Merge branch 'irq-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull irq updates from Thomas Gleixner:
 "The interrupt departement delivers this time:

   - New infrastructure to manage NMIs on platforms which have a sane
     NMI delivery, i.e. identifiable NMI vectors instead of a single
     lump.

   - Simplification of the interrupt affinity management so drivers
     don't have to implement ugly loops around the PCI/MSI enablement.

   - Speedup for interrupt statistics in /proc/stat

   - Provide a function to retrieve the default irq domain

   - A new interrupt controller for the Loongson LS1X platform

   - Affinity support for the SiFive PLIC

   - Better support for the iMX irqsteer driver

   - NUMA aware memory allocations for GICv3

   - The usual small fixes, improvements and cleanups all over the
     place"

* 'irq-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (36 commits)
  irqchip/imx-irqsteer: Add multi output interrupts support
  irqchip/imx-irqsteer: Change to use reg_num instead of irq_group
  dt-bindings: irq: imx-irqsteer: Add multi output interrupts support
  dt-binding: irq: imx-irqsteer: Use irq number instead of group number
  irqchip/brcmstb-l2: Use _irqsave locking variants in non-interrupt code
  irqchip/gicv3-its: Use NUMA aware memory allocation for ITS tables
  irqdomain: Allow the default irq domain to be retrieved
  irqchip/sifive-plic: Implement irq_set_affinity() for SMP host
  irqchip/sifive-plic: Differentiate between PLIC handler and context
  irqchip/sifive-plic: Add warning in plic_init() if handler already present
  irqchip/sifive-plic: Pre-compute context hart base and enable base
  PCI/MSI: Remove obsolete sanity checks for multiple interrupt sets
  genirq/affinity: Remove the leftovers of the original set support
  nvme-pci: Simplify interrupt allocation
  genirq/affinity: Add new callback for (re)calculating interrupt sets
  genirq/affinity: Store interrupt sets size in struct irq_affinity
  genirq/affinity: Code consolidation
  irqchip/irq-sifive-plic: Check and continue in case of an invalid cpuid.
  irqchip/i8259: Fix shutdown order by moving syscore_ops registration
  dt-bindings: interrupt-controller: loongson ls1x intc
  ...

kernel/irq/affinity.c

@@ -9,7 +9,7 @@
 #include <linux/cpu.h>
 
 static void irq_spread_init_one(struct cpumask *irqmsk, struct cpumask *nmsk,
-				int cpus_per_vec)
+				unsigned int cpus_per_vec)
 {
 	const struct cpumask *siblmsk;
 	int cpu, sibl;
@@ -95,15 +95,17 @@ static int get_nodes_in_cpumask(cpumask_var_t *node_to_cpumask,
 }
 
 static int __irq_build_affinity_masks(const struct irq_affinity *affd,
-				      int startvec, int numvecs, int firstvec,
+				      unsigned int startvec,
+				      unsigned int numvecs,
+				      unsigned int firstvec,
 				      cpumask_var_t *node_to_cpumask,
 				      const struct cpumask *cpu_mask,
 				      struct cpumask *nmsk,
 				      struct irq_affinity_desc *masks)
 {
-	int n, nodes, cpus_per_vec, extra_vecs, done = 0;
-	int last_affv = firstvec + numvecs;
-	int curvec = startvec;
+	unsigned int n, nodes, cpus_per_vec, extra_vecs, done = 0;
+	unsigned int last_affv = firstvec + numvecs;
+	unsigned int curvec = startvec;
 	nodemask_t nodemsk = NODE_MASK_NONE;
 
 	if (!cpumask_weight(cpu_mask))
@@ -117,18 +119,16 @@ static int __irq_build_affinity_masks(const struct irq_affinity *affd,
 	 */
 	if (numvecs <= nodes) {
 		for_each_node_mask(n, nodemsk) {
-			cpumask_or(&masks[curvec].mask,
-					&masks[curvec].mask,
-					node_to_cpumask[n]);
+			cpumask_or(&masks[curvec].mask, &masks[curvec].mask,
+				   node_to_cpumask[n]);
 			if (++curvec == last_affv)
 				curvec = firstvec;
 		}
-		done = numvecs;
-		goto out;
+		return numvecs;
 	}
 
 	for_each_node_mask(n, nodemsk) {
-		int ncpus, v, vecs_to_assign, vecs_per_node;
+		unsigned int ncpus, v, vecs_to_assign, vecs_per_node;
 
 		/* Spread the vectors per node */
 		vecs_per_node = (numvecs - (curvec - firstvec)) / nodes;
@@ -163,8 +163,6 @@ static int __irq_build_affinity_masks(const struct irq_affinity *affd,
 			curvec = firstvec;
 		--nodes;
 	}
-
-out:
 	return done;
 }
 
