Skip to contentSkip to conversation

IRQ type flags

RFC In review Russell King 6 participants 16 messages Nov 6, 2005 – Dec 15, 2005

Subsystems (7)
lore
Share & tools
lore
#0 lore raw

RFC 1/1 IRQ type flags

I haven't had any feedback on this patch. akpm - can you add it to -mm please? Here's the sign-off for it, thanks.

----- Forwarded message from Russell King <[email protected]> -----
Date: Fri, 28 Oct 2005 22:57:47 +0100
From: Russell King <[email protected]>
To: Linux Kernel List <[email protected]>
Subject: [RFC] IRQ type flags

Hi,

Some ARM platforms have the ability to program the interrupt controller to detect various interrupt edges and/or levels. For some platforms, this is critical to setup correctly, particularly those which the setting is dependent on the device.

Currently, ARM drivers do (eg) the following:

	err = request_irq(irq, ...);

	set_irq_type(irq, IRQT_RISING);

However, if the interrupt has previously been programmed to be level sensitive (for whatever reason) then this will cause an interrupt storm.

Hence, if we combine set_irq_type() with request_irq(), we can then safely set the type prior to unmasking the interrupt. The unfortunate problem is that in order to support this, these flags need to be visible outside of the ARM architecture - drivers such as smc91x need these flags and they're cross-architecture.

Finally, the SA_TRIGGER_* flag passed to request_irq() should reflect the property that the device would like. The IRQ controller code should do its best to select the most appropriate supported mode.

Comments?

arch/arm/kernel/irq.c +14 -2
@@ -681,10 +681,16 @@ int setup_irq(unsigned int irq, struct i
 	 */
 	desc = irq_desc + irq;
 	spin_lock_irqsave(&irq_controller_lock, flags);
+#define SA_TRIGGER	(SA_TRIGGER_HIGH|SA_TRIGGER_LOW|\
+			 SA_TRIGGER_RISING|SA_TRIGGER_FALLING)
 	p = &desc->action;
 	if ((old = *p) != NULL) {
-		/* Can't share interrupts unless both agree to */
-		if (!(old->flags & new->flags & SA_SHIRQ)) {
+		/*
+		 * Can't share interrupts unless both agree to and are
+		 * the same type.
+		 */
+		if (!(old->flags & new->flags & SA_SHIRQ) ||
+		    (~old->flags & new->flags) & SA_TRIGGER) {
 			spin_unlock_irqrestore(&irq_controller_lock, flags);
 			return -EBUSY;
 		}
@@ -704,6 +710,12 @@ int setup_irq(unsigned int irq, struct i
 		desc->running = 0;
 		desc->pending = 0;
 		desc->disable_depth = 1;
+
+		if (new->flags & SA_TRIGGER) {
+			unsigned int type = new->flags & SA_TRIGGER;
+			desc->chip->set_type(irq, type);
+		}
+
 		if (!desc->noautoenable) {
 			desc->disable_depth = 0;
 			desc->chip->unmask(irq);
include/linux/signal.h +8 -0
@@ -18,6 +18,14 @@
 #define SA_PROBE		SA_ONESHOT
 #define SA_SAMPLE_RANDOM	SA_RESTART
 #define SA_SHIRQ		0x04000000
+/*
+ * As above, these correspond to the __IRQT defines in asm-arm/irq.h
+ * to select the interrupt line behaviour.
+ */
+#define SA_TRIGGER_HIGH		0x00000008
+#define SA_TRIGGER_LOW		0x00000004
+#define SA_TRIGGER_RISING	0x00000002
+#define SA_TRIGGER_FALLING	0x00000001
 
 /*
  * Real Time signals may be queued.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

----- End forwarded message -----

Signature · 3 lines
Russell King
 Linux kernel    2.6 ARM Linux   - http://www.arm.linux.org.uk/
 maintainer of:  2.6 Serial core
#1 lore raw
Quoted 5 lines · Russell King

Finally, the SA_TRIGGER_* flag passed to request_irq() should reflect the property that the device would like. The IRQ controller code should do its best to select the most appropriate supported mode.

Comments?

This is actually true of some x86 hardware in the EISA space where there is a control register for level v edge that we sort of half deal with.

#2 lore raw
Quoted 9 lines · Alan Cox
Quoted 5 lines · Russell King

Finally, the SA_TRIGGER_* flag passed to request_irq() should reflect the property that the device would like. The IRQ controller code should do its best to select the most appropriate supported mode.

Comments?

This is actually true of some x86 hardware in the EISA space where there is a control register for level v edge that we sort of half deal with.

Thanks Alan. Can I assume you're happy with the patch, even if x86 currently ignores the flags?

Signature · 3 lines
Russell King
 Linux kernel    2.6 ARM Linux   - http://www.arm.linux.org.uk/
 maintainer of:  2.6 Serial core
#3 lore raw
Quoted 5 lines · Russell King
Quoted 2 lines

This is actually true of some x86 hardware in the EISA space where there is a control register for level v edge that we sort of half deal with.

Thanks Alan. Can I assume you're happy with the patch, even if x86 currently ignores the flags?

I'm certainly happy with it. Both the APIC and EISA IRQ control registers could even be made to honour it if anyone ever needed to.

Should platforms that don't support the flags be patched to error requests they don't support however ?

Alan

#4 lore raw
Quoted 4 lines · Alan Cox
Quoted 3 lines · Russell King
Quoted 2 lines

This is actually true of some x86 hardware in the EISA space where there is a control register for level v edge that we sort of half deal with.

Quoted 8 lines
Quoted 2 lines

Thanks Alan. Can I assume you're happy with the patch, even if x86 currently ignores the flags?

I'm certainly happy with it. Both the APIC and EISA IRQ control registers could even be made to honour it if anyone ever needed to.

Should platforms that don't support the flags be patched to error requests they don't support however ?

Good question - I'm not sure currently. In the places where set_irq_type is used on ARM, we're mainly interested in setting the input according to rising/falling edge or high/low levels rather than switching between edge and level mode.

