enh: row iter and fetch optimization

This commit is contained in:
Cary Xu 2022-08-10 17:26:44 +08:00
parent 6250c1b157
commit 1d6d0aa2db
7 changed files with 108 additions and 183 deletions

View File

@ -200,8 +200,6 @@ struct STag {
#if 1 //================================================================================================================================================
// Imported since 3.0 and use bitmap to demonstrate None/Null/Norm, while use Null/Norm below 3.0 without of bitmap.
#define TD_SUPPORT_BITMAP
#define TD_SUPPORT_READ2
#define TD_SUPPORT_BACK2 // suppport back compatibility of 2.0
#define TASSERT(x) ASSERT(x)

View File

@ -195,6 +195,12 @@ static FORCE_INLINE void tdRowSetVal(SCellVal *pVal, uint8_t valType, void *val)
pVal->valType = valType;
pVal->val = val;
}
/**
* @brief Primary TS column not included.
*
* @param pRow
* @return FORCE_INLINE
*/
static FORCE_INLINE col_id_t tdRowGetNCols(STSRow *pRow) { return *(col_id_t *)TD_ROW_NCOLS(pRow); }
static FORCE_INLINE void tdRowCpy(void *dst, const STSRow *pRow) { memcpy(dst, pRow, TD_ROW_LEN(pRow)); }
static FORCE_INLINE const char *tdRowEnd(STSRow *pRow) { return (const char *)POINTER_SHIFT(pRow, TD_ROW_LEN(pRow)); }
@ -250,7 +256,7 @@ static FORCE_INLINE void *tdGetBitmapAddrTp(STSRow *pRow, uint32_t flen) {
static FORCE_INLINE void *tdGetBitmapAddrKv(STSRow *pRow, col_id_t nKvCols) {
// The primary TS key is stored separatedly and is Norm value, thus should minus 1 firstly
return POINTER_SHIFT(TD_ROW_COL_IDX(pRow), (--nKvCols) * sizeof(SKvRowIdx));
return POINTER_SHIFT(TD_ROW_COL_IDX(pRow), nKvCols * sizeof(SKvRowIdx));
}
void *tdGetBitmapAddr(STSRow *pRow, uint8_t rowType, uint32_t flen, col_id_t nKvCols);
int32_t tdSetBitmapValType(void *pBitmap, int16_t colIdx, TDRowValT valType, int8_t bitmapMode);
@ -322,10 +328,9 @@ typedef struct {
void tdSTSRowIterReset(STSRowIter *pIter, STSRow *pRow);
void tdSTSRowIterInit(STSRowIter *pIter, STSchema *pSchema);
int32_t tdSTSRowNew(SArray *pArray, STSchema *pTSchema, STSRow **ppRow);
bool tdSTSRowGetVal(STSRowIter *pIter, col_id_t colId, col_type_t colType, SCellVal *pVal);
bool tdGetTpRowDataOfCol(STSRowIter *pIter, col_type_t colType, int32_t offset, SCellVal *pVal);
bool tdGetKvRowValOfColEx(STSRowIter *pIter, col_id_t colId, col_type_t colType, col_id_t *nIdx, SCellVal *pVal);
bool tdSTSRowIterNext(STSRowIter *pIter, col_id_t colId, col_type_t colType, SCellVal *pVal);
bool tdGetKvRowValOfColEx(STSRowIter *pIter, col_id_t colId, col_id_t *nIdx, SCellVal *pVal);
bool tdSTSRowIterNext(STSRowIter *pIter, SCellVal *pVal);
bool tdSTpRowGetVal(STSRow *pRow, col_id_t colId, col_type_t colType, int32_t flen, uint32_t offset, col_id_t colIdx,
SCellVal *pVal);
bool tdSKvRowGetVal(STSRow *pRow, col_id_t colId, col_id_t colIdx, SCellVal *pVal);

View File

@ -330,15 +330,17 @@ void tdSRowPrint(STSRow *row, STSchema *pSchema, const char *tag) {
tdSTSRowIterInit(&iter, pSchema);
tdSTSRowIterReset(&iter, row);
printf("%s >>>type:%d,sver:%d ", tag, (int32_t)TD_ROW_TYPE(row), (int32_t)TD_ROW_SVER(row));
for (int i = 0; i < pSchema->numOfCols; ++i) {
STColumn *stCol = pSchema->columns + i;
SCellVal sVal = {255, NULL};
if (!tdSTSRowIterNext(&iter, stCol->colId, stCol->type, &sVal)) {
STColumn *cols = (STColumn *)&iter.pSchema->columns;
while (true) {
SCellVal sVal = {.valType = 255, NULL};
if (!tdSTSRowIterNext(&iter, &sVal)) {
break;
}
ASSERT(sVal.valType == 0 || sVal.valType == 1 || sVal.valType == 2);
tdSCellValPrint(&sVal, stCol->type);
tdSCellValPrint(&sVal, cols[iter.colIdx - 1].type);
}
printf("\n");
}
@ -425,7 +427,7 @@ bool tdSKvRowGetVal(STSRow *pRow, col_id_t colId, col_id_t colIdx, SCellVal *pVa
tdRowSetVal(pVal, TD_VTYPE_NORM, TD_ROW_KEY_ADDR(pRow));
return true;
}
int16_t nCols = tdRowGetNCols(pRow) - 1;
int16_t nCols = tdRowGetNCols(pRow);
if (nCols <= 0) {
pVal->valType = TD_VTYPE_NONE;
return true;
@ -456,46 +458,41 @@ bool tdSTpRowGetVal(STSRow *pRow, col_id_t colId, col_type_t colType, int32_t fl
return true;
}
bool tdSTSRowIterNext(STSRowIter *pIter, col_id_t colId, col_type_t colType, SCellVal *pVal) {
if (colId == PRIMARYKEY_TIMESTAMP_COL_ID) {
bool tdSTSRowIterNext(STSRowIter *pIter, SCellVal *pVal) {
if (pIter->colIdx >= pIter->pSchema->numOfCols) {
return false;
}
STColumn *pCol = &pIter->pSchema->columns[pIter->colIdx];
if (pCol->colId == PRIMARYKEY_TIMESTAMP_COL_ID) {
pVal->val = &pIter->pRow->ts;
pVal->valType = TD_VTYPE_NORM;
++pIter->colIdx;
return true;
}
if (TD_IS_TP_ROW(pIter->pRow)) {
STColumn *pCol = NULL;
STSchema *pSchema = pIter->pSchema;
while (pIter->colIdx < pSchema->numOfCols) {
pCol = &pSchema->columns[pIter->colIdx]; // 1st column of schema is primary TS key
if (colId == pCol->colId) {
break;
} else if (pCol->colId < colId) {
++pIter->colIdx;
continue;
} else {
return false;
}
}
tdGetTpRowDataOfCol(pIter, pCol->type, pCol->offset - sizeof(TSKEY), pVal);
++pIter->colIdx;
} else if (TD_IS_KV_ROW(pIter->pRow)) {
return tdGetKvRowValOfColEx(pIter, colId, colType, &pIter->kvIdx, pVal);
tdGetKvRowValOfColEx(pIter, pCol->colId, &pIter->kvIdx, pVal);
} else {
pVal->valType = TD_VTYPE_NONE;
terrno = TSDB_CODE_INVALID_PARA;
if (COL_REACH_END(colId, pIter->maxColId)) return false;
ASSERT(0);
}
++pIter->colIdx;
return true;
}
bool tdGetKvRowValOfColEx(STSRowIter *pIter, col_id_t colId, col_type_t colType, col_id_t *nIdx, SCellVal *pVal) {