@@ -174,19 +172,24 @@ out:
  *	2) spread other possible CPUs on these vectors
  */
 static int irq_build_affinity_masks(const struct irq_affinity *affd,
-				    int startvec, int numvecs, int firstvec,
-				    cpumask_var_t *node_to_cpumask,
+				    unsigned int startvec, unsigned int numvecs,
+				    unsigned int firstvec,
 				    struct irq_affinity_desc *masks)
 {
-	int curvec = startvec, nr_present, nr_others;
-	int ret = -ENOMEM;
+	unsigned int curvec = startvec, nr_present, nr_others;
+	cpumask_var_t *node_to_cpumask;
 	cpumask_var_t nmsk, npresmsk;
+	int ret = -ENOMEM;
 
 	if (!zalloc_cpumask_var(&nmsk, GFP_KERNEL))
 		return ret;
 
 	if (!zalloc_cpumask_var(&npresmsk, GFP_KERNEL))
-		goto fail;
+		goto fail_nmsk;
+
+	node_to_cpumask = alloc_node_to_cpumask();
+	if (!node_to_cpumask)
+		goto fail_npresmsk;
 
 	ret = 0;
 	/* Stabilize the cpumasks */
@@ -217,13 +220,22 @@ static int irq_build_affinity_masks(const struct irq_affinity *affd,
 	if (nr_present < numvecs)
 		WARN_ON(nr_present + nr_others < numvecs);
 
+	free_node_to_cpumask(node_to_cpumask);
+
+ fail_npresmsk:
 	free_cpumask_var(npresmsk);
 
- fail:
+ fail_nmsk:
 	free_cpumask_var(nmsk);
 	return ret;
 }
 
+static void default_calc_sets(struct irq_affinity *affd, unsigned int affvecs)
+{
+	affd->nr_sets = 1;
+	affd->set_size[0] = affvecs;
+}
+
 /**
  * irq_create_affinity_masks - Create affinity masks for multiqueue spreading
  * @nvecs:	The total number of vectors
@@ -232,50 +244,62 @@ static int irq_build_affinity_masks(const struct irq_affinity *affd,
  * Returns the irq_affinity_desc pointer or NULL if allocation failed.
  */
 struct irq_affinity_desc *
-irq_create_affinity_masks(int nvecs, const struct irq_affinity *affd)
+irq_create_affinity_masks(unsigned int nvecs, struct irq_affinity *affd)
 {
-	int affvecs = nvecs - affd->pre_vectors - affd->post_vectors;
-	int curvec, usedvecs;
-	cpumask_var_t *node_to_cpumask;
+	unsigned int affvecs, curvec, usedvecs, i;
 	struct irq_affinity_desc *masks = NULL;
-	int i, nr_sets;
 
 	/*
-	 * If there aren't any vectors left after applying the pre/post
-	 * vectors don't bother with assigning affinity.
+	 * Determine the number of vectors which need interrupt affinities
+	 * assigned. If the pre/post request exhausts the available vectors
+	 * then nothing to do here except for invoking the calc_sets()
+	 * callback so the device driver can adjust to the situation. If there
+	 * is only a single vector, then managing the queue is pointless as
+	 * well.
 	 */
-	if (nvecs == affd->pre_vectors + affd->post_vectors)
+	if (nvecs > 1 && nvecs > affd->pre_vectors + affd->post_vectors)
+		affvecs = nvecs - affd->pre_vectors - affd->post_vectors;
+	else
+		affvecs = 0;
+
+	/*
+	 * Simple invocations do not provide a calc_sets() callback. Install
+	 * the generic one.
+	 */
+	if (!affd->calc_sets)
+		affd->calc_sets = default_calc_sets;
+
+	/* Recalculate the sets */
+	affd->calc_sets(affd, affvecs);
+
+	if (WARN_ON_ONCE(affd->nr_sets > IRQ_AFFINITY_MAX_SETS))
 		return NULL;
 
-	node_to_cpumask = alloc_node_to_cpumask();
-	if (!node_to_cpumask)
+	/* Nothing to assign? */
+	if (!affvecs)
 		return NULL;
 
 	masks = kcalloc(nvecs, sizeof(*masks), GFP_KERNEL);
 	if (!masks)
-		goto outnodemsk;
+		return NULL;
 
 	/* Fill out vectors at the beginning that don't need affinity */
 	for (curvec = 0; curvec < affd->pre_vectors; curvec++)
 		cpumask_copy(&masks[curvec].mask, irq_default_affinity);
+
 	/*
 	 * Spread on present CPUs starting from affd->pre_vectors. If we
 	 * have multiple sets, build each sets affinity mask separately.
 	 */
-	nr_sets = affd->nr_sets;
-	if (!nr_sets)
-		nr_sets = 1;
-
-	for (i = 0, usedvecs = 0; i < nr_sets; i++) {
-		int this_vecs = affd->sets ? affd->sets[i] : affvecs;
+	for (i = 0, usedvecs = 0; i < affd->nr_sets; i++) {
+		unsigned int this_vecs = affd->set_size[i];
 		int ret;
 
 		ret = irq_build_affinity_masks(affd, curvec, this_vecs,
-						curvec, node_to_cpumask, masks);
+					       curvec, masks);
 		if (ret) {
 			kfree(masks);
-			masks = NULL;
-			goto outnodemsk;
+			return NULL;
 		}
 		curvec += this_vecs;
 		usedvecs += this_vecs;
@@ -293,8 +317,6 @@ irq_create_affinity_masks(int nvecs, const struct irq_affinity *affd)
 	for (i = affd->pre_vectors; i < nvecs - affd->post_vectors; i++)
 		masks[i].is_managed = 1;
 
