[td-225]opt query perf.

src/client/src/tscSQLParser.c

@@ -2471,7 +2471,7 @@ static bool functionCompatibleCheck(SQueryInfo* pQueryInfo) {
     startIdx++;
   }
 
-  int32_t factor = funcCompatDefList[tscSqlExprGet(pQueryInfo, startIdx)->functionId];
+  int32_t factor = functionCompatList[tscSqlExprGet(pQueryInfo, startIdx)->functionId];
 
   // diff function cannot be executed with other function
   // arithmetic function can be executed with other arithmetic functions
@@ -2489,7 +2489,7 @@ static bool functionCompatibleCheck(SQueryInfo* pQueryInfo) {
       continue;
     }
 
-    if (funcCompatDefList[functionId] != factor) {
+    if (functionCompatList[functionId] != factor) {
       return false;
     }
   }

src/query/inc/qExecutor.h

@@ -121,6 +121,7 @@ typedef struct SQueryCostInfo {
   uint32_t loadBlockStatis;
   uint32_t discardBlocks;
   uint64_t elapsedTime;
+  uint64_t ioTime;
   uint64_t computTime;
 } SQueryCostInfo;
 

src/query/inc/tsqlfunction.h

@@ -224,25 +224,14 @@ int32_t getResultDataInfo(int32_t dataType, int32_t dataBytes, int32_t functionI
 #define IS_SINGLEOUTPUT(x)        (((x)&TSDB_FUNCSTATE_SO) != 0)
 #define IS_OUTER_FORWARD(x)       (((x)&TSDB_FUNCSTATE_OF) != 0)
 
-/*
- * the status of one block, used in metric query. all blocks are mixed together,
- * we need the status to decide if one block is a first/end/inter block of one meter
- */
-enum {
-  BLK_FILE_BLOCK = 0x1,
-  BLK_BLOCK_LOADED = 0x2,
-  BLK_CACHE_BLOCK = 0x4,  // in case of cache block, block must be loaded
-};
-
 /* determine the real data need to calculated the result */
 enum {
-  BLK_DATA_NO_NEEDED = 0x0,
+  BLK_DATA_NO_NEEDED     = 0x0,
   BLK_DATA_STATIS_NEEDED = 0x1,
-  BLK_DATA_ALL_NEEDED = 0x3,
+  BLK_DATA_ALL_NEEDED    = 0x3,
+  BLK_DATA_DISCARD       = 0x4,   // discard current data block since it is not qualified for filter
 };
 
-#define SET_DATA_BLOCK_NOT_LOADED(x) ((x) &= (~BLK_BLOCK_LOADED));
-
 typedef struct STwaInfo {
   TSKEY   lastKey;
   int8_t  hasResult;  // flag to denote has value
@@ -264,12 +253,9 @@ typedef struct STwaInfo {
 /* global sql function array */
 extern struct SQLAggFuncElem aAggs[];
 
-/* compatible check array list */
-extern int32_t funcCompatDefList[];
-
-bool top_bot_datablock_filter(SQLFunctionCtx *pCtx, int32_t functionId, char *minval, char *maxval);
+extern int32_t functionCompatList[]; // compatible check array list
 
-bool stableQueryFunctChanged(int32_t funcId);
+bool topbot_datablock_filter(SQLFunctionCtx *pCtx, int32_t functionId, const char *minval, const char *maxval);
 
 void resetResultInfo(SResultInfo *pResInfo);
 void setResultInfoBuf(SResultInfo *pResInfo, int32_t size, bool superTable, char* buf);

src/query/src/qExecutor.c

@@ -1928,73 +1928,45 @@ char *getPosInResultPage(SQueryRuntimeEnv *pRuntimeEnv, int32_t columnIndex, SWi
          pQuery->pSelectExpr[columnIndex].bytes * realRowId;
 }
 
