net: allow CSubNet of non-IP networks #20852

pull vasild wants to merge 2 commits into bitcoin:master from vasild:non_ip_subnets changing 4 files +123 −19
  1. vasild commented at 5:50 pm on January 4, 2021: member

    Allow creation of valid CSubNet objects of non-IP networks and only match the single address they were created from (like /32 for IPv4 or /128 for IPv6).

    This fixes a deficiency in CConnman::DisconnectNode(const CNetAddr& addr) and in BanMan which assume that creating a subnet from any address using the CSubNet(CNetAddr) constructor would later match that address only. Before this change a non-IP subnet would be invalid and would not match any address.

  2. vasild commented at 5:51 pm on January 4, 2021: member
    If this is merged, then #20849 will not be needed. That PR is valid with or without this one, but this one fixes the issue more broadly.
  3. DrahtBot added the label P2P on Jan 4, 2021
  4. in src/netaddress.cpp:1152 in 06af169cff outdated
    1156+    case NET_CJDNS:
    1157+    case NET_INTERNAL:
    1158+    case NET_UNROUTABLE:
    1159+    case NET_MAX:
    1160+        return network.ToString();
    1161+    }
    


    sipa commented at 6:01 pm on January 4, 2021:
    Control reaches end of non-void function.

    vasild commented at 11:11 am on January 5, 2021:
    Fixed.
  5. sipa added this to the milestone 0.21.0rc5 on Jan 5, 2021
  6. sipa added the label Needs backport (0.21) on Jan 5, 2021
  7. sipa commented at 0:50 am on January 5, 2021: member
    I’ve given this label “Needs backport (0.21)”, as either this or #20849 (or both) should go into rc5. If 20849 is merged and backported, we’re good.
  8. sipa commented at 0:52 am on January 5, 2021: member

    Test is broken, fix:

     0diff --git a/src/test/netbase_tests.cpp b/src/test/netbase_tests.cpp
     1index ac4db3a5b6..87d8bf6ef6 100644
     2--- a/src/test/netbase_tests.cpp
     3+++ b/src/test/netbase_tests.cpp
     4@@ -442,8 +442,7 @@ BOOST_AUTO_TEST_CASE(netbase_dont_resolve_strings_with_embedded_nul_characters)
     5     BOOST_CHECK(!LookupSubNet("1.2.3.0/24\0"s, ret));
     6     BOOST_CHECK(!LookupSubNet("1.2.3.0/24\0example.com"s, ret));
     7     BOOST_CHECK(!LookupSubNet("1.2.3.0/24\0example.com\0"s, ret));
     8-    // We only do subnetting for IPv4 and IPv6
     9-    BOOST_CHECK(!LookupSubNet("5wyqrzbvrdsumnok.onion"s, ret));
    10+    BOOST_CHECK(LookupSubNet("5wyqrzbvrdsumnok.onion"s, ret));
    11     BOOST_CHECK(!LookupSubNet("5wyqrzbvrdsumnok.onion\0"s, ret));
    12     BOOST_CHECK(!LookupSubNet("5wyqrzbvrdsumnok.onion\0example.com"s, ret));
    13     BOOST_CHECK(!LookupSubNet("5wyqrzbvrdsumnok.onion\0example.com\0"s, ret));
    
  9. DrahtBot commented at 5:42 am on January 5, 2021: member

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

    Conflicts

    Reviewers, this pull request conflicts with the following ones:

    • #19825 (rpc: simpler setban and new ban manipulation commands by dhruv)

    If you consider this pull request important, please also help to review the conflicting pull requests. Ideally, start with the one that should be merged first.

  10. laanwj commented at 8:31 am on January 5, 2021: member
    Concept/direction ACK, will review the code later
  11. vasild force-pushed on Jan 5, 2021
  12. vasild commented at 11:11 am on January 5, 2021: member
    06af169cf…6ab8b2b5d: minor tweaks, add comments to CSubNet() constructors, fix and extend tests.
  13. in src/netaddress.h:472 in 6ab8b2b5d4 outdated
    467+         */
    468         CSubNet();
    469+
    470+        /**
    471+         * Construct from a given network start and number of bits (CIDR mask).
    472+         * @param[in] addr Network start. If not IPv4 or IPv6, then a signle-host subnet is created.
    


