microsemi Jaguar2, Luton and Serval interrupt controller that is part of the ICPU. It is connected directly to the MIPS core interrupt controller.

---
 .../bindings/interrupt-controller/mscc,ocelot-icpu-intr.yaml  | 4 ++++
 1 file changed, 4 insertions(+)
Documentation/devicetree/bindings/interrupt-controller/mscc,ocelot-icpu-intr.yaml +4 -0
index 3a537635a859..5483ed7062ba 100644
@@ -21,7 +21,11 @@ properties:
   compatible:
     items:
       - enum:
+          - mscc,jaguar2-icpu-intr
+          - mscc,luton-icpu-intr
           - mscc,ocelot-icpu-intr
+          - mscc,serval-icpu-intr
+
 
   '#interrupt-cells':
     const: 1
Signature · 1 line
2.29.2
#4 lore raw

Re: [PATCH v3 2/5] dt-bindings: interrupt-controller: Add binding for few Microsemi interrupt controllers

Hello,

Quoted 14 lines · Gregory CLEMENT

Add the Device Tree binding documentation for the Microsemi Jaguar2, Luton and Serval interrupt controller that is part of the ICPU. It is connected directly to the MIPS core interrupt controller.

---
 .../bindings/interrupt-controller/mscc,ocelot-icpu-intr.yaml  | 4 ++++
 1 file changed, 4 insertions(+)
Documentation/devicetree/bindings/interrupt-controller/mscc,ocelot-icpu-intr.yaml +0 -0
index 3a537635a859..5483ed7062ba 100644

Documentation/devicetree/bindings/interrupt-controller/mscc,ocelot-icpu-intr.yaml properties: · line 28

Quoted 8 lines
Documentation/devicetree/bindings/interrupt-controller/mscc,ocelot-icpu-intr.yaml +4 -0
@@ -21,7 +21,11 @@ properties:
   compatible:
     items:
       - enum:
+          - mscc,jaguar2-icpu-intr
+          - mscc,luton-icpu-intr
           - mscc,ocelot-icpu-intr
+          - mscc,serval-icpu-intr
+

Spurious blank line

Quoted 6 lines
Documentation/devicetree/bindings/interrupt-controller/mscc,ocelot-icpu-intr.yaml +0 -0
   '#interrupt-cells':
     const: 1
Signature · 1 line
2.29.2
Signature · 3 lines
Alexandre Belloni, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com
#5 lore raw

PATCH v3 3/5 irqchip: ocelot: Add support for Luton platforms

In reply to Gregory CLEMENT

This patch extends irqchip driver for oceleot to be used with an other vcoreiii base platform: Luton.

Based on a larger patch from Lars Povlsen <[email protected]>

---
 drivers/irqchip/irq-mscc-ocelot.c | 145 +++++++++++++++++++++++++-----
 1 file changed, 123 insertions(+), 22 deletions(-)
drivers/irqchip/irq-mscc-ocelot.c +123 -22
index 88143c0b700c..9964800c53c2 100644
@@ -12,39 +12,115 @@
 #include <linux/irqchip/chained_irq.h>
 #include <linux/interrupt.h>
 
-#define ICPU_CFG_INTR_INTR_STICKY	0x10
-#define ICPU_CFG_INTR_INTR_ENA		0x18
-#define ICPU_CFG_INTR_INTR_ENA_CLR	0x1c
-#define ICPU_CFG_INTR_INTR_ENA_SET	0x20
-#define ICPU_CFG_INTR_DST_INTR_IDENT(x)	(0x38 + 0x4 * (x))
-#define ICPU_CFG_INTR_INTR_TRIGGER(x)	(0x5c + 0x4 * (x))
-
-#define OCELOT_NR_IRQ 24
+#define ICPU_CFG_INTR_DST_INTR_IDENT(_p, x)	(_p->reg_off_ident + 0x4 * (x))
+#define ICPU_CFG_INTR_INTR_TRIGGER(_p, x)	(_p->reg_off_trigger + 0x4 * (x))
+
+#define FLAGS_NEED_INIT_ENABLE	BIT(0)
+#define FLAGS_FORCE_LUTON_STYLE	BIT(1)
+#define FLAGS_HAS_TRIGGER	BIT(2)
+
+struct chip_props {
+	u32 flags;
+	u32 reg_off_sticky;
+	u32 reg_off_ena;
+	u32 reg_off_ena_clr;
+	u32 reg_off_ena_set;
+	u32 reg_off_ident;
+	u32 reg_off_trigger;
+	u32 reg_off_force;
+	u32 reg_off_ena_irq0;
+	u32 n_irq;
+};
+
+static const struct chip_props ocelot_props = {
+	.flags			= FLAGS_HAS_TRIGGER,
+	.reg_off_sticky		= 0x10,
+	.reg_off_ena		= 0x18,
+	.reg_off_ena_clr	= 0x1c,
+	.reg_off_ena_set	= 0x20,
+	.reg_off_ident		= 0x38,
+	.reg_off_trigger	= 0x5c,
+	.reg_off_force		= 0xc,
+	.n_irq			= 24,
+};
+
+static const struct chip_props luton_props = {
+	.flags			= FLAGS_NEED_INIT_ENABLE |
+				  FLAGS_FORCE_LUTON_STYLE,
+	.reg_off_sticky		= 0,
+	.reg_off_ena		= 0x4,
+	.reg_off_ena_clr	= 0x8,
+	.reg_off_ena_set	= 0xc,
+	.reg_off_ident		= 0x18,
+	.reg_off_trigger	= 0,
+	.reg_off_force		= 0x38,
+	.reg_off_ena_irq0	= 0x14,
+	.n_irq			= 28,
+};
 
