[PATCH] powerpc: fix trigger handling in the new irq code

This patch slightly reworks the new irq code to fix a small design error.  I
removed the passing of the trigger to the map() calls entirely, it was not a
good idea to have one call do two different things.  It also fixes a couple of
corner cases.

Mapping a linux virtual irq to a physical irq now does only that.  Setting the
trigger is a different action which has a different call.

The main changes are:

- I no longer call host->ops->map() for an already mapped irq, I just return
  the virtual number that was already mapped.  It was called before to give an
  opportunity to change the trigger, but that was causing issues as that could
  happen while the interrupt was in use by a device, and because of the
  trigger change, map would potentially muck around with things in a racy way.
   That was causing much burden on a given's controller implementation of
  map() to get it right.  This is much simpler now.  map() is only called on
  the initial mapping of an irq, meaning that you know that this irq is _not_
  being used.  You can initialize the hardware if you want (though you don't
  have to).

- Controllers that can handle different type of triggers (level/edge/etc...)
  now implement the standard irq_chip->set_type() call as defined by the
  generic code.  That means that you can use the standard set_irq_type() to
  configure an irq line manually if you wish or (though I don't like that
  interface), pass explicit trigger flags to request_irq() as defined by the
  generic kernel interfaces.  Also, using those interfaces guarantees that
  your controller set_type callback is called with the descriptor lock held,
  thus providing locking against activity on the same interrupt (including
  mask/unmask/etc...) automatically.  A result is that, for example, MPIC's
  own map() implementation calls irq_set_type(NONE) to configure the hardware
  to the default triggers.

- To allow the above, the irq_map array entry for the new mapped interrupt
  is now set before map() callback is called for the controller.

- The irq_create_of_mapping() (also used by irq_of_parse_and_map()) function
  for mapping interrupts from the device-tree now also call the separate
  set_irq_type(), and only does so if there is a change in the trigger type.

- While I was at it, I changed pci_read_irq_line() (which is the helper I
  would expect most archs to use in their pcibios_fixup() to get the PCI
  interrupt routing from the device tree) to also handle a fallback when the
  DT mapping fails consisting of reading the PCI_INTERRUPT_PIN to know wether
  the device has an interrupt at all, and the the PCI_INTERRUPT_LINE to get an
  interrupt number from the device.  That number is then mapped using the
  default controller, and the trigger is set to level low.  That default
  behaviour works for several platforms that don't have a proper interrupt
  tree like Pegasos.  If it doesn't work for your platform, then either
  provide a proper interrupt tree from the firmware so that fallback isn't
  needed, or don't call pci_read_irq_line()

- Add back a bit that got dropped by my main rework patch for properly
  clearing pending IPIs on pSeries when using a kexec
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

arch/powerpc/kernel/ibmebus.c

@@ -323,7 +323,7 @@ int ibmebus_request_irq(struct ibmebus_dev *dev,
 			unsigned long irq_flags, const char * devname,
 			void *dev_id)
 {
-	unsigned int irq = irq_create_mapping(NULL, ist, 0);
+	unsigned int irq = irq_create_mapping(NULL, ist);
 	
 	if (irq == NO_IRQ)
 		return -EINVAL;

arch/powerpc/kernel/irq.c

@@ -391,15 +391,14 @@ struct irq_host *irq_alloc_host(unsigned int revmap_type,
 			irq_map[i].host = host;
 			smp_wmb();
 
-			/* Clear some flags */
-			get_irq_desc(i)->status
-				&= ~(IRQ_NOREQUEST | IRQ_LEVEL);
+			/* Clear norequest flags */
+			get_irq_desc(i)->status &= ~IRQ_NOREQUEST;
 
 			/* Legacy flags are left to default at this point,
 			 * one can then use irq_create_mapping() to
 			 * explicitely change them
 			 */
-			ops->map(host, i, i, 0);
+			ops->map(host, i, i);
 		}
 		break;
 	case IRQ_HOST_MAP_LINEAR:
@@ -457,13 +456,11 @@ void irq_set_virq_count(unsigned int count)
 }
 
 unsigned int irq_create_mapping(struct irq_host *host,
-				irq_hw_number_t hwirq,
-				unsigned int flags)
+				irq_hw_number_t hwirq)
 {
 	unsigned int virq, hint;
 
-	pr_debug("irq: irq_create_mapping(0x%p, 0x%lx, 0x%x)\n",
-		 host, hwirq, flags);
+	pr_debug("irq: irq_create_mapping(0x%p, 0x%lx)\n", host, hwirq);
 
 	/* Look for default host if nececssary */
 	if (host == NULL)
@@ -482,7 +479,6 @@ unsigned int irq_create_mapping(struct irq_host *host,
 	virq = irq_find_mapping(host, hwirq);
 	if (virq != IRQ_NONE) {
 		pr_debug("irq: -> existing mapping on virq %d\n", virq);
-		host->ops->map(host, virq, hwirq, flags);
 		return virq;
 	}
 
@@ -504,18 +500,18 @@ unsigned int irq_create_mapping(struct irq_host *host,
 	}
 	pr_debug("irq: -> obtained virq %d\n", virq);
 
-	/* Clear some flags */
-	get_irq_desc(virq)->status &= ~(IRQ_NOREQUEST | IRQ_LEVEL);
+	/* Clear IRQ_NOREQUEST flag */
+	get_irq_desc(virq)->status &= ~IRQ_NOREQUEST;
 