    jonatack commented at 11:26 am on January 5, 2021:
    s/signle/single/ here and line 480

    vasild commented at 4:12 pm on January 6, 2021:
    Fixed.
  14. in src/netaddress.h:485 in 6ab8b2b5d4 outdated
    478+        /**
    479+         * Construct from a given network start and mask.
    480+         * @param[in] addr Network start. If not IPv4 or IPv6, then a signle-host subnet is created.
    481+         * @param[in] mask Network mask, must be of the same type as `addr` and not contain 0-bits
    482+         * followed by 1-bits. Otherwise an invalid subnet is created. Ignored for non-IP `addr`.
    483+         */
    


    jonatack commented at 11:29 am on January 5, 2021:

    readability suggestion

     0         /**
     1          * Construct from a given network start and number of bits (CIDR mask).
     2-         * [@param](/bitcoin-bitcoin/contributor/param/)[in] addr Network start. If not IPv4 or IPv6, then a signle-host subnet is created.
     3-         * [@param](/bitcoin-bitcoin/contributor/param/)[in] mask CIDR mask, must be in [0, 32] for IPv4 addresses and in [0, 128] for IPv6
     4-         * addresses. Otherwise an invalid subnet is created. Ignored for non-IP `addr`.
     5+         * [@param](/bitcoin-bitcoin/contributor/param/)[in] addr  Network start. If not IPv4 or IPv6, then a single-host subnet is created.
     6+         * [@param](/bitcoin-bitcoin/contributor/param/)[in] mask  CIDR mask, must be in [0, 32] for IPv4 addresses and in [0, 128] for IPv6
     7+         *                  addresses. Otherwise an invalid subnet is created. Ignored for non-IP `addr`.
     8          */
     9         CSubNet(const CNetAddr& addr, uint8_t mask);
    10 
    11         /**
    12          * Construct from a given network start and mask.
    13-         * [@param](/bitcoin-bitcoin/contributor/param/)[in] addr Network start. If not IPv4 or IPv6, then a signle-host subnet is created.
    14-         * [@param](/bitcoin-bitcoin/contributor/param/)[in] mask Network mask, must be of the same type as `addr` and not contain 0-bits
    15-         * followed by 1-bits. Otherwise an invalid subnet is created. Ignored for non-IP `addr`.
    16+         * [@param](/bitcoin-bitcoin/contributor/param/)[in] addr  Network start. If not IPv4 or IPv6, then a single-host subnet is created.
    17+         * [@param](/bitcoin-bitcoin/contributor/param/)[in] mask  Network mask, must be of the same type as `addr` and not contain 0-bits
    18+         *                  followed by 1-bits. Otherwise an invalid subnet is created. Ignored for non-IP `addr`.
    19          */
    20         CSubNet(const CNetAddr& addr, const CNetAddr& mask);
    

    vasild commented at 4:13 pm on January 6, 2021:
    This 3-column-layout has a maintainability problem when a parameter with longer name is added or existent one renamed.

    jonatack commented at 11:14 am on January 7, 2021:
    Yes, though I think it’s worth it. Anyway, just a suggestion.
  15. in src/netaddress.h:488 in 6ab8b2b5d4 outdated
    484         CSubNet(const CNetAddr& addr, const CNetAddr& mask);
    485 
    486-        //constructor for single ip subnet (<ipv4>/32 or <ipv6>/128)
    487+        /**
    488+         * Construct a single-host subnet.
    489+         * @param[in] addr the sole address to be contained in the subnet.
    


    jonatack commented at 11:29 am on January 5, 2021:

    readability suggestion

    0         * [@param](/bitcoin-bitcoin/contributor/param/)[in] addr  The sole address to be contained in the subnet.
    

    vasild commented at 4:15 pm on January 6, 2021:
    Took only the capital, without the space :)
  16. MarcoFalke commented at 11:41 am on January 5, 2021: member
     0 test  2021-01-05T11:36:34.542000Z TestFramework (ERROR): JSONRPC error 
     1                                   Traceback (most recent call last):
     2                                     File "/private/var/folders/3y/l0z1x3693dl_8n0qybp4dqwh0000gn/T/cirrus-ci-build/ci/scratch/build/bitcoin-x86_64-apple-darwin18/test/functional/test_framework/test_framework.py", line 126, in main
     3                                       self.run_test()
     4                                     File "/private/var/folders/3y/l0z1x3693dl_8n0qybp4dqwh0000gn/T/cirrus-ci-build/ci/scratch/build/bitcoin-x86_64-apple-darwin18/test/functional/rpc_setban.py", line 39, in run_test
     5                                       self.nodes[1].setban("127.0.0.1", "remove")
     6                                     File "/private/var/folders/3y/l0z1x3693dl_8n0qybp4dqwh0000gn/T/cirrus-ci-build/ci/scratch/build/bitcoin-x86_64-apple-darwin18/test/functional/test_framework/coverage.py", line 47, in __call__
     7                                       return_val = self.auth_service_proxy_instance.__call__(*args, **kwargs)
     8                                     File "/private/var/folders/3y/l0z1x3693dl_8n0qybp4dqwh0000gn/T/cirrus-ci-build/ci/scratch/build/bitcoin-x86_64-apple-darwin18/test/functional/test_framework/authproxy.py", line 146, in __call__
     9                                       raise JSONRPCException(response['error'], status)
    10                                   test_framework.authproxy.JSONRPCException: Error: Unban failed. Requested address/subnet was not previously manually banned. (-30)
    
