net: Add interrupt to pcp retry loop #33338

TheCharlatan commented at 10:20 AM on September 8, 2025: contributor

Without this interrupt bitcoind takes a long time to exit if requested to do so after a failed pcp lookup on startup.

net: Add interrupt to pcp retry loop

Without this interrupt bitcoind takes a long time to exit if requested
to do so after a failed pcp lookup on startup.

188de70c86

DrahtBot added the label P2P on Sep 8, 2025

DrahtBot commented at 10:20 AM on September 8, 2025: contributor

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

Code Coverage & Benchmarks

For details see: https://corecheck.dev/bitcoin/bitcoin/pulls/33338.

Reviews

See the guideline for information on the review process.

Type	Reviewers
ACK	fjahr, hodlinator, achow101

If your review is incorrectly listed, please react with 👎 to this comment and the bot will ignore it on the next update.

TheCharlatan commented at 10:21 AM on September 8, 2025: contributor

Logs:

<details> <summary>Master logs when mapping fails (Click to expand)</summary>

2025-09-08T10:02:52Z msghand thread start
2025-09-08T10:02:52Z Imported mempool transactions from file: 3 succeeded, 0 failed, 0 expired, 0 already there, 0 waiting for initial broadcast
2025-09-08T10:02:52Z initload thread exit
2025-09-08T10:02:52Z [net:warning] pcp: Could not receive response: Connection refused (111)
2025-09-08T10:02:52Z [net] portmap: gateway [IPv6]: fe80::a6ce:daff:fe8f:12a0%4
2025-09-08T10:02:52Z [net] pcp: Requesting port mapping for addr 2a02:1210:4a25:2f00:4bfd:91da:7cba:67cb port 38333 from gateway fe80::a6ce:daff:fe8f:12a0%4
2025-09-08T10:02:52Z [net] pcp: Internal address after connect: 2a02:1210:4a25:2f00:4bfd:91da:7cba:67cb
2025-09-08T10:02:53Z [net] trying v2 connection [2001:e60:1154:ccd9:3515:d56c:bb7:5f4c]:38333 lastseen=58.3hrs
2025-09-08T10:02:53Z [net] pcp: Timeout
2025-09-08T10:02:53Z [net] pcp: Retrying (1)
2025-09-08T10:02:54Z [net] pcp: Timeout
2025-09-08T10:02:54Z [net] pcp: Retrying (2)
^C2025-09-08T10:02:54Z tor: Thread interrupt
2025-09-08T10:02:54Z Shutdown in progress...
2025-09-08T10:02:54Z torcontrol thread exit
2025-09-08T10:02:54Z dnsseed thread exit
2025-09-08T10:02:54Z addcon thread exit
2025-09-08T10:02:54Z net thread exit
2025-09-08T10:02:55Z msghand thread exit
2025-09-08T10:02:55Z [net] pcp: Timeout
2025-09-08T10:02:55Z [net] pcp: Giving up after 3 tries
