Re: [PATCH v4] irqchip/irq-mst: Support polarity configuration
From: Marc Zyngier <maz@kernel.org>
Date: 2021-04-07 12:12:13
Also in:
linux-mediatek, lkml
Subsystem:
irqchip drivers, mstar interrupt controller driver, the rest · Maintainers:
Thomas Gleixner, Mark-PK Tsai, Daniel Palmer, Linus Torvalds
On Mon, 15 Mar 2021 13:18:48 +0000, Mark-PK Tsai [off-list ref] wrote:
quoted hunk ↗ jump to hunk
Support irq polarity configuration and save and restore the config when system suspend and resume. Signed-off-by: Mark-PK Tsai <mark-pk.tsai@mediatek.com> --- drivers/irqchip/irq-mst-intc.c | 94 ++++++++++++++++++++++++++++++++-- 1 file changed, 91 insertions(+), 3 deletions(-)diff --git a/drivers/irqchip/irq-mst-intc.c b/drivers/irqchip/irq-mst-intc.c index 143657b0cf28..a2ab3f837b96 100644 --- a/drivers/irqchip/irq-mst-intc.c +++ b/drivers/irqchip/irq-mst-intc.c@@ -13,15 +13,27 @@ #include <linux/of_irq.h> #include <linux/slab.h> #include <linux/spinlock.h> +#include <linux/syscore_ops.h> -#define INTC_MASK 0x0 -#define INTC_EOI 0x20 +#define MST_INTC_MAX_IRQS 64 + +#define INTC_MASK 0x0 +#define INTC_REV_POLARITY 0x10 +#define INTC_EOI 0x20 + +#ifdef CONFIG_PM_SLEEP +static LIST_HEAD(mst_intc_list); +#endif struct mst_intc_chip_data { raw_spinlock_t lock; unsigned int irq_start, nr_irqs; void __iomem *base; bool no_eoi; +#ifdef CONFIG_PM_SLEEP + struct list_head entry; + u16 saved_polarity_conf[DIV_ROUND_UP(MST_INTC_MAX_IRQS, 16)]; +#endif }; static void mst_set_irq(struct irq_data *d, u32 offset)@@ -78,6 +90,20 @@ static void mst_intc_eoi_irq(struct irq_data *d) irq_chip_eoi_parent(d); } +static int mst_irq_chip_set_type(struct irq_data *data, unsigned int type) +{ + if (type != IRQ_TYPE_EDGE_RISING && type != IRQ_TYPE_EDGE_FALLING && + type != IRQ_TYPE_LEVEL_HIGH && type != IRQ_TYPE_LEVEL_LOW) + return -EINVAL;
All of this amounts to checking for IRQ_TYPE_{NONE,EDGE_BOTH}. Maybe
that's a better option.
+ + if (type == IRQ_TYPE_LEVEL_LOW || type == IRQ_TYPE_EDGE_FALLING) + mst_set_irq(data, INTC_REV_POLARITY);
What happens if the interrupt was set to LEVEL_LOW, then switched to LEVEL_HIGH? You end up with a stuck reversed polarity.
+ + type = IRQ_TYPE_LEVEL_HIGH; + + return irq_chip_set_type_parent(data, type);
I'm going to fix this as below. Shout if you disagree. Thanks, M.
diff --git a/drivers/irqchip/irq-mst-intc.c b/drivers/irqchip/irq-mst-intc.c
index a2ab3f837b96..f6133ae28155 100644
--- a/drivers/irqchip/irq-mst-intc.c
+++ b/drivers/irqchip/irq-mst-intc.c@@ -92,16 +92,20 @@ static void mst_intc_eoi_irq(struct irq_data *d) static int mst_irq_chip_set_type(struct irq_data *data, unsigned int type) { - if (type != IRQ_TYPE_EDGE_RISING && type != IRQ_TYPE_EDGE_FALLING && - type != IRQ_TYPE_LEVEL_HIGH && type != IRQ_TYPE_LEVEL_LOW) - return -EINVAL; - - if (type == IRQ_TYPE_LEVEL_LOW || type == IRQ_TYPE_EDGE_FALLING) + switch (type) { + case IRQ_TYPE_LEVEL_LOW: + case IRQ_TYPE_EDGE_FALLING: mst_set_irq(data, INTC_REV_POLARITY); + break; + case IRQ_TYPE_LEVEL_HIGH: + case IRQ_TYPE_EDGE_RISING: + mst_clear_irq(data, INTC_REV_POLARITY); + break; + default: + return -EINVAL; + } - type = IRQ_TYPE_LEVEL_HIGH; - - return irq_chip_set_type_parent(data, type); + return irq_chip_set_type_parent(data, IRQ_TYPE_LEVEL_HIGH); } static struct irq_chip mst_intc_chip = {
--
Without deviation from the norm, progress is not possible.
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel