p2p: remove tor v2 support

jonatack commented at 9:48 pm on May 24, 2021: member

This patch removes support in Bitcoin Core for Tor v2 onions, which are already removed from the release of Tor 0.4.6.

no longer serialize/deserialize and relay Tor v2 addresses
ignore incoming Tor v2 addresses
remove Tor v2 addresses from the addrman and peers.dat on node launch
update generate-seeds.py to ignore Tor v2 addresses
remove Tor v2 hard-coded seeds

Tested with tor-0.4.6.1-alpha (no v2 support) and 0.4.5.7 (v2 support). With the latest Tor (no v2 support), this removes all the warnings like those reported with current master in #21351

0<bitcoind debug log>
1Socks5() connect to […].onion:8333 failed: general failure
2
3<tor log>
4Invalid hostname [scrubbed]; rejecting

and the addrman no longer has Tor v2 addresses on launching bitcoind.

 0$ ./src/bitcoin-cli -addrinfo
 1{
 2  "addresses_known": {
 3    "ipv4": 44483,
 4    "ipv6": 8467,
 5    "torv2": 0,
 6    "torv3": 2296,
 7    "i2p": 6,
 8    "total": 55252
 9  }
10}

After recompiling back to current master and restarting with either of the two Tor versions (0.4.5.7 or 0.4.6.1), -addrinfo initially returns 0 Tor v2 addresses and then begins finding them again.

Ran nodes on this patch over the past week on mainnet/testnet/signet/regtest after building with DEBUG_ADDRMAN.

Verified that this patch bootstraps an onlynet=onion node from the Tor v3 hardcoded fixed seeds on mainnet and testnet and connects to blocks and v3 onion peers: rm ~/.bitcoin/testnet3/peers.dat ; ./src/bitcoind -testnet -dnsseed=0 -onlynet=onion

Screenshot from 2021-05-28 00-26-17

Tested using addnode, getaddednodeinfo,addpeeraddress, disconnectnode and -addrinfo that a currently valid, connectable Tor v2 peer can no longer be added:

Screenshot from 2021-05-30 11-32-05

Thanks to Vasil Dimov, Carl Dong, and Wladimir J. van der Laan for their work on BIP155 and Tor v3 that got us here.

jonatack commented at 10:01 pm on May 24, 2021: member

A recent IRC discussion for more context. http://www.erisian.com.au/bitcoin-core-dev/log-2021-05-20.html#l-401

DrahtBot added the label P2P on May 24, 2021

DrahtBot added the label Scripts and tools on May 24, 2021

theStack commented at 11:05 pm on May 24, 2021: member

Concept ACK

jonatack force-pushed on May 24, 2021

Rspigler commented at 2:12 am on May 25, 2021: contributor

Concept ACK

jonatack marked this as a draft on May 25, 2021

dunxen commented at 6:53 am on May 25, 2021: contributor

Concept ACK. (Thank you for your service, tor v2)

0xB10C commented at 7:08 am on May 25, 2021: member

Concept ACK

laanwj added this to the milestone 22.0 on May 25, 2021

laanwj commented at 2:53 pm on May 25, 2021: member

Concept ACK, adding 22.0 label Going to test this.

DrahtBot commented at 5:46 pm on May 25, 2021: member

The following sections might be updated with supplementary metadata relevant to reviewers and maintainers.

Conflicts

No conflicts as of last run.

jonatack force-pushed on May 26, 2021

test: update feature_proxy to torv3 0f1c58ae87

p2p: remove torv2/ADDR_TORV2_SIZE from SetTor() f8e94002fc

in contrib/seeds/nodes_main.txt:653 in 7cfec99a26 outdated

649@@ -650,518 +650,6 @@
650 [2a0f:df00:0:254::46]:8333
651 [2c0f:f598:5:1:1001::1]:8333
652 [2c0f:fce8:0:400:b7c::1]:8333
653-226eupdnaouu4h2v.onion:8333

laanwj commented at 11:23 am on May 27, 2021:

Please also run cd contrib/seeds && python3 generate-seeds.py . > ../../src/chainparamsseeds.h to update chainparamsseeds.h (in the same commit?).