 static void ocelot_irq_unmask(struct irq_data *data)
 {
 	struct irq_chip_generic *gc = irq_data_get_irq_chip_data(data);
+	struct irq_domain *d = data->domain;
+	struct chip_props *p = d->host_data;
 	struct irq_chip_type *ct = irq_data_get_chip_type(data);
 	unsigned int mask = data->mask;
 	u32 val;
 
 	irq_gc_lock(gc);
-	val = irq_reg_readl(gc, ICPU_CFG_INTR_INTR_TRIGGER(0)) |
-	      irq_reg_readl(gc, ICPU_CFG_INTR_INTR_TRIGGER(1));
-	if (!(val & mask))
-		irq_reg_writel(gc, mask, ICPU_CFG_INTR_INTR_STICKY);
+	if (p->flags & FLAGS_HAS_TRIGGER) {
+		val = irq_reg_readl(gc, ICPU_CFG_INTR_INTR_TRIGGER(p, 0)) |
+			irq_reg_readl(gc, ICPU_CFG_INTR_INTR_TRIGGER(p, 1));
+		if (!(val & mask))
+			irq_reg_writel(gc, mask, p->reg_off_sticky);
+	}
 
 	*ct->mask_cache &= ~mask;
-	irq_reg_writel(gc, mask, ICPU_CFG_INTR_INTR_ENA_SET);
+	irq_reg_writel(gc, mask, p->reg_off_ena_set);
 	irq_gc_unlock(gc);
 }
 
+static void luton_irq_force(struct irq_data *data,
+			    struct irq_chip_generic *gc,
+			    struct chip_props *p)
+{
+	int off = p->reg_off_force + (data->hwirq * sizeof(u32));
+	u32 val = irq_reg_readl(gc, off);
+
+	irq_reg_writel(gc, val | BIT(3), off);
+}
+
+static int ocelot_irq_force(struct irq_data *data,
+			    enum irqchip_irq_state which, bool state)
+{
+	struct irq_chip_generic *gc = irq_data_get_irq_chip_data(data);
+	struct irq_domain *d = data->domain;
+	struct chip_props *p = d->host_data;
+	int ret = -EINVAL;
+
+	/* Only supports triggering */
+	if ((which == IRQCHIP_STATE_PENDING ||
+	     which == IRQCHIP_STATE_ACTIVE) &&
+	    state && p->reg_off_force) {
+		if (p->flags & FLAGS_FORCE_LUTON_STYLE)
+			/* Config register style */
+			luton_irq_force(data, gc, p);
+		else
+			/* New, bitmask style */
+			irq_reg_writel(gc, data->mask, p->reg_off_force);
+		ret = 0;
+	}
+
+	return ret;
+}
+
 static void ocelot_irq_handler(struct irq_desc *desc)
 {
 	struct irq_chip *chip = irq_desc_get_chip(desc);
 	struct irq_domain *d = irq_desc_get_handler_data(desc);
+	struct chip_props *p = d->host_data;
 	struct irq_chip_generic *gc = irq_get_domain_generic_chip(d, 0);
-	u32 reg = irq_reg_readl(gc, ICPU_CFG_INTR_DST_INTR_IDENT(0));
+	u32 reg = irq_reg_readl(gc, ICPU_CFG_INTR_DST_INTR_IDENT(p, 0));
 
 	chained_irq_enter(chip, desc);
 
@@ -58,25 +134,28 @@ static void ocelot_irq_handler(struct irq_desc *desc)
 	chained_irq_exit(chip, desc);
 }
 
-static int __init ocelot_irq_init(struct device_node *node,
-				  struct device_node *parent)
+static int __init vcoreiii_irq_init(struct device_node *node,
+				    struct device_node *parent,
+				    const struct chip_props *p)
 {
 	struct irq_domain *domain;
 	struct irq_chip_generic *gc;
 	int parent_irq, ret;
 
+	pr_info("%s: Load, %d irqs\n", node->name, p->n_irq);
+
 	parent_irq = irq_of_parse_and_map(node, 0);
 	if (!parent_irq)
 		return -EINVAL;
 
-	domain = irq_domain_add_linear(node, OCELOT_NR_IRQ,
+	domain = irq_domain_add_linear(node, p->n_irq,
 				       &irq_generic_chip_ops, NULL);
 	if (!domain) {
 		pr_err("%pOFn: unable to add irq domain\n", node);
 		return -ENOMEM;
 	}
 
