Re: [PATCH v5 2/5] irqchip: Add RZ/G2L IA55 Interrupt Controller driver

[Marc Zyngier <maz@kernel.org>]

On Sat, 25 Jun 2022 11:54:44 +0100,
"Lad, Prabhakar" <prabhakar.csengg@gmail.com> wrote:
> 
> Hi Marc,
> 
> Thank you for the review.
> 
> On Sat, Jun 25, 2022 at 10:30 AM Marc Zyngier <maz@kernel.org> wrote:
> >
> > On Mon, 23 May 2022 18:42:35 +0100,
> > Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com> wrote:
> > >

[...]

> > > +static int rzg2l_irqc_alloc(struct irq_domain *domain, unsigned int virq,
> > > +                         unsigned int nr_irqs, void *arg)
> > > +{
> > > +     struct rzg2l_irqc_priv *priv = domain->host_data;
> > > +     unsigned long *chip_data = NULL;
> >
> > Why the init to NULL?
> >
> Can be dropped.
> 
> > > +     struct irq_fwspec spec;
> > > +     irq_hw_number_t hwirq;
> > > +     int tint = -EINVAL;
> > > +     unsigned int type;
> > > +     unsigned int i;
> > > +     int ret;
> > > +
> > > +     ret = irq_domain_translate_twocell(domain, arg, &hwirq, &type);
> > > +     if (ret)
> > > +             return ret;
> > > +
> > > +     /*
> > > +      * For TINT interrupts ie where pinctrl driver is child of irqc domain
> > > +      * the hwirq and TINT are encoded in fwspec->param[0].
> > > +      * hwirq for TINT range from 9-40, hwirq is embedded 0-15 bits and TINT
> > > +      * from 16-31 bits. TINT from the pinctrl driver needs to be programmed
> > > +      * in IRQC registers to enable a given gpio pin as interrupt.
> > > +      */
> > > +     if (hwirq > IRQC_IRQ_COUNT) {
> > > +             tint = TINT_EXTRACT_GPIOINT(hwirq);
> > > +             hwirq = TINT_EXTRACT_HWIRQ(hwirq);
> > > +
> > > +             if (hwirq < IRQC_TINT_START)
> > > +                     return -EINVAL;
> > > +     }
> > > +
> > > +     if (hwirq > (IRQC_NUM_IRQ - 1))
> > > +             return -EINVAL;
> > > +
> > > +     chip_data = kzalloc(sizeof(*chip_data), GFP_KERNEL);
> >
> > Are we really allocating an unsigned long for something that already
> > fits in something that is pointer-sized?
> >
> I think I received some feedback to use unsigned long.  Let me know
> what you want me to use here.

I think this is just a waste of memory, but I don't really care.

> 
> > > +     if (!chip_data)
> > > +             return -ENOMEM;
> > > +     *chip_data = tint;
> >
> > So here, *chip_data can be set to -EINVAL if hwirq <= IRQC_IRQ_COUNT?
> > This can't be right.
> >
> Yes *chip_data can be -EINVAL. IRQC block handles IRQ0-7 and
> GPIOINT0-122. So the -EINVAL here is for IRQ0-7 case were dont
> required the chip data in the call backs hence -EINVAL, Whereas for
> GPIOINT0-122 we need chip_data in the callbacks as this value needs to
> be programmed in the hardware registers.

I can't see anything that checks it (let alone the difference in
types). And if it isn't checked, this means that the allocation is
pointless.

> 
> > > +
> > > +     ret = irq_domain_set_hwirq_and_chip(domain, virq, hwirq, &irqc_chip,
> > > +                                         chip_data);
> > > +     if (ret) {
> > > +             kfree(chip_data);
> > > +             return ret;
> > > +     }
> > > +
> > > +     spec.fwnode = domain->parent->fwnode;
> > > +     spec.param_count = priv->map[hwirq].args_count;
> > > +     for (i = 0; i < spec.param_count; i++)
> > > +             spec.param[i] = priv->map[hwirq].args[i];
> >
> > Why isn't that simply:
> >
> >         spec = priv->map[hwirq];
> >
> spec is of type ‘struct irq_fwspec’ and map is of type ‘struct of_phandle_args’.
> 
> > as this really is the interrupt you want to map to?
> >
> Yes.
> 
> > > +
> > > +     ret = irq_domain_alloc_irqs_parent(domain, virq, nr_irqs, &spec);
> >
> > or even better:
> >
> >         ret = irq_domain_alloc_irqs_parent(domain, virq, nr_irqs,
> >                                            &priv->map[hwirq]);
> >
> Does not work as map is of type ‘struct of_phandle_args’.