We could do as you suggest, but my concern would be adding extra complexity to drivers, causing them to do something like:

	ret = request_irq(..., SA_TRIGGER_HIGH, ...);
	if (ret == -E<whatever>)
		ret = request_irq(..., SA_TRIGGER_RISING, ...);

The alternative is:

	ret = request_irq(..., SA_TRIGGER_HIGH | SA_TRIGGER_RISING, ...);

at which point the driver is telling the IRQ layer what it can support, and allowing the IRQ layer to select the most appropriate type given the driver and hardware constraints.

Signature · 3 lines
Russell King
 Linux kernel    2.6 ARM Linux   - http://www.arm.linux.org.uk/
 maintainer of:  2.6 Serial core
#5 lore raw
Quoted 10 lines · Russell King

We could do as you suggest, but my concern would be adding extra complexity to drivers, causing them to do something like:

	ret = request_irq(..., SA_TRIGGER_HIGH, ...);
	if (ret == -E<whatever>)
		ret = request_irq(..., SA_TRIGGER_RISING, ...);

The alternative is:

	ret = request_irq(..., SA_TRIGGER_HIGH | SA_TRIGGER_RISING, ...);

I was thinking that specifying neither would imply 'don't care' or 'system default'. That would mean existing drivers just worked and driver authors who didnt care need take no specific action.

#6 lore raw
Quoted 4 lines

I was thinking that specifying neither would imply 'don't care' or 'system default'. That would mean existing drivers just worked and driver authors who didnt care need take no specific action.

Not only that, but as I wrote in my other mail, on some archs, the system already got all the necessary triggers from the firmware and has setup sane defaults, in which case it's more up to the driver to either not care or adapt to what was set (some IRQ controllers don't support all possible settings for example).

I would also expect PCI drivers to always pass default (rather than explicitly level negative)

Ben.

#7 lore raw
In reply to Alan Cox
Quoted 8 lines · Alan Cox
Quoted 7 lines · Russell King

We could do as you suggest, but my concern would be adding extra complexity to drivers, causing them to do something like:

	ret = request_irq(..., SA_TRIGGER_HIGH, ...);
	if (ret == -E<whatever>)
		ret = request_irq(..., SA_TRIGGER_RISING, ...);
Quoted 7 lines
Quoted 3 lines

The alternative is:

	ret = request_irq(..., SA_TRIGGER_HIGH | SA_TRIGGER_RISING, ...);

I was thinking that specifying neither would imply 'don't care' or 'system default'. That would mean existing drivers just worked and driver authors who didnt care need take no specific action.

Yes, this is exactly what the ARM implementation already does. I'll add a comment to that effect.

As per benh's suggestion, I don't see the point of adding a definition
- not unless we're going to fix up all drivers which call request_irq(). That would be a very big task.

Signature · 3 lines
Russell King
 Linux kernel    2.6 ARM Linux   - http://www.arm.linux.org.uk/
 maintainer of:  2.6 Serial core
#8 lore raw
In reply to Russell King
Quoted 13 lines

Good question - I'm not sure currently. In the places where set_irq_type is used on ARM, we're mainly interested in setting the input according to rising/falling edge or high/low levels rather than switching between edge and level mode.

We could do as you suggest, but my concern would be adding extra complexity to drivers, causing them to do something like:

	ret = request_irq(..., SA_TRIGGER_HIGH, ...);
	if (ret == -E<whatever>)
		ret = request_irq(..., SA_TRIGGER_RISING, ...);
Quoted 7 lines

The alternative is:

	ret = request_irq(..., SA_TRIGGER_HIGH | SA_TRIGGER_RISING, ...);

at which point the driver is telling the IRQ layer what it can support, and allowing the IRQ layer to select the most appropriate type given the driver and hardware constraints.

We have similar things on ppc but dealt differently. The type of interrupt sense supported depends on the PIC (and you can have more than one PIC around). On Open Firmware based machines at least, the device-tree tells us the sense setting to use on all devices in the system, we program our PIC based on that information.

Your proposal though is interesting as it would allow individual to override that setting (it may be broken, firmware occasionally are), and would probably be useful for embedded PPCs as well.

However, I would suggest defining the absence of explicit trigger (0) as a constant rather than 0 (SA_TRIGGER_DEFAULT), just for consistency.That means using whatever the platform considers as a good default for this interrupt. In addition, I would still add a get_irq_trigger() function for a driver to enquire the kind of sense that was set by default by the platfrom for a given irq line.

Ben.

#9 lore raw
In reply to Russell King
Quoted 20 lines · Russell King

I haven't had any feedback on this patch. akpm - can you add it to -mm please? Here's the sign-off for it, thanks.

----- Forwarded message from Russell King <[email protected]> -----
Date: Fri, 28 Oct 2005 22:57:47 +0100
From: Russell King <[email protected]>
To: Linux Kernel List <[email protected]>
Subject: [RFC] IRQ type flags

Hi,

Some ARM platforms have the ability to program the interrupt controller to detect various interrupt edges and/or levels. For some platforms, this is critical to setup correctly, particularly those which the setting is dependent on the device.

Currently, ARM drivers do (eg) the following:

Quoted 7 lines
	err = request_irq(irq, ...);

	set_irq_type(irq, IRQT_RISING);

However, if the interrupt has previously been programmed to be level sensitive (for whatever reason) then this will cause an interrupt storm.

surely, thats the other way around, going from edge to level.

Quoted 5 lines

Hence, if we combine set_irq_type() with request_irq(), we can then safely set the type prior to unmasking the interrupt. The unfortunate problem is that in order to support this, these flags need to be visible outside of the ARM architecture - drivers such as smc91x need these flags and they're cross-architecture.

I agree that the type of IRQ should be considered when registering the IRQ. On the s3c2410, most boards do set the level/edge correctly before startup (bootloader) but occasionally, the bootloader cannot deal with all the cases.

