[...]

quoted

quoted

@@ -1009,6 +1012,16 @@ gic_of_init(struct device_node *node, struct device_node *parent)
         if (of_property_read_u32(node, "cpu-offset", &percpu_offset))
                 percpu_offset = 0;

+       gic_data[gic_cnt].irq_chip = &gic_chip;
+
+       /* Currently, we only support one v2m subnode. */
+       child = of_get_child_by_name(node, "v2m");
+       if (child) {
+               ret = gicv2m_of_init(child, &gic_data[gic_cnt]);
+               if (ret)
+                       return ret;
+       }

I can't see how you'd sanely expand this to multiple children, which was
the main point of having a separate node for the M block.

Give the M block a compatible string and look for children with that
string.

Mark,

I am making change in the struct gic_chip_data to contain "v2m_list" (instead of just
a single struct v2m_data). This way, it is clear on how we should handle multiple v2m nodes
within a GIC.

Ok.

As for the device tree binding, in order to handle multiple v2m nodes within a GIC,
it should not require adding another compatibility ID as it seems too complicate
to have GIC node with multiple compat IDs).

I don't follow.

How does each sub-node having a compatible string complicate the GIC
node?

Mark.