Merge pull request #1953 from kuzudb/bool-bitpacking-fix

Fix bool column chunk buffer size
kuzudb · Aug 28, 2023 · a1041bf · a1041bf
2 parents feabe82 + 62d50f3
commit a1041bf
Show file tree

Hide file tree

Showing 2 changed files with 24 additions and 16 deletions.
diff --git a/src/include/storage/copier/column_chunk.h b/src/include/storage/copier/column_chunk.h
@@ -186,6 +186,16 @@ class BoolColumnChunk : public ColumnChunk {
 
     void append(ColumnChunk* other, common::offset_t startPosInOtherChunk,
         common::offset_t startPosInChunk, uint32_t numValuesToAppend) final;
+
+    void resize(uint64_t capacity) final;
+
+protected:
+    inline uint64_t numBytesForValues(common::offset_t numValues) const {
+        // 8 values per byte, and we need a buffer size which is a multiple of 8 bytes
+        return ceil(numValues / 8.0 / 8.0) * 8;
+    }
+
+    void initialize(common::offset_t capacity) final;
 };
 
 class NullColumnChunk : public BoolColumnChunk {
@@ -195,21 +205,7 @@ class NullColumnChunk : public BoolColumnChunk {
     inline bool isNull(common::offset_t pos) const { return getValue<bool>(pos); }
     inline void setNull(common::offset_t pos, bool isNull) { setValue(isNull, pos); }
 
-    void resize(uint64_t numValues) final;
-
     inline void resetNullBuffer() { memset(buffer.get(), 0 /* non null */, bufferSize); }
-
-protected:
-    inline uint64_t numBytesForValues(common::offset_t numValues) const {
-        // 8 values per byte, and we need a buffer size which is a multiple of 8 bytes
-        return ceil(numValues / 8.0 / 8.0) * 8;
-    }
-    inline void initialize(common::offset_t numValues) final {
-        numBytesPerValue = 0;
-        bufferSize = numBytesForValues(numValues);
-        // Each byte defaults to 0, indicating everything is non-null
-        buffer = std::make_unique<uint8_t[]>(bufferSize);
-    }
 };
 
 class FixedListColumnChunk : public ColumnChunk {

diff --git a/src/storage/copier/column_chunk.cpp b/src/storage/copier/column_chunk.cpp
@@ -542,14 +542,26 @@ void ColumnChunk::copyVectorToBuffer(
     }
 }
 
-void NullColumnChunk::resize(uint64_t numValues) {
-    auto numBytesAfterResize = numBytesForValues(numValues);
+inline void BoolColumnChunk::initialize(common::offset_t capacity) {
+    numBytesPerValue = 0;
+    bufferSize = numBytesForValues(capacity);
+    buffer = std::make_unique<uint8_t[]>(bufferSize);
+    if (nullChunk) {
+        static_cast<BoolColumnChunk*>(nullChunk.get())->initialize(capacity);
+    }
+}
+
+void BoolColumnChunk::resize(uint64_t capacity) {
+    auto numBytesAfterResize = numBytesForValues(capacity);
     assert(numBytesAfterResize > bufferSize);
     auto reservedBuffer = std::make_unique<uint8_t[]>(numBytesAfterResize);
     memset(reservedBuffer.get(), 0 /* non null */, numBytesAfterResize);
     memcpy(reservedBuffer.get(), buffer.get(), bufferSize);
     buffer = std::move(reservedBuffer);
     bufferSize = numBytesAfterResize;
+    if (nullChunk) {
+        nullChunk->resize(capacity);
+    }
 }
 
 } // namespace storage