-outnodemsk:
-	free_node_to_cpumask(node_to_cpumask);
 	return masks;
 }
 
@@ -304,25 +326,22 @@ outnodemsk:
  * @maxvec:	The maximum number of vectors available
  * @affd:	Description of the affinity requirements
  */
-int irq_calc_affinity_vectors(int minvec, int maxvec, const struct irq_affinity *affd)
+unsigned int irq_calc_affinity_vectors(unsigned int minvec, unsigned int maxvec,
+				       const struct irq_affinity *affd)
 {
-	int resv = affd->pre_vectors + affd->post_vectors;
-	int vecs = maxvec - resv;
-	int set_vecs;
+	unsigned int resv = affd->pre_vectors + affd->post_vectors;
+	unsigned int set_vecs;
 
 	if (resv > minvec)
 		return 0;
 
-	if (affd->nr_sets) {
-		int i;
-
-		for (i = 0, set_vecs = 0;  i < affd->nr_sets; i++)
-			set_vecs += affd->sets[i];
+	if (affd->calc_sets) {
+		set_vecs = maxvec - resv;
 	} else {
 		get_online_cpus();
 		set_vecs = cpumask_weight(cpu_possible_mask);
 		put_online_cpus();
 	}
 
-	return resv + min(set_vecs, vecs);
+	return resv + min(set_vecs, maxvec - resv);
 }

kernel/irq/chip.c

@@ -729,6 +729,37 @@ out:
 }
 EXPORT_SYMBOL_GPL(handle_fasteoi_irq);
 
+/**
+ *	handle_fasteoi_nmi - irq handler for NMI interrupt lines
+ *	@desc:	the interrupt description structure for this irq
+ *
+ *	A simple NMI-safe handler, considering the restrictions
+ *	from request_nmi.
+ *
+ *	Only a single callback will be issued to the chip: an ->eoi()
+ *	call when the interrupt has been serviced. This enables support
+ *	for modern forms of interrupt handlers, which handle the flow
+ *	details in hardware, transparently.
+ */
+void handle_fasteoi_nmi(struct irq_desc *desc)
+{
+	struct irq_chip *chip = irq_desc_get_chip(desc);
+	struct irqaction *action = desc->action;
+	unsigned int irq = irq_desc_get_irq(desc);
+	irqreturn_t res;
+
+	trace_irq_handler_entry(irq, action);
+	/*
+	 * NMIs cannot be shared, there is only one action.
+	 */
+	res = action->handler(irq, action->dev_id);
+	trace_irq_handler_exit(irq, action, res);
+
+	if (chip->irq_eoi)
+		chip->irq_eoi(&desc->irq_data);
+}
+EXPORT_SYMBOL_GPL(handle_fasteoi_nmi);
+
 /**
  *	handle_edge_irq - edge type IRQ handler
  *	@desc:	the interrupt description structure for this irq
@@ -855,7 +886,11 @@ void handle_percpu_irq(struct irq_desc *desc)
 {
 	struct irq_chip *chip = irq_desc_get_chip(desc);
 
-	kstat_incr_irqs_this_cpu(desc);
+	/*
+	 * PER CPU interrupts are not serialized. Do not touch
+	 * desc->tot_count.
+	 */
+	__kstat_incr_irqs_this_cpu(desc);
 
 	if (chip->irq_ack)
 		chip->irq_ack(&desc->irq_data);
@@ -884,7 +919,11 @@ void handle_percpu_devid_irq(struct irq_desc *desc)
 	unsigned int irq = irq_desc_get_irq(desc);
 	irqreturn_t res;
 
-	kstat_incr_irqs_this_cpu(desc);
+	/*
+	 * PER CPU interrupts are not serialized. Do not touch
+	 * desc->tot_count.
+	 */
+	__kstat_incr_irqs_this_cpu(desc);
 
 	if (chip->irq_ack)
 		chip->irq_ack(&desc->irq_data);
@@ -908,6 +947,29 @@ void handle_percpu_devid_irq(struct irq_desc *desc)
 		chip->irq_eoi(&desc->irq_data);
 }
 
+/**
+ * handle_percpu_devid_fasteoi_nmi - Per CPU local NMI handler with per cpu
+ *				     dev ids
+ * @desc:	the interrupt description structure for this irq
+ *
+ * Similar to handle_fasteoi_nmi, but handling the dev_id cookie
+ * as a percpu pointer.
+ */
+void handle_percpu_devid_fasteoi_nmi(struct irq_desc *desc)
+{
+	struct irq_chip *chip = irq_desc_get_chip(desc);
+	struct irqaction *action = desc->action;
+	unsigned int irq = irq_desc_get_irq(desc);
+	irqreturn_t res;
+
+	trace_irq_handler_entry(irq, action);
+	res = action->handler(irq, raw_cpu_ptr(action->percpu_dev_id));
+	trace_irq_handler_exit(irq, action, res);
+
+	if (chip->irq_eoi)
+		chip->irq_eoi(&desc->irq_data);
+}
+
 static void
 __irq_do_set_handler(struct irq_desc *desc, irq_flow_handler_t handle,
 		     int is_chained, const char *name)

kernel/irq/debugfs.c

@@ -56,6 +56,7 @@ static const struct irq_bit_descr irqchip_flags[] = {
 	BIT_MASK_DESCR(IRQCHIP_ONESHOT_SAFE),
 	BIT_MASK_DESCR(IRQCHIP_EOI_THREADED),
 	BIT_MASK_DESCR(IRQCHIP_SUPPORTS_LEVEL_MSI),
+	BIT_MASK_DESCR(IRQCHIP_SUPPORTS_NMI),
 };
 
 static void
@@ -140,6 +141,7 @@ static const struct irq_bit_descr irqdesc_istates[] = {
 	BIT_MASK_DESCR(IRQS_WAITING),
 	BIT_MASK_DESCR(IRQS_PENDING),
 	BIT_MASK_DESCR(IRQS_SUSPENDED),
+	BIT_MASK_DESCR(IRQS_NMI),
 };
 
