If you don't break up the first line, the KASSERT will still fit on two lines.

Shouldn't it be 0 <= %d < %d?

And here as well: 0 <= ..

While changing the code, can we start the comment upper case and end in a '.' as well?

One the other hand, these checks violate RFC 4861 I think? (Haven't checked if there is a specified behaviour for IPv4)

A host MUST NOT consider a redirect invalid just because the Target
Address of the redirect is not covered under one of the link's
prefixes.  Part of the semantics of the Redirect message is that the
Target Address is on-link.

Given we are currently not prepared to add an interface route to my understanding,
can we add a comment

/* XXX-BZ RFC 4861 8.1 does not mandate this check. */

and fix it later (unless you do want to fix it as well, but I fear that will require more code changes which are better done in a follow-up?

'.' at the end of comment.

Hmm... I wonder if that should be done for other errors as well?

I keep wondering if I'd swap the two last checks with the first (currently) if (rt != NULL) block and do the flags checks before comparing addresses?

If rt is NULL we don't do anything up to the end and there we (a) call RTFREE_LOCKED() and will have a NULL pointer deref and then (b) return (0);
If we add a check for rt == NULL here and handle it, then we can drop all the rt != NULL on all the conditions further down.

The comment could be re-wrapped to three lines.

Indentation looks too much here.

And as mentioned for the declaration I think the argument order should change?

Please don't assign values in variable declaration. Set it after NET_EPOCH_ASSERT() maybe.

error is 0 here?

Comments start with upper case and end in punctuation.

This is the same "on-link" vs. prefix issue as in the previous function, right? We should add another comment here?

'.' at the end.

'.' at the end.

I think time_second + "expire_sec"; would read better. I am still confused if this is "expire_sec" or actually "lifetime" as it seems to be the lifetime of the redirect whereas rmx_expire then indeed is the "second in which to expire this entry".

, "it " iterates .. in "the " particular ..

Could be one sentence with an 'and'. also "returns a" not "returned" maybe?

Iterates over a routing table specified by @fibnum and @family and
deletes elements marked by @filter_f.

(replacing rtable, fixinf af argument name, removing white space on 2nd line).

remove "the" or add "function" at the end.

Then it could almost be a bool instead of int?

u_int fibnum?

'.' at the end.

Is there a particular reason to break up the sockaddr pointers by putting the int flags in the middle between them?

Normally this would be a __FBSDID(..); in the code after the cdefs include and not the comment.

extra space to remove .. but the 2nd sentence can also be one line easily.

No assignments in variable declarations.

Is this is debug printf which should go? Otherwise at least at a %s: func please.

Here the next_callout = (time_t *)arg; could happen; which means we'd also not do it for previous returns and save some work.

Please move the assignments down before first use.

Isn't this a problem after roughly 24 days?

Same ~24 days issue as above?