x86: make 32 bit to use sparse_irq

but actually irq still needs to be less than NR_IRQS, because
interrupt[NR_IRQS] in entry.S.

need to enable per_cpu vector...

Signed-off-by: Yinghai Lu <yhlu.kernel@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

3 files changed, 78 insertions, 29 deletions

diff --git a/arch/x86/kernel/io_apic_32.c b/arch/x86/kernel/io_apic_32.c
index 59d2e8a273e3..66c0a91362a7 100644
--- a/arch/x86/kernel/io_apic_32.c
+++ b/arch/x86/kernel/io_apic_32.c
@@ -595,7 +595,6 @@ static void set_ioapic_affinity_irq(unsigned int irq, cpumask_t cpumask)
 	struct irq_pin_list *entry;
 	unsigned int apicid_value;
 	cpumask_t tmp;
-	struct irq_desc *desc;
 
 
 	cfg = irq_cfg(irq);
@@ -620,8 +619,7 @@ static void set_ioapic_affinity_irq(unsigned int irq, cpumask_t cpumask)
 			break;
 		entry = entry->next;
 	}
-	desc = irq_to_desc(irq);
-	desc->affinity = cpumask;
+	irq_to_desc(irq)->affinity = cpumask;
 	spin_unlock_irqrestore(&ioapic_lock, flags);
 }
 
@@ -1055,8 +1053,6 @@ static int __assign_irq_vector(int irq)
 	int vector, offset;
 	struct irq_cfg *cfg;
 
-	BUG_ON((unsigned)irq >= nr_irqs);
-
 	cfg = irq_cfg(irq);
 	if (cfg->vector > 0)
 		return cfg->vector;
@@ -1103,7 +1099,12 @@ static void ioapic_register_intr(int irq, int vector, unsigned long trigger)
 {
 	struct irq_desc *desc;
 
-	desc = irq_to_desc(irq);
+	/* first time to use this irq_desc */
+	if (irq < 16)
+		desc = irq_to_desc(irq);
+	else
+		desc = irq_to_desc_alloc(irq);
+
 	if ((trigger == IOAPIC_AUTO && IO_APIC_irq_trigger(irq)) ||
 	    trigger == IOAPIC_LEVEL) {
 		desc->status |= IRQ_LEVEL;
@@ -2330,16 +2331,19 @@ device_initcall(ioapic_init_sysfs);
 /*
  * Dynamic irq allocate and deallocation
  */
-int create_irq(void)
+unsigned int create_irq_nr(unsigned int irq_want)
 {
 	/* Allocate an unused irq */
-	int irq, new, vector = 0;
+	unsigned int irq, new, vector = 0;
 	unsigned long flags;
 	struct irq_cfg *cfg_new;
 
-	irq = -ENOSPC;
+	/* only can use bus/dev/fn.. when per_cpu vector is used */
+	irq_want = nr_irqs - 1;
+
+	irq = 0;
 	spin_lock_irqsave(&vector_lock, flags);
-	for (new = (nr_irqs - 1); new >= 0; new--) {
+	for (new = (nr_irqs - 1); new > 0; new--) {
 		if (platform_legacy_irq(new))
 			continue;
 		cfg_new = irq_cfg(new);
@@ -2354,13 +2358,18 @@ int create_irq(void)
 	}
 	spin_unlock_irqrestore(&vector_lock, flags);
 
-	if (irq >= 0) {
+	if (irq > 0) {
 		set_intr_gate(vector, interrupt[irq]);
 		dynamic_irq_init(irq);
 	}
 	return irq;
 }
 
+int create_irq(void)
+{
+	return create_irq_nr(nr_irqs - 1);
+}
+
 void destroy_irq(unsigned int irq)
 {
 	unsigned long flags;
@@ -2415,7 +2424,6 @@ static void set_msi_irq_affinity(unsigned int irq, cpumask_t mask)
 	unsigned int dest;
 	cpumask_t tmp;
 	int vector;
-	struct irq_desc *desc;
 
 	cpus_and(tmp, mask, cpu_online_map);
 	if (cpus_empty(tmp))
@@ -2435,8 +2443,7 @@ static void set_msi_irq_affinity(unsigned int irq, cpumask_t mask)
 	msg.address_lo |= MSI_ADDR_DEST_ID(dest);
 
 	write_msi_msg(irq, &msg);
-	desc = irq_to_desc(irq);
-	desc->affinity = mask;
+	irq_to_desc(irq)->affinity = mask;
 }
 #endif /* CONFIG_SMP */
 
@@ -2455,13 +2462,31 @@ static struct irq_chip msi_chip = {
 	.retrigger	= ioapic_retrigger_irq,
 };
 