jonatack commented at 9:43 pm on May 27, 2021:

Thanks! Done.

jonatack force-pushed on May 27, 2021

p2p: drop onions from IsAddrV1Compatible(), no longer relay torv2 c56a1c9b18

fuzz: rename CNetAddr/CService deserialize targets

as the changes that follow are incompatible with the inputs.

eba9a94b9f

p2p: remove torv2 from SetNetFromBIP155Network() 7d1769bc45

jonatack force-pushed on May 27, 2021

jonatack commented at 8:18 am on May 28, 2021: member

This seems to be ready. Updated the PR description. Will propose release documentation in a follow-up.

jonatack marked this as ready for review on May 28, 2021

willcl-ark commented at 8:01 pm on May 28, 2021: member

Light tested ACK, rebased on top of master @ 123b401e0acf3b26f149711bdf06d7f6eced1968

Used Tor version 0.4.5.7. with manually configured hidden service and a bitcoin.conf with

0listenonion=1
1onlynet=onion
2proxy=127.0.0.1:9050

Node has not picked up any Tor V2 addresses since startup and has a healthy number of V3 addresses in a short period of time:

Tested bootstrapping without peers.dat and that picked up over 1000 V3 peers in just a few minutes.
Adding a V2 address check back into src/test/net_test.cpp resulted in expected test failure:

test/net_tests.cpp(614): error: in "net_tests/cnetaddr_unserialize_v2": check addr.IsValid() has failed

willcl-ark approved

in contrib/seeds/generate-seeds.py:50 in 5715134414 outdated

45@@ -46,9 +46,7 @@ def name_to_bip155(addr):
46     '''Convert address string to BIP155 (networkID, addr) tuple.'''
47     if addr.endswith('.onion'):
48         vchAddr = b32decode(addr[0:-6], True)
49-        if len(vchAddr) == 10:
50-            return (BIP155Network.TORV2, vchAddr)

benthecarman commented at 8:39 pm on May 28, 2021:

I think an error message specific for v2 not being supported would be better than the defualt error message

laanwj commented at 9:25 pm on May 28, 2021:

I guess makeseeds.py is still producing these? It should probably ignore Torv2’s from the DNS seeder (all of them, at the moment sipa/bitcoin-seeder#92) completely.

jonatack commented at 10:04 am on May 29, 2021:

Thanks, updated generate-seeds.py to ignore torv2 rather than raise.

jonatack force-pushed on May 29, 2021

jonatack commented at 10:10 am on May 29, 2021: member

Changed generate-seeds.py to ignore torv2 addresses rather than raise. (Edit: I don’t know if using None or an optional type would be preferred to -1 or more pythonic, happy to update if so.)

 0diff --git a/contrib/seeds/generate-seeds.py b/contrib/seeds/generate-seeds.py
 1index aaccefcf6f..ae97240203 100755
 2--- a/contrib/seeds/generate-seeds.py
 3+++ b/contrib/seeds/generate-seeds.py
 4@@ -47,8 +47,10 @@ def name_to_bip155(addr):
 5     if addr.endswith('.onion'):
 6         vchAddr = b32decode(addr[0:-6], True)
 7         if len(vchAddr) == 35:
 8-            assert(vchAddr[34] == 3)
 9+            assert vchAddr[34] == 3
10             return (BIP155Network.TORV3, vchAddr[:32])
11+        elif len(vchAddr) == 10:
12+            return (BIP155Network.TORV2, vchAddr)
13         else:
14             raise ValueError('Invalid onion %s' % vchAddr)
15     elif addr.endswith('.b32.i2p'):
16@@ -98,7 +100,10 @@ def parse_spec(s):
17 
18     host = name_to_bip155(host)
19 
20-    return host + (port, )
21+    if host[0] == BIP155Network.TORV2:
22+        return -1  # TORV2 is no longer supported, so we ignore it
23+    else:
24+        return host + (port, )
25 
26 def ser_compact_size(l):
27     r = b""
28@@ -134,6 +139,8 @@ def process_nodes(g, f, structname):
29             continue
30 
31         spec = parse_spec(line)
32+        if spec == -1:  # ignore this entry (e.g. no longer supported addresses like TORV2)
33+            continue
34         blob = bip155_serialize(spec)
35         hoststr = ','.join(('0x%02x' % b) for b in blob)
36         g.write(f'    {hoststr},\n')

