Merge remote-tracking branch 'origin/3.0' into feature/qnode

.gitignore

@@ -89,6 +89,7 @@ tests/examples/JDBC/JDBCDemo/.project
 tests/examples/JDBC/JDBCDemo/.settings/
 source/libs/parser/inc/sql.*
 tests/script/tmqResult.txt
+tests/tmqResult.txt
 
 # Emacs
 # -*- mode: gitignore; -*-

2.0/src/client/inc/tsclient.h

@@ -116,7 +116,7 @@ typedef struct SParsedDataColInfo {
   uint16_t       allNullLen;  // TODO: get from STSchema(base on SDataRow)
   uint16_t       extendedVarLen;
   uint16_t       boundNullLen;    // bound column len with all NULL value(without VarDataOffsetT/SColIdx part)
-  int32_t *      boundedColumns;  // bound column idx according to schema
+  int32_t       *boundColumns;    // bound column idx according to schema
   SBoundColumn * cols;
   SBoundIdxInfo *colIdxInfo;
   int8_t         orderStatus;  // bound columns
@@ -125,7 +125,7 @@ typedef struct SParsedDataColInfo {
 #define IS_DATA_COL_ORDERED(spd) ((spd->orderStatus) == (int8_t)ORDER_STATUS_ORDERED)
 
 typedef struct {
-  uint8_t memRowType;  // default is 0, that is SDataRow
+  uint8_t rowType;  // default is 0, that is SDataRow
   int32_t rowSize;
 } SMemRowBuilder;
 
@@ -137,17 +137,17 @@ void destroyMemRowBuilder(SMemRowBuilder *pBuilder);
 /**
  * @brief
  *
- * @param memRowType
+ * @param rowType
  * @param spd
  * @param idx   the absolute bound index of columns
  * @return FORCE_INLINE
  */
-static FORCE_INLINE void tscGetMemRowAppendInfo(SSchema *pSchema, uint8_t memRowType, SParsedDataColInfo *spd,
+static FORCE_INLINE void tscGetSTSRowAppendInfo(SSchema *pSchema, uint8_t rowType, SParsedDataColInfo *spd, int32_t idx,
-                                                int32_t idx, int32_t *toffset, int16_t *colId) {
+                                                int32_t *toffset, int16_t *colId) {
   int32_t schemaIdx = 0;
   if (IS_DATA_COL_ORDERED(spd)) {
-    schemaIdx = spd->boundedColumns[idx];
+    schemaIdx = spd->boundColumns[idx];
-    if (isDataRowT(memRowType)) {
+    if (isDataRowT(rowType)) {
       *toffset = (spd->cols + schemaIdx)->toffset;  // the offset of firstPart
     } else {
       *toffset = idx * sizeof(SColIdx);  // the offset of SColIdx
@@ -155,7 +155,7 @@ static FORCE_INLINE void tscGetMemRowAppendInfo(SSchema *pSchema, uint8_t memRow
   } else {
     ASSERT(idx == (spd->colIdxInfo + idx)->boundIdx);
     schemaIdx = (spd->colIdxInfo + idx)->schemaColIdx;
-    if (isDataRowT(memRowType)) {
+    if (isDataRowT(rowType)) {
       *toffset = (spd->cols + schemaIdx)->toffset;
     } else {
       *toffset = ((spd->colIdxInfo + idx)->finalIdx) * sizeof(SColIdx);

2.0/src/client/src/tscParseInsert.c

@@ -428,7 +428,7 @@ int tsParseOneRow(char **str, STableDataBlocks *pDataBlocks, int16_t timePrec, i
   // 1. set the parsed value from sql string
   for (int i = 0; i < spd->numOfBound; ++i) {
     // the start position in data block buffer of current value in sql
-    int32_t colIndex = spd->boundedColumns[i];
+    int32_t colIndex = spd->boundColumns[i];
 
     char *start = row + spd->cols[colIndex].offset;
 
