Merge #19534: net: save the network type explicitly in CNetAddr

bcfebb6d5511ad4c156868bc799831ace628a225 net: save the network type explicitly in CNetAddr (Vasil Dimov)
100c64a95b518a6a19241aec4058b866a8872d9b net: document `enum Network` (Vasil Dimov)

Pull request description:

  (chopped off from https://github.com/bitcoin/bitcoin/pull/19031 to ease review)

  Before this change, we would analyze the contents of `CNetAddr::ip[16]`
  in order to tell which type is an address. Change this by introducing a
  new member `CNetAddr::m_net` that explicitly tells the type of the
  address.

  This is necessary because in BIP155 we will not be able to tell the
  address type by just looking at its raw representation (e.g. both TORv3
  and I2P are "seemingly random" 32 bytes).

  As a side effect of this change we no longer need to store IPv4
  addresses encoded as IPv6 addresses - we can store them in proper 4
  bytes (will be done in a separate commit). Also the code gets
  somewhat simplified - instead of
  `memcmp(ip, pchIPv4, sizeof(pchIPv4)) == 0` we can use
  `m_net == NET_IPV4`.

ACKs for top commit:
  troygiorshev:
    reACK bcfebb6d5511ad4c156868bc799831ace628a225 via `git range-diff master 64897c5 bcfebb6`
  jonatack:
    re-ACK bcfebb6 per `git diff 662bb25 bcfebb6`, code review, debug build/tests clean, ran bitcoind.
  laanwj:
    Code review ACK bcfebb6d5511ad4c156868bc799831ace628a225

Tree-SHA512: 9347e2a50feac617a994bfb46a8f77e31c236bde882e4fd4f03eea4766cd5110216f5f3d24dee91d25218bab7f8bb6e1d2d6212a44db9e34594299fd6ff7606b

3 files changed, 113 insertions, 40 deletions

diff --git a/src/netaddress.cpp b/src/netaddress.cpp
index 0aaba440b8..0874b8dcea 100644
--- a/src/netaddress.cpp
+++ b/src/netaddress.cpp
@@ -28,19 +28,35 @@ CNetAddr::CNetAddr()
 
 void CNetAddr::SetIP(const CNetAddr& ipIn)
 {
+    m_net = ipIn.m_net;
     memcpy(ip, ipIn.ip, sizeof(ip));
 }
 
+void CNetAddr::SetLegacyIPv6(const uint8_t ipv6[16])
+{
+    if (memcmp(ipv6, pchIPv4, sizeof(pchIPv4)) == 0) {
+        m_net = NET_IPV4;
+    } else if (memcmp(ipv6, pchOnionCat, sizeof(pchOnionCat)) == 0) {
+        m_net = NET_ONION;
+    } else if (memcmp(ipv6, g_internal_prefix, sizeof(g_internal_prefix)) == 0) {
+        m_net = NET_INTERNAL;
+    } else {
+        m_net = NET_IPV6;
+    }
+    memcpy(ip, ipv6, 16);
+}
+
 void CNetAddr::SetRaw(Network network, const uint8_t *ip_in)
 {
     switch(network)
     {
         case NET_IPV4:
+            m_net = NET_IPV4;
             memcpy(ip, pchIPv4, 12);
             memcpy(ip+12, ip_in, 4);
             break;
         case NET_IPV6:
-            memcpy(ip, ip_in, 16);
+            SetLegacyIPv6(ip_in);
             break;
         default:
             assert(!"invalid network");
@@ -66,6 +82,7 @@ bool CNetAddr::SetInternal(const std::string &name)
     if (name.empty()) {
         return false;
     }
+    m_net = NET_INTERNAL;
     unsigned char hash[32] = {};
     CSHA256().Write((const unsigned char*)name.data(), name.size()).Finalize(hash);
     memcpy(ip, g_internal_prefix, sizeof(g_internal_prefix));
@@ -89,6 +106,7 @@ bool CNetAddr::SetSpecial(const std::string &strName)
         std::vector<unsigned char> vchAddr = DecodeBase32(strName.substr(0, strName.size() - 6).c_str());
         if (vchAddr.size() != 16-sizeof(pchOnionCat))
             return false;
+        m_net = NET_ONION;
         memcpy(ip, pchOnionCat, sizeof(pchOnionCat));
         for (unsigned int i=0; i<16-sizeof(pchOnionCat); i++)
             ip[i + sizeof(pchOnionCat)] = vchAddr[i];
@@ -123,15 +141,9 @@ bool CNetAddr::IsBindAny() const
     return true;
 }
 
