[PATCH 1/1] asix: use ramdom hw addr if the one read is not valid

Sergei Shtylyov [off-list ref] writes:

On 11/21/2012 01:22 PM, Jean-Christophe PLAGNIOL-VILLARD wrote:

quoted

Signed-off-by: Jean-Christophe PLAGNIOL-VILLARD <redacted>
Cc: linux-usb at vger.kernel.org
Cc: netdev at vger.kernel.org
---
 drivers/net/usb/asix_devices.c |   24 +++++++++++++++++++++---
 1 file changed, 21 insertions(+), 3 deletions(-)

quoted

diff --git a/drivers/net/usb/asix_devices.c b/drivers/net/usb/asix_devices.c
index 33ab824..7ebec5b 100644
--- a/drivers/net/usb/asix_devices.c
+++ b/drivers/net/usb/asix_devices.c

@@ -225,7 +225,13 @@ static int ax88172_bind(struct usbnet *dev, struct usb_interface *intf)
 			   ret);
 		goto out;
 	}
-	memcpy(dev->net->dev_addr, buf, ETH_ALEN);
+
+	if (is_valid_ether_addr(buf)) {
+		memcpy(dev->net->dev_addr, buf, ETH_ALEN);
+	} else {
+		netdev_info(dev->net, "invalid hw address, using random\n");
+		eth_hw_addr_random(dev->net);
+	}
 
 	/* Initialize MII structure */
 	dev->mii.dev = dev->net;

[..]

   Repeated thrice, this asks to be put into subroutine...

Yes.  Looking at the driver, this probably goes for most of the three
_bind() functions.  There is a lot of common code there.

But more important wrt the eth_hw_addr_random() change: Does this
actually work with real devices?  The driver implements a
asix_set_mac_address() which writes the address back to the device when
you change it.  I assume there is a reason for doing that.  Why don't
you do it here?


Bj?rn