benthecarman commented at 10:20 am on May 29, 2021: contributor

utACK ef34bda6c52d7a59be9cd3451bc3536430410e36

jonatack commented at 6:38 pm on May 29, 2021: member

Tested using addnode, getaddednodeinfo,addpeeraddress, disconnectnode and -addrinfo that a currently valid, connectable Tor v2 peer can no longer be added:

Screenshot from 2021-05-30 11-32-05

dunxen commented at 6:47 pm on May 29, 2021: contributor

tACK ef34bda

Using Tor v0.4.4.6.
Node originally had established connections with Tor v2 peers, now does not connect to v2 peers on patch.
After a short while, with onlynet=onion, node is connected to 8+ Tor v3 peers exclusively.

No Tor v2 addresses known

0{
1  "addresses_known": {
2    "ipv4": 309,
3    "ipv6": 33,
4    "torv2": 0,
5    "torv3": 3563,
6    "i2p": 0,
7    "total": 3905
8  }
9}

Cannot add v2 address or connect to v2 address with addnode.

Rspigler commented at 0:45 am on May 30, 2021: contributor

Could we add a more detailed error message for someone who misses or doesn’t read the release docs?

Zero-1729 commented at 3:13 am on May 30, 2021: contributor

tACK ef34bda6c52d7a59be9cd3451bc3536430410e36

Using Tor v0.4.5.8 on MacOS 11.3.1 (20E241)

Note: Tor v2 seed (bk7yp6epnmcllq72.onion) used is from this article.

Started bitcoind with -onlynet=onion -listenonion=1 -proxy=127.0.0.1:9050

Before (on master be1c512437f7aa8215d2e05eda5cc8b58c3d3937):

Establishes connection with Tor v2 peers.

 0$ bitcoin-cli -addrinfo
 1{
 2  "addresses_known": {
 3    "ipv4": 272,
 4    "ipv6": 0,
 5    "torv2": 2598,
 6    "torv3": 345,
 7    "i2p": 0,
 8    "total": 3215
 9  }
10}

Successfully adds Tor v2 peer bk7yp6epnmcllq72.onion with addpeeraddress.

After (on torv2-remove-support ef34bda6c52d7a59be9cd3451bc3536430410e36):

No longer connects to Tor v2 peers.

0{
1  "addresses_known": {
2    "ipv4": 271,
3    "ipv6": 0,
4    "torv2": 0,
5    "torv3": 352,
6    "i2p": 0,
7    "total": 623
8  }
9}

Establishing manual connection with bk7yp6epnmcllq72.onion (Tor v2) using addpeeraddress fails.
Adding bk7yp6epnmcllq72.onion via addnode also doesn’t work.

$ bitcoin-cli addpeeraddress bk7yp6epnmcllq72.onion 8333  
{
    "success": false
}
$ bitcoin-cli addnode bk7yp6epnmcllq72.onion onetry   
$ bitcoin-cli disconnectnode bk7yp6epnmcllq72.onion
error code: -29
error message:
Node not found in connected nodes

Can verify connections to Tor v2 peers no longer established, including when using addnode & addpeeraddress after patch (ef34bda6c52d7a59be9cd3451bc3536430410e36) 👍

jonatack commented at 10:04 am on May 30, 2021: member

Updated #22050 (comment) to also mention getaddednodeinfo.

Could we add a more detailed error message for someone who misses or doesn’t read the release docs?

Good question. Are you thinking of addpeeraddress or addnode? The former is a hidden test/debug RPC. The latter can be a valuable anti-Eclipse measure to help ensure a node has at least one honest peer, but it is presumably used by informed users.

I was kind of hoping that updating doc/tor.md and adding an announcement toward the top of the release notes (both in a follow-up), along with communication on twitter and the fediverse (and by the Tor Project, who propose to communicate together for the transition) and maybe resources like the ML and Bitcoin Optech would be enough.

