Jukka Rissanen

> 3\. If we think there is a viable more recent alternative to 20.04, we stop supporting it despite it not being EOL'd. Hmm, please do not remove the support...

Could some reasoning mentioned in the commit message that why we are adding minikube support?

> do you have any comments on the couple of discussions since your last requested change? I think we can have this change, I was just questioning the macro formatting,...

> By the way, don't you think the PR (#68809) already addresses #68793, but just hinders @EricNRS's use case. I have never seen the crash with shell, but if the...

> What about `OPTION3`, since you haven't mentioned it yet? I think the cons you mention i.e., `It limits the option of the backend to just SERIAL/RTT, and later, if...

> I've named it `PR_S`, where 'S' stands for the safe check of the sh pointer. No need to introduce new variants, the existing ones are enough.

Forgot to mention earlier, you should add `Fixes #69145` line to commit message body so that the issue gets linked to the commit (helps when generating release notes).

> So currently we are getting crashes with the default configured correctly, any suggestions? (I am still intently debugging of course) It would be really helpful if you could pin...

> conn going into `tcp_in` is null and it tries to access a member of `conn` which I believe causes the usage fault Hmm, but we access the data behind...

> Can confirm setting the default interface and using the socket option `SO_BINDTODEVICE` selecting between my 3 network interfaces appears to work reliably, thank you so much for your help...