refactor, blockstorage: Generalize GetFirstStoredBlock

GetFirstStoredBlock is generalized to check for any data status with a
status mask that needs to be passed as a parameter. To reflect this the
function is also renamed to GetFirstBlock.

Co-authored-by: stickies-v <[email protected]>

Author: fjahr

src/node/blockstorage.cpp

@@ -595,12 +595,12 @@ bool BlockManager::IsBlockPruned(const CBlockIndex& block)
     return m_have_pruned && !(block.nStatus & BLOCK_HAVE_DATA) && (block.nTx > 0);
 }
 
-const CBlockIndex* BlockManager::GetFirstStoredBlock(const CBlockIndex& upper_block, const CBlockIndex* lower_block)
+const CBlockIndex* BlockManager::GetFirstBlock(const CBlockIndex& upper_block, uint32_t status_mask, const CBlockIndex* lower_block) const
 {
     AssertLockHeld(::cs_main);
     const CBlockIndex* last_block = &upper_block;
-    assert(last_block->nStatus & BLOCK_HAVE_DATA); // 'upper_block' must have data
-    while (last_block->pprev && (last_block->pprev->nStatus & BLOCK_HAVE_DATA)) {
+    assert((last_block->nStatus & status_mask) == status_mask); // 'upper_block' must satisfy the status mask
+    while (last_block->pprev && ((last_block->pprev->nStatus & status_mask) == status_mask)) {
         if (lower_block) {
             // Return if we reached the lower_block
             if (last_block == lower_block) return lower_block;
@@ -617,7 +617,7 @@ const CBlockIndex* BlockManager::GetFirstStoredBlock(const CBlockIndex& upper_bl
 bool BlockManager::CheckBlockDataAvailability(const CBlockIndex& upper_block, const CBlockIndex& lower_block)
 {
     if (!(upper_block.nStatus & BLOCK_HAVE_DATA)) return false;
-    return GetFirstStoredBlock(upper_block, &lower_block) == &lower_block;
+    return GetFirstBlock(upper_block, BLOCK_HAVE_DATA, &lower_block) == &lower_block;
 }
 
 // If we're using -prune with -reindex, then delete block files that will be ignored by the

src/node/blockstorage.h

@@ -335,10 +335,33 @@ class BlockManager
     //! (part of the same chain).
     bool CheckBlockDataAvailability(const CBlockIndex& upper_block LIFETIMEBOUND, const CBlockIndex& lower_block LIFETIMEBOUND) EXCLUSIVE_LOCKS_REQUIRED(::cs_main);
 
-    //! Find the first stored ancestor of start_block immediately after the last
-    //! pruned ancestor. Return value will never be null. Caller is responsible
-    //! for ensuring that start_block has data is not pruned.
-    const CBlockIndex* GetFirstStoredBlock(const CBlockIndex& start_block LIFETIMEBOUND, const CBlockIndex* lower_block=nullptr) EXCLUSIVE_LOCKS_REQUIRED(::cs_main);
+    /**
+     * @brief Returns the earliest block with specified `status_mask` flags set after
+     * the latest block _not_ having those flags.
+     *
+     * This function starts from `upper_block`, which must have all
+     * `status_mask` flags set, and iterates backwards through its ancestors. It
+     * continues as long as each block has all `status_mask` flags set, until
+     * reaching the oldest ancestor or `lower_block`.
+     *
+     * @pre `upper_block` must have all `status_mask` flags set.
+     * @pre `lower_block` must be null or an ancestor of `upper_block`
+     *
+     * @param upper_block The starting block for the search, which must have all
+     *                    `status_mask` flags set.
+     * @param status_mask Bitmask specifying required status flags.
+     * @param lower_block The earliest possible block to return. If null, the
+     *                    search can extend to the genesis block.
+     *
+     * @return A non-null pointer to the earliest block between `upper_block`
+     *         and `lower_block`, inclusive, such that every block between the
+     *         returned block and `upper_block` has `status_mask` flags set.
+     */
+    const CBlockIndex* GetFirstBlock(
+        const CBlockIndex& upper_block LIFETIMEBOUND,
+        uint32_t status_mask,
+        const CBlockIndex* lower_block = nullptr
+    ) const EXCLUSIVE_LOCKS_REQUIRED(::cs_main);
 
