Re: [PATCH net-next] selftests: net: Skip test if IPv6 is not configured

[Willem de Bruijn <willemdebruijn.kernel@gmail.com>]

Breno Leitao wrote:
> Hello Jakub,
> 
> On Thu, Jul 24, 2025 at 06:24:27PM -0700, Jakub Kicinski wrote:
> > On Wed, 23 Jul 2025 10:35:06 -0700 Breno Leitao wrote:
> > > Extend the `check_for_dependencies()` function in `lib_netcons.sh` to check
> > > whether IPv6 is enabled by verifying the existence of
> > > `/proc/net/if_inet6`. Having IPv6 is a now a dependency of netconsole
> > > tests. If the file does not exist, the script will skip the test with an
> > > appropriate message suggesting to verify if `CONFIG_IPV6` is enabled.
> > > 
> > > This prevents the test to misbehave if IPv6 is not configured.
> > 
> > IDK. I think this is related to some of the recent patches?
> 
> Yes, commit 3dc6c76391cbe (“selftests: net: Add IPv6 support to
> netconsole basic tests”) introduced IPv6 support to the netconsole basic
> tests.
> 
> Because the NIPA config enables IPv6, the tests pass in that
> environment. However, if the tests are run somewhere without IPv6
> support such as in a test I was doing regarding another patch, they will
> fail, when it should be skipped.
> 
> > The context would be helpful in the commit message.
> 
> Apologies for not including more context in the commit message.
> 
> > Otherwise, as networking people, I think we are obligated 
> > to respond with hostility to "IPv6 may not be enabled"..
> 
> As for handling systems without IPv6, if IPv6 isn’t available, the
> intention is for the test to be skipped. That’s exactly what this patch
> addresses.

I think there is some consensus that these environments should no
longer exist in 2025. And test failure is the best way to accomplish
that.

Less opinionated: the tests implicitly depends on the config files
in the test directory. Do we have to start making the robust against
situations where CONFIGs in that file are missing?

> I did consider making the test adaptable so it would just run with
> whichever protocol (IPv4 or IPv6) is present, but rejected that
> approach. Allowing the test to “pass” in such cases doesn’t really
> demonstrate meaningful coverage, since the test isn’t actually being
> exercised as intended.
> 
> In short, it seems more appropriate to skip the test entirely if all
> conditions aren’t met, so, you know that your .config needs adjustment.
> 
> Thanks for your review,
> --breno