bool tdGetKvRowValOfColEx(STSRowIter *pIter, col_id_t colId, col_id_t *nIdx, SCellVal *pVal) {
STSRow *pRow = pIter->pRow;
SKvRowIdx *pKvIdx = NULL;
bool colFound = false;
col_id_t kvNCols = tdRowGetNCols(pRow) - 1;
col_id_t kvNCols = tdRowGetNCols(pRow);
void *pColIdx = TD_ROW_COL_IDX(pRow);
while (*nIdx < kvNCols) {
pKvIdx = (SKvRowIdx *)POINTER_SHIFT(TD_ROW_COL_IDX(pRow), *nIdx * sizeof(SKvRowIdx));
pKvIdx = (SKvRowIdx *)POINTER_SHIFT(pColIdx, *nIdx * sizeof(SKvRowIdx));
if (pKvIdx->colId == colId) {
++(*nIdx);
pVal->val = POINTER_SHIFT(pRow, pKvIdx->offset);
@ -518,34 +515,37 @@ bool tdGetKvRowValOfColEx(STSRowIter *pIter, col_id_t colId, col_type_t colType,
}
}
#ifdef TD_SUPPORT_BITMAP
int16_t colIdx = -1;
if (pKvIdx) colIdx = POINTER_DISTANCE(pKvIdx, TD_ROW_COL_IDX(pRow)) / sizeof(SKvRowIdx);
if (tdGetBitmapValType(pIter->pBitmap, colIdx, &pVal->valType, 0) != TSDB_CODE_SUCCESS) {
if (tdGetBitmapValType(pIter->pBitmap, (*nIdx) - 1, &pVal->valType, 0) != TSDB_CODE_SUCCESS) {
pVal->valType = TD_VTYPE_NONE;
}
#else
pVal->valType = isNull(pVal->val, colType) ? TD_VTYPE_NULL : TD_VTYPE_NORM;
#endif
return true;
}
bool tdGetTpRowDataOfCol(STSRowIter *pIter, col_type_t colType, int32_t offset, SCellVal *pVal) {
STSRow *pRow = pIter->pRow;
if (pRow->statis == 0) {
pVal->valType = TD_VTYPE_NORM;
if (IS_VAR_DATA_TYPE(colType)) {
pVal->val = POINTER_SHIFT(pRow, *(VarDataOffsetT *)POINTER_SHIFT(TD_ROW_DATA(pRow), offset));
} else {
pVal->val = POINTER_SHIFT(TD_ROW_DATA(pRow), offset);
}
return TSDB_CODE_SUCCESS;
}
if (tdGetBitmapValType(pIter->pBitmap, pIter->colIdx, &pVal->valType, 0) != TSDB_CODE_SUCCESS) {
pVal->valType = TD_VTYPE_NONE;
return terrno;
}
if (pVal->valType == TD_VTYPE_NORM) {
if (IS_VAR_DATA_TYPE(colType)) {
pVal->val = POINTER_SHIFT(pRow, *(VarDataOffsetT *)POINTER_SHIFT(TD_ROW_DATA(pRow), offset));
} else {
pVal->val = POINTER_SHIFT(TD_ROW_DATA(pRow), offset);
}
#ifdef TD_SUPPORT_BITMAP
if (tdGetBitmapValType(pIter->pBitmap, pIter->colIdx - 1, &pVal->valType, 0) != TSDB_CODE_SUCCESS) {
pVal->valType = TD_VTYPE_NONE;
}
#else
pVal->valType = isNull(pVal->val, colType) ? TD_VTYPE_NULL : TD_VTYPE_NORM;
#endif
return true;
}
@ -673,45 +673,6 @@ int32_t tdSTSRowNew(SArray *pArray, STSchema *pTSchema, STSRow **ppRow) {
return 0;
}
bool tdSTSRowGetVal(STSRowIter *pIter, col_id_t colId, col_type_t colType, SCellVal *pVal) {
if (colId == PRIMARYKEY_TIMESTAMP_COL_ID) {
pVal->val = &pIter->pRow->ts;
pVal->valType = TD_VTYPE_NORM;
