Hi,
This is an alternative patch - but not on git trunk - ist againts the last tag i are allowed to use.
I don't know if the ethaddr should be copied in register_preset_mac_address or after calling.
below is only tested on am335x.
-------SNIP------
Fix missing eheraddr setting. If some modul register an MAC for an network device
like cpsw, the etheraddr of the device was never set.
This is a workaround only. Don't put upstream.
--- barebox-mlo-2014.09.0/net/eth.c 2014-09-04 15:28:52.000000000 +0200
+++ barebox-mlo-2014.09.0/net/eth.c 2014-09-23 17:18:21.592884794 +0200
@@ -358,9 +357,8 @@
list_add_tail(&edev->list, &netdev_list);
ret = eth_get_registered_ethaddr(edev, ethaddr);
- if (!ret)
+ if (!ret && is_valid_ether_addr(ethaddr))
found = 1;
-
if (!found) {
ret = edev->get_ethaddr(edev, ethaddr);
if (!ret)
@@ -368,7 +366,11 @@
}
if (found)
+ {
register_preset_mac_address(edev, ethaddr);
+ /* fix missing mac in edev */
+ memcpy(edev->ethaddr,ethaddr,6);
+ }
if (IS_ENABLED(CONFIG_OFDEVICE) && edev->parent &&
edev->parent->device_node)
-------SNIP------
-----Ursprüngliche Nachricht-----
Von: barebox [mailto:barebox-bounces@xxxxxxxxxxxxxxxxxxx] Im Auftrag von Wadim Egorov
Gesendet: Mittwoch, 10. Juni 2015 08:44
An: Sascha Hauer
Cc: barebox@xxxxxxxxxxxxxxxxxxx
Betreff: Re: [PATCH] net: Set the actual ethaddr in register_preset_mac_address()
Hello Sascha,
On 10.06.2015 06:32, Sascha Hauer wrote:
> Hi Wadim,
>
> On Tue, Jun 09, 2015 at 09:04:25AM +0200, Wadim Egorov wrote:
>> Set the ethaddr for the current edev.
>>
>> Signed-off-by: Wadim Egorov <w.egorov@xxxxxxxxx>
>> ---
>> net/eth.c | 1 +
>> 1 file changed, 1 insertion(+)
>>
>> diff --git a/net/eth.c b/net/eth.c
>> index 89bddba..03e0a2e 100644
>> --- a/net/eth.c
>> +++ b/net/eth.c
>> @@ -49,6 +49,7 @@ static void register_preset_mac_address(struct eth_device *edev, const char *eth
>> ethaddr_to_string(ethaddr, ethaddr_str);
>>
>> if (is_valid_ether_addr(ethaddr)) {
>> + memcpy(edev->ethaddr, ethaddr, 6);
>> dev_info(&edev->dev, "got preset MAC address: %s\n", ethaddr_str);
>> dev_set_param(&edev->dev, "ethaddr", ethaddr_str);
>> }
> In which case is this necessary? Normally a dev_set_param on "ethaddr"
> should already set edev->ethaddr, there should be no need to copy this
> manually.
>
> Sascha
when booting from ethernet on the AM335x, net_new() (called in net_udp_new) will check if ethaddr is valid. This check fails, because ethaddr is not set at this moment and a random MAC will be used.
Wadim
>
_______________________________________________
barebox mailing list
barebox@xxxxxxxxxxxxxxxxxxx
http://lists.infradead.org/mailman/listinfo/barebox
_______________________________________________
barebox mailing list
barebox@xxxxxxxxxxxxxxxxxxx
http://lists.infradead.org/mailman/listinfo/barebox
