libp2p: add prefer option

6 files changed, 121 insertions, 8 deletions

diff --git a/src/config/mod.rs b/src/config/mod.rs
index 2dfefcf..c45f1d0 100644
--- a/src/config/mod.rs
+++ b/src/config/mod.rs
@@ -296,6 +296,13 @@ pub(crate) const OPTIONS: &[OptionSpec] = &[
         "With the libp2p transport: address ranges libp2p leaves out, e.g. 10.244.0.0/16 and 42::/16 (VPNs): addresses there are neither published (but announced ones) nor offered for hole punching nor dialled, and connections there are refused, whoever dials.",
     ),
     option(
+        "libp2p-prefer",
+        Kind::List,
+        "RANGE,[RANGE...]",
+        None,
+        "With the libp2p transport: address ranges that matter most, best first, e.g. the LAN's 10.23.0.0/16: of the many addresses a node listening everywhere has, those there are published first, and of a peer's, kept first and dialled with the others. Without, public addresses come first.",
+    ),
+    option(
         "libp2p-relay",
         Kind::Flag,
         "BOOL",
@@ -507,6 +514,9 @@ pub(crate) struct Settings {
     pub(crate) libp2p_ignore: Vec<String>,
     #[serde(default)]
     #[cfg_attr(not(feature = "libp2p"), allow(dead_code))]
+    pub(crate) libp2p_prefer: Vec<String>,
+    #[serde(default)]
+    #[cfg_attr(not(feature = "libp2p"), allow(dead_code))]
     pub(crate) libp2p_public: bool,
     #[cfg_attr(not(feature = "libp2p"), allow(dead_code))]
     pub(crate) libp2p_key_file: Option<PathBuf>,
diff --git a/src/simtest/identity.rs b/src/simtest/identity.rs
index 6290ba7..5ed2052 100644
--- a/src/simtest/identity.rs
+++ b/src/simtest/identity.rs
@@ -224,6 +224,7 @@ async fn nodes_over_libp2p() {
                 mdns: false,
                 upnp: false,
                 ignore: Vec::new(),
+                prefer: Vec::new(),
                 public: false,
                 memory: true,
             })
diff --git a/src/transport/libp2p/book.rs b/src/transport/libp2p/book.rs
index ceb9d27..b495719 100644
--- a/src/transport/libp2p/book.rs
+++ b/src/transport/libp2p/book.rs
@@ -48,6 +48,8 @@ pub(super) struct Book {
     pub(super) announced: Vec<Multiaddr>,
     /// `Libp2pConfig::ignore`: not published (but announced), not dialled.
     pub(super) ignored: Vec<IpRange>,
+    /// `Libp2pConfig::prefer`: addresses there first (`preference`).
+    pub(super) preferred: Vec<IpRange>,
     /// `Libp2pConfig::relay`: this node relays, and says so.
     pub(super) serving: bool,
     /// Relays named to pick (configured, listed), by peer: picked first.
@@ -265,6 +267,7 @@ impl Book {
             }
         }
         direct.retain(|address| !first.contains(address));
+        direct.sort_by_key(|address| self.preferred_range(address));
         first.extend(direct);
         first.retain(|address| !ignored(&self.ignored, address));
         first
@@ -469,9 +472,11 @@ impl Book {
 
     /// `peer`, connected, identified itself at `listen` (`published`): its direct
     /// addresses are those from now on, while connected (`remember`); those through
-    /// relays and by name stay.
+    /// relays and by name stay. Kept in order of `preference`: of a peer with many,
+    /// the ones that matter stay (`ADDRESSES_PER_PEER`).
     pub(super) fn identified(&mut self, peer: PeerId, listen: Vec<Multiaddr>) {
-        let listen: Vec<Multiaddr> = listen.into_iter().map(without_peer).collect();
+        let mut listen: Vec<Multiaddr> = listen.into_iter().map(without_peer).collect();
+        listen.sort_by_key(|address| self.preference(address));
         if let Some(known) = self.addresses.get_mut(&peer) {
             known.retain(|address| !is_ip(address) || listen.contains(address));
         }
@@ -483,6 +488,63 @@ impl Book {
     }
 }
 