-bool CNetAddr::IsIPv4() const
-{
-    return (memcmp(ip, pchIPv4, sizeof(pchIPv4)) == 0);
-}
+bool CNetAddr::IsIPv4() const { return m_net == NET_IPV4; }
 
-bool CNetAddr::IsIPv6() const
-{
-    return (!IsIPv4() && !IsTor() && !IsInternal());
-}
+bool CNetAddr::IsIPv6() const { return m_net == NET_IPV6; }
 
 bool CNetAddr::IsRFC1918() const
 {
@@ -165,50 +177,54 @@ bool CNetAddr::IsRFC5737() const
 
 bool CNetAddr::IsRFC3849() const
 {
-    return GetByte(15) == 0x20 && GetByte(14) == 0x01 && GetByte(13) == 0x0D && GetByte(12) == 0xB8;
+    return IsIPv6() && GetByte(15) == 0x20 && GetByte(14) == 0x01 &&
+           GetByte(13) == 0x0D && GetByte(12) == 0xB8;
 }
 
 bool CNetAddr::IsRFC3964() const
 {
-    return (GetByte(15) == 0x20 && GetByte(14) == 0x02);
+    return IsIPv6() && GetByte(15) == 0x20 && GetByte(14) == 0x02;
 }
 
 bool CNetAddr::IsRFC6052() const
 {
     static const unsigned char pchRFC6052[] = {0,0x64,0xFF,0x9B,0,0,0,0,0,0,0,0};
-    return (memcmp(ip, pchRFC6052, sizeof(pchRFC6052)) == 0);
+    return IsIPv6() && memcmp(ip, pchRFC6052, sizeof(pchRFC6052)) == 0;
 }
 
 bool CNetAddr::IsRFC4380() const
 {
-    return (GetByte(15) == 0x20 && GetByte(14) == 0x01 && GetByte(13) == 0 && GetByte(12) == 0);
+    return IsIPv6() && GetByte(15) == 0x20 && GetByte(14) == 0x01 && GetByte(13) == 0 &&
+           GetByte(12) == 0;
 }
 
 bool CNetAddr::IsRFC4862() const
 {
     static const unsigned char pchRFC4862[] = {0xFE,0x80,0,0,0,0,0,0};
-    return (memcmp(ip, pchRFC4862, sizeof(pchRFC4862)) == 0);
+    return IsIPv6() && memcmp(ip, pchRFC4862, sizeof(pchRFC4862)) == 0;
 }
 
 bool CNetAddr::IsRFC4193() const
 {
-    return ((GetByte(15) & 0xFE) == 0xFC);
+    return IsIPv6() && (GetByte(15) & 0xFE) == 0xFC;
 }
 
 bool CNetAddr::IsRFC6145() const
 {
     static const unsigned char pchRFC6145[] = {0,0,0,0,0,0,0,0,0xFF,0xFF,0,0};
-    return (memcmp(ip, pchRFC6145, sizeof(pchRFC6145)) == 0);
+    return IsIPv6() && memcmp(ip, pchRFC6145, sizeof(pchRFC6145)) == 0;
 }
 
 bool CNetAddr::IsRFC4843() const
 {
-    return (GetByte(15) == 0x20 && GetByte(14) == 0x01 && GetByte(13) == 0x00 && (GetByte(12) & 0xF0) == 0x10);
+    return IsIPv6() && GetByte(15) == 0x20 && GetByte(14) == 0x01 &&
+           GetByte(13) == 0x00 && (GetByte(12) & 0xF0) == 0x10;
 }
 
 bool CNetAddr::IsRFC7343() const
 {
-    return (GetByte(15) == 0x20 && GetByte(14) == 0x01 && GetByte(13) == 0x00 && (GetByte(12) & 0xF0) == 0x20);
+    return IsIPv6() && GetByte(15) == 0x20 && GetByte(14) == 0x01 &&
+           GetByte(13) == 0x00 && (GetByte(12) & 0xF0) == 0x20;
 }
 
 bool CNetAddr::IsHeNet() const
@@ -222,10 +238,7 @@ bool CNetAddr::IsHeNet() const
  *
  * @see CNetAddr::SetSpecial(const std::string &)
  */
-bool CNetAddr::IsTor() const
-{
-    return (memcmp(ip, pchOnionCat, sizeof(pchOnionCat)) == 0);
-}
+bool CNetAddr::IsTor() const { return m_net == NET_ONION; }
 
 bool CNetAddr::IsLocal() const
 {
@@ -235,7 +248,7 @@ bool CNetAddr::IsLocal() const
 
