Merge bitcoin/bitcoin#28052: blockstorage: XOR blocksdir *.dat files

fa895c7 mingw: Document mode wbx workaround (MarcoFalke)
fa35925 Add -blocksxor boolean option (MarcoFalke)
fa7f7ac Return XOR AutoFile from BlockManager::Open*File() (MarcoFalke)

Pull request description:

  Currently the *.dat files in the blocksdir store the data received from remote peers as-is. This may be problematic when a program other than Bitcoin Core tries to interpret them by accident. For example, an anti-virus program or other program may scan them and move them into quarantine, or delete them, or corrupt them. This may cause Bitcoin Core to fail a reorg, or fail to reply to block requests (via P2P, RPC, REST, ...).

  Fix this, similar to bitcoin/bitcoin#6650, by rolling a random XOR pattern over the dat files when writing or reading them.

  Obviously this can only protect against programs that accidentally and unintentionally are trying to mess with the dat files. Any program that intentionally wants to mess with the dat files can still trivially do so.

  The XOR pattern is only applied when the blocksdir is freshly created, and there is an option to disable it (on creation), so that people can disable it, if needed.

ACKs for top commit:
  achow101:
    ACK fa895c7
  TheCharlatan:
    Re-ACK fa895c7
  hodlinator:
    ACK fa895c7

Tree-SHA512: c92a6a717da83bc33a9b8671a779eeefde2c63b192362ba1d71e6535ee31d08e2802b74acc908345197de9daac6930e4771595ee25b09acd5a67f7ea34854720

Author: achow101

doc/files.md

@@ -47,8 +47,9 @@ Subdirectory       | File(s)               | Description
 -------------------|-----------------------|------------
 `blocks/`          |                       | Blocks directory; can be specified by `-blocksdir` option (except for `blocks/index/`)
 `blocks/index/`    | LevelDB database      | Block index; `-blocksdir` option does not affect this path
-`blocks/`          | `blkNNNNN.dat`<sup>[\[2\]](#note2)</sup> | Actual Bitcoin blocks (in network format, dumped in raw on disk, 128 MiB per file)
+`blocks/`          | `blkNNNNN.dat`<sup>[\[2\]](#note2)</sup> | Actual Bitcoin blocks (dumped in network format, 128 MiB per file)
 `blocks/`          | `revNNNNN.dat`<sup>[\[2\]](#note2)</sup> | Block undo data (custom format)
+`blocks/`          | `xor.dat`             | Rolling XOR pattern for block and undo data files
 `chainstate/`      | LevelDB database      | Blockchain state (a compact representation of all currently unspent transaction outputs (UTXOs) and metadata about the transactions they are from)
 `indexes/txindex/` | LevelDB database      | Transaction index; *optional*, used if `-txindex=1`
 `indexes/blockfilter/basic/db/` | LevelDB database      | Blockfilter index LevelDB database for the basic filtertype; *optional*, used if `-blockfilterindex=basic`

doc/release-notes-28052.md

@@ -0,0 +1,6 @@
+Blockstorage
+============
+
+Block files are now XOR'd by default with a key stored in the blocksdir.
+Previous releases of Bitcoin Core or previous external software will not be able to read the blocksdir with a non-zero XOR-key.
+Refer to the `-blocksxor` help for more details.

src/init.cpp

@@ -469,6 +469,13 @@ void SetupServerArgs(ArgsManager& argsman)
 #endif
     argsman.AddArg("-assumevalid=<hex>", strprintf("If this block is in the chain assume that it and its ancestors are valid and potentially skip their script verification (0 to verify all, default: %s, testnet: %s, signet: %s)", defaultChainParams->GetConsensus().defaultAssumeValid.GetHex(), testnetChainParams->GetConsensus().defaultAssumeValid.GetHex(), signetChainParams->GetConsensus().defaultAssumeValid.GetHex()), ArgsManager::ALLOW_ANY, OptionsCategory::OPTIONS);
     argsman.AddArg("-blocksdir=<dir>", "Specify directory to hold blocks subdirectory for *.dat files (default: <datadir>)", ArgsManager::ALLOW_ANY, OptionsCategory::OPTIONS);
+    argsman.AddArg("-blocksxor",
+                   strprintf("Whether an XOR-key applies to blocksdir *.dat files. "
+                             "The created XOR-key will be zeros for an existing blocksdir or when `-blocksxor=0` is "
+                             "set, and random for a freshly initialized blocksdir. "
+                             "(default: %u)",
+                             kernel::DEFAULT_XOR_BLOCKSDIR),
+                   ArgsManager::ALLOW_ANY, OptionsCategory::OPTIONS);
     argsman.AddArg("-fastprune", "Use smaller block files and lower minimum prune height for testing purposes", ArgsManager::ALLOW_ANY | ArgsManager::DEBUG_ONLY, OptionsCategory::DEBUG_TEST);
 #if HAVE_SYSTEM
     argsman.AddArg("-blocknotify=<cmd>", "Execute command when the best block changes (%s in cmd is replaced by block hash)", ArgsManager::ALLOW_ANY, OptionsCategory::OPTIONS);
@@ -1534,7 +1541,11 @@ bool AppInitMain(NodeContext& node, interfaces::BlockAndHeaderTipInfo* tip_info)
         }
         LogPrintf("* Using %.1f MiB for in-memory UTXO set (plus up to %.1f MiB of unused mempool space)\n", cache_sizes.coins * (1.0 / 1024 / 1024), mempool_opts.max_size_bytes * (1.0 / 1024 / 1024));
 