Which begs the question: why don't you convert it to an irq_fwspec the
first place and be done with it?

> 
> > > +     if (ret)
> > > +             kfree(chip_data);
> > > +
> > > +     return ret;
> > > +}
> > > +
> > > +static void rzg2l_irqc_domain_free(struct irq_domain *domain, unsigned int virq,
> > > +                                unsigned int nr_irqs)
> > > +{
> > > +     struct irq_data *d;
> > > +
> > > +     d = irq_domain_get_irq_data(domain, virq);
> > > +     if (d)
> > > +             kfree(d->chip_data);
> > > +
> > > +     irq_domain_free_irqs_common(domain, virq, nr_irqs);
> > > +}
> > > +
> > > +static const struct irq_domain_ops rzg2l_irqc_domain_ops = {
> > > +     .alloc = rzg2l_irqc_alloc,
> > > +     .free = rzg2l_irqc_domain_free,
> > > +     .translate = irq_domain_translate_twocell,
> > > +};
> > > +
> > > +static int rzg2l_irqc_parse_map(struct rzg2l_irqc_priv *priv,
> > > +                             struct device_node *np)

nit: this function could afford being renamed to something more
correct. It really doesn't map anything, only retrieves the output
interrupts.

> > > +{
> > > +     unsigned int i;
> > > +     int ret;
> > > +
> > > +     for (i = 0; i < IRQC_NUM_IRQ; i++) {
> > > +             ret = of_irq_parse_one(np, i, &priv->map[i]);

Make map an array of irq_fwspec, and use of_phandle_args_to_fwspec()
for the conversion.

> > > +             if (ret)
> > > +                     return ret;
> > > +     }
> > > +
> > > +     return 0;
> > > +}
> > > +
> > > +static int rzg2l_irqc_init(struct device_node *node, struct device_node *parent)
> > > +{
> > > +     struct irq_domain *irq_domain, *parent_domain;
> > > +     struct platform_device *pdev;
> > > +     struct reset_control *resetn;
> > > +     struct rzg2l_irqc_priv *priv;
> > > +     int ret;
> > > +
> > > +     pdev = of_find_device_by_node(node);
> > > +     if (!pdev)
> > > +             return -ENODEV;
> > > +
> > > +     parent_domain = irq_find_host(parent);
> > > +     if (!parent_domain) {
> > > +             dev_err(&pdev->dev, "cannot find parent domain\n");
> > > +             return -ENODEV;
> > > +     }
> > > +
> > > +     priv = devm_kzalloc(&pdev->dev, sizeof(*priv), GFP_KERNEL);
> > > +     if (!priv)
> > > +             return -ENOMEM;
> > > +
> > > +     priv->base = devm_of_iomap(&pdev->dev, pdev->dev.of_node, 0, NULL);
> > > +     if (IS_ERR(priv->base))
> > > +             return PTR_ERR(priv->base);
> > > +
> > > +     ret = rzg2l_irqc_parse_map(priv, node);
> > > +     if (ret) {
> > > +             dev_err(&pdev->dev, "cannot parse interrupts: %d\n", ret);
> > > +             return ret;
> > > +     }
> > > +
> > > +     resetn = devm_reset_control_get_exclusive_by_index(&pdev->dev, 0);
> > > +     if (IS_ERR(resetn))
> > > +             return IS_ERR(resetn);
> > > +
> > > +     ret = reset_control_deassert(resetn);
> > > +     if (ret) {
> > > +             dev_err(&pdev->dev, "failed to deassert resetn pin, %d\n", ret);
> > > +             return ret;
> > > +     }
> > > +
> > > +     pm_runtime_enable(&pdev->dev);
> > > +     ret = pm_runtime_resume_and_get(&pdev->dev);
> > > +     if (ret < 0) {
> > > +             dev_err(&pdev->dev, "pm_runtime_resume_and_get failed: %d\n", ret);
> > > +             goto pm_disable;
> > > +     }
> >
> > If using runtime PM, why isn't the core IRQ code made aware of this
> > dependency by registering the device with irq_domain_set_pm_device()
> > instead of leaving it enabled forever?
> >
> Ouch will add irq_domain_set_pm_device() below.

You'll need a bit more than that. You'll either need to take a PM
reference on each alloc, or improve irq_chip_pm_{get,put}() to talk
the hierarchy.

That's probably a separate patch.

	M.

-- 
Without deviation from the norm, progress is not possible.