From: Marc Zyngier <marc.zyngier@arm.com>
To: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Geert Uytterhoeven <geert+renesas@glider.be>,
Thomas Gleixner <tglx@linutronix.de>,
Jason Cooper <jason@lakedaemon.net>,
Rob Herring <robh+dt@kernel.org>,
Mark Rutland <mark.rutland@arm.com>,
Simon Horman <horms@verge.net.au>,
Magnus Damm <magnus.damm@gmail.com>,
Chris Brandt <chris.brandt@renesas.com>,
"open list:OPEN FIRMWARE AND FLATTENED DEVICE TREE BINDINGS"
<devicetree@vger.kernel.org>,
Linux-Renesas <linux-renesas-soc@vger.kernel.org>,
Linux Kernel Mailing List <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH 2/5] irqchip: Add Renesas RZ/A1 Interrupt Controller driver
Date: Mon, 29 Apr 2019 12:36:47 +0100 [thread overview]
Message-ID: <e6c6dfe7-f67d-90a9-eee5-d30b94c7c978@arm.com> (raw)
In-Reply-To: <CAMuHMdU3i7vqs9hd7rfvYH8QtqvwUB5vgsa_fwo5L4E3DQ_d1Q@mail.gmail.com>
On 29/04/2019 12:21, Geert Uytterhoeven wrote:
> Hi Marc,
>
> On Mon, Apr 29, 2019 at 12:07 PM Marc Zyngier <marc.zyngier@arm.com> wrote:
>> On 29/04/2019 10:36, Geert Uytterhoeven wrote:
>>> Add a driver for the Renesas RZ/A1 Interrupt Controller.
>>>
>>> This supports using up to 8 external interrupts on RZ/A1, with
>>> configurable sense select.
>>>
>>> NMI edge select is not yet supported.
>>>
>>> Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
>
>>> --- /dev/null
>>> +++ b/drivers/irqchip/irq-renesas-rza1.c
>
>>> +static void rza1_irqc_eoi(struct irq_data *d)
>>> +{
>>> + struct rza1_irqc_priv *priv = irq_data_to_priv(d);
>>> + unsigned int bit = BIT(irqd_to_hwirq(d));
>>
>> Please use u32 instead of "unsigned int" for something that operates on
>> HW registers.
>
> Even for 16-bit registers?
Ah, I'm so used to see 32bit accessors everywhere that I missed the fact
that this is a 16bit MMIO. How bizarre! ;-)
In general, try to have types that do match the actual size of the MMIO
access. There are a few exceptions to this rule (using an unsigned long
to be able to use bitmap operations, for example), but that's the
general idea.
>
>>> + u16 tmp;
>>> +
>>> + tmp = readw(priv->base + IRQRR);
>>
>> Same thing here. It's less confusing to use a u32 and mask out the top
>> bits if needed rather than having this implicit cast (applies all over
>> the code).
>
> ... so yes.
To sum it up:
readw/writew -> u16
readl/writel -> u32
>
>>
>>> + if (tmp & bit)
>>> + writew(GENMASK(IRQC_NUM_IRQ - 1, 0) & ~bit, priv->base + IRQRR);
>>
>> Please use the _relaxed accessors all over the driver, you really do not
>> need a DSB on each of these accesses.
>
> OK.
>
>>> +static int rza1_irqc_set_type(struct irq_data *d, unsigned int type)
>>> +{
>>> + struct rza1_irqc_priv *priv = irq_data_to_priv(d);
>>> + unsigned int hw_irq = irqd_to_hwirq(d);
>>> + u16 sense, tmp;
>>> +
>>> + switch (type & IRQ_TYPE_SENSE_MASK) {
>>> + case IRQ_TYPE_LEVEL_LOW:
>>> + sense = ICR1_IRQS_LEVEL_LOW;
>>> + break;
>>> +
>>> + case IRQ_TYPE_EDGE_FALLING:
>>> + sense = ICR1_IRQS_EDGE_FALLING;
>>> + break;
>>> +
>>> + case IRQ_TYPE_EDGE_RISING:
>>> + sense = ICR1_IRQS_EDGE_RISING;
>>> + break;
>>> +
>>> + case IRQ_TYPE_EDGE_BOTH:
>>> + sense = ICR1_IRQS_EDGE_BOTH;
>>> + break;
>>> +
>>> + default:
>>> + return -EINVAL;
>>> + }
>>> +
>>> + tmp = readw(priv->base + ICR1);
>>> + tmp &= ~ICR1_IRQS_MASK(hw_irq);
>>> + tmp |= ICR1_IRQS(hw_irq, sense);
>>> + writew(tmp, priv->base + ICR1);
>>> + return 0;
>>
>> Don't you need to propagate the trigger configuration to the parent irqchip?
>
> No, the line to the parent GIC is always configured for high-level.