-        node.chainman = std::make_unique<ChainstateManager>(*Assert(node.shutdown), chainman_opts, blockman_opts);
+        try {
+            node.chainman = std::make_unique<ChainstateManager>(*Assert(node.shutdown), chainman_opts, blockman_opts);
+        } catch (std::exception& e) {
+            return InitError(strprintf(Untranslated("Failed to initialize ChainstateManager: %s"), e.what()));
+        }
         ChainstateManager& chainman = *node.chainman;
 
         // This is defined and set here instead of inline in validation.h to avoid a hard

src/kernel/blockmanager_opts.h

@@ -14,12 +14,15 @@ class CChainParams;
 
 namespace kernel {
 
+static constexpr bool DEFAULT_XOR_BLOCKSDIR{true};
+
 /**
  * An options struct for `BlockManager`, more ergonomically referred to as
  * `BlockManager::Options` due to the using-declaration in `BlockManager`.
  */
 struct BlockManagerOpts {
     const CChainParams& chainparams;
+    bool use_xor{DEFAULT_XOR_BLOCKSDIR};
     uint64_t prune_target{0};
     bool fast_prune{false};
     const fs::path blocks_dir;

src/node/blockmanager_args.cpp

@@ -16,6 +16,7 @@
 namespace node {
 util::Result<void> ApplyArgsManOptions(const ArgsManager& args, BlockManager::Options& opts)
 {
+    if (auto value{args.GetBoolArg("-blocksxor")}) opts.use_xor = *value;
     // block pruning; get the amount of disk space (in MiB) to allot for block & undo files
     int64_t nPruneArg{args.GetIntArg("-prune", opts.prune_target)};
     if (nPruneArg < 0) {

src/node/blockstorage.cpp

@@ -19,6 +19,7 @@
 #include <pow.h>
 #include <primitives/block.h>
 #include <primitives/transaction.h>
+#include <random.h>
 #include <reverse_iterator.h>
 #include <serialize.h>
 #include <signet.h>
@@ -818,13 +819,13 @@ void BlockManager::UnlinkPrunedFiles(const std::set<int>& setFilesToPrune) const
 
 AutoFile BlockManager::OpenBlockFile(const FlatFilePos& pos, bool fReadOnly) const
 {
-    return AutoFile{m_block_file_seq.Open(pos, fReadOnly)};
+    return AutoFile{m_block_file_seq.Open(pos, fReadOnly), m_xor_key};
 }
 
 /** Open an undo file (rev?????.dat) */
 AutoFile BlockManager::OpenUndoFile(const FlatFilePos& pos, bool fReadOnly) const
 {
-    return AutoFile{m_undo_file_seq.Open(pos, fReadOnly)};
+    return AutoFile{m_undo_file_seq.Open(pos, fReadOnly), m_xor_key};
 }
 
 fs::path BlockManager::GetBlockPosFilename(const FlatFilePos& pos) const
@@ -1144,6 +1145,54 @@ FlatFilePos BlockManager::SaveBlockToDisk(const CBlock& block, int nHeight)
     return blockPos;
 }
 
+static auto InitBlocksdirXorKey(const BlockManager::Options& opts)
+{
+    // Bytes are serialized without length indicator, so this is also the exact
+    // size of the XOR-key file.
+    std::array<std::byte, 8> xor_key{};
+
+    if (opts.use_xor && fs::is_empty(opts.blocks_dir)) {
+        // Only use random fresh key when the boolean option is set and on the
+        // very first start of the program.
+        FastRandomContext{}.fillrand(xor_key);
+    }
+
+    const fs::path xor_key_path{opts.blocks_dir / "xor.dat"};
+    if (fs::exists(xor_key_path)) {
+        // A pre-existing xor key file has priority.
+        AutoFile xor_key_file{fsbridge::fopen(xor_key_path, "rb")};
+        xor_key_file >> xor_key;
+    } else {
+        // Create initial or missing xor key file
+        AutoFile xor_key_file{fsbridge::fopen(xor_key_path,
+#ifdef __MINGW64__
+            "wb" // Temporary workaround for https://github.com/bitcoin/bitcoin/issues/30210
+#else
+            "wbx"
+#endif
+        )};
+        xor_key_file << xor_key;
+    }
+    // If the user disabled the key, it must be zero.
+    if (!opts.use_xor && xor_key != decltype(xor_key){}) {
+        throw std::runtime_error{
+            strprintf("The blocksdir XOR-key can not be disabled when a random key was already stored! "
+                      "Stored key: '%s', stored path: '%s'.",
+                      HexStr(xor_key), fs::PathToString(xor_key_path)),
+        };
+    }
+    LogInfo("Using obfuscation key for blocksdir *.dat files (%s): '%s'\n", fs::PathToString(opts.blocks_dir), HexStr(xor_key));
+    return std::vector<std::byte>{xor_key.begin(), xor_key.end()};
+}
+
+BlockManager::BlockManager(const util::SignalInterrupt& interrupt, Options opts)
+    : m_prune_mode{opts.prune_target > 0},
+      m_xor_key{InitBlocksdirXorKey(opts)},
+      m_opts{std::move(opts)},
+      m_block_file_seq{FlatFileSeq{m_opts.blocks_dir, "blk", m_opts.fast_prune ? 0x4000 /* 16kB */ : BLOCKFILE_CHUNK_SIZE}},
+      m_undo_file_seq{FlatFileSeq{m_opts.blocks_dir, "rev", UNDOFILE_CHUNK_SIZE}},
+      m_interrupt{interrupt} {}
+
 class ImportingNow
 {
     std::atomic<bool>& m_importing;

