FIXME enable 64bit bitmapset and update visimap (#14784)

It includes the portion about enabling 64bit bms and other necessary
changes related to visimap, plus updates of cmockery unit tests.

Author: yaowangm

src/backend/access/appendonly/appendonly_visimap_entry.c

@@ -129,7 +129,10 @@ AppendOnlyVisimap_GetAttrNotNull(HeapTuple t, TupleDesc td, int attr)
 void
 AppendOnlyVisiMapEnty_ReadData(AppendOnlyVisimapEntry *visiMapEntry, size_t dataSize)
 {
-	int			newWordCount;
+	/* the block count of (ondisk) bitstream */
+	int			onDiskBlockCount;
+	/* the word count of in-memory bitmapset */
+	int			bmsWordCount;
 
 	Assert(visiMapEntry);
 	Assert(CurrentMemoryContext == visiMapEntry->memoryContext);
@@ -156,21 +159,23 @@ AppendOnlyVisiMapEnty_ReadData(AppendOnlyVisimapEntry *visiMapEntry, size_t data
 	 * but I think it is reasonable to set it to NULLL to avoid similar issues.
 	 */
 	visiMapEntry->bitmap = NULL;
-	newWordCount =
-		BitmapDecompress_GetBlockCount(&decompressState);
-	if (newWordCount > 0)
+	BitmapDecompress_CalculateBlockCounts(&decompressState,
+										  &onDiskBlockCount,
+										  &bmsWordCount);
+
+	if (onDiskBlockCount > 0)
 	{
 		visiMapEntry->bitmap = palloc0(offsetof(Bitmapset, words) +
-									   (newWordCount * sizeof(bitmapword)));
-		visiMapEntry->bitmap->nwords = newWordCount;
+									   (bmsWordCount * sizeof(bitmapword)));
+		visiMapEntry->bitmap->nwords = bmsWordCount;
 		BitmapDecompress_Decompress(&decompressState,
-									visiMapEntry->bitmap->words,
-									newWordCount);
+									(uint32 *)visiMapEntry->bitmap->words,
+									onDiskBlockCount);
 	}
-	else if (newWordCount != 0)
+	else if (onDiskBlockCount != 0)
 	{
 		elog(ERROR,
-			 "illegal visimap block count: visimap block count %d", newWordCount);
+			 "illegal visimap block count: visimap block count %d", onDiskBlockCount);
 	}
 
 }
@@ -264,23 +269,32 @@ AppendOnlyVisimapEntry_GetHiddenTupleCount(
 void
 AppendOnlyVisimapEntry_WriteData(AppendOnlyVisimapEntry *visiMapEntry)
 {
-	int			bitmapSize,
-				compressedBitmapSize;
+	/* bitmap size, in bytes */
+	int	bitmapSize;
+	int	compressedBitmapSize;
+	/* word count in 64bit or 32bit words for in-memory bms */
+	int	bmsWordCount;
+	/* block count always in 32bit (after conversion if necessary) */
+	int	blockCount;
 
 	Assert(visiMapEntry);
 	Assert(CurrentMemoryContext == visiMapEntry->memoryContext);
 	Assert(AppendOnlyVisimapEntry_IsValid(visiMapEntry));
 
-	bitmapSize = (visiMapEntry->bitmap ? (visiMapEntry->bitmap->nwords * sizeof(bitmapword)) : 0);
+	BitmapCompress_CalculateBlockCounts(visiMapEntry->bitmap,
+										&blockCount,
+										&bmsWordCount);
+	bitmapSize = sizeof(uint32) * blockCount;
 	bitmapSize += BITMAP_COMPRESSION_HEADER_SIZE;
+	Assert(bmsWordCount <= APPENDONLY_VISIMAP_MAX_BITMAP_WORD_COUNT);
 
 	Assert(visiMapEntry->data);
 	Assert(APPENDONLY_VISIMAP_DATA_BUFFER_SIZE >= bitmapSize);
 	visiMapEntry->data->version = 1;
 
 	compressedBitmapSize = Bitmap_Compress(BITMAP_COMPRESSION_TYPE_DEFAULT,
-										   (visiMapEntry->bitmap ? visiMapEntry->bitmap->words : NULL),
-										   (visiMapEntry->bitmap ? visiMapEntry->bitmap->nwords : 0),
+										   (visiMapEntry->bitmap ? (uint32*)visiMapEntry->bitmap->words : NULL),
+										   blockCount,
 										   visiMapEntry->data->data,
 										   bitmapSize);
 	Assert(compressedBitmapSize >= BITMAP_COMPRESSION_HEADER_SIZE);
@@ -499,6 +513,9 @@ AppendOnlyVisimapEntry_GetMinimalSizeToCover(int64 offset)
 	minSize |= minSize >> 8;
 	minSize |= minSize >> 16;
 	minSize++;
+
+	Assert(minSize <= APPENDONLY_VISIMAP_MAX_BITMAP_WORD_COUNT);
+
 	return minSize;
 }
 

src/backend/nodes/bitmapset.c

@@ -29,9 +29,6 @@
 #define WORDNUM(x)	((x) / BITS_PER_BITMAPWORD)
 #define BITNUM(x)	((x) % BITS_PER_BITMAPWORD)
 