  17. in src/netaddress.cpp:1076 in 6ab8b2b5d4 outdated
    1118+    case NET_UNROUTABLE:
    1119+    case NET_MAX:
    1120         return;
    1121     }
    1122 
    1123-    assert(addr.m_addr.size() <= sizeof(netmask));
    


    jonatack commented at 11:42 am on January 5, 2021:

    Should we keep this assert?

    0     case NET_IPV4:
    1     case NET_IPV6:
    2         valid = true;
    3+        assert(addr.m_addr.size() <= sizeof(netmask));
    4         memset(netmask, 0xFF, sizeof(netmask));
    5         break;
    6     case NET_ONION:
    

    vasild commented at 4:19 pm on January 6, 2021:

    The assert was there because in master we wrote addr.m_addr.size() bytes and here it is changed to write sizeof(netmask) bytes so it can’t overflow and thus I removed the assert.

    However, I changed/restored it to write addr.m_addr.size() bytes and also restored the assert, explanation.

  18. MarcoFalke commented at 11:48 am on January 5, 2021: member
    I am not familiar with this part of the codebase, but it seems safer to move this to 0.21.1. This claims to fix a bug how outbound onion addresses are handled by rpc and misbehavior-disconnect. However, it changes the code paths for all networks and all net handling. So it seems a bit risky to push out this late in the rc stage. (There haven’t been any major issues since rc3 other than the macos code signing, so rc5 might very well be a good -final). It could be documented in the release notes, if needed, and a 0.21.1 can be pushed out once this has settled in the master branch (maybe after ~1-2 months)?
  19. jonatack commented at 11:51 am on January 5, 2021: member

    Approach ACK; this works for me manually with disconnectnode on outbound onion peers (edit: but so does master, the issue isn’t hit with it, will look at setban <onion> add) and the netbase_tests run locally, but seeing CI issues.

    It would be good to have a failing test case that this patch fixes.

  20. vasild force-pushed on Jan 6, 2021
  21. vasild commented at 4:10 pm on January 6, 2021: member

    6ab8b2b5d…826b8d9eb: fix failing rpc_setban.py by making sure in-memory representation of CSubNet::netmask[] for IPv4 subnets is the same when deserialized from disk and when created via the constructor.

    For IPv4, deserialize only sets the first 4 bytes of the netmask, leaving the remaining 12 bytes to the default 0. operator== compares all the 16 bytes, so we want a deserialized subnet to equal the same subnet that is created via the CSubNet::CSubNet(CNetAddr) constructor. Thus, in that constructor only set the first 4 bytes and leave the remaining 12 ones to the default 0.

    Yes, operator== need not compare all the 16 bytes for IPv4, it can/should compare just the first 4. Fixing this by modifying operator== I think goes beyond the purpose of this PR which is to allow non-IP subnets. So I restored the code to only set the first 4 bytes (+ the assert to ensure no overflow). This is the same as in master.

    Eventually uint8_t CSubNet::netmask[16] should be changed to uint8_t CSubNet::m_cidr.

  22. vasild force-pushed on Jan 6, 2021
  23. vasild commented at 4:19 pm on January 6, 2021: member
    826b8d9eb…293070adc: fix typos in comments
  24. vasild commented at 4:26 pm on January 6, 2021: member

    @MarcoFalke

    … it changes the code paths for all networks and all net handling …

    Indeed. CSubNet is used in lots of places and changes to it can have subtle effects, as shown by CI failures on this PR.

