Re: [PATCH] pinctrl: sunxi: dt: number the IRQ banks from the first one that has IRQs

[Linus Walleij]

On Tue, Sep 29, 2026 at 10:37 PM Pablo Mazzini <pmazzini@xxxxxxxxx> wrote:

> prepare_function_table() assumes the first pin is in bank 0. On the
> A523, which has no PA pins, every IRQ bank is numbered one too high:
> the last bank writes past the end of irq_array,

Why don't you add a patch to protect against that?

Something like this (completely untested, not even compiled)

diff --git a/drivers/pinctrl/sunxi/pinctrl-sunxi.c
b/drivers/pinctrl/sunxi/pinctrl-sunxi.c
index 31cd142ce0f7..c0c90d9e3215 100644
--- a/drivers/pinctrl/sunxi/pinctrl-sunxi.c
+++ b/drivers/pinctrl/sunxi/pinctrl-sunxi.c
@@ -1632,9 +1632,10 @@ int sunxi_pinctrl_init_with_flags(struct
platform_device *pdev,
        else
                pctl->pow_mod_sel_offset = PIO_POW_MOD_SEL_REG;

+       pctl->irqs = IRQ_PER_BANK * pctl->desc->irq_banks;
        pctl->irq_array = devm_kcalloc(&pdev->dev,
-                                      IRQ_PER_BANK * pctl->desc->irq_banks,
-                                      sizeof(*pctl->irq_array),
+                                      pctl->irqs,
+                                      sizeof(unsigned int),
                                       GFP_KERNEL);
        if (!pctl->irq_array)
                return -ENOMEM;
diff --git a/drivers/pinctrl/sunxi/pinctrl-sunxi.h
b/drivers/pinctrl/sunxi/pinctrl-sunxi.h
index 8bd00c6ff628..9e0ed9c0c1ee 100644
--- a/drivers/pinctrl/sunxi/pinctrl-sunxi.h
+++ b/drivers/pinctrl/sunxi/pinctrl-sunxi.h
@@ -176,7 +176,8 @@ struct sunxi_pinctrl {
        struct sunxi_pinctrl_group      *groups;
        unsigned                        ngroups;
        int                             *irq;
-       unsigned                        *irq_array;
+       unsigned                        irq_array[] __counted_by(irqs);
+       unsigned                        irqs;
        raw_spinlock_t                  lock;
        /*
         * Output latch shadow, one word per bank.  Seeded lockless at


> and interrupts are
> looked up at the wrong offsets.
>
> Count the IRQ banks from the first bank that has interrupts.
(...)
>         irq_bank = 0;
> -       last_bank = 0;
> +       last_bank = -1;
>         for (i = 0; i < npins; i++) {
>                 struct sunxi_desc_pin *pin = &pins[i];
>                 int bank = (pin->pin.number - pin_base) / PINS_PER_BANK;
> @@ -221,16 +221,14 @@ static int prepare_function_table(struct device *dev, struct device_node *pnode,
>                 func[1].muxval = 1;
>
>                 if (irq_mux) {
> -                       if (bank > last_bank)
> +                       if (last_bank >= 0 && bank > last_bank)
>                                 irq_bank++;
>                         func[lastfunc].muxval = irq_mux;
>                         func[lastfunc].irqbank = irq_bank;
>                         func[lastfunc].irqnum = pin->pin.number % PINS_PER_BANK;
>                         func[lastfunc].name = "irq";
> -               }
> -
> -               if (bank > last_bank)
>                         last_bank = bank;
> +               }

This looks reasonable.

Yours,
Linus Walleij