[POWERPC] Add new interrupt mapping core and change platforms to use it

This adds the new irq remapper core and removes the old one.  Because
there are some fundamental conflicts with the old code, like the value
of NO_IRQ which I'm now setting to 0 (as per discussions with Linus),
etc..., this commit also changes the relevant platform and driver code
over to use the new remapper (so as not to cause difficulties later
in bisecting).

This patch removes the old pre-parsing of the open firmware interrupt
tree along with all the bogus assumptions it made to try to renumber
interrupts according to the platform. This is all to be handled by the
new code now.

For the pSeries XICS interrupt controller, a single remapper host is
created for the whole machine regardless of how many interrupt
presentation and source controllers are found, and it's set to match
any device node that isn't a 8259.  That works fine on pSeries and
avoids having to deal with some of the complexities of split source
controllers vs. presentation controllers in the pSeries device trees.

The powerpc i8259 PIC driver now always requests the legacy interrupt
range. It also has the feature of being able to match any device node
(including NULL) if passed no device node as an input. That will help
porting over platforms with broken device-trees like Pegasos who don't
have a proper interrupt tree.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

arch/powerpc/kernel/ibmebus.c

@@ -323,13 +323,11 @@ int ibmebus_request_irq(struct ibmebus_dev *dev,
 			unsigned long irq_flags, const char * devname,
 			void *dev_id)
 {
-	unsigned int irq = virt_irq_create_mapping(ist);
+	unsigned int irq = irq_create_mapping(NULL, ist, 0);
 	
 	if (irq == NO_IRQ)
 		return -EINVAL;
 	
-	irq = irq_offset_up(irq);
-	
 	return request_irq(irq, handler,
 			   irq_flags, devname, dev_id);
 }
@@ -337,12 +335,9 @@ EXPORT_SYMBOL(ibmebus_request_irq);
 
 void ibmebus_free_irq(struct ibmebus_dev *dev, u32 ist, void *dev_id)
 {
-	unsigned int irq = virt_irq_create_mapping(ist);
+	unsigned int irq = irq_find_mapping(NULL, ist);
 	
-	irq = irq_offset_up(irq);
 	free_irq(irq, dev_id);
-	
-	return;
 }
 EXPORT_SYMBOL(ibmebus_free_irq);
 

arch/powerpc/kernel/irq.c

@@ -29,6 +29,8 @@
  * to reduce code space and undefined function references.
  */
 
+#undef DEBUG
+
 #include <linux/module.h>
 #include <linux/threads.h>
 #include <linux/kernel_stat.h>
@@ -46,7 +48,10 @@
 #include <linux/cpumask.h>
 #include <linux/profile.h>
 #include <linux/bitops.h>
-#include <linux/pci.h>
+#include <linux/list.h>
+#include <linux/radix-tree.h>
+#include <linux/mutex.h>
+#include <linux/bootmem.h>
 
 #include <asm/uaccess.h>
 #include <asm/system.h>
@@ -57,6 +62,7 @@
 #include <asm/prom.h>
 #include <asm/ptrace.h>
 #include <asm/machdep.h>
+#include <asm/udbg.h>
 #ifdef CONFIG_PPC_ISERIES
 #include <asm/paca.h>
 #endif
@@ -88,7 +94,6 @@ extern atomic_t ipi_sent;
 EXPORT_SYMBOL(irq_desc);
 
 int distribute_irqs = 1;
-u64 ppc64_interrupt_controller;
 #endif /* CONFIG_PPC64 */
 
 int show_interrupts(struct seq_file *p, void *v)
@@ -181,7 +186,7 @@ void fixup_irqs(cpumask_t map)
 
 void do_IRQ(struct pt_regs *regs)
 {
-	int irq;
+	unsigned int irq;
 #ifdef CONFIG_IRQSTACKS
 	struct thread_info *curtp, *irqtp;
 #endif
@@ -212,7 +217,7 @@ void do_IRQ(struct pt_regs *regs)
 	 */
 	irq = ppc_md.get_irq(regs);
 
-	if (irq >= 0) {
+	if (irq != NO_IRQ && irq != NO_IRQ_IGNORE) {
 #ifdef CONFIG_IRQSTACKS
 		/* Switch to the irq stack to handle this */
 		curtp = current_thread_info();
@@ -231,7 +236,7 @@ void do_IRQ(struct pt_regs *regs)
 		} else
 #endif
 			generic_handle_irq(irq, regs);
-	} else if (irq != -2)
+	} else if (irq != NO_IRQ_IGNORE)
 		/* That's not SMP safe ... but who cares ? */
 		ppc_spurious_interrupts++;
 
@@ -254,123 +259,6 @@ void __init init_IRQ(void)
 #endif
 }
 
-#ifdef CONFIG_PPC64
-/*
- * Virtual IRQ mapping code, used on systems with XICS interrupt controllers.
- */
-
-#define UNDEFINED_IRQ 0xffffffff
-unsigned int virt_irq_to_real_map[NR_IRQS];
-
-/*
- * Don't use virtual irqs 0, 1, 2 for devices.
- * The pcnet32 driver considers interrupt numbers < 2 to be invalid,
- * and 2 is the XICS IPI interrupt.
- * We limit virtual irqs to __irq_offet_value less than virt_irq_max so
- * that when we offset them we don't end up with an interrupt
- * number >= virt_irq_max.
- */
-#define MIN_VIRT_IRQ	3
-
-unsigned int virt_irq_max;
-static unsigned int max_virt_irq;
-static unsigned int nr_virt_irqs;
-
-void
-virt_irq_init(void)
-{
-	int i;
-
-	if ((virt_irq_max == 0) || (virt_irq_max > (NR_IRQS - 1)))
-		virt_irq_max = NR_IRQS - 1;
-	max_virt_irq = virt_irq_max - __irq_offset_value;
-	nr_virt_irqs = max_virt_irq - MIN_VIRT_IRQ + 1;
-
-	for (i = 0; i < NR_IRQS; i++)
-		virt_irq_to_real_map[i] = UNDEFINED_IRQ;
-}
-
-/* Create a mapping for a real_irq if it doesn't already exist.
- * Return the virtual irq as a convenience.
- */
-int virt_irq_create_mapping(unsigned int real_irq)
-{
-	unsigned int virq, first_virq;
-	static int warned;
-
-	if (ppc64_interrupt_controller == IC_OPEN_PIC)
-		return real_irq;	/* no mapping for openpic (for now) */
-
-	if (ppc64_interrupt_controller == IC_CELL_PIC)
-		return real_irq;	/* no mapping for iic either */
-
-	/* don't map interrupts < MIN_VIRT_IRQ */
-	if (real_irq < MIN_VIRT_IRQ) {
-		virt_irq_to_real_map[real_irq] = real_irq;
-		return real_irq;
-	}
-
-	/* map to a number between MIN_VIRT_IRQ and max_virt_irq */
-	virq = real_irq;
-	if (virq > max_virt_irq)
-		virq = (virq % nr_virt_irqs) + MIN_VIRT_IRQ;
-
-	/* search for this number or a free slot */
-	first_virq = virq;
-	while (virt_irq_to_real_map[virq] != UNDEFINED_IRQ) {
-		if (virt_irq_to_real_map[virq] == real_irq)
-			return virq;
-		if (++virq > max_virt_irq)
-			virq = MIN_VIRT_IRQ;
-		if (virq == first_virq)
-			goto nospace;	/* oops, no free slots */
-	}
-
-	virt_irq_to_real_map[virq] = real_irq;
-	return virq;
-
- nospace:
-	if (!warned) {
-		printk(KERN_CRIT "Interrupt table is full\n");
-		printk(KERN_CRIT "Increase virt_irq_max (currently %d) "
-		       "in your kernel sources and rebuild.\n", virt_irq_max);
-		warned = 1;
-	}
-	return NO_IRQ;
-}
-
-/*
- * In most cases will get a hit on the very first slot checked in the
- * virt_irq_to_real_map.  Only when there are a large number of
- * IRQs will this be expensive.
- */
-unsigned int real_irq_to_virt_slowpath(unsigned int real_irq)
-{
-	unsigned int virq;
-	unsigned int first_virq;
-
-	virq = real_irq;
-
-	if (virq > max_virt_irq)
-		virq = (virq % nr_virt_irqs) + MIN_VIRT_IRQ;
-
-	first_virq = virq;
-
-	do {
-		if (virt_irq_to_real_map[virq] == real_irq)
-			return virq;
-
-		virq++;
-
-		if (virq >= max_virt_irq)
-			virq = 0;
-
-	} while (first_virq != virq);
-
-	return NO_IRQ;
-
-}
-#endif /* CONFIG_PPC64 */
 
 #ifdef CONFIG_IRQSTACKS
 struct thread_info *softirq_ctx[NR_CPUS] __read_mostly;
@@ -430,6 +318,503 @@ void do_softirq(void)
 }
 EXPORT_SYMBOL(do_softirq);
 
+
+/*
+ * IRQ controller and virtual interrupts
+ */
+
+#ifdef CONFIG_PPC_MERGE
+
+static LIST_HEAD(irq_hosts);
+static spinlock_t irq_big_lock = SPIN_LOCK_UNLOCKED;
+
+struct irq_map_entry irq_map[NR_IRQS];
+static unsigned int irq_virq_count = NR_IRQS;
+static struct irq_host *irq_default_host;
+
+struct irq_host *irq_alloc_host(unsigned int revmap_type,
+				unsigned int revmap_arg,
+				struct irq_host_ops *ops,
+				irq_hw_number_t inval_irq)
+{
+	struct irq_host *host;
+	unsigned int size = sizeof(struct irq_host);
+	unsigned int i;
+	unsigned int *rmap;
+	unsigned long flags;
+
+	/* Allocate structure and revmap table if using linear mapping */
+	if (revmap_type == IRQ_HOST_MAP_LINEAR)
+		size += revmap_arg * sizeof(unsigned int);
+	if (mem_init_done)
+		host = kzalloc(size, GFP_KERNEL);
+	else {
+		host = alloc_bootmem(size);
+		if (host)
+			memset(host, 0, size);
+	}
+	if (host == NULL)
+		return NULL;
+
+	/* Fill structure */
+	host->revmap_type = revmap_type;
+	host->inval_irq = inval_irq;
+	host->ops = ops;
+
+	spin_lock_irqsave(&irq_big_lock, flags);
+
+	/* If it's a legacy controller, check for duplicates and
+	 * mark it as allocated (we use irq 0 host pointer for that
+	 */
+	if (revmap_type == IRQ_HOST_MAP_LEGACY) {
+		if (irq_map[0].host != NULL) {
+			spin_unlock_irqrestore(&irq_big_lock, flags);
+			/* If we are early boot, we can't free the structure,
+			 * too bad...
+			 * this will be fixed once slab is made available early
+			 * instead of the current cruft
+			 */
+			if (mem_init_done)
+				kfree(host);
+			return NULL;
+		}
+		irq_map[0].host = host;
+	}
+
+	list_add(&host->link, &irq_hosts);
+	spin_unlock_irqrestore(&irq_big_lock, flags);
+
+	/* Additional setups per revmap type */
+	switch(revmap_type) {
+	case IRQ_HOST_MAP_LEGACY:
+		/* 0 is always the invalid number for legacy */
+		host->inval_irq = 0;
+		/* setup us as the host for all legacy interrupts */
+		for (i = 1; i < NUM_ISA_INTERRUPTS; i++) {
+			irq_map[i].hwirq = 0;
+			smp_wmb();
+			irq_map[i].host = host;
+			smp_wmb();
+
+			/* Clear some flags */
+			get_irq_desc(i)->status
+				&= ~(IRQ_NOREQUEST | IRQ_LEVEL);
+
+			/* 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);
+		}
+		break;
+	case IRQ_HOST_MAP_LINEAR:
+		rmap = (unsigned int *)(host + 1);
+		for (i = 0; i < revmap_arg; i++)
+			rmap[i] = IRQ_NONE;
+		host->revmap_data.linear.size = revmap_arg;
+		smp_wmb();
+		host->revmap_data.linear.revmap = rmap;
+		break;
+	default:
+		break;
+	}
+
+	pr_debug("irq: Allocated host of type %d @0x%p\n", revmap_type, host);
+
+	return host;
+}
+
+struct irq_host *irq_find_host(struct device_node *node)
+{
+	struct irq_host *h, *found = NULL;
+	unsigned long flags;
+
+	/* We might want to match the legacy controller last since
+	 * it might potentially be set to match all interrupts in
+	 * the absence of a device node. This isn't a problem so far
+	 * yet though...
+	 */
+	spin_lock_irqsave(&irq_big_lock, flags);
+	list_for_each_entry(h, &irq_hosts, link)
+		if (h->ops->match == NULL || h->ops->match(h, node)) {
+			found = h;
+			break;
+		}
+	spin_unlock_irqrestore(&irq_big_lock, flags);
+	return found;
+}
+EXPORT_SYMBOL_GPL(irq_find_host);
+
+void irq_set_default_host(struct irq_host *host)
+{
+	pr_debug("irq: Default host set to @0x%p\n", host);
+
+	irq_default_host = host;
+}
+
+void irq_set_virq_count(unsigned int count)
+{
+	pr_debug("irq: Trying to set virq count to %d\n", count);
+
+	BUG_ON(count < NUM_ISA_INTERRUPTS);
+	if (count < NR_IRQS)
+		irq_virq_count = count;
+}
+
+unsigned int irq_create_mapping(struct irq_host *host,
+				irq_hw_number_t hwirq,
+				unsigned int flags)
+{
+	unsigned int virq, hint;
+
+	pr_debug("irq: irq_create_mapping(0x%p, 0x%lx, 0x%x)\n",
+		 host, hwirq, flags);
+
+	/* Look for default host if nececssary */
+	if (host == NULL)
+		host = irq_default_host;
+	if (host == NULL) {
+		printk(KERN_WARNING "irq_create_mapping called for"
+		       " NULL host, hwirq=%lx\n", hwirq);
+		WARN_ON(1);
+		return NO_IRQ;
+	}
+	pr_debug("irq: -> using host @%p\n", host);
+
+	/* Check if mapping already exist, if it does, call
+	 * host->ops->map() to update the flags
+	 */
+	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;
+	}
+
+	/* Get a virtual interrupt number */
+	if (host->revmap_type == IRQ_HOST_MAP_LEGACY) {
+		/* Handle legacy */
+		virq = (unsigned int)hwirq;
+		if (virq == 0 || virq >= NUM_ISA_INTERRUPTS)
+			return NO_IRQ;
+		return virq;
+	} else {
+		/* Allocate a virtual interrupt number */
+		hint = hwirq % irq_virq_count;
+		virq = irq_alloc_virt(host, 1, hint);
+		if (virq == NO_IRQ) {
+			pr_debug("irq: -> virq allocation failed\n");
+			return NO_IRQ;
+		}
+	}
+	pr_debug("irq: -> obtained virq %d\n", virq);
+
+	/* Clear some flags */
+	get_irq_desc(virq)->status &= ~(IRQ_NOREQUEST | IRQ_LEVEL);
+
+	/* map it */
+	if (host->ops->map(host, virq, hwirq, flags)) {
+		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);
+
+extern unsigned int irq_create_of_mapping(struct device_node *controller,
+					  u32 *intspec, unsigned int intsize)
+{
+	struct irq_host *host;
+	irq_hw_number_t hwirq;
+	unsigned int flags = IRQ_TYPE_NONE;
+
+	if (controller == NULL)
+		host = irq_default_host;
+	else
+		host = irq_find_host(controller);
+	if (host == NULL)
+		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))
+			return NO_IRQ;
+	}
+
+	return irq_create_mapping(host, hwirq, flags);
+}
+EXPORT_SYMBOL_GPL(irq_create_of_mapping);
+
+unsigned int irq_of_parse_and_map(struct device_node *dev, int index)
+{
+	struct of_irq oirq;
+
+	if (of_irq_map_one(dev, index, &oirq))
+		return NO_IRQ;
+
+	return irq_create_of_mapping(oirq.controller, oirq.specifier,
+				     oirq.size);
+}
+EXPORT_SYMBOL_GPL(irq_of_parse_and_map);
+
+void irq_dispose_mapping(unsigned int virq)
+{
+	struct irq_host *host = irq_map[virq].host;
+	irq_hw_number_t hwirq;
+	unsigned long flags;
+
+	WARN_ON (host == NULL);
+	if (host == NULL)
+		return;
+
+	/* Never unmap legacy interrupts */
+	if (host->revmap_type == IRQ_HOST_MAP_LEGACY)
+		return;
+
+	/* remove chip and handler */
+	set_irq_chip_and_handler(virq, NULL, NULL);
+
+	/* Make sure it's completed */
+	synchronize_irq(virq);
+
+	/* Tell the PIC about it */
+	if (host->ops->unmap)
+		host->ops->unmap(host, virq);
+	smp_mb();
+
+	/* Clear reverse map */
+	hwirq = irq_map[virq].hwirq;
+	switch(host->revmap_type) {
+	case IRQ_HOST_MAP_LINEAR:
+		if (hwirq < host->revmap_data.linear.size)
+			host->revmap_data.linear.revmap[hwirq] = IRQ_NONE;
+		break;
+	case IRQ_HOST_MAP_TREE:
+		/* Check if radix tree allocated yet */
+		if (host->revmap_data.tree.gfp_mask == 0)
+			break;
+		/* XXX radix tree not safe ! remove lock whem it becomes safe
+		 * and use some RCU sync to make sure everything is ok before we
+		 * can re-use that map entry
+		 */
+		spin_lock_irqsave(&irq_big_lock, flags);
+		radix_tree_delete(&host->revmap_data.tree, hwirq);
+		spin_unlock_irqrestore(&irq_big_lock, flags);
+		break;
+	}
+
+	/* Destroy map */
+	smp_mb();
+	irq_map[virq].hwirq = host->inval_irq;
+
+	/* Set some flags */
+	get_irq_desc(virq)->status |= IRQ_NOREQUEST;
+
+	/* Free it */
+	irq_free_virt(virq, 1);
+}
+EXPORT_SYMBOL_GPL(irq_dispose_mapping);
+
+unsigned int irq_find_mapping(struct irq_host *host,
+			      irq_hw_number_t hwirq)
+{
+	unsigned int i;
+	unsigned int hint = hwirq % irq_virq_count;
+
+	/* Look for default host if nececssary */
+	if (host == NULL)
+		host = irq_default_host;
+	if (host == NULL)
+		return NO_IRQ;
+
+	/* legacy -> bail early */
+	if (host->revmap_type == IRQ_HOST_MAP_LEGACY)
+		return hwirq;
+
+	/* Slow path does a linear search of the map */
+	if (hint < NUM_ISA_INTERRUPTS)
+		hint = NUM_ISA_INTERRUPTS;
+	i = hint;
+	do  {
+		if (irq_map[i].host == host &&
+		    irq_map[i].hwirq == hwirq)
+			return i;
+		i++;
+		if (i >= irq_virq_count)
+			i = NUM_ISA_INTERRUPTS;
+	} while(i != hint);
+	return NO_IRQ;
+}
+EXPORT_SYMBOL_GPL(irq_find_mapping);
+
+
+unsigned int irq_radix_revmap(struct irq_host *host,
+			      irq_hw_number_t hwirq)
+{
+	struct radix_tree_root *tree;
+	struct irq_map_entry *ptr;
+	unsigned int virq;
+	unsigned long flags;
+
+	WARN_ON(host->revmap_type != IRQ_HOST_MAP_TREE);
+
+	/* Check if the radix tree exist yet. We test the value of
+	 * the gfp_mask for that. Sneaky but saves another int in the
+	 * structure. If not, we fallback to slow mode
+	 */
+	tree = &host->revmap_data.tree;
+	if (tree->gfp_mask == 0)
+		return irq_find_mapping(host, hwirq);
+
+	/* XXX Current radix trees are NOT SMP safe !!! Remove that lock
+	 * when that is fixed (when Nick's patch gets in
+	 */
+	spin_lock_irqsave(&irq_big_lock, flags);
+
+	/* Now try to resolve */
+	ptr = radix_tree_lookup(tree, hwirq);
+	/* Found it, return */
+	if (ptr) {
+		virq = ptr - irq_map;
+		goto bail;
+	}
+
+	/* If not there, try to insert it */
+	virq = irq_find_mapping(host, hwirq);
+	if (virq != NO_IRQ)
+		radix_tree_insert(tree, virq, &irq_map[virq]);
+ bail:
+	spin_unlock_irqrestore(&irq_big_lock, flags);
+	return virq;
+}
+
+unsigned int irq_linear_revmap(struct irq_host *host,
+			       irq_hw_number_t hwirq)
+{
+	unsigned int *revmap;
+
+	WARN_ON(host->revmap_type != IRQ_HOST_MAP_LINEAR);
+
+	/* Check revmap bounds */
+	if (unlikely(hwirq >= host->revmap_data.linear.size))
+		return irq_find_mapping(host, hwirq);
+
+	/* Check if revmap was allocated */
+	revmap = host->revmap_data.linear.revmap;
+	if (unlikely(revmap == NULL))
+		return irq_find_mapping(host, hwirq);
+
+	/* Fill up revmap with slow path if no mapping found */
+	if (unlikely(revmap[hwirq] == NO_IRQ))
+		revmap[hwirq] = irq_find_mapping(host, hwirq);
+
+	return revmap[hwirq];
+}
+
+unsigned int irq_alloc_virt(struct irq_host *host,
+			    unsigned int count,
+			    unsigned int hint)
+{
+	unsigned long flags;
+	unsigned int i, j, found = NO_IRQ;
+	unsigned int limit = irq_virq_count - count;
+
+	if (count == 0 || count > (irq_virq_count - NUM_ISA_INTERRUPTS))
+		return NO_IRQ;
+
+	spin_lock_irqsave(&irq_big_lock, flags);
+
+	/* Use hint for 1 interrupt if any */
+	if (count == 1 && hint >= NUM_ISA_INTERRUPTS &&
+	    hint < irq_virq_count && irq_map[hint].host == NULL) {
+		found = hint;
+		goto hint_found;
+	}
+
+	/* Look for count consecutive numbers in the allocatable
+	 * (non-legacy) space
+	 */
+	for (i = NUM_ISA_INTERRUPTS; i <= limit; ) {
+		for (j = i; j < (i + count); j++)
+			if (irq_map[j].host != NULL) {
+				i = j + 1;
+				continue;
+			}
+		found = i;
+		break;
+	}
+	if (found == NO_IRQ) {
+		spin_unlock_irqrestore(&irq_big_lock, flags);
+		return NO_IRQ;
+	}
+ hint_found:
+	for (i = found; i < (found + count); i++) {
+		irq_map[i].hwirq = host->inval_irq;
+		smp_wmb();
+		irq_map[i].host = host;
+	}
+	spin_unlock_irqrestore(&irq_big_lock, flags);
+	return found;
+}
+
+void irq_free_virt(unsigned int virq, unsigned int count)
+{
+	unsigned long flags;
+	unsigned int i;
+
+	WARN_ON (virq < NUM_ISA_INTERRUPTS);
+	WARN_ON (count == 0 || (virq + count) > irq_virq_count);
+
+	spin_lock_irqsave(&irq_big_lock, flags);
+	for (i = virq; i < (virq + count); i++) {
+		struct irq_host *host;
+
+		if (i < NUM_ISA_INTERRUPTS ||
+		    (virq + count) > irq_virq_count)
+			continue;
+
+		host = irq_map[i].host;
+		irq_map[i].hwirq = host->inval_irq;
+		smp_wmb();
+		irq_map[i].host = NULL;
+	}
+	spin_unlock_irqrestore(&irq_big_lock, flags);
+}
+
+void irq_early_init(void)
+{
+	unsigned int i;
+
+	for (i = 0; i < NR_IRQS; i++)
+		get_irq_desc(i)->status |= IRQ_NOREQUEST;
+}
+
+/* We need to create the radix trees late */
+static int irq_late_init(void)
+{
+	struct irq_host *h;
+	unsigned long flags;
+
+	spin_lock_irqsave(&irq_big_lock, flags);
+	list_for_each_entry(h, &irq_hosts, link) {
+		if (h->revmap_type == IRQ_HOST_MAP_TREE)
+			INIT_RADIX_TREE(&h->revmap_data.tree, GFP_ATOMIC);
+	}
+	spin_unlock_irqrestore(&irq_big_lock, flags);
+
+	return 0;
+}
+arch_initcall(irq_late_init);
+
+#endif /* CONFIG_PPC_MERGE */
+
 #ifdef CONFIG_PCI_MSI
 int pci_enable_msi(struct pci_dev * pdev)
 {

arch/powerpc/kernel/legacy_serial.c

@@ -28,6 +28,7 @@ static struct legacy_serial_info {
 	struct device_node		*np;
 	unsigned int			speed;
 	unsigned int			clock;
+	int				irq_check_parent;
 	phys_addr_t			taddr;
 } legacy_serial_infos[MAX_LEGACY_SERIAL_PORTS];
 static unsigned int legacy_serial_count;
@@ -36,7 +37,7 @@ static int legacy_serial_console = -1;
 static int __init add_legacy_port(struct device_node *np, int want_index,
 				  int iotype, phys_addr_t base,
 				  phys_addr_t taddr, unsigned long irq,
-				  upf_t flags)
+				  upf_t flags, int irq_check_parent)
 {
 	u32 *clk, *spd, clock = BASE_BAUD * 16;
 	int index;
@@ -68,7 +69,7 @@ static int __init add_legacy_port(struct device_node *np, int want_index,
 	if (legacy_serial_infos[index].np != 0) {
 		/* if we still have some room, move it, else override */
 		if (legacy_serial_count < MAX_LEGACY_SERIAL_PORTS) {
-			printk(KERN_INFO "Moved legacy port %d -> %d\n",
+			printk(KERN_DEBUG "Moved legacy port %d -> %d\n",
 			       index, legacy_serial_count);
 			legacy_serial_ports[legacy_serial_count] =
 				legacy_serial_ports[index];
@@ -76,7 +77,7 @@ static int __init add_legacy_port(struct device_node *np, int want_index,
 				legacy_serial_infos[index];
 			legacy_serial_count++;
 		} else {
-			printk(KERN_INFO "Replacing legacy port %d\n", index);
+			printk(KERN_DEBUG "Replacing legacy port %d\n", index);
 		}
 	}
 
@@ -95,10 +96,11 @@ static int __init add_legacy_port(struct device_node *np, int want_index,
 	legacy_serial_infos[index].np = of_node_get(np);
 	legacy_serial_infos[index].clock = clock;
 	legacy_serial_infos[index].speed = spd ? *spd : 0;
+	legacy_serial_infos[index].irq_check_parent = irq_check_parent;
 
-	printk(KERN_INFO "Found legacy serial port %d for %s\n",
+	printk(KERN_DEBUG "Found legacy serial port %d for %s\n",
 	       index, np->full_name);
-	printk(KERN_INFO "  %s=%llx, taddr=%llx, irq=%lx, clk=%d, speed=%d\n",
+	printk(KERN_DEBUG "  %s=%llx, taddr=%llx, irq=%lx, clk=%d, speed=%d\n",
 	       (iotype == UPIO_PORT) ? "port" : "mem",
 	       (unsigned long long)base, (unsigned long long)taddr, irq,
 	       legacy_serial_ports[index].uartclk,
@@ -132,7 +134,7 @@ static int __init add_legacy_soc_port(struct device_node *np,
 	/* Add port, irq will be dealt with later. We passed a translated
 	 * IO port value. It will be fixed up later along with the irq
 	 */
-	return add_legacy_port(np, -1, UPIO_MEM, addr, addr, NO_IRQ, flags);
+	return add_legacy_port(np, -1, UPIO_MEM, addr, addr, NO_IRQ, flags, 0);
 }
 
 static int __init add_legacy_isa_port(struct device_node *np,
@@ -170,7 +172,7 @@ static int __init add_legacy_isa_port(struct device_node *np,
 
 	/* Add port, irq will be dealt with later */
 	return add_legacy_port(np, index, UPIO_PORT, reg[1], taddr,
-			       NO_IRQ, UPF_BOOT_AUTOCONF);
+			       NO_IRQ, UPF_BOOT_AUTOCONF, 0);
 
 }
 
@@ -242,7 +244,8 @@ static int __init add_legacy_pci_port(struct device_node *np,
 	/* Add port, irq will be dealt with later. We passed a translated
 	 * IO port value. It will be fixed up later along with the irq
 	 */
-	return add_legacy_port(np, index, iotype, base, addr, NO_IRQ, UPF_BOOT_AUTOCONF);
+	return add_legacy_port(np, index, iotype, base, addr, NO_IRQ,
+			       UPF_BOOT_AUTOCONF, np != pci_dev);
 }
 #endif
 
@@ -373,27 +376,22 @@ static void __init fixup_port_irq(int index,
 				  struct device_node *np,
 				  struct plat_serial8250_port *port)
 {
+	unsigned int virq;
+
 	DBG("fixup_port_irq(%d)\n", index);
 
-	/* Check for interrupts in that node */
-	if (np->n_intrs > 0) {
-		port->irq = np->intrs[0].line;
-		DBG(" port %d (%s), irq=%d\n",
-		    index, np->full_name, port->irq);
-		return;
+	virq = irq_of_parse_and_map(np, 0);
+	if (virq == NO_IRQ && legacy_serial_infos[index].irq_check_parent) {
+		np = of_get_parent(np);
+		if (np == NULL)
+			return;
+		virq = irq_of_parse_and_map(np, 0);
+		of_node_put(np);
 	}
-
-	/* Check for interrupts in the parent */
-	np = of_get_parent(np);
-	if (np == NULL)
+	if (virq == NO_IRQ)
 		return;
 
-	if (np->n_intrs > 0) {
-		port->irq = np->intrs[0].line;
-		DBG(" port %d (%s), irq=%d\n",
-		    index, np->full_name, port->irq);
-	}
-	of_node_put(np);
+	port->irq = virq;
 }
 
 static void __init fixup_port_pio(int index,

arch/powerpc/kernel/pci_32.c

@@ -1404,6 +1404,43 @@ pcibios_update_irq(struct pci_dev *dev, int irq)
 	/* XXX FIXME - update OF device tree node interrupt property */
 }
 
+#ifdef CONFIG_PPC_MERGE
+/* XXX This is a copy of the ppc64 version. This is temporary until we start
+ * merging the 2 PCI layers
+ */
+/*
+ * Reads the interrupt pin to determine if interrupt is use by card.
+ * If the interrupt is used, then gets the interrupt line from the
+ * openfirmware and sets it in the pci_dev and pci_config line.
+ */
+int pci_read_irq_line(struct pci_dev *pci_dev)
+{
+	struct of_irq oirq;
+	unsigned int virq;
+
+	DBG("Try to map irq for %s...\n", pci_name(pci_dev));
+
+	if (of_irq_map_pci(pci_dev, &oirq)) {
+		DBG(" -> failed !\n");
+		return -1;
+	}
+
+	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);
+	if(virq == NO_IRQ) {
+		DBG(" -> failed to map !\n");
+		return -1;
+	}
+	pci_dev->irq = virq;
+	pci_write_config_byte(pci_dev, PCI_INTERRUPT_LINE, virq);
+
+	return 0;
+}
+EXPORT_SYMBOL(pci_read_irq_line);
+#endif /* CONFIG_PPC_MERGE */
+
 int pcibios_enable_device(struct pci_dev *dev, int mask)
 {
 	u16 cmd, old_cmd;

arch/powerpc/kernel/pci_64.c

@@ -398,12 +398,8 @@ struct pci_dev *of_create_pci_dev(struct device_node *node,
 	} else {
 		dev->hdr_type = PCI_HEADER_TYPE_NORMAL;
 		dev->rom_base_reg = PCI_ROM_ADDRESS;
+		/* Maybe do a default OF mapping here */
 		dev->irq = NO_IRQ;
-		if (node->n_intrs > 0) {
-			dev->irq = node->intrs[0].line;
-			pci_write_config_byte(dev, PCI_INTERRUPT_LINE,
-					      dev->irq);
-		}
 	}
 
 	pci_parse_of_addrs(node, dev);
@@ -1288,23 +1284,26 @@ EXPORT_SYMBOL(pcibios_fixup_bus);
  */
 int pci_read_irq_line(struct pci_dev *pci_dev)
 {
-	u8 intpin;
-	struct device_node *node;
-
-    	pci_read_config_byte(pci_dev, PCI_INTERRUPT_PIN, &intpin);
-	if (intpin == 0)
-		return 0;
+	struct of_irq oirq;
+	unsigned int virq;
 
-	node = pci_device_to_OF_node(pci_dev);
-	if (node == NULL)
-		return -1;
+	DBG("Try to map irq for %s...\n", pci_name(pci_dev));
 
-	if (node->n_intrs == 0)
+	if (of_irq_map_pci(pci_dev, &oirq)) {
+		DBG(" -> failed !\n");
 		return -1;
+	}
 
-	pci_dev->irq = node->intrs[0].line;
+	DBG(" -> got one, spec %d cells (0x%08x...) on %s\n",
+	    oirq.size, oirq.specifier[0], oirq.controller->full_name);
 
-	pci_write_config_byte(pci_dev, PCI_INTERRUPT_LINE, pci_dev->irq);
+	virq = irq_create_of_mapping(oirq.controller, oirq.specifier, oirq.size);
+	if(virq == NO_IRQ) {
+		DBG(" -> failed to map !\n");
+		return -1;
+	}
+	pci_dev->irq = virq;
+	pci_write_config_byte(pci_dev, PCI_INTERRUPT_LINE, virq);
 
 	return 0;
 }

arch/powerpc/kernel/prom.c

@@ -30,6 +30,7 @@
 #include <linux/module.h>
 #include <linux/kexec.h>
 #include <linux/debugfs.h>
+#include <linux/irq.h>
 
 #include <asm/prom.h>
 #include <asm/rtas.h>
@@ -86,424 +87,6 @@ static DEFINE_RWLOCK(devtree_lock);
 /* export that to outside world */
 struct device_node *of_chosen;
 