 
@@ -203,8 +205,8 @@ static ssize_t irq_debug_write(struct file *file, const char __user *user_buf,
 		chip_bus_lock(desc);
 		raw_spin_lock_irqsave(&desc->lock, flags);
 
-		if (irq_settings_is_level(desc)) {
-			/* Can't do level, sorry */
+		if (irq_settings_is_level(desc) || desc->istate & IRQS_NMI) {
+			/* Can't do level nor NMIs, sorry */
 			err = -EINVAL;
 		} else {
 			desc->istate |= IRQS_PENDING;
@@ -256,8 +258,6 @@ static int __init irq_debugfs_init(void)
 	int irq;
 
 	root_dir = debugfs_create_dir("irq", NULL);
-	if (!root_dir)
-		return -ENOMEM;
 
 	irq_domain_debugfs_init(root_dir);
 

kernel/irq/handle.c

@@ -166,7 +166,7 @@ irqreturn_t __handle_irq_event_percpu(struct irq_desc *desc, unsigned int *flags
 
 			__irq_wake_thread(desc, action);
 
-			/* Fall through to add to randomness */
+			/* Fall through - to add to randomness */
 		case IRQ_HANDLED:
 			*flags |= action->flags;
 			break;

kernel/irq/internals.h

@@ -49,6 +49,7 @@ enum {
  * IRQS_WAITING			- irq is waiting
  * IRQS_PENDING			- irq is pending and replayed later
  * IRQS_SUSPENDED		- irq is suspended
+ * IRQS_NMI			- irq line is used to deliver NMIs
  */
 enum {
 	IRQS_AUTODETECT		= 0x00000001,
@@ -60,6 +61,7 @@ enum {
 	IRQS_PENDING		= 0x00000200,
 	IRQS_SUSPENDED		= 0x00000800,
 	IRQS_TIMINGS		= 0x00001000,
+	IRQS_NMI		= 0x00002000,
 };
 
 #include "debug.h"
@@ -242,12 +244,18 @@ static inline void irq_state_set_masked(struct irq_desc *desc)
 
 #undef __irqd_to_state
 
-static inline void kstat_incr_irqs_this_cpu(struct irq_desc *desc)
+static inline void __kstat_incr_irqs_this_cpu(struct irq_desc *desc)
 {
 	__this_cpu_inc(*desc->kstat_irqs);
 	__this_cpu_inc(kstat.irqs_sum);
 }
 
+static inline void kstat_incr_irqs_this_cpu(struct irq_desc *desc)
+{
+	__kstat_incr_irqs_this_cpu(desc);
+	desc->tot_count++;
+}
+
 static inline int irq_desc_get_node(struct irq_desc *desc)
 {
 	return irq_common_data_get_node(&desc->irq_common_data);

kernel/irq/irqdesc.c

@@ -119,6 +119,7 @@ static void desc_set_defaults(unsigned int irq, struct irq_desc *desc, int node,
 	desc->depth = 1;
 	desc->irq_count = 0;
 	desc->irqs_unhandled = 0;
+	desc->tot_count = 0;
 	desc->name = NULL;
 	desc->owner = owner;
 	for_each_possible_cpu(cpu)
@@ -669,6 +670,41 @@ int __handle_domain_irq(struct irq_domain *domain, unsigned int hwirq,
 	set_irq_regs(old_regs);
 	return ret;
 }
+
+#ifdef CONFIG_IRQ_DOMAIN
+/**
+ * handle_domain_nmi - Invoke the handler for a HW irq belonging to a domain
+ * @domain:	The domain where to perform the lookup
+ * @hwirq:	The HW irq number to convert to a logical one
+ * @regs:	Register file coming from the low-level handling code
+ *
+ * Returns:	0 on success, or -EINVAL if conversion has failed
+ */
+int handle_domain_nmi(struct irq_domain *domain, unsigned int hwirq,
+		      struct pt_regs *regs)
+{
+	struct pt_regs *old_regs = set_irq_regs(regs);
+	unsigned int irq;
+	int ret = 0;
+
+	nmi_enter();
+
+	irq = irq_find_mapping(domain, hwirq);
+
+	/*
+	 * ack_bad_irq is not NMI-safe, just report
+	 * an invalid interrupt.
+	 */
+	if (likely(irq))
+		generic_handle_irq(irq);
+	else
+		ret = -EINVAL;
+
+	nmi_exit();
+	set_irq_regs(old_regs);
+	return ret;
+}
+#endif
 #endif
 