 	/* map it */
-	if (host->ops->map(host, virq, hwirq, flags)) {
+	smp_wmb();
+	irq_map[virq].hwirq = hwirq;
+	smp_mb();
+	if (host->ops->map(host, virq, hwirq)) {
 		pr_debug("irq: -> mapping failed, freeing\n");
 		irq_free_virt(virq, 1);
 		return NO_IRQ;
 	}
-	smp_wmb();
-	irq_map[virq].hwirq = hwirq;
-	smp_mb();
 	return virq;
 }
 EXPORT_SYMBOL_GPL(irq_create_mapping);
@@ -525,25 +521,38 @@ extern unsigned int irq_create_of_mapping(struct device_node *controller,
 {
 	struct irq_host *host;
 	irq_hw_number_t hwirq;
-	unsigned int flags = IRQ_TYPE_NONE;
+	unsigned int type = IRQ_TYPE_NONE;
+	unsigned int virq;
 
 	if (controller == NULL)
 		host = irq_default_host;
 	else
 		host = irq_find_host(controller);
-	if (host == NULL)
+	if (host == NULL) {
+		printk(KERN_WARNING "irq: no irq host found for %s !\n",
+		       controller->full_name);
 		return NO_IRQ;
+	}
 
 	/* If host has no translation, then we assume interrupt line */
 	if (host->ops->xlate == NULL)
 		hwirq = intspec[0];
 	else {
 		if (host->ops->xlate(host, controller, intspec, intsize,
-				     &hwirq, &flags))
+				     &hwirq, &type))
 			return NO_IRQ;
 	}
 
-	return irq_create_mapping(host, hwirq, flags);
+	/* Create mapping */
+	virq = irq_create_mapping(host, hwirq);
+	if (virq == NO_IRQ)
+		return virq;
+
+	/* Set type if specified and different than the current one */
+	if (type != IRQ_TYPE_NONE &&
+	    type != (get_irq_desc(virq)->status & IRQF_TRIGGER_MASK))
+		set_irq_type(virq, type);
+	return virq;
 }
 EXPORT_SYMBOL_GPL(irq_create_of_mapping);
 

arch/powerpc/kernel/pci_32.c

@@ -11,6 +11,7 @@
 #include <linux/sched.h>
 #include <linux/errno.h>
 #include <linux/bootmem.h>
+#include <linux/irq.h>
 
 #include <asm/processor.h>
 #include <asm/io.h>
@@ -18,7 +19,6 @@
 #include <asm/sections.h>
 #include <asm/pci-bridge.h>
 #include <asm/byteorder.h>
-#include <asm/irq.h>
 #include <asm/uaccess.h>
 #include <asm/machdep.h>
 
@@ -1420,15 +1420,37 @@ int pci_read_irq_line(struct pci_dev *pci_dev)
 
 	DBG("Try to map irq for %s...\n", pci_name(pci_dev));
 
+	/* Try to get a mapping from the device-tree */
 	if (of_irq_map_pci(pci_dev, &oirq)) {
-		DBG(" -> failed !\n");
+		u8 line, pin;
+
+		/* If that fails, lets fallback to what is in the config
+		 * space and map that through the default controller. We
+		 * also set the type to level low since that's what PCI
+		 * interrupts are. If your platform does differently, then
+		 * either provide a proper interrupt tree or don't use this
+		 * function.
+		 */
+		if (pci_read_config_byte(pci_dev, PCI_INTERRUPT_PIN, &pin))
+			return -1;
+		if (pin == 0)
+			return -1;
+		if (pci_read_config_byte(pci_dev, PCI_INTERRUPT_LINE, &line) ||
+		    line == 0xff) {
 			return -1;
 		}
+		DBG(" -> no map ! Using irq line %d from PCI config\n", line);
 
+		virq = irq_create_mapping(NULL, line);
+		if (virq != NO_IRQ)
+			set_irq_type(virq, IRQ_TYPE_LEVEL_LOW);
+	} else {
 		DBG(" -> got one, spec %d cells (0x%08x...) on %s\n",
 		    oirq.size, oirq.specifier[0], oirq.controller->full_name);
 
-	virq = irq_create_of_mapping(oirq.controller, oirq.specifier, oirq.size);
+		virq = irq_create_of_mapping(oirq.controller, oirq.specifier,
+					     oirq.size);
+	}
 	if(virq == NO_IRQ) {
 		DBG(" -> failed to map !\n");
 		return -1;

arch/powerpc/kernel/pci_64.c

@@ -21,13 +21,13 @@
 #include <linux/mm.h>
 #include <linux/list.h>
 #include <linux/syscalls.h>
+#include <linux/irq.h>
 
 #include <asm/processor.h>
 #include <asm/io.h>
 #include <asm/prom.h>
 #include <asm/pci-bridge.h>
 #include <asm/byteorder.h>
-#include <asm/irq.h>
 #include <asm/machdep.h>
 #include <asm/ppc-pci.h>
 
@@ -1289,15 +1289,37 @@ int pci_read_irq_line(struct pci_dev *pci_dev)
 
 	DBG("Try to map irq for %s...\n", pci_name(pci_dev));
 
+	/* Try to get a mapping from the device-tree */
 	if (of_irq_map_pci(pci_dev, &oirq)) {
-		DBG(" -> failed !\n");
+		u8 line, pin;
+
+		/* If that fails, lets fallback to what is in the config
+		 * space and map that through the default controller. We
+		 * also set the type to level low since that's what PCI
+		 * interrupts are. If your platform does differently, then
+		 * either provide a proper interrupt tree or don't use this
+		 * function.
+		 */
+		if (pci_read_config_byte(pci_dev, PCI_INTERRUPT_PIN, &pin))
+			return -1;
+		if (pin == 0)
+			return -1;
+		if (pci_read_config_byte(pci_dev, PCI_INTERRUPT_LINE, &line) ||
+		    line == 0xff) {
 			return -1;
 		}
+		DBG(" -> no map ! Using irq line %d from PCI config\n", line);
 
+		virq = irq_create_mapping(NULL, line);
+		if (virq != NO_IRQ)
+			set_irq_type(virq, IRQ_TYPE_LEVEL_LOW);
+	} else {
 		DBG(" -> got one, spec %d cells (0x%08x...) on %s\n",
 		    oirq.size, oirq.specifier[0], oirq.controller->full_name);
 