Quoted 46 lines

Finally, the SA_TRIGGER_* flag passed to request_irq() should reflect the property that the device would like. The IRQ controller code should do its best to select the most appropriate supported mode.

Comments?

arch/arm/kernel/irq.c +14 -2
@@ -681,10 +681,16 @@ int setup_irq(unsigned int irq, struct i
 	 */
 	desc = irq_desc + irq;
 	spin_lock_irqsave(&irq_controller_lock, flags);
+#define SA_TRIGGER	(SA_TRIGGER_HIGH|SA_TRIGGER_LOW|\
+			 SA_TRIGGER_RISING|SA_TRIGGER_FALLING)
 	p = &desc->action;
 	if ((old = *p) != NULL) {
-		/* Can't share interrupts unless both agree to */
-		if (!(old->flags & new->flags & SA_SHIRQ)) {
+		/*
+		 * Can't share interrupts unless both agree to and are
+		 * the same type.
+		 */
+		if (!(old->flags & new->flags & SA_SHIRQ) ||
+		    (~old->flags & new->flags) & SA_TRIGGER) {
 			spin_unlock_irqrestore(&irq_controller_lock, flags);
 			return -EBUSY;
 		}
@@ -704,6 +710,12 @@ int setup_irq(unsigned int irq, struct i
 		desc->running = 0;
 		desc->pending = 0;
 		desc->disable_depth = 1;
+
+		if (new->flags & SA_TRIGGER) {
+			unsigned int type = new->flags & SA_TRIGGER;
+			desc->chip->set_type(irq, type);
+		}
+
 		if (!desc->noautoenable) {
 			desc->disable_depth = 0;
 			desc->chip->unmask(irq);
include/linux/signal.h +0 -0
@@ -18,6 +18,14 @@
 #define SA_PROBE		SA_ONESHOT

include/linux/signal.h line 28

Quoted 10 lines
include/linux/signal.h +8 -0
 #define SA_SAMPLE_RANDOM	SA_RESTART
 #define SA_SHIRQ		0x04000000
+/*
+ * As above, these correspond to the __IRQT defines in asm-arm/irq.h
+ * to select the interrupt line behaviour.
+ */
+#define SA_TRIGGER_HIGH		0x00000008
+#define SA_TRIGGER_LOW		0x00000004
+#define SA_TRIGGER_RISING	0x00000002
+#define SA_TRIGGER_FALLING	0x00000001

How about making these compatible with the
triggers compatible with the flags from
include/linux/ioport.h definitions for the
IRQ resource (IORESOURCE_IRQ_*).

This would make it easier to pass the resource's
flags field to the register irq code, and get
the right IRQ type for the app?

Signature · 3 lines
Ben ([email protected], http://www.fluff.org/)

  'a smiley only costs 4 bytes'
#10 lore raw
Quoted 23 lines · Ben Dooks
Quoted 22 lines · Russell King

I haven't had any feedback on this patch. akpm - can you add it to -mm please? Here's the sign-off for it, thanks.

----- Forwarded message from Russell King <[email protected]> -----
Date: Fri, 28 Oct 2005 22:57:47 +0100
From: Russell King <[email protected]>
To: Linux Kernel List <[email protected]>
Subject: [RFC] IRQ type flags

Hi,

Some ARM platforms have the ability to program the interrupt controller to detect various interrupt edges and/or levels. For some platforms, this is critical to setup correctly, particularly those which the setting is dependent on the device.

Currently, ARM drivers do (eg) the following:

	err = request_irq(irq, ...);
Quoted 7 lines
Quoted 5 lines
	set_irq_type(irq, IRQT_RISING);

However, if the interrupt has previously been programmed to be level sensitive (for whatever reason) then this will cause an interrupt storm.

surely, thats the other way around, going from edge to level.

No. How can an edge triggered interrupt possibly cause a storm?

If the interrupt signal is at logic '0' and it was previously configured for a low level to interrupt, when you call request_irq() with your handler, the interrupt will be enabled.

Because it's level sensitive, it'll immediately interrupt, and the handler will be invoked. When the handler returns, the interrupt signal will still be at logic '0' so it's still active. So you immediately get another interrupt and the handler will be re-invoked. Repeat infinitely.

So it is exactly as I describe.

Quoted 4 lines

How about making these compatible with the
triggers compatible with the flags from
include/linux/ioport.h definitions for the
IRQ resource (IORESOURCE_IRQ_*).

We could do, but I took the set_irq_type() implementation. 8) We could change both to conform.

Signature · 3 lines
Russell King
 Linux kernel    2.6 ARM Linux   - http://www.arm.linux.org.uk/
 maintainer of:  2.6 Serial core
#11 lore raw
In reply to Russell King

Here's an updated patch, taking account of folks comments. The changes from the previous patch are:

* Move SA_TRIGGER into linux/signal.h
* Change SA_TRIGGER_* to match IORESOURCE_IRQ_* definitions
* Change ARM __IRQT_* definitions to match IORESOURCE_IRQ_* definitions
* Add a comment to explain the case where no SA_TRIGGER_* flags are passed to request_irq.

----
Some ARM platforms have the ability to program the interrupt controller to detect various interrupt edges and/or levels. For some platforms, this is critical to setup correctly, particularly those which the setting is dependent on the device.

Currently, ARM drivers do (eg) the following:

	err = request_irq(irq, ...);

	set_irq_type(irq, IRQT_RISING);

However, if the interrupt has previously been programmed to be level sensitive (for whatever reason) then this will cause an interrupt storm.

Hence, if we combine set_irq_type() with request_irq(), we can then safely set the type prior to unmasking the interrupt. The unfortunate problem is that in order to support this, these flags need to be visible outside of the ARM architecture - drivers such as smc91x need these flags and they're cross-architecture.

Finally, the SA_TRIGGER_* flag passed to request_irq() should reflect the property that the device would like. The IRQ controller code should do its best to select the most appropriate supported mode.