@@ -495,7 +495,7 @@ int tsParseOneRow(char **str, STableDataBlocks *pDataBlocks, int16_t timePrec, i
     bool    isPrimaryKey = (colIndex == PRIMARYKEY_TIMESTAMP_COL_INDEX);
     int32_t toffset = -1;
     int16_t colId = -1;
-    tscGetMemRowAppendInfo(schema, pBuilder->memRowType, spd, i, &toffset, &colId);
+    tscGetSTSRowAppendInfo(schema, pBuilder->memRowType, spd, i, &toffset, &colId);
 
     int32_t ret =
         tsParseOneColumnKV(pSchema, &sToken, row, pInsertParam->msg, str, isPrimaryKey, timePrec, toffset, colId);
@@ -630,7 +630,7 @@ void tscSetBoundColumnInfo(SParsedDataColInfo *pColInfo, SSchema *pSchema, int32
   pColInfo->numOfCols = numOfCols;
   pColInfo->numOfBound = numOfCols;
   pColInfo->orderStatus = ORDER_STATUS_ORDERED;  // default is ORDERED for non-bound mode
-  pColInfo->boundedColumns = calloc(pColInfo->numOfCols, sizeof(int32_t));
+  pColInfo->boundColumns = calloc(pColInfo->numOfCols, sizeof(int32_t));
   pColInfo->cols = calloc(pColInfo->numOfCols, sizeof(SBoundColumn));
   pColInfo->colIdxInfo = NULL;
   pColInfo->flen = 0;
@@ -656,7 +656,7 @@ void tscSetBoundColumnInfo(SParsedDataColInfo *pColInfo, SSchema *pSchema, int32
       default:
         break;
     }
-    pColInfo->boundedColumns[i] = i;
+    pColInfo->boundColumns[i] = i;
   }
   pColInfo->allNullLen += pColInfo->flen;
   pColInfo->boundNullLen = pColInfo->allNullLen;  // default set allNullLen
@@ -991,7 +991,7 @@ static int32_t tscCheckIfCreateTable(char **sqlstr, SSqlObj *pSql, char** boundC
     }
 
     for (int i = 0; i < spd.numOfBound; ++i) {
-      SSchema* pSchema = &pTagSchema[spd.boundedColumns[i]];
+      SSchema *pSchema = &pTagSchema[spd.boundColumns[i]];
 
       index = 0;
       sToken = tStrGetToken(sql, &index, true);
@@ -1158,7 +1158,7 @@ static int32_t parseBoundColumns(SInsertStatementParam *pInsertParam, SParsedDat
 
   pColInfo->numOfBound = 0;
   pColInfo->boundNullLen = 0;
-  memset(pColInfo->boundedColumns, 0, sizeof(int32_t) * nCols);
+  memset(pColInfo->boundColumns, 0, sizeof(int32_t) * nCols);
   for (int32_t i = 0; i < nCols; ++i) {
     pColInfo->cols[i].valStat = VAL_STAT_NONE;
   }
@@ -1205,7 +1205,7 @@ static int32_t parseBoundColumns(SInsertStatementParam *pInsertParam, SParsedDat
         }
 
         pColInfo->cols[t].valStat = VAL_STAT_HAS;
-        pColInfo->boundedColumns[pColInfo->numOfBound] = t;
+        pColInfo->boundColumns[pColInfo->numOfBound] = t;
         ++pColInfo->numOfBound;
         switch (pSchema[t].type) {
           case TSDB_DATA_TYPE_BINARY:
@@ -1239,7 +1239,7 @@ static int32_t parseBoundColumns(SInsertStatementParam *pInsertParam, SParsedDat
           }
 
           pColInfo->cols[t].valStat = VAL_STAT_HAS;
-          pColInfo->boundedColumns[pColInfo->numOfBound] = t;
+          pColInfo->boundColumns[pColInfo->numOfBound] = t;
           ++pColInfo->numOfBound;
           switch (pSchema[t].type) {
             case TSDB_DATA_TYPE_BINARY:
@@ -1279,7 +1279,7 @@ static int32_t parseBoundColumns(SInsertStatementParam *pInsertParam, SParsedDat
     }
     SBoundIdxInfo *pColIdx = pColInfo->colIdxInfo;
     for (uint16_t i = 0; i < pColInfo->numOfBound; ++i) {
-      pColIdx[i].schemaColIdx = (uint16_t)pColInfo->boundedColumns[i];
+      pColIdx[i].schemaColIdx = (uint16_t)pColInfo->boundColumns[i];
       pColIdx[i].boundIdx = i;
     }
     qsort(pColIdx, pColInfo->numOfBound, sizeof(SBoundIdxInfo), schemaIdxCompar);
@@ -1289,7 +1289,7 @@ static int32_t parseBoundColumns(SInsertStatementParam *pInsertParam, SParsedDat
     qsort(pColIdx, pColInfo->numOfBound, sizeof(SBoundIdxInfo), boundIdxCompar);
   }
 
