1 files changed, 152 insertions, 29 deletions

diff --git a/kernel/irq_work.c b/kernel/irq_work.c
index eca83965b631..73f7e1fd4ab4 100644
--- a/kernel/irq_work.c
+++ b/kernel/irq_work.c
@@ -18,11 +18,38 @@
 #include <linux/cpu.h>
 #include <linux/notifier.h>
 #include <linux/smp.h>
+#include <linux/smpboot.h>
 #include <asm/processor.h>
+#include <linux/kasan.h>
 
+#include <trace/events/ipi.h>
 
 static DEFINE_PER_CPU(struct llist_head, raised_list);
 static DEFINE_PER_CPU(struct llist_head, lazy_list);
+static DEFINE_PER_CPU(struct task_struct *, irq_workd);
+
+static void wake_irq_workd(void)
+{
+	struct task_struct *tsk = __this_cpu_read(irq_workd);
+
+	if (!llist_empty(this_cpu_ptr(&lazy_list)) && tsk)
+		wake_up_process(tsk);
+}
+
+#ifdef CONFIG_SMP
+static void irq_work_wake(struct irq_work *entry)
+{
+	wake_irq_workd();
+}
+
+static DEFINE_PER_CPU(struct irq_work, irq_work_wakeup) =
+	IRQ_WORK_INIT_HARD(irq_work_wake);
+#endif
+
+static int irq_workd_should_run(unsigned int cpu)
+{
+	return !llist_empty(this_cpu_ptr(&lazy_list));
+}
 
 /*
  * Claim the entry so that no one else will poke at it.
@@ -31,10 +58,10 @@ static bool irq_work_claim(struct irq_work *work)
 {
 	int oflags;
 
-	oflags = atomic_fetch_or(IRQ_WORK_CLAIMED | CSD_TYPE_IRQ_WORK, &work->flags);
+	oflags = atomic_fetch_or(IRQ_WORK_CLAIMED | CSD_TYPE_IRQ_WORK, &work->node.a_flags);
 	/*
 	 * If the work is already pending, no need to raise the IPI.
-	 * The pairing atomic_fetch_andnot() in irq_work_run() makes sure
+	 * The pairing smp_mb() in irq_work_single() makes sure
 	 * everything we did before is visible.
 	 */
 	if (oflags & IRQ_WORK_PENDING)
@@ -49,18 +76,40 @@ void __weak arch_irq_work_raise(void)
 	 */
 }
 
+static __always_inline void irq_work_raise(struct irq_work *work)
+{
+	if (trace_ipi_send_cpu_enabled() && arch_irq_work_has_interrupt())
+		trace_ipi_send_cpu(smp_processor_id(), _RET_IP_, work->func);
+
+	arch_irq_work_raise();
+}
+
 /* Enqueue on current CPU, work must already be claimed and preempt disabled */
 static void __irq_work_queue_local(struct irq_work *work)
 {
+	struct llist_head *list;
+	bool rt_lazy_work = false;
+	bool lazy_work = false;
+	int work_flags;
+
+	work_flags = atomic_read(&work->node.a_flags);
+	if (work_flags & IRQ_WORK_LAZY)
+		lazy_work = true;
+	else if (IS_ENABLED(CONFIG_PREEMPT_RT) &&
+		 !(work_flags & IRQ_WORK_HARD_IRQ))
+		rt_lazy_work = true;
+
+	if (lazy_work || rt_lazy_work)
+		list = this_cpu_ptr(&lazy_list);
+	else
+		list = this_cpu_ptr(&raised_list);
+
+	if (!llist_add(&work->node.llist, list))
+		return;
+
 	/* If the work is "lazy", handle it from next tick if any */
-	if (atomic_read(&work->flags) & IRQ_WORK_LAZY) {
-		if (llist_add(&work->llnode, this_cpu_ptr(&lazy_list)) &&
-		    tick_nohz_tick_stopped())
-			arch_irq_work_raise();
-	} else {
-		if (llist_add(&work->llnode, this_cpu_ptr(&raised_list)))
-			arch_irq_work_raise();
-	}
+	if (!lazy_work || tick_nohz_tick_stopped())
+		irq_work_raise(work);
 }
 
 /* Enqueue the irq work @work on the current CPU */
@@ -98,21 +147,40 @@ bool irq_work_queue_on(struct irq_work *work, int cpu)
 	if (!irq_work_claim(work))
 		return false;
 
+	kasan_record_aux_stack(work);
+
 	preempt_disable();
 	if (cpu != smp_processor_id()) {
 		/* Arch remote IPI send/receive backend aren't NMI safe */
 		WARN_ON_ONCE(in_nmi());
-		__smp_call_single_queue(cpu, &work->llnode);
+
+		/*
+		 * On PREEMPT_RT the items which are not marked as
+		 * IRQ_WORK_HARD_IRQ are added to the lazy list and a HARD work
+		 * item is used on the remote CPU to wake the thread.
+		 */
+		if (IS_ENABLED(CONFIG_PREEMPT_RT) &&
+		    !(atomic_read(&work->node.a_flags) & IRQ_WORK_HARD_IRQ)) {
+
+			if (!llist_add(&work->node.llist, &per_cpu(lazy_list, cpu)))
+				goto out;
+
+			work = &per_cpu(irq_work_wakeup, cpu);
+			if (!irq_work_claim(work))
+				goto out;
+		}
+
+		__smp_call_single_queue(cpu, &work->node.llist);
 	} else {
 		__irq_work_queue_local(work);
 	}
+out:
 	preempt_enable();
 
 	return true;
 #endif /* CONFIG_SMP */
 }
 
-
 bool irq_work_needs_cpu(void)
 {
 	struct llist_head *raised, *lazy;
@@ -136,23 +204,32 @@ void irq_work_single(void *arg)
 	int flags;
 