2025-09-08T10:02:55Z [net] pcp: Requesting port mapping for addr 2a02:1210:4a25:2f00:1386:bc1d:7d89:306d port 38333 from gateway fe80::a6ce:daff:fe8f:12a0%4
2025-09-08T10:02:55Z [net] pcp: Internal address after connect: 2a02:1210:4a25:2f00:1386:bc1d:7d89:306d
2025-09-08T10:02:56Z [net] pcp: Timeout
2025-09-08T10:02:56Z [net] pcp: Retrying (1)
2025-09-08T10:02:57Z [net] pcp: Timeout
2025-09-08T10:02:57Z [net] pcp: Retrying (2)
2025-09-08T10:02:58Z [net] connection attempt to [2001:e60:1154:ccd9:3515:d56c:bb7:5f4c]:38333 timed out
2025-09-08T10:02:58Z opencon thread exit
2025-09-08T10:02:58Z [net] pcp: Timeout
2025-09-08T10:02:58Z [net] pcp: Giving up after 3 tries
2025-09-08T10:02:58Z [net] pcp: Requesting port mapping for addr 2a02:1210:4a25:2f00:17ff:1c49:2591:5195 port 38333 from gateway fe80::a6ce:daff:fe8f:12a0%4
2025-09-08T10:02:58Z [net] pcp: Internal address after connect: 2a02:1210:4a25:2f00:17ff:1c49:2591:5195
2025-09-08T10:02:59Z [net] pcp: Timeout
2025-09-08T10:02:59Z [net] pcp: Retrying (1)
2025-09-08T10:03:00Z [net] pcp: Timeout
2025-09-08T10:03:00Z [net] pcp: Retrying (2)
2025-09-08T10:03:01Z [net] pcp: Timeout
2025-09-08T10:03:01Z [net] pcp: Giving up after 3 tries
2025-09-08T10:03:01Z [net] pcp: Requesting port mapping for addr 2a02:1210:4a25:2f00:5e43:d231:cd0b:5424 port 38333 from gateway fe80::a6ce:daff:fe8f:12a0%4
2025-09-08T10:03:01Z [net] pcp: Internal address after connect: 2a02:1210:4a25:2f00:5e43:d231:cd0b:5424
2025-09-08T10:03:02Z [net] pcp: Timeout
2025-09-08T10:03:02Z [net] pcp: Retrying (1)
2025-09-08T10:03:03Z [net] pcp: Timeout
2025-09-08T10:03:03Z [net] pcp: Retrying (2)
2025-09-08T10:03:04Z [net] pcp: Timeout
2025-09-08T10:03:04Z [net] pcp: Giving up after 3 tries
2025-09-08T10:03:04Z [net] pcp: Requesting port mapping for addr 2a02:1210:4a25:2f00:eb22:5bda:fe93:efb3 port 38333 from gateway fe80::a6ce:daff:fe8f:12a0%4
2025-09-08T10:03:04Z [net] pcp: Internal address after connect: 2a02:1210:4a25:2f00:eb22:5bda:fe93:efb3
2025-09-08T10:03:05Z [net] pcp: Timeout
2025-09-08T10:03:05Z [net] pcp: Retrying (1)
2025-09-08T10:03:06Z [net] pcp: Timeout
2025-09-08T10:03:06Z [net] pcp: Retrying (2)
2025-09-08T10:03:07Z [net] pcp: Timeout
2025-09-08T10:03:07Z [net] pcp: Giving up after 3 tries
2025-09-08T10:03:07Z [net] pcp: Requesting port mapping for addr 2a02:1210:4a25:2f00:ad0f:260a:2c63:7494 port 38333 from gateway fe80::a6ce:daff:fe8f:12a0%4
2025-09-08T10:03:07Z [net] pcp: Internal address after connect: 2a02:1210:4a25:2f00:ad0f:260a:2c63:7494
2025-09-08T10:03:08Z [net] pcp: Timeout