-	ret = irq_alloc_domain_generic_chips(domain, OCELOT_NR_IRQ, 1,
+	ret = irq_alloc_domain_generic_chips(domain, p->n_irq, 1,
 					     "icpu", handle_level_irq,
 					     0, 0, 0);
 	if (ret) {
@@ -92,16 +171,23 @@ static int __init ocelot_irq_init(struct device_node *node,
 		goto err_gc_free;
 	}
 
-	gc->chip_types[0].regs.ack = ICPU_CFG_INTR_INTR_STICKY;
-	gc->chip_types[0].regs.mask = ICPU_CFG_INTR_INTR_ENA_CLR;
+	gc->chip_types[0].regs.ack = p->reg_off_sticky;
+	gc->chip_types[0].regs.mask = p->reg_off_ena_clr;
 	gc->chip_types[0].chip.irq_ack = irq_gc_ack_set_bit;
 	gc->chip_types[0].chip.irq_mask = irq_gc_mask_set_bit;
 	gc->chip_types[0].chip.irq_unmask = ocelot_irq_unmask;
+	gc->chip_types[0].chip.irq_unmask = ocelot_irq_unmask;
+	gc->chip_types[0].chip.irq_set_irqchip_state = ocelot_irq_force;
 
 	/* Mask and ack all interrupts */
-	irq_reg_writel(gc, 0, ICPU_CFG_INTR_INTR_ENA);
-	irq_reg_writel(gc, 0xffffffff, ICPU_CFG_INTR_INTR_STICKY);
+	irq_reg_writel(gc, 0, p->reg_off_ena);
+	irq_reg_writel(gc, 0xffffffff, p->reg_off_sticky);
+
+	/* Overall init */
+	if (p->flags & FLAGS_NEED_INIT_ENABLE)
+		irq_reg_writel(gc, BIT(0), p->reg_off_ena_irq0);
 
+	domain->host_data = (void *)p;
 	irq_set_chained_handler_and_data(parent_irq, ocelot_irq_handler,
 					 domain);
 
@@ -115,4 +201,19 @@ static int __init ocelot_irq_init(struct device_node *node,
 
 	return ret;
 }
+
+static int __init ocelot_irq_init(struct device_node *node,
+				  struct device_node *parent)
+{
+	return vcoreiii_irq_init(node, parent, &ocelot_props);
+}
+
 IRQCHIP_DECLARE(ocelot_icpu, "mscc,ocelot-icpu-intr", ocelot_irq_init);
+
+static int __init luton_irq_init(struct device_node *node,
+				 struct device_node *parent)
+{
+	return vcoreiii_irq_init(node, parent, &luton_props);
+}
+
+IRQCHIP_DECLARE(luton_icpu, "mscc,luton-icpu-intr", luton_irq_init);
Signature · 1 line
2.29.2
#6 lore raw

Re: [PATCH v3 3/5] irqchip: ocelot: Add support for Luton platforms

Hi,

Quoted 16 lines · Gregory CLEMENT

static void ocelot_irq_unmask(struct irq_data *data) {

drivers/irqchip/irq-mscc-ocelot.c +4 -4
 	struct irq_chip_generic *gc = irq_data_get_irq_chip_data(data);
+	struct irq_domain *d = data->domain;
+	struct chip_props *p = d->host_data;
 	struct irq_chip_type *ct = irq_data_get_chip_type(data);
 	unsigned int mask = data->mask;
 	u32 val;
 
 	irq_gc_lock(gc);
-	val = irq_reg_readl(gc, ICPU_CFG_INTR_INTR_TRIGGER(0)) |
-	      irq_reg_readl(gc, ICPU_CFG_INTR_INTR_TRIGGER(1));
-	if (!(val & mask))
-		irq_reg_writel(gc, mask, ICPU_CFG_INTR_INTR_STICKY);
+	if (p->flags & FLAGS_HAS_TRIGGER) {
+		val = irq_reg_readl(gc, ICPU_CFG_INTR_INTR_TRIGGER(p, 0)) |

drivers/irqchip/irq-mscc-ocelot.c line 81

Quoted 10 lines
drivers/irqchip/irq-mscc-ocelot.c +5 -1
+			irq_reg_readl(gc, ICPU_CFG_INTR_INTR_TRIGGER(p, 1));
+		if (!(val & mask))
+			irq_reg_writel(gc, mask, p->reg_off_sticky);
+	}
 
 	*ct->mask_cache &= ~mask;
-	irq_reg_writel(gc, mask, ICPU_CFG_INTR_INTR_ENA_SET);
+	irq_reg_writel(gc, mask, p->reg_off_ena_set);
 	irq_gc_unlock(gc);
 }

Looking at that again, I think you should leave this function as is...