-  memset(&pColInfo->boundedColumns[pColInfo->numOfBound], 0,
+  memset(&pColInfo->boundColumns[pColInfo->numOfBound], 0,
          sizeof(int32_t) * (pColInfo->numOfCols - pColInfo->numOfBound));
 
   return TSDB_CODE_SUCCESS;

2.0/src/client/src/tscUtil.c

@@ -1554,7 +1554,7 @@ void tscFreeSqlObj(SSqlObj* pSql) {
 }
 
 void tscDestroyBoundColumnInfo(SParsedDataColInfo* pColInfo) {
-  tfree(pColInfo->boundedColumns);
+  tfree(pColInfo->boundColumns);
   tfree(pColInfo->cols);
   tfree(pColInfo->colIdxInfo);
 }

example/src/tmq.c

@@ -44,7 +44,7 @@ int32_t init_env() {
 
   pRes = taos_query(pConn, "create stable if not exists st1 (ts timestamp, k int) tags(a int)");
   if (taos_errno(pRes) != 0) {
-    printf("failed to create super table 123_$^), reason:%s\n", taos_errstr(pRes));
+    printf("failed to create super table st1, reason:%s\n", taos_errstr(pRes));
     return -1;
   }
   taos_free_result(pRes);

include/client/taos.h

@@ -213,9 +213,10 @@ typedef void(tmq_commit_cb(tmq_t *, tmq_resp_err_t, tmq_topic_vgroup_list_t *, v
 
 DLL_EXPORT tmq_list_t *tmq_list_new();
 DLL_EXPORT int32_t     tmq_list_append(tmq_list_t *, const char *);
+DLL_EXPORT void        tmq_list_destroy(tmq_list_t *);
 
 DLL_EXPORT tmq_t      *tmq_consumer_new(void *conn, tmq_conf_t *conf, char *errstr, int32_t errstrLen);
-DLL_EXPORT void        tmq_message_destroy(tmq_message_t *tmq_message);
+DLL_EXPORT tmq_t      *tmq_consumer_new1(tmq_conf_t *conf, char *errstr, int32_t errstrLen);
 DLL_EXPORT const char *tmq_err2str(tmq_resp_err_t);
 
 /* ------------------------TMQ CONSUMER INTERFACE------------------------ */
@@ -244,8 +245,8 @@ enum tmq_conf_res_t {
 typedef enum tmq_conf_res_t tmq_conf_res_t;
 
 DLL_EXPORT tmq_conf_t    *tmq_conf_new();
-DLL_EXPORT void           tmq_conf_destroy(tmq_conf_t *conf);
 DLL_EXPORT tmq_conf_res_t tmq_conf_set(tmq_conf_t *conf, const char *key, const char *value);
+DLL_EXPORT void           tmq_conf_destroy(tmq_conf_t *conf);
 DLL_EXPORT void           tmq_conf_set_offset_commit_cb(tmq_conf_t *conf, tmq_commit_cb *cb);
 
 // temporary used function for demo only
@@ -256,6 +257,8 @@ int32_t tmqGetSkipLogNum(tmq_message_t *tmq_message);
 
 DLL_EXPORT TAOS_ROW tmq_get_row(tmq_message_t *message);
 DLL_EXPORT char    *tmq_get_topic_name(tmq_message_t *message);
+DLL_EXPORT void    *tmq_get_topic_schema(tmq_t *tmq, const char *topic);
+DLL_EXPORT void     tmq_message_destroy(tmq_message_t *tmq_message);
 
 /* --------------------TMPORARY INTERFACE FOR TESTING--------------------- */
 DLL_EXPORT TAOS_RES *tmq_create_topic(TAOS *taos, const char *name, const char *sql, int sqlLen);