-/**
- * decrease the refcount for each table involved in this query
- * @param pQInfo
- */
-UNUSED_FUNC void vnodeDecMeterRefcnt(SQInfo *pQInfo) {
-  if (pQInfo != NULL) {
-    //    assert(taosHashGetSize(pQInfo->tableqinfoGroupInfo) >= 1);
-  }
-
-#if 0
-  if (pQInfo == NULL || pQInfo->tableqinfoGroupInfo.numOfTables == 1) {
-    atomic_fetch_sub_32(&pQInfo->pObj->numOfQueries, 1);
-    qDebug("QInfo:%p vid:%d sid:%d meterId:%s, query is over, numOfQueries:%d", pQInfo, pQInfo->pObj->vnode,
-           pQInfo->pObj->sid, pQInfo->pObj->meterId, pQInfo->pObj->numOfQueries);
-  } else {
-    int32_t num = 0;
-    for (int32_t i = 0; i < pQInfo->tableqinfoGroupInfo.numOfTables; ++i) {
-      SMeterObj *pMeter = getMeterObj(pQInfo->tableqinfoGroupInfo, pQInfo->pSidSet->pTableIdList[i]->sid);
-      atomic_fetch_sub_32(&(pMeter->numOfQueries), 1);
-
-      if (pMeter->numOfQueries > 0) {
-        qDebug("QInfo:%p vid:%d sid:%d meterId:%s, query is over, numOfQueries:%d", pQInfo, pMeter->vnode, pMeter->sid,
-               pMeter->meterId, pMeter->numOfQueries);
-        num++;
-      }
-    }
-
-    /*
-     * in order to reduce log output, for all meters of which numOfQueries count are 0,
-     * we do not output corresponding information
-     */
-    num = pQInfo->tableqinfoGroupInfo.numOfTables - num;
-    qDebug("QInfo:%p metric query is over, dec query ref for %d meters, numOfQueries on %d meters are 0", pQInfo,
-           pQInfo->tableqinfoGroupInfo.numOfTables, num);
-  }
-#endif
-}
+#define IS_PREFILTER_TYPE(_t) ((_t) != TSDB_DATA_TYPE_DOUBLE && (_t) != TSDB_DATA_TYPE_FLOAT)
 
-static bool needToLoadDataBlock(SQuery *pQuery, SDataStatis *pDataStatis, SQLFunctionCtx *pCtx,
-                                int32_t numOfTotalPoints) {
-  if (pDataStatis == NULL) {
+static bool needToLoadDataBlock(SQueryRuntimeEnv* pRuntimeEnv, SDataStatis *pDataStatis, SQLFunctionCtx *pCtx,
+    int32_t numOfRows) {
+  SQuery* pQuery = pRuntimeEnv->pQuery;
+  if (pDataStatis == NULL || (pQuery->numOfFilterCols == 0 && (!pRuntimeEnv->topBotQuery))) {
     return true;
   }
 
-#if 0
   for (int32_t k = 0; k < pQuery->numOfFilterCols; ++k) {
     SSingleColumnFilterInfo *pFilterInfo = &pQuery->pFilterInfo[k];
-    int32_t                  colIndex = pFilterInfo->info.colIndex;
 
-    // this column not valid in current data block
-    if (colIndex < 0 || pDataStatis[colIndex].colId != pFilterInfo->info.data.colId) {
+    int32_t index = -1;
+    for(int32_t i = 0; i < pQuery->numOfCols; ++i) {
+      if (pDataStatis[i].colId == pFilterInfo->info.colId) {
+        index = i;
+        break;
+      }
+    }
+
+    if (index == -1) {
       continue;
     }
 
     // not support pre-filter operation on binary/nchar data type
-    if (!vnodeSupportPrefilter(pFilterInfo->info.data.type)) {
+    if (!IS_PREFILTER_TYPE(pFilterInfo->info.type)) {
       continue;
     }
 