Quoted 25 lines
drivers/irqchip/irq-mscc-ocelot.c +24 -0
+static void luton_irq_force(struct irq_data *data,
+			    struct irq_chip_generic *gc,
+			    struct chip_props *p)
+{
+	int off = p->reg_off_force + (data->hwirq * sizeof(u32));
+	u32 val = irq_reg_readl(gc, off);
+
+	irq_reg_writel(gc, val | BIT(3), off);
+}
+
+static int ocelot_irq_force(struct irq_data *data,
+			    enum irqchip_irq_state which, bool state)
+{
+	struct irq_chip_generic *gc = irq_data_get_irq_chip_data(data);
+	struct irq_domain *d = data->domain;
+	struct chip_props *p = d->host_data;
+	int ret = -EINVAL;
+
+	/* Only supports triggering */
+	if ((which == IRQCHIP_STATE_PENDING ||
+	     which == IRQCHIP_STATE_ACTIVE) &&
+	    state && p->reg_off_force) {
+		if (p->flags & FLAGS_FORCE_LUTON_STYLE)
+			/* Config register style */

drivers/irqchip/irq-mscc-ocelot.c line 116

Quoted 10 lines
drivers/irqchip/irq-mscc-ocelot.c +10 -0
+			luton_irq_force(data, gc, p);
+		else
+			/* New, bitmask style */
+			irq_reg_writel(gc, data->mask, p->reg_off_force);
+		ret = 0;
+	}
+
+	return ret;
+}
+

I think the addition of the force function may be separated in an different patch.

drivers/irqchip/irq-mscc-ocelot.c static void ocelot_irq_handler(struct irq_desc *desc) · line 146

Quoted 12 lines
drivers/irqchip/irq-mscc-ocelot.c +5 -2
-static int __init ocelot_irq_init(struct device_node *node,
-				  struct device_node *parent)
+static int __init vcoreiii_irq_init(struct device_node *node,
+				    struct device_node *parent,
+				    const struct chip_props *p)
 {
 	struct irq_domain *domain;
 	struct irq_chip_generic *gc;
 	int parent_irq, ret;
 
+	pr_info("%s: Load, %d irqs\n", node->name, p->n_irq);
+

Is this necessary?

Quoted 19 lines
drivers/irqchip/irq-mscc-ocelot.c +2 -2
 	parent_irq = irq_of_parse_and_map(node, 0);
 	if (!parent_irq)
 		return -EINVAL;
 
-	domain = irq_domain_add_linear(node, OCELOT_NR_IRQ,
+	domain = irq_domain_add_linear(node, p->n_irq,
 				       &irq_generic_chip_ops, NULL);
 	if (!domain) {
 		pr_err("%pOFn: unable to add irq domain\n", node);
 		return -ENOMEM;
 	}
 
-	ret = irq_alloc_domain_generic_chips(domain, OCELOT_NR_IRQ, 1,
+	ret = irq_alloc_domain_generic_chips(domain, p->n_irq, 1,
 					     "icpu", handle_level_irq,
 					     0, 0, 0);
 	if (ret) {
@@ -92,16 +171,23 @@ static int __init ocelot_irq_init(struct device_node *node,
 		goto err_gc_free;

drivers/irqchip/irq-mscc-ocelot.c static int __init ocelot_irq_init(struct device_node *node, · line 179

Quoted 10 lines
drivers/irqchip/irq-mscc-ocelot.c +3 -2
 	}
 
-	gc->chip_types[0].regs.ack = ICPU_CFG_INTR_INTR_STICKY;
-	gc->chip_types[0].regs.mask = ICPU_CFG_INTR_INTR_ENA_CLR;
+	gc->chip_types[0].regs.ack = p->reg_off_sticky;
+	gc->chip_types[0].regs.mask = p->reg_off_ena_clr;
 	gc->chip_types[0].chip.irq_ack = irq_gc_ack_set_bit;
 	gc->chip_types[0].chip.irq_mask = irq_gc_mask_set_bit;
 	gc->chip_types[0].chip.irq_unmask = ocelot_irq_unmask;
+	gc->chip_types[0].chip.irq_unmask = ocelot_irq_unmask;

This is assigned the same member twice.

As said, you can probably leave ocelot_irq_unmask so we avoid having an if in the hot path.

You should test here for triggers and if they are not available, then you can use regs.enable/regs.disable and irq_gc_mask_disable_reg and irq_gc_unmask_enable_reg instead of regs.mask and irq_gc_mask_set_bit/ocelot_irq_unmask

Signature · 3 lines
Alexandre Belloni, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com
#7 lore raw

Re: [PATCH v3 3/5] irqchip: ocelot: Add support for Luton platforms

In reply to Gregory CLEMENT
Quoted 35 lines · Gregory CLEMENT

This patch extends irqchip driver for oceleot to be used with an other vcoreiii base platform: Luton.

Based on a larger patch from Lars Povlsen <[email protected]>

---
 drivers/irqchip/irq-mscc-ocelot.c | 145 +++++++++++++++++++++++++-----
 1 file changed, 123 insertions(+), 22 deletions(-)
drivers/irqchip/irq-mscc-ocelot.c +7 -8
index 88143c0b700c..9964800c53c2 100644
@@ -12,39 +12,115 @@
 #include <linux/irqchip/chained_irq.h>
 #include <linux/interrupt.h>
 
