This adds a mode that makes resolved automatically downgrade from DNSSEC
support to classic non-DNSSEC resolving if the configured DNS server is
not capable of DNSSEC. Enabling this mode increases compatibility with
crappy network equipment, but of course opens up the system to
downgrading attacks.
The new mode can be enabled by setting DNSSEC=downgrade-ok in
resolved.conf. DNSSEC=yes otoh remains a "strict" mode, where DNS
resolving rather fails then allow downgrading.
Downgrading is done:
- when the server does not support EDNS0+DO
- or when the server supports it but does not augment returned RRs with
RRSIGs. The latter is detected when requesting DS or SOA RRs for the
root domain (which is necessary to do proofs for unsigned data)
The setting controls which kind of DNSSEC validation is done: none at
all, trusting the AD bit, or client-side validation.
For now, no validation is implemented, hence the setting doesn't do much
yet, except of toggling the CD bit in the generated messages if full
client-side validation is requested.
This is often needed for proper DNSSEC support, and even to handle AAAA records
without falling back to TCP.
If the path between the client and server is fully compliant, this should always
work, however, that is not the case, and overlarge packets will get mysteriously
lost in some cases.
For that reason, we use a similar fallback mechanism as we do for palin EDNS0,
EDNS0+DO, etc.:
The large UDP size feature is different from the other supported feature, as we
cannot simply verify that it works based on receiving a reply (as the server
will usually send us much smaller packets than what we claim to support, so
simply receiving a reply does not mean much).
For that reason, we keep track of the largest UDP packet we ever received, as this
is the smallest known good size (defaulting to the standard 512 bytes). If
announcing the default large size of 4096 fails (in the same way as the other
features), we fall back to the known good size. The same logic of retrying after a
grace-period applies.
This indicates that we can handle DNSSEC records (per RFC3225), even if
all we do is silently drop them. This feature requires EDNS0 support.
As we do not yet support larger UDP packets, this feature increases the
risk of getting truncated packets.
Similarly to how we fall back to plain UDP if EDNS0 fails, we will fall
back to plain EDNS0 if EDNS0+DO fails (with the same logic of remembering
success and retrying after a grace period after failure).
This is a minimal implementation of RFC6891. Only default values
are used, so in reality this will be a noop.
EDNS0 support is dependent on the current server's feature level,
so appending the OPT pseudo RR is done when the packet is emitted,
rather than when it is assembled. To handle different feature
levels on retransmission, we strip off the OPT RR again after
sending the packet.
Similarly, to how we fall back to TCP if UDP fails, we fall back
to plain UDP if EDNS0 fails (but if EDNS0 ever succeeded we never
fall back again, and after a timeout we will retry EDNS0).
Previously, we would only degrade on packet loss, but when adding EDNS0 support,
we also have to handle the case where the server replies with an explicit error.
This is inspired by the logic in BIND [0], follow-up patches
will implement the reset of that scheme.
If we get a server error back, or if after several attempts we don't
get a reply at all, we switch from UDP to TCP for the given
server for the current and all subsequent requests. However, if
we ever successfully received a reply over UDP, we never fall
back to TCP, and once a grace-period has passed, we try to upgrade
again to using UDP. The grace-period starts off at five minutes
after the current feature level was verified and then grows
exponentially to six hours. This is to mitigate problems due
to temporary lack of network connectivity, but at the same time
avoid flooding the network with retries when the feature attempted
feature level genuinely does not work.
Note that UDP is likely much more commonly supported than TCP,
but depending on the path between the client and the server, we
may have more luck with TCP in case something is wrong. We really
do prefer UDP though, as that is much more lightweight, that is
why TCP is only the last resort.
[0]: <https://kb.isc.org/article/AA-01219/0/Refinements-to-EDNS-fallback-behavior-can-cause-different-outcomes-in-Recursive-Servers.html>
This copies concepts we introduced for the DnsSearchDomain stuff, and
reworks the operations on lists of dns servers to be reusable and
generic for use both with the Link and the Manager object.
Previously, we'd keep adding new dns servers we discover to the end of
our linked list of servers. When we encountered a pre-existing server,
we'd just leave it where it was. In essence that meant that old servers
ended up at the front, and new servers at the end, but not in an order
that would reflect the configuration.
With this change we ensure that every pre-existing server we want to add
again we move to the back of the linked list, so that the order is
stable and in sync with the requested configuration.
Previously, there was a chance of memory corruption, because when
switching to the next DNS server we didn't care whether they linked list
of DNS servers was still valid.
Clean up lifecycle of the dns server logic:
- When a DnsServer object is still in the linked list of DnsServers for
a link or the manager, indicate so with a "linked" boolean field, and
never follow the linked list if that boolean is not set.
- When picking a DnsServer to use for a link ot manager, always
explicitly take a reference.
This also rearranges some logic, to make the tracking of dns servers by
link and globally more alike.
Let's use the same parser when parsing dns server information from
/etc/resolv.conf and our native configuration file.
Also, move all code that manages lists of dns servers to a single place.
resolved-dns-server.c
Rather than fixing this to 5s for unicast DNS and 1s for LLMNR, start
at a tenth of those values and increase exponentially until the old
values are reached. For LLMNR the recommended timeout for IEEE802
networks (which basically means all of the ones we care about) is 100ms,
so that should be uncontroversial. For unicast DNS I have found no
recommended value. However, it seems vastly more likely that hitting a
500ms timeout is casued by a packet loss, rather than the RTT genuinely
being greater than 500ms, so taking this as a startnig value seems
reasonable to me.
In the common case this greatly reduces the latency due to normal packet
loss. Moreover, once we get support for probing for features, this means
that we can send more packets before degrading the feature level whilst
still allowing us to settle on the correct feature level in a reasonable
timeframe.
The timeouts are tracked per server (or per scope for the multicast
protocols), and once a server (or scope) receives a successfull package
the timeout is reset. We also track the largest RTT for the given
server/scope, and always start our timouts at twice the largest
observed RTT.
This patch removes includes that are not used. The removals were found with
include-what-you-use which checks if any of the symbols from a header is
in use.
It is redundant to store 'hash' and 'compare' function pointers in
struct Hashmap separately. The functions always comprise a pair.
Store a single pointer to struct hash_ops instead.
systemd keeps hundreds of hashmaps, so this saves a little bit of
memory.
We now maintain two lists of DNS servers: system servers and fallback
servers.
system servers are used in combination with any per-link servers.
fallback servers are only used if there are no system servers or
per-link servers configured.
The system server list is supposed to be populated from a foreign tool's
/etc/resolv.conf (not implemented yet).
Also adds a configuration switch for LLMNR, that allows configuring
whether LLMNR shall be used simply for resolving or also for responding.
networkd will expose both statically configured DNS servers and servers
receieved over DHCP in sd_network_get_dns(), so no need to keep
the distinction in resolved.
Let's settle on a single type for all address family values, even if
UNIX is very inconsitent on the precise type otherwise. Given that
socket() is the primary entrypoint for the sockets API, and that uses
"int", and "int" is relatively simple and generic, we settle on "int"
for this.
Let's turn resolved into a something truly useful: a fully asynchronous
DNS stub resolver that subscribes to network changes.
(More to come: caching, LLMNR, mDNS/DNS-SD, DNSSEC, IDN, NSS module)
