Hi Joe,
On 1 October 2018 at 16:01, Joe Hershberger joe.hershberger@ni.com wrote:
On Mon, Oct 1, 2018 at 4:27 PM, Simon Glass sjg@chromium.org wrote:
Hi Joe,
On 1 October 2018 at 15:14, Joe Hershberger joe.hershberger@ni.com wrote:
Hi Simon,
On Mon, Oct 1, 2018 at 12:32 PM, Simon Glass sjg@chromium.org wrote:
Hi Joe,
When I start sandbox now I see these errors every time. It is possible to fix this?
$ u-boot -D ... Warning: host_lo MAC addresses don't match: Address in ROM is ea:06:97:67:f4:b6 Address in environment is 00:00:11:22:33:44
Warning: host_eth0 MAC addresses don't match: Address in ROM is ea:06:97:67:f4:b6 Address in environment is 00:00:11:22:33:45
Warning: host_enp0s31f6 using MAC address from ROM
Warning: host_eth6 MAC addresses don't match: Address in ROM is ea:06:97:67:f4:b6 Address in environment is 00:00:11:22:33:46
Warning: host_docker_gwbridge using MAC address from ROM
Warning: host_docker0 MAC addresses don't match: Address in ROM is ca:07:52:28:25:66 Address in environment is 00:00:11:22:33:47
Warning: host_veth1f6bcd2 using MAC address from ROM
These are warnings based on commit b96ced9cdb6 (sandbox: eth-raw: Make random MAC addresses available to eth-raw). Because we don't know the number of interfaces that exist on the given host when we write the sandbox.dts, we override the MAC addresses with random ones. In the normal case (real hardware), we would warn, since the hardware should know its interfaces.
Do you have a proposed different approach? I don't like the idea of adding sandbox checks in the core code that prints this warning, and it also doesn't seem like a great idea to add a hook that doesn't warn just for sandbox.
I'm happy to work on this, but I didn't address it at the time because I didn't see a clear way to do it that didn't degrade the code in order to remove warnings in sandbox.
Can we make it so that the addresses do actually match in sandbox?
The issue is that the eth(N)?addr lives in the default env and the interfaces are discovered by drivers/net/sandbox-raw-bus.c. Currently we make sure we don't "run out" of addresses defined in the env by using a randomly generated one instead. Unfortunately, the env is shared with all modes of sandbox, so while the addresses are not shared when running sandbox.dts, they are required when running test.dts.
Thoughts?
How about a device-tree property that is only present in the test DT, to control the behaviour?
Regards, Simon