-#define ICPU_CFG_INTR_INTR_STICKY	0x10
-#define ICPU_CFG_INTR_INTR_ENA		0x18
-#define ICPU_CFG_INTR_INTR_ENA_CLR	0x1c
-#define ICPU_CFG_INTR_INTR_ENA_SET	0x20
-#define ICPU_CFG_INTR_DST_INTR_IDENT(x)	(0x38 + 0x4 * (x))
-#define ICPU_CFG_INTR_INTR_TRIGGER(x)	(0x5c + 0x4 * (x))
-
-#define OCELOT_NR_IRQ 24
+#define ICPU_CFG_INTR_DST_INTR_IDENT(_p, x)	(_p->reg_off_ident + 0x4 * 
 (x))
+#define ICPU_CFG_INTR_INTR_TRIGGER(_p, x)	(_p->reg_off_trigger + 0x4 * 
 (x))
+
+#define FLAGS_NEED_INIT_ENABLE	BIT(0)
+#define FLAGS_FORCE_LUTON_STYLE	BIT(1)
+#define FLAGS_HAS_TRIGGER	BIT(2)
+

drivers/irqchip/irq-mscc-ocelot.c line 33

Quoted 10 lines
drivers/irqchip/irq-mscc-ocelot.c +10 -0
+struct chip_props {
+	u32 flags;
+	u32 reg_off_sticky;
+	u32 reg_off_ena;
+	u32 reg_off_ena_clr;
+	u32 reg_off_ena_set;
+	u32 reg_off_ident;
+	u32 reg_off_trigger;
+	u32 reg_off_force;
+	u32 reg_off_ena_irq0;

Do all these fields need to be u32s?

Quoted 8 lines
drivers/irqchip/irq-mscc-ocelot.c +8 -0
+	u32 n_irq;
+};
+
+static const struct chip_props ocelot_props = {
+	.flags			= FLAGS_HAS_TRIGGER,
+	.reg_off_sticky		= 0x10,
+	.reg_off_ena		= 0x18,
+	.reg_off_ena_clr	= 0x1c,

drivers/irqchip/irq-mscc-ocelot.c line 49

Quoted 10 lines
drivers/irqchip/irq-mscc-ocelot.c +10 -0
+	.reg_off_ena_set	= 0x20,
+	.reg_off_ident		= 0x38,
+	.reg_off_trigger	= 0x5c,
+	.reg_off_force		= 0xc,
+	.n_irq			= 24,
+};
+
+static const struct chip_props luton_props = {
+	.flags			= FLAGS_NEED_INIT_ENABLE |
+				  FLAGS_FORCE_LUTON_STYLE,

LUTON_STYLE doesn't quite convey the idea of what this implies. Please use a name that actually means something (other than the city north of London). And force what exactly?

drivers/irqchip/irq-mscc-ocelot.c line 55

Quoted 6 lines
drivers/irqchip/irq-mscc-ocelot.c +6 -0
+	.reg_off_sticky		= 0,
+	.reg_off_ena		= 0x4,
+	.reg_off_ena_clr	= 0x8,
+	.reg_off_ena_set	= 0xc,
+	.reg_off_ident		= 0x18,
+	.reg_off_trigger	= 0,

If this field doesn't exist (because the matching flag isn't set), don't list it.

drivers/irqchip/irq-mscc-ocelot.c line 59

Quoted 4 lines
drivers/irqchip/irq-mscc-ocelot.c +4 -0
+	.reg_off_force		= 0x38,
+	.reg_off_ena_irq0	= 0x14,
+	.n_irq			= 28,
+};

Please split this patch in two: rework the Ocelot driver to the "new" style in one patch, then introduce new platform in another.

Quoted 30 lines
drivers/irqchip/irq-mscc-ocelot.c +11 -5
 static void ocelot_irq_unmask(struct irq_data *data)
 {
 	struct irq_chip_generic *gc = irq_data_get_irq_chip_data(data);
+	struct irq_domain *d = data->domain;
+	struct chip_props *p = d->host_data;
 	struct irq_chip_type *ct = irq_data_get_chip_type(data);
 	unsigned int mask = data->mask;
 	u32 val;
 
 	irq_gc_lock(gc);
-	val = irq_reg_readl(gc, ICPU_CFG_INTR_INTR_TRIGGER(0)) |
-	      irq_reg_readl(gc, ICPU_CFG_INTR_INTR_TRIGGER(1));
-	if (!(val & mask))
-		irq_reg_writel(gc, mask, ICPU_CFG_INTR_INTR_STICKY);
+	if (p->flags & FLAGS_HAS_TRIGGER) {
+		val = irq_reg_readl(gc, ICPU_CFG_INTR_INTR_TRIGGER(p, 0)) |
+			irq_reg_readl(gc, ICPU_CFG_INTR_INTR_TRIGGER(p, 1));
+		if (!(val & mask))
+			irq_reg_writel(gc, mask, p->reg_off_sticky);
+	}
 
 	*ct->mask_cache &= ~mask;
-	irq_reg_writel(gc, mask, ICPU_CFG_INTR_INTR_ENA_SET);
+	irq_reg_writel(gc, mask, p->reg_off_ena_set);
 	irq_gc_unlock(gc);
 }
 
