[PATCH 2/3] irqchip: Add the Sophgo SG2042 MSI interrupt controller

[Chen Wang]

On 2024/11/13 14:14, Thomas Gleixner wrote:
> On Mon, Nov 11 2024 at 12:01, Chen Wang wrote:
>> +struct sg2042_msi_data {
>> +	void __iomem	*reg_clr; /* clear reg, see TRM, 10.1.33, GP_INTR0_CLR */
> Please make these tail comments tabular aligned so they actually stand
> out.
>
>    https://www.kernel.org/doc/html/latest/process/maintainer-tip.html#comment-style
Got, will fix this.
>> +
>> +	u64		doorbell_addr; /* see TRM, 10.1.32, GP_INTR0_SET */
>> +
>> +	u32		irq_first; /* The vector number that MSIs starts */
>> +	u32		num_irqs;  /* The number of vectors for MSIs */
>> +
>> +	unsigned long	*msi_map;
>> +	struct mutex	msi_map_lock; /* lock for msi_map */
>> +};
>> +
>> +static int sg2042_msi_allocate_hwirq(struct sg2042_msi_data *priv, int num_req)
>> +{
>> +	int first;
>> +
>> +	mutex_lock(&priv->msi_map_lock);
> Please use
>
>         guard(mutex)(&priv->msi_map_lock);
>
> which removes all the mutex_unlock() hackery and boils this down
Thanks, will double check.
>
>> +
>> +	first = bitmap_find_free_region(priv->msi_map, priv->num_irqs,
>> +					get_count_order(num_req));
>> +	if (first < 0) {
>> +		mutex_unlock(&priv->msi_map_lock);
>> +		return -ENOSPC;
>> +	}
>> +
>> +	mutex_unlock(&priv->msi_map_lock);
>> +
>> +	return priv->irq_first + first;
> to
>
> 	guard(mutex)(&priv->msi_map_lock);
> 	first = bitmap_find_free_region(priv->msi_map, priv->num_irqs,
> 					get_count_order(num_req));
> 	return first >= 0 ? priv->irq_first + first : -ENOSPC;
>
> See?
>
>> +}
>> +
>> +static void sg2042_msi_free_hwirq(struct sg2042_msi_data *priv,
>> +				  int hwirq, int num_req)
>> +{
>> +	int first = hwirq - priv->irq_first;
>> +
>> +	mutex_lock(&priv->msi_map_lock);
> Ditto.
>
>> +	bitmap_release_region(priv->msi_map, first, get_count_order(num_req));
>> +	mutex_unlock(&priv->msi_map_lock);
>> +}
>> +static void sg2042_msi_irq_compose_msi_msg(struct irq_data *data,
>> +					   struct msi_msg *msg)
>> +{
>> +	struct sg2042_msi_data *priv = irq_data_get_irq_chip_data(data);
>> +
>> +	msg->address_hi = upper_32_bits(priv->doorbell_addr);
>> +	msg->address_lo = lower_32_bits(priv->doorbell_addr);
>> +	msg->data = 1 << (data->hwirq - priv->irq_first);
>> +
>> +	pr_debug("%s hwirq[%d]: address_hi[%#x], address_lo[%#x], data[%#x]\n",
>> +		 __func__,
> No point in having this line break. You have 100 characters. Please fix
> this all over the place.
Got.
>
>> +		 (int)data->hwirq, msg->address_hi, msg->address_lo, msg->data);
> (int) ? Why can't you use the proper conversion specifier instead of %d?
Will double-check.
>
>> +static int sg2042_msi_middle_domain_alloc(struct irq_domain *domain,
>> +					  unsigned int virq,
>> +					  unsigned int nr_irqs, void *args)
>> +{
>> +	struct sg2042_msi_data *priv = domain->host_data;
>> +	int hwirq, err, i;
>> +
>> +	hwirq = sg2042_msi_allocate_hwirq(priv, nr_irqs);
>> +	if (hwirq < 0)
>> +		return hwirq;
>> +
>> +	for (i = 0; i < nr_irqs; i++) {
>> +		err = sg2042_msi_parent_domain_alloc(domain, virq + i, hwirq + i);
>> +		if (err)
>> +			goto err_hwirq;
>> +
>> +		pr_debug("%s: virq[%d], hwirq[%d]\n",
>> +			 __func__, virq + i, (int)hwirq + i);
> No line break required.
>
>> +		irq_domain_set_hwirq_and_chip(domain, virq + i, hwirq + i,
>> +					      &sg2042_msi_middle_irq_chip, priv);
>> +	}
>> +static int sg2042_msi_init_domains(struct sg2042_msi_data *priv,
>> +				   struct device_node *node)
>> +{
>> +	struct irq_domain *plic_domain, *middle_domain;
>> +	struct device_node *plic_node;
>> +	struct fwnode_handle *fwnode = of_node_to_fwnode(node);
> https://www.kernel.org/doc/html/latest/process/maintainer-tip.html#variable-declarations
Thanks, will double-check.
>> +	if (!of_find_property(node, "interrupt-parent", NULL)) {
>> +		pr_err("Can't find interrupt-parent!\n");
>> +		return -EINVAL;
>> +	}
>> +
>> +	plic_node = of_irq_find_parent(node);
>> +	if (!plic_node) {
>> +		pr_err("Failed to find the PLIC node!\n");
>> +		return -ENXIO;
>> +	}
>> +
>> +	plic_domain = irq_find_host(plic_node);
>> +	of_node_put(plic_node);
>> +	if (!plic_domain) {
>> +		pr_err("Failed to find the PLIC domain\n");
>> +		return -ENXIO;
>> +	}
>> +
>> +	middle_domain = irq_domain_create_hierarchy(plic_domain, 0, priv->num_irqs,
>> +						    fwnode,
>> +						    &pch_msi_middle_domain_ops,
>> +						    priv);
> So now you have created a domain. How is that supposed to be used by the
> PCI layer?

Here I create the domain and attached it to the fwnode. In PCI driver, 
it can set this msi controller as its ""interrupt-parent" and find the 
domain attached as below:

static int pcie_probe(struct platform_device *pdev)
{
     struct device *dev = &pdev->dev;
     parent_node = of_irq_find_parent(dev->of_node);
     parent_domain = irq_find_host(parent_node);
     ...
}

>> +	if (!middle_domain) {
>> +		pr_err("Failed to create the MSI middle domain\n");
>> +		return -ENOMEM;
>> +	}
>> +
>> +	return 0;
>> +}
>> +static int sg2042_msi_probe(struct platform_device *pdev)
>> +{
> ....
>
>> +	data->msi_map = bitmap_zalloc(data->num_irqs, GFP_KERNEL);
>> +	if (!data->msi_map)
>> +		return -ENOMEM;
>> +
>> +	return sg2042_msi_init_domains(data, pdev->dev.of_node);
> In case of error this leaks data->msi_map, no?
Thanks, I will correct this.
>> +static struct platform_driver sg2042_msi_driver = {
>> +	.driver = {
>> +		.name = "sg2042-msi",
>> +		.of_match_table = of_match_ptr(sg2042_msi_of_match),
>> +	},
>> +	.probe = sg2042_msi_probe,
>> +};
> Please see the documentation I pointed you to above and search for
> struct initializers.
>
> Thanks,
>
>          tglx