-struct device_node *dflt_interrupt_controller;
-int num_interrupt_controllers;
-
-/*
- * Wrapper for allocating memory for various data that needs to be
- * attached to device nodes as they are processed at boot or when
- * added to the device tree later (e.g. DLPAR).  At boot there is
- * already a region reserved so we just increment *mem_start by size;
- * otherwise we call kmalloc.
- */
-static void * prom_alloc(unsigned long size, unsigned long *mem_start)
-{
-	unsigned long tmp;
-
-	if (!mem_start)
-		return kmalloc(size, GFP_KERNEL);
-
-	tmp = *mem_start;
-	*mem_start += size;
-	return (void *)tmp;
-}
-
-/*
- * Find the device_node with a given phandle.
- */
-static struct device_node * find_phandle(phandle ph)
-{
-	struct device_node *np;
-
-	for (np = allnodes; np != 0; np = np->allnext)
-		if (np->linux_phandle == ph)
-			return np;
-	return NULL;
-}
-
-/*
- * Find the interrupt parent of a node.
- */
-static struct device_node * __devinit intr_parent(struct device_node *p)
-{
-	phandle *parp;
-
-	parp = (phandle *) get_property(p, "interrupt-parent", NULL);
-	if (parp == NULL)
-		return p->parent;
-	p = find_phandle(*parp);
-	if (p != NULL)
-		return p;
-	/*
-	 * On a powermac booted with BootX, we don't get to know the
-	 * phandles for any nodes, so find_phandle will return NULL.
-	 * Fortunately these machines only have one interrupt controller
-	 * so there isn't in fact any ambiguity.  -- paulus
-	 */
-	if (num_interrupt_controllers == 1)
-		p = dflt_interrupt_controller;
-	return p;
-}
-
-/*
- * Find out the size of each entry of the interrupts property
- * for a node.
- */
-int __devinit prom_n_intr_cells(struct device_node *np)
-{
-	struct device_node *p;
-	unsigned int *icp;
-
-	for (p = np; (p = intr_parent(p)) != NULL; ) {
-		icp = (unsigned int *)
-			get_property(p, "#interrupt-cells", NULL);
-		if (icp != NULL)
-			return *icp;
-		if (get_property(p, "interrupt-controller", NULL) != NULL
-		    || get_property(p, "interrupt-map", NULL) != NULL) {
-			printk("oops, node %s doesn't have #interrupt-cells\n",
-			       p->full_name);
-			return 1;
-		}
-	}
-#ifdef DEBUG_IRQ
-	printk("prom_n_intr_cells failed for %s\n", np->full_name);
-#endif
-	return 1;
-}
-
-/*
- * Map an interrupt from a device up to the platform interrupt
- * descriptor.
- */
-static int __devinit map_interrupt(unsigned int **irq, struct device_node **ictrler,
-				   struct device_node *np, unsigned int *ints,
-				   int nintrc)
-{
-	struct device_node *p, *ipar;
-	unsigned int *imap, *imask, *ip;
-	int i, imaplen, match;
-	int newintrc = 0, newaddrc = 0;
-	unsigned int *reg;
-	int naddrc;
-
-	reg = (unsigned int *) get_property(np, "reg", NULL);
-	naddrc = prom_n_addr_cells(np);
-	p = intr_parent(np);
-	while (p != NULL) {
-		if (get_property(p, "interrupt-controller", NULL) != NULL)
-			/* this node is an interrupt controller, stop here */
-			break;
-		imap = (unsigned int *)
-			get_property(p, "interrupt-map", &imaplen);
-		if (imap == NULL) {
-			p = intr_parent(p);
-			continue;
-		}
-		imask = (unsigned int *)
-			get_property(p, "interrupt-map-mask", NULL);
-		if (imask == NULL) {
-			printk("oops, %s has interrupt-map but no mask\n",
-			       p->full_name);
-			return 0;
-		}
-		imaplen /= sizeof(unsigned int);
-		match = 0;
-		ipar = NULL;
-		while (imaplen > 0 && !match) {
-			/* check the child-interrupt field */
-			match = 1;
-			for (i = 0; i < naddrc && match; ++i)
-				match = ((reg[i] ^ imap[i]) & imask[i]) == 0;
-			for (; i < naddrc + nintrc && match; ++i)
-				match = ((ints[i-naddrc] ^ imap[i]) & imask[i]) == 0;
-			imap += naddrc + nintrc;
-			imaplen -= naddrc + nintrc;
-			/* grab the interrupt parent */
-			ipar = find_phandle((phandle) *imap++);
-			--imaplen;
-			if (ipar == NULL && num_interrupt_controllers == 1)
-				/* cope with BootX not giving us phandles */
-				ipar = dflt_interrupt_controller;
-			if (ipar == NULL) {
-				printk("oops, no int parent %x in map of %s\n",
-				       imap[-1], p->full_name);
-				return 0;
-			}
-			/* find the parent's # addr and intr cells */
-			ip = (unsigned int *)
-				get_property(ipar, "#interrupt-cells", NULL);
-			if (ip == NULL) {
-				printk("oops, no #interrupt-cells on %s\n",
-				       ipar->full_name);
-				return 0;
-			}
-			newintrc = *ip;
-			ip = (unsigned int *)
-				get_property(ipar, "#address-cells", NULL);
-			newaddrc = (ip == NULL)? 0: *ip;
-			imap += newaddrc + newintrc;
-			imaplen -= newaddrc + newintrc;
-		}
-		if (imaplen < 0) {
-			printk("oops, error decoding int-map on %s, len=%d\n",
-			       p->full_name, imaplen);
-			return 0;
-		}
-		if (!match) {
-#ifdef DEBUG_IRQ
-			printk("oops, no match in %s int-map for %s\n",
-			       p->full_name, np->full_name);
-#endif
-			return 0;
-		}
-		p = ipar;
-		naddrc = newaddrc;
-		nintrc = newintrc;
-		ints = imap - nintrc;
-		reg = ints - naddrc;
-	}
-	if (p == NULL) {
-#ifdef DEBUG_IRQ
-		printk("hmmm, int tree for %s doesn't have ctrler\n",
-		       np->full_name);
-#endif
-		return 0;
-	}
-	*irq = ints;
-	*ictrler = p;
-	return nintrc;
-}
-
-static unsigned char map_isa_senses[4] = {
-	IRQ_SENSE_LEVEL | IRQ_POLARITY_NEGATIVE,
-	IRQ_SENSE_LEVEL | IRQ_POLARITY_POSITIVE,
-	IRQ_SENSE_EDGE  | IRQ_POLARITY_NEGATIVE,
-	IRQ_SENSE_EDGE  | IRQ_POLARITY_POSITIVE
-};
-
-static unsigned char map_mpic_senses[4] = {
-	IRQ_SENSE_EDGE  | IRQ_POLARITY_POSITIVE,
-	IRQ_SENSE_LEVEL | IRQ_POLARITY_NEGATIVE,
-	/* 2 seems to be used for the 8259 cascade... */
-	IRQ_SENSE_LEVEL | IRQ_POLARITY_POSITIVE,
-	IRQ_SENSE_EDGE  | IRQ_POLARITY_NEGATIVE,
-};
-
-static int __devinit finish_node_interrupts(struct device_node *np,
-					    unsigned long *mem_start,
-					    int measure_only)
-{
-	unsigned int *ints;
-	int intlen, intrcells, intrcount;
-	int i, j, n, sense;
-	unsigned int *irq, virq;
-	struct device_node *ic;
-	int trace = 0;
-
-	//#define TRACE(fmt...) do { if (trace) { printk(fmt); mdelay(1000); } } while(0)
-#define TRACE(fmt...)
-
-	if (!strcmp(np->name, "smu-doorbell"))
-		trace = 1;
-
-	TRACE("Finishing SMU doorbell ! num_interrupt_controllers = %d\n",
-	      num_interrupt_controllers);
-
-	if (num_interrupt_controllers == 0) {
-		/*
-		 * Old machines just have a list of interrupt numbers
-		 * and no interrupt-controller nodes.
-		 */
-		ints = (unsigned int *) get_property(np, "AAPL,interrupts",
-						     &intlen);
-		/* XXX old interpret_pci_props looked in parent too */
-		/* XXX old interpret_macio_props looked for interrupts
-		   before AAPL,interrupts */
-		if (ints == NULL)
-			ints = (unsigned int *) get_property(np, "interrupts",
-							     &intlen);
-		if (ints == NULL)
-			return 0;
-
-		np->n_intrs = intlen / sizeof(unsigned int);
-		np->intrs = prom_alloc(np->n_intrs * sizeof(np->intrs[0]),
-				       mem_start);
-		if (!np->intrs)
-			return -ENOMEM;
-		if (measure_only)
-			return 0;
-
-		for (i = 0; i < np->n_intrs; ++i) {
-			np->intrs[i].line = *ints++;
-			np->intrs[i].sense = IRQ_SENSE_LEVEL
-				| IRQ_POLARITY_NEGATIVE;
-		}
-		return 0;
-	}
-
-	ints = (unsigned int *) get_property(np, "interrupts", &intlen);
-	TRACE("ints=%p, intlen=%d\n", ints, intlen);
-	if (ints == NULL)
-		return 0;
-	intrcells = prom_n_intr_cells(np);
-	intlen /= intrcells * sizeof(unsigned int);
-	TRACE("intrcells=%d, new intlen=%d\n", intrcells, intlen);
-	np->intrs = prom_alloc(intlen * sizeof(*(np->intrs)), mem_start);
-	if (!np->intrs)
-		return -ENOMEM;
-
-	if (measure_only)
-		return 0;
-
-	intrcount = 0;
-	for (i = 0; i < intlen; ++i, ints += intrcells) {
-		n = map_interrupt(&irq, &ic, np, ints, intrcells);
-		TRACE("map, irq=%d, ic=%p, n=%d\n", irq, ic, n);
-		if (n <= 0)
-			continue;
-
-		/* don't map IRQ numbers under a cascaded 8259 controller */
-		if (ic && device_is_compatible(ic, "chrp,iic")) {
-			np->intrs[intrcount].line = irq[0];
-			sense = (n > 1)? (irq[1] & 3): 3;
-			np->intrs[intrcount].sense = map_isa_senses[sense];
-		} else {
-			virq = virt_irq_create_mapping(irq[0]);
-			TRACE("virq=%d\n", virq);
-#ifdef CONFIG_PPC64
-			if (virq == NO_IRQ) {
-				printk(KERN_CRIT "Could not allocate interrupt"
-				       " number for %s\n", np->full_name);
-				continue;
-			}
-#endif
-			np->intrs[intrcount].line = irq_offset_up(virq);
-			sense = (n > 1)? (irq[1] & 3): 1;
-
-			/* Apple uses bits in there in a different way, let's
-			 * only keep the real sense bit on macs
-			 */
-			if (machine_is(powermac))
-				sense &= 0x1;
-			np->intrs[intrcount].sense = map_mpic_senses[sense];
-		}
-
-#ifdef CONFIG_PPC64
-		/* We offset irq numbers for the u3 MPIC by 128 in PowerMac */
-		if (machine_is(powermac) && ic && ic->parent) {
-			char *name = get_property(ic->parent, "name", NULL);
-			if (name && !strcmp(name, "u3"))
-				np->intrs[intrcount].line += 128;
-			else if (!(name && (!strcmp(name, "mac-io") ||
-					    !strcmp(name, "u4"))))
-				/* ignore other cascaded controllers, such as
-				   the k2-sata-root */
-				break;
-		}
-#endif /* CONFIG_PPC64 */
-		if (n > 2) {
-			printk("hmmm, got %d intr cells for %s:", n,
-			       np->full_name);
-			for (j = 0; j < n; ++j)
-				printk(" %d", irq[j]);
-			printk("\n");
-		}
-		++intrcount;
-	}
-	np->n_intrs = intrcount;
-
-	return 0;
-}
-
-static int __devinit finish_node(struct device_node *np,
-				 unsigned long *mem_start,
-				 int measure_only)
-{
-	struct device_node *child;
-	int rc = 0;
-
-	rc = finish_node_interrupts(np, mem_start, measure_only);
-	if (rc)
-		goto out;
-
-	for (child = np->child; child != NULL; child = child->sibling) {
-		rc = finish_node(child, mem_start, measure_only);
-		if (rc)
-			goto out;
-	}
-out:
-	return rc;
-}
-
-static void __init scan_interrupt_controllers(void)
-{
-	struct device_node *np;
-	int n = 0;
-	char *name, *ic;
-	int iclen;
-
-	for (np = allnodes; np != NULL; np = np->allnext) {
-		ic = get_property(np, "interrupt-controller", &iclen);
-		name = get_property(np, "name", NULL);
-		/* checking iclen makes sure we don't get a false
-		   match on /chosen.interrupt_controller */
-		if ((name != NULL
-		     && strcmp(name, "interrupt-controller") == 0)
-		    || (ic != NULL && iclen == 0
-			&& strcmp(name, "AppleKiwi"))) {
-			if (n == 0)
-				dflt_interrupt_controller = np;
-			++n;
-		}
-	}
-	num_interrupt_controllers = n;
-}
-
-/**
- * finish_device_tree is called once things are running normally
- * (i.e. with text and data mapped to the address they were linked at).
- * It traverses the device tree and fills in some of the additional,
- * fields in each node like {n_}addrs and {n_}intrs, the virt interrupt
- * mapping is also initialized at this point.
- */
-void __init finish_device_tree(void)
-{
-	unsigned long start, end, size = 0;
-
-	DBG(" -> finish_device_tree\n");
-
-#ifdef CONFIG_PPC64
-	/* Initialize virtual IRQ map */
-	virt_irq_init();
-#endif
-	scan_interrupt_controllers();
-
-	/*
-	 * Finish device-tree (pre-parsing some properties etc...)
-	 * We do this in 2 passes. One with "measure_only" set, which
-	 * will only measure the amount of memory needed, then we can
-	 * allocate that memory, and call finish_node again. However,
-	 * we must be careful as most routines will fail nowadays when
-	 * prom_alloc() returns 0, so we must make sure our first pass
-	 * doesn't start at 0. We pre-initialize size to 16 for that
-	 * reason and then remove those additional 16 bytes
-	 */
-	size = 16;
-	finish_node(allnodes, &size, 1);
-	size -= 16;
-
-	if (0 == size)
-		end = start = 0;
-	else
-		end = start = (unsigned long)__va(lmb_alloc(size, 128));
-
-	finish_node(allnodes, &end, 0);
-	BUG_ON(end != start + size);
-
-	DBG(" <- finish_device_tree\n");
-}
-
 static inline char *find_flat_dt_string(u32 offset)
 {
 	return ((char *)initial_boot_params) +
@@ -1388,27 +971,6 @@ prom_n_size_cells(struct device_node* np)
 }
 EXPORT_SYMBOL(prom_n_size_cells);
 
-/**
- * Work out the sense (active-low level / active-high edge)
- * of each interrupt from the device tree.
- */
-void __init prom_get_irq_senses(unsigned char *senses, int off, int max)
-{
-	struct device_node *np;
-	int i, j;
-
-	/* default to level-triggered */
-	memset(senses, IRQ_SENSE_LEVEL | IRQ_POLARITY_NEGATIVE, max - off);
-
-	for (np = allnodes; np != 0; np = np->allnext) {
-		for (j = 0; j < np->n_intrs; j++) {
-			i = np->intrs[j].line;
-			if (i >= off && i < max)
-				senses[i-off] = np->intrs[j].sense;
-		}
-	}
-}
-
 /**
  * Construct and return a list of the device_nodes with a given name.
  */
@@ -1808,7 +1370,6 @@ static void of_node_release(struct kref *kref)
 			node->deadprops = NULL;
 		}
 	}
-	kfree(node->intrs);
 	kfree(node->full_name);
 	kfree(node->data);
 	kfree(node);
@@ -1881,13 +1442,7 @@ void of_detach_node(const struct device_node *np)
 #ifdef CONFIG_PPC_PSERIES
 /*
  * Fix up the uninitialized fields in a new device node:
- * name, type, n_addrs, addrs, n_intrs, intrs, and pci-specific fields
- *
- * A lot of boot-time code is duplicated here, because functions such
- * as finish_node_interrupts, interpret_pci_props, etc. cannot use the
- * slab allocator.
- *
- * This should probably be split up into smaller chunks.
+ * name, type and pci-specific fields
  */
 
 static int of_finish_dynamic_node(struct device_node *node)
@@ -1928,8 +1483,6 @@ static int prom_reconfig_notifier(struct notifier_block *nb,
 	switch (action) {
 	case PSERIES_RECONFIG_ADD:
 		err = of_finish_dynamic_node(node);
-		if (!err)
-			finish_node(node, NULL, 0);
 		if (err < 0) {
 			printk(KERN_ERR "finish_node returned %d\n", err);
 			err = NOTIFY_BAD;

arch/powerpc/kernel/rtas_pci.c

@@ -297,19 +297,9 @@ unsigned long __init find_and_init_phbs(void)
 	struct device_node *node;
 	struct pci_controller *phb;
 	unsigned int index;
-	unsigned int root_size_cells = 0;
-	unsigned int *opprop = NULL;
 	struct device_node *root = of_find_node_by_path("/");
 
-	if (ppc64_interrupt_controller == IC_OPEN_PIC) {
-		opprop = (unsigned int *)get_property(root,
-				"platform-open-pic", NULL);
-	}
-
-	root_size_cells = prom_n_size_cells(root);
-
 	index = 0;
-
 	for (node = of_get_next_child(root, NULL);
 	     node != NULL;
 	     node = of_get_next_child(root, node)) {
@@ -324,13 +314,6 @@ unsigned long __init find_and_init_phbs(void)
 		setup_phb(node, phb);
 		pci_process_bridge_OF_ranges(phb, node, 0);
 		pci_setup_phb_io(phb, index == 0);
-#ifdef CONFIG_PPC_PSERIES
-		/* XXX This code need serious fixing ... --BenH */
-		if (ppc64_interrupt_controller == IC_OPEN_PIC && pSeries_mpic) {
-			int addr = root_size_cells * (index + 2) - 1;
-			mpic_assign_isu(pSeries_mpic, index, opprop[addr]);
-		}
-#endif
 		index++;
 	}
 

arch/powerpc/kernel/setup_32.c

@@ -239,7 +239,6 @@ void __init setup_arch(char **cmdline_p)
 		ppc_md.init_early();
 
 	find_legacy_serial_ports();
-	finish_device_tree();
 
 	smp_setup_cpu_maps();
 

arch/powerpc/kernel/setup_64.c

@@ -361,12 +361,15 @@ void __init setup_system(void)
 
 	/*
 	 * Fill the ppc64_caches & systemcfg structures with informations
-	 * retrieved from the device-tree. Need to be called before
-	 * finish_device_tree() since the later requires some of the
-	 * informations filled up here to properly parse the interrupt tree.
+ 	 * retrieved from the device-tree.
 	 */
 	initialize_cache_info();
 
+	/*
+	 * Initialize irq remapping subsystem
+	 */
+	irq_early_init();
+
 #ifdef CONFIG_PPC_RTAS
 	/*
 	 * Initialize RTAS if available
@@ -393,12 +396,6 @@ void __init setup_system(void)
 	 */
 	find_legacy_serial_ports();
 
-	/*
-	 * "Finish" the device-tree, that is do the actual parsing of
-	 * some of the properties like the interrupt map
-	 */
-	finish_device_tree();
-
 	/*
 	 * Initialize xmon
 	 */
@@ -427,8 +424,6 @@ void __init setup_system(void)
 
 	printk("-----------------------------------------------------\n");
 	printk("ppc64_pft_size                = 0x%lx\n", ppc64_pft_size);
-	printk("ppc64_interrupt_controller    = 0x%ld\n",
-	       ppc64_interrupt_controller);
 	printk("physicalMemorySize            = 0x%lx\n", lmb_phys_mem_size());
 	printk("ppc64_caches.dcache_line_size = 0x%x\n",
 	       ppc64_caches.dline_size);

arch/powerpc/kernel/vio.c

@@ -218,7 +218,6 @@ struct vio_dev * __devinit vio_register_device_node(struct device_node *of_node)
 {
 	struct vio_dev *viodev;
 	unsigned int *unit_address;
-	unsigned int *irq_p;
 
 	/* we need the 'device_type' property, in order to match with drivers */
 	if (of_node->type == NULL) {
@@ -243,16 +242,7 @@ struct vio_dev * __devinit vio_register_device_node(struct device_node *of_node)
 
 	viodev->dev.platform_data = of_node_get(of_node);
 
-	viodev->irq = NO_IRQ;
-	irq_p = (unsigned int *)get_property(of_node, "interrupts", NULL);
-	if (irq_p) {
-		int virq = virt_irq_create_mapping(*irq_p);
-		if (virq == NO_IRQ) {
-			printk(KERN_ERR "Unable to allocate interrupt "
-			       "number for %s\n", of_node->full_name);
-		} else
-			viodev->irq = irq_offset_up(virq);
-	}
+	viodev->irq = irq_of_parse_and_map(of_node, 0);
 
 	snprintf(viodev->dev.bus_id, BUS_ID_SIZE, "%x", *unit_address);
 	viodev->name = of_node->name;

arch/powerpc/platforms/cell/interrupt.c

@@ -1,6 +1,9 @@
 /*
  * Cell Internal Interrupt Controller
  *
+ * Copyright (C) 2006 Benjamin Herrenschmidt (benh@kernel.crashing.org)
+ *                    IBM, Corp.
+ *
  * (C) Copyright IBM Deutschland Entwicklung GmbH 2005
  *
  * Author: Arnd Bergmann <arndb@de.ibm.com>
@@ -25,11 +28,13 @@
 #include <linux/module.h>
 #include <linux/percpu.h>
 #include <linux/types.h>
+#include <linux/ioport.h>
 
 #include <asm/io.h>
 #include <asm/pgtable.h>
 #include <asm/prom.h>
 #include <asm/ptrace.h>
+#include <asm/machdep.h>
 
 #include "interrupt.h"
 #include "cbe_regs.h"
@@ -39,9 +44,25 @@ struct iic {
 	u8 target_id;
 	u8 eoi_stack[16];
 	int eoi_ptr;
+	struct irq_host *host;
 };
 
 static DEFINE_PER_CPU(struct iic, iic);
+#define IIC_NODE_COUNT	2
+static struct irq_host *iic_hosts[IIC_NODE_COUNT];
+
+/* Convert between "pending" bits and hw irq number */
+static irq_hw_number_t iic_pending_to_hwnum(struct cbe_iic_pending_bits bits)
+{
+	unsigned char unit = bits.source & 0xf;
+
+	if (bits.flags & CBE_IIC_IRQ_IPI)
+		return IIC_IRQ_IPI0 | (bits.prio >> 4);
+	else if (bits.class <= 3)
+		return (bits.class << 4) | unit;
+	else
+		return IIC_IRQ_INVALID;
+}
 
 static void iic_mask(unsigned int irq)
 {
@@ -65,197 +86,21 @@ static struct irq_chip iic_chip = {
 	.eoi = iic_eoi,
 };
 
-/* XXX All of this has to be reworked completely. We need to assign a real
- * interrupt numbers to the external interrupts and remove all the hard coded
- * interrupt maps (rely on the device-tree whenever possible).
- *
- * Basically, my scheme is to define the "pendings" bits to be the HW interrupt
- * number (ignoring the data and flags here). That means we can sort-of split
- * external sources based on priority, and we can use request_irq() on pretty
- * much anything.
- *
- * For spider or axon, they have their own interrupt space. spider will just have
- * local "hardward" interrupts 0...xx * node stride. The node stride is not
- * necessary (separate interrupt chips will have separate HW number space), but
- * will allow to be compatible with existing device-trees.
- *
- * All of thise little world will get a standard remapping scheme to map those HW
- * numbers into the linux flat irq number space.
-*/
-static int iic_external_get_irq(struct cbe_iic_pending_bits pending)
-{
-	int irq;
-	unsigned char node, unit;
-
-	node = pending.source >> 4;
-	unit = pending.source & 0xf;
-	irq = -1;
-
-	/*
-	 * This mapping is specific to the Cell Broadband
-	 * Engine. We might need to get the numbers
-	 * from the device tree to support future CPUs.
-	 */
-	switch (unit) {
-	case 0x00:
-	case 0x0b:
-		/*
-		 * One of these units can be connected
-		 * to an external interrupt controller.
-		 */
-		if (pending.class != 2)
-			break;
-		/* TODO: We might want to silently ignore cascade interrupts
-		 * when no cascade handler exist yet
-		 */
-		irq = IIC_EXT_CASCADE + node * IIC_NODE_STRIDE;
-		break;
-	case 0x01 ... 0x04:
-	case 0x07 ... 0x0a:
-		/*
-		 * These units are connected to the SPEs
-		 */
-		if (pending.class > 2)
-			break;
-		irq = IIC_SPE_OFFSET
-			+ pending.class * IIC_CLASS_STRIDE
-			+ node * IIC_NODE_STRIDE
-			+ unit;
-		break;
-	}
-	if (irq == -1)
-		printk(KERN_WARNING "Unexpected interrupt class %02x, "
-			"source %02x, prio %02x, cpu %02x\n", pending.class,
-			pending.source, pending.prio, smp_processor_id());
-	return irq;
-}
-
 /* Get an IRQ number from the pending state register of the IIC */
-int iic_get_irq(struct pt_regs *regs)
-{
-	struct iic *iic;
-	int irq;
-	struct cbe_iic_pending_bits pending;
-
-	iic = &__get_cpu_var(iic);
-	*(unsigned long *) &pending = 
-		in_be64((unsigned long __iomem *) &iic->regs->pending_destr);
-	iic->eoi_stack[++iic->eoi_ptr] = pending.prio;
-	BUG_ON(iic->eoi_ptr > 15);
-
-	irq = -1;
-	if (pending.flags & CBE_IIC_IRQ_VALID) {
-		if (pending.flags & CBE_IIC_IRQ_IPI) {
-			irq = IIC_IPI_OFFSET + (pending.prio >> 4);
-/*
-			if (irq > 0x80)
-				printk(KERN_WARNING "Unexpected IPI prio %02x"
-					"on CPU %02x\n", pending.prio,
-							smp_processor_id());
-*/
-		} else {
-			irq = iic_external_get_irq(pending);
-		}
-	}
-	return irq;
-}
-
-/* hardcoded part to be compatible with older firmware */
-
-static int __init setup_iic_hardcoded(void)
-{
-	struct device_node *np;
-	int nodeid, cpu;
-	unsigned long regs;
-	struct iic *iic;
-
-	for_each_possible_cpu(cpu) {
-		iic = &per_cpu(iic, cpu);
-		nodeid = cpu/2;
-
-		for (np = of_find_node_by_type(NULL, "cpu");
-		     np;
-		     np = of_find_node_by_type(np, "cpu")) {
-			if (nodeid == *(int *)get_property(np, "node-id", NULL))
-				break;
-			}
-
-		if (!np) {
-			printk(KERN_WARNING "IIC: CPU %d not found\n", cpu);
-			iic->regs = NULL;
-			iic->target_id = 0xff;
-			return -ENODEV;
-			}
-
-		regs = *(long *)get_property(np, "iic", NULL);
-
-		/* hack until we have decided on the devtree info */
-		regs += 0x400;
-		if (cpu & 1)
-			regs += 0x20;
-
-		printk(KERN_INFO "IIC for CPU %d at %lx\n", cpu, regs);
-		iic->regs = ioremap(regs, sizeof(struct cbe_iic_thread_regs));
-		iic->target_id = (nodeid << 4) + ((cpu & 1) ? 0xf : 0xe);
-		iic->eoi_stack[0] = 0xff;
-	}
-
-	return 0;
-}
-
-static int __init setup_iic(void)
+static unsigned int iic_get_irq(struct pt_regs *regs)
 {
-	struct device_node *dn;
-	unsigned long *regs;
-	char *compatible;
- 	unsigned *np, found = 0;
-	struct iic *iic = NULL;
-
-	for (dn = NULL; (dn = of_find_node_by_name(dn, "interrupt-controller"));) {
-		compatible = (char *)get_property(dn, "compatible", NULL);
-
-		if (!compatible) {
-			printk(KERN_WARNING "no compatible property found !\n");
-			continue;
-		}
-
- 		if (strstr(compatible, "IBM,CBEA-Internal-Interrupt-Controller"))
- 			regs = (unsigned long *)get_property(dn,"reg", NULL);
- 		else
-			continue;
-
- 		if (!regs)
- 			printk(KERN_WARNING "IIC: no reg property\n");
-
- 		np = (unsigned int *)get_property(dn, "ibm,interrupt-server-ranges", NULL);
-
- 		if (!np) {
-			printk(KERN_WARNING "IIC: CPU association not found\n");
-			iic->regs = NULL;
-			iic->target_id = 0xff;
-			return -ENODEV;
-		}
-
- 		iic = &per_cpu(iic, np[0]);
- 		iic->regs = ioremap(regs[0], sizeof(struct cbe_iic_thread_regs));
-		iic->target_id = ((np[0] & 2) << 3) + ((np[0] & 1) ? 0xf : 0xe);
-		iic->eoi_stack[0] = 0xff;
- 		printk("IIC for CPU %d at %lx mapped to %p\n", np[0], regs[0], iic->regs);
-
- 		iic = &per_cpu(iic, np[1]);
- 		iic->regs = ioremap(regs[2], sizeof(struct cbe_iic_thread_regs));
-		iic->target_id = ((np[1] & 2) << 3) + ((np[1] & 1) ? 0xf : 0xe);
-		iic->eoi_stack[0] = 0xff;
-
- 		printk("IIC for CPU %d at %lx mapped to %p\n", np[1], regs[2], iic->regs);
-
-		found++;
-  	}
-
-	if (found)
-		return 0;
-	else
-		return -ENODEV;
+  	struct cbe_iic_pending_bits pending;
+ 	struct iic *iic;
+
+ 	iic = &__get_cpu_var(iic);
+ 	*(unsigned long *) &pending =
+ 		in_be64((unsigned long __iomem *) &iic->regs->pending_destr);
+ 	iic->eoi_stack[++iic->eoi_ptr] = pending.prio;
+ 	BUG_ON(iic->eoi_ptr > 15);
+	if (pending.flags & CBE_IIC_IRQ_VALID)
+		return irq_linear_revmap(iic->host,
+ 					 iic_pending_to_hwnum(pending));
+	return NO_IRQ;
 }
 
 #ifdef CONFIG_SMP
@@ -263,12 +108,12 @@ static int __init setup_iic(void)
 /* Use the highest interrupt priorities for IPI */
 static inline int iic_ipi_to_irq(int ipi)
 {
-	return IIC_IPI_OFFSET + IIC_NUM_IPIS - 1 - ipi;
+	return IIC_IRQ_IPI0 + IIC_NUM_IPIS - 1 - ipi;
 }
 
 static inline int iic_irq_to_ipi(int irq)
 {
-	return IIC_NUM_IPIS - 1 - (irq - IIC_IPI_OFFSET);
+	return IIC_NUM_IPIS - 1 - (irq - IIC_IRQ_IPI0);
 }
 
 void iic_setup_cpu(void)
@@ -287,22 +132,51 @@ u8 iic_get_target_id(int cpu)
 }
 EXPORT_SYMBOL_GPL(iic_get_target_id);
 
+struct irq_host *iic_get_irq_host(int node)
+{
+	if (node < 0 || node >= IIC_NODE_COUNT)
+		return NULL;
+	return iic_hosts[node];
+}
+EXPORT_SYMBOL_GPL(iic_get_irq_host);
+
+
 static irqreturn_t iic_ipi_action(int irq, void *dev_id, struct pt_regs *regs)
 {
-	smp_message_recv(iic_irq_to_ipi(irq), regs);
+	int ipi = (int)(long)dev_id;
+
+	smp_message_recv(ipi, regs);
+
 	return IRQ_HANDLED;
 }
 
 static void iic_request_ipi(int ipi, const char *name)
 {
-	int irq;
+	int node, virq;
 
-	irq = iic_ipi_to_irq(ipi);
-
-	/* IPIs are marked IRQF_DISABLED as they must run with irqs
-	 * disabled */
- 	set_irq_chip_and_handler(irq, &iic_chip, handle_percpu_irq);
-	request_irq(irq, iic_ipi_action, IRQF_DISABLED, name, NULL);
+	for (node = 0; node < IIC_NODE_COUNT; node++) {
+		char *rname;
+		if (iic_hosts[node] == NULL)
+			continue;
+		virq = irq_create_mapping(iic_hosts[node],
+					  iic_ipi_to_irq(ipi), 0);
+		if (virq == NO_IRQ) {
+			printk(KERN_ERR
+			       "iic: failed to map IPI %s on node %d\n",
+			       name, node);
+			continue;
+		}
+		rname = kzalloc(strlen(name) + 16, GFP_KERNEL);
+		if (rname)
+			sprintf(rname, "%s node %d", name, node);
+		else
+			rname = (char *)name;
+		if (request_irq(virq, iic_ipi_action, IRQF_DISABLED,
+				rname, (void *)(long)ipi))
+			printk(KERN_ERR
+			       "iic: failed to request IPI %s on node %d\n",
+			       name, node);
+	}
 }
 
 void iic_request_IPIs(void)
@@ -313,41 +187,119 @@ void iic_request_IPIs(void)
 	iic_request_ipi(PPC_MSG_DEBUGGER_BREAK, "IPI-debug");
 #endif /* CONFIG_DEBUGGER */
 }
+
 #endif /* CONFIG_SMP */
 
-static void __init iic_setup_builtin_handlers(void)
+
+static int iic_host_match(struct irq_host *h, struct device_node *node)
+{
+	return h->host_data != NULL && node == h->host_data;
+}
+
+static int iic_host_map(struct irq_host *h, unsigned int virq,
+			irq_hw_number_t hw, unsigned int flags)
+{
+	if (hw < IIC_IRQ_IPI0)
+		set_irq_chip_and_handler(virq, &iic_chip, handle_fasteoi_irq);
+	else
+		set_irq_chip_and_handler(virq, &iic_chip, handle_percpu_irq);
+	return 0;
+}
+
+static int iic_host_xlate(struct irq_host *h, struct device_node *ct,
+			   u32 *intspec, unsigned int intsize,
+			   irq_hw_number_t *out_hwirq, unsigned int *out_flags)
+
 {
-	int be, isrc;
+	/* Currently, we don't translate anything. That needs to be fixed as
+	 * we get better defined device-trees. iic interrupts have to be
+	 * explicitely mapped by whoever needs them
+	 */
+	return -ENODEV;
+}
+
+static struct irq_host_ops iic_host_ops = {
+	.match = iic_host_match,
+	.map = iic_host_map,
+	.xlate = iic_host_xlate,
+};
+
+static void __init init_one_iic(unsigned int hw_cpu, unsigned long addr,
+				struct irq_host *host)
+{
+	/* XXX FIXME: should locate the linux CPU number from the HW cpu
+	 * number properly. We are lucky for now
+	 */
+	struct iic *iic = &per_cpu(iic, hw_cpu);
 
