Switch net.inet6.ip6.use_stableaddr to on by default

[madpilot]

Summary

As suggested by pouria on the current mailing list, I'm proposing
apatch to switch the ddefault for RFC 7217 stableaddress support
to on.

I also wrote an UPDATING note, I hope the wording is correct and
clear enough.

[pouria]

LGTM

[glebius]

No objections from me, I rather agree. But I am not enough expert in IPv6 to make a strong judgement on defaults change.

[zlei]

This is a behavior change. Please also add a meta Relnotes: yes to the commit message.

[madpilot]

The patch I submitted was missing a part. This is the updated and correct version.

@zlei Thanks for the suggestion, I'll add that to the commit message.

[zlei]

sys/netinet6/in6_proto.c

314 ↗

(On Diff #171404)

This change is not necessary but harmless. I'd suggest you do not change it, to avoid potential confusion.

The val argument of SYSCTL_BOOL function only when the argument ptr is NULL. If not then that argument does nothing.

See the comment above SYSCTL_BOOL in sys/sys/sysctl.h,

/* Oid for a bool.  If ptr is NULL, val is returned. */
#define SYSCTL_NULL_BOOL_PTR ((bool *)NULL)
#define SYSCTL_BOOL(parent, nbr, name, access, ptr, val, descr) \
        SYSCTL_OID(parent, nbr, name,                           \
            CTLTYPE_U8 | CTLFLAG_MPSAFE | (access),             \
            ptr, val, sysctl_handle_bool, "CU", descr);         \
        CTASSERT(((access) & CTLTYPE) == 0 &&                   \
            sizeof(bool) == sizeof(*(ptr)))

[madpilot]

sys/netinet6/in6_proto.c

314 ↗

(On Diff #171404)

This change is not necessary but harmless. I'd suggest you do not change it, to avoid potential confusion.

The val argument of SYSCTL_BOOL function only when the argument ptr is NULL. If not then that argument does nothing.

Thanks for the explanation, I was, in fact, confused by this duplication.

Fixing the review shortly.

[madpilot]

Removed unneeded change indicated by @zlei