x86/apic: Convert to IRQCHIP_MOVE_DEFERRED

Instead of marking individual interrupts as safe to be migrated in
arbitrary contexts, mark the interrupt chips, which require the interrupt
to be moved in actual interrupt context, with the new IRQCHIP_MOVE_DEFERRED
flag. This makes more sense because this is a per interrupt chip property
and not restricted to individual interrupts.

That flips the logic from the historical opt-out to a opt-in model. This is
simpler to handle for other architectures, which default to unrestricted
affinity setting. It also allows to cleanup the redundant core logic
significantly.

All interrupt chips, which belong to a top-level domain sitting directly on
top of the x86 vector domain are marked accordingly, unless the related
setup code marks the interrupts with IRQ_MOVE_PCNTXT, i.e. XEN.

No functional change intended.

Signed-off-by: Thomas Gleixner <[email protected]>
Reviewed-by: Steve Wahl <[email protected]>
Acked-by: Wei Liu <[email protected]>
Link: https://lore.kernel.org/all/[email protected]

Author: KAGA-KOKO

arch/x86/Kconfig

@@ -173,6 +173,7 @@ config X86
 	select GENERIC_IRQ_RESERVATION_MODE
 	select GENERIC_IRQ_SHOW
 	select GENERIC_PENDING_IRQ		if SMP
+	select GENERIC_PENDING_IRQ_CHIPFLAGS	if SMP
 	select GENERIC_PTDUMP
 	select GENERIC_SMP_IDLE_THREAD
 	select GENERIC_TIME_VSYSCALL

arch/x86/hyperv/irqdomain.c

@@ -304,7 +304,7 @@ static struct irq_chip hv_pci_msi_controller = {
 	.irq_retrigger		= irq_chip_retrigger_hierarchy,
 	.irq_compose_msi_msg	= hv_irq_compose_msi_msg,
 	.irq_set_affinity	= msi_domain_set_affinity,
-	.flags			= IRQCHIP_SKIP_SET_WAKE,
+	.flags			= IRQCHIP_SKIP_SET_WAKE | IRQCHIP_MOVE_DEFERRED,
 };
 
 static struct msi_domain_ops pci_msi_domain_ops = {

arch/x86/kernel/apic/io_apic.c

@@ -1861,7 +1861,7 @@ static struct irq_chip ioapic_chip __read_mostly = {
 	.irq_set_affinity	= ioapic_set_affinity,
 	.irq_retrigger		= irq_chip_retrigger_hierarchy,
 	.irq_get_irqchip_state	= ioapic_irq_get_chip_state,
-	.flags			= IRQCHIP_SKIP_SET_WAKE |
+	.flags			= IRQCHIP_SKIP_SET_WAKE | IRQCHIP_MOVE_DEFERRED |
 				  IRQCHIP_AFFINITY_PRE_STARTUP,
 };
 

arch/x86/kernel/apic/msi.c

@@ -214,6 +214,7 @@ static bool x86_init_dev_msi_info(struct device *dev, struct irq_domain *domain,
 		if (WARN_ON_ONCE(domain != real_parent))
 			return false;
 		info->chip->irq_set_affinity = msi_set_affinity;
+		info->chip->flags |= IRQCHIP_MOVE_DEFERRED;
 		break;
 	case DOMAIN_BUS_DMAR:
 	case DOMAIN_BUS_AMDVI:
@@ -315,7 +316,7 @@ static struct irq_chip dmar_msi_controller = {
 	.irq_retrigger		= irq_chip_retrigger_hierarchy,
 	.irq_compose_msi_msg	= dmar_msi_compose_msg,
 	.irq_write_msi_msg	= dmar_msi_write_msg,
-	.flags			= IRQCHIP_SKIP_SET_WAKE |
+	.flags			= IRQCHIP_SKIP_SET_WAKE | IRQCHIP_MOVE_DEFERRED |
 				  IRQCHIP_AFFINITY_PRE_STARTUP,
 };
 

arch/x86/kernel/hpet.c

@@ -516,22 +516,14 @@ static int hpet_msi_init(struct irq_domain *domain,
 			 struct msi_domain_info *info, unsigned int virq,
 			 irq_hw_number_t hwirq, msi_alloc_info_t *arg)
 {
-	irq_set_status_flags(virq, IRQ_MOVE_PCNTXT);
 	irq_domain_set_info(domain, virq, arg->hwirq, info->chip, NULL,
 			    handle_edge_irq, arg->data, "edge");
 