+static unsigned int build_irq_for_pci_dev(struct pci_dev *dev)
+{
+	unsigned int irq;
+
+	irq = dev->bus->number;
+	irq <<= 8;
+	irq |= dev->devfn;
+	irq <<= 12;
+
+	return irq;
+}
+
 int arch_setup_msi_irq(struct pci_dev *dev, struct msi_desc *desc)
 {
 	struct msi_msg msg;
 	int irq, ret;
-	irq = create_irq();
-	if (irq < 0)
-		return irq;
+
+	unsigned int irq_want;
+
+	irq_want = build_irq_for_pci_dev(dev) + 0x100;
+
+	irq = create_irq_nr(irq_want);
+
+	if (irq == 0)
+		return -1;
 
 	ret = msi_compose_msg(dev, irq, &msg);
 	if (ret < 0) {
@@ -2510,7 +2535,6 @@ static void set_ht_irq_affinity(unsigned int irq, cpumask_t mask)
 {
 	unsigned int dest;
 	cpumask_t tmp;
-	struct irq_desc *desc;
 
 	cpus_and(tmp, mask, cpu_online_map);
 	if (cpus_empty(tmp))
@@ -2521,8 +2545,7 @@ static void set_ht_irq_affinity(unsigned int irq, cpumask_t mask)
 	dest = cpu_mask_to_apicid(mask);
 
 	target_ht_irq(irq, dest);
-	desc = irq_to_desc(irq);
-	desc->affinity = mask;
+	irq_to_desc(irq)->affinity = mask;
 }
 #endif
 
diff --git a/arch/x86/kernel/irq_32.c b/arch/x86/kernel/irq_32.c
index 576c5df6cad8..0a57e39159a8 100644
--- a/arch/x86/kernel/irq_32.c
+++ b/arch/x86/kernel/irq_32.c
@@ -224,9 +224,10 @@ unsigned int do_IRQ(struct pt_regs *regs)
 	struct pt_regs *old_regs;
 	/* high bit used in ret_from_ code */
 	int overflow, irq = ~regs->orig_ax;
-	struct irq_desc *desc = irq_to_desc(irq);
+	struct irq_desc *desc;
 
-	if (unlikely((unsigned)irq >= nr_irqs)) {
+	desc = irq_to_desc(irq);
+	if (unlikely(!desc)) {
 		printk(KERN_EMERG "%s: cannot handle IRQ %d\n",
 					__func__, irq);
 		BUG();
@@ -263,6 +264,24 @@ int show_interrupts(struct seq_file *p, void *v)
 	int i = *(loff_t *) v, j;
 	struct irqaction * action;
 	unsigned long flags;
+	unsigned int entries;
+	struct irq_desc *desc;
+	int tail = 0;
+
+#ifdef CONFIG_HAVE_SPARSE_IRQ
+	desc = (struct irq_desc *)v;
+	entries = -1U;
+	i = desc->irq;
+	if (!desc->next)
+		tail = 1;
+#else
+	entries = nr_irqs - 1;
+	i = *(loff_t *) v;
+	if (i == nr_irqs)
+		tail = 1;
+	else
+		desc = irq_to_desc(i);
+#endif
 
 	if (i == 0) {
 		seq_printf(p, "           ");
@@ -271,9 +290,8 @@ int show_interrupts(struct seq_file *p, void *v)
 		seq_putc(p, '\n');
 	}
 
-	if (i < nr_irqs) {
+	if (i <= entries) {
 		unsigned any_count = 0;
-		struct irq_desc *desc = irq_to_desc(i);
 
 		spin_lock_irqsave(&desc->lock, flags);
 #ifndef CONFIG_SMP
@@ -285,7 +303,7 @@ int show_interrupts(struct seq_file *p, void *v)
 		action = desc->action;
 		if (!action && !any_count)
 			goto skip;
-		seq_printf(p, "%3d: ",i);
+		seq_printf(p, "%#x: ",i);
 #ifndef CONFIG_SMP
 		seq_printf(p, "%10u ", kstat_irqs(i));
 #else
@@ -304,7 +322,9 @@ int show_interrupts(struct seq_file *p, void *v)
 		seq_putc(p, '\n');
 skip:
 		spin_unlock_irqrestore(&desc->lock, flags);
-	} else if (i == nr_irqs) {
+	}
+
+	if (tail) {
 		seq_printf(p, "NMI: ");
 		for_each_online_cpu(j)
 			seq_printf(p, "%10u ", nmi_count(j));
@@ -396,15 +416,14 @@ void fixup_irqs(cpumask_t map)
 {
 	unsigned int irq;
 	static int warned;
+	struct irq_desc *desc;
 
-	for (irq = 0; irq < nr_irqs; irq++) {
+	for_each_irq_desc(irq, desc) {
 		cpumask_t mask;
-		struct irq_desc *desc;
 
 		if (irq == 2)
 			continue;
 
-		desc = irq_to_desc(irq);
 		cpus_and(mask, desc->affinity, map);
 		if (any_online_cpu(mask) == NR_CPUS) {
 			printk("Breaking affinity for irq %i\n", irq);
diff --git a/arch/x86/kernel/irqinit_32.c b/arch/x86/kernel/irqinit_32.c
index 65c1c9507707..ded09ac2642e 100644
--- a/arch/x86/kernel/irqinit_32.c
+++ b/arch/x86/kernel/irqinit_32.c
@@ -69,6 +69,13 @@ void __init init_ISA_irqs (void)
 	 * 16 old-style INTA-cycle interrupts:
 	 */
 	for (i = 0; i < 16; i++) {
+		/* first time call this irq_desc */
+		struct irq_desc *desc = irq_to_desc_alloc(i);
+
+		desc->status = IRQ_DISABLED;
+		desc->action = NULL;
+		desc->depth = 1;
+
 		set_irq_chip_and_handler_name(i, &i8259A_chip,
 					      handle_level_irq, "XT");
 	}