+static void luton_irq_force(struct irq_data *data,
+			    struct irq_chip_generic *gc,

drivers/irqchip/irq-mscc-ocelot.c line 94

Quoted 10 lines
drivers/irqchip/irq-mscc-ocelot.c +10 -0
+			    struct chip_props *p)
+{
+	int off = p->reg_off_force + (data->hwirq * sizeof(u32));
+	u32 val = irq_reg_readl(gc, off);
+
+	irq_reg_writel(gc, val | BIT(3), off);
+}
+
+static int ocelot_irq_force(struct irq_data *data,
+			    enum irqchip_irq_state which, bool state)

Please use a name that matches the same of the API.

drivers/irqchip/irq-mscc-ocelot.c line 103

Quoted 9 lines
drivers/irqchip/irq-mscc-ocelot.c +9 -0
+{
+	struct irq_chip_generic *gc = irq_data_get_irq_chip_data(data);
+	struct irq_domain *d = data->domain;
+	struct chip_props *p = d->host_data;
+	int ret = -EINVAL;
+
+	/* Only supports triggering */
+	if ((which == IRQCHIP_STATE_PENDING ||
+	     which == IRQCHIP_STATE_ACTIVE) &&

PENDING and ACTIVE are two very different things, and yet you handle them the same way. Which one does your interrupt controller support?

Quoted 30 lines
drivers/irqchip/irq-mscc-ocelot.c +15 -2
+	    state && p->reg_off_force) {
+		if (p->flags & FLAGS_FORCE_LUTON_STYLE)
+			/* Config register style */
+			luton_irq_force(data, gc, p);
+		else
+			/* New, bitmask style */
+			irq_reg_writel(gc, data->mask, p->reg_off_force);
+		ret = 0;
+	}
+
+	return ret;
+}
+
 static void ocelot_irq_handler(struct irq_desc *desc)
 {
 	struct irq_chip *chip = irq_desc_get_chip(desc);
 	struct irq_domain *d = irq_desc_get_handler_data(desc);
+	struct chip_props *p = d->host_data;
 	struct irq_chip_generic *gc = irq_get_domain_generic_chip(d, 0);
-	u32 reg = irq_reg_readl(gc, ICPU_CFG_INTR_DST_INTR_IDENT(0));
+	u32 reg = irq_reg_readl(gc, ICPU_CFG_INTR_DST_INTR_IDENT(p, 0));
 
 	chained_irq_enter(chip, desc);
@@ -58,25 +134,28 @@ static void ocelot_irq_handler(struct irq_desc 
 *desc)
 	chained_irq_exit(chip, desc);
 }
 
-static int __init ocelot_irq_init(struct device_node *node,

drivers/irqchip/irq-mscc-ocelot.c static void ocelot_irq_handler(struct irq_desc · line 146

Quoted 10 lines
drivers/irqchip/irq-mscc-ocelot.c +4 -1
-				  struct device_node *parent)
+static int __init vcoreiii_irq_init(struct device_node *node,
+				    struct device_node *parent,
+				    const struct chip_props *p)
 {
 	struct irq_domain *domain;
 	struct irq_chip_generic *gc;
 	int parent_irq, ret;
 
+	pr_info("%s: Load, %d irqs\n", node->name, p->n_irq);

If you must print something, use %pOFn, just like in the rest of the driver.

Quoted 34 lines
drivers/irqchip/irq-mscc-ocelot.c +3 -2
+
 	parent_irq = irq_of_parse_and_map(node, 0);
 	if (!parent_irq)
 		return -EINVAL;
 
-	domain = irq_domain_add_linear(node, OCELOT_NR_IRQ,
+	domain = irq_domain_add_linear(node, p->n_irq,
 				       &irq_generic_chip_ops, NULL);
 	if (!domain) {
 		pr_err("%pOFn: unable to add irq domain\n", node);
 		return -ENOMEM;
 	}
 
-	ret = irq_alloc_domain_generic_chips(domain, OCELOT_NR_IRQ, 1,
+	ret = irq_alloc_domain_generic_chips(domain, p->n_irq, 1,
 					     "icpu", handle_level_irq,
 					     0, 0, 0);
drivers/irqchip/irq-mscc-ocelot.c +4 -2
 	if (ret) {
@@ -92,16 +171,23 @@ static int __init ocelot_irq_init(struct 
 device_node *node,
 		goto err_gc_free;
 	}
 
-	gc->chip_types[0].regs.ack = ICPU_CFG_INTR_INTR_STICKY;
-	gc->chip_types[0].regs.mask = ICPU_CFG_INTR_INTR_ENA_CLR;
+	gc->chip_types[0].regs.ack = p->reg_off_sticky;
+	gc->chip_types[0].regs.mask = p->reg_off_ena_clr;
 	gc->chip_types[0].chip.irq_ack = irq_gc_ack_set_bit;
 	gc->chip_types[0].chip.irq_mask = irq_gc_mask_set_bit;
 	gc->chip_types[0].chip.irq_unmask = ocelot_irq_unmask;
+	gc->chip_types[0].chip.irq_unmask = ocelot_irq_unmask;
+	gc->chip_types[0].chip.irq_set_irqchip_state = ocelot_irq_force;
 
 	/* Mask and ack all interrupts */

drivers/irqchip/irq-mscc-ocelot.c static int __init ocelot_irq_init(struct · line 191

Quoted 10 lines
drivers/irqchip/irq-mscc-ocelot.c +7 -2
-	irq_reg_writel(gc, 0, ICPU_CFG_INTR_INTR_ENA);
-	irq_reg_writel(gc, 0xffffffff, ICPU_CFG_INTR_INTR_STICKY);
+	irq_reg_writel(gc, 0, p->reg_off_ena);
+	irq_reg_writel(gc, 0xffffffff, p->reg_off_sticky);
+
+	/* Overall init */
+	if (p->flags & FLAGS_NEED_INIT_ENABLE)
+		irq_reg_writel(gc, BIT(0), p->reg_off_ena_irq0);
 
+	domain->host_data = (void *)p;

Useless cast.