-#define BITMAPSET_SIZE(nwords)	\
-	(offsetof(Bitmapset, words) + (nwords) * sizeof(bitmapword))
-
 /*----------
  * This is a well-known cute trick for isolating the rightmost one-bit
  * in a word.  It assumes two's complement arithmetic.  Consider any

src/backend/utils/misc/bitmap_compression.c

@@ -15,6 +15,7 @@
 #include "utils/bitmap_compression.h"
 #include "utils/bitstream.h"
 #include "utils/guc.h"
+#include "access/appendonly_visimap.h"
 
 typedef enum BitmapCompressionFlag
 {
@@ -80,9 +81,15 @@ BitmapDecompress_HasError(
 }
 
 /*
- * Performs the bitmap decompression.
+ * Perform bitmap decompression into in-memory buffer
  *
- * bitmapDataSize in uint32-words.
+ * bitmap: caller-allocated buffer that can hold state->blockCount
+ * number of 32-bit on-disk bitmap words.
+ *
+ * For both 32-bit and 64-bit in-memory bitmap word sizes, we write
+ * 32-bit words into the in-memory buffer contiguously. This is safe
+ * to do as we interpret two contiguous 32-bit words as one 64-bit
+ * word.
  */
 void
 BitmapDecompress_Decompress(BitmapDecompressState *state,
@@ -105,7 +112,7 @@ BitmapDecompress_Decompress(BitmapDecompressState *state,
 
 	if (state->compressionType == BITMAP_COMPRESSION_TYPE_NO)
 	{
-		memcpy(bitmap, 
+		memcpy(bitmap,
 				Bitstream_GetAlignedData(&state->bitstream, 16), 
 				state->blockCount * sizeof(uint32));
 	}
@@ -357,3 +364,104 @@ Bitmap_Compress(
 			return 0;
 	}
 }
+
+/*
+ * Calculate two counts for decompress:
+ * 1. 'onDiskBlockCount': the block count of (ondisk) bitstream
+ * 2. 'bmsWordCount': the word count of in-memory bitmapset
+ */
+void BitmapDecompress_CalculateBlockCounts(BitmapDecompressState *decompressState,
+										   int *onDiskBlockCount,
+										   int *bmsWordCount)
+{
+	*onDiskBlockCount =
+		BitmapDecompress_GetBlockCount(decompressState);
+
+	/* The on-disk bitmap representation always uses 32-bit block size
+	 * (for backward compatibility). Depending on the environment, we
+	 * may be using either 64-bit words or 32-bit words for the
+	 * in-memory representation.
+	 * So, if (in-memory) bitmapset uses 64 bit words, we can use half
+	 * of the on-disk bitmap block count.
+	 */
+	if (BITS_PER_BITMAPWORD == 64)
+	{
+		/*
+		 * Number of on-disk blocks is always 0, 1 or even.
+		 * See resizing logic in AppendOnlyVisimapEntry_HideTuple()
+		 */
+		if (*onDiskBlockCount == 1)
+			*bmsWordCount = 1;
+		else
+		{
+			Assert(*onDiskBlockCount % 2 == 0);
+			*bmsWordCount = *onDiskBlockCount / 2;
+		}
+	}
+	else
+	{
+		Assert(BITS_PER_BITMAPWORD == 32);
+		*bmsWordCount = *onDiskBlockCount;
+	}
+	Assert(*bmsWordCount <= APPENDONLY_VISIMAP_MAX_BITMAP_WORD_COUNT);
+	Assert(*bmsWordCount >= 0);
+}
+
+/*
+ * Calculate two counts for compress:
+ * 1. 'onDiskBlockCount': the block count of (ondisk) bitstream
+ * 2. 'bmsWordCount': the word count of in-memory bitmapset
+ */
+void BitmapCompress_CalculateBlockCounts(Bitmapset *bitmap,
+										 int *onDiskBlockCount,
+										 int *bmsWordCount)
+{
+	*onDiskBlockCount = 0;
+	*bmsWordCount = 0;
+
+	if (bitmap)
+	{
+		*bmsWordCount = bitmap->nwords;
+
+		/*
+		 * On 64bit env, there is a conflict: in-memory bms is in 64bit word,
+		 * but on-disk block is in 32bit word to keep consistency. We need to
+		 * provide 32bit block count to Bitmap_Compress() after kind of
+		 * conversion.
+		 */
+		if (BITS_PER_BITMAPWORD == 64)
+		{
+			/*
+			 * On 64bit env, if there is only one 64 bit word in memory, and the
+			 * 32 higher order bits of that word are all zero, it implies that
+			 * there is only one 32 bit word. We can always assume that the 32
+			 * higher order bits for a 64 bit bitmap word is zeroed out - this
+			 * is ensured by routines such as bms_add_member() and
+			 * AppendOnlyVisiMapEnty_ReadData().
+			 */
+			if (*bmsWordCount == 1
+				&& (bitmap->words[0] >> 32) == 0)
+			{
+				*onDiskBlockCount = 1;
+			}
+			else
+			{
+				/*
+				 * onDiskBlockCount required by Bitmap_Compress() is always in
+				 * uint32-words. So, if bitmapset uses 64 bit words, double
+				 * the value of bmsWordCount.
+				 */
+				*onDiskBlockCount = bitmap->nwords * 2;
+			}
+		}
+		else
+		{
+			Assert(BITS_PER_BITMAPWORD == 32);
+
+			/*
+			 * On 32bit env, onDiskBlockCount is always equal to bmsWordCount.
+			 */
+			*onDiskBlockCount = bitmap->nwords;
+		}
+	}
+}