     /** True if any block files have ever been pruned. */
     bool m_have_pruned = false;

src/rpc/blockchain.cpp

@@ -782,6 +782,24 @@ static RPCHelpMan getblock()
     };
 }
 
+//! Return height of highest block that has been pruned, or std::nullopt if no blocks have been pruned
+std::optional<int> GetPruneHeight(const BlockManager& blockman, const CChain& chain) {
+    AssertLockHeld(::cs_main);
+
+    const CBlockIndex* chain_tip{chain.Tip()};
+    if (!(chain_tip->nStatus & BLOCK_HAVE_DATA)) return chain_tip->nHeight;
+
+    // Get first block with data, after the last block without data.
+    // This is the start of the unpruned range of blocks.
+    const auto& first_unpruned{*Assert(blockman.GetFirstBlock(*chain_tip, /*status_mask=*/BLOCK_HAVE_DATA))};
+    if (!first_unpruned.pprev) {
+       // No block before the first unpruned block means nothing is pruned.
+       return std::nullopt;
+    }
+    // Block before the first unpruned block is the last pruned block.
+    return Assert(first_unpruned.pprev)->nHeight;
+}
+
 static RPCHelpMan pruneblockchain()
 {
     return RPCHelpMan{"pruneblockchain", "",
@@ -834,8 +852,7 @@ static RPCHelpMan pruneblockchain()
     }
 
     PruneBlockFilesManual(active_chainstate, height);
-    const CBlockIndex& block{*CHECK_NONFATAL(active_chain.Tip())};
-    return block.nStatus & BLOCK_HAVE_DATA ? active_chainstate.m_blockman.GetFirstStoredBlock(block)->nHeight - 1 : block.nHeight;
+    return GetPruneHeight(chainman.m_blockman, active_chain).value_or(-1);
 },
     };
 }
@@ -1288,8 +1305,8 @@ RPCHelpMan getblockchaininfo()
     obj.pushKV("size_on_disk", chainman.m_blockman.CalculateCurrentUsage());
     obj.pushKV("pruned", chainman.m_blockman.IsPruneMode());
     if (chainman.m_blockman.IsPruneMode()) {
-        bool has_tip_data = tip.nStatus & BLOCK_HAVE_DATA;
-        obj.pushKV("pruneheight", has_tip_data ? chainman.m_blockman.GetFirstStoredBlock(tip)->nHeight : tip.nHeight + 1);
+        const auto prune_height{GetPruneHeight(chainman.m_blockman, active_chainstate.m_chain)};
+        obj.pushKV("pruneheight", prune_height ? prune_height.value() + 1 : 0);
 
         const bool automatic_pruning{chainman.m_blockman.GetPruneTarget() != BlockManager::PRUNE_TARGET_MANUAL};
         obj.pushKV("automatic_pruning",  automatic_pruning);

src/rpc/blockchain.h

@@ -21,6 +21,7 @@ class CBlockIndex;
 class Chainstate;
 class UniValue;
 namespace node {
+class BlockManager;
 struct NodeContext;
 } // namespace node
 
@@ -57,4 +58,7 @@ UniValue CreateUTXOSnapshot(
     const fs::path& path,
     const fs::path& tmppath);
 
+//! Return height of highest block that has been pruned, or std::nullopt if no blocks have been pruned
+std::optional<int> GetPruneHeight(const node::BlockManager& blockman, const CChain& chain) EXCLUSIVE_LOCKS_REQUIRED(::cs_main);
+
 #endif // BITCOIN_RPC_BLOCKCHAIN_H

src/test/blockchain_tests.cpp

@@ -5,7 +5,9 @@
 #include <boost/test/unit_test.hpp>
 
 #include <chain.h>
+#include <node/blockstorage.h>
 #include <rpc/blockchain.h>
+#include <sync.h>
 #include <test/util/setup_common.h>
 #include <util/string.h>
 