-	virq = irq_create_of_mapping(oirq.controller, oirq.specifier, oirq.size);
+		virq = irq_create_of_mapping(oirq.controller, oirq.specifier,
+					     oirq.size);
+	}
 	if(virq == NO_IRQ) {
 		DBG(" -> failed to map !\n");
 		return -1;

arch/powerpc/platforms/cell/interrupt.c

@@ -159,7 +159,7 @@ static void iic_request_ipi(int ipi, const char *name)
 		if (iic_hosts[node] == NULL)
 			continue;
 		virq = irq_create_mapping(iic_hosts[node],
-					  iic_ipi_to_irq(ipi), 0);
+					  iic_ipi_to_irq(ipi));
 		if (virq == NO_IRQ) {
 			printk(KERN_ERR
 			       "iic: failed to map IPI %s on node %d\n",
@@ -197,7 +197,7 @@ static int iic_host_match(struct irq_host *h, struct device_node *node)
 }
 
 static int iic_host_map(struct irq_host *h, unsigned int virq,
-			irq_hw_number_t hw, unsigned int flags)
+			irq_hw_number_t hw)
 {
 	if (hw < IIC_IRQ_IPI0)
 		set_irq_chip_and_handler(virq, &iic_chip, handle_fasteoi_irq);

arch/powerpc/platforms/cell/spider-pic.c

@@ -85,9 +85,6 @@ static void spider_unmask_irq(unsigned int virq)
 	struct spider_pic *pic = spider_virq_to_pic(virq);
 	void __iomem *cfg = spider_get_irq_config(pic, irq_map[virq].hwirq);
 
-	/* We use no locking as we should be covered by the descriptor lock
-	 * for access to invidual source configuration registers
-	 */
 	out_be32(cfg, in_be32(cfg) | 0x30000000u);
 }
 
@@ -96,9 +93,6 @@ static void spider_mask_irq(unsigned int virq)
 	struct spider_pic *pic = spider_virq_to_pic(virq);
 	void __iomem *cfg = spider_get_irq_config(pic, irq_map[virq].hwirq);
 
-	/* We use no locking as we should be covered by the descriptor lock
-	 * for access to invidual source configuration registers
-	 */
 	out_be32(cfg, in_be32(cfg) & ~0x30000000u);
 }
 
@@ -120,26 +114,14 @@ static void spider_ack_irq(unsigned int virq)
 	out_be32(pic->regs + TIR_EDC, 0x100 | (src & 0xf));
 }
 
-static struct irq_chip spider_pic = {
-	.typename = " SPIDER   ",
-	.unmask = spider_unmask_irq,
-	.mask = spider_mask_irq,
-	.ack = spider_ack_irq,
-};
-
-static int spider_host_match(struct irq_host *h, struct device_node *node)
-{
-	struct spider_pic *pic = h->host_data;
-	return node == pic->of_node;
-}
-
-static int spider_host_map(struct irq_host *h, unsigned int virq,
-			irq_hw_number_t hw, unsigned int flags)
+static int spider_set_irq_type(unsigned int virq, unsigned int type)
 {
-	unsigned int sense = flags & IRQ_TYPE_SENSE_MASK;
-	struct spider_pic *pic = h->host_data;
+	unsigned int sense = type & IRQ_TYPE_SENSE_MASK;
+	struct spider_pic *pic = spider_virq_to_pic(virq);
+	unsigned int hw = irq_map[virq].hwirq;
 	void __iomem *cfg = spider_get_irq_config(pic, hw);
-	int level = 0;
+	struct irq_desc *desc = get_irq_desc(virq);
+	u32 old_mask;
 	u32 ic;
 
 	/* Note that only level high is supported for most interrupts */
@@ -157,29 +139,57 @@ static int spider_host_map(struct irq_host *h, unsigned int virq,
 		break;
 	case IRQ_TYPE_LEVEL_LOW:
 		ic = 0x0;
-		level = 1;
 		break;
 	case IRQ_TYPE_LEVEL_HIGH:
 	case IRQ_TYPE_NONE:
 		ic = 0x1;
-		level = 1;
 		break;
 	default:
 		return -EINVAL;
 	}
 
+	/* Update irq_desc */
+	desc->status &= ~(IRQ_TYPE_SENSE_MASK | IRQ_LEVEL);
+	desc->status |= type & IRQ_TYPE_SENSE_MASK;
+	if (type & (IRQ_TYPE_LEVEL_HIGH | IRQ_TYPE_LEVEL_LOW))
+		desc->status |= IRQ_LEVEL;
+
 	/* Configure the source. One gross hack that was there before and
 	 * that I've kept around is the priority to the BE which I set to
 	 * be the same as the interrupt source number. I don't know wether
 	 * that's supposed to make any kind of sense however, we'll have to
 	 * decide that, but for now, I'm not changing the behaviour.
 	 */
-	out_be32(cfg, (ic << 24) | (0x7 << 16) | (pic->node_id << 4) | 0xe);
+	old_mask = in_be32(cfg) & 0x30000000u;
+	out_be32(cfg, old_mask | (ic << 24) | (0x7 << 16) |
+		 (pic->node_id << 4) | 0xe);
 	out_be32(cfg + 4, (0x2 << 16) | (hw & 0xff));
 
-	if (level)
-		get_irq_desc(virq)->status |= IRQ_LEVEL;
+	return 0;
+}
+
+static struct irq_chip spider_pic = {
+	.typename = " SPIDER   ",
+	.unmask = spider_unmask_irq,
+	.mask = spider_mask_irq,
+	.ack = spider_ack_irq,
+	.set_type = spider_set_irq_type,
+};
+
+static int spider_host_match(struct irq_host *h, struct device_node *node)
+{
+	struct spider_pic *pic = h->host_data;
+	return node == pic->of_node;
+}
+
+static int spider_host_map(struct irq_host *h, unsigned int virq,
+			irq_hw_number_t hw)
+{
 	set_irq_chip_and_handler(virq, &spider_pic, handle_level_irq);
+
+	/* Set default irq type */
+	set_irq_type(virq, IRQ_TYPE_NONE);
+
 	return 0;
 }
 