Quoted 17 lines
drivers/irqchip/irq-mscc-ocelot.c +7 -0
 	irq_set_chained_handler_and_data(parent_irq, ocelot_irq_handler,
 					 domain);
@@ -115,4 +201,19 @@ static int __init ocelot_irq_init(struct 
 device_node *node,
 
 	return ret;
 }
+
+static int __init ocelot_irq_init(struct device_node *node,
+				  struct device_node *parent)
+{
+	return vcoreiii_irq_init(node, parent, &ocelot_props);
+}
+
 IRQCHIP_DECLARE(ocelot_icpu, "mscc,ocelot-icpu-intr", 

ocelot_irq_init);

drivers/irqchip/irq-mscc-ocelot.c static int __init ocelot_irq_init(struct device_node *node, · line 219

Quoted 8 lines
drivers/irqchip/irq-mscc-ocelot.c +8 -0
+
+static int __init luton_irq_init(struct device_node *node,
+				 struct device_node *parent)
+{
+	return vcoreiii_irq_init(node, parent, &luton_props);
+}
+
+IRQCHIP_DECLARE(luton_icpu, "mscc,luton-icpu-intr", luton_irq_init);

Thanks,

        M.
Signature · 1 line
Jazz is not dead. It just smells funny...
#8 lore raw

PATCH v3 4/5 irqchip: ocelot: Add support for Serval platforms

In reply to Gregory CLEMENT

This patch extends irqchip driver for ocelot to be used with an other vcoreiii base platform: Serval.

Based on a larger patch from Lars Povlsen <[email protected]>

---
 drivers/irqchip/irq-mscc-ocelot.c | 20 ++++++++++++++++++++
 1 file changed, 20 insertions(+)
drivers/irqchip/irq-mscc-ocelot.c +20 -0
index 9964800c53c2..584af3b0a9e2 100644
@@ -44,6 +44,18 @@ static const struct chip_props ocelot_props = {
 	.n_irq			= 24,
 };
 
+static const struct chip_props serval_props = {
+	.flags			= FLAGS_HAS_TRIGGER,
+	.reg_off_sticky		= 0xc,
+	.reg_off_ena		= 0x14,
+	.reg_off_ena_clr	= 0x18,
+	.reg_off_ena_set	= 0x1c,
+	.reg_off_ident		= 0x20,
+	.reg_off_trigger	= 0x4,
+	.reg_off_force		= 0x8,
+	.n_irq			= 24,
+};
+
 static const struct chip_props luton_props = {
 	.flags			= FLAGS_NEED_INIT_ENABLE |
 				  FLAGS_FORCE_LUTON_STYLE,
@@ -210,6 +222,14 @@ static int __init ocelot_irq_init(struct device_node *node,
 
 IRQCHIP_DECLARE(ocelot_icpu, "mscc,ocelot-icpu-intr", ocelot_irq_init);
 
+static int __init serval_irq_init(struct device_node *node,
+				  struct device_node *parent)
+{
+	return vcoreiii_irq_init(node, parent, &serval_props);
+}
+
+IRQCHIP_DECLARE(serval_icpu, "mscc,serval-icpu-intr", serval_irq_init);
+
 static int __init luton_irq_init(struct device_node *node,
 				 struct device_node *parent)
 {
Signature · 1 line
2.29.2
Acked-by on v3 4/5
#9 lore raw

Re: [PATCH v3 4/5] irqchip: ocelot: Add support for Serval platforms

Quoted 5 lines · Gregory CLEMENT

This patch extends irqchip driver for ocelot to be used with an other vcoreiii base platform: Serval.

Based on a larger patch from Lars Povlsen <[email protected]>

Quoted 45 lines
---
 drivers/irqchip/irq-mscc-ocelot.c | 20 ++++++++++++++++++++
 1 file changed, 20 insertions(+)
drivers/irqchip/irq-mscc-ocelot.c +20 -0
index 9964800c53c2..584af3b0a9e2 100644
@@ -44,6 +44,18 @@ static const struct chip_props ocelot_props = {
 	.n_irq			= 24,
 };
 