 /* Dynamic interrupt handling */
@@ -919,11 +955,15 @@ unsigned int kstat_irqs_cpu(unsigned int irq, int cpu)
 unsigned int kstat_irqs(unsigned int irq)
 {
 	struct irq_desc *desc = irq_to_desc(irq);
-	int cpu;
 	unsigned int sum = 0;
+	int cpu;
 
 	if (!desc || !desc->kstat_irqs)
 		return 0;
+	if (!irq_settings_is_per_cpu_devid(desc) &&
+	    !irq_settings_is_per_cpu(desc))
+	    return desc->tot_count;
+
 	for_each_possible_cpu(cpu)
 		sum += *per_cpu_ptr(desc->kstat_irqs, cpu);
 	return sum;

kernel/irq/irqdomain.c

@@ -458,6 +458,20 @@ void irq_set_default_host(struct irq_domain *domain)
 }
 EXPORT_SYMBOL_GPL(irq_set_default_host);
 
+/**
+ * irq_get_default_host() - Retrieve the "default" irq domain
+ *
+ * Returns: the default domain, if any.
+ *
+ * Modern code should never use this. This should only be used on
+ * systems that cannot implement a firmware->fwnode mapping (which
+ * both DT and ACPI provide).
+ */
+struct irq_domain *irq_get_default_host(void)
+{
+	return irq_default_domain;
+}
+
 static void irq_domain_clear_mapping(struct irq_domain *domain,
 				     irq_hw_number_t hwirq)
 {
@@ -1749,8 +1763,6 @@ void __init irq_domain_debugfs_init(struct dentry *root)
 	struct irq_domain *d;
 
 	domain_dir = debugfs_create_dir("domains", root);
-	if (!domain_dir)
-		return;
 
 	debugfs_create_file("default", 0444, domain_dir, NULL,
 			    &irq_domain_debug_fops);

kernel/irq/manage.c

@@ -341,7 +341,7 @@ irq_set_affinity_notifier(unsigned int irq, struct irq_affinity_notify *notify)
 	/* The release function is promised process context */
 	might_sleep();
 
-	if (!desc)
+	if (!desc || desc->istate & IRQS_NMI)
 		return -EINVAL;
 
 	/* Complete initialisation of *notify */
@@ -553,6 +553,21 @@ bool disable_hardirq(unsigned int irq)
 }
 EXPORT_SYMBOL_GPL(disable_hardirq);
 
+/**
+ *	disable_nmi_nosync - disable an nmi without waiting
+ *	@irq: Interrupt to disable
+ *
+ *	Disable the selected interrupt line. Disables and enables are
+ *	nested.
+ *	The interrupt to disable must have been requested through request_nmi.
+ *	Unlike disable_nmi(), this function does not ensure existing
+ *	instances of the IRQ handler have completed before returning.
+ */
+void disable_nmi_nosync(unsigned int irq)
+{
+	disable_irq_nosync(irq);
+}
+
 void __enable_irq(struct irq_desc *desc)
 {
 	switch (desc->depth) {
@@ -609,6 +624,20 @@ out:
 }
 EXPORT_SYMBOL(enable_irq);
 
+/**
+ *	enable_nmi - enable handling of an nmi
+ *	@irq: Interrupt to enable
+ *
+ *	The interrupt to enable must have been requested through request_nmi.
+ *	Undoes the effect of one call to disable_nmi(). If this
+ *	matches the last disable, processing of interrupts on this
+ *	IRQ line is re-enabled.
+ */
+void enable_nmi(unsigned int irq)
+{
+	enable_irq(irq);
+}
+
 static int set_irq_wake_real(unsigned int irq, unsigned int on)
 {
 	struct irq_desc *desc = irq_to_desc(irq);
@@ -644,6 +673,12 @@ int irq_set_irq_wake(unsigned int irq, unsigned int on)
 	if (!desc)
 		return -EINVAL;
 
+	/* Don't use NMIs as wake up interrupts please */
+	if (desc->istate & IRQS_NMI) {
+		ret = -EINVAL;
+		goto out_unlock;
+	}
+
 	/* wakeup-capable irqs can be shared between drivers that
 	 * don't need to have the same sleep mode behaviors.
 	 */
@@ -666,6 +701,8 @@ int irq_set_irq_wake(unsigned int irq, unsigned int on)
 				irqd_clear(&desc->irq_data, IRQD_WAKEUP_STATE);
 		}
 	}
+
+out_unlock:
 	irq_put_desc_busunlock(desc, flags);
 	return ret;
 }
@@ -726,6 +763,7 @@ int __irq_set_trigger(struct irq_desc *desc, unsigned long flags)
 	case IRQ_SET_MASK_OK_DONE:
 		irqd_clear(&desc->irq_data, IRQD_TRIGGER_MASK);
 		irqd_set(&desc->irq_data, flags);
+		/* fall through */
 
 	case IRQ_SET_MASK_OK_NOCOPY:
 		flags = irqd_get_trigger_type(&desc->irq_data);
@@ -1128,6 +1166,39 @@ static void irq_release_resources(struct irq_desc *desc)
 		c->irq_release_resources(d);
 }
 