return true;
}
STSRow *pRow = pIter->pRow;
int16_t colIdx = -1;
if (TD_IS_TP_ROW(pRow)) {
STSchema *pSchema = pIter->pSchema;
STColumn *pCol =
(STColumn *)taosbsearch(&colId, pSchema->columns, pSchema->numOfCols, sizeof(STColumn), tdCompareColId, TD_EQ);
if (!pCol) {
pVal->valType = TD_VTYPE_NONE;
if (COL_REACH_END(colId, pIter->maxColId)) return false;
return true;
}
#ifdef TD_SUPPORT_BITMAP
colIdx = POINTER_DISTANCE(pCol, pSchema->columns) / sizeof(STColumn);
#endif
tdGetTpRowValOfCol(pVal, pRow, pIter->pBitmap, pCol->type, pCol->offset - sizeof(TSKEY), colIdx - 1);
} else if (TD_IS_KV_ROW(pRow)) {
SKvRowIdx *pIdx = (SKvRowIdx *)taosbsearch(&colId, TD_ROW_COL_IDX(pRow), tdRowGetNCols(pRow), sizeof(SKvRowIdx),
compareKvRowColId, TD_EQ);
#ifdef TD_SUPPORT_BITMAP
if (pIdx) {
colIdx = POINTER_DISTANCE(pIdx, TD_ROW_COL_IDX(pRow)) / sizeof(SKvRowIdx);
}
#endif
tdGetKvRowValOfCol(pVal, pRow, pIter->pBitmap, pIdx ? pIdx->offset : -1, colIdx);
} else {
if (COL_REACH_END(colId, pIter->maxColId)) return false;
pVal->valType = TD_VTYPE_NONE;
}
return true;
}
static int32_t tdCompareColId(const void *arg1, const void *arg2) {
int32_t colId = *(int32_t *)arg1;
STColumn *pCol = (STColumn *)arg2;
@ -853,7 +814,7 @@ int32_t tdSetBitmapValTypeI(void *pBitmap, int16_t colIdx, TDRowValT valType) {
int32_t tdGetKvRowValOfCol(SCellVal *output, STSRow *pRow, void *pBitmap, int32_t offset, int16_t colIdx) {
#ifdef TD_SUPPORT_BITMAP
TASSERT(colIdx < tdRowGetNCols(pRow) - 1);
TASSERT(colIdx < tdRowGetNCols(pRow));
if (tdGetBitmapValType(pBitmap, colIdx, &output->valType, 0) != TSDB_CODE_SUCCESS) {
output->valType = TD_VTYPE_NONE;
return terrno;
@ -938,7 +899,7 @@ int32_t tdAppendColValToRow(SRowBuilder *pBuilder, col_id_t colId, int8_t colTyp
break;
case TD_VTYPE_NONE:
if (!pBuilder->hasNone) pBuilder->hasNone = true;
break;
return TSDB_CODE_SUCCESS;
default:
ASSERT(0);
break;
@ -970,13 +931,11 @@ int32_t tdAppendColValToKvRow(SRowBuilder *pBuilder, TDRowValT valType, const vo
STSRow *row = pBuilder->pBuf;
// No need to store None/Null values.
if (valType == TD_VTYPE_NORM) {
// ts key stored in STSRow.ts
SKvRowIdx *pColIdx = (SKvRowIdx *)POINTER_SHIFT(TD_ROW_COL_IDX(row), offset);
char *ptr = (char *)POINTER_SHIFT(row, TD_ROW_LEN(row));
pColIdx->colId = colId;
pColIdx->offset = TD_ROW_LEN(row); // the offset include the TD_ROW_HEAD_LEN
if (valType == TD_VTYPE_NORM) {
char *ptr = (char *)POINTER_SHIFT(row, TD_ROW_LEN(row));
if (IS_VAR_DATA_TYPE(colType)) {
if (isCopyVarData) {
