On Fri, Mar 29, 2024 at 01:07:31PM +0100, Krzysztof Kozlowski wrote:

On 29/03/2024 12:42, Stefano Garzarella wrote:

quoted

quoted

};

-int register_virtio_driver(struct virtio_driver *driver)
+int __register_virtio_driver(struct virtio_driver *driver, struct module *owner)
{
	/* Catch this early. */
	BUG_ON(driver->feature_table_size && !driver->feature_table);
	driver->driver.bus = &virtio_bus;
+	driver->driver.owner = owner;
+

`.driver.name =  KBUILD_MODNAME` also seems very common, should we put
that in the macro as well?

This is a bit different thing. Every driver is expected to set owner to
itself (THIS_MODULE), but is every driver name KBUILD_MODNAME?

Nope, IIUC we have 2 exceptions:
- drivers/firmware/arm_scmi/virtio.c
- arch/um/drivers/virt-pci.c

Remember that this overrides whatever driver actually put there.

They can call __register_virtio_driver() where we can add the `name`
parameter. That said, I don't have a strong opinion, we can leave it
as it is.

Thanks,
Stefano