Merge #11458: Don't process unrequested, low-work blocks

01b52ce Add comment explaining forced processing of compact blocks (Suhas Daftuar)
08fd822 qa: add test for minchainwork use in acceptblock (Suhas Daftuar)
ce8cd7a Don't process unrequested, low-work blocks (Suhas Daftuar)

Pull request description:

  A peer could try to waste our resources by sending us unrequested blocks with
  low work (eg to fill up our disk).  Since e265200 we no longer request blocks until we
  know we're on a chain with more than nMinimumChainWork (our anti-DoS
  threshold), but we would still process unrequested blocks that had more work
  than our tip (which generally has low-work during IBD), even though we may not
  yet have found a headers chain with sufficient work.

  Fix this and add a test.

Tree-SHA512: 1a4fb0bbd78054b84683f995c8c3194dd44fa914dc351ae4379c7c1a6f83224f609f8b9c2d9dde28741426c6af008ffffea836d21aa31a5ebaa00f8e0f81229e
This commit is contained in:
Wladimir J. van der Laan 2017-10-21 11:13:46 +02:00 committed by Alexander Block
parent 9938dd83d4
commit 35d60de2b5
3 changed files with 54 additions and 13 deletions

View File

@ -2449,7 +2449,16 @@ bool static ProcessMessage(CNode* pfrom, const std::string& strCommand, CDataStr
mapBlockSource.emplace(pblock->GetHash(), std::make_pair(pfrom->GetId(), false));
}
bool fNewBlock = false;
ProcessNewBlock(chainparams, pblock, true, &fNewBlock);
// Setting fForceProcessing to true means that we bypass some of
// our anti-DoS protections in AcceptBlock, which filters
// unrequested blocks that might be trying to waste our resources
// (eg disk space). Because we only try to reconstruct blocks when
// we're close to caught up (via the CanDirectFetch() requirement
// above, combined with the behavior of not requesting blocks until
// we have a chain with at least nMinimumChainWork), and we ignore
// compact blocks with less work than our tip, it is safe to treat
// reconstructed compact blocks as having been requested.
ProcessNewBlock(chainparams, pblock, /*fForceProcessing=*/true, &fNewBlock);
if (fNewBlock) {
pfrom->nLastBlockTime = GetTime();
} else {
@ -2529,7 +2538,11 @@ bool static ProcessMessage(CNode* pfrom, const std::string& strCommand, CDataStr
bool fNewBlock = false;
// Since we requested this block (it was in mapBlocksInFlight), force it to be processed,
// even if it would not be a candidate for new tip (missing previous block, chain not long enough, etc)
ProcessNewBlock(chainparams, pblock, true, &fNewBlock);
// This bypasses some anti-DoS logic in AcceptBlock (eg to prevent
// disk-space attacks), but this should be safe due to the
// protections in the compact block handler -- see related comment
// in compact block optimistic reconstruction handling.
ProcessNewBlock(chainparams, pblock, /*fForceProcessing=*/true, &fNewBlock);
if (fNewBlock) {
pfrom->nLastBlockTime = GetTime();
} else {

View File

@ -3464,6 +3464,12 @@ static bool AcceptBlock(const std::shared_ptr<const CBlock>& pblock, CValidation
if (pindex->nTx != 0) return true; // This is a previously-processed block that was pruned
if (!fHasMoreWork) return true; // Don't process less-work chains
if (fTooFarAhead) return true; // Block height is too high
// Protect against DoS attacks from low-work chains.
// If our tip is behind, a peer could try to send us
// low-work blocks on a fake chain that we would never
// request; don't process these.
if (pindex->nChainWork < nMinimumChainWork) return true;
}
if (fNewBlock) *fNewBlock = true;