     // all points in current column are NULL, no need to check its boundary value
-    if (pDataStatis[colIndex].numOfNull == numOfTotalPoints) {
+    if (pDataStatis[index].numOfNull == numOfRows) {
       continue;
     }
 
-    if (pFilterInfo->info.info.type == TSDB_DATA_TYPE_FLOAT) {
-      float minval = *(double *)(&pDataStatis[colIndex].min);
-      float maxval = *(double *)(&pDataStatis[colIndex].max);
+    SDataStatis* pDataBlockst = &pDataStatis[index];
+
+    if (pFilterInfo->info.type == TSDB_DATA_TYPE_FLOAT) {
+      float minval = *(double *)(&pDataBlockst->min);
+      float maxval = *(double *)(&pDataBlockst->max);
 
       for (int32_t i = 0; i < pFilterInfo->numOfFilters; ++i) {
         if (pFilterInfo->pFilters[i].fp(&pFilterInfo->pFilters[i], (char *)&minval, (char *)&maxval)) {
@@ -2003,53 +1975,50 @@ static bool needToLoadDataBlock(SQuery *pQuery, SDataStatis *pDataStatis, SQLFun
       }
     } else {
       for (int32_t i = 0; i < pFilterInfo->numOfFilters; ++i) {
-        if (pFilterInfo->pFilters[i].fp(&pFilterInfo->pFilters[i], (char *)&pDataStatis[colIndex].min,
-                                        (char *)&pDataStatis[colIndex].max)) {
+        if (pFilterInfo->pFilters[i].fp(&pFilterInfo->pFilters[i], (char *)&pDataBlockst->min, (char *)&pDataBlockst->max)) {
           return true;
         }
       }
     }
   }
 
-  // todo disable this opt code block temporarily
-  //  for (int32_t i = 0; i < pQuery->numOfOutput; ++i) {
-  //    int32_t functId = pQuery->pSelectExpr[i].base.functionId;
-  //    if (functId == TSDB_FUNC_TOP || functId == TSDB_FUNC_BOTTOM) {
-  //      return top_bot_datablock_filter(&pCtx[i], functId, (char *)&pField[i].min, (char *)&pField[i].max);
-  //    }
-  //  }
+  if (pRuntimeEnv->topBotQuery) {
+    for (int32_t i = 0; i < pQuery->numOfOutput; ++i) {
+      int32_t functionId = pQuery->pSelectExpr[i].base.functionId;
+      if (functionId == TSDB_FUNC_TOP || functionId == TSDB_FUNC_BOTTOM) {
+        return topbot_datablock_filter(&pCtx[i], functionId, (char *)&pDataStatis[i].min, (char *)&pDataStatis[i].max);
+      }
+    }
+  }
 
-#endif
-  return true;
+  return false;
 }
 
-SArray *loadDataBlockOnDemand(SQueryRuntimeEnv *pRuntimeEnv, void* pQueryHandle, SDataBlockInfo* pBlockInfo, SDataStatis **pStatis) {
+int32_t loadDataBlockOnDemand(SQueryRuntimeEnv *pRuntimeEnv, void* pQueryHandle, SDataBlockInfo* pBlockInfo, SDataStatis **pStatis, SArray** pDataBlock) {
   SQuery *pQuery = pRuntimeEnv->pQuery;
 
-  uint32_t r = 0;
-  SArray * pDataBlock = NULL;
-
+  uint32_t status = 0;
   if (pQuery->numOfFilterCols > 0) {
-    r = BLK_DATA_ALL_NEEDED;
-  } else {
-    // check if this data block is required to load
+    status = BLK_DATA_ALL_NEEDED;
+  } else { // check if this data block is required to load
     for (int32_t i = 0; i < pQuery->numOfOutput; ++i) {
       SSqlFuncMsg* pSqlFunc = &pQuery->pSelectExpr[i].base;
 