-	/* XXX FIXME: Assume two threads per BE are present */
-	for (be=0; be < num_present_cpus() / 2; be++) {
-		int irq;
+	iic->regs = ioremap(addr, sizeof(struct cbe_iic_thread_regs));
+	BUG_ON(iic->regs == NULL);
 
-		/* setup SPE chip and handlers */
-		for (isrc = 0; isrc < IIC_CLASS_STRIDE * 3; isrc++) {
-			irq = IIC_NODE_STRIDE * be + IIC_SPE_OFFSET + isrc;
-			set_irq_chip_and_handler(irq, &iic_chip, handle_fasteoi_irq);
+	iic->target_id = ((hw_cpu & 2) << 3) | ((hw_cpu & 1) ? 0xf : 0xe);
+	iic->eoi_stack[0] = 0xff;
+	iic->host = host;
+	out_be64(&iic->regs->prio, 0);
+
+	printk(KERN_INFO "IIC for CPU %d at %lx mapped to %p, target id 0x%x\n",
+	       hw_cpu, addr, iic->regs, iic->target_id);
+}
+
+static int __init setup_iic(void)
+{
+	struct device_node *dn;
+	struct resource r0, r1;
+	struct irq_host *host;
+	int found = 0;
+ 	u32 *np;
+
+	for (dn = NULL;
+	     (dn = of_find_node_by_name(dn,"interrupt-controller")) != NULL;) {
+		if (!device_is_compatible(dn,
+				     "IBM,CBEA-Internal-Interrupt-Controller"))
+			continue;
+ 		np = (u32 *)get_property(dn, "ibm,interrupt-server-ranges",
+					 NULL);
+ 		if (np == NULL) {
+			printk(KERN_WARNING "IIC: CPU association not found\n");
+			of_node_put(dn);
+			return -ENODEV;
 		}
-		/* setup cascade chip */
-		irq = IIC_EXT_CASCADE + be * IIC_NODE_STRIDE;
-		set_irq_chip_and_handler(irq, &iic_chip, handle_fasteoi_irq);
+		if (of_address_to_resource(dn, 0, &r0) ||
+		    of_address_to_resource(dn, 1, &r1)) {
+			printk(KERN_WARNING "IIC: Can't resolve addresses\n");
+			of_node_put(dn);
+			return -ENODEV;
+		}
+		host = NULL;
+		if (found < IIC_NODE_COUNT) {
+			host = irq_alloc_host(IRQ_HOST_MAP_LINEAR,
+					      IIC_SOURCE_COUNT,
+					      &iic_host_ops,
+					      IIC_IRQ_INVALID);
+			iic_hosts[found] = host;
+			BUG_ON(iic_hosts[found] == NULL);
+			iic_hosts[found]->host_data = of_node_get(dn);
+			found++;
+		}
+		init_one_iic(np[0], r0.start, host);
+		init_one_iic(np[1], r1.start, host);
 	}
+
+	if (found)
+		return 0;
+	else
+		return -ENODEV;
 }
 
 void __init iic_init_IRQ(void)
 {
-	int cpu, irq_offset;
-	struct iic *iic;
-
+	/* Discover and initialize iics */
 	if (setup_iic() < 0)
-		setup_iic_hardcoded();
+		panic("IIC: Failed to initialize !\n");
 
-	irq_offset = 0;
-	for_each_possible_cpu(cpu) {
-		iic = &per_cpu(iic, cpu);
-		if (iic->regs)
-			out_be64(&iic->regs->prio, 0xff);
-	}
-	iic_setup_builtin_handlers();
+	/* Set master interrupt handling function */
+	ppc_md.get_irq = iic_get_irq;
 
+	/* Enable on current CPU */
+	iic_setup_cpu();
 }

arch/powerpc/platforms/cell/interrupt.h

@@ -37,23 +37,22 @@
  */
 
 enum {
-	IIC_EXT_OFFSET   = 0x00, /* Start of south bridge IRQs */
-	IIC_EXT_CASCADE  = 0x20, /* There is no interrupt 32 on spider */
-	IIC_NUM_EXT      = 0x40, /* Number of south bridge IRQs */
-	IIC_SPE_OFFSET   = 0x40, /* Start of SPE interrupts */
-	IIC_CLASS_STRIDE = 0x10, /* SPE IRQs per class    */
-	IIC_IPI_OFFSET   = 0x70, /* Start of IPI IRQs */
-	IIC_NUM_IPIS     = 0x10, /* IRQs reserved for IPI */
-	IIC_NODE_STRIDE  = 0x80, /* Total IRQs per node   */
+	IIC_IRQ_INVALID		= 0xff,
+	IIC_IRQ_MAX		= 0x3f,
+	IIC_IRQ_EXT_IOIF0	= 0x20,
+	IIC_IRQ_EXT_IOIF1	= 0x2b,
+	IIC_IRQ_IPI0		= 0x40,
+	IIC_NUM_IPIS    	= 0x10, /* IRQs reserved for IPI */
+	IIC_SOURCE_COUNT	= 0x50,
 };
 
 extern void iic_init_IRQ(void);
-extern int  iic_get_irq(struct pt_regs *regs);
 extern void iic_cause_IPI(int cpu, int mesg);
 extern void iic_request_IPIs(void);
 extern void iic_setup_cpu(void);
 
 extern u8 iic_get_target_id(int cpu);
+extern struct irq_host *iic_get_irq_host(int node);
 
 extern void spider_init_IRQ(void);
 

arch/powerpc/platforms/cell/setup.c

@@ -80,6 +80,14 @@ static void cell_progress(char *s, unsigned short hex)
 	printk("*** %04x : %s\n", hex, s ? s : "");
 }
 
+static void __init cell_pcibios_fixup(void)
+{
+	struct pci_dev *dev = NULL;
+
+	for_each_pci_dev(dev)
+		pci_read_irq_line(dev);
+}
+
 static void __init cell_init_irq(void)
 {
 	iic_init_IRQ();
@@ -130,8 +138,6 @@ static void __init cell_init_early(void)
 
 	cell_init_iommu();
 
-	ppc64_interrupt_controller = IC_CELL_PIC;
-
 	DBG(" <- cell_init_early()\n");
 }
 
@@ -178,8 +184,7 @@ define_machine(cell) {
 	.check_legacy_ioport	= cell_check_legacy_ioport,
 	.progress		= cell_progress,
 	.init_IRQ       	= cell_init_irq,
-	.get_irq        	= iic_get_irq,
-
+	.pcibios_fixup		= cell_pcibios_fixup,
 #ifdef CONFIG_KEXEC
 	.machine_kexec		= default_machine_kexec,
 	.machine_kexec_prepare	= default_machine_kexec_prepare,

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

@@ -22,6 +22,7 @@
 
 #include <linux/interrupt.h>
 #include <linux/irq.h>
+#include <linux/ioport.h>
 
 #include <asm/pgtable.h>
 #include <asm/prom.h>
@@ -56,58 +57,67 @@ enum {
 	REISWAITEN	= 0x508, /* Reissue Wait Control*/
 };
 
-static void __iomem *spider_pics[4];
+#define SPIDER_CHIP_COUNT	4
+#define SPIDER_SRC_COUNT	64
+#define SPIDER_IRQ_INVALID	63
 
-static void __iomem *spider_get_pic(int irq)
-{
-	int node = irq / IIC_NODE_STRIDE;
-	irq %= IIC_NODE_STRIDE;
-
-	if (irq >= IIC_EXT_OFFSET &&
-	    irq < IIC_EXT_OFFSET + IIC_NUM_EXT &&
-	    spider_pics)
-		return spider_pics[node];
-	return NULL;
-}
+struct spider_pic {
+	struct irq_host		*host;
+	struct device_node	*of_node;
+	void __iomem		*regs;
+	unsigned int		node_id;
+};
+static struct spider_pic spider_pics[SPIDER_CHIP_COUNT];
 
-static int spider_get_nr(unsigned int irq)
+static struct spider_pic *spider_virq_to_pic(unsigned int virq)
 {
-	return (irq % IIC_NODE_STRIDE) - IIC_EXT_OFFSET;
+	return irq_map[virq].host->host_data;
 }
 
-static void __iomem *spider_get_irq_config(int irq)
+static void __iomem *spider_get_irq_config(struct spider_pic *pic,
+					   unsigned int src)
 {
-	void __iomem *pic;
-	pic = spider_get_pic(irq);
-	return pic + TIR_CFGA + 8 * spider_get_nr(irq);
+	return pic->regs + TIR_CFGA + 8 * src;
 }
 
-static void spider_unmask_irq(unsigned int irq)
+static void spider_unmask_irq(unsigned int virq)
 {
-	int nodeid = (irq / IIC_NODE_STRIDE) * 0x10;
-	void __iomem *cfg = spider_get_irq_config(irq);
-	irq = spider_get_nr(irq);
+	struct spider_pic *pic = spider_virq_to_pic(virq);
+	void __iomem *cfg = spider_get_irq_config(pic, irq_map[virq].hwirq);
 
-	/* FIXME: Most of that is configuration and has nothing to do with enabling/disable,
-	 * besides, it's also partially bogus.
+	/* 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) & ~0xf0)| 0x3107000eu | nodeid);
-	out_be32(cfg + 4, in_be32(cfg + 4) | 0x00020000u | irq);
+	out_be32(cfg, in_be32(cfg) | 0x30000000u);
 }
 
-static void spider_mask_irq(unsigned int irq)
+static void spider_mask_irq(unsigned int virq)
 {
-	void __iomem *cfg = spider_get_irq_config(irq);
-	irq = spider_get_nr(irq);
+	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);
 }
 
-static void spider_ack_irq(unsigned int irq)
+static void spider_ack_irq(unsigned int virq)
 {
-	/* Should reset edge detection logic but we don't configure any edge interrupt
-	 * at the moment.
+	struct spider_pic *pic = spider_virq_to_pic(virq);
+	unsigned int src = irq_map[virq].hwirq;
+
+	/* Reset edge detection logic if necessary
 	 */
+	if (get_irq_desc(virq)->status & IRQ_LEVEL)
+		return;
+
+	/* Only interrupts 47 to 50 can be set to edge */
+	if (src < 47 || src > 50)
+		return;
+
+	/* Perform the clear of the edge logic */
+	out_be32(pic->regs + TIR_EDC, 0x100 | (src & 0xf));
 }
 
 static struct irq_chip spider_pic = {
@@ -117,102 +127,243 @@ static struct irq_chip spider_pic = {
 	.ack = spider_ack_irq,
 };
 
-static int spider_get_irq(int node)
+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)
 {
-	unsigned long cs;
-	void __iomem *regs = spider_pics[node];
+	unsigned int sense = flags & IRQ_TYPE_SENSE_MASK;
+	struct spider_pic *pic = h->host_data;
+	void __iomem *cfg = spider_get_irq_config(pic, hw);
+	int level = 0;
+	u32 ic;
+
+	/* Note that only level high is supported for most interrupts */
+	if (sense != IRQ_TYPE_NONE && sense != IRQ_TYPE_LEVEL_HIGH &&
+	    (hw < 47 || hw > 50))
+		return -EINVAL;
+
+	/* Decode sense type */
+	switch(sense) {
+	case IRQ_TYPE_EDGE_RISING:
+		ic = 0x3;
+		break;
+	case IRQ_TYPE_EDGE_FALLING:
+		ic = 0x2;
+		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;
+	}
 
-	cs = in_be32(regs + TIR_CS) >> 24;
+	/* 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);
+	out_be32(cfg + 4, (0x2 << 16) | (hw & 0xff));
 
-	if (cs == 63)
-		return -1;
-	else
-		return cs;
+	if (level)
+		get_irq_desc(virq)->status |= IRQ_LEVEL;
+	set_irq_chip_and_handler(virq, &spider_pic, handle_level_irq);
+	return 0;
+}
+
+static int spider_host_xlate(struct irq_host *h, struct device_node *ct,
+			   u32 *intspec, unsigned int intsize,
+			   irq_hw_number_t *out_hwirq, unsigned int *out_flags)
+
+{
+	/* Spider interrupts have 2 cells, first is the interrupt source,
+	 * second, well, I don't know for sure yet ... We mask the top bits
+	 * because old device-trees encode a node number in there
+	 */
+	*out_hwirq = intspec[0] & 0x3f;
+	*out_flags = IRQ_TYPE_LEVEL_HIGH;
+	return 0;
 }
 
+static struct irq_host_ops spider_host_ops = {
+	.match = spider_host_match,
+	.map = spider_host_map,
+	.xlate = spider_host_xlate,
+};
+
 static void spider_irq_cascade(unsigned int irq, struct irq_desc *desc,
 			       struct pt_regs *regs)
 {
-	int node = (int)(long)desc->handler_data;
-	int cascade_irq;
+	struct spider_pic *pic = desc->handler_data;
+	unsigned int cs, virq;
 
-	cascade_irq = spider_get_irq(node);
-	generic_handle_irq(cascade_irq, regs);
+	cs = in_be32(pic->regs + TIR_CS) >> 24;
+	if (cs == SPIDER_IRQ_INVALID)
+		virq = NO_IRQ;
+	else
+		virq = irq_linear_revmap(pic->host, cs);
+	if (virq != NO_IRQ)
+		generic_handle_irq(virq, regs);
 	desc->chip->eoi(irq);
 }
 
-/* hardcoded part to be compatible with older firmware */
+/* For hooking up the cascace we have a problem. Our device-tree is
+ * crap and we don't know on which BE iic interrupt we are hooked on at
+ * least not the "standard" way. We can reconstitute it based on two
+ * informations though: which BE node we are connected to and wether
+ * we are connected to IOIF0 or IOIF1. Right now, we really only care
+ * about the IBM cell blade and we know that its firmware gives us an
+ * interrupt-map property which is pretty strange.
+ */
+static unsigned int __init spider_find_cascade_and_node(struct spider_pic *pic)
+{
+	unsigned int virq;
+	u32 *imap, *tmp;
+	int imaplen, intsize, unit;
+	struct device_node *iic;
+	struct irq_host *iic_host;
+
+#if 0 /* Enable that when we have a way to retreive the node as well */
+	/* First, we check wether we have a real "interrupts" in the device
+	 * tree in case the device-tree is ever fixed
+	 */
+	struct of_irq oirq;
+	if (of_irq_map_one(pic->of_node, 0, &oirq) == 0) {
+		virq = irq_create_of_mapping(oirq.controller, oirq.specifier,
+					     oirq.size);
+		goto bail;
+	}
+#endif
+
+	/* Now do the horrible hacks */
+	tmp = (u32 *)get_property(pic->of_node, "#interrupt-cells", NULL);
+	if (tmp == NULL)
+		return NO_IRQ;
+	intsize = *tmp;
+	imap = (u32 *)get_property(pic->of_node, "interrupt-map", &imaplen);
+	if (imap == NULL || imaplen < (intsize + 1))
+		return NO_IRQ;
+	iic = of_find_node_by_phandle(imap[intsize]);
+	if (iic == NULL)
+		return NO_IRQ;
+	imap += intsize + 1;
+	tmp = (u32 *)get_property(iic, "#interrupt-cells", NULL);
+	if (tmp == NULL)
+		return NO_IRQ;
+	intsize = *tmp;
+	/* Assume unit is last entry of interrupt specifier */
+	unit = imap[intsize - 1];
+	/* Ok, we have a unit, now let's try to get the node */
+	tmp = (u32 *)get_property(iic, "ibm,interrupt-server-ranges", NULL);
+	if (tmp == NULL) {
+		of_node_put(iic);
+		return NO_IRQ;
+	}
+	/* ugly as hell but works for now */
+	pic->node_id = (*tmp) >> 1;
+	of_node_put(iic);
+
+	/* Ok, now let's get cracking. You may ask me why I just didn't match
+	 * the iic host from the iic OF node, but that way I'm still compatible
+	 * with really really old old firmwares for which we don't have a node
+	 */
+	iic_host = iic_get_irq_host(pic->node_id);
+	if (iic_host == NULL)
+		return NO_IRQ;
+	/* Manufacture an IIC interrupt number of class 2 */
+	virq = irq_create_mapping(iic_host, 0x20 | unit, 0);
+	if (virq == NO_IRQ)
+		printk(KERN_ERR "spider_pic: failed to map cascade !");
+	return virq;
+}
+
 
-static void __init spider_init_one(int node, unsigned long addr)
+static void __init spider_init_one(struct device_node *of_node, int chip,
+				   unsigned long addr)
 {
-	int n, irq;
+	struct spider_pic *pic = &spider_pics[chip];
+	int i, virq;
 
-	spider_pics[node] = ioremap(addr, 0x800);
-	if (spider_pics[node] == NULL)
+	/* Map registers */
+	pic->regs = ioremap(addr, 0x1000);
+	if (pic->regs == NULL)
 		panic("spider_pic: can't map registers !");
 
-	printk(KERN_INFO "spider_pic: mapped for node %d, addr: 0x%lx mapped to %p\n",
-	       node, addr, spider_pics[node]);
+	/* Allocate a host */
+	pic->host = irq_alloc_host(IRQ_HOST_MAP_LINEAR, SPIDER_SRC_COUNT,
+				   &spider_host_ops, SPIDER_IRQ_INVALID);
+	if (pic->host == NULL)
+		panic("spider_pic: can't allocate irq host !");
+	pic->host->host_data = pic;
 
-	for (n = 0; n < IIC_NUM_EXT; n++) {
-		if (n == IIC_EXT_CASCADE)
-			continue;
-		irq = n + IIC_EXT_OFFSET + node * IIC_NODE_STRIDE;
-		set_irq_chip_and_handler(irq, &spider_pic, handle_level_irq);
-		get_irq_desc(irq)->status |= IRQ_LEVEL;
+	/* Fill out other bits */
+	pic->of_node = of_node_get(of_node);
+
+	/* Go through all sources and disable them */
+	for (i = 0; i < SPIDER_SRC_COUNT; i++) {
+		void __iomem *cfg = pic->regs + TIR_CFGA + 8 * i;
+		out_be32(cfg, in_be32(cfg) & ~0x30000000u);
 	}
 
 	/* do not mask any interrupts because of level */
-	out_be32(spider_pics[node] + TIR_MSK, 0x0);
-
-	/* disable edge detection clear */
-	/* out_be32(spider_pics[node] + TIR_EDC, 0x0); */
+	out_be32(pic->regs + TIR_MSK, 0x0);
 
 	/* enable interrupt packets to be output */
-	out_be32(spider_pics[node] + TIR_PIEN,
-		 in_be32(spider_pics[node] + TIR_PIEN) | 0x1);
+	out_be32(pic->regs + TIR_PIEN, in_be32(pic->regs + TIR_PIEN) | 0x1);
 
-	/* Hook up cascade */
-	irq = IIC_EXT_CASCADE + node * IIC_NODE_STRIDE;
-	set_irq_data(irq, (void *)(long)node);
-	set_irq_chained_handler(irq, spider_irq_cascade);
+	/* Hook up the cascade interrupt to the iic and nodeid */
+	virq = spider_find_cascade_and_node(pic);
+	if (virq == NO_IRQ)
+		return;
+	set_irq_data(virq, pic);
+	set_irq_chained_handler(virq, spider_irq_cascade);
+
+	printk(KERN_INFO "spider_pic: node %d, addr: 0x%lx %s\n",
+	       pic->node_id, addr, of_node->full_name);
 
 	/* Enable the interrupt detection enable bit. Do this last! */
-	out_be32(spider_pics[node] + TIR_DEN,
-		 in_be32(spider_pics[node] + TIR_DEN) | 0x1);
+	out_be32(pic->regs + TIR_DEN, in_be32(pic->regs + TIR_DEN) | 0x1);
 }
 
 void __init spider_init_IRQ(void)
 {
-	unsigned long *spider_reg;
+	struct resource r;
 	struct device_node *dn;
-	char *compatible;
-	int node = 0;
-
-	/* XXX node numbers are totally bogus. We _hope_ we get the device nodes in the right
-	 * order here but that's definitely not guaranteed, we need to get the node from the
-	 * device tree instead. There is currently no proper property for it (but our whole
-	 * device-tree is bogus anyway) so all we can do is pray or maybe test the address
-	 * and deduce the node-id
+	int chip = 0;
+
+	/* XXX node numbers are totally bogus. We _hope_ we get the device
+	 * nodes in the right order here but that's definitely not guaranteed,
+	 * we need to get the node from the device tree instead.
+	 * There is currently no proper property for it (but our whole
+	 * device-tree is bogus anyway) so all we can do is pray or maybe test
+	 * the address and deduce the node-id
 	 */
-	for (dn = NULL; (dn = of_find_node_by_name(dn, "interrupt-controller"));) {
-		compatible = (char *)get_property(dn, "compatible", NULL);
-
-		if (!compatible)
-			continue;
-
- 		if (strstr(compatible, "CBEA,platform-spider-pic"))
-			spider_reg = (unsigned long *)get_property(dn, "reg", NULL);
-		else if (strstr(compatible, "sti,platform-spider-pic") && (node < 2)) {
-			static long hard_coded_pics[] = { 0x24000008000, 0x34000008000 };
-			spider_reg = &hard_coded_pics[node];
+	for (dn = NULL;
+	     (dn = of_find_node_by_name(dn, "interrupt-controller"));) {
+		if (device_is_compatible(dn, "CBEA,platform-spider-pic")) {
+			if (of_address_to_resource(dn, 0, &r)) {
+				printk(KERN_WARNING "spider-pic: Failed\n");
+				continue;
+			}
+		} else if (device_is_compatible(dn, "sti,platform-spider-pic")
+			   && (chip < 2)) {
+			static long hard_coded_pics[] =
+				{ 0x24000008000, 0x34000008000 };
+			r.start = hard_coded_pics[chip];
 		} else
 			continue;
-
-		if (spider_reg == NULL)
-			printk(KERN_ERR "spider_pic: No address for node %d\n", node);
-
-		spider_init_one(node, *spider_reg);
-		node++;
+		spider_init_one(dn, chip++, r.start);
 	}
 }

arch/powerpc/platforms/cell/spu_base.c

@@ -264,51 +264,57 @@ spu_irq_class_2(int irq, void *data, struct pt_regs *regs)
 	return stat ? IRQ_HANDLED : IRQ_NONE;
 }
 
-static int
-spu_request_irqs(struct spu *spu)
+static int spu_request_irqs(struct spu *spu)
 {
-	int ret;
-	int irq_base;
-
-	irq_base = IIC_NODE_STRIDE * spu->node + IIC_SPE_OFFSET;
-
-	snprintf(spu->irq_c0, sizeof (spu->irq_c0), "spe%02d.0", spu->number);
-	ret = request_irq(irq_base + spu->isrc,
-		 spu_irq_class_0, IRQF_DISABLED, spu->irq_c0, spu);
-	if (ret)
-		goto out;
-
-	snprintf(spu->irq_c1, sizeof (spu->irq_c1), "spe%02d.1", spu->number);
-	ret = request_irq(irq_base + IIC_CLASS_STRIDE + spu->isrc,
-		 spu_irq_class_1, IRQF_DISABLED, spu->irq_c1, spu);
-	if (ret)
-		goto out1;
+	int ret = 0;
 
-	snprintf(spu->irq_c2, sizeof (spu->irq_c2), "spe%02d.2", spu->number);
-	ret = request_irq(irq_base + 2*IIC_CLASS_STRIDE + spu->isrc,
-		 spu_irq_class_2, IRQF_DISABLED, spu->irq_c2, spu);
-	if (ret)
-		goto out2;
-	goto out;
+	if (spu->irqs[0] != NO_IRQ) {
+		snprintf(spu->irq_c0, sizeof (spu->irq_c0), "spe%02d.0",
+			 spu->number);
+		ret = request_irq(spu->irqs[0], spu_irq_class_0,
+				  IRQF_DISABLED,
+				  spu->irq_c0, spu);
+		if (ret)
+			goto bail0;
+	}
+	if (spu->irqs[1] != NO_IRQ) {
+		snprintf(spu->irq_c1, sizeof (spu->irq_c1), "spe%02d.1",
+			 spu->number);
+		ret = request_irq(spu->irqs[1], spu_irq_class_1,
+				  IRQF_DISABLED,
+				  spu->irq_c1, spu);
+		if (ret)
+			goto bail1;
+	}
+	if (spu->irqs[2] != NO_IRQ) {
+		snprintf(spu->irq_c2, sizeof (spu->irq_c2), "spe%02d.2",
+			 spu->number);
+		ret = request_irq(spu->irqs[2], spu_irq_class_2,
+				  IRQF_DISABLED,
+				  spu->irq_c2, spu);
+		if (ret)
+			goto bail2;
+	}
+	return 0;
 
-out2:
-	free_irq(irq_base + IIC_CLASS_STRIDE + spu->isrc, spu);
-out1:
-	free_irq(irq_base + spu->isrc, spu);
-out:
+bail2:
+	if (spu->irqs[1] != NO_IRQ)
+		free_irq(spu->irqs[1], spu);
+bail1:
+	if (spu->irqs[0] != NO_IRQ)
+		free_irq(spu->irqs[0], spu);
+bail0:
 	return ret;
 }
 
-static void
-spu_free_irqs(struct spu *spu)
+static void spu_free_irqs(struct spu *spu)
 {
-	int irq_base;
-
-	irq_base = IIC_NODE_STRIDE * spu->node + IIC_SPE_OFFSET;
-
-	free_irq(irq_base + spu->isrc, spu);
-	free_irq(irq_base + IIC_CLASS_STRIDE + spu->isrc, spu);
-	free_irq(irq_base + 2*IIC_CLASS_STRIDE + spu->isrc, spu);
+	if (spu->irqs[0] != NO_IRQ)
+		free_irq(spu->irqs[0], spu);
+	if (spu->irqs[1] != NO_IRQ)
+		free_irq(spu->irqs[1], spu);
+	if (spu->irqs[2] != NO_IRQ)
+		free_irq(spu->irqs[2], spu);
 }
 
 static LIST_HEAD(spu_list);
@@ -559,17 +565,38 @@ static void spu_unmap(struct spu *spu)
 	iounmap((u8 __iomem *)spu->local_store);
 }
 
+/* This function shall be abstracted for HV platforms */
+static int __init spu_map_interrupts(struct spu *spu, struct device_node *np)
+{
+	struct irq_host *host;
+	unsigned int isrc;
+	u32 *tmp;
+
+	host = iic_get_irq_host(spu->node);
+	if (host == NULL)
+		return -ENODEV;
+
+	/* Get the interrupt source from the device-tree */
+	tmp = (u32 *)get_property(np, "isrc", NULL);
+	if (!tmp)
+		return -ENODEV;
+	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);
+
+	/* Right now, we only fail if class 2 failed */
+	return spu->irqs[2] == NO_IRQ ? -EINVAL : 0;
+}
+
 static int __init spu_map_device(struct spu *spu, struct device_node *node)
 {
 	char *prop;
 	int ret;
 
 	ret = -ENODEV;
-	prop = get_property(node, "isrc", NULL);
-	if (!prop)
-		goto out;
-	spu->isrc = *(unsigned int *)prop;
-
 	spu->name = get_property(node, "name", NULL);
 	if (!spu->name)
 		goto out;
@@ -636,7 +663,8 @@ static int spu_create_sysdev(struct spu *spu)
 		return ret;
 	}
 
-	sysdev_create_file(&spu->sysdev, &attr_isrc);
+	if (spu->isrc != 0)
+		sysdev_create_file(&spu->sysdev, &attr_isrc);
 	sysfs_add_device_to_node(&spu->sysdev, spu->nid);
 
 	return 0;
@@ -668,6 +696,9 @@ static int __init create_spu(struct device_node *spe)
 	spu->nid = of_node_to_nid(spe);
 	if (spu->nid == -1)
 		spu->nid = 0;
+	ret = spu_map_interrupts(spu, spe);
+	if (ret)
+		goto out_unmap;
 	spin_lock_init(&spu->register_lock);
 	spu_mfc_sdr_set(spu, mfspr(SPRN_SDR1));
 	spu_mfc_sr1_set(spu, 0x33);

arch/powerpc/platforms/chrp/pci.c

@@ -18,7 +18,6 @@
 #include <asm/machdep.h>
 #include <asm/sections.h>
 #include <asm/pci-bridge.h>
-#include <asm/open_pic.h>
 #include <asm/grackle.h>
 #include <asm/rtas.h>
 
@@ -161,15 +160,9 @@ void __init
 chrp_pcibios_fixup(void)
 {
 	struct pci_dev *dev = NULL;
-	struct device_node *np;
 
-	/* PCI interrupts are controlled by the OpenPIC */
-	for_each_pci_dev(dev) {
-		np = pci_device_to_OF_node(dev);
-		if ((np != 0) && (np->n_intrs > 0) && (np->intrs[0].line != 0))
-			dev->irq = np->intrs[0].line;
-		pci_write_config_byte(dev, PCI_INTERRUPT_LINE, dev->irq);
-	}
+	for_each_pci_dev(dev)
+		pci_read_irq_line(dev);
 }
 
 #define PRG_CL_RESET_VALID 0x00010000

arch/powerpc/platforms/chrp/setup.c

@@ -59,7 +59,7 @@ void rtas_indicator_progress(char *, unsigned short);
 int _chrp_type;
 EXPORT_SYMBOL(_chrp_type);
 
-struct mpic *chrp_mpic;
+static struct mpic *chrp_mpic;
 
 /* Used for doing CHRP event-scans */
 DEFINE_PER_CPU(struct timer_list, heartbeat_timer);
@@ -315,19 +315,13 @@ chrp_event_scan(unsigned long unused)
 		  jiffies + event_scan_interval);
 }
 
-void chrp_8259_cascade(unsigned int irq, struct irq_desc *desc,
-			struct pt_regs *regs)
+static void chrp_8259_cascade(unsigned int irq, struct irq_desc *desc,
+			      struct pt_regs *regs)
 {
-	unsigned int max = 100;
-
-	while(max--) {
-		int irq = i8259_irq(regs);
-		if (max == 99)
-			desc->chip->eoi(irq);
-		if (irq < 0)
-			break;
-		generic_handle_irq(irq, regs);
-	};
+	unsigned int cascade_irq = i8259_irq(regs);
+	if (cascade_irq != NO_IRQ)
+		generic_handle_irq(cascade_irq, regs);
+	desc->chip->eoi(irq);
 }
 
 /*
@@ -336,18 +330,17 @@ void chrp_8259_cascade(unsigned int irq, struct irq_desc *desc,
 static void __init chrp_find_openpic(void)
 {
 	struct device_node *np, *root;
-	int len, i, j, irq_count;
+	int len, i, j;
 	int isu_size, idu_size;
 	unsigned int *iranges, *opprop = NULL;
 	int oplen = 0;
 	unsigned long opaddr;
 	int na = 1;
-	unsigned char init_senses[NR_IRQS - NUM_8259_INTERRUPTS];
 
-	np = find_type_devices("open-pic");
+	np = of_find_node_by_type(NULL, "open-pic");
 	if (np == NULL)
 		return;
-	root = find_path_device("/");
+	root = of_find_node_by_path("/");
 	if (root) {
 		opprop = (unsigned int *) get_property
 			(root, "platform-open-pic", &oplen);
@@ -358,19 +351,15 @@ static void __init chrp_find_openpic(void)
 		oplen /= na * sizeof(unsigned int);
 	} else {
 		struct resource r;
-		if (of_address_to_resource(np, 0, &r))
-			return;
+		if (of_address_to_resource(np, 0, &r)) {
+			goto bail;
+		}
 		opaddr = r.start;
 		oplen = 0;
 	}
 
 	printk(KERN_INFO "OpenPIC at %lx\n", opaddr);
 
-	irq_count = NR_IRQS - NUM_ISA_INTERRUPTS - 4; /* leave room for IPIs */
-	prom_get_irq_senses(init_senses, NUM_ISA_INTERRUPTS, NR_IRQS - 4);
-	/* i8259 cascade is always positive level */
-	init_senses[0] = IRQ_SENSE_LEVEL | IRQ_POLARITY_POSITIVE;
-
 	iranges = (unsigned int *) get_property(np, "interrupt-ranges", &len);
 	if (iranges == NULL)
 		len = 0;	/* non-distributed mpic */
@@ -397,15 +386,12 @@ static void __init chrp_find_openpic(void)
 	if (len > 1)
 		isu_size = iranges[3];
 
-	chrp_mpic = mpic_alloc(opaddr, MPIC_PRIMARY,
-			       isu_size, NUM_ISA_INTERRUPTS, irq_count,
-			       NR_IRQS - 4, init_senses, irq_count,
-			       " MPIC    ");
+	chrp_mpic = mpic_alloc(np, opaddr, MPIC_PRIMARY,
+			       isu_size, 0, " MPIC    ");
 	if (chrp_mpic == NULL) {
 		printk(KERN_ERR "Failed to allocate MPIC structure\n");
-		return;
+		goto bail;
 	}
-
 	j = na - 1;
 	for (i = 1; i < len; ++i) {
 		iranges += 2;
@@ -417,7 +403,10 @@ static void __init chrp_find_openpic(void)
 	}
 
 	mpic_init(chrp_mpic);
-	set_irq_chained_handler(NUM_ISA_INTERRUPTS, chrp_8259_cascade);
+	ppc_md.get_irq = mpic_get_irq;
+ bail:
+	of_node_put(root);
+	of_node_put(np);
 }
 
 #if defined(CONFIG_VT) && defined(CONFIG_INPUT_ADBHID) && defined(XMON)
@@ -428,14 +417,34 @@ static struct irqaction xmon_irqaction = {
 };
 #endif
 
-void __init chrp_init_IRQ(void)
+static void __init chrp_find_8259(void)
 {
-	struct device_node *np;
+	struct device_node *np, *pic = NULL;
 	unsigned long chrp_int_ack = 0;
-#if defined(CONFIG_VT) && defined(CONFIG_INPUT_ADBHID) && defined(XMON)
-	struct device_node *kbd;
-#endif
+	unsigned int cascade_irq;
 
