Re: [PATCH 4/4] pinctrl: rockchip: Protect read-modify-write with the spinlock

[Doug Anderson]

Hi,

On Mon, Oct 20, 2014 at 4:27 PM, Doug Anderson <dianders@xxxxxxxxxxxx> wrote:
> @@ -1464,15 +1474,20 @@ static int rockchip_irq_set_type(struct irq_data *d, unsigned int type)
>         if (ret < 0)
>                 return ret;
>
> +       spin_lock_irqsave(&bank->slock, flags);
> +
>         data = readl_relaxed(bank->reg_base + GPIO_SWPORT_DDR);
>         data &= ~mask;
>         writel_relaxed(data, bank->reg_base + GPIO_SWPORT_DDR);
>
> +       spin_unlock_irqrestore(&bank->slock, flags);
> +
>         if (type & IRQ_TYPE_EDGE_BOTH)
>                 __irq_set_handler_locked(d->irq, handle_edge_irq);
>         else
>                 __irq_set_handler_locked(d->irq, handle_level_irq);
>
> +       spin_lock_irqsave(&bank->slock, flags);
>         irq_gc_lock(gc);
>
>         level = readl_relaxed(gc->reg_base + GPIO_INTTYPE_LEVEL);
> @@ -1522,6 +1537,7 @@ static int rockchip_irq_set_type(struct irq_data *d, unsigned int type)
>         writel_relaxed(polarity, gc->reg_base + GPIO_INT_POLARITY);
>
>         irq_gc_unlock(gc);
> +       spin_unlock_irqrestore(&bank->slock, flags);

I just reviewed this again myself and I realized that there was an
error path case that I missed.  I'll plan to spin tomorrow and include
any extra feedback people have.

I'll also note that I'd love any extra scrutiny people have on this
patch.  I'm submitting it totally based on code inspection and am
nowhere near an expert on this driver.  I mostly just looked for all
of the read/modify/wites using writel and added the lock...

-Doug
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/