@@ -283,7 +293,7 @@ static unsigned int __init spider_find_cascade_and_node(struct spider_pic *pic)
 	if (iic_host == NULL)
 		return NO_IRQ;
 	/* Manufacture an IIC interrupt number of class 2 */
-	virq = irq_create_mapping(iic_host, 0x20 | unit, 0);
+	virq = irq_create_mapping(iic_host, 0x20 | unit);
 	if (virq == NO_IRQ)
 		printk(KERN_ERR "spider_pic: failed to map cascade !");
 	return virq;

arch/powerpc/platforms/cell/spu_base.c

@@ -583,9 +583,9 @@ static int __init spu_map_interrupts(struct spu *spu, struct device_node *np)
 	spu->isrc = isrc = tmp[0];
 
 	/* Now map interrupts of all 3 classes */
-	spu->irqs[0] = irq_create_mapping(host, 0x00 | isrc, 0);
-	spu->irqs[1] = irq_create_mapping(host, 0x10 | isrc, 0);
-	spu->irqs[2] = irq_create_mapping(host, 0x20 | isrc, 0);
+	spu->irqs[0] = irq_create_mapping(host, 0x00 | isrc);
+	spu->irqs[1] = irq_create_mapping(host, 0x10 | isrc);
+	spu->irqs[2] = irq_create_mapping(host, 0x20 | isrc);
 
 	/* Right now, we only fail if class 2 failed */
 	return spu->irqs[2] == NO_IRQ ? -EINVAL : 0;

arch/powerpc/platforms/iseries/irq.c

@@ -300,7 +300,7 @@ int __init iSeries_allocate_IRQ(HvBusNumber bus,
 	realirq = (((((sub_bus << 8) + (bus - 1)) << 3) + (idsel - 1)) << 3)
 		+ function;
 
-	return irq_create_mapping(NULL, realirq, IRQ_TYPE_NONE);
+	return irq_create_mapping(NULL, realirq);
 }
 
 #endif /* CONFIG_PCI */
@@ -341,7 +341,7 @@ unsigned int iSeries_get_irq(struct pt_regs *regs)
 }
 
 static int iseries_irq_host_map(struct irq_host *h, unsigned int virq,
-				irq_hw_number_t hw, unsigned int flags)
+				irq_hw_number_t hw)
 {
 	set_irq_chip_and_handler(virq, &iseries_pic, handle_fasteoi_irq);
 

arch/powerpc/platforms/powermac/pci.c

@@ -16,6 +16,7 @@
 #include <linux/string.h>
 #include <linux/init.h>
 #include <linux/bootmem.h>
+#include <linux/irq.h>
 
 #include <asm/sections.h>
 #include <asm/io.h>
@@ -24,10 +25,7 @@
 #include <asm/machdep.h>
 #include <asm/pmac_feature.h>
 #include <asm/grackle.h>
-#ifdef CONFIG_PPC64
-//#include <asm/iommu.h>
 #include <asm/ppc-pci.h>
-#endif
 
 #undef DEBUG
 
@@ -46,7 +44,6 @@ static int has_uninorth;
 static struct pci_controller *u3_agp;
 static struct pci_controller *u4_pcie;
 static struct pci_controller *u3_ht;
-#define has_second_ohare 0
 #else
 static int has_second_ohare;
 #endif /* CONFIG_PPC64 */
@@ -993,6 +990,7 @@ void __init pmac_pcibios_fixup(void)
 		/* Read interrupt from the device-tree */
 		pci_read_irq_line(dev);
 
+#ifdef CONFIG_PPC32
 		/* Fixup interrupt for the modem/ethernet combo controller.
 		 * on machines with a second ohare chip.
 		 * The number in the device tree (27) is bogus (correct for
@@ -1002,8 +1000,11 @@ void __init pmac_pcibios_fixup(void)
 		 */
 		if (has_second_ohare &&
 		    dev->vendor == PCI_VENDOR_ID_DEC &&
-		    dev->device == PCI_DEVICE_ID_DEC_TULIP_PLUS)
-			dev->irq = irq_create_mapping(NULL, 60, 0);
+		    dev->device == PCI_DEVICE_ID_DEC_TULIP_PLUS) {
+			dev->irq = irq_create_mapping(NULL, 60);
+			set_irq_type(dev->irq, IRQ_TYPE_LEVEL_LOW);
+		}
+#endif /* CONFIG_PPC32 */
 	}
 }
 

arch/powerpc/platforms/powermac/pic.c

@@ -291,7 +291,7 @@ static int pmac_pic_host_match(struct irq_host *h, struct device_node *node)
 }
 
 static int pmac_pic_host_map(struct irq_host *h, unsigned int virq,
-			     irq_hw_number_t hw, unsigned int flags)
+			     irq_hw_number_t hw)
 {
 	struct irq_desc *desc = get_irq_desc(virq);
 	int level;
@@ -318,6 +318,7 @@ static int pmac_pic_host_xlate(struct irq_host *h, struct device_node *ct,
 			       unsigned int *out_flags)
 
 {
+	*out_flags = IRQ_TYPE_NONE;
 	*out_hwirq = *intspec;
 	return 0;
 }
@@ -434,7 +435,7 @@ static void __init pmac_pic_probe_oldstyle(void)
 
 	printk(KERN_INFO "irq: System has %d possible interrupts\n", max_irqs);
 #ifdef CONFIG_XMON
-	setup_irq(irq_create_mapping(NULL, 20, 0), &xmon_action);
+	setup_irq(irq_create_mapping(NULL, 20), &xmon_action);
 #endif
 }
 #endif /* CONFIG_PPC32 */
@@ -579,9 +580,10 @@ void __init pmac_pic_init(void)
 		flags |= OF_IMAP_OLDWORLD_MAC;
 	if (get_property(of_chosen, "linux,bootx", NULL) != NULL)
 		flags |= OF_IMAP_NO_PHANDLE;
-	of_irq_map_init(flags);
 #endif /* CONFIG_PPC_32 */
 
