On Tue, Sep 10 2024 at 18:18, Stanimir Varbanov wrote:

+
+struct mip_priv {
+	/* used to protect bitmap alloc/free */
+	spinlock_t lock;
+	void __iomem *base;
+	u64 msg_addr;
+	u32 msi_base;
+	u32 num_msis;
+	unsigned long *bitmap;
+	struct irq_domain *parent;

https://www.kernel.org/doc/html/latest/process/maintainer-tip.html#struct-declarations-and-initializers

And please read the rest of the document too.

+};
+
+static void mip_mask_msi_irq(struct irq_data *d)
+{
+	pci_msi_mask_irq(d);
+	irq_chip_mask_parent(d);
+}
+
+static void mip_unmask_msi_irq(struct irq_data *d)
+{
+	pci_msi_unmask_irq(d);
+	irq_chip_unmask_parent(d);

This is asymmetric vs. mask(), but that's just the usual copy & pasta
problem.

+}
+static int mip_init_domains(struct mip_priv *priv, struct device_node *np)
+{
+	struct irq_domain *middle_domain, *msi_domain;
+
+	middle_domain = irq_domain_add_hierarchy(priv->parent, 0,
+						 priv->num_msis, np,
+						 &mip_middle_domain_ops,
+						 priv);
+	if (!middle_domain)
+		return -ENOMEM;
+
+	msi_domain = pci_msi_create_irq_domain(of_node_to_fwnode(np),
+					       &mip_msi_domain_info,
+					       middle_domain);
+	if (!msi_domain) {
+		irq_domain_remove(middle_domain);
+		return -ENOMEM;
+	}

This is not much different. Please convert this to a proper MSI parent
domain and let the PCI/MSI core handle the PCI/MSI part.

See

https://git.kernel.org/pub/scm/linux/kernel/git/tip/tip.git/log/?h=irq-msi-2024-07-22

for reference.

Thanks,

        tglx