+static const struct chip_props serval_props = {
+	.flags			= FLAGS_HAS_TRIGGER,
+	.reg_off_sticky		= 0xc,
+	.reg_off_ena		= 0x14,
+	.reg_off_ena_clr	= 0x18,
+	.reg_off_ena_set	= 0x1c,
+	.reg_off_ident		= 0x20,
+	.reg_off_trigger	= 0x4,
+	.reg_off_force		= 0x8,
+	.n_irq			= 24,
+};
+
 static const struct chip_props luton_props = {
 	.flags			= FLAGS_NEED_INIT_ENABLE |
 				  FLAGS_FORCE_LUTON_STYLE,
@@ -210,6 +222,14 @@ static int __init ocelot_irq_init(struct device_node *node,
 
 IRQCHIP_DECLARE(ocelot_icpu, "mscc,ocelot-icpu-intr", ocelot_irq_init);
 
+static int __init serval_irq_init(struct device_node *node,
+				  struct device_node *parent)
+{
+	return vcoreiii_irq_init(node, parent, &serval_props);
+}
+
+IRQCHIP_DECLARE(serval_icpu, "mscc,serval-icpu-intr", serval_irq_init);
+
 static int __init luton_irq_init(struct device_node *node,
 				 struct device_node *parent)
 {
Signature · 1 line
2.29.2
Signature · 3 lines
Alexandre Belloni, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com
#10 lore raw

PATCH v3 5/5 irqchip: ocelot: Add support for Jaguar2 platforms

In reply to Gregory CLEMENT

This patch extends irqchip driver for ocelot to be used with an other vcoreiii base platform: Jaguar2.

Based on a larger patch from Lars Povlsen <[email protected]>

---
 drivers/irqchip/irq-mscc-ocelot.c | 20 ++++++++++++++++++++
 1 file changed, 20 insertions(+)
drivers/irqchip/irq-mscc-ocelot.c +20 -0
index 584af3b0a9e2..0dfea8771172 100644
@@ -70,6 +70,18 @@ static const struct chip_props luton_props = {
 	.n_irq			= 28,
 };
 
+static const struct chip_props jaguar2_props = {
+	.flags			= FLAGS_HAS_TRIGGER,
+	.reg_off_sticky		= 0x10,
+	.reg_off_ena		= 0x18,
+	.reg_off_ena_clr	= 0x1c,
+	.reg_off_ena_set	= 0x20,
+	.reg_off_ident		= 0x38,
+	.reg_off_trigger	= 0x5c,
+	.reg_off_force		= 0xc,
+	.n_irq			= 29,
+};
+
 static void ocelot_irq_unmask(struct irq_data *data)
 {
 	struct irq_chip_generic *gc = irq_data_get_irq_chip_data(data);
@@ -237,3 +249,11 @@ static int __init luton_irq_init(struct device_node *node,
 }
 
 IRQCHIP_DECLARE(luton_icpu, "mscc,luton-icpu-intr", luton_irq_init);
+
+static int __init jaguar2_irq_init(struct device_node *node,
+				   struct device_node *parent)
+{
+	return vcoreiii_irq_init(node, parent, &jaguar2_props);
+}
+
+IRQCHIP_DECLARE(jaguar2_icpu, "mscc,jaguar2-icpu-intr", jaguar2_irq_init);
Signature · 1 line
2.29.2
Acked-by on v3 5/5
#11 lore raw

Re: [PATCH v3 5/5] irqchip: ocelot: Add support for Jaguar2 platforms

Quoted 5 lines · Gregory CLEMENT

This patch extends irqchip driver for ocelot to be used with an other vcoreiii base platform: Jaguar2.

Based on a larger patch from Lars Povlsen <[email protected]>

Quoted 42 lines
---
 drivers/irqchip/irq-mscc-ocelot.c | 20 ++++++++++++++++++++
 1 file changed, 20 insertions(+)
drivers/irqchip/irq-mscc-ocelot.c +20 -0
index 584af3b0a9e2..0dfea8771172 100644
@@ -70,6 +70,18 @@ static const struct chip_props luton_props = {
 	.n_irq			= 28,
 };
 
+static const struct chip_props jaguar2_props = {
+	.flags			= FLAGS_HAS_TRIGGER,
+	.reg_off_sticky		= 0x10,
+	.reg_off_ena		= 0x18,
+	.reg_off_ena_clr	= 0x1c,
+	.reg_off_ena_set	= 0x20,
+	.reg_off_ident		= 0x38,
+	.reg_off_trigger	= 0x5c,
+	.reg_off_force		= 0xc,
+	.n_irq			= 29,
+};
+
 static void ocelot_irq_unmask(struct irq_data *data)
 {
 	struct irq_chip_generic *gc = irq_data_get_irq_chip_data(data);
@@ -237,3 +249,11 @@ static int __init luton_irq_init(struct device_node *node,
 }
 
 IRQCHIP_DECLARE(luton_icpu, "mscc,luton-icpu-intr", luton_irq_init);
+
+static int __init jaguar2_irq_init(struct device_node *node,
+				   struct device_node *parent)
+{
+	return vcoreiii_irq_init(node, parent, &jaguar2_props);
+}
+
+IRQCHIP_DECLARE(jaguar2_icpu, "mscc,jaguar2-icpu-intr", jaguar2_irq_init);
Signature · 1 line
2.29.2
Signature · 3 lines
Alexandre Belloni, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com