+	of_irq_map_init(flags);
+
 	/* We first try to detect Apple's new Core99 chipset, since mac-io
 	 * is quite different on those machines and contains an IBM MPIC2.
 	 */

arch/powerpc/platforms/pseries/ras.c

@@ -93,8 +93,7 @@ static void request_ras_irqs(struct device_node *np,
 		for (i = 0; i < opicplen; i++) {
 			if (count > 15)
 				break;
-			virqs[count] = irq_create_mapping(NULL, *(opicprop++),
-							 IRQ_TYPE_NONE);
+			virqs[count] = irq_create_mapping(NULL, *(opicprop++));
 			if (virqs[count] == NO_IRQ)
 				printk(KERN_ERR "Unable to allocate interrupt "
 				       "number for %s\n", np->full_name);

arch/powerpc/platforms/pseries/xics.c

@@ -502,16 +502,9 @@ static int xics_host_match(struct irq_host *h, struct device_node *node)
 }
 
 static int xics_host_map_direct(struct irq_host *h, unsigned int virq,
-				irq_hw_number_t hw, unsigned int flags)
+				irq_hw_number_t hw)
 {
-	unsigned int sense = flags & IRQ_TYPE_SENSE_MASK;
-
-	pr_debug("xics: map_direct virq %d, hwirq 0x%lx, flags: 0x%x\n",
-		 virq, hw, flags);
-
-	if (sense && sense != IRQ_TYPE_LEVEL_LOW)
-		printk(KERN_WARNING "xics: using unsupported sense 0x%x"
-		       " for irq %d (h: 0x%lx)\n", flags, virq, hw);
+	pr_debug("xics: map_direct virq %d, hwirq 0x%lx\n", virq, hw);
 
 	get_irq_desc(virq)->status |= IRQ_LEVEL;
 	set_irq_chip_and_handler(virq, &xics_pic_direct, handle_fasteoi_irq);
@@ -519,16 +512,9 @@ static int xics_host_map_direct(struct irq_host *h, unsigned int virq,
 }
 
 static int xics_host_map_lpar(struct irq_host *h, unsigned int virq,
-			      irq_hw_number_t hw, unsigned int flags)
+			      irq_hw_number_t hw)
 {
-	unsigned int sense = flags & IRQ_TYPE_SENSE_MASK;
-
-	pr_debug("xics: map_lpar virq %d, hwirq 0x%lx, flags: 0x%x\n",
-		 virq, hw, flags);
-
-	if (sense && sense != IRQ_TYPE_LEVEL_LOW)
-		printk(KERN_WARNING "xics: using unsupported sense 0x%x"
-		       " for irq %d (h: 0x%lx)\n", flags, virq, hw);
+	pr_debug("xics: map_direct virq %d, hwirq 0x%lx\n", virq, hw);
 
 	get_irq_desc(virq)->status |= IRQ_LEVEL;
 	set_irq_chip_and_handler(virq, &xics_pic_lpar, handle_fasteoi_irq);
@@ -757,7 +743,7 @@ void xics_request_IPIs(void)
 {
 	unsigned int ipi;
 
-	ipi = irq_create_mapping(xics_host, XICS_IPI, 0);
+	ipi = irq_create_mapping(xics_host, XICS_IPI);
 	BUG_ON(ipi == NO_IRQ);
 
 	/*
@@ -782,6 +768,14 @@ void xics_teardown_cpu(int secondary)
 
 	xics_set_cpu_priority(cpu, 0);
 
+	/*
+	 * Clear IPI
+	 */
+	if (firmware_has_feature(FW_FEATURE_LPAR))
+		lpar_qirr_info(cpu, 0xff);
+	else
+		direct_qirr_info(cpu, 0xff);
+
 	/*
 	 * we need to EOI the IPI if we got here from kexec down IPI
 	 *
@@ -795,7 +789,7 @@ void xics_teardown_cpu(int secondary)
 		return;
 	desc = get_irq_desc(ipi);
 	if (desc->chip && desc->chip->eoi)
-		desc->chip->eoi(XICS_IPI);
+		desc->chip->eoi(ipi);
 
 	/*
 	 * Some machines need to have at least one cpu in the GIQ,

arch/powerpc/sysdev/i8259.c

@@ -169,7 +169,7 @@ static int i8259_host_match(struct irq_host *h, struct device_node *node)
 }
 
 static int i8259_host_map(struct irq_host *h, unsigned int virq,
-			  irq_hw_number_t hw, unsigned int flags)
+			  irq_hw_number_t hw)
 {
 	pr_debug("i8259_host_map(%d, 0x%lx)\n", virq, hw);
 
@@ -177,7 +177,7 @@ static int i8259_host_map(struct irq_host *h, unsigned int virq,
 	if (hw == 2)
 		get_irq_desc(virq)->status |= IRQ_NOREQUEST;
 
-	/* We use the level stuff only for now, we might want to
+	/* We use the level handler only for now, we might want to
 	 * be more cautious here but that works for now
 	 */
 	get_irq_desc(virq)->status |= IRQ_LEVEL;

arch/powerpc/sysdev/mpic.c

@@ -337,6 +337,17 @@ static void __init mpic_scan_ht_pics(struct mpic *mpic)
 	}
 }
 
+#else /* CONFIG_MPIC_BROKEN_U3 */
+
+static inline int mpic_is_ht_interrupt(struct mpic *mpic, unsigned int source)
+{
+	return 0;
+}
+
+static void __init mpic_scan_ht_pics(struct mpic *mpic)
+{
+}
+
 #endif /* CONFIG_MPIC_BROKEN_U3 */
 
 
@@ -405,11 +416,9 @@ static void mpic_unmask_irq(unsigned int irq)
 	unsigned int loops = 100000;
 	struct mpic *mpic = mpic_from_irq(irq);
 	unsigned int src = mpic_irq_to_hw(irq);
-	unsigned long flags;
 