+static bool irq_supports_nmi(struct irq_desc *desc)
+{
+	struct irq_data *d = irq_desc_get_irq_data(desc);
+
+#ifdef CONFIG_IRQ_DOMAIN_HIERARCHY
+	/* Only IRQs directly managed by the root irqchip can be set as NMI */
+	if (d->parent_data)
+		return false;
+#endif
+	/* Don't support NMIs for chips behind a slow bus */
+	if (d->chip->irq_bus_lock || d->chip->irq_bus_sync_unlock)
+		return false;
+
+	return d->chip->flags & IRQCHIP_SUPPORTS_NMI;
+}
+
+static int irq_nmi_setup(struct irq_desc *desc)
+{
+	struct irq_data *d = irq_desc_get_irq_data(desc);
+	struct irq_chip *c = d->chip;
+
+	return c->irq_nmi_setup ? c->irq_nmi_setup(d) : -EINVAL;
+}
+
+static void irq_nmi_teardown(struct irq_desc *desc)
+{
+	struct irq_data *d = irq_desc_get_irq_data(desc);
+	struct irq_chip *c = d->chip;
+
+	if (c->irq_nmi_teardown)
+		c->irq_nmi_teardown(d);
+}
+
 static int
 setup_irq_thread(struct irqaction *new, unsigned int irq, bool secondary)
 {
@@ -1302,9 +1373,17 @@ __setup_irq(unsigned int irq, struct irq_desc *desc, struct irqaction *new)
 		 * fields must have IRQF_SHARED set and the bits which
 		 * set the trigger type must match. Also all must
 		 * agree on ONESHOT.
+		 * Interrupt lines used for NMIs cannot be shared.
 		 */
 		unsigned int oldtype;
 
+		if (desc->istate & IRQS_NMI) {
+			pr_err("Invalid attempt to share NMI for %s (irq %d) on irqchip %s.\n",
+				new->name, irq, desc->irq_data.chip->name);
+			ret = -EINVAL;
+			goto out_unlock;
+		}
+
 		/*
 		 * If nobody did set the configuration before, inherit
 		 * the one provided by the requester.
@@ -1756,6 +1835,59 @@ const void *free_irq(unsigned int irq, void *dev_id)
 }
 EXPORT_SYMBOL(free_irq);
 
+/* This function must be called with desc->lock held */
+static const void *__cleanup_nmi(unsigned int irq, struct irq_desc *desc)
+{
+	const char *devname = NULL;
+
+	desc->istate &= ~IRQS_NMI;
+
+	if (!WARN_ON(desc->action == NULL)) {
+		irq_pm_remove_action(desc, desc->action);
+		devname = desc->action->name;
+		unregister_handler_proc(irq, desc->action);
+
+		kfree(desc->action);
+		desc->action = NULL;
+	}
+
+	irq_settings_clr_disable_unlazy(desc);
+	irq_shutdown(desc);
+
+	irq_release_resources(desc);
+
+	irq_chip_pm_put(&desc->irq_data);
+	module_put(desc->owner);
+
+	return devname;
+}
+
+const void *free_nmi(unsigned int irq, void *dev_id)
+{
+	struct irq_desc *desc = irq_to_desc(irq);
+	unsigned long flags;
+	const void *devname;
+
+	if (!desc || WARN_ON(!(desc->istate & IRQS_NMI)))
+		return NULL;
+
+	if (WARN_ON(irq_settings_is_per_cpu_devid(desc)))
+		return NULL;
+
+	/* NMI still enabled */
+	if (WARN_ON(desc->depth == 0))
+		disable_nmi_nosync(irq);
+
+	raw_spin_lock_irqsave(&desc->lock, flags);
+
+	irq_nmi_teardown(desc);
+	devname = __cleanup_nmi(irq, desc);
+
+	raw_spin_unlock_irqrestore(&desc->lock, flags);
+
+	return devname;
+}
+
 /**
  *	request_threaded_irq - allocate an interrupt line
  *	@irq: Interrupt line to allocate
@@ -1925,6 +2057,101 @@ int request_any_context_irq(unsigned int irq, irq_handler_t handler,
 }
 EXPORT_SYMBOL_GPL(request_any_context_irq);
 