---

 arch/arm/kernel/irq.c  |   14 ++++++++++++--
 include/asm-arm/irq.h  |   12 ++++++++----
 include/linux/signal.h |   13 +++++++++++++
 3 files changed, 33 insertions(+), 6 deletions(-)
arch/arm/kernel/irq.c +12 -2
@@ -684,8 +684,12 @@ int setup_irq(unsigned int irq, struct i
 	spin_lock_irqsave(&irq_controller_lock, flags);
 	p = &desc->action;
 	if ((old = *p) != NULL) {
-		/* Can't share interrupts unless both agree to */
-		if (!(old->flags & new->flags & SA_SHIRQ)) {
+		/*
+		 * Can't share interrupts unless both agree to and are
+		 * the same type.
+		 */
+		if (!(old->flags & new->flags & SA_SHIRQ) ||
+		    (~old->flags & new->flags) & SA_TRIGGER_MASK) {
 			spin_unlock_irqrestore(&irq_controller_lock, flags);
 			return -EBUSY;
 		}
@@ -705,6 +709,12 @@ int setup_irq(unsigned int irq, struct i
 		desc->running = 0;
 		desc->pending = 0;
 		desc->disable_depth = 1;
+
+		if (new->flags & SA_TRIGGER_MASK) {
+			unsigned int type = new->flags & SA_TRIGGER;
+			desc->chip->set_type(irq, type);
+		}
+
 		if (!desc->noautoenable) {
 			desc->disable_depth = 0;
 			desc->chip->unmask(irq);
include/asm-arm/irq.h +8 -4
@@ -25,10 +25,14 @@ extern void disable_irq_nosync(unsigned 
 extern void disable_irq(unsigned int);
 extern void enable_irq(unsigned int);
 
-#define __IRQT_FALEDGE	(1 << 0)
-#define __IRQT_RISEDGE	(1 << 1)
-#define __IRQT_LOWLVL	(1 << 2)
-#define __IRQT_HIGHLVL	(1 << 3)
+/*
+ * These correspond with the SA_TRIGGER_* defines, and therefore the
+ * IRQRESOURCE_IRQ_* defines.
+ */
+#define __IRQT_RISEDGE	(1 << 0)
+#define __IRQT_FALEDGE	(1 << 1)
+#define __IRQT_HIGHLVL	(1 << 2)
+#define __IRQT_LOWLVL	(1 << 3)
 
 #define IRQT_NOEDGE	(0)
 #define IRQT_RISING	(__IRQT_RISEDGE)
include/linux/signal.h +13 -0
@@ -18,6 +18,19 @@
 #define SA_PROBE		SA_ONESHOT
 #define SA_SAMPLE_RANDOM	SA_RESTART
 #define SA_SHIRQ		0x04000000
+/*
+ * As above, these correspond to the IORESOURCE_IRQ_* defines in
+ * linux/ioport.h to select the interrupt line behaviour.  When
+ * requesting an interrupt without specifying a SA_TRIGGER, the
+ * setting should be assumed to be "as already configured", which
+ * may be as per machine or firmware initialisation.
+ */
+#define SA_TRIGGER_LOW		0x00000008
+#define SA_TRIGGER_HIGH		0x00000004
+#define SA_TRIGGER_FALLING	0x00000002
+#define SA_TRIGGER_RISING	0x00000001
+#define SA_TRIGGER_MASK	(SA_TRIGGER_HIGH|SA_TRIGGER_LOW|\
+				 SA_TRIGGER_RISING|SA_TRIGGER_FALLING)
 
 /*
  * Real Time signals may be queued.
Signature · 3 lines
Russell King
 Linux kernel    2.6 ARM Linux   - http://www.arm.linux.org.uk/
 maintainer of:  2.6 Serial core
#12 lore raw

Arrange for drivers which do:

	request_irq(irq, ...);
	set_irq_type(irq, ...);

or vice versa to use the new SA_TRIGGER flags with request_irq().

---

 arch/arm/mach-omap1/serial.c       |    3 +--
 arch/arm/mach-pxa/corgi.c          |    7 +++----
 arch/arm/mach-pxa/poodle.c         |    7 +++----
 arch/arm/mach-pxa/spitz.c          |    7 +++----
 arch/arm/mach-s3c2410/usb-simtec.c |    6 +++---
 drivers/i2c/chips/tps65010.c       |   11 ++++++-----
 drivers/input/keyboard/corgikbd.c  |    6 ++----
 drivers/input/keyboard/spitzkbd.c  |   27 ++++++++++++++-------------
 drivers/mfd/ucb1x00-core.c         |    5 ++---
 drivers/net/smc91x.c               |    5 +----
 drivers/net/smc91x.h               |   18 +++++++++---------
 11 files changed, 47 insertions(+), 55 deletions(-)
arch/arm/mach-omap1/serial.c +1 -2
@@ -252,9 +252,8 @@ static void __init omap_serial_set_port_
 		return;
 	}
 	omap_set_gpio_direction(gpio_nr, 1);
-	set_irq_type(OMAP_GPIO_IRQ(gpio_nr), IRQT_RISING);
 	ret = request_irq(OMAP_GPIO_IRQ(gpio_nr), &omap_serial_wake_interrupt,
-			  0, "serial wakeup", NULL);
+			  SA_TRIGGER_RISING, "serial wakeup", NULL);
 	if (ret) {
 		omap_free_gpio(gpio_nr);
 		printk(KERN_ERR "No interrupt for UART wake GPIO: %i\n",
arch/arm/mach-pxa/corgi.c +3 -4
@@ -213,15 +213,14 @@ static int corgi_mci_init(struct device 
 
 	corgi_mci_platform_data.detect_delay = msecs_to_jiffies(250);
 