+	/* Look for cascade */
+	for_each_node_by_type(np, "interrupt-controller")
+		if (device_is_compatible(np, "chrp,iic")) {
+			pic = np;
+			break;
+		}
+	/* Ok, 8259 wasn't found. We need to handle the case where
+	 * we have a pegasos that claims to be chrp but doesn't have
+	 * a proper interrupt tree
+	 */
+	if (pic == NULL && chrp_mpic != NULL) {
+		printk(KERN_ERR "i8259: Not found in device-tree"
+		       " assuming no legacy interrupts\n");
+		return;
+	}
+
+	/* Look for intack. In a perfect world, we would look for it on
+	 * the ISA bus that holds the 8259 but heh... Works that way. If
+	 * we ever see a problem, we can try to re-use the pSeries code here.
+	 * Also, Pegasos-type platforms don't have a proper node to start
+	 * from anyway
+	 */
 	for (np = find_devices("pci"); np != NULL; np = np->next) {
 		unsigned int *addrp = (unsigned int *)
 			get_property(np, "8259-interrupt-acknowledge", NULL);
@@ -446,11 +455,29 @@ void __init chrp_init_IRQ(void)
 		break;
 	}
 	if (np == NULL)
-		printk(KERN_ERR "Cannot find PCI interrupt acknowledge address\n");
+		printk(KERN_WARNING "Cannot find PCI interrupt acknowledge"
+		       " address, polling\n");
+
+	i8259_init(pic, chrp_int_ack);
+	if (ppc_md.get_irq == NULL)
+		ppc_md.get_irq = i8259_irq;
+	if (chrp_mpic != NULL) {
+		cascade_irq = irq_of_parse_and_map(pic, 0);
+		if (cascade_irq == NO_IRQ)
+			printk(KERN_ERR "i8259: failed to map cascade irq\n");
+		else
+			set_irq_chained_handler(cascade_irq,
+						chrp_8259_cascade);
+	}
+}
 
+void __init chrp_init_IRQ(void)
+{
+#if defined(CONFIG_VT) && defined(CONFIG_INPUT_ADBHID) && defined(XMON)
+	struct device_node *kbd;
+#endif
 	chrp_find_openpic();
-
-	i8259_init(chrp_int_ack, 0);
+	chrp_find_8259();
 
 	if (_chrp_type == _CHRP_Pegasos)
 		ppc_md.get_irq        = i8259_irq;
@@ -535,10 +562,6 @@ static int __init chrp_probe(void)
 	DMA_MODE_READ = 0x44;
 	DMA_MODE_WRITE = 0x48;
 	isa_io_base = CHRP_ISA_IO_BASE;		/* default value */
-	ppc_do_canonicalize_irqs = 1;
-
-	/* Assume we have an 8259... */
-	__irq_offset_value = NUM_ISA_INTERRUPTS;
 
 	return 1;
 }
@@ -550,7 +573,6 @@ define_machine(chrp) {
 	.init			= chrp_init2,
 	.show_cpuinfo		= chrp_show_cpuinfo,
 	.init_IRQ		= chrp_init_IRQ,
-	.get_irq		= mpic_get_irq,
 	.pcibios_fixup		= chrp_pcibios_fixup,
 	.restart		= rtas_restart,
 	.power_off		= rtas_power_off,

arch/powerpc/platforms/chrp/smp.c

@@ -29,7 +29,6 @@
 #include <asm/smp.h>
 #include <asm/residual.h>
 #include <asm/time.h>
-#include <asm/open_pic.h>
 #include <asm/machdep.h>
 #include <asm/smp.h>
 #include <asm/mpic.h>

arch/powerpc/platforms/iseries/irq.c

@@ -162,27 +162,6 @@ static void pci_event_handler(struct HvLpEvent *event, struct pt_regs *regs)
 		printk(KERN_ERR "pci_event_handler: NULL event received\n");
 }
 
-/*
- * This is called by init_IRQ.  set in ppc_md.init_IRQ by iSeries_setup.c
- * It must be called before the bus walk.
- */
-void __init iSeries_init_IRQ(void)
-{
-	/* Register PCI event handler and open an event path */
-	int ret;
-
-	ret = HvLpEvent_registerHandler(HvLpEvent_Type_PciIo,
-			&pci_event_handler);
-	if (ret == 0) {
-		ret = HvLpEvent_openPath(HvLpEvent_Type_PciIo, 0);
-		if (ret != 0)
-			printk(KERN_ERR "iseries_init_IRQ: open event path "
-					"failed with rc 0x%x\n", ret);
-	} else
-		printk(KERN_ERR "iseries_init_IRQ: register handler "
-				"failed with rc 0x%x\n", ret);
-}
-
 #define REAL_IRQ_TO_SUBBUS(irq)	(((irq) >> 14) & 0xff)
 #define REAL_IRQ_TO_BUS(irq)	((((irq) >> 6) & 0xff) + 1)
 #define REAL_IRQ_TO_IDSEL(irq)	((((irq) >> 3) & 7) + 1)
@@ -196,7 +175,7 @@ static void iseries_enable_IRQ(unsigned int irq)
 {
 	u32 bus, dev_id, function, mask;
 	const u32 sub_bus = 0;
-	unsigned int rirq = virt_irq_to_real_map[irq];
+	unsigned int rirq = (unsigned int)irq_map[irq].hwirq;
 
 	/* The IRQ has already been locked by the caller */
 	bus = REAL_IRQ_TO_BUS(rirq);
@@ -213,7 +192,7 @@ static unsigned int iseries_startup_IRQ(unsigned int irq)
 {
 	u32 bus, dev_id, function, mask;
 	const u32 sub_bus = 0;
-	unsigned int rirq = virt_irq_to_real_map[irq];
+	unsigned int rirq = (unsigned int)irq_map[irq].hwirq;
 
 	bus = REAL_IRQ_TO_BUS(rirq);
 	function = REAL_IRQ_TO_FUNC(rirq);
@@ -254,7 +233,7 @@ static void iseries_shutdown_IRQ(unsigned int irq)
 {
 	u32 bus, dev_id, function, mask;
 	const u32 sub_bus = 0;
-	unsigned int rirq = virt_irq_to_real_map[irq];
+	unsigned int rirq = (unsigned int)irq_map[irq].hwirq;
 
 	/* irq should be locked by the caller */
 	bus = REAL_IRQ_TO_BUS(rirq);
@@ -277,7 +256,7 @@ static void iseries_disable_IRQ(unsigned int irq)
 {
 	u32 bus, dev_id, function, mask;
 	const u32 sub_bus = 0;
-	unsigned int rirq = virt_irq_to_real_map[irq];
+	unsigned int rirq = (unsigned int)irq_map[irq].hwirq;
 
 	/* The IRQ has already been locked by the caller */
 	bus = REAL_IRQ_TO_BUS(rirq);
@@ -291,7 +270,7 @@ static void iseries_disable_IRQ(unsigned int irq)
 
 static void iseries_end_IRQ(unsigned int irq)
 {
-	unsigned int rirq = virt_irq_to_real_map[irq];
+	unsigned int rirq = (unsigned int)irq_map[irq].hwirq;
 
 	HvCallPci_eoi(REAL_IRQ_TO_BUS(rirq), REAL_IRQ_TO_SUBBUS(rirq),
 		(REAL_IRQ_TO_IDSEL(rirq) << 4) + REAL_IRQ_TO_FUNC(rirq));
@@ -314,16 +293,14 @@ static struct irq_chip iseries_pic = {
 int __init iSeries_allocate_IRQ(HvBusNumber bus,
 		HvSubBusNumber sub_bus, u32 bsubbus)
 {
-	int virtirq;
 	unsigned int realirq;
 	u8 idsel = ISERIES_GET_DEVICE_FROM_SUBBUS(bsubbus);
 	u8 function = ISERIES_GET_FUNCTION_FROM_SUBBUS(bsubbus);
 
 	realirq = (((((sub_bus << 8) + (bus - 1)) << 3) + (idsel - 1)) << 3)
 		+ function;
-	virtirq = virt_irq_create_mapping(realirq);
-	set_irq_chip_and_handler(virtirq, &iseries_pic, handle_fasteoi_irq);
-	return virtirq;
+
+	return irq_create_mapping(NULL, realirq, IRQ_TYPE_NONE);
 }
 
 #endif /* CONFIG_PCI */
@@ -331,10 +308,9 @@ int __init iSeries_allocate_IRQ(HvBusNumber bus,
 /*
  * Get the next pending IRQ.
  */
-int iSeries_get_irq(struct pt_regs *regs)
+unsigned int iSeries_get_irq(struct pt_regs *regs)
 {
-	/* -2 means ignore this interrupt */
-	int irq = -2;
+	int irq = NO_IRQ_IGNORE;
 
 #ifdef CONFIG_SMP
 	if (get_lppaca()->int_dword.fields.ipi_cnt) {
@@ -357,9 +333,57 @@ int iSeries_get_irq(struct pt_regs *regs)
 		}
 		spin_unlock(&pending_irqs_lock);
 		if (irq >= NR_IRQS)
-			irq = -2;
+			irq = NO_IRQ_IGNORE;
 	}
 #endif
 
 	return irq;
 }
+
+static int iseries_irq_host_map(struct irq_host *h, unsigned int virq,
+				irq_hw_number_t hw, unsigned int flags)
+{
+	set_irq_chip_and_handler(virq, &iseries_pic, handle_fasteoi_irq);
+
+	return 0;
+}
+
+static struct irq_host_ops iseries_irq_host_ops = {
+	.map = iseries_irq_host_map,
+};
+
+/*
+ * This is called by init_IRQ.  set in ppc_md.init_IRQ by iSeries_setup.c
+ * It must be called before the bus walk.
+ */
+void __init iSeries_init_IRQ(void)
+{
+	/* Register PCI event handler and open an event path */
+	struct irq_host *host;
+	int ret;
+
+	/*
+	 * The Hypervisor only allows us up to 256 interrupt
+	 * sources (the irq number is passed in a u8).
+	 */
+	irq_set_virq_count(256);
+
+	/* Create irq host. No need for a revmap since HV will give us
+	 * back our virtual irq number
+	 */
+	host = irq_alloc_host(IRQ_HOST_MAP_NOMAP, 0, &iseries_irq_host_ops, 0);
+	BUG_ON(host == NULL);
+	irq_set_default_host(host);
+
+	ret = HvLpEvent_registerHandler(HvLpEvent_Type_PciIo,
+			&pci_event_handler);
+	if (ret == 0) {
+		ret = HvLpEvent_openPath(HvLpEvent_Type_PciIo, 0);
+		if (ret != 0)
+			printk(KERN_ERR "iseries_init_IRQ: open event path "
+					"failed with rc 0x%x\n", ret);
+	} else
+		printk(KERN_ERR "iseries_init_IRQ: register handler "
+				"failed with rc 0x%x\n", ret);
+}
+

arch/powerpc/platforms/iseries/irq.h

@@ -4,6 +4,6 @@
 extern void iSeries_init_IRQ(void);
 extern int  iSeries_allocate_IRQ(HvBusNumber, HvSubBusNumber, u32);
 extern void iSeries_activate_IRQs(void);
-extern int iSeries_get_irq(struct pt_regs *);
+extern unsigned int iSeries_get_irq(struct pt_regs *);
 
 #endif /* _ISERIES_IRQ_H */

arch/powerpc/platforms/iseries/setup.c

@@ -294,8 +294,6 @@ static void __init iSeries_init_early(void)
 {
 	DBG(" -> iSeries_init_early()\n");
 
-	ppc64_interrupt_controller = IC_ISERIES;
-
 #if defined(CONFIG_BLK_DEV_INITRD)
 	/*
 	 * If the init RAM disk has been configured and there is
@@ -659,12 +657,6 @@ static int __init iseries_probe(void)
 	powerpc_firmware_features |= FW_FEATURE_ISERIES;
 	powerpc_firmware_features |= FW_FEATURE_LPAR;
 
-	/*
-	 * The Hypervisor only allows us up to 256 interrupt
-	 * sources (the irq number is passed in a u8).
-	 */
-	virt_irq_max = 255;
-
 	hpte_init_iSeries();
 
 	return 1;

arch/powerpc/platforms/maple/pci.c

@@ -443,18 +443,23 @@ void __init maple_pci_init(void)
 int maple_pci_get_legacy_ide_irq(struct pci_dev *pdev, int channel)
 {
 	struct device_node *np;
-	int irq = channel ? 15 : 14;
+	unsigned int defirq = channel ? 15 : 14;
+	unsigned int irq;
 
 	if (pdev->vendor != PCI_VENDOR_ID_AMD ||
 	    pdev->device != PCI_DEVICE_ID_AMD_8111_IDE)
-		return irq;
+		return defirq;
 
 	np = pci_device_to_OF_node(pdev);
 	if (np == NULL)
-		return irq;
-	if (np->n_intrs < 2)
-		return irq;
-	return np->intrs[channel & 0x1].line;
+		return defirq;
+	irq = irq_of_parse_and_map(np, channel & 0x1);
+	if (irq == NO_IRQ) {
+		printk("Failed to map onboard IDE interrupt for channel %d\n",
+		       channel);
+		return defirq;
+	}
+	return irq;
 }
 
 /* XXX: To remove once all firmwares are ok */

arch/powerpc/platforms/maple/setup.c

@@ -198,50 +198,81 @@ static void __init maple_init_early(void)
 {
 	DBG(" -> maple_init_early\n");
 
-	/* Setup interrupt mapping options */
-	ppc64_interrupt_controller = IC_OPEN_PIC;
-
 	iommu_init_early_dart();
 
 	DBG(" <- maple_init_early\n");
 }
 
-
-static __init void maple_init_IRQ(void)
+/*
+ * This is almost identical to pSeries and CHRP. We need to make that
+ * code generic at one point, with appropriate bits in the device-tree to
+ * identify the presence of an HT APIC
+ */
+static void __init maple_init_IRQ(void)
 {
-	struct device_node *root;
+	struct device_node *root, *np, *mpic_node = NULL;
 	unsigned int *opprop;
-	unsigned long opic_addr;
+	unsigned long openpic_addr = 0;
+	int naddr, n, i, opplen, has_isus = 0;
 	struct mpic *mpic;
-	unsigned char senses[128];
-	int n;
+	unsigned int flags = MPIC_PRIMARY;
 
-	DBG(" -> maple_init_IRQ\n");
+	/* Locate MPIC in the device-tree. Note that there is a bug
+	 * in Maple device-tree where the type of the controller is
+	 * open-pic and not interrupt-controller
+	 */
+	for_each_node_by_type(np, "open-pic") {
+		mpic_node = np;
+		break;
+	}
+	if (mpic_node == NULL) {
+		printk(KERN_ERR
+		       "Failed to locate the MPIC interrupt controller\n");
+		return;
+	}
 
-	/* XXX: Non standard, replace that with a proper openpic/mpic node
-	 * in the device-tree. Find the Open PIC if present */
+	/* Find address list in /platform-open-pic */
 	root = of_find_node_by_path("/");
-	opprop = (unsigned int *) get_property(root,
-				"platform-open-pic", NULL);
-	if (opprop == 0)
-		panic("OpenPIC not found !\n");
-
-	n = prom_n_addr_cells(root);
-	for (opic_addr = 0; n > 0; --n)
-		opic_addr = (opic_addr << 32) + *opprop++;
+	naddr = prom_n_addr_cells(root);
+	opprop = (unsigned int *) get_property(root, "platform-open-pic",
+					       &opplen);
+	if (opprop != 0) {
+		openpic_addr = of_read_number(opprop, naddr);
+		has_isus = (opplen > naddr);
+		printk(KERN_DEBUG "OpenPIC addr: %lx, has ISUs: %d\n",
+		       openpic_addr, has_isus);
+	}
 	of_node_put(root);
 
-	/* Obtain sense values from device-tree */
-	prom_get_irq_senses(senses, 0, 128);
+	BUG_ON(openpic_addr == 0);
 
-	mpic = mpic_alloc(opic_addr,
-			  MPIC_PRIMARY | MPIC_BIG_ENDIAN |
-			  MPIC_BROKEN_U3 | MPIC_WANTS_RESET,
-			  0, 0, 128, 128, senses, 128, "U3-MPIC");
+	/* Check for a big endian MPIC */
+	if (get_property(np, "big-endian", NULL) != NULL)
+		flags |= MPIC_BIG_ENDIAN;
+
+	/* XXX Maple specific bits */
+	flags |= MPIC_BROKEN_U3 | MPIC_WANTS_RESET;
+
+	/* Setup the openpic driver. More device-tree junks, we hard code no
+	 * ISUs for now. I'll have to revisit some stuffs with the folks doing
+	 * the firmware for those
+	 */
+	mpic = mpic_alloc(mpic_node, openpic_addr, flags,
+			  /*has_isus ? 16 :*/ 0, 0, " MPIC     ");
 	BUG_ON(mpic == NULL);
-	mpic_init(mpic);
 
-	DBG(" <- maple_init_IRQ\n");
+	/* Add ISUs */
+	opplen /= sizeof(u32);
+	for (n = 0, i = naddr; i < opplen; i += naddr, n++) {
+		unsigned long isuaddr = of_read_number(opprop + i, naddr);
+		mpic_assign_isu(mpic, n, isuaddr);
+	}
+
+	/* All ISUs are setup, complete initialization */
+	mpic_init(mpic);
+	ppc_md.get_irq = mpic_get_irq;
+	of_node_put(mpic_node);
+	of_node_put(root);
 }
 
 static void __init maple_progress(char *s, unsigned short hex)
@@ -279,7 +310,6 @@ define_machine(maple_md) {
 	.setup_arch		= maple_setup_arch,
 	.init_early		= maple_init_early,
 	.init_IRQ		= maple_init_IRQ,
-	.get_irq		= mpic_get_irq,
 	.pcibios_fixup		= maple_pcibios_fixup,
 	.pci_get_legacy_ide_irq	= maple_pci_get_legacy_ide_irq,
 	.restart		= maple_restart,

arch/powerpc/platforms/powermac/bootx_init.c

@@ -162,6 +162,8 @@ static void __init bootx_add_chosen_props(unsigned long base,
 {
 	u32 val;
 
+	bootx_dt_add_prop("linux,bootx", NULL, 0, mem_end);
+
 	if (bootx_info->kernelParamsOffset) {
 		char *args = (char *)((unsigned long)bootx_info) +
 			bootx_info->kernelParamsOffset;
@@ -228,7 +230,7 @@ static void __init bootx_scan_dt_build_strings(unsigned long base,
 
 	if (!strcmp(namep, "/chosen")) {
 		DBG(" detected /chosen ! adding properties names !\n");
-		bootx_dt_add_string("linux,platform", mem_end);
+		bootx_dt_add_string("linux,bootx", mem_end);
 		bootx_dt_add_string("linux,stdout-path", mem_end);
 		bootx_dt_add_string("linux,initrd-start", mem_end);
 		bootx_dt_add_string("linux,initrd-end", mem_end);

arch/powerpc/platforms/powermac/low_i2c.c

@@ -522,10 +522,11 @@ static struct pmac_i2c_host_kw *__init kw_i2c_host_init(struct device_node *np)
 		host->speed = KW_I2C_MODE_25KHZ;
 		break;
 	}	
-	if (np->n_intrs > 0)
-		host->irq = np->intrs[0].line;
-	else
-		host->irq = NO_IRQ;
+	host->irq = irq_of_parse_and_map(np, 0);
+	if (host->irq == NO_IRQ)
+		printk(KERN_WARNING
+		       "low_i2c: Failed to map interrupt for %s\n",
+		       np->full_name);
 
 	host->base = ioremap((*addrp), 0x1000);
 	if (host->base == NULL) {

arch/powerpc/platforms/powermac/nvram.c

@@ -29,6 +29,8 @@
 #include <asm/machdep.h>
 #include <asm/nvram.h>
 
+#include "pmac.h"
+
 #define DEBUG
 
 #ifdef DEBUG
@@ -80,9 +82,6 @@ static int nvram_partitions[3];
 // XXX Turn that into a sem
 static DEFINE_SPINLOCK(nv_lock);
 
-extern int pmac_newworld;
-extern int system_running;
-
 static int (*core99_write_bank)(int bank, u8* datas);
 static int (*core99_erase_bank)(int bank);
 

arch/powerpc/platforms/powermac/pci.c

@@ -46,6 +46,9 @@ 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 */
 
 extern u8 pci_cache_line_size;
@@ -647,6 +650,33 @@ static void __init init_p2pbridge(void)
 	early_write_config_word(hose, bus, devfn, PCI_BRIDGE_CONTROL, val);
 }
 
+static void __init init_second_ohare(void)
+{
+	struct device_node *np = of_find_node_by_name(NULL, "pci106b,7");
+	unsigned char bus, devfn;
+	unsigned short cmd;
+
+	if (np == NULL)
+		return;
+
+	/* This must run before we initialize the PICs since the second
+	 * ohare hosts a PIC that will be accessed there.
+	 */
+	if (pci_device_from_OF_node(np, &bus, &devfn) == 0) {
+		struct pci_controller* hose =
+			pci_find_hose_for_OF_device(np);
+		if (!hose) {
+			printk(KERN_ERR "Can't find PCI hose for OHare2 !\n");
+			return;
+		}
+		early_read_config_word(hose, bus, devfn, PCI_COMMAND, &cmd);
+		cmd |= PCI_COMMAND_MEMORY | PCI_COMMAND_MASTER;
+		cmd &= ~PCI_COMMAND_IO;
+		early_write_config_word(hose, bus, devfn, PCI_COMMAND, cmd);
+	}
+	has_second_ohare = 1;
+}
+
 /*
  * Some Apple desktop machines have a NEC PD720100A USB2 controller
  * on the motherboard. Open Firmware, on these, will disable the
@@ -688,9 +718,6 @@ static void __init fixup_nec_usb2(void)
 			       " EHCI, fixing up...\n");
 			data &= ~1UL;
 			early_write_config_dword(hose, bus, devfn, 0xe4, data);
-			early_write_config_byte(hose, bus,
-						devfn | 2, PCI_INTERRUPT_LINE,
-				nec->intrs[0].line);
 		}
 	}
 }
@@ -958,32 +985,28 @@ static int __init add_bridge(struct device_node *dev)
 	return 0;
 }
 
-static void __init pcibios_fixup_OF_interrupts(void)
+void __init pmac_pcibios_fixup(void)
 {
 	struct pci_dev* dev = NULL;
 
-	/*
-	 * Open Firmware often doesn't initialize the
-	 * PCI_INTERRUPT_LINE config register properly, so we
-	 * should find the device node and apply the interrupt
-	 * obtained from the OF device-tree
-	 */
 	for_each_pci_dev(dev) {
-		struct device_node *node;
-		node = pci_device_to_OF_node(dev);
-		/* this is the node, see if it has interrupts */
-		if (node && node->n_intrs > 0)
-			dev->irq = node->intrs[0].line;
-		pci_write_config_byte(dev, PCI_INTERRUPT_LINE, dev->irq);
+		/* Read interrupt from the device-tree */
+		pci_read_irq_line(dev);
+
+		/* 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
+		 * the ethernet-only board but not the combo ethernet/modem
+		 * board). The real interrupt is 28 on the second controller
+		 * -> 28+32 = 60.
+		 */
+		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);
 	}
 }
 
-void __init pmac_pcibios_fixup(void)
-{
-	/* Fixup interrupts according to OF tree */
-	pcibios_fixup_OF_interrupts();
-}
-
 #ifdef CONFIG_PPC64
 static void __init pmac_fixup_phb_resources(void)
 {
@@ -1071,6 +1094,7 @@ void __init pmac_pci_init(void)
 
 #else /* CONFIG_PPC64 */
 	init_p2pbridge();
+	init_second_ohare();
 	fixup_nec_usb2();
 
 	/* We are still having some issues with the Xserve G4, enabling

arch/powerpc/platforms/powermac/pfunc_base.c

@@ -24,19 +24,18 @@ static irqreturn_t macio_gpio_irq(int irq, void *data, struct pt_regs *regs)
 
 static int macio_do_gpio_irq_enable(struct pmf_function *func)
 {
-	if (func->node->n_intrs < 1)
+	unsigned int irq = irq_of_parse_and_map(func->node, 0);
+	if (irq == NO_IRQ)
 		return -EINVAL;
-
-	return request_irq(func->node->intrs[0].line, macio_gpio_irq, 0,
-			   func->node->name, func);
+	return request_irq(irq, macio_gpio_irq, 0, func->node->name, func);
 }
 
 static int macio_do_gpio_irq_disable(struct pmf_function *func)
 {
-	if (func->node->n_intrs < 1)
+	unsigned int irq = irq_of_parse_and_map(func->node, 0);
+	if (irq == NO_IRQ)
 		return -EINVAL;
-
-	free_irq(func->node->intrs[0].line, func);
+	free_irq(irq, func);
 	return 0;
 }
 

arch/powerpc/platforms/powermac/pic.c

@@ -65,13 +65,11 @@ static u32 level_mask[4];
 
 static DEFINE_SPINLOCK(pmac_pic_lock);
 
-#define GATWICK_IRQ_POOL_SIZE        10
-static struct interrupt_info gatwick_int_pool[GATWICK_IRQ_POOL_SIZE];
-
 #define NR_MASK_WORDS	((NR_IRQS + 31) / 32)
 static unsigned long ppc_lost_interrupts[NR_MASK_WORDS];
 static unsigned long ppc_cached_irq_mask[NR_MASK_WORDS];
 static int pmac_irq_cascade = -1;
+static struct irq_host *pmac_pic_host;
 
 static void __pmac_retrigger(unsigned int irq_nr)
 {
@@ -86,18 +84,16 @@ static void __pmac_retrigger(unsigned int irq_nr)
 	}
 }
 
-static void pmac_mask_and_ack_irq(unsigned int irq_nr)
+static void pmac_mask_and_ack_irq(unsigned int virq)
 {
-        unsigned long bit = 1UL << (irq_nr & 0x1f);
-        int i = irq_nr >> 5;
+	unsigned int src = irq_map[virq].hwirq;
+        unsigned long bit = 1UL << (virq & 0x1f);
+        int i = virq >> 5;
         unsigned long flags;
 
-        if ((unsigned)irq_nr >= max_irqs)
-                return;
-
 	spin_lock_irqsave(&pmac_pic_lock, flags);
-        __clear_bit(irq_nr, ppc_cached_irq_mask);
-        if (__test_and_clear_bit(irq_nr, ppc_lost_interrupts))
+        __clear_bit(src, ppc_cached_irq_mask);
+        if (__test_and_clear_bit(src, ppc_lost_interrupts))
                 atomic_dec(&ppc_n_lost_interrupts);
         out_le32(&pmac_irq_hw[i]->enable, ppc_cached_irq_mask[i]);
         out_le32(&pmac_irq_hw[i]->ack, bit);
@@ -110,17 +106,15 @@ static void pmac_mask_and_ack_irq(unsigned int irq_nr)
 	spin_unlock_irqrestore(&pmac_pic_lock, flags);
 }
 
-static void pmac_ack_irq(unsigned int irq_nr)
+static void pmac_ack_irq(unsigned int virq)
 {
-        unsigned long bit = 1UL << (irq_nr & 0x1f);
-        int i = irq_nr >> 5;
+	unsigned int src = irq_map[virq].hwirq;
+        unsigned long bit = 1UL << (src & 0x1f);
+        int i = src >> 5;
         unsigned long flags;
 
-        if ((unsigned)irq_nr >= max_irqs)
-                return;
-
   	spin_lock_irqsave(&pmac_pic_lock, flags);
-	if (__test_and_clear_bit(irq_nr, ppc_lost_interrupts))
+	if (__test_and_clear_bit(src, ppc_lost_interrupts))
                 atomic_dec(&ppc_n_lost_interrupts);
         out_le32(&pmac_irq_hw[i]->ack, bit);
         (void)in_le32(&pmac_irq_hw[i]->ack);
@@ -157,48 +151,51 @@ static void __pmac_set_irq_mask(unsigned int irq_nr, int nokicklost)
 /* When an irq gets requested for the first client, if it's an
  * edge interrupt, we clear any previous one on the controller
  */
-static unsigned int pmac_startup_irq(unsigned int irq_nr)
+static unsigned int pmac_startup_irq(unsigned int virq)
 {
 	unsigned long flags;
-        unsigned long bit = 1UL << (irq_nr & 0x1f);
-        int i = irq_nr >> 5;
+	unsigned int src = irq_map[virq].hwirq;
+        unsigned long bit = 1UL << (src & 0x1f);
+        int i = src >> 5;
 
   	spin_lock_irqsave(&pmac_pic_lock, flags);
-	if ((irq_desc[irq_nr].status & IRQ_LEVEL) == 0)
+	if ((irq_desc[virq].status & IRQ_LEVEL) == 0)
 		out_le32(&pmac_irq_hw[i]->ack, bit);
-        __set_bit(irq_nr, ppc_cached_irq_mask);
-        __pmac_set_irq_mask(irq_nr, 0);
+        __set_bit(src, ppc_cached_irq_mask);
+        __pmac_set_irq_mask(src, 0);
   	spin_unlock_irqrestore(&pmac_pic_lock, flags);
 
 	return 0;
 }
 
-static void pmac_mask_irq(unsigned int irq_nr)
+static void pmac_mask_irq(unsigned int virq)
 {
 	unsigned long flags;
+	unsigned int src = irq_map[virq].hwirq;
 
   	spin_lock_irqsave(&pmac_pic_lock, flags);
-        __clear_bit(irq_nr, ppc_cached_irq_mask);
-        __pmac_set_irq_mask(irq_nr, 0);
+        __clear_bit(src, ppc_cached_irq_mask);
+        __pmac_set_irq_mask(src, 0);
   	spin_unlock_irqrestore(&pmac_pic_lock, flags);
 }
 
-static void pmac_unmask_irq(unsigned int irq_nr)
+static void pmac_unmask_irq(unsigned int virq)
 {
 	unsigned long flags;
+	unsigned int src = irq_map[virq].hwirq;
 
 	spin_lock_irqsave(&pmac_pic_lock, flags);
-	__set_bit(irq_nr, ppc_cached_irq_mask);
-        __pmac_set_irq_mask(irq_nr, 0);
+	__set_bit(src, ppc_cached_irq_mask);
+        __pmac_set_irq_mask(src, 0);
   	spin_unlock_irqrestore(&pmac_pic_lock, flags);
 }
 
-static int pmac_retrigger(unsigned int irq_nr)
+static int pmac_retrigger(unsigned int virq)
 {
 	unsigned long flags;
 
   	spin_lock_irqsave(&pmac_pic_lock, flags);
-	__pmac_retrigger(irq_nr);
+	__pmac_retrigger(irq_map[virq].hwirq);
   	spin_unlock_irqrestore(&pmac_pic_lock, flags);
 	return 1;
 }
@@ -238,7 +235,7 @@ static irqreturn_t gatwick_action(int cpl, void *dev_id, struct pt_regs *regs)
 	return rc;
 }
 
-static int pmac_get_irq(struct pt_regs *regs)
+static unsigned int pmac_pic_get_irq(struct pt_regs *regs)
 {
 	int irq;
 	unsigned long bits = 0;
@@ -250,7 +247,7 @@ static int pmac_get_irq(struct pt_regs *regs)
        	/* IPI's are a hack on the powersurge -- Cort */
        	if ( smp_processor_id() != 0 ) {
 		psurge_smp_message_recv(regs);
-		return -2;	/* ignore, already handled */
+		return NO_IRQ_IGNORE;	/* ignore, already handled */
         }
 #endif /* CONFIG_SMP */
   	spin_lock_irqsave(&pmac_pic_lock, flags);
@@ -266,133 +263,9 @@ static int pmac_get_irq(struct pt_regs *regs)
 		break;
 	}
   	spin_unlock_irqrestore(&pmac_pic_lock, flags);
-
-	return irq;
-}
-
-/* This routine will fix some missing interrupt values in the device tree
- * on the gatwick mac-io controller used by some PowerBooks
- *
- * Walking of OF nodes could use a bit more fixing up here, but it's not
- * very important as this is all boot time code on static portions of the
- * device-tree.
- *
- * However, the modifications done to "intrs" will have to be removed and
- * replaced with proper updates of the "interrupts" properties or
- * AAPL,interrupts, yet to be decided, once the dynamic parsing is there.
- */
-static void __init pmac_fix_gatwick_interrupts(struct device_node *gw,
-					       int irq_base)
-{
-	struct device_node *node;
-	int count;
-
-	memset(gatwick_int_pool, 0, sizeof(gatwick_int_pool));
-	count = 0;
-	for (node = NULL; (node = of_get_next_child(gw, node)) != NULL;) {
-		/* Fix SCC */
-		if ((strcasecmp(node->name, "escc") == 0) && node->child) {
-			if (node->child->n_intrs < 3) {
-				node->child->intrs = &gatwick_int_pool[count];
-				count += 3;
-			}
-			node->child->n_intrs = 3;
-			node->child->intrs[0].line = 15+irq_base;
-			node->child->intrs[1].line =  4+irq_base;
-			node->child->intrs[2].line =  5+irq_base;
-			printk(KERN_INFO "irq: fixed SCC on gatwick"
-			       " (%d,%d,%d)\n",
-			       node->child->intrs[0].line,
-			       node->child->intrs[1].line,
-			       node->child->intrs[2].line);
-		}
-		/* Fix media-bay & left SWIM */
-		if (strcasecmp(node->name, "media-bay") == 0) {
-			struct device_node* ya_node;
-
-			if (node->n_intrs == 0)
-				node->intrs = &gatwick_int_pool[count++];
-			node->n_intrs = 1;
-			node->intrs[0].line = 29+irq_base;
-			printk(KERN_INFO "irq: fixed media-bay on gatwick"
-			       " (%d)\n", node->intrs[0].line);
-
-			ya_node = node->child;
-			while(ya_node) {
-				if (strcasecmp(ya_node->name, "floppy") == 0) {
-					if (ya_node->n_intrs < 2) {
-						ya_node->intrs = &gatwick_int_pool[count];
-						count += 2;
-					}
-					ya_node->n_intrs = 2;
-					ya_node->intrs[0].line = 19+irq_base;
-					ya_node->intrs[1].line =  1+irq_base;
-					printk(KERN_INFO "irq: fixed floppy on second controller (%d,%d)\n",
-						ya_node->intrs[0].line, ya_node->intrs[1].line);
-				}
-				if (strcasecmp(ya_node->name, "ata4") == 0) {
-					if (ya_node->n_intrs < 2) {
-						ya_node->intrs = &gatwick_int_pool[count];
-						count += 2;
-					}
-					ya_node->n_intrs = 2;
-					ya_node->intrs[0].line = 14+irq_base;
-					ya_node->intrs[1].line =  3+irq_base;
-					printk(KERN_INFO "irq: fixed ide on second controller (%d,%d)\n",
-						ya_node->intrs[0].line, ya_node->intrs[1].line);
-				}
-				ya_node = ya_node->sibling;
-			}
-		}
-	}
-	if (count > 10) {
-		printk("WARNING !! Gatwick interrupt pool overflow\n");
-		printk("  GATWICK_IRQ_POOL_SIZE = %d\n", GATWICK_IRQ_POOL_SIZE);
-		printk("              requested = %d\n", count);
-	}
-}
-
-/*
- * The PowerBook 3400/2400/3500 can have a combo ethernet/modem
- * card which includes an ohare chip that acts as a second interrupt
- * controller.  If we find this second ohare, set it up and fix the
- * interrupt value in the device tree for the ethernet chip.
- */
-static void __init enable_second_ohare(struct device_node *np)
-{
-	unsigned char bus, devfn;
-	unsigned short cmd;
-	struct device_node *ether;
-
-	/* This code doesn't strictly belong here, it could be part of
-	 * either the PCI initialisation or the feature code. It's kept
-	 * here for historical reasons.
-	 */
-	if (pci_device_from_OF_node(np, &bus, &devfn) == 0) {
-		struct pci_controller* hose =
-			pci_find_hose_for_OF_device(np);
-		if (!hose) {
-			printk(KERN_ERR "Can't find PCI hose for OHare2 !\n");
-			return;
-		}
-		early_read_config_word(hose, bus, devfn, PCI_COMMAND, &cmd);
-		cmd |= PCI_COMMAND_MEMORY | PCI_COMMAND_MASTER;
-		cmd &= ~PCI_COMMAND_IO;
-		early_write_config_word(hose, bus, devfn, PCI_COMMAND, cmd);
-	}
-
-	/* Fix interrupt for the modem/ethernet combo controller. The number
-	 * in the device tree (27) is bogus (correct for the ethernet-only
-	 * board but not the combo ethernet/modem board).
-	 * The real interrupt is 28 on the second controller -> 28+32 = 60.
-	 */
-	ether = of_find_node_by_name(NULL, "pci1011,14");
-	if (ether && ether->n_intrs > 0) {
-		ether->intrs[0].line = 60;
-		printk(KERN_INFO "irq: Fixed ethernet IRQ to %d\n",
-		       ether->intrs[0].line);
-	}
-	of_node_put(ether);
+	if (unlikely(irq < 0))
+		return NO_IRQ;
+	return irq_linear_revmap(pmac_pic_host, irq);
 }
 
