precomputed block hashes are now in blocks of N (currently 256)

This shaves a lot of space off binaries

10 files changed, 180 insertions, 12 deletions

diff --git a/src/blockchain_utilities/blockchain_import.cpp b/src/blockchain_utilities/blockchain_import.cpp
index 635a70b10..3841911fe 100644
--- a/src/blockchain_utilities/blockchain_import.cpp
+++ b/src/blockchain_utilities/blockchain_import.cpp
@@ -169,6 +169,26 @@ int check_flush(cryptonote::core &core, std::list<block_complete_entry> &blocks,
   if (!force && blocks.size() < db_batch_size)
     return 0;
 
+  // wait till we can verify a full HOH without extra, for speed
+  uint64_t new_height = core.get_blockchain_storage().get_db().height() + blocks.size();
+  if (!force && new_height % HASH_OF_HASHES_STEP)
+    return 0;
+
+  std::list<crypto::hash> hashes;
+  for (const auto &b: blocks)
+  {
+    cryptonote::block block;
+    if (!parse_and_validate_block_from_blob(b.block, block))
+    {
+      MERROR("Failed to parse block: "
+          << epee::string_tools::pod_to_hex(get_blob_hash(b.block)));
+      core.cleanup_handle_incoming_blocks();
+      return 1;
+    }
+    hashes.push_back(cryptonote::get_block_hash(block));
+  }
+  core.prevalidate_block_hashes(core.get_blockchain_storage().get_db().height(), hashes);
+
   core.prepare_handle_incoming_blocks(blocks);
 
   for(const block_complete_entry& block_entry: blocks)
diff --git a/src/blockchain_utilities/blocksdat_file.cpp b/src/blockchain_utilities/blocksdat_file.cpp
index 63224225f..32e93345e 100644
--- a/src/blockchain_utilities/blocksdat_file.cpp
+++ b/src/blockchain_utilities/blocksdat_file.cpp
@@ -82,7 +82,7 @@ bool BlocksdatFile::open_writer(const boost::filesystem::path& file_path, uint64
 
 bool BlocksdatFile::initialize_file(uint64_t block_stop)
 {
-  const uint32_t nblocks = block_stop + 1;
+  const uint32_t nblocks = (block_stop + 1) / HASH_OF_HASHES_STEP;
   unsigned char nblocksc[4];
 
   nblocksc[0] = nblocks & 0xff;
@@ -101,8 +101,16 @@ bool BlocksdatFile::initialize_file(uint64_t block_stop)
 
 void BlocksdatFile::write_block(const crypto::hash& block_hash)
 {
-  const std::string data(block_hash.data, sizeof(block_hash));
-  *m_raw_data_file << data;
+  m_hashes.push_back(block_hash);
+  while (m_hashes.size() >= HASH_OF_HASHES_STEP)
+  {
+    crypto::hash hash;
+    crypto::cn_fast_hash(m_hashes.data(), HASH_OF_HASHES_STEP * sizeof(crypto::hash), hash);
+    memmove(m_hashes.data(), m_hashes.data() + HASH_OF_HASHES_STEP * sizeof(crypto::hash), (m_hashes.size() - HASH_OF_HASHES_STEP) * sizeof(crypto::hash));
+    m_hashes.resize(m_hashes.size() - HASH_OF_HASHES_STEP);
+    const std::string data(hash.data, sizeof(hash));
+    *m_raw_data_file << data;
+  }
 }
 
 bool BlocksdatFile::close()
diff --git a/src/blockchain_utilities/blocksdat_file.h b/src/blockchain_utilities/blocksdat_file.h
index 0a5913058..d43811772 100644
--- a/src/blockchain_utilities/blocksdat_file.h
+++ b/src/blockchain_utilities/blocksdat_file.h
@@ -76,4 +76,5 @@ protected:
 private:
 
   uint64_t m_cur_height; // tracks current height during export
+  std::vector<crypto::hash> m_hashes;
 };
diff --git a/src/blocks/checkpoints.dat b/src/blocks/checkpoints.dat
index 15fa042cf..c12c3d8a0 100644
--- a/src/blocks/checkpoints.dat
+++ b/src/blocks/checkpoints.dat
diff --git a/src/cryptonote_config.h b/src/cryptonote_config.h
index cdb46dda9..f38528d58 100644
--- a/src/cryptonote_config.h
+++ b/src/cryptonote_config.h
@@ -137,6 +137,9 @@
 