 	DBG("%p: %s: enable_irq: %d (src %d)\n", mpic, mpic->name, irq, src);
 
-	spin_lock_irqsave(&mpic_lock, flags);
 	mpic_irq_write(src, MPIC_IRQ_VECTOR_PRI,
 		       mpic_irq_read(src, MPIC_IRQ_VECTOR_PRI) &
 		       ~MPIC_VECPRI_MASK);
@@ -420,7 +429,6 @@ static void mpic_unmask_irq(unsigned int irq)
 			break;
 		}
 	} while(mpic_irq_read(src, MPIC_IRQ_VECTOR_PRI) & MPIC_VECPRI_MASK);
-	spin_unlock_irqrestore(&mpic_lock, flags);
 }
 
 static void mpic_mask_irq(unsigned int irq)
@@ -428,11 +436,9 @@ static void mpic_mask_irq(unsigned int irq)
 	unsigned int loops = 100000;
 	struct mpic *mpic = mpic_from_irq(irq);
 	unsigned int src = mpic_irq_to_hw(irq);
-	unsigned long flags;
 
 	DBG("%s: disable_irq: %d (src %d)\n", mpic->name, irq, src);
 
-	spin_lock_irqsave(&mpic_lock, flags);
 	mpic_irq_write(src, MPIC_IRQ_VECTOR_PRI,
 		       mpic_irq_read(src, MPIC_IRQ_VECTOR_PRI) |
 		       MPIC_VECPRI_MASK);
@@ -444,7 +450,6 @@ static void mpic_mask_irq(unsigned int irq)
 			break;
 		}
 	} while(!(mpic_irq_read(src, MPIC_IRQ_VECTOR_PRI) & MPIC_VECPRI_MASK));
-	spin_unlock_irqrestore(&mpic_lock, flags);
 }
 
 static void mpic_end_irq(unsigned int irq)
@@ -512,8 +517,7 @@ static void mpic_end_ht_irq(unsigned int irq)
 		mpic_ht_end_irq(mpic, src);
 	mpic_eoi(mpic);
 }
-
-#endif /* CONFIG_MPIC_BROKEN_U3 */
+#endif /* !CONFIG_MPIC_BROKEN_U3 */
 
 #ifdef CONFIG_SMP
 
@@ -560,40 +564,67 @@ static void mpic_set_affinity(unsigned int irq, cpumask_t cpumask)
 		       mpic_physmask(cpus_addr(tmp)[0]));	
 }
 
-static unsigned int mpic_flags_to_vecpri(unsigned int flags, int *level)
+static unsigned int mpic_type_to_vecpri(unsigned int type)
 {
-	unsigned int vecpri;
-
 	/* Now convert sense value */
-	switch(flags & IRQ_TYPE_SENSE_MASK) {
+	switch(type & IRQ_TYPE_SENSE_MASK) {
 	case IRQ_TYPE_EDGE_RISING:
-		vecpri = MPIC_VECPRI_SENSE_EDGE |
-			MPIC_VECPRI_POLARITY_POSITIVE;
-		*level = 0;
-		break;
+		return MPIC_VECPRI_SENSE_EDGE | MPIC_VECPRI_POLARITY_POSITIVE;
 	case IRQ_TYPE_EDGE_FALLING:
-		vecpri = MPIC_VECPRI_SENSE_EDGE |
-			MPIC_VECPRI_POLARITY_NEGATIVE;
-		*level = 0;
-		break;
+	case IRQ_TYPE_EDGE_BOTH:
+		return MPIC_VECPRI_SENSE_EDGE | MPIC_VECPRI_POLARITY_NEGATIVE;
 	case IRQ_TYPE_LEVEL_HIGH:
-		vecpri = MPIC_VECPRI_SENSE_LEVEL |
-			MPIC_VECPRI_POLARITY_POSITIVE;
-		*level = 1;
-		break;
+		return MPIC_VECPRI_SENSE_LEVEL | MPIC_VECPRI_POLARITY_POSITIVE;
 	case IRQ_TYPE_LEVEL_LOW:
 	default:
-		vecpri = MPIC_VECPRI_SENSE_LEVEL |
-			MPIC_VECPRI_POLARITY_NEGATIVE;
-		*level = 1;
+		return MPIC_VECPRI_SENSE_LEVEL | MPIC_VECPRI_POLARITY_NEGATIVE;
 	}
-	return vecpri;
+}
+
+static int mpic_set_irq_type(unsigned int virq, unsigned int flow_type)
+{
+	struct mpic *mpic = mpic_from_irq(virq);
+	unsigned int src = mpic_irq_to_hw(virq);
+	struct irq_desc *desc = get_irq_desc(virq);
+	unsigned int vecpri, vold, vnew;
+
+	pr_debug("mpic: set_irq_type(mpic:@%p,virq:%d,src:%d,type:0x%x)\n",
+		 mpic, virq, src, flow_type);
+
+	if (src >= mpic->irq_count)
+		return -EINVAL;
+
+	if (flow_type == IRQ_TYPE_NONE)
+		if (mpic->senses && src < mpic->senses_count)
+			flow_type = mpic->senses[src];
+	if (flow_type == IRQ_TYPE_NONE)
+		flow_type = IRQ_TYPE_LEVEL_LOW;
+
+	desc->status &= ~(IRQ_TYPE_SENSE_MASK | IRQ_LEVEL);
+	desc->status |= flow_type & IRQ_TYPE_SENSE_MASK;
+	if (flow_type & (IRQ_TYPE_LEVEL_HIGH | IRQ_TYPE_LEVEL_LOW))
+		desc->status |= IRQ_LEVEL;
+
+	if (mpic_is_ht_interrupt(mpic, src))
+		vecpri = MPIC_VECPRI_POLARITY_POSITIVE |
+			MPIC_VECPRI_SENSE_EDGE;
+	else
+		vecpri = mpic_type_to_vecpri(flow_type);
+
+	vold = mpic_irq_read(src, MPIC_IRQ_VECTOR_PRI);
+	vnew = vold & ~(MPIC_VECPRI_POLARITY_MASK | MPIC_VECPRI_SENSE_MASK);
+	vnew |= vecpri;
+	if (vold != vnew)
+		mpic_irq_write(src, MPIC_IRQ_VECTOR_PRI, vnew);
+
+	return 0;
 }
 
 static struct irq_chip mpic_irq_chip = {
 	.mask		= mpic_mask_irq,
 	.unmask		= mpic_unmask_irq,
 	.eoi		= mpic_end_irq,
+	.set_type	= mpic_set_irq_type,
 };
 