  25. in src/netaddress.cpp:1009 in 293070adc3 outdated
    1006+        valid = mask <= ADDR_IPV6_SIZE * 8;
    1007+        break;
    1008+    case NET_ONION:
    1009+    case NET_I2P:
    1010+    case NET_CJDNS:
    1011+        valid = true;
    


    luke-jr commented at 5:14 pm on January 6, 2021:
    We shouldn’t just ignore mask… Might be best to only change the single-address form?

    vasild commented at 12:13 pm on January 7, 2021:
  26. laanwj removed this from the milestone 0.21.0rc5 on Jan 7, 2021
  27. laanwj added this to the milestone 0.21.0 on Jan 7, 2021
  28. jonatack commented at 11:28 am on January 7, 2021: member
    ACK 293070adc34f979bee1e9315cc70973a168c3af2 code review, manually verified that setban OUTBOUND_TOR_V3_PEER add succeeds with this patch and fails on master, cherry-picked the unit test change in 98140bc855b037d2c552153496e8e4e2c2efdad9 that fails on master, and it passes with this patch
  29. vasild force-pushed on Jan 7, 2021
  30. vasild commented at 12:13 pm on January 7, 2021: member
    293070adc…1a54bd7d0: as per suggestion, reverted the two constructors that take also a “netmask” argument to create an invalid subnet, like in master without this PR.
  31. in src/netaddress.cpp:1086 in 1a54bd7d00 outdated
    1083+        valid = true;
    1084+        break;
    1085+    case NET_INTERNAL:
    1086+    case NET_UNROUTABLE:
    1087+    case NET_MAX:
    1088         return;
    


    laanwj commented at 2:13 pm on January 7, 2021:
    I think valid is left uninitialized here. It’s not initialized in class CSubNet declaration nor here.

    vasild commented at 2:58 pm on January 7, 2021:

    It is set to false in the default constructor which is called by this constructor:

    CSubNet::CSubNet(const CNetAddr& addr) : CSubNet()

    netmask is also set there.


    laanwj commented at 9:39 am on January 8, 2021:
    Thanks for illuminating. That’s kinda sneaky, would prefer to do initialization in the class declaration itself, but no need to do so in this PR.

    jonatack commented at 4:12 pm on January 8, 2021:
    Maybe add valid = false; here as belt and suspenders.
  32. vasild force-pushed on Jan 7, 2021
  33. vasild commented at 2:59 pm on January 7, 2021: member
    1a54bd7d0…5e95ce679: add a functional test that fails without this fix and passes with it
  34. in test/functional/rpc_setban.py:53 in 5e95ce6797 outdated
    48+        for ban_entry in self.nodes[1].listbanned():
    49+            if ban_entry['address'] == tor_addr:
    50+                is_banned = True
    51+                break
    52+        assert(is_banned)
    53+        self.nodes[1].setban(tor_addr, "remove")
    


    jonatack commented at 3:52 pm on January 8, 2021:

    The test sees only one entry in listbanned, the added one. If it is moved above the previous test, there will also be a second entry in listbanned, e.g.

    0[
    1  {'address': '127.0.0.1/32', ...},
    2  {'address': 'pg6mmjiyjmcrsslvykfwnntlaru7p5svn6y2ymmju6nubxndf4pscryd.onion', ...}
    3]
    

    Here’s a suggestion to do that, add logging, and also assert pre-add and post-remove (with the same number of lines):

     0+TOR_ADDR = "pg6mmjiyjmcrsslvykfwnntlaru7p5svn6y2ymmju6nubxndf4pscryd.onion"
     1+
     2 class SetBanTests(BitcoinTestFramework):
     3 
     4+    def banned_tor_addr(self, *, tor_addr=TOR_ADDR):
     5+        return list(filter(lambda ban_entry: ban_entry["address"] == tor_addr, self.nodes[1].listbanned()))
     6+
     7     def run_test(self):
     8@@ -35,6 +40,13 @@ class SetBanTests(BitcoinTestFramework):
     9         peerinfo = self.nodes[1].getpeerinfo()[0]
    10         assert('noban' in peerinfo['permissions'])
    11 
    12+        self.log.info("Test that a Tor v3 address can be banned/unbanned")
    13+        assert(not self.banned_tor_addr())
    14+        self.nodes[1].setban(TOR_ADDR, "add")
    15+        assert(self.banned_tor_addr())
    16+        self.nodes[1].setban(TOR_ADDR, "remove")
    17+        assert(not self.banned_tor_addr())
    18+
    19         # If we remove the ban, Node 0 should be able to reconnect even without noban permission
    20@@ -42,15 +54,5 @@ class SetBanTests(BitcoinTestFramework):
    21         assert(not 'noban' in peerinfo['permissions'])
    22 
    23-        tor_addr = "pg6mmjiyjmcrsslvykfwnntlaru7p5svn6y2ymmju6nubxndf4pscryd.onion"
    24-        self.nodes[1].setban(tor_addr, "add")
    25-        is_banned = False
    26-        for ban_entry in self.nodes[1].listbanned():
    27-            if ban_entry['address'] == tor_addr:
    28-                is_banned = True
    29-                break
    30-        assert(is_banned)
    31-        self.nodes[1].setban(tor_addr, "remove")
    32
    33 if __name__ == '__main__':
    