+impl Book {
+    /// Where a peer's `address` comes, first first: in a range `libp2p-prefer` names
+    /// (by the range's place), then public, then through a relay or by name, then a
+    /// private one (another LAN, a container bridge).
+    fn preference(&self, address: &Multiaddr) -> (u8, usize) {
+        let range = self.preferred_range(address);
+        if range != usize::MAX {
+            return (0, range);
+        }
+        match ip_of(address).filter(|_| is_ip(address)) {
+            Some(ip) if is_public(ip) => (1, 0),
+            Some(_) => (3, 0),
+            None => (2, 0),
+        }
+    }
+
+    /// The place of the first range of `libp2p-prefer` that has `address`; for an
+    /// address none has, `usize::MAX`: after them.
+    fn preferred_range(&self, address: &Multiaddr) -> usize {
+        ip_of(address)
+            .filter(|_| is_ip(address))
+            .and_then(|ip| self.preferred.iter().position(|range| range.contains(ip)))
+            .unwrap_or(usize::MAX)
+    }
+}
+
+/// The IP address `address` starts with, if any.
+fn ip_of(address: &Multiaddr) -> Option<std::net::IpAddr> {
+    match address.iter().next() {
+        Some(Protocol::Ip4(ip)) => Some(ip.into()),
+        Some(Protocol::Ip6(ip)) => Some(ip.into()),
+        _ => None,
+    }
+}
+
+/// Whether `ip` can be reached from the internet, as far as its range tells.
+fn is_public(ip: std::net::IpAddr) -> bool {
+    use std::net::IpAddr;
+    match ip {
+        IpAddr::V4(ip) => {
+            // 100.64.0.0/10: carrier-grade NAT.
+            let shared = ip.octets()[0] == 100 && (ip.octets()[1] & 0xc0) == 64;
+            !(ip.is_private()
+                || ip.is_loopback()
+                || ip.is_link_local()
+                || ip.is_unspecified()
+                || shared)
+        }
+        IpAddr::V6(ip) => {
+            !(ip.is_loopback()
+                || ip.is_unspecified()
+                || ip.is_unique_local()
+                || ip.is_unicast_link_local())
+        }
+    }
+}
+
 /// Whether `address` is a direct one at an IP address (not through a relay, nor a name).
 pub(super) fn is_ip(address: &Multiaddr) -> bool {
     matches!(
diff --git a/src/transport/libp2p/config.rs b/src/transport/libp2p/config.rs
index 719f2e8..8b172ba 100644
--- a/src/transport/libp2p/config.rs
+++ b/src/transport/libp2p/config.rs
@@ -23,6 +23,8 @@ pub(crate) struct Libp2pConfig {
     pub(crate) upnp: bool,
     /// Address ranges neither published nor dialled, nor connected over (`Ignoring`).
     pub(crate) ignore: Vec<IpRange>,
+    /// Address ranges that matter most, best first (`Book::preference`).
+    pub(crate) prefer: Vec<IpRange>,
     /// Use public infrastructure: the IPFS DHT, public relays.
     pub(crate) public: bool,
     /// libp2p's in-memory transport instead of TCP and QUIC: for tests.
@@ -41,11 +43,11 @@ pub(super) fn parse_addresses(option: &str, addresses: &[String]) -> Result<Vec<
         .collect()
 }
 
-/// The ranges of `libp2p-ignore`.
-pub(super) fn parse_ranges(ranges: &[String]) -> Result<Vec<IpRange>> {
+/// The ranges of option `option`.
+pub(super) fn parse_ranges(option: &str, ranges: &[String]) -> Result<Vec<IpRange>> {
     ranges
         .iter()
-        .map(|range| range.parse().context("libp2p-ignore"))
+        .map(|range| range.parse().with_context(|| option.to_owned()))
         .collect()
 }
 
