[RFC PATCH v2 04/10] genirq: Add handle_strict_flow_irq() flow handler

[Valentin Schneider]

The GIC family of irqchips have been so far treated as "fasteoi"
chips. As handle_fasteoi_irq() states, this implies:

 *	Only a single callback will be issued to the chip: an ->eoi()
 *	call when the interrupt has been serviced.

However, the GICs have an operating mode (EOImode=1) which requires an
additional chip interaction during the IRQ handling. Said operating mode
already has some uses with virtualization, but could also be leveraged to
slightly optimize ONESHOT IRQs.

This extra interaction is currently hidden away in the drivers, but further
exploiting its effects (see IRQD_IRQ_FLOW_MASKED) requires lifting it from
the driver code into core code. It so happens that it fits the role of
->irq_ack(); unfortunately, the GICs require both callbacks to be strictly
paired with one another: for a given IRQ activation, there must be a single
->irq_ack() followed by a single ->irq_eoi(). No more, no less, and in that
order.

Introduce a new flow handler which guarantees said ack / eoi pairing. Note
that it is strikingly similar to handle_fasteoi_mask_irq() for now, but
will be further modified in later patches

Signed-off-by: Valentin Schneider <valentin.schneider@xxxxxxx>
---
 include/linux/irq.h |  1 +
 kernel/irq/chip.c   | 48 +++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 49 insertions(+)

diff --git a/include/linux/irq.h b/include/linux/irq.h
index ef179245a642..37075929e329 100644
--- a/include/linux/irq.h
+++ b/include/linux/irq.h
@@ -663,6 +663,7 @@ extern void handle_edge_irq(struct irq_desc *desc);
 extern void handle_edge_eoi_irq(struct irq_desc *desc);
 extern void handle_simple_irq(struct irq_desc *desc);
 extern void handle_untracked_irq(struct irq_desc *desc);
+extern void handle_strict_flow_irq(struct irq_desc *desc);
 extern void handle_percpu_irq(struct irq_desc *desc);
 extern void handle_percpu_devid_irq(struct irq_desc *desc);
 extern void handle_bad_irq(struct irq_desc *desc);
diff --git a/kernel/irq/chip.c b/kernel/irq/chip.c
index 4d3bde55c5d9..699e70b51aae 100644
--- a/kernel/irq/chip.c
+++ b/kernel/irq/chip.c
@@ -896,6 +896,54 @@ void handle_edge_eoi_irq(struct irq_desc *desc)
 }
 #endif
 
+/**
+ *	handle_strict_flow_irq - irq handler for strict controllers
+ *	@desc:	the interrupt description structure for this irq
+ *
+ *      Ensures strict pairing of ->ack() and ->eoi() for any IRQ passing
+ *      through here. The ->eoi() may be deferred to the tail of the IRQ thread
+ *      for ONESHOT IRQs.
+ */
+void handle_strict_flow_irq(struct irq_desc *desc)
+{
+	struct irq_chip *chip = desc->irq_data.chip;
+
+	raw_spin_lock(&desc->lock);
+	mask_ack_irq(desc);
+
+	if (!irq_may_run(desc))
+		goto out;
+
+	desc->istate &= ~(IRQS_REPLAY | IRQS_WAITING);
+
+	/*
+	 * If it's disabled or no action available then keep it masked and
+	 * get out of here:
+	 */
+	if (unlikely(!desc->action || irqd_irq_disabled(&desc->irq_data))) {
+		desc->istate |= IRQS_PENDING;
+		goto out;
+	}
+
+	kstat_incr_irqs_this_cpu(desc);
+
+	handle_irq_event(desc);
+
+	cond_unmask_eoi_irq(desc, chip);
+
+	raw_spin_unlock(&desc->lock);
+	return;
+out:
+	/*
+	 * XXX: this is where IRQCHIP_EOI_IF_HANDLED would be checked, but
+	 * it's conceptually incompatible with this handler (it breaks the
+	 * strict pairing)
+	 */
+	eoi_irq(desc);
+	raw_spin_unlock(&desc->lock);
+}
+EXPORT_SYMBOL_GPL(handle_strict_flow_irq);
+
 /**
  *	handle_percpu_irq - Per CPU local irq handler
  *	@desc:	the interrupt description structure for this irq
-- 
2.25.1