 #define PER_KB_FEE_QUANTIZATION_DECIMALS        8
 
+#define HASH_OF_HASHES_STEP                     256
+
+
 // New constants are intended to go here
 namespace config
 {
diff --git a/src/cryptonote_core/blockchain.cpp b/src/cryptonote_core/blockchain.cpp
index 619dbdc07..ece5c46aa 100644
--- a/src/cryptonote_core/blockchain.cpp
+++ b/src/cryptonote_core/blockchain.cpp
@@ -3206,13 +3206,21 @@ leave:
   if (m_db->height() < m_blocks_hash_check.size())
   {
     auto hash = get_block_hash(bl);
-    if (memcmp(&hash, &m_blocks_hash_check[m_db->height()], sizeof(hash)) != 0)
+    const auto &expected_hash = m_blocks_hash_check[m_db->height()];
+    if (expected_hash != cryptonote::null_hash)
     {
-      MERROR_VER("Block with id is INVALID: " << id);
-      bvc.m_verifivation_failed = true;
-      goto leave;
+      if (memcmp(&hash, &expected_hash, sizeof(hash)) != 0)
+      {
+        MERROR_VER("Block with id is INVALID: " << id);
+        bvc.m_verifivation_failed = true;
+        goto leave;
+      }
+      fast_check = true;
+    }
+    else
+    {
+      MCINFO("verify", "No pre-validated hash at height " << m_db->height() << ", verifying fully");
     }
-    fast_check = true;
   }
   else
 #endif
@@ -3657,6 +3665,14 @@ bool Blockchain::cleanup_handle_incoming_blocks(bool force_sync)
   m_blocks_txs_check.clear();
   m_check_txin_table.clear();
 
+  // when we're well clear of the precomputed hashes, free the memory
+  if (!m_blocks_hash_check.empty() && m_db->height() > m_blocks_hash_check.size() + 4096)
+  {
+    MINFO("Dumping block hashes, we're now 4k past " << m_blocks_hash_check.size());
+    m_blocks_hash_check.clear();
+    m_blocks_hash_check.shrink_to_fit();
+  }
+
   CRITICAL_REGION_END();
   m_tx_pool.unlock();
 
@@ -3681,6 +3697,98 @@ void Blockchain::output_scan_worker(const uint64_t amount, const std::vector<uin
   }
 }
 