include/common/tcommon.h

@@ -127,7 +127,7 @@ static FORCE_INLINE int32_t tEncodeSMqPollRsp(void** buf, const SMqPollRsp* pRsp
   tlen += taosEncodeFixedI32(buf, pRsp->skipLogNum);
   tlen += taosEncodeFixedI32(buf, pRsp->numOfTopics);
   if (pRsp->numOfTopics == 0) return tlen;
-  tlen += tEncodeSSchemaWrapper(buf, pRsp->schema);
+  tlen += taosEncodeSSchemaWrapper(buf, pRsp->schema);
   if (pRsp->pBlockData) {
     sz = taosArrayGetSize(pRsp->pBlockData);
   }
@@ -149,7 +149,7 @@ static FORCE_INLINE void* tDecodeSMqPollRsp(void* buf, SMqPollRsp* pRsp) {
   if (pRsp->numOfTopics == 0) return buf;
   pRsp->schema = (SSchemaWrapper*)taosMemoryCalloc(1, sizeof(SSchemaWrapper));
   if (pRsp->schema == NULL) return NULL;
-  buf = tDecodeSSchemaWrapper(buf, pRsp->schema);
+  buf = taosDecodeSSchemaWrapper(buf, pRsp->schema);
   buf = taosDecodeFixedI32(buf, &sz);
   pRsp->pBlockData = taosArrayInit(sz, sizeof(SSDataBlock));
   for (int32_t i = 0; i < sz; i++) {

include/common/tdataformat.h

@@ -59,12 +59,15 @@ extern "C" {
   } while (0);
 
 // ----------------- TSDB COLUMN DEFINITION
+#pragma pack(push, 1)
 typedef struct {
-  int8_t   type;    // Column type
+  col_id_t colId;        // column ID(start from PRIMARYKEY_TIMESTAMP_COL_ID(1))
-  col_id_t colId;   // column ID(start from PRIMARYKEY_TIMESTAMP_COL_ID(1))
+  int32_t  type : 8;     // column type
-  int16_t  bytes;   // column bytes (restore to int16_t in case of misuse)
+  int32_t  bytes : 24;   // column bytes (restore to int32_t in case of misuse)
-  uint16_t offset;  // point offset in STpRow after the header part.
+  int32_t  sma : 8;      // block SMA: 0, no SMA, 1, sum/min/max, 2, ...
+  int32_t  offset : 24;  // point offset in STpRow after the header part.
 } STColumn;
+#pragma pack(pop)
 
 #define colType(col)   ((col)->type)
 #define colColId(col)  ((col)->colId)
@@ -136,7 +139,7 @@ typedef struct {
 int32_t   tdInitTSchemaBuilder(STSchemaBuilder *pBuilder, int32_t version);
 void      tdDestroyTSchemaBuilder(STSchemaBuilder *pBuilder);
 void      tdResetTSchemaBuilder(STSchemaBuilder *pBuilder, int32_t version);
-int32_t   tdAddColToSchema(STSchemaBuilder *pBuilder, int8_t type, int16_t colId, int16_t bytes);
+int32_t   tdAddColToSchema(STSchemaBuilder *pBuilder, int8_t type, col_id_t colId, col_bytes_t bytes);
 STSchema *tdGetSchemaFromBuilder(STSchemaBuilder *pBuilder);
 
 // ----------------- Semantic timestamp key definition
@@ -590,7 +593,7 @@ void    tdDestroyKVRowBuilder(SKVRowBuilder *pBuilder);
 void    tdResetKVRowBuilder(SKVRowBuilder *pBuilder);
 SKVRow  tdGetKVRowFromBuilder(SKVRowBuilder *pBuilder);
 
-static FORCE_INLINE int32_t tdAddColToKVRow(SKVRowBuilder *pBuilder, int16_t colId, int8_t type, const void *value) {
+static FORCE_INLINE int32_t tdAddColToKVRow(SKVRowBuilder *pBuilder, col_id_t colId, int8_t type, const void *value) {
   if (pBuilder->nCols >= pBuilder->tCols) {
     pBuilder->tCols *= 2;
     SColIdx *pColIdx = (SColIdx *)taosMemoryRealloc((void *)(pBuilder->pColIdx), sizeof(SColIdx) * pBuilder->tCols);

include/common/tmsg.h

@@ -208,11 +208,11 @@ typedef struct {
 typedef struct SSubmitBlk {
   int64_t uid;        // table unique id
   int64_t suid;       // stable id
-  int32_t padding;    // TODO just for padding here
   int32_t sversion;   // data schema version
   int32_t dataLen;    // data part length, not including the SSubmitBlk head
   int32_t schemaLen;  // schema length, if length is 0, no schema exists
   int16_t numOfRows;  // total number of rows in current submit block
+  int16_t padding;    // TODO just for padding here
   char    data[];
 } SSubmitBlk;
 
