[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <CAMuHMdXqG_qp0XL7OLLdHKpYJ=xAG+hTMxPctAbeBfUV9EpodQ@mail.gmail.com>
Date: Mon, 24 Jan 2022 10:57:44 +0100
From: Geert Uytterhoeven <geert@...ux-m68k.org>
To: Nikita Yushchenko <nikita.yoush@...entembedded.com>
Cc: Thomas Gleixner <tglx@...utronix.de>,
Marc Zyngier <maz@...nel.org>,
Eugeniu Rosca <erosca@...adit-jv.com>,
Linux Kernel Mailing List <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] drivers: irqchip: add irq-type-changer
Hi Nikita,
On Wed, Jan 19, 2022 at 9:17 PM Nikita Yushchenko
<nikita.yoush@...entembedded.com> wrote:
> Irq type changer is a virtual irqchip useful to support boards that
> change (e.g. invert) interrupt signal between producer and consumer.
>
> Usage example, for Kingfisher extension board for Renesas Gen-3 Soc,
> that has WiFi interrupt delivered over inverting level-shifter:
>
> / {
> gpio1_25_inverted: inverter {
> compatible = "linux,irq-type-changer";
> interrupt-controller;
> #interrupt-cells = <2>;
> interrupt-parent = <&gpio1>;
> interrupts = <25 IRQ_TYPE_EDGE_FALLING>;
> };
> };
>
> &wlcore {
> interrupt-parent = <&gpio1_25_inverted>;
> interrupts = <0 IRQ_TYPE_EDGE_RISING>;
> };
>
> Then, wlcore driver observes IRQ_TYPE_EDGE_RISING trigger type and
> configures interrupt output as such. At the same time, gpio-rcar driver
> gets interrupt configured for IRQ_TYPE_EDGE_FALLING.
>
> This version uses hierarchical irq_domain API, and works only with
> parent interrupt domains compatible with that API.
>
> Signed-off-by: Nikita Yushchenko <nikita.yoush@...entembedded.com>
Thanks for your patch!
> --- /dev/null
> +++ b/drivers/irqchip/irq-type-changer.c
> @@ -0,0 +1,162 @@
> +// SPDX-License-Identifier: GPL-2.0
> +
> +#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
> +
> +#include <linux/irqchip.h>
> +#include <linux/irqdomain.h>
> +#include <linux/of_irq.h>
> +
> +struct changer {
> + unsigned long count;
> + struct {
> + struct irq_fwspec fwspec;
> + unsigned int type;
> + } out[0];
Please use [] instead of [0] for flexible arrays, else the compiler
doesn't realize it is a flexible array, and won't complain if you add
more members below.
> +};
> +static int __init changer_of_init(struct device_node *node,
> + struct device_node *parent)
> +{
> + struct irq_domain *domain, *parent_domain;
> + int count, i, ret;
> + struct changer *ch;
> + struct of_phandle_args pargs;
> + irq_hw_number_t unused;
> +
> + if (!parent) {
> + pr_err("%pOF: no parent node\n", node);
> + return -EINVAL;
> + }
> +
> + parent_domain = irq_find_host(parent);
> + if (!parent_domain) {
> + pr_err("%pOF: no parent domain\n", node);
> + return -EINVAL;
> + }
> +
> + if (WARN_ON(!parent_domain->ops->translate))
> + return -EINVAL;
> +
> + count = of_irq_count(node);
> + if (count < 1) {
> + pr_err("%pOF: no interrupts defined\n", node);
> + return -EINVAL;
> + }
> +
> + ch = kzalloc(GFP_KERNEL, sizeof(*ch) + count * sizeof(ch->out[0]));
Oops, wrong parameter order, as detected by the kernel test robot.
Please use struct_size() to simplify and harden the size calculation.
Gr{oetje,eeting}s,
Geert
--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@...ux-m68k.org
In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds
Powered by blists - more mailing lists