       int32_t functionId = pSqlFunc->functionId;
       int32_t colId = pSqlFunc->colInfo.colId;
-      r |= aAggs[functionId].dataReqFunc(&pRuntimeEnv->pCtx[i], pQuery->window.skey, pQuery->window.ekey, colId);
+      status |= aAggs[functionId].dataReqFunc(&pRuntimeEnv->pCtx[i], pBlockInfo->window.skey, pBlockInfo->window.ekey, colId);
     }
 
     if (pRuntimeEnv->pTSBuf > 0 || QUERY_IS_INTERVAL_QUERY(pQuery)) {
-      r |= BLK_DATA_ALL_NEEDED;
+      status |= BLK_DATA_ALL_NEEDED;
     }
   }
 
-  if (r == BLK_DATA_NO_NEEDED) {
-    qDebug("QInfo:%p data block discard, rows:%d", GET_QINFO_ADDR(pRuntimeEnv), pBlockInfo->rows);
+  if (status == BLK_DATA_NO_NEEDED) {
+    qDebug("QInfo:%p data block discard, brange:%"PRId64 "-%"PRId64", rows:%d", GET_QINFO_ADDR(pRuntimeEnv),
+           pBlockInfo->window.skey, pBlockInfo->window.ekey, pBlockInfo->rows);
     pRuntimeEnv->summary.discardBlocks += 1;
-  } else if (r == BLK_DATA_STATIS_NEEDED) {
+  } else if (status == BLK_DATA_STATIS_NEEDED) {
     if (tsdbRetrieveDataBlockStatisInfo(pQueryHandle, pStatis) != TSDB_CODE_SUCCESS) {
       //        return DISK_DATA_LOAD_FAILED;
     }
@@ -2057,32 +2026,34 @@ SArray *loadDataBlockOnDemand(SQueryRuntimeEnv *pRuntimeEnv, void* pQueryHandle,
     pRuntimeEnv->summary.loadBlockStatis += 1;
   
     if (*pStatis == NULL) { // data block statistics does not exist, load data block
-      pDataBlock = tsdbRetrieveDataBlock(pQueryHandle, NULL);
+      *pDataBlock = tsdbRetrieveDataBlock(pQueryHandle, NULL);
       pRuntimeEnv->summary.totalCheckedRows += pBlockInfo->rows;
     }
   } else {
-    assert(r == BLK_DATA_ALL_NEEDED);
+    assert(status == BLK_DATA_ALL_NEEDED);
   
     // load the data block statistics to perform further filter
-    pRuntimeEnv->summary.loadBlockStatis +=1;
+    pRuntimeEnv->summary.loadBlockStatis += 1;
     if (tsdbRetrieveDataBlockStatisInfo(pQueryHandle, pStatis) != TSDB_CODE_SUCCESS) {
     }
     
-    if (!needToLoadDataBlock(pQuery,*pStatis, pRuntimeEnv->pCtx, pBlockInfo->rows)) {
+    if (!needToLoadDataBlock(pRuntimeEnv, *pStatis, pRuntimeEnv->pCtx, pBlockInfo->rows)) {
 #if defined(_DEBUG_VIEW)
       qDebug("QInfo:%p block discarded by per-filter", GET_QINFO_ADDR(pRuntimeEnv));
 #endif
       // current block has been discard due to filter applied
       pRuntimeEnv->summary.discardBlocks += 1;
-      //        return DISK_DATA_DISCARDED;
+      qDebug("QInfo:%p data block discard, brange:%"PRId64 "-%"PRId64", rows:%d", GET_QINFO_ADDR(pRuntimeEnv),
+          pBlockInfo->window.skey, pBlockInfo->window.ekey, pBlockInfo->rows);
+      return BLK_DATA_DISCARD;
     }
   
     pRuntimeEnv->summary.totalCheckedRows += pBlockInfo->rows;
     pRuntimeEnv->summary.loadBlocks += 1;
-    pDataBlock = tsdbRetrieveDataBlock(pQueryHandle, NULL);
+    *pDataBlock = tsdbRetrieveDataBlock(pQueryHandle, NULL);
   }
 