This is a one-off event and soon the Tor v2 network will no longer be operating at all; according to the Tor AMA event last Friday, the v2 net is already less robust than the v3 one. This should soon be visible in the metrics at https://metrics.torproject.org/versions.html.

Rspigler commented at 10:24 pm on May 30, 2021: contributor

I was kind of hoping that updating doc/tor.md and adding an announcement toward the top of the release notes (both in a follow-up), along with communication on twitter and the fediverse (and by the Tor Project, who propose to communicate together for the transition) and maybe resources like the ML and Bitcoin Optech would be enough.

Good point, I think if we do all that, it should be enough

in contrib/seeds/generate-seeds.py:104 in ef34bda6c5 outdated

 99@@ -100,7 +100,10 @@ def parse_spec(s):
100 
101     host = name_to_bip155(host)
102 
103-    return host + (port, )
104+    if host[0] == BIP155Network.TORV2:
105+        return -1  # TORV2 is no longer supported, so we ignore it

kiminuo commented at 9:21 am on June 1, 2021:

Would it be possibly better to return None here?

jonatack commented at 9:51 am on June 1, 2021:

Yes, I asked this in #22050 (comment), happy to update if reviewers think it’s worth changing.

kiminuo commented at 4:31 pm on June 1, 2021:

Ah, sorry. None +1 from me :)

jonatack commented at 8:16 pm on June 1, 2021:

Ok, I have a couple minor refactorings ready that aren’t essential to removing tor v2 support and propose to insert this change into that commit, unless I need to update this patch for another reason.

jonatack commented at 11:30 am on June 3, 2021:

done

in contrib/seeds/generate-seeds.py:53 in ef34bda6c5 outdated

52-            assert(vchAddr[34] == 3)
53+        if len(vchAddr) == 35:
54+            assert vchAddr[34] == 3
55             return (BIP155Network.TORV3, vchAddr[:32])
56+        elif len(vchAddr) == 10:
57+            return (BIP155Network.TORV2, vchAddr)

kiminuo commented at 9:24 am on June 1, 2021:

This re-ordering change looks unnecessary now. The diff would be smaller when you revert this 4 line change. :-)

jonatack commented at 9:51 am on June 1, 2021:

Most frequent path first.

in src/netaddress.h:337 in ef34bda6c5 outdated

329@@ -334,31 +330,22 @@ class CNetAddr
330                 memcpy(arr, IPV4_IN_IPV6_PREFIX.data(), prefix_size);
331                 memcpy(arr + prefix_size, m_addr.data(), m_addr.size());
332                 return;
333-            case NET_ONION:

kiminuo commented at 9:37 am on June 1, 2021:

L301 mentions that GetBIP155Network() Must not be called for IsInternal() objects.

Would it make sense to add a note after L301 about Tor v2 not being supported anymore? The reason why it may make sense is simply because BIP155 is written with Tor v2 in mind (and that won’t change, or will it?) and the function name is GetBIP155Network. So we are about to deviate from BIP155. For a good reason but still.

jonatack commented at 9:53 am on June 1, 2021:

I’m not sure if you’re confusing “addrv2” and “torv2”. What comment would you suggest?

kiminuo commented at 6:22 pm on June 1, 2021:

The idea was to change:

0/**
1 * Get the BIP155 network id of this address.
2 * Must not be called for IsInternal() objects.
3 * [@returns](/bitcoin-bitcoin/contributor/returns/) BIP155 network id
4 */
5BIP155Network GetBIP155Network() const;

to something like

0/**
1 * Get the BIP155 network id of this address.
2 * Must not be called for IsInternal() objects.
3 * [@returns](/bitcoin-bitcoin/contributor/returns/) BIP155 network id except [BIP155Network::]TORV2 which is no longer supported
4 */
5BIP155Network GetBIP155Network() const;

but it’s probably an overkill.

jonatack commented at 8:19 pm on June 1, 2021:

SGTM, will add to the follow-up refactoring commit unless this patch needs a more substantial update.