 #ifdef CONFIG_XMON
@@ -411,6 +284,50 @@ static struct irqaction gatwick_cascade_action = {
 	.name		= "cascade",
 };
 
+static int pmac_pic_host_match(struct irq_host *h, struct device_node *node)
+{
+	/* We match all, we don't always have a node anyway */
+	return 1;
+}
+
+static int pmac_pic_host_map(struct irq_host *h, unsigned int virq,
+			     irq_hw_number_t hw, unsigned int flags)
+{
+	struct irq_desc *desc = get_irq_desc(virq);
+	int level;
+
+	if (hw >= max_irqs)
+		return -EINVAL;
+
+	/* Mark level interrupts, set delayed disable for edge ones and set
+	 * handlers
+	 */
+	level = !!(level_mask[hw >> 5] & (1UL << (hw & 0x1f)));
+	if (level)
+		desc->status |= IRQ_LEVEL;
+	else
+		desc->status |= IRQ_DELAYED_DISABLE;
+	set_irq_chip_and_handler(virq, &pmac_pic, level ?
+				 handle_level_irq : handle_edge_irq);
+	return 0;
+}
+
+static int pmac_pic_host_xlate(struct irq_host *h, struct device_node *ct,
+			       u32 *intspec, unsigned int intsize,
+			       irq_hw_number_t *out_hwirq,
+			       unsigned int *out_flags)
+
+{
+	*out_hwirq = *intspec;
+	return 0;
+}
+
+static struct irq_host_ops pmac_pic_host_ops = {
+	.match = pmac_pic_host_match,
+	.map = pmac_pic_host_map,
+	.xlate = pmac_pic_host_xlate,
+};
+
 static void __init pmac_pic_probe_oldstyle(void)
 {
         int i;
@@ -420,7 +337,7 @@ static void __init pmac_pic_probe_oldstyle(void)
 	struct resource r;
 
 	/* Set our get_irq function */
-	ppc_md.get_irq = pmac_get_irq;
+	ppc_md.get_irq = pmac_pic_get_irq;
 
 	/*
 	 * Find the interrupt controller type & node
@@ -438,7 +355,6 @@ static void __init pmac_pic_probe_oldstyle(void)
 		if (slave) {
 			max_irqs = 64;
 			level_mask[1] = OHARE_LEVEL_MASK;
-			enable_second_ohare(slave);
 		}
 	} else if ((master = of_find_node_by_name(NULL, "mac-io")) != NULL) {
 		max_irqs = max_real_irqs = 64;
@@ -462,21 +378,18 @@ static void __init pmac_pic_probe_oldstyle(void)
 			max_irqs = 128;
 			level_mask[2] = HEATHROW_LEVEL_MASK;
 			level_mask[3] = 0;
-			pmac_fix_gatwick_interrupts(slave, max_real_irqs);
 		}
 	}
 	BUG_ON(master == NULL);
 
-	/* Mark level interrupts and set handlers */
-	for (i = 0; i < max_irqs; i++) {
-		int level = !!(level_mask[i >> 5] & (1UL << (i & 0x1f)));
-		if (level)
-			irq_desc[i].status |= IRQ_LEVEL;
-		else
-			irq_desc[i].status |= IRQ_DELAYED_DISABLE;
-		set_irq_chip_and_handler(i, &pmac_pic, level ?
-					 handle_level_irq : handle_edge_irq);
-	}
+	/*
+	 * Allocate an irq host
+	 */
+	pmac_pic_host = irq_alloc_host(IRQ_HOST_MAP_LINEAR, max_irqs,
+				       &pmac_pic_host_ops,
+				       max_irqs);
+	BUG_ON(pmac_pic_host == NULL);
+	irq_set_default_host(pmac_pic_host);
 
 	/* Get addresses of first controller if we have a node for it */
 	BUG_ON(of_address_to_resource(master, 0, &r));
@@ -503,7 +416,7 @@ static void __init pmac_pic_probe_oldstyle(void)
 			pmac_irq_hw[i++] =
 				(volatile struct pmac_irq_hw __iomem *)
 				(addr + 0x10);
-		pmac_irq_cascade = slave->intrs[0].line;
+		pmac_irq_cascade = irq_of_parse_and_map(slave, 0);
 
 		printk(KERN_INFO "irq: Found slave Apple PIC %s for %d irqs"
 		       " cascade: %d\n", slave->full_name,
@@ -516,12 +429,12 @@ static void __init pmac_pic_probe_oldstyle(void)
 		out_le32(&pmac_irq_hw[i]->enable, 0);
 
 	/* Hookup cascade irq */
-	if (slave)
+	if (slave && pmac_irq_cascade != NO_IRQ)
 		setup_irq(pmac_irq_cascade, &gatwick_cascade_action);
 
 	printk(KERN_INFO "irq: System has %d possible interrupts\n", max_irqs);
 #ifdef CONFIG_XMON
-	setup_irq(20, &xmon_action);
+	setup_irq(irq_create_mapping(NULL, 20, 0), &xmon_action);
 #endif
 }
 #endif /* CONFIG_PPC32 */
@@ -530,16 +443,11 @@ static void pmac_u3_cascade(unsigned int irq, struct irq_desc *desc,
 			    struct pt_regs *regs)
 {
 	struct mpic *mpic = desc->handler_data;
-	unsigned int max = 100;
 
-	while(max--) {
-		int cascade_irq = mpic_get_one_irq(mpic, regs);
-		if (max == 99)
-			desc->chip->eoi(irq);
-		if (irq < 0)
-			break;
+	unsigned int cascade_irq = mpic_get_one_irq(mpic, regs);
+	if (cascade_irq != NO_IRQ)
 		generic_handle_irq(cascade_irq, regs);
-	};
+	desc->chip->eoi(irq);
 }
 
 static void __init pmac_pic_setup_mpic_nmi(struct mpic *mpic)
@@ -549,21 +457,20 @@ static void __init pmac_pic_setup_mpic_nmi(struct mpic *mpic)
 	int nmi_irq;
 
 	pswitch = of_find_node_by_name(NULL, "programmer-switch");
-	if (pswitch && pswitch->n_intrs) {
-		nmi_irq = pswitch->intrs[0].line;
-		mpic_irq_set_priority(nmi_irq, 9);
-		setup_irq(nmi_irq, &xmon_action);
+	if (pswitch) {
+		nmi_irq = irq_of_parse_and_map(pswitch, 0);
+		if (nmi_irq != NO_IRQ) {
+			mpic_irq_set_priority(nmi_irq, 9);
+			setup_irq(nmi_irq, &xmon_action);
+		}
+		of_node_put(pswitch);
 	}
-	of_node_put(pswitch);
 #endif	/* defined(CONFIG_XMON) && defined(CONFIG_PPC32) */
 }
 
 static struct mpic * __init pmac_setup_one_mpic(struct device_node *np,
 						int master)
 {
-	unsigned char senses[128];
-	int offset = master ? 0 : 128;
-	int count = master ? 128 : 124;
 	const char *name = master ? " MPIC 1   " : " MPIC 2   ";
 	struct resource r;
 	struct mpic *mpic;
@@ -576,8 +483,6 @@ static struct mpic * __init pmac_setup_one_mpic(struct device_node *np,
 
 	pmac_call_feature(PMAC_FTR_ENABLE_MPIC, np, 0, 0);
 
-	prom_get_irq_senses(senses, offset, offset + count);
-
 	flags |= MPIC_WANTS_RESET;
 	if (get_property(np, "big-endian", NULL))
 		flags |= MPIC_BIG_ENDIAN;
@@ -588,8 +493,7 @@ static struct mpic * __init pmac_setup_one_mpic(struct device_node *np,
 	if (master && (flags & MPIC_BIG_ENDIAN))
 		flags |= MPIC_BROKEN_U3;
 
-	mpic = mpic_alloc(r.start, flags, 0, offset, count, master ? 252 : 0,
-			  senses, count, name);
+	mpic = mpic_alloc(np, r.start, flags, 0, 0, name);
 	if (mpic == NULL)
 		return NULL;
 
@@ -602,6 +506,7 @@ static int __init pmac_pic_probe_mpic(void)
 {
 	struct mpic *mpic1, *mpic2;
 	struct device_node *np, *master = NULL, *slave = NULL;
+	unsigned int cascade;
 
 	/* We can have up to 2 MPICs cascaded */
 	for (np = NULL; (np = of_find_node_by_type(np, "open-pic"))
@@ -638,8 +543,15 @@ static int __init pmac_pic_probe_mpic(void)
 	of_node_put(master);
 
 	/* No slave, let's go out */
-	if (slave == NULL || slave->n_intrs < 1)
+	if (slave == NULL)
+		return 0;
+
+	/* Get/Map slave interrupt */
+	cascade = irq_of_parse_and_map(slave, 0);
+	if (cascade == NO_IRQ) {
+		printk(KERN_ERR "Failed to map cascade IRQ\n");
 		return 0;
+	}
 
 	mpic2 = pmac_setup_one_mpic(slave, 0);
 	if (mpic2 == NULL) {
@@ -647,8 +559,8 @@ static int __init pmac_pic_probe_mpic(void)
 		of_node_put(slave);
 		return 0;
 	}
-	set_irq_data(slave->intrs[0].line, mpic2);
-	set_irq_chained_handler(slave->intrs[0].line, pmac_u3_cascade);
+	set_irq_data(cascade, mpic2);
+	set_irq_chained_handler(cascade, pmac_u3_cascade);
 
 	of_node_put(slave);
 	return 0;
@@ -657,6 +569,19 @@ static int __init pmac_pic_probe_mpic(void)
 
 void __init pmac_pic_init(void)
 {
+	unsigned int flags = 0;
+
+	/* We configure the OF parsing based on our oldworld vs. newworld
+	 * platform type and wether we were booted by BootX.
+	 */
+#ifdef CONFIG_PPC32
+	if (!pmac_newworld)
+		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 */
+
 	/* We first try to detect Apple's new Core99 chipset, since mac-io
 	 * is quite different on those machines and contains an IBM MPIC2.
 	 */
@@ -679,6 +604,7 @@ unsigned long sleep_save_mask[2];
 
 /* This used to be passed by the PMU driver but that link got
  * broken with the new driver model. We use this tweak for now...
+ * We really want to do things differently though...
  */
 static int pmacpic_find_viaint(void)
 {
@@ -692,7 +618,7 @@ static int pmacpic_find_viaint(void)
 	np = of_find_node_by_name(NULL, "via-pmu");
 	if (np == NULL)
 		goto not_found;
-	viaint = np->intrs[0].line;
+	viaint = irq_of_parse_and_map(np, 0);;
 #endif /* CONFIG_ADB_PMU */
 
 not_found:

arch/powerpc/platforms/powermac/pmac.h

@@ -12,6 +12,8 @@
 
 struct rtc_time;
 
+extern int pmac_newworld;
+
 extern long pmac_time_init(void);
 extern unsigned long pmac_get_boot_time(void);
 extern void pmac_get_rtc_time(struct rtc_time *);

arch/powerpc/platforms/powermac/setup.c

@@ -613,9 +613,6 @@ static void __init pmac_init_early(void)
 	udbg_adb_init(!!strstr(cmd_line, "btextdbg"));
 
 #ifdef CONFIG_PPC64
-	/* Setup interrupt mapping options */
-	ppc64_interrupt_controller = IC_OPEN_PIC;
-
 	iommu_init_early_dart();
 #endif
 }

arch/powerpc/platforms/pseries/ras.c

@@ -72,32 +72,62 @@ static irqreturn_t ras_error_interrupt(int irq, void *dev_id,
 
 /* #define DEBUG */
 
-static void request_ras_irqs(struct device_node *np, char *propname,
+
+static void request_ras_irqs(struct device_node *np,
 			irqreturn_t (*handler)(int, void *, struct pt_regs *),
 			const char *name)
 {
-	unsigned int *ireg, len, i;
-	int virq, n_intr;
-
-	ireg = (unsigned int *)get_property(np, propname, &len);
-	if (ireg == NULL)
-		return;
-	n_intr = prom_n_intr_cells(np);
-	len /= n_intr * sizeof(*ireg);
-
-	for (i = 0; i < len; i++) {
-		virq = virt_irq_create_mapping(*ireg);
-		if (virq == NO_IRQ) {
-			printk(KERN_ERR "Unable to allocate interrupt "
-			       "number for %s\n", np->full_name);
-			return;
+	int i, index, count = 0;
+	struct of_irq oirq;
+	u32 *opicprop;
+	unsigned int opicplen;
+	unsigned int virqs[16];
+
+	/* Check for obsolete "open-pic-interrupt" property. If present, then
+	 * map those interrupts using the default interrupt host and default
+	 * trigger
+	 */
+	opicprop = (u32 *)get_property(np, "open-pic-interrupt", &opicplen);
+	if (opicprop) {
+		opicplen /= sizeof(u32);
+		for (i = 0; i < opicplen; i++) {
+			if (count > 15)
+				break;
+			virqs[count] = irq_create_mapping(NULL, *(opicprop++),
+							 IRQ_TYPE_NONE);
+			if (virqs[count] == NO_IRQ)
+				printk(KERN_ERR "Unable to allocate interrupt "
+				       "number for %s\n", np->full_name);
+			else
+				count++;
+
 		}
-		if (request_irq(irq_offset_up(virq), handler, 0, name, NULL)) {
+	}
+	/* Else use normal interrupt tree parsing */
+	else {
+		/* First try to do a proper OF tree parsing */
+		for (index = 0; of_irq_map_one(np, index, &oirq) == 0;
+		     index++) {
+			if (count > 15)
+				break;
+			virqs[count] = irq_create_of_mapping(oirq.controller,
+							    oirq.specifier,
+							    oirq.size);
+			if (virqs[count] == NO_IRQ)
+				printk(KERN_ERR "Unable to allocate interrupt "
+				       "number for %s\n", np->full_name);
+			else
+				count++;
+		}
+	}
+
+	/* Now request them */
+	for (i = 0; i < count; i++) {
+		if (request_irq(virqs[i], handler, 0, name, NULL)) {
 			printk(KERN_ERR "Unable to request interrupt %d for "
-			       "%s\n", irq_offset_up(virq), np->full_name);
+			       "%s\n", virqs[i], np->full_name);
 			return;
 		}
-		ireg += n_intr;
 	}
 }
 
@@ -115,20 +145,14 @@ static int __init init_ras_IRQ(void)
 	/* Internal Errors */
 	np = of_find_node_by_path("/event-sources/internal-errors");
 	if (np != NULL) {
-		request_ras_irqs(np, "open-pic-interrupt", ras_error_interrupt,
-				 "RAS_ERROR");
-		request_ras_irqs(np, "interrupts", ras_error_interrupt,
-				 "RAS_ERROR");
+		request_ras_irqs(np, ras_error_interrupt, "RAS_ERROR");
 		of_node_put(np);
 	}
 
 	/* EPOW Events */
 	np = of_find_node_by_path("/event-sources/epow-events");
 	if (np != NULL) {
-		request_ras_irqs(np, "open-pic-interrupt", ras_epow_interrupt,
-				 "RAS_EPOW");
-		request_ras_irqs(np, "interrupts", ras_epow_interrupt,
-				 "RAS_EPOW");
+		request_ras_irqs(np, ras_epow_interrupt, "RAS_EPOW");
 		of_node_put(np);
 	}
 
@@ -162,7 +186,7 @@ ras_epow_interrupt(int irq, void *dev_id, struct pt_regs * regs)
 
 	status = rtas_call(ras_check_exception_token, 6, 1, NULL,
 			   RAS_VECTOR_OFFSET,
-			   virt_irq_to_real(irq_offset_down(irq)),
+			   irq_map[irq].hwirq,
 			   RTAS_EPOW_WARNING | RTAS_POWERMGM_EVENTS,
 			   critical, __pa(&ras_log_buf),
 				rtas_get_error_log_max());
@@ -198,7 +222,7 @@ ras_error_interrupt(int irq, void *dev_id, struct pt_regs * regs)
 
 	status = rtas_call(ras_check_exception_token, 6, 1, NULL,
 			   RAS_VECTOR_OFFSET,
-			   virt_irq_to_real(irq_offset_down(irq)),
+			   irq_map[irq].hwirq,
 			   RTAS_INTERNAL_ERROR, 1 /*Time Critical */,
 			   __pa(&ras_log_buf),
 				rtas_get_error_log_max());

arch/powerpc/platforms/pseries/setup.c

@@ -76,6 +76,9 @@
 #define DBG(fmt...)
 #endif
 
+/* move those away to a .h */
+extern void smp_init_pseries_mpic(void);
+extern void smp_init_pseries_xics(void);
 extern void find_udbg_vterm(void);
 
 int fwnmi_active;  /* TRUE if an FWNMI handler is present */
@@ -83,7 +86,7 @@ int fwnmi_active;  /* TRUE if an FWNMI handler is present */
 static void pseries_shared_idle_sleep(void);
 static void pseries_dedicated_idle_sleep(void);
 
-struct mpic *pSeries_mpic;
+static struct device_node *pSeries_mpic_node;
 
 static void pSeries_show_cpuinfo(struct seq_file *m)
 {
@@ -118,78 +121,92 @@ static void __init fwnmi_init(void)
 		fwnmi_active = 1;
 }
 
-void pSeries_8259_cascade(unsigned int irq, struct irq_desc *desc,
+void pseries_8259_cascade(unsigned int irq, struct irq_desc *desc,
 			  struct pt_regs *regs)
 {
-	unsigned int max = 100;
-
-	while(max--) {
-		int cascade_irq = i8259_irq(regs);
-		if (max == 99)
-			desc->chip->eoi(irq);
-		if (cascade_irq < 0)
-			break;
+	unsigned int cascade_irq = i8259_irq(regs);
+	if (cascade_irq != NO_IRQ)
 		generic_handle_irq(cascade_irq, regs);
-	};
+	desc->chip->eoi(irq);
 }
 
-static void __init pSeries_init_mpic(void)
+static void __init pseries_mpic_init_IRQ(void)
 {
+	struct device_node *np, *old, *cascade = NULL;
         unsigned int *addrp;
-	struct device_node *np;
 	unsigned long intack = 0;
-
-	/* All ISUs are setup, complete initialization */
-	mpic_init(pSeries_mpic);
-
-	/* Check what kind of cascade ACK we have */
-        if (!(np = of_find_node_by_name(NULL, "pci"))
-            || !(addrp = (unsigned int *)
-                 get_property(np, "8259-interrupt-acknowledge", NULL)))
-                printk(KERN_ERR "Cannot find pci to get ack address\n");
-        else
-		intack = addrp[prom_n_addr_cells(np)-1];
-	of_node_put(np);
-
-	/* Setup the legacy interrupts & controller */
-	i8259_init(intack, 0);
-
-	/* Hook cascade to mpic */
-	set_irq_chained_handler(NUM_ISA_INTERRUPTS, pSeries_8259_cascade);
-}
-
-static void __init pSeries_setup_mpic(void)
-{
 	unsigned int *opprop;
 	unsigned long openpic_addr = 0;
-        unsigned char senses[NR_IRQS - NUM_ISA_INTERRUPTS];
-        struct device_node *root;
-	int irq_count;
+	unsigned int cascade_irq;
+	int naddr, n, i, opplen;
+	struct mpic *mpic;
 
-	/* Find the Open PIC if present */
-	root = of_find_node_by_path("/");
-	opprop = (unsigned int *) get_property(root, "platform-open-pic", NULL);
+	np = of_find_node_by_path("/");
+	naddr = prom_n_addr_cells(np);
+	opprop = (unsigned int *) get_property(np, "platform-open-pic", &opplen);
 	if (opprop != 0) {
-		int n = prom_n_addr_cells(root);
-
-		for (openpic_addr = 0; n > 0; --n)
-			openpic_addr = (openpic_addr << 32) + *opprop++;
+		openpic_addr = of_read_number(opprop, naddr);
 		printk(KERN_DEBUG "OpenPIC addr: %lx\n", openpic_addr);
 	}
-	of_node_put(root);
+	of_node_put(np);
 
 	BUG_ON(openpic_addr == 0);
 
-	/* Get the sense values from OF */
-	prom_get_irq_senses(senses, NUM_ISA_INTERRUPTS, NR_IRQS);
-	
 	/* Setup the openpic driver */
-	irq_count = NR_IRQS - NUM_ISA_INTERRUPTS - 4; /* leave room for IPIs */
-	pSeries_mpic = mpic_alloc(openpic_addr, MPIC_PRIMARY,
-				  16, 16, irq_count, /* isu size, irq offset, irq count */ 
-				  NR_IRQS - 4, /* ipi offset */
-				  senses, irq_count, /* sense & sense size */
-				  " MPIC     ");
+	mpic = mpic_alloc(pSeries_mpic_node, openpic_addr,
+			  MPIC_PRIMARY,
+			  16, 250, /* isu size, irq count */
+			  " MPIC     ");
+	BUG_ON(mpic == NULL);
+
+	/* Add ISUs */
+	opplen /= sizeof(u32);
+	for (n = 0, i = naddr; i < opplen; i += naddr, n++) {
+		unsigned long isuaddr = of_read_number(opprop + i, naddr);
+		mpic_assign_isu(mpic, n, isuaddr);
+	}
+
+	/* All ISUs are setup, complete initialization */
+	mpic_init(mpic);
+
+	/* Look for cascade */
+	for_each_node_by_type(np, "interrupt-controller")
+		if (device_is_compatible(np, "chrp,iic")) {
+			cascade = np;
+			break;
+		}
+	if (cascade == NULL)
+		return;
+
+	cascade_irq = irq_of_parse_and_map(cascade, 0);
+	if (cascade == NO_IRQ) {
+		printk(KERN_ERR "xics: failed to map cascade interrupt");
+		return;
+	}
+
+	/* Check ACK type */
+	for (old = of_node_get(cascade); old != NULL ; old = np) {
+		np = of_get_parent(old);
+		of_node_put(old);
+		if (np == NULL)
+			break;
+		if (strcmp(np->name, "pci") != 0)
+			continue;
+		addrp = (u32 *)get_property(np, "8259-interrupt-acknowledge",
+					    NULL);
+		if (addrp == NULL)
+			continue;
+		naddr = prom_n_addr_cells(np);
+		intack = addrp[naddr-1];
+		if (naddr > 1)
+			intack |= ((unsigned long)addrp[naddr-2]) << 32;
+	}
+	if (intack)
+		printk(KERN_DEBUG "mpic: PCI 8259 intack at 0x%016lx\n",
+		       intack);
+	i8259_init(cascade, intack);
+	of_node_put(cascade);
+	set_irq_chained_handler(cascade_irq, pseries_8259_cascade);
 }
 
 static void pseries_lpar_enable_pmcs(void)
@@ -207,21 +224,67 @@ static void pseries_lpar_enable_pmcs(void)
 		get_lppaca()->pmcregs_in_use = 1;
 }
 
-static void __init pSeries_setup_arch(void)
+#ifdef CONFIG_KEXEC
+static void pseries_kexec_cpu_down_mpic(int crash_shutdown, int secondary)
 {
-	/* Fixup ppc_md depending on the type of interrupt controller */
-	if (ppc64_interrupt_controller == IC_OPEN_PIC) {
-		ppc_md.init_IRQ       = pSeries_init_mpic;
-		ppc_md.get_irq        = mpic_get_irq;
-		/* Allocate the mpic now, so that find_and_init_phbs() can
-		 * fill the ISUs */
-		pSeries_setup_mpic();
-	} else
-		ppc_md.init_IRQ       = xics_init_IRQ;
+	mpic_teardown_this_cpu(secondary);
+}
 
+static void pseries_kexec_cpu_down_xics(int crash_shutdown, int secondary)
+{
+	/* Don't risk a hypervisor call if we're crashing */
+	if (firmware_has_feature(FW_FEATURE_SPLPAR) && !crash_shutdown) {
+		unsigned long vpa = __pa(get_lppaca());
+
+		if (unregister_vpa(hard_smp_processor_id(), vpa)) {
+			printk("VPA deregistration of cpu %u (hw_cpu_id %d) "
+					"failed\n", smp_processor_id(),
+					hard_smp_processor_id());
+		}
+	}
+	xics_teardown_cpu(secondary);
+}
+#endif /* CONFIG_KEXEC */
+
+static void __init pseries_discover_pic(void)
+{
+	struct device_node *np;
+	char *typep;
+
+	for (np = NULL; (np = of_find_node_by_name(np,
+						   "interrupt-controller"));) {
+		typep = (char *)get_property(np, "compatible", NULL);
+		if (strstr(typep, "open-pic")) {
+			pSeries_mpic_node = of_node_get(np);
+			ppc_md.init_IRQ       = pseries_mpic_init_IRQ;
+			ppc_md.get_irq        = mpic_get_irq;
+#ifdef CONFIG_KEXEC
+			ppc_md.kexec_cpu_down = pseries_kexec_cpu_down_mpic;
+#endif
 #ifdef CONFIG_SMP
-	smp_init_pSeries();
+			smp_init_pseries_mpic();
 #endif
+			return;
+		} else if (strstr(typep, "ppc-xicp")) {
+			ppc_md.init_IRQ       = xics_init_IRQ;
+#ifdef CONFIG_KEXEC
+			ppc_md.kexec_cpu_down = pseries_kexec_cpu_down_xics;
+#endif
+#ifdef CONFIG_SMP
+			smp_init_pseries_xics();
+#endif
+			return;
+		}
+	}
+	printk(KERN_ERR "pSeries_discover_pic: failed to recognize"
+	       " interrupt-controller\n");
+}
+
+static void __init pSeries_setup_arch(void)
+{
+	/* Discover PIC type and setup ppc_md accordingly */
+	pseries_discover_pic();
+
 	/* openpic global configuration register (64-bit format). */
 	/* openpic Interrupt Source Unit pointer (64-bit format). */
 	/* python0 facility area (mmio) (64-bit format) REAL address. */
@@ -273,33 +336,6 @@ static int __init pSeries_init_panel(void)
 }
 arch_initcall(pSeries_init_panel);
 
-static  void __init pSeries_discover_pic(void)
-{
-	struct device_node *np;
-	char *typep;
-
-	/*
-	 * Setup interrupt mapping options that are needed for finish_device_tree
-	 * to properly parse the OF interrupt tree & do the virtual irq mapping
-	 */
-	__irq_offset_value = NUM_ISA_INTERRUPTS;
-	ppc64_interrupt_controller = IC_INVALID;
-	for (np = NULL; (np = of_find_node_by_name(np, "interrupt-controller"));) {
-		typep = (char *)get_property(np, "compatible", NULL);
-		if (strstr(typep, "open-pic")) {
-			ppc64_interrupt_controller = IC_OPEN_PIC;
-			break;
-		} else if (strstr(typep, "ppc-xicp")) {
-			ppc64_interrupt_controller = IC_PPC_XIC;
-			break;
-		}
-	}
-	if (ppc64_interrupt_controller == IC_INVALID)
-		printk("pSeries_discover_pic: failed to recognize"
-			" interrupt-controller\n");
-
-}
-
 static void pSeries_mach_cpu_die(void)
 {
 	local_irq_disable();
@@ -342,8 +378,6 @@ static void __init pSeries_init_early(void)
 
 	iommu_init_early_pSeries();
 
-	pSeries_discover_pic();
-
 	DBG(" <- pSeries_init_early()\n");
 }
 
@@ -515,27 +549,6 @@ static int pSeries_pci_probe_mode(struct pci_bus *bus)
 	return PCI_PROBE_NORMAL;
 }
 
-#ifdef CONFIG_KEXEC
-static void pseries_kexec_cpu_down(int crash_shutdown, int secondary)
-{
-	/* Don't risk a hypervisor call if we're crashing */
-	if (firmware_has_feature(FW_FEATURE_SPLPAR) && !crash_shutdown) {
-		unsigned long vpa = __pa(get_lppaca());
-
-		if (unregister_vpa(hard_smp_processor_id(), vpa)) {
-			printk("VPA deregistration of cpu %u (hw_cpu_id %d) "
-					"failed\n", smp_processor_id(),
-					hard_smp_processor_id());
-		}
-	}
-
-	if (ppc64_interrupt_controller == IC_OPEN_PIC)
-		mpic_teardown_this_cpu(secondary);
-	else
-		xics_teardown_cpu(secondary);
-}
-#endif
-
 define_machine(pseries) {
 	.name			= "pSeries",
 	.probe			= pSeries_probe,
@@ -560,7 +573,6 @@ define_machine(pseries) {
 	.system_reset_exception = pSeries_system_reset_exception,
 	.machine_check_exception = pSeries_machine_check_exception,
 #ifdef CONFIG_KEXEC
-	.kexec_cpu_down		= pseries_kexec_cpu_down,
 	.machine_kexec		= default_machine_kexec,
 	.machine_kexec_prepare	= default_machine_kexec_prepare,
 	.machine_crash_shutdown	= default_machine_crash_shutdown,

arch/powerpc/platforms/pseries/smp.c

@@ -416,27 +416,12 @@ static struct smp_ops_t pSeries_xics_smp_ops = {
 #endif
 
 /* This is called very early */
-void __init smp_init_pSeries(void)
+static void __init smp_init_pseries(void)
 {
 	int i;
 
 	DBG(" -> smp_init_pSeries()\n");
 
-	switch (ppc64_interrupt_controller) {
-#ifdef CONFIG_MPIC
-	case IC_OPEN_PIC:
-		smp_ops = &pSeries_mpic_smp_ops;
-		break;
-#endif
-#ifdef CONFIG_XICS
-	case IC_PPC_XIC:
-		smp_ops = &pSeries_xics_smp_ops;
-		break;
-#endif
-	default:
-		panic("Invalid interrupt controller");
-	}
-
 #ifdef CONFIG_HOTPLUG_CPU
 	smp_ops->cpu_disable = pSeries_cpu_disable;
 	smp_ops->cpu_die = pSeries_cpu_die;
@@ -471,3 +456,18 @@ void __init smp_init_pSeries(void)
 	DBG(" <- smp_init_pSeries()\n");
 }
 
+#ifdef CONFIG_MPIC
+void __init smp_init_pseries_mpic(void)
+{
+	smp_ops = &pSeries_mpic_smp_ops;
+
+	smp_init_pseries();
+}
+#endif
+
+void __init smp_init_pseries_xics(void)
+{
+	smp_ops = &pSeries_xics_smp_ops;
+
+	smp_init_pseries();
+}

arch/powerpc/platforms/pseries/xics.c

@@ -8,6 +8,9 @@
  *  as published by the Free Software Foundation; either version
  *  2 of the License, or (at your option) any later version.
  */
+
+#undef DEBUG
+
 #include <linux/types.h>
 #include <linux/threads.h>
 #include <linux/kernel.h>
@@ -19,6 +22,7 @@
 #include <linux/gfp.h>
 #include <linux/radix-tree.h>
 #include <linux/cpu.h>
+
 #include <asm/firmware.h>
 #include <asm/prom.h>
 #include <asm/io.h>
@@ -31,9 +35,6 @@
 
 #include "xics.h"
 
-/* This is used to map real irq numbers to virtual */
-static struct radix_tree_root irq_map = RADIX_TREE_INIT(GFP_ATOMIC);
-
 #define XICS_IPI		2
 #define XICS_IRQ_SPURIOUS	0
 
@@ -64,12 +65,12 @@ struct xics_ipl {
 
 static struct xics_ipl __iomem *xics_per_cpu[NR_CPUS];
 
-static int xics_irq_8259_cascade = 0;
-static int xics_irq_8259_cascade_real = 0;
 static unsigned int default_server = 0xFF;
 static unsigned int default_distrib_server = 0;
 static unsigned int interrupt_server_size = 8;
 