-  return pDataBlock;
+  return TSDB_CODE_SUCCESS;
 }
 
 int32_t binarySearchForKey(char *pValue, int num, TSKEY key, int order) {
@@ -2225,6 +2196,7 @@ static int64_t doScanAllDataBlocks(SQueryRuntimeEnv *pRuntimeEnv) {
          pQuery->order.order);
 
   TsdbQueryHandleT pQueryHandle = IS_MASTER_SCAN(pRuntimeEnv)? pRuntimeEnv->pQueryHandle : pRuntimeEnv->pSecQueryHandle;
+  int32_t step = GET_FORWARD_DIRECTION_FACTOR(pQuery->order.order);
 
   SDataBlockInfo blockInfo = SDATA_BLOCK_INITIALIZER;
   while (tsdbNextDataBlock(pQueryHandle)) {
@@ -2259,7 +2231,11 @@ static int64_t doScanAllDataBlocks(SQueryRuntimeEnv *pRuntimeEnv) {
     ensureOutputBuffer(pRuntimeEnv, &blockInfo);
 
     SDataStatis *pStatis = NULL;
-    SArray *pDataBlock = loadDataBlockOnDemand(pRuntimeEnv, pQueryHandle, &blockInfo, &pStatis);
+    SArray *pDataBlock   = NULL;
+    if (loadDataBlockOnDemand(pRuntimeEnv, pQueryHandle, &blockInfo, &pStatis, &pDataBlock) == BLK_DATA_DISCARD) {
+      pQuery->current->lastKey = QUERY_IS_ASC_QUERY(pQuery)? blockInfo.window.ekey + step:blockInfo.window.skey + step;
+      continue;
+    }
 
     // query start position can not move into tableApplyFunctionsOnBlock due to limit/offset condition
     pQuery->pos = QUERY_IS_ASC_QUERY(pQuery)? 0 : blockInfo.rows - 1;
@@ -2282,8 +2258,6 @@ static int64_t doScanAllDataBlocks(SQueryRuntimeEnv *pRuntimeEnv) {
 
   if (QUERY_IS_INTERVAL_QUERY(pQuery) && IS_MASTER_SCAN(pRuntimeEnv)) {
     if (Q_STATUS_EQUAL(pQuery->status, QUERY_COMPLETED)) {
-//      int32_t step = QUERY_IS_ASC_QUERY(pQuery) ? QUERY_ASC_FORWARD_STEP : QUERY_DESC_FORWARD_STEP;
-
       closeAllTimeWindow(&pRuntimeEnv->windowResInfo);
 //      removeRedundantWindow(&pRuntimeEnv->windowResInfo, pTableQueryInfo->lastKey - step, step);
       pRuntimeEnv->windowResInfo.curIndex = pRuntimeEnv->windowResInfo.size - 1;  // point to the last time window
@@ -3700,7 +3674,7 @@ static void updateWindowResNumOfRes(SQueryRuntimeEnv *pRuntimeEnv) {
   }
 }
 
-void stableApplyFunctionsOnBlock(SQueryRuntimeEnv *pRuntimeEnv, SDataBlockInfo *pDataBlockInfo, SDataStatis *pStatis,
+static void stableApplyFunctionsOnBlock(SQueryRuntimeEnv *pRuntimeEnv, SDataBlockInfo *pDataBlockInfo, SDataStatis *pStatis,
     SArray *pDataBlock, __block_search_fn_t searchFn) {
   SQuery *         pQuery = pRuntimeEnv->pQuery;
   STableQueryInfo* pTableQueryInfo = pQuery->current;
@@ -3859,9 +3833,10 @@ static void queryCostStatis(SQInfo *pQInfo) {
 //      pQInfo, pSummary->readDiskBlocks, pSummary->totalBlockSize, pSummary->loadBlocksUs / 1000.0,
 //      pSummary->skippedFileBlocks, pSummary->totalGenData);
   
