Re: [PATCH 03/14] net: ks8851: Pass device pointer into ks8851_init_mac()

On Tue, Mar 24, 2020 at 02:06:22AM +0100, Andrew Lunn wrote:

On Tue, Mar 24, 2020 at 12:42:52AM +0100, Marek Vasut wrote:

quoted

Since the driver probe function already has a struct device *dev pointer,
pass it as a parameter to ks8851_init_mac() to avoid fishing it out via
ks->spidev. This is the only reference to spidev in the function, so get
rid of it. This is done in preparation for unifying the KS8851 SPI and
parallel drivers.

[...]

quoted

-static void ks8851_init_mac(struct ks8851_net *ks)
+static void ks8851_init_mac(struct ks8851_net *ks, struct device *ddev)
 {
 	struct net_device *dev = ks->netdev;
 	const u8 *mac_addr;
 
-	mac_addr = of_get_mac_address(ks->spidev->dev.of_node);
+	mac_addr = of_get_mac_address(ddev->of_node);

The name ddev is a bit odd. Looking at the code, i see why. dev is
normally a struct net_device, which this function already has.

You could avoid this oddness by directly passing of_node.

Actually after adding the invocation of of_get_mac_address() with
commit 566bd54b067d ("net: ks8851: Support DT-provided MAC address")
I've had regrets that I should have used device_get_mac_address()
instead since it's platform-agnostic, hence would work with ACPI
as well as DT-based systems.

device_get_mac_address() needs a struct device, so I'd prefer
using that instead of passing an of_node.

I agree that "ddev" is somewhat odd.  Some drivers name it "device"
or "pdev" (which however collides with the naming of platform_devices).
Another idea would be to move the handy ndev_to_dev() static inline
from apm/xgene/xgene_enet_main.h to include/linux/netdevice.h and
use that with "struct net_device *dev", which we already have in
ks8851_init_mac().

Thanks,

Lukas