+uint64_t Blockchain::prevalidate_block_hashes(uint64_t height, const std::list<crypto::hash> &hashes)
+{
+  // new: . . . . . X X X X X . . . . . .
+  // pre: A A A A B B B B C C C C D D D D
+
+  // easy case: height >= hashes
+  if (height >= m_blocks_hash_of_hashes.size() * HASH_OF_HASHES_STEP)
+    return hashes.size();
+
+  // find hashes encompassing those block
+  size_t first_index = height / HASH_OF_HASHES_STEP;
+  size_t last_index = (height + hashes.size() - 1) / HASH_OF_HASHES_STEP;
+  MDEBUG("Blocks " << height << " - " << (height + hashes.size() - 1) << " start at " << first_index << " and end at " << last_index);
+
+  // case of not enough to calculate even a single hash
+  if (first_index == last_index && hashes.size() < HASH_OF_HASHES_STEP && (height + hashes.size()) % HASH_OF_HASHES_STEP)
+    return hashes.size();
+
+  // build hashes vector to hash hashes together
+  std::vector<crypto::hash> data;
+  data.reserve(hashes.size() + HASH_OF_HASHES_STEP - 1); // may be a bit too much
+
+  // we expect height to be either equal or a bit below db height
+  bool disconnected = (height > m_db->height());
+  size_t pop;
+  if (disconnected && height % HASH_OF_HASHES_STEP)
+  {
+    ++first_index;
+    pop = HASH_OF_HASHES_STEP - height % HASH_OF_HASHES_STEP;
+  }
+  else
+  {
+    // we might need some already in the chain for the first part of the first hash
+    for (uint64_t h = first_index * HASH_OF_HASHES_STEP; h < height; ++h)
+    {
+      data.push_back(m_db->get_block_hash_from_height(h));
+    }
+    pop = 0;
+  }
+
+  // push the data to check
+  for (const auto &h: hashes)
+  {
+    if (pop)
+      --pop;
+    else
+      data.push_back(h);
+  }
+
+  // hash and check
+  uint64_t usable = first_index * HASH_OF_HASHES_STEP - height; // may start negative, but unsigned under/overflow is not UB
+  for (size_t n = first_index; n <= last_index; ++n)
+  {
+    if (n < m_blocks_hash_of_hashes.size())
+    {
+      // if the last index isn't fully filled, we can't tell if valid
+      if (data.size() < (n - first_index) * HASH_OF_HASHES_STEP + HASH_OF_HASHES_STEP)
+        break;
+
+      crypto::hash hash;
+      cn_fast_hash(data.data() + (n - first_index) * HASH_OF_HASHES_STEP, HASH_OF_HASHES_STEP * sizeof(crypto::hash), hash);
+      bool valid = hash == m_blocks_hash_of_hashes[n];
+
+      // add to the known hashes array
+      if (!valid)
+      {
+        MWARNING("invalid hash for blocks " << n * HASH_OF_HASHES_STEP << " - " << (n * HASH_OF_HASHES_STEP + HASH_OF_HASHES_STEP - 1));
+        break;
+      }
+
+      size_t end = n * HASH_OF_HASHES_STEP + HASH_OF_HASHES_STEP;
+      for (size_t i = n * HASH_OF_HASHES_STEP; i < end; ++i)
+      {
+        CHECK_AND_ASSERT_MES(m_blocks_hash_check[i] == cryptonote::null_hash || m_blocks_hash_check[i] == data[i - first_index * HASH_OF_HASHES_STEP],
+            0, "Consistency failure in m_blocks_hash_check construction");
+        m_blocks_hash_check[i] = data[i - first_index * HASH_OF_HASHES_STEP];
+      }
+      usable += HASH_OF_HASHES_STEP;
+    }
+    else
+    {
+      // if after the end of the precomputed blocks, accept anything
+      usable += HASH_OF_HASHES_STEP;
+      if (usable > hashes.size())
+        usable = hashes.size();
+    }
+  }
+  MDEBUG("usable: " << usable << " / " << hashes.size());
+  CHECK_AND_ASSERT_MES(usable < std::numeric_limits<uint64_t>::max() / 2, 0, "usable is negative");
+  return usable;
+}
+
 //------------------------------------------------------------------
 // ND: Speedups:
 // 1. Thread long_hash computations if possible (m_max_prepare_blocks_threads = nthreads, default = 4)
@@ -4144,7 +4252,7 @@ void Blockchain::cancel()
 }
 
 #if defined(PER_BLOCK_CHECKPOINT)
-static const char expected_block_hashes_hash[] = "d3ca80d50661684cde0e715d46d7c19704d2e216b21ed088af9fd4ef37ed4d65";
+static const char expected_block_hashes_hash[] = "4b553162ee4e7af3c53666506591489c68560b9175e6e941dc96c89f96f0e35c";
 void Blockchain::load_compiled_in_block_hashes()
 {
   if (m_fast_sync && get_blocks_dat_start(m_testnet) != nullptr && get_blocks_dat_size(m_testnet) > 0)
@@ -4180,16 +4288,18 @@ void Blockchain::load_compiled_in_block_hashes()
       const unsigned char *p = get_blocks_dat_start(m_testnet);
       const uint32_t nblocks = *p | ((*(p+1))<<8) | ((*(p+2))<<16) | ((*(p+3))<<24);
       const size_t size_needed = 4 + nblocks * sizeof(crypto::hash);
-      if(nblocks > 0 && nblocks > m_db->height() && get_blocks_dat_size(m_testnet) >= size_needed)
+      if(nblocks > 0 && nblocks * HASH_OF_HASHES_STEP > m_db->height() && get_blocks_dat_size(m_testnet) >= size_needed)
       {
         p += sizeof(uint32_t);
+        m_blocks_hash_of_hashes.reserve(nblocks);
         for (uint32_t i = 0; i < nblocks; i++)
         {
           crypto::hash hash;
           memcpy(hash.data, p, sizeof(hash.data));
           p += sizeof(hash.data);
-          m_blocks_hash_check.push_back(hash);
+          m_blocks_hash_of_hashes.push_back(hash);
         }
+        m_blocks_hash_check.resize(m_blocks_hash_of_hashes.size() * HASH_OF_HASHES_STEP, cryptonote::null_hash);
         MINFO(nblocks << " block hashes loaded");
 