-  qDebug("QInfo:%p :cost summary: elpased time:%"PRId64" us, total blocks:%d, use block statis:%d, use block data:%d, "
-         "total rows:%"PRId64 ", check rows:%"PRId64, pQInfo, pSummary->elapsedTime, pSummary->totalBlocks,
-         pSummary->loadBlockStatis, pSummary->loadBlocks, pSummary->totalRows, pSummary->totalCheckedRows);
+  qDebug("QInfo:%p :cost summary: elapsed time:%"PRId64" us, io time:%"PRId64" us, total blocks:%d, load block statis:%d,"
+         " load data block:%d, total rows:%"PRId64 ", check rows:%"PRId64,
+         pQInfo, pSummary->elapsedTime, pSummary->ioTime, pSummary->totalBlocks, pSummary->loadBlockStatis,
+         pSummary->loadBlocks, pSummary->totalRows, pSummary->totalCheckedRows);
 
 //  qDebug("QInfo:%p cost: temp file:%d Bytes", pQInfo, pSummary->tmpBufferInDisk);
 //
@@ -4247,10 +4222,11 @@ static int64_t scanMultiTableDataBlocks(SQInfo *pQInfo) {
   TsdbQueryHandleT pQueryHandle = IS_MASTER_SCAN(pRuntimeEnv)? pRuntimeEnv->pQueryHandle : pRuntimeEnv->pSecQueryHandle;
   SDataBlockInfo blockInfo = SDATA_BLOCK_INITIALIZER;
 
+  int32_t step = GET_FORWARD_DIRECTION_FACTOR(pQuery->order.order);
+
   while (tsdbNextDataBlock(pQueryHandle)) {
     summary->totalBlocks += 1;
     if (IS_QUERY_KILLED(pQInfo)) {
-      finalizeQueryResult(pRuntimeEnv); // clean up allocated resource during query
       longjmp(pRuntimeEnv->env, TSDB_CODE_TSC_QUERY_CANCELLED);
     }
 
@@ -4263,12 +4239,8 @@ static int64_t scanMultiTableDataBlocks(SQInfo *pQInfo) {
     assert(*pTableQueryInfo != NULL);
     SET_CURRENT_QUERY_TABLE_INFO(pRuntimeEnv, *pTableQueryInfo);
 
-    SDataStatis *pStatis = NULL;
-    SArray *pDataBlock = loadDataBlockOnDemand(pRuntimeEnv, pQueryHandle, &blockInfo, &pStatis);
-
     if (!pRuntimeEnv->groupbyNormalCol) {
       if (!QUERY_IS_INTERVAL_QUERY(pQuery)) {
-        int32_t step = QUERY_IS_ASC_QUERY(pQuery)? 1:-1;
         setExecutionContext(pQInfo, (*pTableQueryInfo)->groupIndex, blockInfo.window.ekey + step);
       } else {  // interval query
         TSKEY nextKey = blockInfo.window.skey;
@@ -4280,6 +4252,13 @@ static int64_t scanMultiTableDataBlocks(SQInfo *pQInfo) {
       }
     }
 
+    SDataStatis *pStatis = NULL;
+    SArray *pDataBlock = NULL;
+    if (loadDataBlockOnDemand(pRuntimeEnv, pQueryHandle, &blockInfo, &pStatis, &pDataBlock) == BLK_DATA_DISCARD) {
+      pQuery->current->lastKey = QUERY_IS_ASC_QUERY(pQuery)? blockInfo.window.ekey + step:blockInfo.window.skey + step;
+      continue;
+    }
+
     summary->totalRows += blockInfo.rows;
     stableApplyFunctionsOnBlock(pRuntimeEnv, &blockInfo, pStatis, pDataBlock, binarySearchForKey);