jonatack commented at 11:35 am on June 3, 2021:

Updated

0-         * [@returns](/bitcoin-bitcoin/contributor/returns/) BIP155 network id
1+         * [@returns](/bitcoin-bitcoin/contributor/returns/) BIP155 network id, except TORV2 which is no longer supported.

kiminuo commented at 9:46 am on June 1, 2021: contributor

@jonatack $ ./src/bitcoin-cli -addrinfo reports torv2 entry as mentioned in OP. The plan is to remove this later, I guess, right?

jonatack commented at 9:56 am on June 1, 2021: member

@jonatack $ ./src/bitcoin-cli -addrinfo reports torv2 entry as mentioned in OP. The plan is to remove this later, I guess, right?

Indeed yes, the reason for -addrinfo returning torv2 and torv3 is for this transition period so the user can see if their node knows enough torv3 peers. When torv2 peers are gone, we can simplify -addrinfo to return only “onion” instead.

reemuru commented at 7:50 pm on June 1, 2021: none

tACK ef34bda. Run on mainnet, Fedora 34 stable.

0[user@fedora bitcoin]$ tor --version
1Jun 01 14:44:30.057 [warn] Tor was compiled with zstd 1.4.9, but is running with zstd 1.5.0. For safety, we'll avoid using advanced zstd functionality.
2Tor version 0.4.5.8.

[user@fedora ~]$ bitcoin/src/bitcoin-cli -addrinfo
{
  "addresses_known": {
    "ipv4": 39776,
    "ipv6": 7431,
    "torv2": 17,
    "torv3": 8,
    "i2p": 0,
    "total": 47232
  }
}

[user@fedora ~]$ bitcoin/src/bitcoin-cli -addrinfo
{
  "addresses_known": {
  "ipv4": 39729,
  "ipv6": 7425,
  "torv2": 0,
  "torv3": 416,
  "i2p": 0,
  "total": 47570
  }
}

reemuru approved

MarcoFalke requested review from laanwj on Jun 2, 2021

MarcoFalke commented at 5:02 am on June 2, 2021: member

cc @laanwj from #22050 (comment)

in contrib/seeds/generate-seeds.py:142 in ef34bda6c5 outdated

138@@ -136,6 +139,8 @@ def process_nodes(g, f, structname):
139             continue
140 
141         spec = parse_spec(line)
142+        if spec == -1:  # ignore this entry (e.g. no longer supported addresses like TORV2)

laanwj commented at 7:18 am on June 3, 2021:

I would prefer using the sentinel value None instead of -1 then check with spec is None (just a small nit though, no need to change if no other changes are necessary).

jonatack commented at 11:29 am on June 3, 2021:

done

in src/netaddress.cpp:163 in ef34bda6c5 outdated

144@@ -160,10 +145,6 @@ void CNetAddr::SetLegacyIPv6(Span<const uint8_t> ipv6)
145         // IPv4-in-IPv6
146         m_net = NET_IPV4;
147         skip = sizeof(IPV4_IN_IPV6_PREFIX);
148-    } else if (HasPrefix(ipv6, TORV2_IN_IPV6_PREFIX)) {

laanwj commented at 7:19 am on June 3, 2021:

Don’t we need some kind of check here to be able to filter out (ignore) torv2-in-ipv6 in old style addr messages?

jonatack commented at 10:53 am on June 3, 2021:

Yes, I thought they would be invalid (so ignored) without the removed code, but it may be better to be explicit about it with something like:

0    } else if (HasPrefix(ipv6, TORV2_IN_IPV6_PREFIX)) {
1        // TORv2-in-IPv6 (unsupported). Unserialize as !IsValid(), thus ignoring them.
2        // Mimic a default-constructed CNetAddr object which is !IsValid() and thus
3        // will not be gossiped, but continue reading next addresses from the stream.
4        m_net = NET_IPV6;
5        m_addr.assign(ADDR_IPV6_SIZE, 0x0);
6        return;

jonatack commented at 11:36 am on June 3, 2021:

done

p2p: remove BIP155Network::TORV2 from GetBIP155Network() 1d631e956f

p2p: ignore torv2-in-ipv6 addresses in SetLegacyIPv6() 4192a74413

p2p: remove pre-addrv2 onions from SerializeV1Array() cff5ec477a

p2p: remove torv2 in SetIP() and ADDR_TORV2_SIZE constant 3d39042144

p2p: remove torv2 from CNetAddr::ToStringIP() 5f9d3c09b4

p2p, refactor: extract OnionToString() from CNetAddr::ToStringIp() 8be56f0f8e

jonatack force-pushed on Jun 3, 2021

jonatack commented at 11:51 am on June 3, 2021: member

Changed the torv2-in-ipv6 case to deserialize explicitly to an invalid address which will be ignored, changed the ignore value in generate-seeds.py from -1 to None, extracted the onion-to-string code to a helper, and updated a Doxygen comment.

 0diff --git a/contrib/seeds/generate-seeds.py b/contrib/seeds/generate-seeds.py
 1index ae97240203..0ce868bfcc 100755
 2--- a/contrib/seeds/generate-seeds.py
 3+++ b/contrib/seeds/generate-seeds.py
 4@@ -101,7 +101,7 @@ def parse_spec(s):
 5     host = name_to_bip155(host)
 6 
 7     if host[0] == BIP155Network.TORV2:
 8-        return -1  # TORV2 is no longer supported, so we ignore it
 9+        return None  # TORV2 is no longer supported, so we ignore it
10     else:
11         return host + (port, )
12 
13@@ -139,7 +139,7 @@ def process_nodes(g, f, structname):
14             continue
15 
16         spec = parse_spec(line)
17-        if spec == -1:  # ignore this entry (e.g. no longer supported addresses like TORV2)
18+        if spec is None:  # ignore this entry (e.g. no longer supported addresses like TORV2)
19             continue
20         blob = bip155_serialize(spec)
21         hoststr = ','.join(('0x%02x' % b) for b in blob)
22diff --git a/src/netaddress.cpp b/src/netaddress.cpp
23index e46f9c2568..0ae8dd0698 100644
24--- a/src/netaddress.cpp
25+++ b/src/netaddress.cpp
26@@ -145,6 +145,13 @@ void CNetAddr::SetLegacyIPv6(Span<const uint8_t> ipv6)
27         // IPv4-in-IPv6
28         m_net = NET_IPV4;
29         skip = sizeof(IPV4_IN_IPV6_PREFIX);
30+    } else if (HasPrefix(ipv6, TORV2_IN_IPV6_PREFIX)) {
31+        // TORv2-in-IPv6 (unsupported). Unserialize as !IsValid(), thus ignoring them.
32+        // Mimic a default-constructed CNetAddr object which is !IsValid() and thus
33+        // will not be gossiped, but continue reading next addresses from the stream.
34+        m_net = NET_IPV6;
35+        m_addr.assign(ADDR_IPV6_SIZE, 0x0);
36+        return;
37     } else if (HasPrefix(ipv6, INTERNAL_IN_IPV6_PREFIX)) {
38         // Internal-in-IPv6
39         m_net = NET_INTERNAL;
40@@ -587,31 +594,26 @@ static std::string IPv6ToString(Span<const uint8_t> a, uint32_t scope_id)
41     return r;
42 }
43 
44+static std::string OnionToString(const Span<const uint8_t>& addr)
45+{
46+    uint8_t checksum[torv3::CHECKSUM_LEN];
47+    torv3::Checksum(addr, checksum);
48+    // TORv3 onion_address = base32(PUBKEY | CHECKSUM | VERSION) + ".onion"
49+    prevector<torv3::TOTAL_LEN, uint8_t> address{addr.begin(), addr.end()};
50+    address.insert(address.end(), checksum, checksum + torv3::CHECKSUM_LEN);
51+    address.insert(address.end(), torv3::VERSION, torv3::VERSION + sizeof(torv3::VERSION));
52+    return EncodeBase32(address) + ".onion";
53+}
54+
55 std::string CNetAddr::ToStringIP() const
56 {
57     switch (m_net) {
58     case NET_IPV4:
59         return IPv4ToString(m_addr);
60-    case NET_IPV6: {
61+    case NET_IPV6:
62         return IPv6ToString(m_addr, m_scope_id);
63-    }
64     case NET_ONION:
65-        switch (m_addr.size()) {
66-        case ADDR_TORV3_SIZE: {
67-
68-            uint8_t checksum[torv3::CHECKSUM_LEN];
69-            torv3::Checksum(m_addr, checksum);
70-
71-            // TORv3 onion_address = base32(PUBKEY | CHECKSUM | VERSION) + ".onion"
72-            prevector<torv3::TOTAL_LEN, uint8_t> address{m_addr.begin(), m_addr.end()};
73-            address.insert(address.end(), checksum, checksum + torv3::CHECKSUM_LEN);
74-            address.insert(address.end(), torv3::VERSION, torv3::VERSION + sizeof(torv3::VERSION));
75-
76-            return EncodeBase32(address) + ".onion";
77-        }
78-        default:
79-            assert(false);
80-        }
81+        return OnionToString(m_addr);
82     case NET_I2P:
83         return EncodeBase32(m_addr, false /* don't pad with = */) + ".b32.i2p";
84     case NET_CJDNS:
85diff --git a/src/netaddress.h b/src/netaddress.h
86index d539bd51ff..0d04ab88fb 100644
87--- a/src/netaddress.h
88+++ b/src/netaddress.h
89@@ -299,7 +299,7 @@ class CNetAddr
90         /**
91          * Get the BIP155 network id of this address.
92          * Must not be called for IsInternal() objects.
93-         * [@returns](/bitcoin-bitcoin/contributor/returns/) BIP155 network id
94+         * [@returns](/bitcoin-bitcoin/contributor/returns/) BIP155 network id, except TORV2 which is no longer supported.
95          */
96         BIP155Network GetBIP155Network() const;