@@ -259,10 +259,10 @@ typedef struct {
 } SSubmitRsp;
 
 typedef struct SSchema {
-  int8_t  type;
+  int8_t   type;
-  int32_t colId;
+  col_id_t colId;
-  int32_t bytes;
+  int32_t  bytes;
-  char    name[TSDB_COL_NAME_LEN];
+  char     name[TSDB_COL_NAME_LEN];
 } SSchema;
 
 typedef struct {
@@ -438,8 +438,8 @@ typedef struct {
  */
 typedef struct {
   union {
-    int16_t colId;
+    col_id_t colId;
-    int16_t slotId;
+    int16_t  slotId;
   };
 
   int16_t type;
@@ -1901,7 +1901,7 @@ static FORCE_INLINE int32_t taosEncodeSSchema(void** buf, const SSchema* pSchema
   int32_t tlen = 0;
   tlen += taosEncodeFixedI8(buf, pSchema->type);
   tlen += taosEncodeFixedI32(buf, pSchema->bytes);
-  tlen += taosEncodeFixedI32(buf, pSchema->colId);
+  tlen += taosEncodeFixedI16(buf, pSchema->colId);
   tlen += taosEncodeString(buf, pSchema->name);
   return tlen;
 }
@@ -1909,7 +1909,7 @@ static FORCE_INLINE int32_t taosEncodeSSchema(void** buf, const SSchema* pSchema
 static FORCE_INLINE void* taosDecodeSSchema(void* buf, SSchema* pSchema) {
   buf = taosDecodeFixedI8(buf, &pSchema->type);
   buf = taosDecodeFixedI32(buf, &pSchema->bytes);
-  buf = taosDecodeFixedI32(buf, &pSchema->colId);
+  buf = taosDecodeFixedI16(buf, &pSchema->colId);
   buf = taosDecodeStringTo(buf, pSchema->name);
   return buf;
 }
@@ -1917,7 +1917,7 @@ static FORCE_INLINE void* taosDecodeSSchema(void* buf, SSchema* pSchema) {
 static FORCE_INLINE int32_t tEncodeSSchema(SCoder* pEncoder, const SSchema* pSchema) {
   if (tEncodeI8(pEncoder, pSchema->type) < 0) return -1;
   if (tEncodeI32(pEncoder, pSchema->bytes) < 0) return -1;
-  if (tEncodeI32(pEncoder, pSchema->colId) < 0) return -1;
+  if (tEncodeI16(pEncoder, pSchema->colId) < 0) return -1;
   if (tEncodeCStr(pEncoder, pSchema->name) < 0) return -1;
   return 0;
 }
@@ -1925,12 +1925,12 @@ static FORCE_INLINE int32_t tEncodeSSchema(SCoder* pEncoder, const SSchema* pSch
 static FORCE_INLINE int32_t tDecodeSSchema(SCoder* pDecoder, SSchema* pSchema) {
   if (tDecodeI8(pDecoder, &pSchema->type) < 0) return -1;
   if (tDecodeI32(pDecoder, &pSchema->bytes) < 0) return -1;
-  if (tDecodeI32(pDecoder, &pSchema->colId) < 0) return -1;
+  if (tDecodeI16(pDecoder, &pSchema->colId) < 0) return -1;
   if (tDecodeCStrTo(pDecoder, pSchema->name) < 0) return -1;
   return 0;
 }
 
-static FORCE_INLINE int32_t tEncodeSSchemaWrapper(void** buf, const SSchemaWrapper* pSW) {
+static FORCE_INLINE int32_t taosEncodeSSchemaWrapper(void** buf, const SSchemaWrapper* pSW) {
   int32_t tlen = 0;
   tlen += taosEncodeFixedU32(buf, pSW->nCols);
   for (int32_t i = 0; i < pSW->nCols; i++) {
@@ -1939,7 +1939,7 @@ static FORCE_INLINE int32_t tEncodeSSchemaWrapper(void** buf, const SSchemaWrapp
   return tlen;
 }
 
-static FORCE_INLINE void* tDecodeSSchemaWrapper(void* buf, SSchemaWrapper* pSW) {
+static FORCE_INLINE void* taosDecodeSSchemaWrapper(void* buf, SSchemaWrapper* pSW) {
   buf = taosDecodeFixedU32(buf, &pSW->nCols);
   pSW->pSchema = (SSchema*)taosMemoryCalloc(pSW->nCols, sizeof(SSchema));
   if (pSW->pSchema == NULL) {
@@ -1952,6 +1952,27 @@ static FORCE_INLINE void* tDecodeSSchemaWrapper(void* buf, SSchemaWrapper* pSW)
   return buf;
 }
 