+static struct irq_host *xics_host;
+
 /*
  * XICS only has a single IPI, so encode the messages per CPU
  */
@@ -85,7 +86,7 @@ static int ibm_int_off;
 /* Direct HW low level accessors */
 
 
-static inline int direct_xirr_info_get(int n_cpu)
+static inline unsigned int direct_xirr_info_get(int n_cpu)
 {
 	return in_be32(&xics_per_cpu[n_cpu]->xirr.word);
 }
@@ -130,7 +131,7 @@ static inline long plpar_xirr(unsigned long *xirr_ret)
 	return plpar_hcall(H_XIRR, 0, 0, 0, 0, xirr_ret, &dummy, &dummy);
 }
 
-static inline int lpar_xirr_info_get(int n_cpu)
+static inline unsigned int lpar_xirr_info_get(int n_cpu)
 {
 	unsigned long lpar_rc;
 	unsigned long return_value;
@@ -138,7 +139,7 @@ static inline int lpar_xirr_info_get(int n_cpu)
 	lpar_rc = plpar_xirr(&return_value);
 	if (lpar_rc != H_SUCCESS)
 		panic(" bad return code xirr - rc = %lx \n", lpar_rc);
-	return (int)return_value;
+	return (unsigned int)return_value;
 }
 
 static inline void lpar_xirr_info_set(int n_cpu, int value)
@@ -175,11 +176,11 @@ static inline void lpar_qirr_info(int n_cpu , u8 value)
 
 
 #ifdef CONFIG_SMP
-static int get_irq_server(unsigned int irq)
+static int get_irq_server(unsigned int virq)
 {
 	unsigned int server;
 	/* For the moment only implement delivery to all cpus or one cpu */
-	cpumask_t cpumask = irq_desc[irq].affinity;
+	cpumask_t cpumask = irq_desc[virq].affinity;
 	cpumask_t tmp = CPU_MASK_NONE;
 
 	if (!distribute_irqs)
@@ -200,7 +201,7 @@ static int get_irq_server(unsigned int irq)
 
 }
 #else
-static int get_irq_server(unsigned int irq)
+static int get_irq_server(unsigned int virq)
 {
 	return default_server;
 }
@@ -213,9 +214,11 @@ static void xics_unmask_irq(unsigned int virq)
 	int call_status;
 	unsigned int server;
 
-	irq = virt_irq_to_real(irq_offset_down(virq));
-	WARN_ON(irq == NO_IRQ);
-	if (irq == XICS_IPI || irq == NO_IRQ)
+	pr_debug("xics: unmask virq %d\n", virq);
+
+	irq = (unsigned int)irq_map[virq].hwirq;
+	pr_debug(" -> map to hwirq 0x%x\n", irq);
+	if (irq == XICS_IPI || irq == XICS_IRQ_SPURIOUS)
 		return;
 
 	server = get_irq_server(virq);
@@ -267,75 +270,57 @@ static void xics_mask_irq(unsigned int virq)
 {
 	unsigned int irq;
 
-	irq = virt_irq_to_real(irq_offset_down(virq));
-	WARN_ON(irq == NO_IRQ);
-	if (irq != NO_IRQ)
-		xics_mask_real_irq(irq);
+	pr_debug("xics: mask virq %d\n", virq);
+
+	irq = (unsigned int)irq_map[virq].hwirq;
+	if (irq == XICS_IPI || irq == XICS_IRQ_SPURIOUS)
+		return;
+	xics_mask_real_irq(irq);
 }
 
-static void xics_set_irq_revmap(unsigned int virq)
+static unsigned int xics_startup(unsigned int virq)
 {
 	unsigned int irq;
 
-	irq = irq_offset_down(virq);
-	if (radix_tree_insert(&irq_map, virt_irq_to_real(irq),
-			      &virt_irq_to_real_map[irq]) == -ENOMEM)
-		printk(KERN_CRIT "Out of memory creating real -> virtual"
-		       " IRQ mapping for irq %u (real 0x%x)\n",
-		       virq, virt_irq_to_real(irq));
-}
+	/* force a reverse mapping of the interrupt so it gets in the cache */
+	irq = (unsigned int)irq_map[virq].hwirq;
+	irq_radix_revmap(xics_host, irq);
 
-static unsigned int xics_startup(unsigned int virq)
-{
-	xics_set_irq_revmap(virq);
+	/* unmask it */
 	xics_unmask_irq(virq);
 	return 0;
 }
 
-static unsigned int real_irq_to_virt(unsigned int real_irq)
-{
-	unsigned int *ptr;
-
-	ptr = radix_tree_lookup(&irq_map, real_irq);
-	if (ptr == NULL)
-		return NO_IRQ;
-	return ptr - virt_irq_to_real_map;
-}
-
-static void xics_eoi_direct(unsigned int irq)
+static void xics_eoi_direct(unsigned int virq)
 {
 	int cpu = smp_processor_id();
+	unsigned int irq = (unsigned int)irq_map[virq].hwirq;
 
 	iosync();
-	direct_xirr_info_set(cpu, ((0xff << 24) |
-				   (virt_irq_to_real(irq_offset_down(irq)))));
+	direct_xirr_info_set(cpu, (0xff << 24) | irq);
 }
 
 
-static void xics_eoi_lpar(unsigned int irq)
+static void xics_eoi_lpar(unsigned int virq)
 {
 	int cpu = smp_processor_id();
+	unsigned int irq = (unsigned int)irq_map[virq].hwirq;
 
 	iosync();
-	lpar_xirr_info_set(cpu, ((0xff << 24) |
-				 (virt_irq_to_real(irq_offset_down(irq)))));
-
+	lpar_xirr_info_set(cpu, (0xff << 24) | irq);
 }
 
-static inline int xics_remap_irq(int vec)
+static inline unsigned int xics_remap_irq(unsigned int vec)
 {
-	int irq;
+	unsigned int irq;
 
 	vec &= 0x00ffffff;
 
 	if (vec == XICS_IRQ_SPURIOUS)
 		return NO_IRQ;
-
-	irq = real_irq_to_virt(vec);
-	if (irq == NO_IRQ)
-		irq = real_irq_to_virt_slowpath(vec);
+	irq = irq_radix_revmap(xics_host, vec);
 	if (likely(irq != NO_IRQ))
-		return irq_offset_up(irq);
+		return irq;
 
 	printk(KERN_ERR "Interrupt %u (real) is invalid,"
 	       " disabling it.\n", vec);
@@ -343,14 +328,14 @@ static inline int xics_remap_irq(int vec)
 	return NO_IRQ;
 }
 
-static int xics_get_irq_direct(struct pt_regs *regs)
+static unsigned int xics_get_irq_direct(struct pt_regs *regs)
 {
 	unsigned int cpu = smp_processor_id();
 
 	return xics_remap_irq(direct_xirr_info_get(cpu));
 }
 
-static int xics_get_irq_lpar(struct pt_regs *regs)
+static unsigned int xics_get_irq_lpar(struct pt_regs *regs)
 {
 	unsigned int cpu = smp_processor_id();
 
@@ -437,8 +422,8 @@ static void xics_set_affinity(unsigned int virq, cpumask_t cpumask)
 	unsigned long newmask;
 	cpumask_t tmp = CPU_MASK_NONE;
 
-	irq = virt_irq_to_real(irq_offset_down(virq));
-	if (irq == XICS_IPI || irq == NO_IRQ)
+	irq = (unsigned int)irq_map[virq].hwirq;
+	if (irq == XICS_IPI || irq == XICS_IRQ_SPURIOUS)
 		return;
 
 	status = rtas_call(ibm_get_xive, 1, 3, xics_status, irq);
@@ -469,6 +454,24 @@ static void xics_set_affinity(unsigned int virq, cpumask_t cpumask)
 	}
 }
 
+void xics_setup_cpu(void)
+{
+	int cpu = smp_processor_id();
+
+	xics_set_cpu_priority(cpu, 0xff);
+
+	/*
+	 * Put the calling processor into the GIQ.  This is really only
+	 * necessary from a secondary thread as the OF start-cpu interface
+	 * performs this function for us on primary threads.
+	 *
+	 * XXX: undo of teardown on kexec needs this too, as may hotplug
+	 */
+	rtas_set_indicator(GLOBAL_INTERRUPT_QUEUE,
+		(1UL << interrupt_server_size) - 1 - default_distrib_server, 1);
+}
+
+
 static struct irq_chip xics_pic_direct = {
 	.typename = " XICS     ",
 	.startup = xics_startup,
@@ -489,90 +492,245 @@ static struct irq_chip xics_pic_lpar = {
 };
 
 
-void xics_setup_cpu(void)
+static int xics_host_match(struct irq_host *h, struct device_node *node)
 {
-	int cpu = smp_processor_id();
+	/* IBM machines have interrupt parents of various funky types for things
+	 * like vdevices, events, etc... The trick we use here is to match
+	 * everything here except the legacy 8259 which is compatible "chrp,iic"
+	 */
+	return !device_is_compatible(node, "chrp,iic");
+}
 
-	xics_set_cpu_priority(cpu, 0xff);
+static int xics_host_map_direct(struct irq_host *h, unsigned int virq,
+				irq_hw_number_t hw, unsigned int flags)
+{
+	unsigned int sense = flags & IRQ_TYPE_SENSE_MASK;
 
-	/*
-	 * Put the calling processor into the GIQ.  This is really only
-	 * necessary from a secondary thread as the OF start-cpu interface
-	 * performs this function for us on primary threads.
-	 *
-	 * XXX: undo of teardown on kexec needs this too, as may hotplug
+	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);
+
+	get_irq_desc(virq)->status |= IRQ_LEVEL;
+	set_irq_chip_and_handler(virq, &xics_pic_direct, handle_fasteoi_irq);
+	return 0;
+}
+
+static int xics_host_map_lpar(struct irq_host *h, unsigned int virq,
+			      irq_hw_number_t hw, unsigned int flags)
+{
+	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);
+
+	get_irq_desc(virq)->status |= IRQ_LEVEL;
+	set_irq_chip_and_handler(virq, &xics_pic_lpar, handle_fasteoi_irq);
+	return 0;
+}
+
+static int xics_host_xlate(struct irq_host *h, struct device_node *ct,
+			   u32 *intspec, unsigned int intsize,
+			   irq_hw_number_t *out_hwirq, unsigned int *out_flags)
+
+{
+	/* Current xics implementation translates everything
+	 * to level. It is not technically right for MSIs but this
+	 * is irrelevant at this point. We might get smarter in the future
 	 */
-	rtas_set_indicator(GLOBAL_INTERRUPT_QUEUE,
-		(1UL << interrupt_server_size) - 1 - default_distrib_server, 1);
+	*out_hwirq = intspec[0];
+	*out_flags = IRQ_TYPE_LEVEL_LOW;
+
+	return 0;
+}
+
+static struct irq_host_ops xics_host_direct_ops = {
+	.match = xics_host_match,
+	.map = xics_host_map_direct,
+	.xlate = xics_host_xlate,
+};
+
+static struct irq_host_ops xics_host_lpar_ops = {
+	.match = xics_host_match,
+	.map = xics_host_map_lpar,
+	.xlate = xics_host_xlate,
+};
+
+static void __init xics_init_host(void)
+{
+	struct irq_host_ops *ops;
+
+	if (firmware_has_feature(FW_FEATURE_LPAR))
+		ops = &xics_host_lpar_ops;
+	else
+		ops = &xics_host_direct_ops;
+	xics_host = irq_alloc_host(IRQ_HOST_MAP_TREE, 0, ops,
+				   XICS_IRQ_SPURIOUS);
+	BUG_ON(xics_host == NULL);
+	irq_set_default_host(xics_host);
 }
 
-void xics_init_IRQ(void)
+static void __init xics_map_one_cpu(int hw_id, unsigned long addr,
+				     unsigned long size)
 {
+#ifdef CONFIG_SMP
 	int i;
-	unsigned long intr_size = 0;
-	struct device_node *np;
-	uint *ireg, ilen, indx = 0;
-	unsigned long intr_base = 0;
-	struct xics_interrupt_node {
-		unsigned long addr;
-		unsigned long size;
-	} intnodes[NR_CPUS];
-	struct irq_chip *chip;
 
-	ppc64_boot_msg(0x20, "XICS Init");
+	/* This may look gross but it's good enough for now, we don't quite
+	 * have a hard -> linux processor id matching.
+	 */
+	for_each_possible_cpu(i) {
+		if (!cpu_present(i))
+			continue;
+		if (hw_id == get_hard_smp_processor_id(i)) {
+			xics_per_cpu[i] = ioremap(addr, size);
+			return;
+		}
+	}
+#else
+	if (hw_id != 0)
+		return;
+	xics_per_cpu[0] = ioremap(addr, size);
+#endif /* CONFIG_SMP */
+}
 
-	ibm_get_xive = rtas_token("ibm,get-xive");
-	ibm_set_xive = rtas_token("ibm,set-xive");
-	ibm_int_on  = rtas_token("ibm,int-on");
-	ibm_int_off = rtas_token("ibm,int-off");
+static void __init xics_init_one_node(struct device_node *np,
+				      unsigned int *indx)
+{
+	unsigned int ilen;
+	u32 *ireg;
 
-	np = of_find_node_by_type(NULL, "PowerPC-External-Interrupt-Presentation");
-	if (!np)
-		panic("xics_init_IRQ: can't find interrupt presentation");
+	/* This code does the theorically broken assumption that the interrupt
+	 * server numbers are the same as the hard CPU numbers.
+	 * This happens to be the case so far but we are playing with fire...
+	 * should be fixed one of these days. -BenH.
+	 */
+	ireg = (u32 *)get_property(np, "ibm,interrupt-server-ranges", NULL);
 
-nextnode:
-	ireg = (uint *)get_property(np, "ibm,interrupt-server-ranges", NULL);
+	/* Do that ever happen ? we'll know soon enough... but even good'old
+	 * f80 does have that property ..
+	 */
+	WARN_ON(ireg == NULL);
 	if (ireg) {
 		/*
 		 * set node starting index for this node
 		 */
-		indx = *ireg;
+		*indx = *ireg;
 	}
-
-	ireg = (uint *)get_property(np, "reg", &ilen);
+	ireg = (u32 *)get_property(np, "reg", &ilen);
 	if (!ireg)
 		panic("xics_init_IRQ: can't find interrupt reg property");
 
-	while (ilen) {
-		intnodes[indx].addr = (unsigned long)*ireg++ << 32;
-		ilen -= sizeof(uint);
-		intnodes[indx].addr |= *ireg++;
-		ilen -= sizeof(uint);
-		intnodes[indx].size = (unsigned long)*ireg++ << 32;
-		ilen -= sizeof(uint);
-		intnodes[indx].size |= *ireg++;
-		ilen -= sizeof(uint);
-		indx++;
-		if (indx >= NR_CPUS) break;
+	while (ilen >= (4 * sizeof(u32))) {
+		unsigned long addr, size;
+
+		/* XXX Use proper OF parsing code here !!! */
+		addr = (unsigned long)*ireg++ << 32;
+		ilen -= sizeof(u32);
+		addr |= *ireg++;
+		ilen -= sizeof(u32);
+		size = (unsigned long)*ireg++ << 32;
+		ilen -= sizeof(u32);
+		size |= *ireg++;
+		ilen -= sizeof(u32);
+		xics_map_one_cpu(*indx, addr, size);
+		(*indx)++;
+	}
+}
+
+
+static void __init xics_setup_8259_cascade(void)
+{
+	struct device_node *np, *old, *found = NULL;
+	int cascade, naddr;
+	u32 *addrp;
+	unsigned long intack = 0;
+
+	for_each_node_by_type(np, "interrupt-controller")
+		if (device_is_compatible(np, "chrp,iic")) {
+			found = np;
+			break;
+		}
+	if (found == NULL) {
+		printk(KERN_DEBUG "xics: no ISA interrupt controller\n");
+		return;
 	}
+	cascade = irq_of_parse_and_map(found, 0);
+	if (cascade == NO_IRQ) {
+		printk(KERN_ERR "xics: failed to map cascade interrupt");
+		return;
+	}
+	pr_debug("xics: cascade mapped to irq %d\n", cascade);
+
+	for (old = of_node_get(found); old != NULL ; old = np) {
+		np = of_get_parent(old);
+		of_node_put(old);
+		if (np == NULL)
+			break;
+		if (strcmp(np->name, "pci") != 0)
+			continue;
+		addrp = (u32 *)get_property(np, "8259-interrupt-acknowledge", NULL);
+		if (addrp == NULL)
+			continue;
+		naddr = prom_n_addr_cells(np);
+		intack = addrp[naddr-1];
+		if (naddr > 1)
+			intack |= ((unsigned long)addrp[naddr-2]) << 32;
+	}
+	if (intack)
+		printk(KERN_DEBUG "xics: PCI 8259 intack at 0x%016lx\n", intack);
+	i8259_init(found, intack);
+	of_node_put(found);
+	set_irq_chained_handler(cascade, pseries_8259_cascade);
+}
 
-	np = of_find_node_by_type(np, "PowerPC-External-Interrupt-Presentation");
-	if ((indx < NR_CPUS) && np) goto nextnode;
+void __init xics_init_IRQ(void)
+{
+	int i;
+	struct device_node *np;
+	u32 *ireg, ilen, indx = 0;
+	int found = 0;
+
+	ppc64_boot_msg(0x20, "XICS Init");
+
+	ibm_get_xive = rtas_token("ibm,get-xive");
+	ibm_set_xive = rtas_token("ibm,set-xive");
+	ibm_int_on  = rtas_token("ibm,int-on");
+	ibm_int_off = rtas_token("ibm,int-off");
+
+	for_each_node_by_type(np, "PowerPC-External-Interrupt-Presentation") {
+		found = 1;
+		if (firmware_has_feature(FW_FEATURE_LPAR))
+			break;
+		xics_init_one_node(np, &indx);
+	}
+	if (found == 0)
+		return;
+
+	xics_init_host();
 
 	/* Find the server numbers for the boot cpu. */
 	for (np = of_find_node_by_type(NULL, "cpu");
 	     np;
 	     np = of_find_node_by_type(np, "cpu")) {
-		ireg = (uint *)get_property(np, "reg", &ilen);
+		ireg = (u32 *)get_property(np, "reg", &ilen);
 		if (ireg && ireg[0] == get_hard_smp_processor_id(boot_cpuid)) {
-			ireg = (uint *)get_property(np, "ibm,ppc-interrupt-gserver#s",
-						    &ilen);
+			ireg = (u32 *)get_property(np,
+						  "ibm,ppc-interrupt-gserver#s",
+						   &ilen);
 			i = ilen / sizeof(int);
 			if (ireg && i > 0) {
 				default_server = ireg[0];
-				default_distrib_server = ireg[i-1]; /* take last element */
+				/* take last element */
+				default_distrib_server = ireg[i-1];
 			}
-			ireg = (uint *)get_property(np,
+			ireg = (u32 *)get_property(np,
 					"ibm,interrupt-server#-size", NULL);
 			if (ireg)
 				interrupt_server_size = *ireg;
@@ -581,102 +739,48 @@ nextnode:
 	}
 	of_node_put(np);
 
-	intr_base = intnodes[0].addr;
-	intr_size = intnodes[0].size;
-
- 	if (firmware_has_feature(FW_FEATURE_LPAR)) {
- 		ppc_md.get_irq = xics_get_irq_lpar;
- 		chip = &xics_pic_lpar;
-  	} else {
-#ifdef CONFIG_SMP
-		for_each_possible_cpu(i) {
-			int hard_id;
-
-			/* FIXME: Do this dynamically! --RR */
-			if (!cpu_present(i))
-				continue;
-
-			hard_id = get_hard_smp_processor_id(i);
-			xics_per_cpu[i] = ioremap(intnodes[hard_id].addr,
-						  intnodes[hard_id].size);
-		}
-#else
-		xics_per_cpu[0] = ioremap(intr_base, intr_size);
-#endif /* CONFIG_SMP */
+	if (firmware_has_feature(FW_FEATURE_LPAR))
+		ppc_md.get_irq = xics_get_irq_lpar;
+	else
 		ppc_md.get_irq = xics_get_irq_direct;
-		chip = &xics_pic_direct;
-
-	}
-
-	for (i = irq_offset_value(); i < NR_IRQS; ++i) {
-		/* All IRQs on XICS are level for now. MSI code may want to modify
-		 * that for reporting purposes
-		 */
-		get_irq_desc(i)->status |= IRQ_LEVEL;
-		set_irq_chip_and_handler(i, chip, handle_fasteoi_irq);
-	}
 
 	xics_setup_cpu();
 
-	ppc64_boot_msg(0x21, "XICS Done");
-}
+	xics_setup_8259_cascade();
 
-static int xics_setup_8259_cascade(void)
-{
-	struct device_node *np;
-	uint *ireg;
-
-	np = of_find_node_by_type(NULL, "interrupt-controller");
-	if (np == NULL) {
-		printk(KERN_WARNING "xics: no ISA interrupt controller\n");
-		xics_irq_8259_cascade_real = -1;
-		xics_irq_8259_cascade = -1;
-		return 0;
-	}
-
-	ireg = (uint *) get_property(np, "interrupts", NULL);
-	if (!ireg)
-		panic("xics_init_IRQ: can't find ISA interrupts property");
-
-	xics_irq_8259_cascade_real = *ireg;
-	xics_irq_8259_cascade = irq_offset_up
-		(virt_irq_create_mapping(xics_irq_8259_cascade_real));
-	i8259_init(0, 0);
-	of_node_put(np);
-
-	xics_set_irq_revmap(xics_irq_8259_cascade);
-	set_irq_chained_handler(xics_irq_8259_cascade, pSeries_8259_cascade);
-
-	return 0;
+	ppc64_boot_msg(0x21, "XICS Done");
 }
-arch_initcall(xics_setup_8259_cascade);
 
 
 #ifdef CONFIG_SMP
 void xics_request_IPIs(void)
 {
-	virt_irq_to_real_map[XICS_IPI] = XICS_IPI;
+	unsigned int ipi;
+
+	ipi = irq_create_mapping(xics_host, XICS_IPI, 0);
+	BUG_ON(ipi == NO_IRQ);
 
 	/*
 	 * IPIs are marked IRQF_DISABLED as they must run with irqs
 	 * disabled
 	 */
-	set_irq_handler(irq_offset_up(XICS_IPI), handle_percpu_irq);
+	set_irq_handler(ipi, handle_percpu_irq);
 	if (firmware_has_feature(FW_FEATURE_LPAR))
-		request_irq(irq_offset_up(XICS_IPI), xics_ipi_action_lpar,
-			    SA_INTERRUPT, "IPI", NULL);
+		request_irq(ipi, xics_ipi_action_lpar, IRQF_DISABLED,
+			    "IPI", NULL);
 	else
-		request_irq(irq_offset_up(XICS_IPI), xics_ipi_action_direct,
-			    SA_INTERRUPT, "IPI", NULL);
+		request_irq(ipi, xics_ipi_action_direct, IRQF_DISABLED,
+			    "IPI", NULL);
 }
 #endif /* CONFIG_SMP */
 
 void xics_teardown_cpu(int secondary)
 {
-	struct irq_desc *desc = get_irq_desc(irq_offset_up(XICS_IPI));
 	int cpu = smp_processor_id();
+	unsigned int ipi;
+	struct irq_desc *desc;
 
- 	xics_set_cpu_priority(cpu, 0);
+	xics_set_cpu_priority(cpu, 0);
 
 	/*
 	 * we need to EOI the IPI if we got here from kexec down IPI
@@ -685,6 +789,11 @@ void xics_teardown_cpu(int secondary)
 	 * should we be flagging idle loop instead?
 	 * or creating some task to be scheduled?
 	 */
+
+	ipi = irq_find_mapping(xics_host, XICS_IPI);
+	if (ipi == XICS_IRQ_SPURIOUS)
+		return;
+	desc = get_irq_desc(ipi);
 	if (desc->chip && desc->chip->eoi)
 		desc->chip->eoi(XICS_IPI);
 
@@ -694,8 +803,8 @@ void xics_teardown_cpu(int secondary)
 	 */
 	if (secondary)
 		rtas_set_indicator(GLOBAL_INTERRUPT_QUEUE,
-			(1UL << interrupt_server_size) - 1 -
-			default_distrib_server, 0);
+				   (1UL << interrupt_server_size) - 1 -
+				   default_distrib_server, 0);
 }
 
 #ifdef CONFIG_HOTPLUG_CPU
@@ -723,15 +832,15 @@ void xics_migrate_irqs_away(void)
 		unsigned long flags;
 
 		/* We cant set affinity on ISA interrupts */
-		if (virq < irq_offset_value())
+		if (virq < NUM_ISA_INTERRUPTS)
 			continue;
-
-		desc = get_irq_desc(virq);
-		irq = virt_irq_to_real(irq_offset_down(virq));
-
+		if (irq_map[virq].host != xics_host)
+			continue;
+		irq = (unsigned int)irq_map[virq].hwirq;
 		/* We need to get IPIs still. */
-		if (irq == XICS_IPI || irq == NO_IRQ)
+		if (irq == XICS_IPI || irq == XICS_IRQ_SPURIOUS)
 			continue;
+		desc = get_irq_desc(virq);
 
 		/* We only need to migrate enabled IRQS */
 		if (desc == NULL || desc->chip == NULL

arch/powerpc/platforms/pseries/xics.h

@@ -31,7 +31,7 @@ struct xics_ipi_struct {
 extern struct xics_ipi_struct xics_ipi_message[NR_CPUS] __cacheline_aligned;
 
 struct irq_desc;
-extern void pSeries_8259_cascade(unsigned int irq, struct irq_desc *desc,
+extern void pseries_8259_cascade(unsigned int irq, struct irq_desc *desc,
 				 struct pt_regs *regs);
 
 #endif /* _POWERPC_KERNEL_XICS_H */

arch/powerpc/sysdev/i8259.c

@@ -6,11 +6,16 @@
  * as published by the Free Software Foundation; either version
  * 2 of the License, or (at your option) any later version.
  */
+#undef DEBUG
+
 #include <linux/init.h>
 #include <linux/ioport.h>
 #include <linux/interrupt.h>
+#include <linux/kernel.h>
+#include <linux/delay.h>
 #include <asm/io.h>
 #include <asm/i8259.h>
+#include <asm/prom.h>
 
 static volatile void __iomem *pci_intack; /* RO, gives us the irq vector */
 
@@ -20,7 +25,8 @@ static unsigned char cached_8259[2] = { 0xff, 0xff };
 
 static DEFINE_SPINLOCK(i8259_lock);
 
-static int i8259_pic_irq_offset;
+static struct device_node *i8259_node;
+static struct irq_host *i8259_host;
 
 /*
  * Acknowledge the IRQ using either the PCI host bridge's interrupt
@@ -28,16 +34,18 @@ static int i8259_pic_irq_offset;
  * which is called.  It should be noted that polling is broken on some
  * IBM and Motorola PReP boxes so we must use the int-ack feature on them.
  */
-int i8259_irq(struct pt_regs *regs)
+unsigned int i8259_irq(struct pt_regs *regs)
 {
 	int irq;
-
-	spin_lock(&i8259_lock);
+	int lock = 0;
 
 	/* Either int-ack or poll for the IRQ */
 	if (pci_intack)
 		irq = readb(pci_intack);
 	else {
+		spin_lock(&i8259_lock);
+		lock = 1;
+
 		/* Perform an interrupt acknowledge cycle on controller 1. */
 		outb(0x0C, 0x20);		/* prepare for poll */
 		irq = inb(0x20) & 7;
@@ -62,11 +70,13 @@ int i8259_irq(struct pt_regs *regs)
 		if (!pci_intack)
 			outb(0x0B, 0x20);	/* ISR register */
 		if(~inb(0x20) & 0x80)
-			irq = -1;
-	}
+			irq = NO_IRQ;
+	} else if (irq == 0xff)
+		irq = NO_IRQ;
 
-	spin_unlock(&i8259_lock);
-	return irq + i8259_pic_irq_offset;
+	if (lock)
+		spin_unlock(&i8259_lock);
+	return irq;
 }
 
 static void i8259_mask_and_ack_irq(unsigned int irq_nr)
@@ -74,7 +84,6 @@ static void i8259_mask_and_ack_irq(unsigned int irq_nr)
 	unsigned long flags;
 
 	spin_lock_irqsave(&i8259_lock, flags);
-	irq_nr -= i8259_pic_irq_offset;
 	if (irq_nr > 7) {
 		cached_A1 |= 1 << (irq_nr-8);
 		inb(0xA1); 	/* DUMMY */
@@ -100,8 +109,9 @@ static void i8259_mask_irq(unsigned int irq_nr)
 {
 	unsigned long flags;
 
+	pr_debug("i8259_mask_irq(%d)\n", irq_nr);
+
 	spin_lock_irqsave(&i8259_lock, flags);
-	irq_nr -= i8259_pic_irq_offset;
 	if (irq_nr < 8)
 		cached_21 |= 1 << irq_nr;
 	else
@@ -114,8 +124,9 @@ static void i8259_unmask_irq(unsigned int irq_nr)
 {
 	unsigned long flags;
 
+	pr_debug("i8259_unmask_irq(%d)\n", irq_nr);
+
 	spin_lock_irqsave(&i8259_lock, flags);
-	irq_nr -= i8259_pic_irq_offset;
 	if (irq_nr < 8)
 		cached_21 &= ~(1 << irq_nr);
 	else
@@ -152,25 +163,84 @@ static struct resource pic_edgectrl_iores = {
 	.flags = IORESOURCE_BUSY,
 };
 
-static struct irqaction i8259_irqaction = {
-	.handler = no_action,
-	.flags = IRQF_DISABLED,
-	.mask = CPU_MASK_NONE,
-	.name = "82c59 secondary cascade",
+static int i8259_host_match(struct irq_host *h, struct device_node *node)
+{
+	return i8259_node == NULL || i8259_node == node;
+}
+
+static int i8259_host_map(struct irq_host *h, unsigned int virq,
+			  irq_hw_number_t hw, unsigned int flags)
+{
+	pr_debug("i8259_host_map(%d, 0x%lx)\n", virq, hw);
+
+	/* We block the internal cascade */
+	if (hw == 2)
+		get_irq_desc(virq)->status |= IRQ_NOREQUEST;
+
+	/* We use the level stuff only for now, we might want to
+	 * be more cautious here but that works for now
+	 */
+	get_irq_desc(virq)->status |= IRQ_LEVEL;
+	set_irq_chip_and_handler(virq, &i8259_pic, handle_level_irq);
+	return 0;
+}
+
+static void i8259_host_unmap(struct irq_host *h, unsigned int virq)
+{
+	/* Make sure irq is masked in hardware */
+	i8259_mask_irq(virq);
+
+	/* remove chip and handler */
+	set_irq_chip_and_handler(virq, NULL, NULL);
+
+	/* Make sure it's completed */
+	synchronize_irq(virq);
+}
+
+static int i8259_host_xlate(struct irq_host *h, struct device_node *ct,
+			    u32 *intspec, unsigned int intsize,
+			    irq_hw_number_t *out_hwirq, unsigned int *out_flags)
+{
+	static unsigned char map_isa_senses[4] = {
+		IRQ_TYPE_LEVEL_LOW,
+		IRQ_TYPE_LEVEL_HIGH,
+		IRQ_TYPE_EDGE_FALLING,
+		IRQ_TYPE_EDGE_RISING,
+	};
+
+	*out_hwirq = intspec[0];
+	if (intsize > 1 && intspec[1] < 4)
+		*out_flags = map_isa_senses[intspec[1]];
+	else
+		*out_flags = IRQ_TYPE_NONE;
+
+	return 0;
+}
+
+static struct irq_host_ops i8259_host_ops = {
+	.match = i8259_host_match,
+	.map = i8259_host_map,
+	.unmap = i8259_host_unmap,
+	.xlate = i8259_host_xlate,
 };
 
-/*
- * i8259_init()
- * intack_addr - PCI interrupt acknowledge (real) address which will return
- *               the active irq from the 8259
+/****
+ * i8259_init - Initialize the legacy controller
+ * @node: device node of the legacy PIC (can be NULL, but then, it will match
+ *        all interrupts, so beware)
+ * @intack_addr: PCI interrupt acknowledge (real) address which will return
+ *             	 the active irq from the 8259
  */
-void __init i8259_init(unsigned long intack_addr, int offset)
+void i8259_init(struct device_node *node, unsigned long intack_addr)
 {
 	unsigned long flags;
-	int i;
 
+	/* initialize the controller */
 	spin_lock_irqsave(&i8259_lock, flags);
-	i8259_pic_irq_offset = offset;
+
+	/* Mask all first */
+	outb(0xff, 0xA1);
+	outb(0xff, 0x21);
 
 	/* init master interrupt controller */
 	outb(0x11, 0x20); /* Start init sequence */
@@ -184,24 +254,36 @@ void __init i8259_init(unsigned long intack_addr, int offset)
 	outb(0x02, 0xA1); /* edge triggered, Cascade (slave) on IRQ2 */
 	outb(0x01, 0xA1); /* Select 8086 mode */
 
+	/* That thing is slow */
+	udelay(100);
+
 	/* always read ISR */
 	outb(0x0B, 0x20);
 	outb(0x0B, 0xA0);
 
-	/* Mask all interrupts */
+	/* Unmask the internal cascade */
+	cached_21 &= ~(1 << 2);
+
+	/* Set interrupt masks */
 	outb(cached_A1, 0xA1);
 	outb(cached_21, 0x21);
 
 	spin_unlock_irqrestore(&i8259_lock, flags);
 
-	for (i = 0; i < NUM_ISA_INTERRUPTS; ++i) {
-		set_irq_chip_and_handler(offset + i, &i8259_pic,
-					 handle_level_irq);
-		irq_desc[offset + i].status |= IRQ_LEVEL;
+	/* create a legacy host */
+	if (node)
+		i8259_node = of_node_get(node);
+	i8259_host = irq_alloc_host(IRQ_HOST_MAP_LEGACY, 0, &i8259_host_ops, 0);
+	if (i8259_host == NULL) {
+		printk(KERN_ERR "i8259: failed to allocate irq host !\n");
+		return;
 	}
 