-	err = request_irq(CORGI_IRQ_GPIO_nSD_DETECT, corgi_detect_int, SA_INTERRUPT,
-			     "MMC card detect", data);
+	err = request_irq(CORGI_IRQ_GPIO_nSD_DETECT, corgi_detect_int,
+			  SA_INTERRUPT | SA_TRIGGER_RISING | SA_TRIGGER_FALLING,
+			  "MMC card detect", data);
 	if (err) {
 		printk(KERN_ERR "corgi_mci_init: MMC/SD: can't request MMC card detect IRQ\n");
 		return -1;
 	}
 
-	set_irq_type(CORGI_IRQ_GPIO_nSD_DETECT, IRQT_BOTHEDGE);
-
 	return 0;
 }
 
arch/arm/mach-pxa/poodle.c +3 -4
@@ -146,15 +146,14 @@ static int poodle_mci_init(struct device
 
 	poodle_mci_platform_data.detect_delay = msecs_to_jiffies(250);
 
-	err = request_irq(POODLE_IRQ_GPIO_nSD_DETECT, poodle_detect_int, SA_INTERRUPT,
-			     "MMC card detect", data);
+	err = request_irq(POODLE_IRQ_GPIO_nSD_DETECT, poodle_detect_int,
+			  SA_INTERRUPT | SA_TRIGGER_RISING | SA_TRIGGER_FALLING,
+			  "MMC card detect", data);
 	if (err) {
 		printk(KERN_ERR "poodle_mci_init: MMC/SD: can't request MMC card detect IRQ\n");
 		return -1;
 	}
 
-	set_irq_type(POODLE_IRQ_GPIO_nSD_DETECT, IRQT_BOTHEDGE);
-
 	return 0;
 }
 
arch/arm/mach-pxa/spitz.c +3 -4
@@ -293,15 +293,14 @@ static int spitz_mci_init(struct device 
 
 	spitz_mci_platform_data.detect_delay = msecs_to_jiffies(250);
 
-	err = request_irq(SPITZ_IRQ_GPIO_nSD_DETECT, spitz_detect_int, SA_INTERRUPT,
-			     "MMC card detect", data);
+	err = request_irq(SPITZ_IRQ_GPIO_nSD_DETECT, spitz_detect_int,
+			  SA_INTERRUPT | SA_TRIGGER_RISING | SA_TRIGGER_FALLING,
+			  "MMC card detect", data);
 	if (err) {
 		printk(KERN_ERR "spitz_mci_init: MMC/SD: can't request MMC card detect IRQ\n");
 		return -1;
 	}
 
-	set_irq_type(SPITZ_IRQ_GPIO_nSD_DETECT, IRQT_BOTHEDGE);
-
 	return 0;
 }
 
arch/arm/mach-s3c2410/usb-simtec.c +3 -3
@@ -84,13 +84,13 @@ static void usb_simtec_enableoc(struct s
 	int ret;
 
 	if (on) {
-		ret = request_irq(IRQ_USBOC, usb_simtec_ocirq, SA_INTERRUPT,
+		ret = request_irq(IRQ_USBOC, usb_simtec_ocirq,
+				  SA_INTERRUPT | SA_TRIGGER_RISING |
+				   SA_TRIGGER_FALLING,
 				  "USB Over-current", info);
 		if (ret != 0) {
 			printk(KERN_ERR "failed to request usb oc irq\n");
 		}
-
-		set_irq_type(IRQ_USBOC, IRQT_BOTHEDGE);
 	} else {
 		free_irq(IRQ_USBOC, info);
 	}
drivers/i2c/chips/tps65010.c +6 -5
@@ -494,6 +494,7 @@ tps65010_probe(struct i2c_adapter *bus, 
 {
 	struct tps65010		*tps;
 	int			status;
+	unsigned long		irqflags;
 
 	if (the_tps) {
 		dev_dbg(&bus->dev, "only one %s for now\n", DRIVER_NAME);
@@ -520,13 +521,14 @@ tps65010_probe(struct i2c_adapter *bus, 
 	}
 
 #ifdef	CONFIG_ARM
+	irqflags = SA_SAMPLE_RANDOM | SA_TRIGGER_LOW;
 	if (machine_is_omap_h2()) {
 		tps->model = TPS65010;
 		omap_cfg_reg(W4_GPIO58);
 		tps->irq = OMAP_GPIO_IRQ(58);
 		omap_request_gpio(58);
 		omap_set_gpio_direction(58, 1);
-		set_irq_type(tps->irq, IRQT_FALLING);
+		irqflags |= SA_TRIGGER_FALLING;
 	}
 	if (machine_is_omap_osk()) {
 		tps->model = TPS65010;
@@ -534,7 +536,7 @@ tps65010_probe(struct i2c_adapter *bus, 
 		tps->irq = OMAP_GPIO_IRQ(OMAP_MPUIO(1));
 		omap_request_gpio(OMAP_MPUIO(1));
 		omap_set_gpio_direction(OMAP_MPUIO(1), 1);
-		set_irq_type(tps->irq, IRQT_FALLING);
+		irqflags |= SA_TRIGGER_FALLING;
 	}
 	if (machine_is_omap_h3()) {
 		tps->model = TPS65013;
@@ -542,13 +544,12 @@ tps65010_probe(struct i2c_adapter *bus, 
 		// FIXME set up this board's IRQ ...
 	}
 #else
-#define set_irq_type(num,trigger)	do{}while(0)
+	irqflags = SA_SAMPLE_RANDOM;
 #endif
 
 	if (tps->irq > 0) {
-		set_irq_type(tps->irq, IRQT_LOW);
 		status = request_irq(tps->irq, tps65010_irq,
-			SA_SAMPLE_RANDOM, DRIVER_NAME, tps);
+			irqflags, DRIVER_NAME, tps);
 		if (status < 0) {
 			dev_dbg(&tps->client.dev, "can't get IRQ %d, err %d\n",
 					tps->irq, status);
drivers/input/keyboard/corgikbd.c +2 -4
@@ -19,7 +19,6 @@
 #include <linux/jiffies.h>
 #include <linux/module.h>
 #include <linux/slab.h>
-#include <asm/irq.h>
 