src/node/blockstorage.h

@@ -240,6 +240,8 @@ class BlockManager
 
     const bool m_prune_mode;
 
+    const std::vector<std::byte> m_xor_key;
+
     /** Dirty block index entries. */
     std::set<CBlockIndex*> m_dirty_blockindex;
 
@@ -264,12 +266,7 @@ class BlockManager
 public:
     using Options = kernel::BlockManagerOpts;
 
-    explicit BlockManager(const util::SignalInterrupt& interrupt, Options opts)
-        : m_prune_mode{opts.prune_target > 0},
-          m_opts{std::move(opts)},
-          m_block_file_seq{FlatFileSeq{m_opts.blocks_dir, "blk", m_opts.fast_prune ? 0x4000 /* 16kB */ : BLOCKFILE_CHUNK_SIZE}},
-          m_undo_file_seq{FlatFileSeq{m_opts.blocks_dir, "rev", UNDOFILE_CHUNK_SIZE}},
-          m_interrupt{interrupt} {}
+    explicit BlockManager(const util::SignalInterrupt& interrupt, Options opts);
 
     const util::SignalInterrupt& m_interrupt;
     std::atomic<bool> m_importing{false};

src/test/streams_tests.cpp

@@ -30,8 +30,7 @@ BOOST_AUTO_TEST_CASE(xor_file)
     }
     {
 #ifdef __MINGW64__
-        // Our usage of mingw-w64 and the msvcrt runtime does not support
-        // the x modifier for the _wfopen().
+        // Temporary workaround for https://github.com/bitcoin/bitcoin/issues/30210
         const char* mode = "wb";
 #else
         const char* mode = "wbx";

test/functional/feature_loadblock.py

@@ -26,6 +26,10 @@ def set_test_params(self):
         self.setup_clean_chain = True
         self.num_nodes = 2
         self.supports_cli = False
+        self.extra_args = [
+            ["-blocksxor=0"],  # TODO: The linearize scripts should be adjusted to apply any XOR
+            [],
+        ]
 
     def run_test(self):
         self.nodes[1].setnetworkactive(state=False)

test/functional/feature_reindex.py

@@ -39,9 +39,19 @@ def out_of_order(self):
         # we're generating them rather than getting them from peers), so to
         # test out-of-order handling, swap blocks 1 and 2 on disk.
         blk0 = self.nodes[0].blocks_path / "blk00000.dat"
+        with open(self.nodes[0].blocks_path / "xor.dat", "rb") as xor_f:
+            NUM_XOR_BYTES = 8 # From InitBlocksdirXorKey::xor_key.size()
+            xor_dat = xor_f.read(NUM_XOR_BYTES)
+
+        def util_xor(data, key, *, offset):
+            data = bytearray(data)
+            for i in range(len(data)):
+                data[i] ^= key[(i + offset) % len(key)]
+            return bytes(data)
+
         with open(blk0, 'r+b') as bf:
             # Read at least the first few blocks (including genesis)
-            b = bf.read(2000)
+            b = util_xor(bf.read(2000), xor_dat, offset=0)
 
             # Find the offsets of blocks 2, 3, and 4 (the first 3 blocks beyond genesis)
             # by searching for the regtest marker bytes (see pchMessageStart).
@@ -55,12 +65,12 @@ def find_block(b, start):
             b4_start = find_block(b, b3_start)
 
             # Blocks 2 and 3 should be the same size.
-            assert_equal(b3_start-b2_start, b4_start-b3_start)
+            assert_equal(b3_start - b2_start, b4_start - b3_start)
 
             # Swap the second and third blocks (don't disturb the genesis block).
             bf.seek(b2_start)
-            bf.write(b[b3_start:b4_start])
-            bf.write(b[b2_start:b3_start])
+            bf.write(util_xor(b[b3_start:b4_start], xor_dat, offset=b2_start))
+            bf.write(util_xor(b[b2_start:b3_start], xor_dat, offset=b3_start))
 
         # The reindexing code should detect and accommodate out of order blocks.
         with self.nodes[0].assert_debug_log([