2025-09-08T10:03:08Z [net] pcp: Retrying (1)
2025-09-08T10:03:09Z [net] pcp: Timeout
2025-09-08T10:03:09Z [net] pcp: Retrying (2)
2025-09-08T10:03:10Z [net] pcp: Timeout
2025-09-08T10:03:10Z [net] pcp: Giving up after 3 tries
2025-09-08T10:03:10Z [net] pcp: Requesting port mapping for addr 2a02:1210:4a25:2f00:65cf:20cb:5921:45bf port 38333 from gateway fe80::a6ce:daff:fe8f:12a0%4
2025-09-08T10:03:10Z [net] pcp: Internal address after connect: 2a02:1210:4a25:2f00:65cf:20cb:5921:45bf
2025-09-08T10:03:11Z [net] pcp: Timeout
2025-09-08T10:03:11Z [net] pcp: Retrying (1)
2025-09-08T10:03:12Z [net] pcp: Timeout
2025-09-08T10:03:12Z [net] pcp: Retrying (2)
2025-09-08T10:03:13Z [net] pcp: Timeout
2025-09-08T10:03:13Z [net] pcp: Giving up after 3 tries
2025-09-08T10:03:13Z [net] pcp: Requesting port mapping for addr 2a02:1210:4a25:2f00:5f6b:1db9:7ba0:44a port 38333 from gateway fe80::a6ce:daff:fe8f:12a0%4
2025-09-08T10:03:13Z [net] pcp: Internal address after connect: 2a02:1210:4a25:2f00:5f6b:1db9:7ba0:44a
2025-09-08T10:03:14Z [net] pcp: Timeout
2025-09-08T10:03:14Z [net] pcp: Retrying (1)
2025-09-08T10:03:15Z [net] pcp: Timeout
2025-09-08T10:03:15Z [net] pcp: Retrying (2)
2025-09-08T10:03:16Z [net] pcp: Timeout
2025-09-08T10:03:16Z [net] pcp: Giving up after 3 tries
2025-09-08T10:03:16Z [net] pcp: Requesting port mapping for addr 1111:1111::1 port 38333 from gateway fe80::a6ce:daff:fe8f:12a0%4
2025-09-08T10:03:16Z [net] pcp: Internal address after connect: 1111:1111::1
2025-09-08T10:03:17Z [net] pcp: Timeout
2025-09-08T10:03:17Z [net] pcp: Retrying (1)
2025-09-08T10:03:18Z [net] pcp: Timeout
2025-09-08T10:03:18Z [net] pcp: Retrying (2)
2025-09-08T10:03:19Z [net] pcp: Timeout
2025-09-08T10:03:19Z [net] pcp: Giving up after 3 tries
2025-09-08T10:03:19Z mapport thread exit
2025-09-08T10:03:19Z [net] Flushed 13682 addresses to peers.dat  22ms
2025-09-08T10:03:19Z DumpAnchors: Flush 0 outbound block-relay-only peer addresses to anchors.dat started
2025-09-08T10:03:19Z DumpAnchors: Flush 0 outbound block-relay-only peer addresses to anchors.dat completed (0.01s)
2025-09-08T10:03:19Z scheduler thread exit
2025-09-08T10:03:19Z Writing 3 mempool transactions to file...
2025-09-08T10:03:19Z Writing 0 unbroadcast transactions to file.
2025-09-08T10:03:19Z Dumped mempool: 0.000s to copy, 0.008s to dump, 1011 bytes dumped to file
2025-09-08T10:03:19Z Flushed fee estimates to fee_estimates.dat.
2025-09-08T10:03:19Z Shutdown done

