[PATCH] net: eth-uclass: Do not return error when no MAC is found
On some i.MX8QXP MEK boards with no MAC address stored, the following hang is seen:
Error: ethernet@5b040000 address not set.
(Board hangs)
One way to avoid this issue is to select CONFIG_NET_RANDOM_ETHADDR, so that a random MAC is provided and boot proceeds, but the lack of MAC should not hang the entire system.
Fix the hang by removing the propagation of error when no MAC is found.
In this case there are already messages indicating that something is not correct:
Error: ethernet@5b040000 address not set.
Error: ethernet@5b040000 address 00:00:00:00:00:00 illegal value eth0: ethernet@5b040000 Error: ethernet@5b040000 address 00:00:00:00:00:00 illegal value
Signed-off-by: Fabio Estevam festevam@gmail.com --- net/eth-uclass.c | 1 - 1 file changed, 1 deletion(-)
diff --git a/net/eth-uclass.c b/net/eth-uclass.c index ed81cbd537..bc2b0aac00 100644 --- a/net/eth-uclass.c +++ b/net/eth-uclass.c @@ -554,7 +554,6 @@ static int eth_post_probe(struct udevice *dev) #else printf("\nError: %s address not set.\n", dev->name); - return -EINVAL; #endif }
Hi Joe,
On Thu, Jan 9, 2020 at 3:28 PM Fabio Estevam festevam@gmail.com wrote:
On some i.MX8QXP MEK boards with no MAC address stored, the following hang is seen:
Error: ethernet@5b040000 address not set.
(Board hangs)
One way to avoid this issue is to select CONFIG_NET_RANDOM_ETHADDR, so that a random MAC is provided and boot proceeds, but the lack of MAC should not hang the entire system.
Fix the hang by removing the propagation of error when no MAC is found.
In this case there are already messages indicating that something is not correct:
Error: ethernet@5b040000 address not set.
Error: ethernet@5b040000 address 00:00:00:00:00:00 illegal value eth0: ethernet@5b040000 Error: ethernet@5b040000 address 00:00:00:00:00:00 illegal value
Signed-off-by: Fabio Estevam festevam@gmail.com
Any comments?
On Thu, Jan 9, 2020 at 12:29 PM Fabio Estevam festevam@gmail.com wrote:
On some i.MX8QXP MEK boards with no MAC address stored, the following hang is seen:
Error: ethernet@5b040000 address not set.
(Board hangs)
One way to avoid this issue is to select CONFIG_NET_RANDOM_ETHADDR, so that a random MAC is provided and boot proceeds, but the lack of MAC should not hang the entire system.
Fix the hang by removing the propagation of error when no MAC is found.
In this case there are already messages indicating that something is not correct:
Error: ethernet@5b040000 address not set.
Error: ethernet@5b040000 address 00:00:00:00:00:00 illegal value eth0: ethernet@5b040000 Error: ethernet@5b040000 address 00:00:00:00:00:00 illegal value
Signed-off-by: Fabio Estevam festevam@gmail.com
Acked-by: Joe Hershberger joe.hershberger@ni.com
Hi Fabio,
On Thu, Jan 9, 2020 at 12:29 PM Fabio Estevam festevam@gmail.com wrote:
On some i.MX8QXP MEK boards with no MAC address stored, the following hang is seen:
Error: ethernet@5b040000 address not set.
(Board hangs)
One way to avoid this issue is to select CONFIG_NET_RANDOM_ETHADDR, so that a random MAC is provided and boot proceeds, but the lack of MAC should not hang the entire system.
Fix the hang by removing the propagation of error when no MAC is found.
In this case there are already messages indicating that something is not correct:
Error: ethernet@5b040000 address not set.
Error: ethernet@5b040000 address 00:00:00:00:00:00 illegal value eth0: ethernet@5b040000 Error: ethernet@5b040000 address 00:00:00:00:00:00 illegal value
Signed-off-by: Fabio Estevam festevam@gmail.com
I tried to apply this but it breaks the unit tests. I guess you didn't try "ut dm" in sandbox?
https://travis-ci.org/jhershbe/u-boot/builds/660395782
Please test those and update the tests to pass with your change.
Thanks, -Joe
Hi Joe,
On Tue, Mar 10, 2020 at 2:45 PM Joe Hershberger joe.hershberger@ni.com wrote:
I tried to apply this but it breaks the unit tests. I guess you didn't try "ut dm" in sandbox?
https://travis-ci.org/jhershbe/u-boot/builds/660395782
Please test those and update the tests to pass with your change.
Ok, I will revisit this patch after 2020.04.
I think we should probably let net_random_ethaddr() run unconditionally:
--- a/net/eth-uclass.c +++ b/net/eth-uclass.c @@ -547,15 +547,9 @@ static int eth_post_probe(struct udevice *dev) dev->name); } else if (is_zero_ethaddr(pdata->enetaddr) || !is_valid_ethaddr(pdata->enetaddr)) { -#ifdef CONFIG_NET_RANDOM_ETHADDR net_random_ethaddr(pdata->enetaddr); printf("\nWarning: %s (eth%d) using random MAC address - %pM\n", dev->name, dev->seq, pdata->enetaddr); -#else - printf("\nError: %s address not set.\n", - dev->name); - return -EINVAL; -#endif }
eth_write_hwaddr(dev);
Thanks
On Tue, Mar 10, 2020 at 1:34 PM Fabio Estevam festevam@gmail.com wrote:
Hi Joe,
On Tue, Mar 10, 2020 at 2:45 PM Joe Hershberger joe.hershberger@ni.com wrote:
I tried to apply this but it breaks the unit tests. I guess you didn't try "ut dm" in sandbox?
https://travis-ci.org/jhershbe/u-boot/builds/660395782
Please test those and update the tests to pass with your change.
Ok, I will revisit this patch after 2020.04.
I think we should probably let net_random_ethaddr() run unconditionally:
That may be a better solution... with the warning of course.
Cheers, -Joe
--- a/net/eth-uclass.c +++ b/net/eth-uclass.c @@ -547,15 +547,9 @@ static int eth_post_probe(struct udevice *dev) dev->name); } else if (is_zero_ethaddr(pdata->enetaddr) || !is_valid_ethaddr(pdata->enetaddr)) { -#ifdef CONFIG_NET_RANDOM_ETHADDR net_random_ethaddr(pdata->enetaddr); printf("\nWarning: %s (eth%d) using random MAC address - %pM\n", dev->name, dev->seq, pdata->enetaddr); -#else
printf("\nError: %s address not set.\n",dev->name);return -EINVAL;-#endif }
eth_write_hwaddr(dev);Thanks
Hi Joe,
On Wed, Mar 11, 2020 at 2:15 PM Joe Hershberger joe.hershberger@ni.com wrote:
That may be a better solution... with the warning of course.
Thanks for the feedback.
I will submit this solution after 2020.04 is out.
Thanks
participants (2)
-
Fabio Estevam -
Joe Hershberger