View File

@ -8,17 +8,22 @@ Since behavior differs when receiving unrequested blocks from whitelisted peers
versus non-whitelisted peers, this tests the behavior of both (effectively two
separate tests running in parallel).
Setup: two nodes, node0 and node1, not connected to each other. Node0 does not
Setup: three nodes, node0+node1+node2, not connected to each other. Node0 does not
whitelist localhost, but node1 does. They will each be on their own chain for
this test.
this test. Node2 will have nMinimumChainWork set to 0x10, so it won't process
low-work unrequested blocks.
We have one NodeConn connection to each, test_node and white_node respectively.
We have one NodeConn connection to each, test_node, white_node, and min_work_node,
respectively.
The test:
1. Generate one block on each node, to leave IBD.
2. Mine a new block on each tip, and deliver to each node from node's peer.
The tip should advance.
The tip should advance for node0 and node1, but node2 should skip processing
due to nMinimumChainWork.
Node2 is unused in tests 3-7:
3. Mine a block that forks the previous block, and deliver to each node from
corresponding peer.
@ -46,6 +51,10 @@ The test:
7. Send Node0 the missing block again.
Node0 should process and the tip should advance.
8. Test Node2 is able to sync when connected to node0 (which should have sufficient
work on its chain).
"""
from test_framework.mininode import *
@ -62,52 +71,60 @@ class AcceptBlockTest(BitcoinTestFramework):
def set_test_params(self):
self.setup_clean_chain = True
self.num_nodes = 2
self.extra_args = [[], ["-whitelist=127.0.0.1"]]
self.num_nodes = 3
self.extra_args = [[], ["-whitelist=127.0.0.1"], ["-minimumchainwork=0x10"]]
def setup_network(self):
# Node0 will be used to test behavior of processing unrequested blocks
# from peers which are not whitelisted, while Node1 will be used for
# the whitelisted case.
# Node2 will be used for non-whitelisted peers to test the interaction
# with nMinimumChainWork.
self.setup_nodes()
def run_test(self):
# Setup the p2p connections and start up the network thread.
test_node = NodeConnCB() # connects to node0 (not whitelisted)
white_node = NodeConnCB() # connects to node1 (whitelisted)
min_work_node = NodeConnCB() # connects to node2 (not whitelisted)
connections = []
connections.append(NodeConn('127.0.0.1', p2p_port(0), self.nodes[0], test_node))
connections.append(NodeConn('127.0.0.1', p2p_port(1), self.nodes[1], white_node))
connections.append(NodeConn('127.0.0.1', p2p_port(2), self.nodes[2], min_work_node))
test_node.add_connection(connections[0])
white_node.add_connection(connections[1])
min_work_node.add_connection(connections[2])
NetworkThread().start() # Start up network handling in another thread
# Test logic begins here
test_node.wait_for_verack()
white_node.wait_for_verack()
min_work_node.wait_for_verack()
# 1. Have both nodes mine a block (leave IBD)
# 1. Have nodes mine a block (nodes1/2 leave IBD)
[ n.generate(1) for n in self.nodes ]
tips = [ int("0x" + n.getbestblockhash(), 0) for n in self.nodes ]
# 2. Send one block that builds on each tip.
# This should be accepted.
# This should be accepted by nodes 1/2
blocks_h2 = [] # the height 2 blocks on each node's chain
block_time = self.mocktime + 1
for i in range(2):
for i in range(3):
blocks_h2.append(create_block(tips[i], create_coinbase(2), block_time + 1))
blocks_h2[i].solve()
block_time += 1
test_node.send_message(msg_block(blocks_h2[0]))
white_node.send_message(msg_block(blocks_h2[1]))
min_work_node.send_message(msg_block(blocks_h2[2]))
for x in [test_node, white_node]:
for x in [test_node, white_node, min_work_node]:
x.sync_with_ping()
assert_equal(self.nodes[0].getblockcount(), 2)
assert_equal(self.nodes[1].getblockcount(), 2)
self.log.info("First height 2 block accepted by both nodes")
assert_equal(self.nodes[2].getblockcount(), 1)
self.log.info("First height 2 block accepted by node0/node1; correctly rejected by node2")
# 3. Send another block that builds on the original tip.
blocks_h2f = [] # Blocks at height 2 that fork off the main chain
@ -221,6 +238,11 @@ class AcceptBlockTest(BitcoinTestFramework):
assert_equal(self.nodes[0].getblockcount(), 290)
self.log.info("Successfully reorged to longer chain from non-whitelisted peer")
# 8. Connect node2 to node0 and ensure it is able to sync
connect_nodes(self.nodes[0], 2)
sync_blocks([self.nodes[0], self.nodes[2]])
self.log.info("Successfully synced nodes 2 and 0")
[ c.disconnect_node() for c in connections ]
if __name__ == '__main__':