>
>>> +static int rza1_irqc_alloc(struct irq_domain *domain, unsigned int virq,
>>> + unsigned int nr_irqs, void *arg)
>>> +{
>>> + struct rza1_irqc_priv *priv = domain->host_data;
>>> + struct irq_fwspec *fwspec = arg;
>>> + struct irq_fwspec spec;
>>> + int ret;
>>> +
>>> + ret = irq_domain_set_hwirq_and_chip(domain, virq, fwspec->param[0],
>>> + &priv->chip, priv);
>>> + if (ret)
>>> + return ret;
>>> +
>>> + spec.fwnode = &priv->dev->of_node->fwnode;
>>> + spec.param_count = 3;
>>> + spec.param[0] = GIC_SPI;
>>> + spec.param[1] = priv->gic_spi_base + fwspec->param[0];
>>> + spec.param[2] = IRQ_TYPE_LEVEL_HIGH;
>>
>> This is related to my earlier question: Does this block turn everything
>> into level interrupts?
>
> That is my understanding of the hardware:
> - Low-level interrupts are cleared when input becomes high again,
> - Rising/falling/both edge interrupts are cleared by reading+writing IRQRR.
>
> FTR, the Hardware User Manual is available from
> https://www.renesas.com/eu/en/products/microcontrollers-microprocessors/rz/rza/rza1h.html#documents
> (Section 7. Interrupt Controller).
OK, thanks for the detailed explanation.
>
>>> +static int rza1_irqc_probe(struct platform_device *pdev)
>>> +{
>
>>> + priv->chip.name = dev_name(dev);
>>
>> name should normally be used to identify the overall "class" of
>
> OK, replacing by "rza1-irqc".
>
>> interrupt. .device is what should be used for the device itself.
>
> You mean .parent_device?
> Been there, done that: if I fill that in with "dev", it fails with
>
> gpio-keys keyboard: Unable to claim irq 41; error -13
> gpio-keys: probe of keyboard failed with error -13
>
> due to the call to pm_runtime_get_sync() in irq_chip_pm_get() failing.
> This driver doesn't have (and doesn't need) Runtime PM.
OK, fair enough. Who needs PM anyway? ;-)
>
>>> +struct rza1_irqc_info rza1_irqc_info = {
>>> + .gic_spi_base = 0,
>>> +};
>>
>> To answer your question in the cover letter, I'd rather this came from
>> DT. And otherwise, it should be be static.
>
> (Oops, forget the "static const")
>
> Using a custom property, or derived from 8 interrupt specifiers in the
> interrupts property?
A custom property is fine by me (everybody does that anyway).
>
>> It otherwise looks good to me. If you respin it quickly enough, I'm
>> happy to take it for 5.2.
>
> Thanks, will do tomorrow, so Chris (in NC.US; let's hope he doesn't
> celebrate Golden Week) has a chance to comment...
Thanks,
M.
--
Jazz is not dead. It just smells funny...
next prev parent reply other threads:[~2019-04-29 11:36 UTC|newest]
Thread overview: 16+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-04-29 9:36 [PATCH 0/5] ARM: rskrza1: Add RZ/A1 IRQC and input switches Geert Uytterhoeven
2019-04-29 9:36 ` [PATCH 1/5] dt-bindings: interrupt-controller: Add Renesas RZ/A1 Interrupt Controller Geert Uytterhoeven
2019-04-29 9:36 ` [PATCH 2/5] irqchip: Add Renesas RZ/A1 Interrupt Controller driver Geert Uytterhoeven
2019-04-29 10:06 ` Marc Zyngier
2019-04-29 11:21 ` Geert Uytterhoeven
2019-04-29 11:36 ` Marc Zyngier [this message]
2019-04-29 17:31 ` Chris Brandt
2019-04-29 12:25 ` Chris Brandt
2019-04-29 9:36 ` [PATCH 3/5] soc: renesas: Enable RZ/A1 IRQC on RZ/A1H Geert Uytterhoeven
2019-04-29 12:58 ` Simon Horman
2019-04-29 9:36 ` [PATCH 4/5] ARM: dts: r7s72100: Add IRQC device node Geert Uytterhoeven
2019-04-29 13:11 ` Simon Horman
2019-04-29 9:36 ` [PATCH 5/5] ARM: dts: rskrza1: Add input switches Geert Uytterhoeven
2019-04-29 12:21 ` [PATCH 0/5] ARM: rskrza1: Add RZ/A1 IRQC and " Chris Brandt
2019-04-29 12:49 ` Geert Uytterhoeven
2019-04-29 13:14 ` Chris Brandt
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=e6c6dfe7-f67d-90a9-eee5-d30b94c7c978@arm.com \
--to=marc.zyngier@arm.com \
--cc=chris.brandt@renesas.com \
--cc=devicetree@vger.kernel.org \
--cc=geert+renesas@glider.be \
--cc=geert@linux-m68k.org \
--cc=horms@verge.net.au \
--cc=jason@lakedaemon.net \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-renesas-soc@vger.kernel.org \
--cc=magnus.damm@gmail.com \
--cc=mark.rutland@arm.com \
--cc=robh+dt@kernel.org \
--cc=tglx@linutronix.de \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).