Re: [PATCH 05/16] drivers/fsi: Add fake master driver

From: Mark Rutland
Date: Wed Dec 07 2016 - 07:09:58 EST


On Tue, Dec 06, 2016 at 06:14:26PM -0600, Chris Bostic wrote:
> From: Jeremy Kerr <jk@xxxxxxxxxx>
>
> For debugging, add a fake master driver, that only supports reads,
> returning a fixed set of data.

> +config FSI_MASTER_FAKE
> + tristate "Fake FSI master"
> + depends on FSI
> + ---help---
> + This option enables a fake FSI master driver for debugging.
> +endif

> +static const struct of_device_id fsi_master_fake_match[] = {
> + { .compatible = "ibm,fsi-master-fake" },
> + { },
> +};

NAK.

DT should be treated as an ABI, and should describe the HW explicitly.
This makes no sense. This is also missing a binding document.

Have your module take a module parameter allowing you to bind it to
arbitrary devices, or do something like what PCI does where you can
bind/unbind arbitrary drivers to devices using sysfs.

> +
> +static struct platform_driver fsi_master_fake_driver = {
> + .driver = {
> + .name = "fsi-master-fake",
> + .of_match_table = fsi_master_fake_match,
> + },
> + .probe = fsi_master_fake_probe,
> +};
> +
> +static int __init fsi_master_fake_init(void)
> +{
> + struct device_node *np;
> +
> + platform_driver_register(&fsi_master_fake_driver);
> +
> + for_each_compatible_node(np, NULL, "ibm,fsi-master-fake")
> + of_platform_device_create(np, NULL, NULL);

As a general note, please use for_each_matching_node in situations like
this. That way you can reuse your existing of_device_id table, and not
reproduce the string.

That said, this is not necessary. The platform driver has an
of_match_table, so presumes the parent bus registers children, and hence
they should already have platform devices.

Thanks,
Mark.