Thread (34 messages) 34 messages, 6 authors, 2019-08-22

Re: [PATCH 06/19] irqchip/mmp: add missing chained_irq_{enter,exit}()

From: Marc Zyngier <maz@kernel.org>
Date: 2019-08-09 10:56:26
Also in: linux-arm-kernel, linux-clk, lkml

On 09/08/2019 10:31, Lubomir Rintel wrote:
quoted hunk ↗ jump to hunk
The lack of chained_irq_exit() leaves the muxed interrupt masked on MMP3.
For reasons unknown this is not a problem on MMP2.

Signed-off-by: Lubomir Rintel <lkundrak@v3.sk>
---
 drivers/irqchip/irq-mmp.c | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)
diff --git a/drivers/irqchip/irq-mmp.c b/drivers/irqchip/irq-mmp.c
index af9cba4a51c2e..cd8d2253f56d1 100644
--- a/drivers/irqchip/irq-mmp.c
+++ b/drivers/irqchip/irq-mmp.c
@@ -13,6 +13,7 @@
 #include <linux/init.h>
 #include <linux/irq.h>
 #include <linux/irqchip.h>
+#include <linux/irqchip/chained_irq.h>
 #include <linux/irqdomain.h>
 #include <linux/io.h>
 #include <linux/ioport.h>
@@ -132,11 +133,14 @@ struct irq_chip icu_irq_chip = {
 static void icu_mux_irq_demux(struct irq_desc *desc)
 {
 	unsigned int irq = irq_desc_get_irq(desc);
+	struct irq_chip *chip = irq_get_chip(irq);
Consider using irq_desc_get_chip() instead, which avoids going through
the irq->desc again.
quoted hunk ↗ jump to hunk
 	struct irq_domain *domain;
 	struct icu_chip_data *data;
 	int i;
 	unsigned long mask, status, n;
 
+	chained_irq_enter(chip, desc);
+
 	for (i = 1; i < max_icu_nr; i++) {
 		if (irq == icu_data[i].cascade_irq) {
 			domain = icu_data[i].domain;
@@ -146,7 +150,7 @@ static void icu_mux_irq_demux(struct irq_desc *desc)
 	}
 	if (i >= max_icu_nr) {
 		pr_err("Spurious irq %d in MMP INTC\n", irq);
-		return;
+		goto out;
 	}
 
 	mask = readl_relaxed(data->reg_mask);
@@ -158,6 +162,9 @@ static void icu_mux_irq_demux(struct irq_desc *desc)
 			generic_handle_irq(icu_data[i].virq_base + n);
 		}
 	}
+
+out:
+	chained_irq_exit(chip, desc);
 }
 
 static int mmp_irq_domain_map(struct irq_domain *d, unsigned int irq,
Otherwise looks OK.

	M.
-- 
Jazz is not dead, it just smells funny...
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help