 #ifdef CONFIG_SMP
@@ -611,6 +642,7 @@ static struct irq_chip mpic_irq_ht_chip = {
 	.mask		= mpic_mask_irq,
 	.unmask		= mpic_unmask_ht_irq,
 	.eoi		= mpic_end_ht_irq,
+	.set_type	= mpic_set_irq_type,
 };
 #endif /* CONFIG_MPIC_BROKEN_U3 */
 
@@ -624,18 +656,12 @@ static int mpic_host_match(struct irq_host *h, struct device_node *node)
 }
 
 static int mpic_host_map(struct irq_host *h, unsigned int virq,
-			 irq_hw_number_t hw, unsigned int flags)
+			 irq_hw_number_t hw)
 {
-	struct irq_desc *desc = get_irq_desc(virq);
-	struct irq_chip *chip;
 	struct mpic *mpic = h->host_data;
-	u32 v, vecpri = MPIC_VECPRI_SENSE_LEVEL |
-		MPIC_VECPRI_POLARITY_NEGATIVE;
-	int level;
-	unsigned long iflags;
+	struct irq_chip *chip;
 
-	pr_debug("mpic: map virq %d, hwirq 0x%lx, flags: 0x%x\n",
-		 virq, hw, flags);
+	pr_debug("mpic: map virq %d, hwirq 0x%lx\n", virq, hw);
 
 	if (hw == MPIC_VEC_SPURRIOUS)
 		return -EINVAL;
@@ -654,44 +680,23 @@ static int mpic_host_map(struct irq_host *h, unsigned int virq,
 	if (hw >= mpic->irq_count)
 		return -EINVAL;
 
-	/* If no sense provided, check default sense array */
-	if (((flags & IRQ_TYPE_SENSE_MASK) == IRQ_TYPE_NONE) &&
-	    mpic->senses && hw < mpic->senses_count)
-		flags |= mpic->senses[hw];
-
-	vecpri = mpic_flags_to_vecpri(flags, &level);
-	if (level)
-		desc->status |= IRQ_LEVEL;
+	/* Default chip */
 	chip = &mpic->hc_irq;
 
 #ifdef CONFIG_MPIC_BROKEN_U3
 	/* Check for HT interrupts, override vecpri */
-	if (mpic_is_ht_interrupt(mpic, hw)) {
-		vecpri &= ~(MPIC_VECPRI_SENSE_MASK |
-			    MPIC_VECPRI_POLARITY_MASK);
-		vecpri |= MPIC_VECPRI_POLARITY_POSITIVE;
+	if (mpic_is_ht_interrupt(mpic, hw))
 		chip = &mpic->hc_ht_irq;
-	}
-#endif
+#endif /* CONFIG_MPIC_BROKEN_U3 */
 
-	/* Reconfigure irq. We must preserve the mask bit as we can be called
-	 * while the interrupt is still active (This may change in the future
-	 * but for now, it is the case).
-	 */
-	spin_lock_irqsave(&mpic_lock, iflags);
-	v = mpic_irq_read(hw, MPIC_IRQ_VECTOR_PRI);
-	vecpri = (v &
-		~(MPIC_VECPRI_POLARITY_MASK | MPIC_VECPRI_SENSE_MASK)) |
-		vecpri;
-	if (vecpri != v)
-		mpic_irq_write(hw, MPIC_IRQ_VECTOR_PRI, vecpri);
-	spin_unlock_irqrestore(&mpic_lock, iflags);
-
-	pr_debug("mpic: mapping as IRQ, vecpri = 0x%08x (was 0x%08x)\n",
-		 vecpri, v);
+	pr_debug("mpic: mapping to irq chip @%p\n", chip);
 
 	set_irq_chip_data(virq, mpic);
 	set_irq_chip_and_handler(virq, chip, handle_fasteoi_irq);
+
+	/* Set default irq type */
+	set_irq_type(virq, IRQ_TYPE_NONE);
+
 	return 0;
 }
 
@@ -906,40 +911,15 @@ void __init mpic_init(struct mpic *mpic)
 	if (mpic->irq_count == 0)
 		mpic->irq_count = mpic->num_sources;
 
-#ifdef CONFIG_MPIC_BROKEN_U3
 	/* Do the HT PIC fixups on U3 broken mpic */
 	DBG("MPIC flags: %x\n", mpic->flags);
 	if ((mpic->flags & MPIC_BROKEN_U3) && (mpic->flags & MPIC_PRIMARY))
  		mpic_scan_ht_pics(mpic);
-#endif /* CONFIG_MPIC_BROKEN_U3 */
 
 	for (i = 0; i < mpic->num_sources; i++) {
 		/* start with vector = source number, and masked */
-		u32 vecpri = MPIC_VECPRI_MASK | i | (8 << MPIC_VECPRI_PRIORITY_SHIFT);
-		int level = 1;
-		
-		/* do senses munging */
-		if (mpic->senses && i < mpic->senses_count)
-			vecpri |= mpic_flags_to_vecpri(mpic->senses[i],
-						       &level);
-		else
-			vecpri |= MPIC_VECPRI_SENSE_LEVEL;
-
-		/* deal with broken U3 */
-		if (mpic->flags & MPIC_BROKEN_U3) {
-#ifdef CONFIG_MPIC_BROKEN_U3
-			if (mpic_is_ht_interrupt(mpic, i)) {
-				vecpri &= ~(MPIC_VECPRI_SENSE_MASK |
-					    MPIC_VECPRI_POLARITY_MASK);
-				vecpri |= MPIC_VECPRI_POLARITY_POSITIVE;
-			}
-#else
-			printk(KERN_ERR "mpic: BROKEN_U3 set, but CONFIG doesn't match\n");
-#endif
-		}
-
-		DBG("setup source %d, vecpri: %08x, level: %d\n", i, vecpri,
-		    (level != 0));
+		u32 vecpri = MPIC_VECPRI_MASK | i |
+			(8 << MPIC_VECPRI_PRIORITY_SHIFT);
 		
