linux-kernel - Re: [PATCH v2] irqchip/irq-mst: Support polarity configuration

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20210311161140.32678-1-mark-pk.tsai@mediatek.com>
Date:   Fri, 12 Mar 2021 00:11:40 +0800
From:   Mark-PK Tsai <mark-pk.tsai@...iatek.com>
To:     <daniel@...f.com>, Mark-PK Tsai <mark-pk.tsai@...iatek.com>
CC:     <daniel@...ngy.jp>, <linux-arm-kernel@...ts.infradead.org>,
        <linux-kernel@...r.kernel.org>,
        <linux-mediatek@...ts.infradead.org>, <matthias.bgg@...il.com>,
        <maz@...nel.org>, <tglx@...utronix.de>, <yj.chiang@...iatek.com>
Subject: Re: [PATCH v2] irqchip/irq-mst: Support polarity configuration

From: Daniel Palmer <daniel@...f.com>

> On Thu, 11 Mar 2021 at 12:12, Mark-PK Tsai <mark-pk.tsai@...iatek.com> wrote:
> > For a fiq controller, the input edge signal will be convert to level and
> > keep the interrupt status until we do EOI operation.
> > That means if a rising edge input if trigger the ouput line will keep high
> > until we clear the interrupt status.
> 
> I think maybe the fiq is always edge triggered?
> It seems like it latches on an edge and holds it's output to the GIC
> high until it is reset by eoi and then only triggers again on another
> edge.
> I can experiment to confirm that's what it actually does for the chips I have.
> 
> Then it seems like the irq version is almost just a configurable
> inverter that passes either the input signal or the inverted input
> signal to the GIC.
> 
> So maybe fiq should only accept edge type interrupts and irq could
> accept either?

Why irq could accept either?
And if it's a fiq controller, we can just remove 'mstar,intc-no-eoi' from the
device node.
Whether the controller is fiq or irq, the interrupt request from mst-intc to
parent GIC is level sensitive. Because if the source is edge triggered, mst-intc
always latch it.

> 
> >  static int mst_irq_chip_set_type(struct irq_data *data, unsigned int type)
> >  {
> > -       if (type != IRQ_TYPE_LEVEL_LOW && type != IRQ_TYPE_LEVEL_HIGH)
> > -               return -EINVAL;
> > -
> > -       if (type == IRQ_TYPE_LEVEL_LOW) {
> > +       if (type == IRQ_TYPE_EDGE_FALLING) {
> > +               mst_set_irq(data, INTC_REV_POLARITY);
> > +               type = IRQ_TYPE_EDGE_RISING;

The interrupt triggered type of the parent GIC should be active high,
I will modify it.

> > +       } else if (type == IRQ_TYPE_LEVEL_LOW) {
> >                 mst_set_irq(data, INTC_REV_POLARITY);
> >                 type = IRQ_TYPE_LEVEL_HIGH;
> >         }
> 
> I think this still needs the logic to check that type is something we
> can handle (not IRQ_TYPE_EDGE_BOTH) and maybe if the fiq controller

Agree.

> can only do edge interrupts level types should return -EINVAL?

No matter what triggered type of an irq we write in dts is edge or level triggered,
the init and handling flow is the same except we need to do eoi if the
controller doesn't have the property 'mstar,intc-no-eoi'.
So maybe we don't need to do extra work to check the type for an fiq or irq controller?

And I will update the patch as following:

diff --git a/drivers/irqchip/irq-mst-intc.c b/drivers/irqchip/irq-mst-intc.c
index 841b9b1c2699..f46ade7b1775 100644
--- a/drivers/irqchip/irq-mst-intc.c
+++ b/drivers/irqchip/irq-mst-intc.c
@@ -90,13 +90,14 @@ 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_LEVEL_LOW && type != IRQ_TYPE_LEVEL_HIGH)
+	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) {
+	if (type == IRQ_TYPE_LEVEL_LOW || type == IRQ_TYPE_EDGE_FALLING)
 		mst_set_irq(data, INTC_REV_POLARITY);
-		type = IRQ_TYPE_LEVEL_HIGH;
-	}
+
+	type = IRQ_TYPE_LEVEL_HIGH;
 
 	return irq_chip_set_type_parent(data, type);
 }
@@ -219,11 +220,12 @@ static int mst_intc_domain_alloc(struct irq_domain *domain, unsigned int virq,
 	parent_fwspec.param[1] = cd->irq_start + hwirq;
 
 	/*
-	 * If the irq signal is active low, configure it to active high
-	 * to meet GIC SPI spec in mst_irq_chip_set_type via REV_POLARITY bit
+	 * mst-intc latch the interrupt request if it's edge triggered,
+	 * so the output signal to parent GIC is always level sensitive.
+	 * And if the irq signal is active low, configure it to active high
+	 * to meet GIC SPI spec in mst_irq_chip_set_type via REV_POLARITY bit.
 	 */
-	if (fwspec->param[2] == IRQ_TYPE_LEVEL_LOW)
-		parent_fwspec.param[2] = IRQ_TYPE_LEVEL_HIGH;
+	parent_fwspec.param[2] = IRQ_TYPE_LEVEL_HIGH;
 
 	return irq_domain_alloc_irqs_parent(domain, virq, nr_irqs, &parent_fwspec);
 }