</details>

<details> <summary>This PR's logs when mapping fails (Click to expand)</summary>

2025-09-08T10:07:03Z msghand thread start
2025-09-08T10:07:03Z addcon thread start
2025-09-08T10:07:03Z initload thread exit
2025-09-08T10:07:03Z [net:warning] pcp: Could not receive response: Connection refused (111)
2025-09-08T10:07:03Z [net] portmap: gateway [IPv6]: fe80::a6ce:daff:fe8f:12a0%4
2025-09-08T10:07:03Z [net] pcp: Requesting port mapping for addr 2a02:1210:4a25:2f00:4bfd:91da:7cba:67cb port 38333 from gateway fe80::a6ce:daff:fe8f:12a0%4
2025-09-08T10:07:03Z [net] pcp: Internal address after connect: 2a02:1210:4a25:2f00:4bfd:91da:7cba:67cb
^C2025-09-08T10:07:03Z tor: Thread interrupt
2025-09-08T10:07:03Z Shutdown in progress...
2025-09-08T10:07:03Z torcontrol thread exit
2025-09-08T10:07:03Z addcon thread exit
2025-09-08T10:07:03Z opencon thread exit
2025-09-08T10:07:03Z dnsseed thread exit
2025-09-08T10:07:03Z net thread exit
2025-09-08T10:07:03Z msghand thread exit
2025-09-08T10:07:04Z [net] pcp: Timeout
2025-09-08T10:07:04Z [net] pcp: Retrying (1)
2025-09-08T10:07:04Z [net] pcp: Requesting port mapping for addr 2a02:1210:4a25:2f00:1386:bc1d:7d89:306d port 38333 from gateway fe80::a6ce:daff:fe8f:12a0%4
2025-09-08T10:07:04Z [net] pcp: Internal address after connect: 2a02:1210:4a25:2f00:1386:bc1d:7d89:306d
2025-09-08T10:07:04Z [net] pcp: Requesting port mapping for addr 2a02:1210:4a25:2f00:17ff:1c49:2591:5195 port 38333 from gateway fe80::a6ce:daff:fe8f:12a0%4
2025-09-08T10:07:04Z [net] pcp: Internal address after connect: 2a02:1210:4a25:2f00:17ff:1c49:2591:5195
2025-09-08T10:07:04Z [net] pcp: Requesting port mapping for addr 2a02:1210:4a25:2f00:5e43:d231:cd0b:5424 port 38333 from gateway fe80::a6ce:daff:fe8f:12a0%4
2025-09-08T10:07:04Z [net] pcp: Internal address after connect: 2a02:1210:4a25:2f00:5e43:d231:cd0b:5424
2025-09-08T10:07:04Z [net] pcp: Requesting port mapping for addr 2a02:1210:4a25:2f00:eb22:5bda:fe93:efb3 port 38333 from gateway fe80::a6ce:daff:fe8f:12a0%4
2025-09-08T10:07:04Z [net] pcp: Internal address after connect: 2a02:1210:4a25:2f00:eb22:5bda:fe93:efb3
2025-09-08T10:07:04Z [net] pcp: Requesting port mapping for addr 2a02:1210:4a25:2f00:ad0f:260a:2c63:7494 port 38333 from gateway fe80::a6ce:daff:fe8f:12a0%4
2025-09-08T10:07:04Z [net] pcp: Internal address after connect: 2a02:1210:4a25:2f00:ad0f:260a:2c63:7494
2025-09-08T10:07:04Z [net] pcp: Requesting port mapping for addr 2a02:1210:4a25:2f00:65cf:20cb:5921:45bf port 38333 from gateway fe80::a6ce:daff:fe8f:12a0%4
2025-09-08T10:07:04Z [net] pcp: Internal address after connect: 2a02:1210:4a25:2f00:65cf:20cb:5921:45bf
2025-09-08T10:07:04Z [net] pcp: Requesting port mapping for addr 2a02:1210:4a25:2f00:5f6b:1db9:7ba0:44a port 38333 from gateway fe80::a6ce:daff:fe8f:12a0%4
2025-09-08T10:07:04Z [net] pcp: Internal address after connect: 2a02:1210:4a25:2f00:5f6b:1db9:7ba0:44a
2025-09-08T10:07:04Z [net] pcp: Requesting port mapping for addr 1111:1111::1 port 38333 from gateway fe80::a6ce:daff:fe8f:12a0%4
2025-09-08T10:07:04Z [net] pcp: Internal address after connect: 1111:1111::1
2025-09-08T10:07:04Z mapport thread exit
2025-09-08T10:07:04Z [net] Flushed 13692 addresses to peers.dat  20ms
2025-09-08T10:07:04Z DumpAnchors: Flush 0 outbound block-relay-only peer addresses to anchors.dat started
2025-09-08T10:07:04Z DumpAnchors: Flush 0 outbound block-relay-only peer addresses to anchors.dat completed (0.01s)
2025-09-08T10:07:04Z scheduler thread exit
2025-09-08T10:07:04Z Writing 3 mempool transactions to file...
2025-09-08T10:07:04Z Writing 0 unbroadcast transactions to file.
2025-09-08T10:07:04Z Dumped mempool: 0.000s to copy, 0.006s to dump, 1011 bytes dumped to file
2025-09-08T10:07:04Z Flushed fee estimates to fee_estimates.dat.
2025-09-08T10:07:04Z Shutdown done

</details>

fanquake added this to the milestone 30.0 on Sep 8, 2025

in src/common/pcp.cpp:222 in 188de70c86

 217 | @@ -217,7 +218,8 @@ CNetAddr PCPUnwrapAddress(std::span<const uint8_t> wrapped_addr)
 218 |  //! PCP or NAT-PMP send-receive loop.
 219 |  std::optional<std::vector<uint8_t>> PCPSendRecv(Sock &sock, const std::string &protocol, std::span<const uint8_t> request, int num_tries,
 220 |          std::chrono::milliseconds timeout_per_try,
 221 | -        std::function<bool(std::span<const uint8_t>)> check_packet)
 222 | +        std::function<bool(std::span<const uint8_t>)> check_packet,
 223 | +        CThreadInterrupt& interrupt)

hodlinator commented at 1:04 PM on September 8, 2025:

nit: Would prefer keeping the lambda as the last parameter and also avoiding to mix &-placement within the same argument list.