 		/* init hw */
 		mpic_irq_write(i, MPIC_IRQ_VECTOR_PRI, vecpri);
@@ -1154,7 +1134,7 @@ void mpic_request_ipis(void)
 
 	for (i = 0; i < 4; i++) {
 		unsigned int vipi = irq_create_mapping(mpic->irqhost,
-						       MPIC_VEC_IPI_0 + i, 0);
+						       MPIC_VEC_IPI_0 + i);
 		if (vipi == NO_IRQ) {
 			printk(KERN_ERR "Failed to map IPI %d\n", i);
 			break;

drivers/char/hvsi.c

@@ -1299,7 +1299,7 @@ static int __init hvsi_console_init(void)
 		hp->inbuf_end = hp->inbuf;
 		hp->state = HVSI_CLOSED;
 		hp->vtermno = *vtermno;
-		hp->virq = irq_create_mapping(NULL, irq[0], 0);
+		hp->virq = irq_create_mapping(NULL, irq[0]);
 		if (hp->virq == NO_IRQ) {
 			printk(KERN_ERR "%s: couldn't create irq mapping for 0x%x\n",
 				__FUNCTION__, irq[0]);

drivers/macintosh/macio_asic.c

@@ -330,7 +330,7 @@ static void macio_create_fixup_irq(struct macio_dev *dev, int index,
 {
 	unsigned int irq;
 
-	irq = irq_create_mapping(NULL, line, 0);
+	irq = irq_create_mapping(NULL, line);
 	if (irq != NO_IRQ) {
 		dev->interrupt[index].start = irq;
 		dev->interrupt[index].flags = IORESOURCE_IRQ;

include/asm-powerpc/irq.h

@@ -83,25 +83,24 @@ struct irq_host_ops {
 	int (*match)(struct irq_host *h, struct device_node *node);
 
 	/* Create or update a mapping between a virtual irq number and a hw
-	 * irq number. This can be called several times for the same mapping
-	 * but with different flags, though unmap shall always be called
-	 * before the virq->hw mapping is changed.
+	 * irq number. This is called only once for a given mapping.
 	 */
-	int (*map)(struct irq_host *h, unsigned int virq,
-		   irq_hw_number_t hw, unsigned int flags);
+	int (*map)(struct irq_host *h, unsigned int virq, irq_hw_number_t hw);
 
 	/* Dispose of such a mapping */
 	void (*unmap)(struct irq_host *h, unsigned int virq);
 
 	/* Translate device-tree interrupt specifier from raw format coming
 	 * from the firmware to a irq_hw_number_t (interrupt line number) and
-	 * trigger flags that can be passed to irq_create_mapping().
-	 * If no translation is provided, raw format is assumed to be one cell
-	 * for interrupt line and default sense.
+	 * type (sense) that can be passed to set_irq_type(). In the absence
+	 * of this callback, irq_create_of_mapping() and irq_of_parse_and_map()
+	 * will return the hw number in the first cell and IRQ_TYPE_NONE for
+	 * the type (which amount to keeping whatever default value the
+	 * interrupt controller has for that line)
 	 */
 	int (*xlate)(struct irq_host *h, struct device_node *ctrler,
 		     u32 *intspec, unsigned int intsize,
-		     irq_hw_number_t *out_hwirq, unsigned int *out_flags);
+		     irq_hw_number_t *out_hwirq, unsigned int *out_type);
 };
 
 struct irq_host {
@@ -193,25 +192,14 @@ extern void irq_set_virq_count(unsigned int count);
  * irq_create_mapping - Map a hardware interrupt into linux virq space
  * @host: host owning this hardware interrupt or NULL for default host
  * @hwirq: hardware irq number in that host space
- * @flags: flags passed to the controller. contains the trigger type among
- *         others. Use IRQ_TYPE_* defined in include/linux/irq.h
  *
  * Only one mapping per hardware interrupt is permitted. Returns a linux
- * virq number. The flags can be used to provide sense information to the
- * controller (typically extracted from the device-tree). If no information
- * is passed, the controller defaults will apply (for example, xics can only
- * do edge so flags are irrelevant for some pseries specific irqs).
- *
- * The device-tree generally contains the trigger info in an encoding that is
- * specific to a given type of controller. In that case, you can directly use
- * host->ops->trigger_xlate() to translate that.
- *
- * It is recommended that new PICs that don't have existing OF bindings chose
- * to use a representation of triggers identical to linux.
+ * virq number.
+ * If the sense/trigger is to be specified, set_irq_type() should be called
+ * on the number returned from that call.
  */
 extern unsigned int irq_create_mapping(struct irq_host *host,
-				       irq_hw_number_t hwirq,
-				       unsigned int flags);
+				       irq_hw_number_t hwirq);
 
 
 /***
@@ -295,7 +283,7 @@ extern void irq_free_virt(unsigned int virq, unsigned int count);
  *
  * This function is identical to irq_create_mapping except that it takes
  * as input informations straight from the device-tree (typically the results
- * of the of_irq_map_*() functions
+ * of the of_irq_map_*() functions.
  */
 extern unsigned int irq_create_of_mapping(struct device_node *controller,
 					  u32 *intspec, unsigned int intsize);