+/**
+ *	request_nmi - allocate an interrupt line for NMI delivery
+ *	@irq: Interrupt line to allocate
+ *	@handler: Function to be called when the IRQ occurs.
+ *		  Threaded handler for threaded interrupts.
+ *	@irqflags: Interrupt type flags
+ *	@name: An ascii name for the claiming device
+ *	@dev_id: A cookie passed back to the handler function
+ *
+ *	This call allocates interrupt resources and enables the
+ *	interrupt line and IRQ handling. It sets up the IRQ line
+ *	to be handled as an NMI.
+ *
+ *	An interrupt line delivering NMIs cannot be shared and IRQ handling
+ *	cannot be threaded.
+ *
+ *	Interrupt lines requested for NMI delivering must produce per cpu
+ *	interrupts and have auto enabling setting disabled.
+ *
+ *	Dev_id must be globally unique. Normally the address of the
+ *	device data structure is used as the cookie. Since the handler
+ *	receives this value it makes sense to use it.
+ *
+ *	If the interrupt line cannot be used to deliver NMIs, function
+ *	will fail and return a negative value.
+ */
+int request_nmi(unsigned int irq, irq_handler_t handler,
+		unsigned long irqflags, const char *name, void *dev_id)
+{
+	struct irqaction *action;
+	struct irq_desc *desc;
+	unsigned long flags;
+	int retval;
+
+	if (irq == IRQ_NOTCONNECTED)
+		return -ENOTCONN;
+
+	/* NMI cannot be shared, used for Polling */
+	if (irqflags & (IRQF_SHARED | IRQF_COND_SUSPEND | IRQF_IRQPOLL))
+		return -EINVAL;
+
+	if (!(irqflags & IRQF_PERCPU))
+		return -EINVAL;
+
+	if (!handler)
+		return -EINVAL;
+
+	desc = irq_to_desc(irq);
+
+	if (!desc || irq_settings_can_autoenable(desc) ||
+	    !irq_settings_can_request(desc) ||
+	    WARN_ON(irq_settings_is_per_cpu_devid(desc)) ||
+	    !irq_supports_nmi(desc))
+		return -EINVAL;
+
+	action = kzalloc(sizeof(struct irqaction), GFP_KERNEL);
+	if (!action)
+		return -ENOMEM;
+
+	action->handler = handler;
+	action->flags = irqflags | IRQF_NO_THREAD | IRQF_NOBALANCING;
+	action->name = name;
+	action->dev_id = dev_id;
+
+	retval = irq_chip_pm_get(&desc->irq_data);
+	if (retval < 0)
+		goto err_out;
+
+	retval = __setup_irq(irq, desc, action);
+	if (retval)
+		goto err_irq_setup;
+
+	raw_spin_lock_irqsave(&desc->lock, flags);
+
+	/* Setup NMI state */
+	desc->istate |= IRQS_NMI;
+	retval = irq_nmi_setup(desc);
+	if (retval) {
+		__cleanup_nmi(irq, desc);
+		raw_spin_unlock_irqrestore(&desc->lock, flags);
+		return -EINVAL;
+	}
+
+	raw_spin_unlock_irqrestore(&desc->lock, flags);
+
+	return 0;
+
+err_irq_setup:
+	irq_chip_pm_put(&desc->irq_data);
+err_out:
+	kfree(action);
+
+	return retval;
+}
+
 void enable_percpu_irq(unsigned int irq, unsigned int type)
 {
 	unsigned int cpu = smp_processor_id();
@@ -1959,6 +2186,11 @@ out:
 }
 EXPORT_SYMBOL_GPL(enable_percpu_irq);
 
+void enable_percpu_nmi(unsigned int irq, unsigned int type)
+{
+	enable_percpu_irq(irq, type);
+}
+
 /**
  * irq_percpu_is_enabled - Check whether the per cpu irq is enabled
  * @irq:	Linux irq number to check for
@@ -1998,6 +2230,11 @@ void disable_percpu_irq(unsigned int irq)
 }
 EXPORT_SYMBOL_GPL(disable_percpu_irq);
 
+void disable_percpu_nmi(unsigned int irq)
+{
+	disable_percpu_irq(irq);
+}
+
 /*
  * Internal function to unregister a percpu irqaction.
  */
@@ -2029,6 +2266,8 @@ static struct irqaction *__free_percpu_irq(unsigned int irq, void __percpu *dev_
 	/* Found it - now remove it from the list of entries: */
 	desc->action = NULL;
 
+	desc->istate &= ~IRQS_NMI;
+
 	raw_spin_unlock_irqrestore(&desc->lock, flags);
 
 	unregister_handler_proc(irq, action);
@@ -2082,6 +2321,19 @@ void free_percpu_irq(unsigned int irq, void __percpu *dev_id)
 }
 EXPORT_SYMBOL_GPL(free_percpu_irq);
 
+void free_percpu_nmi(unsigned int irq, void __percpu *dev_id)
+{
+	struct irq_desc *desc = irq_to_desc(irq);
+
+	if (!desc || !irq_settings_is_per_cpu_devid(desc))
+		return;
+
+	if (WARN_ON(!(desc->istate & IRQS_NMI)))
+		return;
+
+	kfree(__free_percpu_irq(irq, dev_id));
+}
+
 /**
  *	setup_percpu_irq - setup a per-cpu interrupt
  *	@irq: Interrupt line to setup
@@ -2171,6 +2423,158 @@ int __request_percpu_irq(unsigned int irq, irq_handler_t handler,
 }
 EXPORT_SYMBOL_GPL(__request_percpu_irq);
 