     // IPv6 loopback (::1/128)
     static const unsigned char pchLocal[16] = {0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1};
-    if (memcmp(ip, pchLocal, 16) == 0)
+    if (IsIPv6() && memcmp(ip, pchLocal, 16) == 0)
         return true;
 
     return false;
@@ -259,12 +272,12 @@ bool CNetAddr::IsValid() const
     // header20 vectorlen3 addr26 addr26 addr26 header20 vectorlen3 addr26 addr26 addr26...
     // so if the first length field is garbled, it reads the second batch
     // of addr misaligned by 3 bytes.
-    if (memcmp(ip, pchIPv4+3, sizeof(pchIPv4)-3) == 0)
+    if (IsIPv6() && memcmp(ip, pchIPv4+3, sizeof(pchIPv4)-3) == 0)
         return false;
 
     // unspecified IPv6 address (::/128)
     unsigned char ipNone6[16] = {};
-    if (memcmp(ip, ipNone6, 16) == 0)
+    if (IsIPv6() && memcmp(ip, ipNone6, 16) == 0)
         return false;
 
     // documentation IPv6 address
@@ -311,7 +324,7 @@ bool CNetAddr::IsRoutable() const
  */
 bool CNetAddr::IsInternal() const
 {
-   return memcmp(ip, g_internal_prefix, sizeof(g_internal_prefix)) == 0;
+   return m_net == NET_INTERNAL;
 }
 
 enum Network CNetAddr::GetNetwork() const
@@ -322,13 +335,7 @@ enum Network CNetAddr::GetNetwork() const
     if (!IsRoutable())
         return NET_UNROUTABLE;
 
-    if (IsIPv4())
-        return NET_IPV4;
-
-    if (IsTor())
-        return NET_ONION;
-
-    return NET_IPV6;
+    return m_net;
 }
 
 std::string CNetAddr::ToStringIP() const
@@ -362,12 +369,12 @@ std::string CNetAddr::ToString() const
 
 bool operator==(const CNetAddr& a, const CNetAddr& b)
 {
-    return (memcmp(a.ip, b.ip, 16) == 0);
+    return a.m_net == b.m_net && memcmp(a.ip, b.ip, 16) == 0;
 }
 
 bool operator<(const CNetAddr& a, const CNetAddr& b)
 {
-    return (memcmp(a.ip, b.ip, 16) < 0);
+    return a.m_net < b.m_net || (a.m_net == b.m_net && memcmp(a.ip, b.ip, 16) < 0);
 }
 
 /**
@@ -813,7 +820,7 @@ CSubNet::CSubNet(const CNetAddr &addr):
  */
 bool CSubNet::Match(const CNetAddr &addr) const
 {
-    if (!valid || !addr.IsValid())
+    if (!valid || !addr.IsValid() || network.m_net != addr.m_net)
         return false;
     for(int x=0; x<16; ++x)
         if ((addr.ip[x] & netmask[x]) != network.ip[x])
diff --git a/src/netaddress.h b/src/netaddress.h
index f2daad7fb6..0365907d44 100644
--- a/src/netaddress.h
+++ b/src/netaddress.h
@@ -16,21 +16,50 @@
 #include <string>
 #include <vector>
 
+/**
+ * A network type.
+ * @note An address may belong to more than one network, for example `10.0.0.1`
+ * belongs to both `NET_UNROUTABLE` and `NET_IPV4`.
+ * Keep these sequential starting from 0 and `NET_MAX` as the last entry.
+ * We have loops like `for (int i = 0; i < NET_MAX; i++)` that expect to iterate
+ * over all enum values and also `GetExtNetwork()` "extends" this enum by
+ * introducing standalone constants starting from `NET_MAX`.
+ */
 enum Network
 {
+    /// Addresses from these networks are not publicly routable on the global Internet.
     NET_UNROUTABLE = 0,
+
+    /// IPv4
     NET_IPV4,
+
+    /// IPv6
     NET_IPV6,
+
+    /// TORv2
     NET_ONION,
+
+    /// A set of dummy addresses that map a name to an IPv6 address. These
+    /// addresses belong to RFC4193's fc00::/7 subnet (unique-local addresses).
+    /// We use them to map a string or FQDN to an IPv6 address in CAddrMan to
+    /// keep track of which DNS seeds were used.
     NET_INTERNAL,
 
+    /// Dummy value to indicate the number of NET_* constants.
     NET_MAX,
 };
 