        CThreadInterrupt &interrupt,
        std::function<bool(std::span<const uint8_t>)> check_packet)

fjahr commented at 7:29 PM on September 8, 2025:

CThreadInterrupt& interrupt is our preferred style though (see PointerAlignment: Left in src/.clang-format), so maybe we rather want to fix the other parameter to keep it consistent.

in src/common/pcp.cpp:243 in 188de70c86

 239 | @@ -238,6 +240,7 @@ std::optional<std::vector<uint8_t>> PCPSendRecv(Sock &sock, const std::string &p
 240 |          auto cur_time = time_point_cast<milliseconds>(MockableSteadyClock::now());
 241 |          auto deadline = cur_time + timeout_per_try;
 242 |          while ((cur_time = time_point_cast<milliseconds>(MockableSteadyClock::now())) < deadline) {
 243 | +            if (interrupt) return std::nullopt;

fjahr commented at 2:51 PM on September 8, 2025:

nit: comment above lists all the conditions for the wait to exit so it should list interrupt as well.

in src/test/pcp_tests.cpp:18 in 188de70c86

  14 | @@ -15,6 +15,8 @@
  15 |  
  16 |  using namespace std::literals;
  17 |  
  18 | +static CThreadInterrupt g_interrupt;

fjahr commented at 2:52 PM on September 8, 2025:

nit: Could add a simple test where the interrupt is instantly called to make sure the line is covered somehow.

fjahr commented at 2:54 PM on September 8, 2025: contributor

utACK 188de70c86414b8b2ad5143f5c607b67686526ea

Looks good to me, happy to re-review if you decide to address the nits.

hodlinator approved

hodlinator commented at 7:15 PM on September 8, 2025: contributor

utACK 188de70c86414b8b2ad5143f5c607b67686526ea

Change just passes down interrupt for checking in inner loops.

Didn't figure out a way of reproducing the behavior PR-author describes in #33338 (comment).

Wasn't able to bring to (zombie) life a local IPv6 gateway.
Only have one ethernet interface on my local machine. Seems the author may have a handful.

Reproduced something similar anyway through hacking this in:

diff --git a/src/common/pcp.cpp b/src/common/pcp.cpp
index ed6624cb89..33b4c653fa 100644
--- a/src/common/pcp.cpp
+++ b/src/common/pcp.cpp
@@ -246,9 +246,11 @@ std::optional<std::vector<uint8_t>> PCPSendRecv(Sock &sock, const std::string &p
                 LogPrintLevel(BCLog::NET, BCLog::Level::Warning, "%s: Could not wait on socket: %s\n", protocol, NetworkErrorString(WSAGetLastError()));
                 return std::nullopt; // Network-level error, probably no use retrying.
             }
-            if (!occurred) {
+            std::this_thread::sleep_for(3s);
+            //if (!occurred) {
                 LogPrintLevel(BCLog::NET, BCLog::Level::Debug, "%s: Timeout\n", protocol);
                 break; // Retry.
+            /*
             }

             // Receive response.
@@ -263,6 +265,7 @@ std::optional<std::vector<uint8_t>> PCPSendRecv(Sock &sock, const std::string &p
                 got_response = true; // Got expected response, break from receive loop as well as from retry loop.
                 break;
             }
+            */
         }
     }
     if (!got_response) {

...and testing Ctrl+C with/without interrupt/PR.

achow101 commented at 8:29 PM on September 8, 2025: member

ACK 188de70c86414b8b2ad5143f5c607b67686526ea

achow101 merged this on Sep 8, 2025

achow101 closed this on Sep 8, 2025