@@ -57,7 +59,8 @@ pub(crate) fn check(settings: &crate::config::Settings) -> Result<()> {
     parse_addresses("libp2p-listen", &settings.libp2p_listen)?;
     parse_addresses("libp2p-announce", &settings.libp2p_announce)?;
     parse_addresses("libp2p-relays", &settings.libp2p_relays)?;
-    parse_ranges(&settings.libp2p_ignore)?;
+    parse_ranges("libp2p-ignore", &settings.libp2p_ignore)?;
+    parse_ranges("libp2p-prefer", &settings.libp2p_prefer)?;
     if settings.relay_only {
         if settings.libp2p_listen.is_empty() {
             bail!("relay-only needs libp2p-listen: where nodes reach the relay");
@@ -88,7 +91,8 @@ impl Libp2pConfig {
             continuum: !relay_only,
             mdns: settings.libp2p_mdns,
             upnp: settings.libp2p_upnp,
-            ignore: parse_ranges(&settings.libp2p_ignore)?,
+            ignore: parse_ranges("libp2p-ignore", &settings.libp2p_ignore)?,
+            prefer: parse_ranges("libp2p-prefer", &settings.libp2p_prefer)?,
             public: settings.libp2p_public,
             memory: false,
         })
diff --git a/src/transport/libp2p/mod.rs b/src/transport/libp2p/mod.rs
index f1006e3..0e15a09 100644
--- a/src/transport/libp2p/mod.rs
+++ b/src/transport/libp2p/mod.rs
@@ -45,7 +45,8 @@
 //!
 //! With `--libp2p-ignore`, address ranges are left out (VPNs): not published, not told
 //! peers by identify (`Identifying`), not offered for hole punching, not dialled, and
-//! connections there refused (`Ignoring`).
+//! connections there refused (`Ignoring`). With `--libp2p-prefer`, addresses in the
+//! ranges named come first, where only so many are kept (`Book::preference`).
 //!
 //! With `--libp2p-upnp`, a node asks its router to forward its listen ports (UPnP), and
 //! publishes the router's public address for them with its own (`Book::direct`): then
@@ -239,6 +240,7 @@ impl Libp2pTransport {
         let mut book = Book {
             announced: config.announce,
             ignored: config.ignore.clone(),
+            preferred: config.prefer.clone(),
             serving: config.relay,
             ..Book::default()
         };
diff --git a/src/transport/libp2p/tests.rs b/src/transport/libp2p/tests.rs
index 12e4fce..9eb2808 100644
--- a/src/transport/libp2p/tests.rs
+++ b/src/transport/libp2p/tests.rs
@@ -32,6 +32,7 @@ fn memory_node_with(
         mdns: false,
         upnp: false,
         ignore: Vec::new(),
+        prefer: Vec::new(),
         public: false,
         memory: true,
     })
@@ -179,6 +180,7 @@ async fn announced_addresses_are_published_instead() {
         mdns: false,
         upnp: false,
         ignore: Vec::new(),
+        prefer: Vec::new(),
         public: false,
         memory: true,
     })
@@ -384,6 +386,7 @@ async fn a_relay_that_is_no_node() {
         mdns: false,
         upnp: false,
         ignore: Vec::new(),
+        prefer: Vec::new(),
         public: false,
         memory: true,
     })
@@ -601,6 +604,37 @@ fn identify_replaces_direct_addresses() {
     assert!(book.knows(them, &[old]));
 }
 
+/// Of a peer with more addresses than are kept, those that matter stay: in the ranges
+/// `libp2p-prefer` names first, then public ones, then through relays, then other
+/// private ones.
+#[test]
+fn identified_addresses_preferred_first() {
+    let mut book = Book {
+        preferred: vec!["10.23.0.0/16".parse().unwrap()],
+        ..Book::default()
+    };
+    let (relay, them) = (peer(), peer());
+    let lan: Multiaddr = "/ip4/10.23.7.7/tcp/9661".parse().unwrap();
+    let public: Multiaddr = "/ip4/198.51.100.7/tcp/9661".parse().unwrap();
+    let relayed: Multiaddr = format!("/ip4/192.0.2.1/tcp/9661/p2p/{relay}/p2p-circuit")
+        .parse()
+        .unwrap();
+    let bridges: Vec<Multiaddr> = (0..ADDRESSES_PER_PEER)
+        .map(|i| format!("/ip4/172.17.{i}.1/tcp/9661").parse().unwrap())
+        .collect();
+    let mut listen = bridges.clone();
+    listen.extend([relayed.clone(), public.clone(), lan.clone()]);
+
+    book.identified(them, listen);
+    let known = &book.addresses[&them];
+    assert_eq!(known.len(), ADDRESSES_PER_PEER);
+    assert_eq!(known[..3], [lan.clone(), public, relayed]);
+
+    // Our own: those listened on in the preferred ranges published first.
+    book.listening = vec![bridges[0].clone(), lan.clone()];
+    assert_eq!(book.direct(), [lan, bridges[0].clone()]);
+}
+
 /// Peers nobody mentions any more are forgotten, but connected ones.
 #[test]
 fn stale_peers_are_forgotten() {