contrib: update generate-seeds.py to ignore torv2 addresses 5f7e086dac

contrib: remove torv2 seed nodes 5d82a57db4

jonatack force-pushed on Jun 3, 2021

laanwj commented at 1:00 pm on June 3, 2021: member

Changes look good to me now. Code review ACK 5d82a57db4f67506a4e80d186ba76f3a8665e147 Tested ACK of an earlier commit, which I tested for a while on one of my nodes. I noticed no difference in behavior except for no longer connecting to Tor v2 nodes.

laanwj merged this on Jun 3, 2021

laanwj closed this on Jun 3, 2021

jonatack deleted the branch on Jun 3, 2021

sidhujag referenced this in commit 31269453d9 on Jun 3, 2021

in src/netaddress.cpp:82 in 5d82a57db4

70-            m_net = NET_ONION;
71-            return true;
72-        }
73-        throw std::ios_base::failure(
74-            strprintf("BIP155 TORv2 address with length %u (should be %u)", address_size,
75-                      ADDR_TORV2_SIZE));

vasild commented at 9:31 am on June 4, 2021:

(can be ignored)

This change in CNetAddr::SetNetFromBIP155Network() will now accept and ignore a Tor v2 addresses with bogus length, e.g. 13. whereas we know it should be exactly 10. The comment near the bottom of this function: // Don't throw on addresses with unknown network ids (maybe from the future). is somewhat incorrect now.

jonatack commented at 10:25 am on June 19, 2021:

Thanks, I’ll have a look.

vasild commented at 11:08 am on June 7, 2021: member

ACK 5d82a57db4f67506a4e80d186ba76f3a8665e147

Some followups in #22179.

laanwj referenced this in commit 4bff49bbb1 on Jun 12, 2021

laanwj referenced this in commit 0de84b7328 on Sep 16, 2021

gwillen referenced this in commit 97ea1dabe4 on Jun 1, 2022

DrahtBot locked this on Aug 18, 2022

p2p: remove tor v2 support #22050

Conflicts

Before (on master be1c512437f7aa8215d2e05eda5cc8b58c3d3937):

After (on torv2-remove-support ef34bda6c52d7a59be9cd3451bc3536430410e36):