 #include <asm/arch/corgi.h>
 #include <asm/arch/hardware.h>
@@ -343,10 +342,9 @@ static int __init corgikbd_probe(struct 
 	for (i = 0; i < CORGI_KEY_SENSE_NUM; i++) {
 		pxa_gpio_mode(CORGI_GPIO_KEY_SENSE(i) | GPIO_IN);
 		if (request_irq(CORGI_IRQ_GPIO_KEY_SENSE(i), corgikbd_interrupt,
-						SA_INTERRUPT, "corgikbd", corgikbd))
+				SA_INTERRUPT | SA_TRIGGER_RISING,
+				"corgikbd", corgikbd))
 			printk(KERN_WARNING "corgikbd: Can't get IRQ: %d!\n", i);
-		else
-			set_irq_type(CORGI_IRQ_GPIO_KEY_SENSE(i),IRQT_RISING);
 	}
 
 	/* Set Strobe lines as outputs - set high */
drivers/input/keyboard/spitzkbd.c +14 -13
@@ -19,7 +19,6 @@
 #include <linux/jiffies.h>
 #include <linux/module.h>
 #include <linux/slab.h>
-#include <asm/irq.h>
 
 #include <asm/arch/spitz.h>
 #include <asm/arch/hardware.h>
@@ -407,10 +406,9 @@ static int __init spitzkbd_probe(struct 
 	for (i = 0; i < SPITZ_KEY_SENSE_NUM; i++) {
 		pxa_gpio_mode(spitz_senses[i] | GPIO_IN);
 		if (request_irq(IRQ_GPIO(spitz_senses[i]), spitzkbd_interrupt,
-						SA_INTERRUPT, "Spitzkbd Sense", spitzkbd))
+				SA_INTERRUPT|SA_TRIGGER_RISING,
+				"Spitzkbd Sense", spitzkbd))
 			printk(KERN_WARNING "spitzkbd: Can't get Sense IRQ: %d!\n", i);
-		else
-			set_irq_type(IRQ_GPIO(spitz_senses[i]),IRQT_RISING);
 	}
 
 	/* Set Strobe lines as outputs - set high */
@@ -422,15 +420,18 @@ static int __init spitzkbd_probe(struct 
 	pxa_gpio_mode(SPITZ_GPIO_SWA | GPIO_IN);
 	pxa_gpio_mode(SPITZ_GPIO_SWB | GPIO_IN);
 
-	request_irq(SPITZ_IRQ_GPIO_SYNC, spitzkbd_interrupt, SA_INTERRUPT, "Spitzkbd Sync", spitzkbd);
-	request_irq(SPITZ_IRQ_GPIO_ON_KEY, spitzkbd_interrupt, SA_INTERRUPT, "Spitzkbd PwrOn", spitzkbd);
-	request_irq(SPITZ_IRQ_GPIO_SWA, spitzkbd_hinge_isr, SA_INTERRUPT, "Spitzkbd SWA", spitzkbd);
-	request_irq(SPITZ_IRQ_GPIO_SWB, spitzkbd_hinge_isr, SA_INTERRUPT, "Spitzkbd SWB", spitzkbd);
-
-	set_irq_type(SPITZ_IRQ_GPIO_SYNC, IRQT_BOTHEDGE);
-	set_irq_type(SPITZ_IRQ_GPIO_ON_KEY, IRQT_BOTHEDGE);
-	set_irq_type(SPITZ_IRQ_GPIO_SWA, IRQT_BOTHEDGE);
-	set_irq_type(SPITZ_IRQ_GPIO_SWB, IRQT_BOTHEDGE);
+	request_irq(SPITZ_IRQ_GPIO_SYNC, spitzkbd_interrupt,
+		    SA_INTERRUPT | SA_TRIGGER_RISING | SA_TRIGGER_FALLING,
+		    "Spitzkbd Sync", spitzkbd);
+	request_irq(SPITZ_IRQ_GPIO_ON_KEY, spitzkbd_interrupt,
+		    SA_INTERRUPT | SA_TRIGGER_RISING | SA_TRIGGER_FALLING,
+		    "Spitzkbd PwrOn", spitzkbd);
+	request_irq(SPITZ_IRQ_GPIO_SWA, spitzkbd_hinge_isr,
+		    SA_INTERRUPT | SA_TRIGGER_RISING | SA_TRIGGER_FALLING,
+		    "Spitzkbd SWA", spitzkbd);
+	request_irq(SPITZ_IRQ_GPIO_SWB, spitzkbd_hinge_isr,
+		    SA_INTERRUPT | SA_TRIGGER_RISING | SA_TRIGGER_FALLING,
+		    "Spitzkbd SWB", spitzkbd);
 
 	printk(KERN_INFO "input: Spitz Keyboard Registered\n");
 
drivers/mfd/ucb1x00-core.c +2 -3
@@ -27,7 +27,6 @@
 
 #include <asm/dma.h>
 #include <asm/hardware.h>
-#include <asm/irq.h>
 
 #include "ucb1x00.h"
 
@@ -507,14 +506,14 @@ static int ucb1x00_probe(struct mcp *mcp
 		goto err_free;
 	}
 
-	ret = request_irq(ucb->irq, ucb1x00_irq, 0, "UCB1x00", ucb);
+	ret = request_irq(ucb->irq, ucb1x00_irq, SA_TRIGGER_RISING,
+			  "UCB1x00", ucb);
 	if (ret) {
 		printk(KERN_ERR "ucb1x00: unable to grab irq%d: %d\n",
 			ucb->irq, ret);
 		goto err_free;
 	}
 
-	set_irq_type(ucb->irq, IRQT_RISING);
 	mcp_set_drvdata(mcp, ucb);
 
 	ret = class_device_register(&ucb->cdev);
drivers/net/smc91x.c +1 -4
@@ -88,7 +88,6 @@ static const char version[] =
 #include <linux/skbuff.h>
 
 #include <asm/io.h>
-#include <asm/irq.h>
 
 #include "smc91x.h"
 