 	/* reserve our resources */
-	setup_irq(offset + 2, &i8259_irqaction);
+	/* XXX should we continue doing that ? it seems to cause problems
+	 * with further requesting of PCI IO resources for that range...
+	 * need to look into it.
+	 */
 	request_resource(&ioport_resource, &pic1_iores);
 	request_resource(&ioport_resource, &pic2_iores);
 	request_resource(&ioport_resource, &pic_edgectrl_iores);
@@ -209,4 +291,5 @@ void __init i8259_init(unsigned long intack_addr, int offset)
 	if (intack_addr != 0)
 		pci_intack = ioremap(intack_addr, 1);
 
+	printk(KERN_INFO "i8259 legacy interrupt controller initialized\n");
 }

arch/powerpc/sysdev/mpic.c

@@ -340,27 +340,19 @@ static void __init mpic_scan_ht_pics(struct mpic *mpic)
 #endif /* CONFIG_MPIC_BROKEN_U3 */
 
 
+#define mpic_irq_to_hw(virq)	((unsigned int)irq_map[virq].hwirq)
+
 /* Find an mpic associated with a given linux interrupt */
 static struct mpic *mpic_find(unsigned int irq, unsigned int *is_ipi)
 {
-	struct mpic *mpic = mpics;
-
-	while(mpic) {
-		/* search IPIs first since they may override the main interrupts */
-		if (irq >= mpic->ipi_offset && irq < (mpic->ipi_offset + 4)) {
-			if (is_ipi)
-				*is_ipi = 1;
-			return mpic;
-		}
-		if (irq >= mpic->irq_offset &&
-		    irq < (mpic->irq_offset + mpic->irq_count)) {
-			if (is_ipi)
-				*is_ipi = 0;
-			return mpic;
-		}
-		mpic = mpic -> next;
-	}
-	return NULL;
+	unsigned int src = mpic_irq_to_hw(irq);
+
+	if (irq < NUM_ISA_INTERRUPTS)
+		return NULL;
+	if (is_ipi)
+		*is_ipi = (src >= MPIC_VEC_IPI_0 && src <= MPIC_VEC_IPI_3);
+
+	return irq_desc[irq].chip_data;
 }
 
 /* Convert a cpu mask from logical to physical cpu numbers. */
@@ -398,9 +390,7 @@ static inline void mpic_eoi(struct mpic *mpic)
 #ifdef CONFIG_SMP
 static irqreturn_t mpic_ipi_action(int irq, void *dev_id, struct pt_regs *regs)
 {
-	struct mpic *mpic = dev_id;
-
-	smp_message_recv(irq - mpic->ipi_offset, regs);
+	smp_message_recv(mpic_irq_to_hw(irq) - MPIC_VEC_IPI_0, regs);
 	return IRQ_HANDLED;
 }
 #endif /* CONFIG_SMP */
@@ -414,7 +404,7 @@ static void mpic_unmask_irq(unsigned int irq)
 {
 	unsigned int loops = 100000;
 	struct mpic *mpic = mpic_from_irq(irq);
-	unsigned int src = irq - mpic->irq_offset;
+	unsigned int src = mpic_irq_to_hw(irq);
 
 	DBG("%p: %s: enable_irq: %d (src %d)\n", mpic, mpic->name, irq, src);
 
@@ -435,7 +425,7 @@ static void mpic_mask_irq(unsigned int irq)
 {
 	unsigned int loops = 100000;
 	struct mpic *mpic = mpic_from_irq(irq);
-	unsigned int src = irq - mpic->irq_offset;
+	unsigned int src = mpic_irq_to_hw(irq);
 
 	DBG("%s: disable_irq: %d (src %d)\n", mpic->name, irq, src);
 
@@ -472,7 +462,7 @@ static void mpic_end_irq(unsigned int irq)
 static void mpic_unmask_ht_irq(unsigned int irq)
 {
 	struct mpic *mpic = mpic_from_irq(irq);
-	unsigned int src = irq - mpic->irq_offset;
+	unsigned int src = mpic_irq_to_hw(irq);
 
 	mpic_unmask_irq(irq);
 
@@ -483,7 +473,7 @@ static void mpic_unmask_ht_irq(unsigned int irq)
 static unsigned int mpic_startup_ht_irq(unsigned int irq)
 {
 	struct mpic *mpic = mpic_from_irq(irq);
-	unsigned int src = irq - mpic->irq_offset;
+	unsigned int src = mpic_irq_to_hw(irq);
 
 	mpic_unmask_irq(irq);
 	mpic_startup_ht_interrupt(mpic, src, irq_desc[irq].status);
@@ -494,7 +484,7 @@ static unsigned int mpic_startup_ht_irq(unsigned int irq)
 static void mpic_shutdown_ht_irq(unsigned int irq)
 {
 	struct mpic *mpic = mpic_from_irq(irq);
-	unsigned int src = irq - mpic->irq_offset;
+	unsigned int src = mpic_irq_to_hw(irq);
 
 	mpic_shutdown_ht_interrupt(mpic, src, irq_desc[irq].status);
 	mpic_mask_irq(irq);
@@ -503,7 +493,7 @@ static void mpic_shutdown_ht_irq(unsigned int irq)
 static void mpic_end_ht_irq(unsigned int irq)
 {
 	struct mpic *mpic = mpic_from_irq(irq);
-	unsigned int src = irq - mpic->irq_offset;
+	unsigned int src = mpic_irq_to_hw(irq);
 
 #ifdef DEBUG_IRQ
 	DBG("%s: end_irq: %d\n", mpic->name, irq);
@@ -525,7 +515,7 @@ static void mpic_end_ht_irq(unsigned int irq)
 static void mpic_unmask_ipi(unsigned int irq)
 {
 	struct mpic *mpic = mpic_from_ipi(irq);
-	unsigned int src = irq - mpic->ipi_offset;
+	unsigned int src = mpic_irq_to_hw(irq) - MPIC_VEC_IPI_0;
 
 	DBG("%s: enable_ipi: %d (ipi %d)\n", mpic->name, irq, src);
 	mpic_ipi_write(src, mpic_ipi_read(src) & ~MPIC_VECPRI_MASK);
@@ -555,15 +545,46 @@ static void mpic_end_ipi(unsigned int irq)
 static void mpic_set_affinity(unsigned int irq, cpumask_t cpumask)
 {
 	struct mpic *mpic = mpic_from_irq(irq);
+	unsigned int src = mpic_irq_to_hw(irq);
 
 	cpumask_t tmp;
 
 	cpus_and(tmp, cpumask, cpu_online_map);
 
-	mpic_irq_write(irq - mpic->irq_offset, MPIC_IRQ_DESTINATION,
+	mpic_irq_write(src, MPIC_IRQ_DESTINATION,
 		       mpic_physmask(cpus_addr(tmp)[0]));	
 }
 
+static unsigned int mpic_flags_to_vecpri(unsigned int flags, int *level)
+{
+	unsigned int vecpri;
+
+	/* Now convert sense value */
+	switch(flags & IRQ_TYPE_SENSE_MASK) {
+	case IRQ_TYPE_EDGE_RISING:
+		vecpri = MPIC_VECPRI_SENSE_EDGE |
+			MPIC_VECPRI_POLARITY_POSITIVE;
+		*level = 0;
+		break;
+	case IRQ_TYPE_EDGE_FALLING:
+		vecpri = MPIC_VECPRI_SENSE_EDGE |
+			MPIC_VECPRI_POLARITY_NEGATIVE;
+		*level = 0;
+		break;
+	case IRQ_TYPE_LEVEL_HIGH:
+		vecpri = MPIC_VECPRI_SENSE_LEVEL |
+			MPIC_VECPRI_POLARITY_POSITIVE;
+		*level = 1;
+		break;
+	case IRQ_TYPE_LEVEL_LOW:
+	default:
+		vecpri = MPIC_VECPRI_SENSE_LEVEL |
+			MPIC_VECPRI_POLARITY_NEGATIVE;
+		*level = 1;
+	}
+	return vecpri;
+}
+
 static struct irq_chip mpic_irq_chip = {
 	.mask	= mpic_mask_irq,
 	.unmask	= mpic_unmask_irq,
@@ -589,19 +610,111 @@ static struct irq_chip mpic_irq_ht_chip = {
 #endif /* CONFIG_MPIC_BROKEN_U3 */
 
 
+static int mpic_host_match(struct irq_host *h, struct device_node *node)
+{
+	struct mpic *mpic = h->host_data;
+
+	/* Exact match, unless mpic node is NULL */
+	return mpic->of_node == NULL || mpic->of_node == node;
+}
+
+static int mpic_host_map(struct irq_host *h, unsigned int virq,
+			 irq_hw_number_t hw, unsigned int flags)
+{
+	struct irq_desc *desc = get_irq_desc(virq);
+	struct irq_chip *chip;
+	struct mpic *mpic = h->host_data;
+	unsigned int vecpri = MPIC_VECPRI_SENSE_LEVEL |
+		MPIC_VECPRI_POLARITY_NEGATIVE;
+	int level;
+
+	pr_debug("mpic: map virq %d, hwirq 0x%lx, flags: 0x%x\n",
+		 virq, hw, flags);
+
+	if (hw == MPIC_VEC_SPURRIOUS)
+		return -EINVAL;
+#ifdef CONFIG_SMP
+	else if (hw >= MPIC_VEC_IPI_0) {
+		WARN_ON(!(mpic->flags & MPIC_PRIMARY));
+
+		pr_debug("mpic: mapping as IPI\n");
+		set_irq_chip_data(virq, mpic);
+		set_irq_chip_and_handler(virq, &mpic->hc_ipi,
+					 handle_percpu_irq);
+		return 0;
+	}
+#endif /* CONFIG_SMP */
+
+	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;
+	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;
+		chip = &mpic->hc_ht_irq;
+	}
+#endif
+
+	/* Reconfigure irq */
+	vecpri |= MPIC_VECPRI_MASK | hw | (8 << MPIC_VECPRI_PRIORITY_SHIFT);
+	mpic_irq_write(hw, MPIC_IRQ_VECTOR_PRI, vecpri);
+
+	pr_debug("mpic: mapping as IRQ\n");
+
+	set_irq_chip_data(virq, mpic);
+	set_irq_chip_and_handler(virq, chip, handle_fasteoi_irq);
+	return 0;
+}
+
+static int mpic_host_xlate(struct irq_host *h, struct device_node *ct,
+			   u32 *intspec, unsigned int intsize,
+			   irq_hw_number_t *out_hwirq, unsigned int *out_flags)
+
+{
+	static unsigned char map_mpic_senses[4] = {
+		IRQ_TYPE_EDGE_RISING,
+		IRQ_TYPE_LEVEL_LOW,
+		IRQ_TYPE_LEVEL_HIGH,
+		IRQ_TYPE_EDGE_FALLING,
+	};
+
+	*out_hwirq = intspec[0];
+	if (intsize > 1 && intspec[1] < 4)
+		*out_flags = map_mpic_senses[intspec[1]];
+	else
+		*out_flags = IRQ_TYPE_NONE;
+
+	return 0;
+}
+
+static struct irq_host_ops mpic_host_ops = {
+	.match = mpic_host_match,
+	.map = mpic_host_map,
+	.xlate = mpic_host_xlate,
+};
+
 /*
  * Exported functions
  */
 
-
-struct mpic * __init mpic_alloc(unsigned long phys_addr,
+struct mpic * __init mpic_alloc(struct device_node *node,
+				unsigned long phys_addr,
 				unsigned int flags,
 				unsigned int isu_size,
-				unsigned int irq_offset,
 				unsigned int irq_count,
-				unsigned int ipi_offset,
-				unsigned char *senses,
-				unsigned int senses_count,
 				const char *name)
 {
 	struct mpic	*mpic;
@@ -613,10 +726,19 @@ struct mpic * __init mpic_alloc(unsigned long phys_addr,
 	if (mpic == NULL)
 		return NULL;
 	
-
 	memset(mpic, 0, sizeof(struct mpic));
 	mpic->name = name;
+	mpic->of_node = node ? of_node_get(node) : NULL;
 
+	mpic->irqhost = irq_alloc_host(IRQ_HOST_MAP_LINEAR, 256,
+				       &mpic_host_ops,
+				       MPIC_VEC_SPURRIOUS);
+	if (mpic->irqhost == NULL) {
+		of_node_put(node);
+		return NULL;
+	}
+
+	mpic->irqhost->host_data = mpic;
 	mpic->hc_irq = mpic_irq_chip;
 	mpic->hc_irq.typename = name;
 	if (flags & MPIC_PRIMARY)
@@ -628,18 +750,14 @@ struct mpic * __init mpic_alloc(unsigned long phys_addr,
 		mpic->hc_ht_irq.set_affinity = mpic_set_affinity;
 #endif /* CONFIG_MPIC_BROKEN_U3 */
 #ifdef CONFIG_SMP
-	mpic->hc_ipi.typename = name;
 	mpic->hc_ipi = mpic_ipi_chip;
+	mpic->hc_ipi.typename = name;
 #endif /* CONFIG_SMP */
 
 	mpic->flags = flags;
 	mpic->isu_size = isu_size;
-	mpic->irq_offset = irq_offset;
 	mpic->irq_count = irq_count;
-	mpic->ipi_offset = ipi_offset;
 	mpic->num_sources = 0; /* so far */
-	mpic->senses = senses;
-	mpic->senses_count = senses_count;
 
 	/* Map the global registers */
 	mpic->gregs = ioremap(phys_addr + MPIC_GREG_BASE, 0x1000);
@@ -707,8 +825,10 @@ struct mpic * __init mpic_alloc(unsigned long phys_addr,
 	mpic->next = mpics;
 	mpics = mpic;
 
-	if (flags & MPIC_PRIMARY)
+	if (flags & MPIC_PRIMARY) {
 		mpic_primary = mpic;
+		irq_set_default_host(mpic->irqhost);
+	}
 
 	return mpic;
 }
@@ -725,11 +845,22 @@ void __init mpic_assign_isu(struct mpic *mpic, unsigned int isu_num,
 		mpic->num_sources = isu_first + mpic->isu_size;
 }
 
+void __init mpic_set_default_senses(struct mpic *mpic, u8 *senses, int count)
+{
+	mpic->senses = senses;
+	mpic->senses_count = count;
+}
+
 void __init mpic_init(struct mpic *mpic)
 {
 	int i;
 
 	BUG_ON(mpic->num_sources == 0);
+	WARN_ON(mpic->num_sources > MPIC_VEC_IPI_0);
+
+	/* Sanitize source count */
+	if (mpic->num_sources > MPIC_VEC_IPI_0)
+		mpic->num_sources = MPIC_VEC_IPI_0;
 
 	printk(KERN_INFO "mpic: Initializing for %d sources\n", mpic->num_sources);
 
@@ -753,14 +884,6 @@ void __init mpic_init(struct mpic *mpic)
 			       MPIC_VECPRI_MASK |
 			       (10 << MPIC_VECPRI_PRIORITY_SHIFT) |
 			       (MPIC_VEC_IPI_0 + i));
-#ifdef CONFIG_SMP
-		if (!(mpic->flags & MPIC_PRIMARY))
-			continue;
-		set_irq_chip_data(mpic->ipi_offset+i, mpic);
-		set_irq_chip_and_handler(mpic->ipi_offset+i,
-					 &mpic->hc_ipi,
-					 handle_percpu_irq);
-#endif /* CONFIG_SMP */
 	}
 
 	/* Initialize interrupt sources */
@@ -777,25 +900,15 @@ void __init mpic_init(struct mpic *mpic)
 	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 = 0;
+		int level = 1;
 		
-		/* if it's an IPI, we skip it */
-		if ((mpic->irq_offset + i) >= (mpic->ipi_offset + i) &&
-		    (mpic->irq_offset + i) <  (mpic->ipi_offset + i + 4))
-			continue;
-
 		/* do senses munging */
-		if (mpic->senses && i < mpic->senses_count) {
-			if (mpic->senses[i] & IRQ_SENSE_LEVEL)
-				vecpri |= MPIC_VECPRI_SENSE_LEVEL;
-			if (mpic->senses[i] & IRQ_POLARITY_POSITIVE)
-				vecpri |= MPIC_VECPRI_POLARITY_POSITIVE;
-		} else
+		if (mpic->senses && i < mpic->senses_count)
+			vecpri = mpic_flags_to_vecpri(mpic->senses[i],
+						      &level);
+		else
 			vecpri |= MPIC_VECPRI_SENSE_LEVEL;
 
-		/* remember if it was a level interrupts */
-		level = (vecpri & MPIC_VECPRI_SENSE_LEVEL);
-
 		/* deal with broken U3 */
 		if (mpic->flags & MPIC_BROKEN_U3) {
 #ifdef CONFIG_MPIC_BROKEN_U3
@@ -816,21 +929,6 @@ void __init mpic_init(struct mpic *mpic)
 		mpic_irq_write(i, MPIC_IRQ_VECTOR_PRI, vecpri);
 		mpic_irq_write(i, MPIC_IRQ_DESTINATION,
 			       1 << hard_smp_processor_id());
-
-		/* init linux descriptors */
-		if (i < mpic->irq_count) {
-			struct irq_chip *chip = &mpic->hc_irq;
-
-			irq_desc[mpic->irq_offset+i].status |=
-				level ? IRQ_LEVEL : 0;
-#ifdef CONFIG_MPIC_BROKEN_U3
-			if (mpic_is_ht_interrupt(mpic, i))
-				chip = &mpic->hc_ht_irq;
-#endif /* CONFIG_MPIC_BROKEN_U3 */
-			set_irq_chip_data(mpic->irq_offset+i, mpic);
-			set_irq_chip_and_handler(mpic->irq_offset+i, chip,
-						 handle_fasteoi_irq);
-		}
 	}
 	
 	/* Init spurrious vector */
@@ -871,19 +969,20 @@ void mpic_irq_set_priority(unsigned int irq, unsigned int pri)
 {
 	int is_ipi;
 	struct mpic *mpic = mpic_find(irq, &is_ipi);
+	unsigned int src = mpic_irq_to_hw(irq);
 	unsigned long flags;
 	u32 reg;
 
 	spin_lock_irqsave(&mpic_lock, flags);
 	if (is_ipi) {
-		reg = mpic_ipi_read(irq - mpic->ipi_offset) &
+		reg = mpic_ipi_read(src - MPIC_VEC_IPI_0) &
 			~MPIC_VECPRI_PRIORITY_MASK;
-		mpic_ipi_write(irq - mpic->ipi_offset,
+		mpic_ipi_write(src - MPIC_VEC_IPI_0,
 			       reg | (pri << MPIC_VECPRI_PRIORITY_SHIFT));
 	} else {
-		reg = mpic_irq_read(irq - mpic->irq_offset,MPIC_IRQ_VECTOR_PRI)
+		reg = mpic_irq_read(src, MPIC_IRQ_VECTOR_PRI)
 			& ~MPIC_VECPRI_PRIORITY_MASK;
-		mpic_irq_write(irq - mpic->irq_offset, MPIC_IRQ_VECTOR_PRI,
+		mpic_irq_write(src, MPIC_IRQ_VECTOR_PRI,
 			       reg | (pri << MPIC_VECPRI_PRIORITY_SHIFT));
 	}
 	spin_unlock_irqrestore(&mpic_lock, flags);
@@ -893,14 +992,15 @@ unsigned int mpic_irq_get_priority(unsigned int irq)
 {
 	int is_ipi;
 	struct mpic *mpic = mpic_find(irq, &is_ipi);
+	unsigned int src = mpic_irq_to_hw(irq);
 	unsigned long flags;
 	u32 reg;
 
 	spin_lock_irqsave(&mpic_lock, flags);
 	if (is_ipi)
-		reg = mpic_ipi_read(irq - mpic->ipi_offset);
+		reg = mpic_ipi_read(src = MPIC_VEC_IPI_0);
 	else
-		reg = mpic_irq_read(irq - mpic->irq_offset, MPIC_IRQ_VECTOR_PRI);
+		reg = mpic_irq_read(src, MPIC_IRQ_VECTOR_PRI);
 	spin_unlock_irqrestore(&mpic_lock, flags);
 	return (reg & MPIC_VECPRI_PRIORITY_MASK) >> MPIC_VECPRI_PRIORITY_SHIFT;
 }
@@ -995,29 +1095,20 @@ void mpic_send_ipi(unsigned int ipi_no, unsigned int cpu_mask)
 		       mpic_physmask(cpu_mask & cpus_addr(cpu_online_map)[0]));
 }
 
-int mpic_get_one_irq(struct mpic *mpic, struct pt_regs *regs)
+unsigned int mpic_get_one_irq(struct mpic *mpic, struct pt_regs *regs)
 {
-	u32 irq;
+	u32 src;
 
-	irq = mpic_cpu_read(MPIC_CPU_INTACK) & MPIC_VECPRI_VECTOR_MASK;
+	src = mpic_cpu_read(MPIC_CPU_INTACK) & MPIC_VECPRI_VECTOR_MASK;
 #ifdef DEBUG_LOW
-	DBG("%s: get_one_irq(): %d\n", mpic->name, irq);
-#endif
-	if (unlikely(irq == MPIC_VEC_SPURRIOUS))
-		return -1;
-	if (irq < MPIC_VEC_IPI_0) {
-#ifdef DEBUG_IRQ
-		DBG("%s: irq %d\n", mpic->name, irq + mpic->irq_offset);
+	DBG("%s: get_one_irq(): %d\n", mpic->name, src);
 #endif
-		return irq + mpic->irq_offset;
-	}
-#ifdef DEBUG_IPI
-       	DBG("%s: ipi %d !\n", mpic->name, irq - MPIC_VEC_IPI_0);
-#endif
-	return irq - MPIC_VEC_IPI_0 + mpic->ipi_offset;
+	if (unlikely(src == MPIC_VEC_SPURRIOUS))
+		return NO_IRQ;
+	return irq_linear_revmap(mpic->irqhost, src);
 }
 
-int mpic_get_irq(struct pt_regs *regs)
+unsigned int mpic_get_irq(struct pt_regs *regs)
 {
 	struct mpic *mpic = mpic_primary;
 
@@ -1031,25 +1122,27 @@ int mpic_get_irq(struct pt_regs *regs)
 void mpic_request_ipis(void)
 {
 	struct mpic *mpic = mpic_primary;
-
+	int i;
+	static char *ipi_names[] = {
+		"IPI0 (call function)",
+		"IPI1 (reschedule)",
+		"IPI2 (unused)",
+		"IPI3 (debugger break)",
+	};
 	BUG_ON(mpic == NULL);
-	
-	printk("requesting IPIs ... \n");
 
-	/*
-	 * IPIs are marked IRQF_DISABLED as they must run with irqs
-	 * disabled
-	 */
-	request_irq(mpic->ipi_offset+0, mpic_ipi_action, IRQF_DISABLED,
-		    "IPI0 (call function)", mpic);
-	request_irq(mpic->ipi_offset+1, mpic_ipi_action, IRQF_DISABLED,
-		   "IPI1 (reschedule)", mpic);
-	request_irq(mpic->ipi_offset+2, mpic_ipi_action, IRQF_DISABLED,
-		   "IPI2 (unused)", mpic);
-	request_irq(mpic->ipi_offset+3, mpic_ipi_action, IRQF_DISABLED,
-		   "IPI3 (debugger break)", mpic);
-
-	printk("IPIs requested... \n");
+	printk(KERN_INFO "mpic: requesting IPIs ... \n");
+
+	for (i = 0; i < 4; i++) {
+		unsigned int vipi = irq_create_mapping(mpic->irqhost,
+						       MPIC_VEC_IPI_0 + i, 0);
+		if (vipi == NO_IRQ) {
+			printk(KERN_ERR "Failed to map IPI %d\n", i);
+			break;
+		}
+		request_irq(vipi, mpic_ipi_action, IRQF_DISABLED,
+			    ipi_names[i], mpic);
+	}
 }
 
 void smp_mpic_message_pass(int target, int msg)

drivers/char/hvsi.c

@@ -1299,13 +1299,12 @@ static int __init hvsi_console_init(void)
 		hp->inbuf_end = hp->inbuf;
 		hp->state = HVSI_CLOSED;
 		hp->vtermno = *vtermno;
-		hp->virq = virt_irq_create_mapping(irq[0]);
+		hp->virq = irq_create_mapping(NULL, irq[0], 0);
 		if (hp->virq == NO_IRQ) {
 			printk(KERN_ERR "%s: couldn't create irq mapping for 0x%x\n",
-				__FUNCTION__, hp->virq);
+				__FUNCTION__, irq[0]);
 			continue;
-		} else
-			hp->virq = irq_offset_up(hp->virq);
+		}
 
 		hvsi_count++;
 	}

drivers/macintosh/macio-adb.c

@@ -90,22 +90,12 @@ int macio_init(void)
 {
 	struct device_node *adbs;
 	struct resource r;
+	unsigned int irq;
 
 	adbs = find_compatible_devices("adb", "chrp,adb0");
 	if (adbs == 0)
 		return -ENXIO;
 
-#if 0
-	{ int i = 0;
-
-	printk("macio_adb_init: node = %p, addrs =", adbs->node);
-	while(!of_address_to_resource(adbs, i, &r))
-		printk(" %x(%x)", r.start, r.end - r.start);
-	printk(", intrs =");
-	for (i = 0; i < adbs->n_intrs; ++i)
-		printk(" %x", adbs->intrs[i].line);
-	printk("\n"); }
-#endif
 	if (of_address_to_resource(adbs, 0, &r))
 		return -ENXIO;
 	adb = ioremap(r.start, sizeof(struct adb_regs));
@@ -117,10 +107,9 @@ int macio_init(void)
 	out_8(&adb->active_lo.r, 0xff);
 	out_8(&adb->autopoll.r, APE);
 
-	if (request_irq(adbs->intrs[0].line, macio_adb_interrupt,
-			0, "ADB", (void *)0)) {
-		printk(KERN_ERR "ADB: can't get irq %d\n",
-		       adbs->intrs[0].line);
+	irq = irq_of_parse_and_map(adbs, 0);
+	if (request_irq(irq, macio_adb_interrupt, 0, "ADB", (void *)0)) {
+		printk(KERN_ERR "ADB: can't get irq %d\n", irq);
 		return -EAGAIN;
 	}
 	out_8(&adb->intr_enb.r, DFB | TAG);

drivers/macintosh/macio_asic.c

@@ -280,75 +280,128 @@ static void macio_release_dev(struct device *dev)
 static int macio_resource_quirks(struct device_node *np, struct resource *res,
 				 int index)
 {
-	if (res->flags & IORESOURCE_MEM) {
-		/* Grand Central has too large resource 0 on some machines */
-		if (index == 0 && !strcmp(np->name, "gc"))
-			res->end = res->start + 0x1ffff;
+	/* Only quirks for memory resources for now */
+	if ((res->flags & IORESOURCE_MEM) == 0)
+		return 0;
+
+	/* Grand Central has too large resource 0 on some machines */
+	if (index == 0 && !strcmp(np->name, "gc"))
+		res->end = res->start + 0x1ffff;
 
-		/* Airport has bogus resource 2 */
-		if (index >= 2 && !strcmp(np->name, "radio"))
-			return 1;
+	/* Airport has bogus resource 2 */
+	if (index >= 2 && !strcmp(np->name, "radio"))
+		return 1;
 
 #ifndef CONFIG_PPC64
-		/* DBDMAs may have bogus sizes */
-		if ((res->start & 0x0001f000) == 0x00008000)
-			res->end = res->start + 0xff;
+	/* DBDMAs may have bogus sizes */
+	if ((res->start & 0x0001f000) == 0x00008000)
+		res->end = res->start + 0xff;
 #endif /* CONFIG_PPC64 */
 
-		/* ESCC parent eats child resources. We could have added a
-		 * level of hierarchy, but I don't really feel the need
-		 * for it
-		 */
-		if (!strcmp(np->name, "escc"))
-			return 1;
-
-		/* ESCC has bogus resources >= 3 */
-		if (index >= 3 && !(strcmp(np->name, "ch-a") &&
-				    strcmp(np->name, "ch-b")))
-			return 1;
-
-		/* Media bay has too many resources, keep only first one */
-		if (index > 0 && !strcmp(np->name, "media-bay"))
-			return 1;
-
-		/* Some older IDE resources have bogus sizes */
-		if (!(strcmp(np->name, "IDE") && strcmp(np->name, "ATA") &&
-		      strcmp(np->type, "ide") && strcmp(np->type, "ata"))) {
-			if (index == 0 && (res->end - res->start) > 0xfff)
-				res->end = res->start + 0xfff;
-			if (index == 1 && (res->end - res->start) > 0xff)
-				res->end = res->start + 0xff;
-		}
+	/* ESCC parent eats child resources. We could have added a
+	 * level of hierarchy, but I don't really feel the need
+	 * for it
+	 */
+	if (!strcmp(np->name, "escc"))
+		return 1;
+
+	/* ESCC has bogus resources >= 3 */
+	if (index >= 3 && !(strcmp(np->name, "ch-a") &&
+			    strcmp(np->name, "ch-b")))
+		return 1;
+
+	/* Media bay has too many resources, keep only first one */
+	if (index > 0 && !strcmp(np->name, "media-bay"))
+		return 1;
+
+	/* Some older IDE resources have bogus sizes */
+	if (!(strcmp(np->name, "IDE") && strcmp(np->name, "ATA") &&
+	      strcmp(np->type, "ide") && strcmp(np->type, "ata"))) {
+		if (index == 0 && (res->end - res->start) > 0xfff)
+			res->end = res->start + 0xfff;
+		if (index == 1 && (res->end - res->start) > 0xff)
+			res->end = res->start + 0xff;
 	}
 	return 0;
 }
 
+static void macio_create_fixup_irq(struct macio_dev *dev, int index,
+				   unsigned int line)
+{
+	unsigned int irq;
 
-static void macio_setup_interrupts(struct macio_dev *dev)
+	irq = irq_create_mapping(NULL, line, 0);
+	if (irq != NO_IRQ) {
+		dev->interrupt[index].start = irq;
+		dev->interrupt[index].flags = IORESOURCE_IRQ;
+		dev->interrupt[index].name = dev->ofdev.dev.bus_id;
+	}
+	if (dev->n_interrupts <= index)
+		dev->n_interrupts = index + 1;
+}
+
+static void macio_add_missing_resources(struct macio_dev *dev)
 {
 	struct device_node *np = dev->ofdev.node;
-	int i,j;
+	unsigned int irq_base;
+
+	/* Gatwick has some missing interrupts on child nodes */
+	if (dev->bus->chip->type != macio_gatwick)
+		return;
 
-	/* For now, we use pre-parsed entries in the device-tree for
-	 * interrupt routing and addresses, but we should change that
-	 * to dynamically parsed entries and so get rid of most of the
-	 * clutter in struct device_node
+	/* irq_base is always 64 on gatwick. I have no cleaner way to get
+	 * that value from here at this point
 	 */
-	for (i = j = 0; i < np->n_intrs; i++) {
+	irq_base = 64;
+
+	/* Fix SCC */
+	if (strcmp(np->name, "ch-a") == 0) {
+		macio_create_fixup_irq(dev, 0, 15 + irq_base);
+		macio_create_fixup_irq(dev, 1,  4 + irq_base);
+		macio_create_fixup_irq(dev, 2,  5 + irq_base);
+		printk(KERN_INFO "macio: fixed SCC irqs on gatwick\n");
+	}
+
+	/* Fix media-bay */
+	if (strcmp(np->name, "media-bay") == 0) {
+		macio_create_fixup_irq(dev, 0, 29 + irq_base);
+		printk(KERN_INFO "macio: fixed media-bay irq on gatwick\n");
+	}
+
+	/* Fix left media bay childs */
+	if (dev->media_bay != NULL && strcmp(np->name, "floppy") == 0) {
+		macio_create_fixup_irq(dev, 0, 19 + irq_base);
+		macio_create_fixup_irq(dev, 1,  1 + irq_base);
+		printk(KERN_INFO "macio: fixed left floppy irqs\n");
+	}
+	if (dev->media_bay != NULL && strcasecmp(np->name, "ata4") == 0) {
+		macio_create_fixup_irq(dev, 0, 14 + irq_base);
+		macio_create_fixup_irq(dev, 0,  3 + irq_base);
+		printk(KERN_INFO "macio: fixed left ide irqs\n");
+	}
+}
+
+static void macio_setup_interrupts(struct macio_dev *dev)
+{
+	struct device_node *np = dev->ofdev.node;
+	unsigned int irq;
+	int i = 0, j = 0;
+
+	for (;;) {
 		struct resource *res = &dev->interrupt[j];
 
 		if (j >= MACIO_DEV_COUNT_IRQS)
 			break;
-		res->start = np->intrs[i].line;
-		res->flags = IORESOURCE_IO;
-		if (np->intrs[j].sense)
-			res->flags |= IORESOURCE_IRQ_LOWLEVEL;
-		else
-			res->flags |= IORESOURCE_IRQ_HIGHEDGE;
+		irq = irq_of_parse_and_map(np, i++);
+		if (irq == NO_IRQ)
+			break;
+		res->start = irq;
+		res->flags = IORESOURCE_IRQ;
 		res->name = dev->ofdev.dev.bus_id;
-		if (macio_resource_quirks(np, res, i))
+		if (macio_resource_quirks(np, res, i - 1)) {
 			memset(res, 0, sizeof(struct resource));
-		else
+			continue;
+		} else
 			j++;
 	}
 	dev->n_interrupts = j;
@@ -445,6 +498,7 @@ static struct macio_dev * macio_add_one_device(struct macio_chip *chip,
 	/* Setup interrupts & resources */
 	macio_setup_interrupts(dev);
 	macio_setup_resources(dev, parent_res);
+	macio_add_missing_resources(dev);
 
 	/* Register with core */
 	if (of_device_register(&dev->ofdev) != 0) {

drivers/macintosh/smu.c

@@ -497,8 +497,7 @@ int __init smu_init (void)
 	smu->doorbell = *data;
 	if (smu->doorbell < 0x50)
 		smu->doorbell += 0x50;
-	if (np->n_intrs > 0)
-		smu->db_irq = np->intrs[0].line;
+	smu->db_irq = irq_of_parse_and_map(np, 0);
 
 	of_node_put(np);
 
@@ -515,8 +514,7 @@ int __init smu_init (void)
 		smu->msg = *data;
 		if (smu->msg < 0x50)
 			smu->msg += 0x50;
-		if (np->n_intrs > 0)
-			smu->msg_irq = np->intrs[0].line;
+		smu->msg_irq = irq_of_parse_and_map(np, 0);
 		of_node_put(np);
 	} while(0);
 

drivers/macintosh/via-cuda.c

@@ -34,13 +34,6 @@
 static volatile unsigned char __iomem *via;
 static DEFINE_SPINLOCK(cuda_lock);
 
