tests: exercise unix `connectat(2)` with a socket peer descriptor

[inquire_JohnEricson.me]

Summary

Cover the new fd-direct connect path: stream connect and data passing,
the peer address reported by getpeername(2), datagram to an unbound
peer, the EINVAL/ENOTSOCK/EPROTOTYPE/ECONNREFUSED error matrix,
and the Capsicum token semantics — a descriptor limited to
CAP_CONNECTAT can be connected to but not listened on, accepted from,
or read, and one lacking CAP_CONNECTAT cannot be a connect target.

Stream listeners are always bound: uipc_listen() refuses unbound
sockets with EDESTADDRREQ, so an unbound fd-direct listener is not
reachable even with this feature.

Signed-off-by: John Ericson <[e-mail]>
Assisted-by: Claude Code (Claude Opus 4.8 and Fable 5)

[markj]

These tests don't exercise fdescfs lookups, nor handling of O_PATH sockets. Were you planning to add something for that? It should be fairly easy to create fdescfs mounts from within the tests themselves, tests/sys/kern/jail_lookup_root.c has some examples of this.

tests/sys/kern/unix_connectat.c

98

Not hugely important, but I think it's good form to check for errors from close(), doing so can catch test bugs (and, rarely, kernel bugs).

[inquire_JohnEricson.me]

These tests don't exercise fdescfs lookups, nor handling of O_PATH sockets. Were you planning to add something for that?

There is some leniency in here so that the tests pass with or without fdescfs, but I didn't know it was OK to mount it within the tests. I'll go do that instead; that's definitely better! And I'll make sure we get O_PATH tested too.

It should be fairly easy to create fdescfs mounts from within the tests themselves, tests/sys/kern/jail_lookup_root.c has some examples of this.

Thanks, that will be useful to make sure I get the idiom correct.

tests/sys/kern/unix_connectat.c

98

Oh oops I admit I didn't notice that --- I absolutely agree. Will fix.

[inquire_JohnEricson.me]

Test all the fdescfs combinations explicitly