         // FIXME: clear tx_pool because the process might have been
@@ -4220,7 +4330,7 @@ void Blockchain::load_compiled_in_block_hashes()
 bool Blockchain::is_within_compiled_block_hash_area(uint64_t height) const
 {
 #if defined(PER_BLOCK_CHECKPOINT)
-  return height < m_blocks_hash_check.size();
+  return height < m_blocks_hash_of_hashes.size() * HASH_OF_HASHES_STEP;
 #else
   return false;
 #endif
diff --git a/src/cryptonote_core/blockchain.h b/src/cryptonote_core/blockchain.h
index e2da535cd..4f3262878 100644
--- a/src/cryptonote_core/blockchain.h
+++ b/src/cryptonote_core/blockchain.h
@@ -955,6 +955,7 @@ namespace cryptonote
 
     bool is_within_compiled_block_hash_area(uint64_t height) const;
     bool is_within_compiled_block_hash_area() const { return is_within_compiled_block_hash_area(m_db->height()); }
+    uint64_t prevalidate_block_hashes(uint64_t height, const std::list<crypto::hash> &hashes);
 
     void lock();
     void unlock();
@@ -995,6 +996,7 @@ namespace cryptonote
     std::unordered_map<crypto::hash, std::unordered_map<crypto::key_image, bool>> m_check_txin_table;
 
     // SHA-3 hashes for each block and for fast pow checking
+    std::vector<crypto::hash> m_blocks_hash_of_hashes;
     std::vector<crypto::hash> m_blocks_hash_check;
     std::vector<crypto::hash> m_blocks_txs_check;
 
diff --git a/src/cryptonote_core/cryptonote_core.cpp b/src/cryptonote_core/cryptonote_core.cpp
index 56485aedf..293fb42e9 100644
--- a/src/cryptonote_core/cryptonote_core.cpp
+++ b/src/cryptonote_core/cryptonote_core.cpp
@@ -1400,6 +1400,11 @@ namespace cryptonote
     return m_target_blockchain_height;
   }
   //-----------------------------------------------------------------------------------------------
+  uint64_t core::prevalidate_block_hashes(uint64_t height, const std::list<crypto::hash> &hashes)
+  {
+    return get_blockchain_storage().prevalidate_block_hashes(height, hashes);
+  }
+  //-----------------------------------------------------------------------------------------------
   std::time_t core::get_start_time() const
   {
     return start_time;
diff --git a/src/cryptonote_core/cryptonote_core.h b/src/cryptonote_core/cryptonote_core.h
index 4ea33dbe1..f0ba16bea 100644
--- a/src/cryptonote_core/cryptonote_core.h
+++ b/src/cryptonote_core/cryptonote_core.h
@@ -752,6 +752,13 @@ namespace cryptonote
       */
      bool fluffy_blocks_enabled() const { return m_fluffy_blocks_enabled; }
 
+     /**
+      * @brief check a set of hashes against the precompiled hash set
+      *
+      * @return number of usable blocks
+      */
+     uint64_t prevalidate_block_hashes(uint64_t height, const std::list<crypto::hash> &hashes);
+
    private:
 
      /**
diff --git a/src/cryptonote_protocol/cryptonote_protocol_handler.inl b/src/cryptonote_protocol/cryptonote_protocol_handler.inl
index 803d948cc..f599e0c8d 100644
--- a/src/cryptonote_protocol/cryptonote_protocol_handler.inl
+++ b/src/cryptonote_protocol/cryptonote_protocol_handler.inl
@@ -1583,10 +1583,22 @@ skip:
       return 1;
     }
 
+    uint64_t n_use_blocks = m_core.prevalidate_block_hashes(arg.start_height, arg.m_block_ids);
+    if (n_use_blocks == 0)
+    {
+      LOG_ERROR_CCONTEXT("Peer yielded no usable blocks, dropping connection");
+      drop_connection(context, false, false);
+      return 1;
+    }
+
+    uint64_t added = 0;
     for(auto& bl_id: arg.m_block_ids)
     {
       context.m_needed_objects.push_back(bl_id);
+      if (++added == n_use_blocks)
+        break;
     }
+    context.m_last_response_height -= arg.m_block_ids.size() - n_use_blocks;
 
     if (!request_missing_objects(context, false))
     {