@@ -2007,12 +2006,10 @@ static int __init smc_probe(struct net_d
 	}
 
 	/* Grab the IRQ */
-      	retval = request_irq(dev->irq, &smc_interrupt, 0, dev->name, dev);
+      	retval = request_irq(dev->irq, &smc_interrupt, SMC_IRQ_FLAGS, dev->name, dev);
       	if (retval)
       		goto err_out;
 
-	set_irq_type(dev->irq, SMC_IRQ_TRIGGER_TYPE);
-
 #ifdef SMC_USE_PXA_DMA
 	{
 		int dma = pxa_request_dma(dev->name, DMA_PRIO_LOW,
drivers/net/smc91x.h +9 -9
@@ -90,7 +90,7 @@
 			__l--;						\
 		}							\
 	} while (0)
-#define set_irq_type(irq, type)
+#define SMC_IRQ_FLAGS		(0)
 
 #elif defined(CONFIG_SA1100_PLEB)
 /* We can only do 16-bit reads and writes in the static memory space. */
@@ -109,7 +109,7 @@
 #define SMC_outw(v, a, r)	writew(v, (a) + (r))
 #define SMC_outsw(a, r, p, l)	writesw((a) + (r), p, l)
 
-#define set_irq_type(irq, type) do {} while (0)
+#define SMC_IRQ_FLAGS		(0)
 
 #elif defined(CONFIG_SA1100_ASSABET)
 
@@ -185,11 +185,11 @@ SMC_outw(u16 val, void __iomem *ioaddr, 
 #include <asm/mach-types.h>
 #include <asm/arch/cpu.h>
 
-#define	SMC_IRQ_TRIGGER_TYPE (( \
+#define	SMC_IRQ_FLAGS (( \
 		   machine_is_omap_h2() \
 		|| machine_is_omap_h3() \
 		|| (machine_is_omap_innovator() && !cpu_is_omap1510()) \
-	) ? IRQT_FALLING : IRQT_RISING)
+	) ? SA_TRIGGER_FALLING : SA_TRIGGER_RISING)
 
 
 #elif	defined(CONFIG_SH_SH4202_MICRODEV)
@@ -209,7 +209,7 @@ SMC_outw(u16 val, void __iomem *ioaddr, 
 #define SMC_insw(a, r, p, l)	insw((a) + (r) - 0xa0000000, p, l)
 #define SMC_outsw(a, r, p, l)	outsw((a) + (r) - 0xa0000000, p, l)
 
-#define set_irq_type(irq, type)	do {} while(0)
+#define SMC_IRQ_FLAGS		(0)
 
 #elif	defined(CONFIG_ISA)
 
@@ -237,7 +237,7 @@ SMC_outw(u16 val, void __iomem *ioaddr, 
 #define SMC_insw(a, r, p, l)	insw(((u32)a) + (r), p, l)
 #define SMC_outsw(a, r, p, l)	outsw(((u32)a) + (r), p, l)
 
-#define set_irq_type(irq, type)	do {} while(0)
+#define SMC_IRQ_FLAGS		(0)
 
 #define RPC_LSA_DEFAULT		RPC_LED_TX_RX
 #define RPC_LSB_DEFAULT		RPC_LED_100_10
@@ -319,7 +319,7 @@ static inline void SMC_outsw (unsigned l
 			au_writew(*_p++ , _a); \
 	} while(0)
 
-#define set_irq_type(irq, type) do {} while (0)
+#define SMC_IRQ_FLAGS		(0)
 
 #else
 
@@ -342,8 +342,8 @@ static inline void SMC_outsw (unsigned l
 
 #endif
 
-#ifndef	SMC_IRQ_TRIGGER_TYPE
-#define	SMC_IRQ_TRIGGER_TYPE	IRQT_RISING
+#ifndef	SMC_IRQ_FLAGS
+#define	SMC_IRQ_FLAGS		SA_TRIGGER_RISING
 #endif
 
 #ifdef SMC_USE_PXA_DMA
Signature · 3 lines
Russell King
 Linux kernel    2.6 ARM Linux   - http://www.arm.linux.org.uk/
 maintainer of:  2.6 Serial core
#13 lore raw
In reply to Russell King
Quoted 77 lines · Russell King

Here's an updated patch, taking account of folks comments. The changes from the previous patch are:

* Move SA_TRIGGER into linux/signal.h
* Change SA_TRIGGER_* to match IORESOURCE_IRQ_* definitions
* Change ARM __IRQT_* definitions to match IORESOURCE_IRQ_* definitions
* Add a comment to explain the case where no SA_TRIGGER_* flags are passed to request_irq.

----
Some ARM platforms have the ability to program the interrupt controller to detect various interrupt edges and/or levels. For some platforms, this is critical to setup correctly, particularly those which the setting is dependent on the device.

Currently, ARM drivers do (eg) the following:

	err = request_irq(irq, ...);

	set_irq_type(irq, IRQT_RISING);

However, if the interrupt has previously been programmed to be level sensitive (for whatever reason) then this will cause an interrupt storm.

Hence, if we combine set_irq_type() with request_irq(), we can then safely set the type prior to unmasking the interrupt. The unfortunate problem is that in order to support this, these flags need to be visible outside of the ARM architecture - drivers such as smc91x need these flags and they're cross-architecture.

Finally, the SA_TRIGGER_* flag passed to request_irq() should reflect the property that the device would like. The IRQ controller code should do its best to select the most appropriate supported mode.