memcpy(ptr, val, varDataTLen(val));
@ -987,26 +946,6 @@ int32_t tdAppendColValToKvRow(SRowBuilder *pBuilder, TDRowValT valType, const vo
TD_ROW_LEN(row) += TYPE_BYTES[colType];
}
}
#ifdef TD_SUPPORT_BACK2
// NULL/None value
else {
SKvRowIdx *pColIdx = (SKvRowIdx *)POINTER_SHIFT(TD_ROW_COL_IDX(row), offset);
char *ptr = (char *)POINTER_SHIFT(row, TD_ROW_LEN(row));
pColIdx->colId = colId;
pColIdx->offset = TD_ROW_LEN(row); // the offset include the TD_ROW_HEAD_LEN
const void *nullVal = getNullValue(colType);
if (IS_VAR_DATA_TYPE(colType)) {
if (isCopyVarData) {
memcpy(ptr, nullVal, varDataTLen(nullVal));
}
TD_ROW_LEN(row) += varDataTLen(nullVal);
} else {
memcpy(ptr, nullVal, TYPE_BYTES[colType]);
TD_ROW_LEN(row) += TYPE_BYTES[colType];
}
}
#endif
return 0;
}
@ -1044,24 +983,6 @@ int32_t tdAppendColValToTpRow(SRowBuilder *pBuilder, TDRowValT valType, const vo
memcpy(POINTER_SHIFT(TD_ROW_DATA(row), offset), val, TYPE_BYTES[colType]);
}
}
#ifdef TD_SUPPORT_BACK2
// NULL/None value
else {
// TODO: Null value for new data types imported since 3.0 need to be defined.
const void *nullVal = getNullValue(colType);
if (IS_VAR_DATA_TYPE(colType)) {
// ts key stored in STSRow.ts
*(VarDataOffsetT *)POINTER_SHIFT(TD_ROW_DATA(row), offset) = TD_ROW_LEN(row);
if (isCopyVarData) {
memcpy(POINTER_SHIFT(row, TD_ROW_LEN(row)), nullVal, varDataTLen(nullVal));
}
TD_ROW_LEN(row) += varDataTLen(nullVal);
} else {
memcpy(POINTER_SHIFT(TD_ROW_DATA(row), offset), nullVal, TYPE_BYTES[colType]);
}
}
#endif
return 0;
}
@ -1133,14 +1054,14 @@ int32_t tdSRowResetBuf(SRowBuilder *pBuilder, void *pBuf) {
break;
case TD_ROW_KV:
#ifdef TD_SUPPORT_BITMAP
pBuilder->pBitmap = tdGetBitmapAddrKv(pBuilder->pBuf, pBuilder->nBoundCols);
pBuilder->pBitmap = tdGetBitmapAddrKv(pBuilder->pBuf, pBuilder->nBoundCols - 1);
memset(pBuilder->pBitmap, TD_VTYPE_NONE_BYTE_II, pBuilder->nBoundBitmaps);
#endif
len = TD_ROW_HEAD_LEN + TD_ROW_NCOLS_LEN + (pBuilder->nBoundCols - 1) * sizeof(SKvRowIdx) +
pBuilder->nBoundBitmaps; // add
TD_ROW_SET_LEN(pBuilder->pBuf, len);
TD_ROW_SET_SVER(pBuilder->pBuf, pBuilder->sver);
TD_ROW_SET_NCOLS(pBuilder->pBuf, pBuilder->nBoundCols);
TD_ROW_SET_NCOLS(pBuilder->pBuf, pBuilder->nBoundCols - 1);
break;
default:
TASSERT(0);
@ -1329,7 +1250,7 @@ void tdSTSRowIterReset(STSRowIter *pIter, STSRow *pRow) {
pIter->pRow = pRow;
pIter->pBitmap = tdGetBitmapAddr(pRow, pRow->type, pIter->pSchema->flen, tdRowGetNCols(pRow));
pIter->offset = 0;
pIter->colIdx = PRIMARYKEY_TIMESTAMP_COL_ID;
pIter->colIdx = 0;
pIter->kvIdx = 0;
}

