highperfocused/bitcoin
1
0
Fork 0
mirror of https://github.com/bitcoin/bitcoin.git synced 2026-05-31 08:13:52 +02:00
Code Issues Packages Projects Releases Wiki Activity

Merge bitcoin/bitcoin#30413: p2p: Lazy init some bloom filters; fuzz version handshake

Browse Source 
afd237bb5d [fuzz] Harness for version handshake (dergoegge)
a90ab4aec9 scripted-diff: Rename lazily initialized bloom filters (dergoegge)
82de1bc478 [net processing] Lazily initialize m_recent_confirmed_transactions (dergoegge)
fa0c87f19c [net processing] Lazily initialize m_recent_rejects_reconsiderable (dergoegge)
662e8db2d3 [net processing] Lazily initialize m_recent_rejects (dergoegge)

Pull request description:

  This adds a fuzzing harness dedicated to the version handshake. To avoid determinism issues, the harness creates necessary components each iteration (addrman, peerman, etc). A harness like this would have easily caught https://bitcoincore.org/en/2024/07/03/disclose-timestamp-overflow/.

  As a performance optimization, this PR includes a change to `PeerManager` to lazily initialize various filters (to avoid large unnecessary memory allocations each iteration).

ACKs for top commit:
  brunoerg:
    ACK afd237bb5d
  marcofleon:
    Tested ACK afd237bb5d. I compared the coverage  of `net_processing` from this harness to the `process_message` and `process_messages` harnesses to see the differences. This target hits more specific parts of the version handshake. The stability looks good as well, at about 94%.
  glozow:
    utACK afd237bb5d lazy blooms look ok
  mzumsande:
    Code Review ACK afd237bb5d

Tree-SHA512: 62bba20aec0cd220e62368354891f9790b81ad75e8adf7b22a76a6d4663bd26aedc4cae8083658a75ea9043d60aad3f0e58ad36bd7bbbf93ff1d16e317bf15cc
This commit is contained in:
glozow
2024-08-01 09:33:44 +01:00
parent 66e82dc90c afd237bb5d
commit b8755164cf
6 changed files with 193 additions and 47 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
test/functional/mempool_reorg.py
View File

@@ -162,7 +162,7 @@ class MempoolCoinbaseTest(BitcoinTestFramework):
self.log.info("Generate a block")
last_block = self.generate(self.nodes[0], 1)
# generate() implicitly syncs blocks, so that peer 1 gets the block before timelock_tx
# Otherwise, peer 1 would put the timelock_tx in m_recent_rejects
# Otherwise, peer 1 would put the timelock_tx in m_lazy_recent_rejects
self.log.info("The time-locked transaction can now be spent")
timelock_tx_id = self.nodes[0].sendrawtransaction(timelock_tx)

2
test/functional/p2p_permissions.py
View File

@@ -121,7 +121,7 @@ class P2PPermissionsTests(BitcoinTestFramework):
tx.vout[0].nValue += 1
txid = tx.rehash()
# Send the transaction twice. The first time, it'll be rejected by ATMP because it conflicts
# with a mempool transaction. The second time, it'll be in the m_recent_rejects filter.
# with a mempool transaction. The second time, it'll be in the m_lazy_recent_rejects filter.
p2p_rebroadcast_wallet.send_txs_and_test(
[tx],
self.nodes[1],