---

 arch/arm/kernel/irq.c  |   14 ++++++++++++--
 include/asm-arm/irq.h  |   12 ++++++++----
 include/linux/signal.h |   13 +++++++++++++
 3 files changed, 33 insertions(+), 6 deletions(-)
arch/arm/kernel/irq.c +12 -2
@@ -684,8 +684,12 @@ int setup_irq(unsigned int irq, struct i
 	spin_lock_irqsave(&irq_controller_lock, flags);
 	p = &desc->action;
 	if ((old = *p) != NULL) {
-		/* Can't share interrupts unless both agree to */
-		if (!(old->flags & new->flags & SA_SHIRQ)) {
+		/*
+		 * Can't share interrupts unless both agree to and are
+		 * the same type.
+		 */
+		if (!(old->flags & new->flags & SA_SHIRQ) ||
+		    (~old->flags & new->flags) & SA_TRIGGER_MASK) {
 			spin_unlock_irqrestore(&irq_controller_lock, flags);
 			return -EBUSY;
 		}
@@ -705,6 +709,12 @@ int setup_irq(unsigned int irq, struct i
 		desc->running = 0;
 		desc->pending = 0;
 		desc->disable_depth = 1;
+
+		if (new->flags & SA_TRIGGER_MASK) {
+			unsigned int type = new->flags & SA_TRIGGER;
+			desc->chip->set_type(irq, type);
+		}
+
 		if (!desc->noautoenable) {
 			desc->disable_depth = 0;
 			desc->chip->unmask(irq);

include/asm-arm/irq.h extern void disable_irq_nosync(unsigned · line 30

Quoted 10 lines
include/asm-arm/irq.h +3 -4
@@ -25,10 +25,14 @@ extern void disable_irq_nosync(unsigned
 extern void disable_irq(unsigned int);
 extern void enable_irq(unsigned int);
 
-#define __IRQT_FALEDGE	(1 << 0)
-#define __IRQT_RISEDGE	(1 << 1)
-#define __IRQT_LOWLVL	(1 << 2)
-#define __IRQT_HIGHLVL	(1 << 3)
+/*
+ * These correspond with the SA_TRIGGER_* defines, and therefore the
+ * IRQRESOURCE_IRQ_* defines.

comment nit. Should be IORESOURCE_IRQ_* not IRQRESOURCE_IRQ_*

Quoted 18 lines
include/asm-arm/irq.h +5 -0
+ */
+#define __IRQT_RISEDGE	(1 << 0)
+#define __IRQT_FALEDGE	(1 << 1)
+#define __IRQT_HIGHLVL	(1 << 2)
+#define __IRQT_LOWLVL	(1 << 3)
 
 #define IRQT_NOEDGE	(0)
 #define IRQT_RISING	(__IRQT_RISEDGE)
include/linux/signal.h +3 -0
@@ -18,6 +18,19 @@
 #define SA_PROBE		SA_ONESHOT
 #define SA_SAMPLE_RANDOM	SA_RESTART
 #define SA_SHIRQ		0x04000000
+/*
+ * As above, these correspond to the IORESOURCE_IRQ_* defines in
+ * linux/ioport.h to select the interrupt line behaviour.  When

include/linux/signal.h line 33

Quoted 10 lines
include/linux/signal.h +10 -0
+ * requesting an interrupt without specifying a SA_TRIGGER, the
+ * setting should be assumed to be "as already configured", which
+ * may be as per machine or firmware initialisation.
+ */
+#define SA_TRIGGER_LOW		0x00000008
+#define SA_TRIGGER_HIGH		0x00000004
+#define SA_TRIGGER_FALLING	0x00000002
+#define SA_TRIGGER_RISING	0x00000001
+#define SA_TRIGGER_MASK	(SA_TRIGGER_HIGH|SA_TRIGGER_LOW|\
+				 SA_TRIGGER_RISING|SA_TRIGGER_FALLING)

Do you mind expand the comment to convey that LOW/HIGH are related to level sensitive interrupts and FALLING/RISING are for edge. This is different naming that I'm used to with PowerPC and it can get a little confusing.

- kumar

#14 lore raw
Quoted 4 lines · Kumar Gala
Quoted 2 lines

+ * These correspond with the SA_TRIGGER_* defines, and therefore the
+ * IRQRESOURCE_IRQ_* defines.

comment nit. Should be IORESOURCE_IRQ_* not IRQRESOURCE_IRQ_*

Added, thanks.

Quoted 8 lines
Quoted 7 lines

+/*
+ * As above, these correspond to the IORESOURCE_IRQ_* defines in
+ * linux/ioport.h to select the interrupt line behaviour. When
+ * requesting an interrupt without specifying a SA_TRIGGER, the
+ * setting should be assumed to be "as already configured", which
+ * may be as per machine or firmware initialisation.
+ */

Quoted 4 lines

Do you mind expand the comment to convey that LOW/HIGH are related to level sensitive interrupts and FALLING/RISING are for edge. This is different naming that I'm used to with PowerPC and it can get a little confusing.

Also added. Thanks.

Signature · 3 lines
Russell King
 Linux kernel    2.6 ARM Linux   - http://www.arm.linux.org.uk/
 maintainer of:  2.6 Serial core
#15 lore raw
In reply to Russell King

Hi Russell,

Quoted 1 line · Russell King

* Move SA_TRIGGER into linux/signal.h

...

arch/arm/kernel/irq.c int setup_irq(unsigned int irq, struct i · line 714

Quoted 7 lines
arch/arm/kernel/irq.c +3 -0
@@ -705,6 +709,12 @@ int setup_irq(unsigned int irq, struct i
 		desc->running = 0;
 		desc->pending = 0;
 		desc->disable_depth = 1;
+
+		if (new->flags & SA_TRIGGER_MASK) {
+			unsigned int type = new->flags & SA_TRIGGER;

Hmm, where is SA_TRIGGER defined?

include/linux/signal.h line 33

Quoted 6 lines
include/linux/signal.h +6 -0
+#define SA_TRIGGER_LOW		0x00000008
+#define SA_TRIGGER_HIGH		0x00000004
+#define SA_TRIGGER_FALLING	0x00000002
+#define SA_TRIGGER_RISING	0x00000001
+#define SA_TRIGGER_MASK	(SA_TRIGGER_HIGH|SA_TRIGGER_LOW|\
+				 SA_TRIGGER_RISING|SA_TRIGGER_FALLING)

Thanks