-#ifdef CONFIG_MAC
-#define CUDA_IRQ IRQ_MAC_ADB
-#define eieio()
-#else
-#define CUDA_IRQ vias->intrs[0].line
-#endif
-
 /* VIA registers - spaced 0x200 bytes apart */
 #define RS		0x200		/* skip between registers */
 #define B		0		/* B-side data */
@@ -189,11 +182,24 @@ int __init find_via_cuda(void)
 
 static int __init via_cuda_start(void)
 {
+    unsigned int irq;
+
     if (via == NULL)
 	return -ENODEV;
 
-    if (request_irq(CUDA_IRQ, cuda_interrupt, 0, "ADB", cuda_interrupt)) {
-	printk(KERN_ERR "cuda_init: can't get irq %d\n", CUDA_IRQ);
+#ifdef CONFIG_MAC
+    irq = IRQ_MAC_ADB;
+#else /* CONFIG_MAC */
+    irq = irq_of_parse_and_map(vias, 0);
+    if (irq == NO_IRQ) {
+	printk(KERN_ERR "via-cuda: can't map interrupts for %s\n",
+	       vias->full_name);
+	return -ENODEV;
+    }
+#endif /* CONFIG_MAP */
+
+    if (request_irq(irq, cuda_interrupt, 0, "ADB", cuda_interrupt)) {
+	printk(KERN_ERR "via-cuda: can't request irq %d\n", irq);
 	return -EAGAIN;
     }
 

drivers/macintosh/via-pmu.c

@@ -64,10 +64,6 @@
 #include <asm/backlight.h>
 #endif
 
-#ifdef CONFIG_PPC32
-#include <asm/open_pic.h>
-#endif
-
 #include "via-pmu-event.h"
 
 /* Some compile options */
@@ -151,7 +147,7 @@ static int pmu_fully_inited = 0;
 static int pmu_has_adb;
 static struct device_node *gpio_node;
 static unsigned char __iomem *gpio_reg = NULL;
-static int gpio_irq = -1;
+static int gpio_irq = NO_IRQ;
 static int gpio_irq_enabled = -1;
 static volatile int pmu_suspended = 0;
 static spinlock_t pmu_lock;
@@ -403,22 +399,21 @@ static int __init pmu_init(void)
  */
 static int __init via_pmu_start(void)
 {
+	unsigned int irq;
+
 	if (vias == NULL)
 		return -ENODEV;
 
 	batt_req.complete = 1;
 
-#ifndef CONFIG_PPC_MERGE
-	if (pmu_kind == PMU_KEYLARGO_BASED)
-		openpic_set_irq_priority(vias->intrs[0].line,
-					 OPENPIC_PRIORITY_DEFAULT + 1);
-#endif
-
-	if (request_irq(vias->intrs[0].line, via_pmu_interrupt, 0, "VIA-PMU",
-			(void *)0)) {
-		printk(KERN_ERR "VIA-PMU: can't get irq %d\n",
-		       vias->intrs[0].line);
-		return -EAGAIN;
+	irq = irq_of_parse_and_map(vias, 0);
+	if (irq == NO_IRQ) {
+		printk(KERN_ERR "via-pmu: can't map interruptn");
+		return -ENODEV;
+	}
+	if (request_irq(irq, via_pmu_interrupt, 0, "VIA-PMU", (void *)0)) {
+		printk(KERN_ERR "via-pmu: can't request irq %d\n", irq);
+		return -ENODEV;
 	}
 
 	if (pmu_kind == PMU_KEYLARGO_BASED) {
@@ -426,10 +421,10 @@ static int __init via_pmu_start(void)
 		if (gpio_node == NULL)
 			gpio_node = of_find_node_by_name(NULL,
 							 "pmu-interrupt");
-		if (gpio_node && gpio_node->n_intrs > 0)
-			gpio_irq = gpio_node->intrs[0].line;
+		if (gpio_node)
+			gpio_irq = irq_of_parse_and_map(gpio_node, 0);
 
-		if (gpio_irq != -1) {
+		if (gpio_irq != NO_IRQ) {
 			if (request_irq(gpio_irq, gpio1_interrupt, 0,
 					"GPIO1 ADB", (void *)0))
 				printk(KERN_ERR "pmu: can't get irq %d"

drivers/net/mace.c

@@ -242,12 +242,12 @@ static int __devinit mace_probe(struct macio_dev *mdev, const struct of_device_i
 	}
 	rc = request_irq(mp->tx_dma_intr, mace_txdma_intr, 0, "MACE-txdma", dev);
 	if (rc) {
-		printk(KERN_ERR "MACE: can't get irq %d\n", mace->intrs[1].line);
+		printk(KERN_ERR "MACE: can't get irq %d\n", mp->tx_dma_intr);
 		goto err_free_irq;
 	}
 	rc = request_irq(mp->rx_dma_intr, mace_rxdma_intr, 0, "MACE-rxdma", dev);
 	if (rc) {
-		printk(KERN_ERR "MACE: can't get irq %d\n", mace->intrs[2].line);
+		printk(KERN_ERR "MACE: can't get irq %d\n", mp->rx_dma_intr);
 		goto err_free_tx_irq;
 	}
 

drivers/serial/pmac_zilog.c

@@ -1443,8 +1443,8 @@ static int __init pmz_init_port(struct uart_pmac_port *uap)
 			uap->flags &= ~PMACZILOG_FLAG_HAS_DMA;
 			goto no_dma;
 		}
-		uap->tx_dma_irq = np->intrs[1].line;
-		uap->rx_dma_irq = np->intrs[2].line;
+		uap->tx_dma_irq = irq_of_parse_and_map(np, 1);
+		uap->rx_dma_irq = irq_of_parse_and_map(np, 2);
 	}
 no_dma:
 
@@ -1491,7 +1491,7 @@ no_dma:
 	 * Init remaining bits of "port" structure
 	 */
 	uap->port.iotype = UPIO_MEM;
-	uap->port.irq = np->intrs[0].line;
+	uap->port.irq = irq_of_parse_and_map(np, 0);
 	uap->port.uartclk = ZS_CLOCK;
 	uap->port.fifosize = 1;
 	uap->port.ops = &pmz_pops;

include/asm-powerpc/i8259.h

@@ -4,8 +4,13 @@
 
 #include <linux/irq.h>
 
+#ifdef CONFIG_PPC_MERGE
+extern void i8259_init(struct device_node *node, unsigned long intack_addr);
+extern unsigned int i8259_irq(struct pt_regs *regs);
+#else
 extern void i8259_init(unsigned long intack_addr, int offset);
 extern int i8259_irq(struct pt_regs *regs);
+#endif
 
 #endif /* __KERNEL__ */
 #endif /* _ASM_POWERPC_I8259_H */

include/asm-powerpc/irq.h

@@ -9,26 +9,14 @@
  * 2 of the License, or (at your option) any later version.
  */
 
+#include <linux/config.h>
 #include <linux/threads.h>
+#include <linux/list.h>
+#include <linux/radix-tree.h>
 
 #include <asm/types.h>
 #include <asm/atomic.h>
 
-/* this number is used when no interrupt has been assigned */
-#define NO_IRQ			(-1)
-
-/*
- * These constants are used for passing information about interrupt
- * signal polarity and level/edge sensing to the low-level PIC chip
- * drivers.
- */
-#define IRQ_SENSE_MASK		0x1
-#define IRQ_SENSE_LEVEL		0x1	/* interrupt on active level */
-#define IRQ_SENSE_EDGE		0x0	/* interrupt triggered by edge */
-
-#define IRQ_POLARITY_MASK	0x2
-#define IRQ_POLARITY_POSITIVE	0x2	/* high level or low->high edge */
-#define IRQ_POLARITY_NEGATIVE	0x0	/* low level or high->low edge */
 
 #define get_irq_desc(irq) (&irq_desc[(irq)])
 
@@ -36,50 +24,325 @@
 #define for_each_irq(i) \
 	for ((i) = 0; (i) < NR_IRQS; ++(i))
 
-#ifdef CONFIG_PPC64
+extern atomic_t ppc_n_lost_interrupts;
 
-/*
- * Maximum number of interrupt sources that we can handle.
+#ifdef CONFIG_PPC_MERGE
+
+/* This number is used when no interrupt has been assigned */
+#define NO_IRQ			(0)
+
+/* This is a special irq number to return from get_irq() to tell that
+ * no interrupt happened _and_ ignore it (don't count it as bad). Some
+ * platforms like iSeries rely on that.
  */
+#define NO_IRQ_IGNORE		((unsigned int)-1)
+
+/* Total number of virq in the platform (make it a CONFIG_* option ? */
 #define NR_IRQS		512
 
-/* Interrupt numbers are virtual in case they are sparsely
- * distributed by the hardware.
+/* Number of irqs reserved for the legacy controller */
+#define NUM_ISA_INTERRUPTS	16
+
+/* This type is the placeholder for a hardware interrupt number. It has to
+ * be big enough to enclose whatever representation is used by a given
+ * platform.
+ */
+typedef unsigned long irq_hw_number_t;
+
+/* Interrupt controller "host" data structure. This could be defined as a
+ * irq domain controller. That is, it handles the mapping between hardware
+ * and virtual interrupt numbers for a given interrupt domain. The host
+ * structure is generally created by the PIC code for a given PIC instance
+ * (though a host can cover more than one PIC if they have a flat number
+ * model). It's the host callbacks that are responsible for setting the
+ * irq_chip on a given irq_desc after it's been mapped.
+ *
+ * The host code and data structures are fairly agnostic to the fact that
+ * we use an open firmware device-tree. We do have references to struct
+ * device_node in two places: in irq_find_host() to find the host matching
+ * a given interrupt controller node, and of course as an argument to its
+ * counterpart host->ops->match() callback. However, those are treated as
+ * generic pointers by the core and the fact that it's actually a device-node
+ * pointer is purely a convention between callers and implementation. This
+ * code could thus be used on other architectures by replacing those two
+ * by some sort of arch-specific void * "token" used to identify interrupt
+ * controllers.
+ */
+struct irq_host;
+struct radix_tree_root;
+
+/* Functions below are provided by the host and called whenever a new mapping
+ * is created or an old mapping is disposed. The host can then proceed to
+ * whatever internal data structures management is required. It also needs
+ * to setup the irq_desc when returning from map().
+ */
+struct irq_host_ops {
+	/* Match an interrupt controller device node to a host, returns
+	 * 1 on a match
+	 */
+	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.
+	 */
+	int (*map)(struct irq_host *h, unsigned int virq,
+		   irq_hw_number_t hw, unsigned int flags);
+
+	/* 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.
+	 */
+	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);
+};
+
+struct irq_host {
+	struct list_head	link;
+
+	/* type of reverse mapping technique */
+	unsigned int		revmap_type;
+#define IRQ_HOST_MAP_LEGACY     0 /* legacy 8259, gets irqs 1..15 */
+#define IRQ_HOST_MAP_NOMAP	1 /* no fast reverse mapping */
+#define IRQ_HOST_MAP_LINEAR	2 /* linear map of interrupts */
+#define IRQ_HOST_MAP_TREE	3 /* radix tree */
+	union {
+		struct {
+			unsigned int size;
+			unsigned int *revmap;
+		} linear;
+		struct radix_tree_root tree;
+	} revmap_data;
+	struct irq_host_ops	*ops;
+	void			*host_data;
+	irq_hw_number_t		inval_irq;
+};
+
+/* The main irq map itself is an array of NR_IRQ entries containing the
+ * associate host and irq number. An entry with a host of NULL is free.
+ * An entry can be allocated if it's free, the allocator always then sets
+ * hwirq first to the host's invalid irq number and then fills ops.
+ */
+struct irq_map_entry {
+	irq_hw_number_t	hwirq;
+	struct irq_host	*host;
+};
+
+extern struct irq_map_entry irq_map[NR_IRQS];
+
+
+/***
+ * irq_alloc_host - Allocate a new irq_host data structure
+ * @node: device-tree node of the interrupt controller
+ * @revmap_type: type of reverse mapping to use
+ * @revmap_arg: for IRQ_HOST_MAP_LINEAR linear only: size of the map
+ * @ops: map/unmap host callbacks
+ * @inval_irq: provide a hw number in that host space that is always invalid
+ *
+ * Allocates and initialize and irq_host structure. Note that in the case of
+ * IRQ_HOST_MAP_LEGACY, the map() callback will be called before this returns
+ * for all legacy interrupts except 0 (which is always the invalid irq for
+ * a legacy controller). For a IRQ_HOST_MAP_LINEAR, the map is allocated by
+ * this call as well. For a IRQ_HOST_MAP_TREE, the radix tree will be allocated
+ * later during boot automatically (the reverse mapping will use the slow path
+ * until that happens).
+ */
+extern struct irq_host *irq_alloc_host(unsigned int revmap_type,
+				       unsigned int revmap_arg,
+				       struct irq_host_ops *ops,
+				       irq_hw_number_t inval_irq);
+
+
+/***
+ * irq_find_host - Locates a host for a given device node
+ * @node: device-tree node of the interrupt controller
+ */
+extern struct irq_host *irq_find_host(struct device_node *node);
+
+
+/***
+ * irq_set_default_host - Set a "default" host
+ * @host: default host pointer
+ *
+ * For convenience, it's possible to set a "default" host that will be used
+ * whenever NULL is passed to irq_create_mapping(). It makes life easier for
+ * platforms that want to manipulate a few hard coded interrupt numbers that
+ * aren't properly represented in the device-tree.
+ */
+extern void irq_set_default_host(struct irq_host *host);
+
+
+/***
+ * irq_set_virq_count - Set the maximum number of virt irqs
+ * @count: number of linux virtual irqs, capped with NR_IRQS
+ *
+ * This is mainly for use by platforms like iSeries who want to program
+ * the virtual irq number in the controller to avoid the reverse mapping
+ */
+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.
+ */
+extern unsigned int irq_create_mapping(struct irq_host *host,
+				       irq_hw_number_t hwirq,
+				       unsigned int flags);
+
+
+/***
+ * irq_dispose_mapping - Unmap an interrupt
+ * @virq: linux virq number of the interrupt to unmap
  */
-extern unsigned int virt_irq_to_real_map[NR_IRQS];
+extern void irq_dispose_mapping(unsigned int virq);
 
-/* The maximum virtual IRQ number that we support.  This
- * can be set by the platform and will be reduced by the
- * value of __irq_offset_value.  It defaults to and is
- * capped by (NR_IRQS - 1).
+/***
+ * irq_find_mapping - Find a linux virq from an hw irq number.
+ * @host: host owning this hardware interrupt
+ * @hwirq: hardware irq number in that host space
+ *
+ * This is a slow path, for use by generic code. It's expected that an
+ * irq controller implementation directly calls the appropriate low level
+ * mapping function.
  */
-extern unsigned int virt_irq_max;
+extern unsigned int irq_find_mapping(struct irq_host *host,
+				     irq_hw_number_t hwirq);
 
-/* Create a mapping for a real_irq if it doesn't already exist.
- * Return the virtual irq as a convenience.
+
+/***
+ * irq_radix_revmap - Find a linux virq from a hw irq number.
+ * @host: host owning this hardware interrupt
+ * @hwirq: hardware irq number in that host space
+ *
+ * This is a fast path, for use by irq controller code that uses radix tree
+ * revmaps
+ */
+extern unsigned int irq_radix_revmap(struct irq_host *host,
+				     irq_hw_number_t hwirq);
+
+/***
+ * irq_linear_revmap - Find a linux virq from a hw irq number.
+ * @host: host owning this hardware interrupt
+ * @hwirq: hardware irq number in that host space
+ *
+ * This is a fast path, for use by irq controller code that uses linear
+ * revmaps. It does fallback to the slow path if the revmap doesn't exist
+ * yet and will create the revmap entry with appropriate locking
+ */
+
+extern unsigned int irq_linear_revmap(struct irq_host *host,
+				      irq_hw_number_t hwirq);
+
+
+
+/***
+ * irq_alloc_virt - Allocate virtual irq numbers
+ * @host: host owning these new virtual irqs
+ * @count: number of consecutive numbers to allocate
+ * @hint: pass a hint number, the allocator will try to use a 1:1 mapping
+ *
+ * This is a low level function that is used internally by irq_create_mapping()
+ * and that can be used by some irq controllers implementations for things
+ * like allocating ranges of numbers for MSIs. The revmaps are left untouched.
  */
-int virt_irq_create_mapping(unsigned int real_irq);
-void virt_irq_init(void);
+extern unsigned int irq_alloc_virt(struct irq_host *host,
+				   unsigned int count,
+				   unsigned int hint);
+
+/***
+ * irq_free_virt - Free virtual irq numbers
+ * @virq: virtual irq number of the first interrupt to free
+ * @count: number of interrupts to free
+ *
+ * This function is the opposite of irq_alloc_virt. It will not clear reverse
+ * maps, this should be done previously by unmap'ing the interrupt. In fact,
+ * all interrupts covered by the range being freed should have been unmapped
+ * prior to calling this.
+ */
+extern void irq_free_virt(unsigned int virq, unsigned int count);
+
+
+/* -- OF helpers -- */
+
+/* irq_create_of_mapping - Map a hardware interrupt into linux virq space
+ * @controller: Device node of the interrupt controller
+ * @inspec: Interrupt specifier from the device-tree
+ * @intsize: Size of the interrupt specifier from the device-tree
+ *
+ * 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
+ */
+extern unsigned int irq_create_of_mapping(struct device_node *controller,
+					  u32 *intspec, unsigned int intsize);
+
+
+/* irq_of_parse_and_map - Parse nad Map an interrupt into linux virq space
+ * @device: Device node of the device whose interrupt is to be mapped
+ * @index: Index of the interrupt to map
+ *
+ * This function is a wrapper that chains of_irq_map_one() and
+ * irq_create_of_mapping() to make things easier to callers
+ */
+extern unsigned int irq_of_parse_and_map(struct device_node *dev, int index);
+
+/* -- End OF helpers -- */
 
-static inline unsigned int virt_irq_to_real(unsigned int virt_irq)
+/***
+ * irq_early_init - Init irq remapping subsystem
+ */
+extern void irq_early_init(void);
+
+static __inline__ int irq_canonicalize(int irq)
 {
-	return virt_irq_to_real_map[virt_irq];
+	return irq;
 }
 
-extern unsigned int real_irq_to_virt_slowpath(unsigned int real_irq);
+
+#else /* CONFIG_PPC_MERGE */
+
+/* This number is used when no interrupt has been assigned */
+#define NO_IRQ			(-1)
+#define NO_IRQ_IGNORE		(-2)
+
 
 /*
- * List of interrupt controllers.
+ * These constants are used for passing information about interrupt
+ * signal polarity and level/edge sensing to the low-level PIC chip
+ * drivers.
  */
-#define IC_INVALID    0
-#define IC_OPEN_PIC   1
-#define IC_PPC_XIC    2
-#define IC_CELL_PIC   3
-#define IC_ISERIES    4
+#define IRQ_SENSE_MASK		0x1
+#define IRQ_SENSE_LEVEL		0x1	/* interrupt on active level */
+#define IRQ_SENSE_EDGE		0x0	/* interrupt triggered by edge */
 
-extern u64 ppc64_interrupt_controller;
+#define IRQ_POLARITY_MASK	0x2
+#define IRQ_POLARITY_POSITIVE	0x2	/* high level or low->high edge */
+#define IRQ_POLARITY_NEGATIVE	0x0	/* low level or high->low edge */
 
-#else /* 32-bit */
 
 #if defined(CONFIG_40x)
 #include <asm/ibm4xx.h>
@@ -512,19 +775,11 @@ extern u64 ppc64_interrupt_controller;
 
 #endif /* CONFIG_8260 */
 
-#endif
+#endif /* Whatever way too big #ifdef */
 
-#ifndef CONFIG_PPC_MERGE
 #define NR_MASK_WORDS	((NR_IRQS + 31) / 32)
 /* pedantic: these are long because they are used with set_bit --RR */
 extern unsigned long ppc_cached_irq_mask[NR_MASK_WORDS];
-#endif
-
-extern atomic_t ppc_n_lost_interrupts;
-
-#define virt_irq_create_mapping(x)	(x)
-
-#endif
 
 /*
  * Because many systems have two overlapping names spaces for
@@ -563,6 +818,7 @@ static __inline__ int irq_canonicalize(int irq)
 		irq = 9;
 	return irq;
 }
+#endif /* CONFIG_PPC_MERGE */
 
 extern int distribute_irqs;
 

include/asm-powerpc/machdep.h

@@ -97,7 +97,7 @@ struct machdep_calls {
 	void		(*show_percpuinfo)(struct seq_file *m, int i);
 
 	void		(*init_IRQ)(void);
-	int		(*get_irq)(struct pt_regs *);
+	unsigned int	(*get_irq)(struct pt_regs *);
 #ifdef CONFIG_KEXEC
 	void		(*kexec_cpu_down)(int crash_shutdown, int secondary);
 #endif

include/asm-powerpc/mpic.h

@@ -129,6 +129,12 @@ struct mpic_irq_fixup
 /* The instance data of a given MPIC */
 struct mpic
 {
+	/* The device node of the interrupt controller */
+	struct device_node	*of_node;
+
+	/* The remapper for this MPIC */
+	struct irq_host		*irqhost;
+
 	/* The "linux" controller struct */
 	struct irq_chip		hc_irq;
 #ifdef CONFIG_MPIC_BROKEN_U3
@@ -144,16 +150,12 @@ struct mpic
 	unsigned int		isu_size;
 	unsigned int		isu_shift;
 	unsigned int		isu_mask;
-	/* Offset of irq vector numbers */
-	unsigned int		irq_offset;	
 	unsigned int		irq_count;
-	/* Offset of ipi vector numbers */
-	unsigned int		ipi_offset;
 	/* Number of sources */
 	unsigned int		num_sources;
 	/* Number of CPUs */
 	unsigned int		num_cpus;
-	/* senses array */
+	/* default senses array */
 	unsigned char		*senses;
 	unsigned int		senses_count;
 
@@ -209,14 +211,11 @@ struct mpic
  * The values in the array start at the first source of the MPIC,
  * that is senses[0] correspond to linux irq "irq_offset".
  */
-extern struct mpic *mpic_alloc(unsigned long phys_addr,
+extern struct mpic *mpic_alloc(struct device_node *node,
+			       unsigned long phys_addr,
 			       unsigned int flags,
 			       unsigned int isu_size,
-			       unsigned int irq_offset,
 			       unsigned int irq_count,
-			       unsigned int ipi_offset,
-			       unsigned char *senses,
-			       unsigned int senses_num,
 			       const char *name);
 
 /* Assign ISUs, to call before mpic_init()
@@ -228,6 +227,22 @@ extern struct mpic *mpic_alloc(unsigned long phys_addr,
 extern void mpic_assign_isu(struct mpic *mpic, unsigned int isu_num,
 			    unsigned long phys_addr);
 
+/* Set default sense codes
+ *
+ * @mpic:	controller
+ * @senses:	array of sense codes
+ * @count:	size of above array
+ *
+ * Optionally provide an array (indexed on hardware interrupt numbers
+ * for this MPIC) of default sense codes for the chip. Those are linux
+ * sense codes IRQ_TYPE_*
+ *
+ * The driver gets ownership of the pointer, don't dispose of it or
+ * anything like that. __init only.
+ */
+extern void mpic_set_default_senses(struct mpic *mpic, u8 *senses, int count);
+
+
 /* Initialize the controller. After this has been called, none of the above
  * should be called again for this mpic
  */
@@ -269,9 +284,9 @@ extern void mpic_send_ipi(unsigned int ipi_no, unsigned int cpu_mask);
 void smp_mpic_message_pass(int target, int msg);
 
 /* Fetch interrupt from a given mpic */
-extern int mpic_get_one_irq(struct mpic *mpic, struct pt_regs *regs);
+extern unsigned int mpic_get_one_irq(struct mpic *mpic, struct pt_regs *regs);
 /* This one gets to the primary mpic */
-extern int mpic_get_irq(struct pt_regs *regs);
+extern unsigned int mpic_get_irq(struct pt_regs *regs);
 
 /* Set the EPIC clock ratio */
 void mpic_set_clk_ratio(struct mpic *mpic, u32 clock_ratio);
@@ -279,8 +294,5 @@ void mpic_set_clk_ratio(struct mpic *mpic, u32 clock_ratio);
 /* Enable/Disable EPIC serial interrupt mode */
 void mpic_set_serial_int(struct mpic *mpic, int enable);
 
-/* global mpic for pSeries */
-extern struct mpic *pSeries_mpic;
-
 #endif /* __KERNEL__ */
 #endif	/* _ASM_POWERPC_MPIC_H */

include/asm-powerpc/prom.h

@@ -64,11 +64,6 @@ struct boot_param_header
 typedef u32 phandle;
 typedef u32 ihandle;
 
-struct interrupt_info {
-	int	line;
-	int	sense;		/* +ve/-ve logic, edge or level, etc. */
-};
-
 struct property {
 	char	*name;
 	int	length;
@@ -81,8 +76,6 @@ struct device_node {
 	char	*type;
 	phandle	node;
 	phandle linux_phandle;
-	int	n_intrs;
-	struct	interrupt_info *intrs;
 	char	*full_name;
 
 	struct	property *properties;

include/asm-powerpc/spu.h

@@ -117,6 +117,7 @@ struct spu {
 	struct list_head sched_list;
 	int number;
 	int nid;
+	unsigned int irqs[3];
 	u32 isrc;
 	u32 node;
 	u64 flags;

sound/aoa/core/snd-aoa-gpio-feature.c

@@ -112,12 +112,7 @@ static struct device_node *get_gpio(char *name,
 
 static void get_irq(struct device_node * np, int *irqptr)
 {
-	*irqptr = -1;
-	if (!np)
-		return;
-	if (np->n_intrs != 1)
-		return;
-	*irqptr = np->intrs[0].line;
+	*irqptr = irq_of_parse_and_map(np, 0);
 }
 
 /* 0x4 is outenable, 0x1 is out, thus 4 or 5 */

sound/aoa/soundbus/i2sbus/i2sbus-core.c

@@ -129,7 +129,7 @@ static int i2sbus_add_dev(struct macio_dev *macio,
 	if (strncmp(np->name, "i2s-", 4))
 		return 0;
 
-	if (np->n_intrs != 3)
+	if (macio_irq_count(macio) != 3)
 		return 0;
 
 	dev = kzalloc(sizeof(struct i2sbus_dev), GFP_KERNEL);
@@ -183,9 +183,10 @@ static int i2sbus_add_dev(struct macio_dev *macio,
 		snprintf(dev->rnames[i], sizeof(dev->rnames[i]), rnames[i], np->name);
 	}
 	for (i=0;i<3;i++) {
-		if (request_irq(np->intrs[i].line, ints[i], 0, dev->rnames[i], dev))
+		if (request_irq(macio_irq(macio, i), ints[i], 0,
+				dev->rnames[i], dev))
 			goto err;
-		dev->interrupts[i] = np->intrs[i].line;
+		dev->interrupts[i] = macio_irq(macio, i);
 	}
 
 	for (i=0;i<3;i++) {

sound/oss/dmasound/dmasound_awacs.c

@@ -374,10 +374,7 @@ setup_audio_gpio(const char *name, const char* compatible, int *gpio_addr, int*
 		*gpio_pol = *pp;
 	else
 		*gpio_pol = 1;
-	if (np->n_intrs > 0)
-		return np->intrs[0].line;
-	
-	return 0;
+	return irq_of_parse_and_map(np, 0);
 }
 
 static inline void
@@ -2864,14 +2861,13 @@ printk("dmasound_pmac: couldn't find a Codec we can handle\n");
 	 * other info if necessary (early AWACS we want to read chip ids)
 	 */
 
-	if (of_get_address(io, 2, NULL, NULL) == NULL || io->n_intrs < 3) {
+	if (of_get_address(io, 2, NULL, NULL) == NULL) {
 		/* OK - maybe we need to use the 'awacs' node (on earlier
 		 * machines).
 		 */
 		if (awacs_node) {
 			io = awacs_node ;
-			if (of_get_address(io, 2, NULL, NULL) == NULL ||
-			    io->n_intrs < 3) {
+			if (of_get_address(io, 2, NULL, NULL) == NULL) {
 				printk("dmasound_pmac: can't use %s\n",
 				       io->full_name);
 				return -ENODEV;
@@ -2940,9 +2936,9 @@ printk("dmasound_pmac: couldn't find a Codec we can handle\n");
 	if (awacs_revision == AWACS_SCREAMER && awacs)
 		awacs_recalibrate();
 
-	awacs_irq = io->intrs[0].line;
-	awacs_tx_irq = io->intrs[1].line;
-	awacs_rx_irq = io->intrs[2].line;
+	awacs_irq = irq_of_parse_and_map(io, 0);
+	awacs_tx_irq = irq_of_parse_and_map(io, 1);
+	awacs_rx_irq = irq_of_parse_and_map(io, 2);
 
 	/* Hack for legacy crap that will be killed someday */
 	awacs_node = io;

sound/ppc/pmac.c

@@ -1120,6 +1120,7 @@ int __init snd_pmac_new(struct snd_card *card, struct snd_pmac **chip_return)
 	struct snd_pmac *chip;
 	struct device_node *np;
 	int i, err;
+	unsigned int irq;
 	unsigned long ctrl_addr, txdma_addr, rxdma_addr;
 	static struct snd_device_ops ops = {
 		.dev_free =	snd_pmac_dev_free,
@@ -1153,10 +1154,6 @@ int __init snd_pmac_new(struct snd_card *card, struct snd_pmac **chip_return)
 	if (chip->is_k2) {
 		static char *rnames[] = {
 			"Sound Control", "Sound DMA" };
-		if (np->n_intrs < 3) {
-			err = -ENODEV;
-			goto __error;
-		}
 		for (i = 0; i < 2; i ++) {
 			if (of_address_to_resource(np->parent, i,
 						   &chip->rsrc[i])) {
@@ -1185,10 +1182,6 @@ int __init snd_pmac_new(struct snd_card *card, struct snd_pmac **chip_return)
 	} else {
 		static char *rnames[] = {
 			"Sound Control", "Sound Tx DMA", "Sound Rx DMA" };
-		if (np->n_intrs < 3) {
-			err = -ENODEV;
-			goto __error;
-		}
 		for (i = 0; i < 3; i ++) {
 			if (of_address_to_resource(np, i,
 						   &chip->rsrc[i])) {
@@ -1220,28 +1213,30 @@ int __init snd_pmac_new(struct snd_card *card, struct snd_pmac **chip_return)
 	chip->playback.dma = ioremap(txdma_addr, 0x100);
 	chip->capture.dma = ioremap(rxdma_addr, 0x100);
 	if (chip->model <= PMAC_BURGUNDY) {
-		if (request_irq(np->intrs[0].line, snd_pmac_ctrl_intr, 0,
+		irq = irq_of_parse_and_map(np, 0);
+		if (request_irq(irq, snd_pmac_ctrl_intr, 0,
 				"PMac", (void*)chip)) {
-			snd_printk(KERN_ERR "pmac: unable to grab IRQ %d\n", np->intrs[0].line);
+			snd_printk(KERN_ERR "pmac: unable to grab IRQ %d\n",
+				   irq);
 			err = -EBUSY;
 			goto __error;
 		}
-		chip->irq = np->intrs[0].line;
+		chip->irq = irq;
 	}
-	if (request_irq(np->intrs[1].line, snd_pmac_tx_intr, 0,
-			"PMac Output", (void*)chip)) {
-		snd_printk(KERN_ERR "pmac: unable to grab IRQ %d\n", np->intrs[1].line);
+	irq = irq_of_parse_and_map(np, 1);
+	if (request_irq(irq, snd_pmac_tx_intr, 0, "PMac Output", (void*)chip)){
+		snd_printk(KERN_ERR "pmac: unable to grab IRQ %d\n", irq);
 		err = -EBUSY;
 		goto __error;
 	}
-	chip->tx_irq = np->intrs[1].line;
-	if (request_irq(np->intrs[2].line, snd_pmac_rx_intr, 0,
-			"PMac Input", (void*)chip)) {
-		snd_printk(KERN_ERR "pmac: unable to grab IRQ %d\n", np->intrs[2].line);
+	chip->tx_irq = irq;
+	irq = irq_of_parse_and_map(np, 2);
+	if (request_irq(irq, snd_pmac_rx_intr, 0, "PMac Input", (void*)chip)) {
+		snd_printk(KERN_ERR "pmac: unable to grab IRQ %d\n", irq);
 		err = -EBUSY;
 		goto __error;
 	}
-	chip->rx_irq = np->intrs[2].line;
+	chip->rx_irq = irq;
 
 	snd_pmac_sound_feature(chip, 1);
 

sound/ppc/tumbler.c

@@ -1121,7 +1121,7 @@ static long tumbler_find_device(const char *device, const char *platform,
 	DBG("(I) GPIO device %s found, offset: %x, active state: %d !\n",
 	    device, gp->addr, gp->active_state);
 
-	return (node->n_intrs > 0) ? node->intrs[0].line : 0;
+	return irq_of_parse_and_map(node, 0);
 }
 
 /* reset audio */
@@ -1264,16 +1264,16 @@ static int __init tumbler_init(struct snd_pmac *chip)
 				    &mix->line_mute, 1);
 	irq = tumbler_find_device("headphone-detect",
 				  NULL, &mix->hp_detect, 0);
-	if (irq < 0)
+	if (irq <= NO_IRQ)
 		irq = tumbler_find_device("headphone-detect",
 					  NULL, &mix->hp_detect, 1);
-	if (irq < 0)
+	if (irq <= NO_IRQ)
 		irq = tumbler_find_device("keywest-gpio15",
 					  NULL, &mix->hp_detect, 1);
 	mix->headphone_irq = irq;
  	irq = tumbler_find_device("line-output-detect",
 				  NULL, &mix->line_detect, 0);
- 	if (irq < 0)
+ 	if (irq <= NO_IRQ)
 		irq = tumbler_find_device("line-output-detect",
 					  NULL, &mix->line_detect, 1);
 	mix->lineout_irq = irq;