 	/*
-	 * Clear the PENDING bit, after this point the @work
-	 * can be re-used.
-	 * Make it immediately visible so that other CPUs trying
-	 * to claim that work don't rely on us to handle their data
-	 * while we are in the middle of the func.
+	 * Clear the PENDING bit, after this point the @work can be re-used.
+	 * The PENDING bit acts as a lock, and we own it, so we can clear it
+	 * without atomic ops.
 	 */
-	flags = atomic_fetch_andnot(IRQ_WORK_PENDING, &work->flags);
+	flags = atomic_read(&work->node.a_flags);
+	flags &= ~IRQ_WORK_PENDING;
+	atomic_set(&work->node.a_flags, flags);
 
-	lockdep_irq_work_enter(work);
+	/*
+	 * See irq_work_claim().
+	 */
+	smp_mb();
+
+	lockdep_irq_work_enter(flags);
 	work->func(work);
-	lockdep_irq_work_exit(work);
+	lockdep_irq_work_exit(flags);
+
 	/*
-	 * Clear the BUSY bit and return to the free state if
-	 * no-one else claimed it meanwhile.
+	 * Clear the BUSY bit, if set, and return to the free state if no-one
+	 * else claimed it meanwhile.
 	 */
-	flags &= ~IRQ_WORK_PENDING;
-	(void)atomic_cmpxchg(&work->flags, flags, flags & ~IRQ_WORK_BUSY);
+	(void)atomic_cmpxchg(&work->node.a_flags, flags, flags & ~IRQ_WORK_BUSY);
+
+	if ((IS_ENABLED(CONFIG_PREEMPT_RT) && !irq_work_is_hard(work)) ||
+	    !arch_irq_work_has_interrupt())
+		rcuwait_wake_up(&work->irqwait);
 }
 
 static void irq_work_run_list(struct llist_head *list)
@@ -160,13 +237,18 @@ static void irq_work_run_list(struct llist_head *list)
 	struct irq_work *work, *tmp;
 	struct llist_node *llnode;
 
-	BUG_ON(!irqs_disabled());
+	/*
+	 * On PREEMPT_RT IRQ-work which is not marked as HARD will be processed
+	 * in a per-CPU thread in preemptible context. Only the items which are
+	 * marked as IRQ_WORK_HARD_IRQ will be processed in hardirq context.
+	 */
+	BUG_ON(!irqs_disabled() && !IS_ENABLED(CONFIG_PREEMPT_RT));
 
 	if (llist_empty(list))
 		return;
 
 	llnode = llist_del_all(list);
-	llist_for_each_entry_safe(work, tmp, llnode, llnode)
+	llist_for_each_entry_safe(work, tmp, llnode, node.llist)
 		irq_work_single(work);
 }
 
@@ -177,7 +259,10 @@ static void irq_work_run_list(struct llist_head *list)
 void irq_work_run(void)
 {
 	irq_work_run_list(this_cpu_ptr(&raised_list));
-	irq_work_run_list(this_cpu_ptr(&lazy_list));
+	if (!IS_ENABLED(CONFIG_PREEMPT_RT))
+		irq_work_run_list(this_cpu_ptr(&lazy_list));
+	else
+		wake_irq_workd();
 }
 EXPORT_SYMBOL_GPL(irq_work_run);
 
@@ -187,7 +272,11 @@ void irq_work_tick(void)
 
 	if (!llist_empty(raised) && !arch_irq_work_has_interrupt())
 		irq_work_run_list(raised);
-	irq_work_run_list(this_cpu_ptr(&lazy_list));
+
+	if (!IS_ENABLED(CONFIG_PREEMPT_RT))
+		irq_work_run_list(this_cpu_ptr(&lazy_list));
+	else
+		wake_irq_workd();
 }
 
 /*
@@ -197,8 +286,42 @@ void irq_work_tick(void)
 void irq_work_sync(struct irq_work *work)
 {
 	lockdep_assert_irqs_enabled();
+	might_sleep();
 
-	while (atomic_read(&work->flags) & IRQ_WORK_BUSY)
+	if ((IS_ENABLED(CONFIG_PREEMPT_RT) && !irq_work_is_hard(work)) ||
+	    !arch_irq_work_has_interrupt()) {
+		rcuwait_wait_event(&work->irqwait, !irq_work_is_busy(work),
+				   TASK_UNINTERRUPTIBLE);
+		return;
+	}
+
+	while (irq_work_is_busy(work))
 		cpu_relax();
 }
 EXPORT_SYMBOL_GPL(irq_work_sync);
+
+static void run_irq_workd(unsigned int cpu)
+{
+	irq_work_run_list(this_cpu_ptr(&lazy_list));
+}
+
+static void irq_workd_setup(unsigned int cpu)
+{
+	sched_set_fifo_low(current);
+}
+
+static struct smp_hotplug_thread irqwork_threads = {
+	.store                  = &irq_workd,
+	.setup			= irq_workd_setup,
+	.thread_should_run      = irq_workd_should_run,
+	.thread_fn              = run_irq_workd,
+	.thread_comm            = "irq_work/%u",
+};
+
+static __init int irq_work_init_threads(void)
+{
+	if (IS_ENABLED(CONFIG_PREEMPT_RT))
+		BUG_ON(smpboot_register_percpu_thread(&irqwork_threads));
+	return 0;
+}
+early_initcall(irq_work_init_threads);