@@ -74,4 +76,36 @@ BOOST_AUTO_TEST_CASE(get_difficulty_for_very_high_target)
     TestDifficulty(0x12345678, 5913134931067755359633408.0);
 }
 
+//! Prune chain from height down to genesis block and check that
+//! GetPruneHeight returns the correct value
+static void CheckGetPruneHeight(node::BlockManager& blockman, CChain& chain, int height) EXCLUSIVE_LOCKS_REQUIRED(::cs_main)
+{
+    AssertLockHeld(::cs_main);
+
+    // Emulate pruning all blocks from `height` down to the genesis block
+    // by unsetting the `BLOCK_HAVE_DATA` flag from `nStatus`
+    for (CBlockIndex* it{chain[height]}; it != nullptr && it->nHeight > 0; it = it->pprev) {
+        it->nStatus &= ~BLOCK_HAVE_DATA;
+    }
+
+    const auto prune_height{GetPruneHeight(blockman, chain)};
+    BOOST_REQUIRE(prune_height.has_value());
+    BOOST_CHECK_EQUAL(*prune_height, height);
+}
+
+BOOST_FIXTURE_TEST_CASE(get_prune_height, TestChain100Setup)
+{
+    LOCK(::cs_main);
+    auto& chain = m_node.chainman->ActiveChain();
+    auto& blockman = m_node.chainman->m_blockman;
+
+    // Fresh chain of 100 blocks without any pruned blocks, so std::nullopt should be returned
+    BOOST_CHECK(!GetPruneHeight(blockman, chain).has_value());
+
+    // Start pruning
+    CheckGetPruneHeight(blockman, chain, 1);
+    CheckGetPruneHeight(blockman, chain, 99);
+    CheckGetPruneHeight(blockman, chain, 100);
+}
+
 BOOST_AUTO_TEST_SUITE_END()

src/test/blockmanager_tests.cpp

@@ -2,6 +2,7 @@
 // Distributed under the MIT software license, see the accompanying
 // file COPYING or http://www.opensource.org/licenses/mit-license.php.
 
+#include <chain.h>
 #include <chainparams.h>
 #include <clientversion.h>
 #include <node/blockstorage.h>
@@ -113,7 +114,7 @@ BOOST_FIXTURE_TEST_CASE(blockmanager_block_data_availability, TestChain100Setup)
     };
 
     // 1) Return genesis block when all blocks are available
-    BOOST_CHECK_EQUAL(blockman.GetFirstStoredBlock(tip), chainman->ActiveChain()[0]);
+    BOOST_CHECK_EQUAL(blockman.GetFirstBlock(tip, BLOCK_HAVE_DATA), chainman->ActiveChain()[0]);
     BOOST_CHECK(blockman.CheckBlockDataAvailability(tip, *chainman->ActiveChain()[0]));
 
     // 2) Check lower_block when all blocks are available
@@ -127,7 +128,7 @@ BOOST_FIXTURE_TEST_CASE(blockmanager_block_data_availability, TestChain100Setup)
     func_prune_blocks(last_pruned_block);
 
     // 3) The last block not pruned is in-between upper-block and the genesis block
-    BOOST_CHECK_EQUAL(blockman.GetFirstStoredBlock(tip), first_available_block);
+    BOOST_CHECK_EQUAL(blockman.GetFirstBlock(tip, BLOCK_HAVE_DATA), first_available_block);
     BOOST_CHECK(blockman.CheckBlockDataAvailability(tip, *first_available_block));
     BOOST_CHECK(!blockman.CheckBlockDataAvailability(tip, *last_pruned_block));
 }