Merge #15305: [validation] Crash if disconnecting a block fails

a47df13 [qa] Test disconnect block failure -> shutdown (Suhas Daftuar)
4433ed0 [validation] Crash if disconnecting a block fails (Suhas Daftuar)

Pull request description:

  If we're unable to disconnect a block during normal operation, then that is a
  failure of our local system (such as disk failure) or the chain that we are on
  (eg CVE-2018-17144), but cannot be due to failure of the (more work) chain that
  we're trying to validate.

  We should abort rather than stay on a less work chain.

  Fixes #14341.

ACKs for top commit:
  practicalswift:
    utACK a47df13
  TheBlueMatt:
    utACK a47df13. Didn't bother to review the test in detail, it looked fine. Debated whether invalidateblock should ever crash the node, but *not* crashing in the case of hitting a pruned block (which is the only change here) is clearly better, even if there are other cases I'd argue we should crash in.
  ryanofsky:
    utACK a47df13. Only change since last review is new comment.
  promag:
    ACK a47df13, it takes awhile to quit (RPC connection timeouts) but that's unrelated - hope to fix that soon.
  fanquake:
    ACK a47df13

Tree-SHA512: 4dec8cef6e7dbbe513c138fc5821a7ceab855e603ece3c16185b51a3830ab7ebbc844a28827bf64e75326f45325991dcb672f13bd7baede53304f27289c4af8d

Author: fanquake

src/validation.cpp

@@ -2185,7 +2185,7 @@ bool CChainState::DisconnectTip(CValidationState& state, const CChainParams& cha
     std::shared_ptr<CBlock> pblock = std::make_shared<CBlock>();
     CBlock& block = *pblock;
     if (!ReadBlockFromDisk(block, pindexDelete, chainparams.GetConsensus()))
-        return AbortNode(state, "Failed to read block");
+        return error("DisconnectTip(): Failed to read block");
     // Apply the block atomically to the chain state.
     int64_t nStart = GetTimeMicros();
     {
@@ -2442,6 +2442,11 @@ bool CChainState::ActivateBestChainStep(CValidationState& state, const CChainPar
             // This is likely a fatal error, but keep the mempool consistent,
             // just in case. Only remove from the mempool in this case.
             UpdateMempoolForReorg(disconnectpool, false);
+
+            // If we're unable to disconnect a block during normal operation,
+            // then that is a failure of our local system -- we should abort
+            // rather than stay on a less work chain.
+            AbortNode(state, "Failed to disconnect block; see debug.log for details");
             return false;
         }
         fBlocksDisconnected = true;

test/functional/feature_abortnode.py

@@ -0,0 +1,48 @@
+#!/usr/bin/env python3
+# Copyright (c) 2019 The Bitcoin Core developers
+# Distributed under the MIT software license, see the accompanying
+# file COPYING or http://www.opensource.org/licenses/mit-license.php.
+"""Test bitcoind aborts if can't disconnect a block.
+
+- Start a single node and generate 3 blocks.
+- Delete the undo data.
+- Mine a fork that requires disconnecting the tip.
+- Verify that bitcoind AbortNode's.
+"""
+
+from test_framework.test_framework import BitcoinTestFramework
+from test_framework.util import wait_until, get_datadir_path, connect_nodes
+import os
+
+class AbortNodeTest(BitcoinTestFramework):
+
+    def set_test_params(self):
+        self.setup_clean_chain = True
+        self.num_nodes = 2
+
+    def setup_network(self):
+        self.setup_nodes()
+        # We'll connect the nodes later
+
+    def run_test(self):
+        self.nodes[0].generate(3)
+        datadir = get_datadir_path(self.options.tmpdir, 0)
+
+        # Deleting the undo file will result in reorg failure
+        os.unlink(os.path.join(datadir, 'regtest', 'blocks', 'rev00000.dat'))
+
+        # Connecting to a node with a more work chain will trigger a reorg
+        # attempt.
+        self.nodes[1].generate(3)
+        with self.nodes[0].assert_debug_log(["Failed to disconnect block"]):
+            connect_nodes(self.nodes[0], 1)
+            self.nodes[1].generate(1)
+
+            # Check that node0 aborted
+            self.log.info("Waiting for crash")
+            wait_until(lambda: self.nodes[0].is_node_stopped(), timeout=60)
+        self.log.info("Node crashed - now verifying restart fails")
+        self.nodes[0].assert_start_raises_init_error()
+
+if __name__ == '__main__':
+    AbortNodeTest().main()

test/functional/test_runner.py

@@ -107,6 +107,7 @@
     'feature_bip68_sequence.py',
     'p2p_feefilter.py',
     'feature_reindex.py',
+    'feature_abortnode.py',
     # vv Tests less than 30s vv
     'wallet_keypool_topup.py',
     'interface_zmq.py',