+/**
+ *	request_percpu_nmi - allocate a percpu interrupt line for NMI delivery
+ *	@irq: Interrupt line to allocate
+ *	@handler: Function to be called when the IRQ occurs.
+ *	@name: An ascii name for the claiming device
+ *	@dev_id: A percpu cookie passed back to the handler function
+ *
+ *	This call allocates interrupt resources for a per CPU NMI. Per CPU NMIs
+ *	have to be setup on each CPU by calling prepare_percpu_nmi() before
+ *	being enabled on the same CPU by using enable_percpu_nmi().
+ *
+ *	Dev_id must be globally unique. It is a per-cpu variable, and
+ *	the handler gets called with the interrupted CPU's instance of
+ *	that variable.
+ *
+ *	Interrupt lines requested for NMI delivering should have auto enabling
+ *	setting disabled.
+ *
+ *	If the interrupt line cannot be used to deliver NMIs, function
+ *	will fail returning a negative value.
+ */
+int request_percpu_nmi(unsigned int irq, irq_handler_t handler,
+		       const char *name, void __percpu *dev_id)
+{
+	struct irqaction *action;
+	struct irq_desc *desc;
+	unsigned long flags;
+	int retval;
+
+	if (!handler)
+		return -EINVAL;
+
+	desc = irq_to_desc(irq);
+
+	if (!desc || !irq_settings_can_request(desc) ||
+	    !irq_settings_is_per_cpu_devid(desc) ||
+	    irq_settings_can_autoenable(desc) ||
+	    !irq_supports_nmi(desc))
+		return -EINVAL;
+
+	/* The line cannot already be NMI */
+	if (desc->istate & IRQS_NMI)
+		return -EINVAL;
+
+	action = kzalloc(sizeof(struct irqaction), GFP_KERNEL);
+	if (!action)
+		return -ENOMEM;
+
+	action->handler = handler;
+	action->flags = IRQF_PERCPU | IRQF_NO_SUSPEND | IRQF_NO_THREAD
+		| IRQF_NOBALANCING;
+	action->name = name;
+	action->percpu_dev_id = dev_id;
+
+	retval = irq_chip_pm_get(&desc->irq_data);
+	if (retval < 0)
+		goto err_out;
+
+	retval = __setup_irq(irq, desc, action);
+	if (retval)
+		goto err_irq_setup;
+
+	raw_spin_lock_irqsave(&desc->lock, flags);
+	desc->istate |= IRQS_NMI;
+	raw_spin_unlock_irqrestore(&desc->lock, flags);
+
+	return 0;
+
+err_irq_setup:
+	irq_chip_pm_put(&desc->irq_data);
+err_out:
+	kfree(action);
+
+	return retval;
+}
+
+/**
+ *	prepare_percpu_nmi - performs CPU local setup for NMI delivery
+ *	@irq: Interrupt line to prepare for NMI delivery
+ *
+ *	This call prepares an interrupt line to deliver NMI on the current CPU,
+ *	before that interrupt line gets enabled with enable_percpu_nmi().
+ *
+ *	As a CPU local operation, this should be called from non-preemptible
+ *	context.
+ *
+ *	If the interrupt line cannot be used to deliver NMIs, function
+ *	will fail returning a negative value.
+ */
+int prepare_percpu_nmi(unsigned int irq)
+{
+	unsigned long flags;
+	struct irq_desc *desc;
+	int ret = 0;
+
+	WARN_ON(preemptible());
+
+	desc = irq_get_desc_lock(irq, &flags,
+				 IRQ_GET_DESC_CHECK_PERCPU);
+	if (!desc)
+		return -EINVAL;
+
+	if (WARN(!(desc->istate & IRQS_NMI),
+		 KERN_ERR "prepare_percpu_nmi called for a non-NMI interrupt: irq %u\n",
+		 irq)) {
+		ret = -EINVAL;
+		goto out;
+	}
+
+	ret = irq_nmi_setup(desc);
+	if (ret) {
+		pr_err("Failed to setup NMI delivery: irq %u\n", irq);
+		goto out;
+	}
+
+out:
+	irq_put_desc_unlock(desc, flags);
+	return ret;
+}
+
+/**
+ *	teardown_percpu_nmi - undoes NMI setup of IRQ line
+ *	@irq: Interrupt line from which CPU local NMI configuration should be
+ *	      removed
+ *
+ *	This call undoes the setup done by prepare_percpu_nmi().
+ *
+ *	IRQ line should not be enabled for the current CPU.
+ *
+ *	As a CPU local operation, this should be called from non-preemptible
+ *	context.
+ */
+void teardown_percpu_nmi(unsigned int irq)
+{
+	unsigned long flags;
+	struct irq_desc *desc;
+
+	WARN_ON(preemptible());
+
+	desc = irq_get_desc_lock(irq, &flags,
+				 IRQ_GET_DESC_CHECK_PERCPU);
+	if (!desc)
+		return;
+
+	if (WARN_ON(!(desc->istate & IRQS_NMI)))
+		goto out;
+
+	irq_nmi_teardown(desc);
+out:
+	irq_put_desc_unlock(desc, flags);
+}
+
 /**
  *	irq_get_irqchip_state - returns the irqchip state of a interrupt.
  *	@irq: Interrupt line that is forwarded to a VM

kernel/kthread.c

@@ -101,6 +101,12 @@ bool kthread_should_stop(void)
 }
 EXPORT_SYMBOL(kthread_should_stop);
 
+bool __kthread_should_park(struct task_struct *k)
+{
+	return test_bit(KTHREAD_SHOULD_PARK, &to_kthread(k)->flags);
+}
+EXPORT_SYMBOL_GPL(__kthread_should_park);
+
 /**
  * kthread_should_park - should this kthread park now?
  *
@@ -114,7 +120,7 @@ EXPORT_SYMBOL(kthread_should_stop);
  */
 bool kthread_should_park(void)
 {
-	return test_bit(KTHREAD_SHOULD_PARK, &to_kthread(current)->flags);
+	return __kthread_should_park(current);
 }
 EXPORT_SYMBOL_GPL(kthread_should_park);
 

kernel/softirq.c

@@ -89,7 +89,8 @@ static bool ksoftirqd_running(unsigned long pending)
 
 	if (pending & SOFTIRQ_NOW_MASK)
 		return false;
-	return tsk && (tsk->state == TASK_RUNNING);
+	return tsk && (tsk->state == TASK_RUNNING) &&
+		!__kthread_should_park(tsk);
 }
 
 /*