-/** IP address (IPv6, or IPv4 using mapped IPv6 range (::FFFF:0:0/96)) */
+/**
+ * Network address.
+ */
 class CNetAddr
 {
     protected:
+        /**
+         * Network to which this address belongs.
+         */
+        Network m_net{NET_IPV6};
+
         unsigned char ip[16]; // in network byte order
         uint32_t scopeId{0}; // for scoped/link-local ipv6 addresses
 
@@ -40,6 +69,14 @@ class CNetAddr
         void SetIP(const CNetAddr& ip);
 
         /**
+         * Set from a legacy IPv6 address.
+         * Legacy IPv6 address may be a normal IPv6 address, or another address
+         * (e.g. IPv4) disguised as IPv6. This encoding is used in the legacy
+         * `addr` encoding.
+         */
+        void SetLegacyIPv6(const uint8_t ipv6[16]);
+
+        /**
          * Set raw IPv4 or IPv6 address (in network byte order)
          * @note Only NET_IPV4 and NET_IPV6 are allowed for network.
          */
@@ -100,7 +137,27 @@ class CNetAddr
         friend bool operator!=(const CNetAddr& a, const CNetAddr& b) { return !(a == b); }
         friend bool operator<(const CNetAddr& a, const CNetAddr& b);
 
-        SERIALIZE_METHODS(CNetAddr, obj) { READWRITE(obj.ip); }
+        /**
+         * Serialize to a stream.
+         */
+        template <typename Stream>
+        void Serialize(Stream& s) const
+        {
+            s << ip;
+        }
+
+        /**
+         * Unserialize from a stream.
+         */
+        template <typename Stream>
+        void Unserialize(Stream& s)
+        {
+            unsigned char ip_temp[sizeof(ip)];
+            s >> ip_temp;
+            // Use SetLegacyIPv6() so that m_net is set correctly. For example
+            // ::FFFF:0102:0304 should be set as m_net=NET_IPV4 (1.2.3.4).
+            SetLegacyIPv6(ip_temp);
+        }
 
         friend class CSubNet;
 };
diff --git a/src/test/netbase_tests.cpp b/src/test/netbase_tests.cpp
index ea3e633cc2..591b4ce49a 100644
--- a/src/test/netbase_tests.cpp
+++ b/src/test/netbase_tests.cpp
@@ -138,6 +138,14 @@ BOOST_AUTO_TEST_CASE(onioncat_test)
 
 }
 
+BOOST_AUTO_TEST_CASE(embedded_test)
+{
+    CNetAddr addr1(ResolveIP("1.2.3.4"));
+    CNetAddr addr2(ResolveIP("::FFFF:0102:0304"));
+    BOOST_CHECK(addr2.IsIPv4());
+    BOOST_CHECK_EQUAL(addr1.ToString(), addr2.ToString());
+}
+
 BOOST_AUTO_TEST_CASE(subnet_test)
 {
 
@@ -158,12 +166,13 @@ BOOST_AUTO_TEST_CASE(subnet_test)
     BOOST_CHECK(ResolveSubNet("1.2.2.1/24").Match(ResolveIP("1.2.2.4")));
     BOOST_CHECK(ResolveSubNet("1.2.2.110/31").Match(ResolveIP("1.2.2.111")));
     BOOST_CHECK(ResolveSubNet("1.2.2.20/26").Match(ResolveIP("1.2.2.63")));
-    // All-Matching IPv6 Matches arbitrary IPv4 and IPv6
+    // All-Matching IPv6 Matches arbitrary IPv6
     BOOST_CHECK(ResolveSubNet("::/0").Match(ResolveIP("1:2:3:4:5:6:7:1234")));
     // But not `::` or `0.0.0.0` because they are considered invalid addresses
     BOOST_CHECK(!ResolveSubNet("::/0").Match(ResolveIP("::")));
     BOOST_CHECK(!ResolveSubNet("::/0").Match(ResolveIP("0.0.0.0")));
-    BOOST_CHECK(ResolveSubNet("::/0").Match(ResolveIP("1.2.3.4")));
+    // Addresses from one network (IPv4) don't belong to subnets of another network (IPv6)
+    BOOST_CHECK(!ResolveSubNet("::/0").Match(ResolveIP("1.2.3.4")));
     // All-Matching IPv4 does not Match IPv6
     BOOST_CHECK(!ResolveSubNet("0.0.0.0/0").Match(ResolveIP("1:2:3:4:5:6:7:1234")));
     // Invalid subnets Match nothing (not even invalid addresses)