    vasild commented at 2:54 pm on January 10, 2021:
    Added something like that in 39b43298d, thanks!
  35. jonatack commented at 4:18 pm on January 8, 2021: member

    ACK 5e95ce6797a286aa23022141188f9139315bff70 code review, manually verified that setban OUTBOUND_TOR_V3_PEER add succeeds with this patch and fails on master, cherry-picked the unit test change in 98140bc855b037d2c552153496e8e4e2c2efdad9 that fails on master, and it passes with this patch.

    Happy to re-review if you update per comments above.

  36. net: allow CSubNet of non-IP networks
    Allow creation of valid `CSubNet` objects of non-IP networks and only
    match the single address they were created from (like /32 for IPv4 or
    /128 for IPv6).
    
    This fixes a deficiency in `CConnman::DisconnectNode(const CNetAddr& addr)`
    and in `BanMan` which assume that creating a subnet from any address
    using the `CSubNet(CNetAddr)` constructor would later match that address
    only. Before this change a non-IP subnet would be invalid and would not
    match any address.
    94d335da7f
  37. test: add test for banning of non-IP addresses
    Co-authored-by: Jon Atack <jon@atack.com>
    39b43298d9
  38. vasild force-pushed on Jan 10, 2021
  39. vasild commented at 2:55 pm on January 10, 2021: member
    5e95ce679…39b43298d: extend the functional test, #20852 (review)
  40. jonatack commented at 9:43 am on January 11, 2021: member
    Code review re-ACK 39b43298d9c54f9c18bef36f3d5934f57aefd088 per git diff 5e95ce6 39b4329; only change since last review is improvements to the functional test; verified the test fails on master @ 616eace0 where expected (assert(self.is_banned(node, tor_addr)) fails and unban unfails)
  41. MarcoFalke removed this from the milestone 0.21.0 on Jan 11, 2021
  42. MarcoFalke added this to the milestone 0.21.1 on Jan 11, 2021
  43. MarcoFalke commented at 10:15 am on January 11, 2021: member
    Moved to 0.21.1 milestone according to last weeks meeting. Let me know if this is not applicable.
  44. laanwj commented at 10:25 am on January 11, 2021: member

    code review ACK 39b43298d9c54f9c18bef36f3d5934f57aefd088

    Moved to 0.21.1 milestone according to last weeks meeting. Let me know if this is not applicable.

    Agreed.

  45. laanwj merged this on Jan 11, 2021
  46. laanwj closed this on Jan 11, 2021

  47. MarcoFalke referenced this in commit c33fbab25c on Jan 11, 2021
  48. MarcoFalke referenced this in commit bdce029191 on Jan 11, 2021
  49. MarcoFalke commented at 11:49 am on January 11, 2021: member

    Would be nice to summarize the changes that are expected here. Am I right that this …

    • Breaks compatibility of the serialized banfile? I.e. opening a banfile written by this version with a previous version may drop some addresses?
    • Fixes banning of onion outbounds via rpc.
    • Fixes misbheavior disconnect of onion outbounds.

    I tried testing the changes and restarting the node will corrupt the listbanned output:

    0./src/bitcoin-cli -signet  listbanned
    1[
    2  {
    3    "address": "::/0",
    4    "banned_until": 1610451878,
    5    "ban_created": 1610365478
    6  }
    7]
    8 
    
  50. MarcoFalke commented at 11:58 am on January 11, 2021: member

    Would probably be good to propose some release notes based on the changes. Added a draft here: https://github.com/bitcoin-core/bitcoin-devwiki/wiki/%2320852-Release-notes-snippet

  51. MarcoFalke removed the label Needs backport (0.21) on Jan 11, 2021
  52. vasild commented at 12:49 pm on January 11, 2021: member

    Would be nice to summarize the changes that are expected here. Am I right that this …

    * Breaks compatibility of the serialized banfile? I.e. opening a banfile written by this version with a previous version may drop some addresses?
    