View File

@ -325,7 +325,7 @@ int32_t tqRetrieveDataBlock(SSDataBlock* pBlock, STqReader* pReader) {
for (int32_t i = 0; i < colActual; i++) {
SColumnInfoData* pColData = taosArrayGet(pBlock->pDataBlock, i);
SCellVal sVal = {0};
if (!tdSTSRowIterNext(&iter, pColData->info.colId, pColData->info.type, &sVal)) {
if (!tdSTSRowIterNext(&iter, &sVal)) {
break;
}
if (colDataAppend(pColData, curRow, sVal.val, sVal.valType != TD_VTYPE_NORM) < 0) {

View File

@ -182,7 +182,7 @@ static int32_t setColumnIdSlotList(STsdbReader* pReader, SSDataBlock* pBlock) {
if (IS_VAR_DATA_TYPE(pCol->info.type)) {
pSupInfo->buildBuf[i] = taosMemoryMalloc(pCol->info.bytes);
tsdbInfo("-------------------%d\n", pCol->info.bytes);
// tsdbInfo("-------------------%d\n", pCol->info.bytes);
}
}

View File

@ -13,8 +13,8 @@
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include "os.h"
#include "tsimplehash.h"
#include "os.h"
#include "taoserror.h"
#define SHASH_DEFAULT_LOAD_FACTOR 0.75
@ -25,7 +25,6 @@
#define GET_SHASH_NODE_DATA(_n) ((char *)(_n) + sizeof(SHNode))
#define HASH_INDEX(v, c) ((v) & ((c)-1))
#define HASH_NEED_RESIZE(_h) ((_h)->size >= (_h)->capacity * SHASH_DEFAULT_LOAD_FACTOR)
#define FREE_HASH_NODE(_n) \
do { \
@ -108,7 +107,7 @@ static SHNode *doCreateHashNode(const void *key, size_t keyLen, const void *pDat
}
static void taosHashTableResize(SSHashObj *pHashObj) {
if (!HASH_NEED_RESIZE(pHashObj)) {
if (!SHASH_NEED_RESIZE(pHashObj)) {
return;
}
@ -134,13 +133,14 @@ static void taosHashTableResize(SSHashObj *pHashObj) {
for (int32_t idx = 0; idx < pHashObj->capacity; ++idx) {
SHNode *pNode = pHashObj->hashList[idx];
SHNode *pNext;
SHNode *pPrev = NULL;
if (pNode == NULL) {
continue;
}
SHNode *pNext;
SHNode *pPrev = NULL;
while (pNode != NULL) {
void *key = GET_SHASH_NODE_KEY(pNode, pHashObj->dataLen);
uint32_t hashVal = (*pHashObj->hashFp)(key, (uint32_t)pHashObj->dataLen);
@ -166,7 +166,8 @@ static void taosHashTableResize(SSHashObj *pHashObj) {
int64_t et = taosGetTimestampUs();
// uDebug("hash table resize completed, new capacity:%d, load factor:%f, elapsed time:%fms", (int32_t)pHashObj->capacity,
// uDebug("hash table resize completed, new capacity:%d, load factor:%f, elapsed time:%fms",
// (int32_t)pHashObj->capacity,
// ((double)pHashObj->size) / pHashObj->capacity, (et - st) / 1000.0);
}
@ -230,9 +231,7 @@ static FORCE_INLINE SHNode *doSearchInEntryList(SSHashObj *pHashObj, const void
return pNode;
}
static FORCE_INLINE bool taosHashTableEmpty(const SSHashObj *pHashObj) {
return tSimpleHashGetSize(pHashObj) == 0;
}
static FORCE_INLINE bool taosHashTableEmpty(const SSHashObj *pHashObj) { return tSimpleHashGetSize(pHashObj) == 0; }
void *tSimpleHashGet(SSHashObj *pHashObj, const void *key) {
if (pHashObj == NULL || taosHashTableEmpty(pHashObj) || key == NULL) {

View File

@ -1192,7 +1192,10 @@ static int parseOneRow(SInsertParseContext* pCxt, STableDataBlocks* pDataBlocks,
pBuilder->hasNone = true;
}
tdSRowEnd(pBuilder);
*gotRow = true;
#ifdef TD_DEBUG_PRINT_ROW
STSchema* pSTSchema = tdGetSTSChemaFromSSChema(schema, spd->numOfCols, 1);
tdSRowPrint(row, pSTSchema, __func__);
@ -1201,7 +1204,6 @@ static int parseOneRow(SInsertParseContext* pCxt, STableDataBlocks* pDataBlocks,
}
// *len = pBuilder->extendedRowSize;
tdSRowEnd(pBuilder);
return TSDB_CODE_SUCCESS;
}