 	return 0;
 }
 
-static void hpet_msi_free(struct irq_domain *domain,
-			  struct msi_domain_info *info, unsigned int virq)
-{
-	irq_clear_status_flags(virq, IRQ_MOVE_PCNTXT);
-}
-
 static struct msi_domain_ops hpet_msi_domain_ops = {
 	.msi_init	= hpet_msi_init,
-	.msi_free	= hpet_msi_free,
 };
 
 static struct msi_domain_info hpet_msi_domain_info = {

arch/x86/platform/uv/uv_irq.c

@@ -92,8 +92,6 @@ static int uv_domain_alloc(struct irq_domain *domain, unsigned int virq,
 	if (ret >= 0) {
 		if (info->uv.limit == UV_AFFINITY_CPU)
 			irq_set_status_flags(virq, IRQ_NO_BALANCING);
-		else
-			irq_set_status_flags(virq, IRQ_MOVE_PCNTXT);
 
 		chip_data->pnode = uv_blade_to_pnode(info->uv.blade);
 		chip_data->offset = info->uv.offset;
@@ -113,7 +111,6 @@ static void uv_domain_free(struct irq_domain *domain, unsigned int virq,
 
 	BUG_ON(nr_irqs != 1);
 	kfree(irq_data->chip_data);
-	irq_clear_status_flags(virq, IRQ_MOVE_PCNTXT);
 	irq_clear_status_flags(virq, IRQ_NO_BALANCING);
 	irq_domain_free_irqs_top(domain, virq, nr_irqs);
 }

drivers/iommu/amd/init.c

@@ -2332,7 +2332,7 @@ static struct irq_chip intcapxt_controller = {
 	.irq_retrigger		= irq_chip_retrigger_hierarchy,
 	.irq_set_affinity       = intcapxt_set_affinity,
 	.irq_set_wake		= intcapxt_set_wake,
-	.flags			= IRQCHIP_MASK_ON_SUSPEND,
+	.flags			= IRQCHIP_MASK_ON_SUSPEND | IRQCHIP_MOVE_DEFERRED,
 };
 
 static const struct irq_domain_ops intcapxt_domain_ops = {

drivers/iommu/amd/iommu.c

@@ -3532,7 +3532,6 @@ static int irq_remapping_alloc(struct irq_domain *domain, unsigned int virq,
 		irq_data->chip_data = data;
 		irq_data->chip = &amd_ir_chip;
 		irq_remapping_prepare_irte(data, cfg, info, devid, index, i);
-		irq_set_status_flags(virq + i, IRQ_MOVE_PCNTXT);
 	}
 
 	return 0;

drivers/iommu/intel/irq_remapping.c

@@ -1463,7 +1463,6 @@ static int intel_irq_remapping_alloc(struct irq_domain *domain,
 		else
 			irq_data->chip = &intel_ir_chip;
 		intel_irq_remapping_prepare_irte(ird, irq_cfg, info, index, i);
-		irq_set_status_flags(virq + i, IRQ_MOVE_PCNTXT);
 	}
 	return 0;
 

drivers/pci/controller/pci-hyperv.c

@@ -2053,6 +2053,7 @@ static struct irq_chip hv_msi_irq_chip = {
 	.irq_set_affinity	= irq_chip_set_affinity_parent,
 #ifdef CONFIG_X86
 	.irq_ack		= irq_chip_ack_parent,
+	.flags			= IRQCHIP_MOVE_DEFERRED,
 #elif defined(CONFIG_ARM64)
 	.irq_eoi		= irq_chip_eoi_parent,
 #endif

drivers/xen/events/events_base.c

@@ -722,12 +722,6 @@ static struct irq_info *xen_irq_init(unsigned int irq)
 		INIT_RCU_WORK(&info->rwork, delayed_free_irq);
 
 		set_info_for_irq(irq, info);
-		/*
-		 * Interrupt affinity setting can be immediate. No point
-		 * in delaying it until an interrupt is handled.
-		 */
-		irq_set_status_flags(irq, IRQ_MOVE_PCNTXT);
-
 		INIT_LIST_HEAD(&info->eoi_list);
 		list_add_tail(&info->list, &xen_irq_list_head);
 	}