    The serialized format has not been changed. If this version writes a ban entry consisting of the newly allowed tor-one-host-subnet, an old version would deserialize that as an invalid subnet:

    https://github.com/bitcoin/bitcoin/blob/5cce60710525e16cac30a855e5cf1d12f429b09b/src/netaddress.h#L518

    old code, before this PR: https://github.com/bitcoin/bitcoin/blob/616eace02acdc5f9ce775f6bf82fc8964980162f/src/netaddress.cpp#L1121-L1123

    * Fixes banning of onion outbounds via rpc.
    

    Yes.

    * Fixes misbheavior disconnect of onion outbounds.
    

    Yes.

    I tried testing the changes and restarting the node will corrupt the listbanned output:

    This looks strange, how to reproduce?

  53. vasild deleted the branch on Jan 11, 2021
  54. MarcoFalke commented at 12:56 pm on January 11, 2021: member

    This looks strange, how to reproduce?

    setban foo.onion add, listbanned (correct), restart, listbanned (corrupt)

  55. MarcoFalke commented at 2:04 pm on January 11, 2021: member

    Steps to reproduce:

     0# ./src/bitcoind --version && ./src/bitcoind  -signet  -printtoconsole=0 -noconnect & 
     1Bitcoin Core version v21.99.0-5cce60710
     2
     3# ./src/bitcoin-cli  -signet listbanned
     4[
     5  {
     6    "address": "::/0",
     7    "banned_until": 1610459838,
     8    "ban_created": 1610373438
     9  }
    10]
    11
    12# ./src/bitcoin-cli  -signet setban v7ajjeirttkbnt32wpy3c6w3emwnfr3fkla7hpxcfokr3ysd3kqtzmqd.onion add
    13# ./src/bitcoin-cli  -signet setban fij5zbz4dui7tqolfmkvrj6uag7axoliif4dywhj5ghndnusxzlho2yd.onion add
    14# ./src/bitcoin-cli  -signet listbanned
    15[
    16  {
    17    "address": "::/0",
    18    "banned_until": 1610459838,
    19    "ban_created": 1610373438
    20  },
    21  {
    22    "address": "fij5zbz4dui7tqolfmkvrj6uag7axoliif4dywhj5ghndnusxzlho2yd.onion",
    23    "banned_until": 1610460063,
    24    "ban_created": 1610373663
    25  },
    26  {
    27    "address": "v7ajjeirttkbnt32wpy3c6w3emwnfr3fkla7hpxcfokr3ysd3kqtzmqd.onion",
    28    "banned_until": 1610460060,
    29    "ban_created": 1610373660
    30  }
    31]
    32
    33
    34# ./src/bitcoin-cli  -signet stop
    35Bitcoin Core stopping
    36
    37
    38# ./src/bitcoind  -signet  -printtoconsole=0 -noconnect &
    39
    40
    41# ./src/bitcoin-cli  -signet listbanned
    42[
    43  {
    44    "address": "::/0",
    45    "banned_until": 1610459838,
    46    "ban_created": 1610373438
    47  }
    48]
    
  56. sidhujag referenced this in commit 490b302b84 on Jan 11, 2021
  57. vasild referenced this in commit 10b3e429a4 on Jan 11, 2021
  58. vasild commented at 4:43 pm on January 11, 2021: member

    @MarcoFalke the problem you describe above is fixed in #20904, but that requires a banlist.dat format change.

    Maybe revert this PR and merge it together with #20904 as one PR?

  59. vasild referenced this in commit 07aefe77ea on Jan 13, 2021
  60. vasild referenced this in commit e61023506e on Jan 13, 2021
  61. MarcoFalke commented at 7:45 am on June 23, 2021: member
    Looks like this was never backported? I guess not worth it at this point.
  62. knst referenced this in commit c6eeee341d on Aug 6, 2022
  63. knst referenced this in commit c5ea4d1a0e on Aug 6, 2022
  64. DrahtBot locked this on Aug 16, 2022

github-metadata-mirror

This is a metadata mirror of the GitHub repository bitcoin/bitcoin. This site is not affiliated with GitHub. Content is generated from a GitHub metadata backup.
generated: 2024-07-05 22:12 UTC

This site is hosted by @0xB10C
More mirrored repositories can be found on mirror.b10c.me