I want to get some perspectives from embedded engineers with considerably more experience (than me) looking at kernel boot logs. I am relatively new to embedded Linux. My i.MX53 board's boot environment is in the context of kernel via tftp and root over nfs.
When I see:
IP-Config: Failed to open eth0
I wonder, what are the possible conditions for this error to be reported. Perhaps I should be reading linux source code to be sure. But that could take awhile! cscope here I come! :)
Is it possible for this error message to appear when network conditions cause a problem for the network driver (e.g. not a problem associated to the ethernet driver failing due to wrong driver or bug in driver).
Could any previously failed dhcp-options negotiation with the DHCP server, at an earlier stage of the boot process, cause this error later after the kernel has booted?
This is one of those "this was working fine last week" problems. I think nothing was changed; certainly checking for change.
Today, having intermittent success in getting completely booted. Kernel transfers OK over network. I am seeing about 1 out of 8 attempts booting all the way to init. I see this error after kernel loads and just before the attempt to get the root fs. I would not be persisting with the same configuration if I was not seeing an occasional successful boot of root fs over nfs and the command prompt of my Linux distribution (hosted on the NFS box).
BTW, my lab's network was reconfigured and it is considerably more complex than it was before... so I am suspicious of the network. (new routers and managed switches, new subnets and new fiber links; hosts are on same logical network but way WAY different physical network. lol!; hence more complex and more opportunity for failures, me thinks!). That's why I ask if an external problem (network environment) could generate the "Failed to open eth0" error.
For comparison, on a successful boot of another i.MX53 board, I see the following in the same spot in the logs:
Thx in advance for any and all comments!