+static FORCE_INLINE int32_t tEncodeSSchemaWrapper(SCoder* pEncoder, const SSchemaWrapper* pSW) {
+  if (tEncodeU32(pEncoder, pSW->nCols) < 0) return -1;
+  for (int32_t i = 0; i < pSW->nCols; i++) {
+    if (tEncodeSSchema(pEncoder, &pSW->pSchema[i]) < 0) return -1;
+  }
+  return pEncoder->pos;
+}
+
+static FORCE_INLINE int32_t tDecodeSSchemaWrapper(SCoder* pDecoder, SSchemaWrapper* pSW) {
+  if (tDecodeU32(pDecoder, &pSW->nCols) < 0) return -1;
+  void* ptr = taosMemoryRealloc(pSW->pSchema, pSW->nCols * sizeof(SSchema));
+  if (ptr == NULL) {
+    return -1;
+  }
+  pSW->pSchema = (SSchema*)ptr;
+  for (int32_t i = 0; i < pSW->nCols; i++) {
+    if (tDecodeSSchema(pDecoder, &pSW->pSchema[i]) < 0) return -1;
+  }
+  return 0;
+}
+
 typedef struct {
   char    name[TSDB_TABLE_FNAME_LEN];
   char    stb[TSDB_TABLE_FNAME_LEN];

include/common/trow.h

@@ -671,8 +671,9 @@ static FORCE_INLINE int32_t tdAppendColValToKvRow(SRowBuilder *pBuilder, TDRowVa
  * @param colIdx sorted column index, start from 0
  * @return FORCE_INLINE
  */
-static FORCE_INLINE int32_t tdAppendColValToRow(SRowBuilder *pBuilder, int16_t colId, int8_t colType, TDRowValT valType,
+static FORCE_INLINE int32_t tdAppendColValToRow(SRowBuilder *pBuilder, col_id_t colId, int8_t colType,
-                                                const void *val, bool isCopyVarData, int32_t offset, int16_t colIdx) {
+                                                TDRowValT valType, const void *val, bool isCopyVarData, int32_t offset,
+                                                col_id_t colIdx) {
   STSRow *pRow = pBuilder->pBuf;
   if (!val) {
 #ifdef TD_SUPPORT_BITMAP

include/common/ttypes.h

@@ -29,6 +29,7 @@ typedef uint32_t TDRowLenT;
 typedef uint8_t  TDRowValT;
 typedef int16_t  col_id_t;
 typedef int8_t   col_type_t;
+typedef int32_t  col_bytes_t;
 
 #pragma pack(push, 1)
 typedef struct {

include/libs/nodes/nodes.h

@@ -120,6 +120,7 @@ typedef enum ENodeType {
   QUERY_NODE_LOGIC_PLAN_VNODE_MODIF,
   QUERY_NODE_LOGIC_PLAN_EXCHANGE,
   QUERY_NODE_LOGIC_PLAN_WINDOW,
+  QUERY_NODE_LOGIC_PLAN_SORT,
   QUERY_NODE_LOGIC_SUBPLAN,
   QUERY_NODE_LOGIC_PLAN,
 

include/libs/nodes/plannodes.h

@@ -66,6 +66,7 @@ typedef struct SAggLogicNode {
 typedef struct SProjectLogicNode {
   SLogicNode node;
   SNodeList* pProjections;
+  char stmtName[TSDB_TABLE_NAME_LEN]; 
 } SProjectLogicNode;
 
 typedef struct SVnodeModifLogicNode {
@@ -97,8 +98,14 @@ typedef struct SWindowLogicNode {
   int8_t  slidingUnit;
   SFillNode* pFill;
   int64_t sessionGap;
+  SNode* pTspk;
 } SWindowLogicNode;
 
+typedef struct SSortLogicNode {
+  SLogicNode node;
+  SNodeList* pSortKeys;
+} SSortLogicNode;
+
 typedef enum ESubplanType {
   SUBPLAN_TYPE_MERGE = 1,
   SUBPLAN_TYPE_PARTIAL,
@@ -200,7 +207,7 @@ typedef struct SJoinPhysiNode {
 typedef struct SAggPhysiNode {
   SPhysiNode node;
   SNodeList* pExprs;   // these are expression list of group_by_clause and parameter expression of aggregate function
-  SNodeList* pGroupKeys; // SColumnRefNode list
+  SNodeList* pGroupKeys;
   SNodeList* pAggFuncs;
 } SAggPhysiNode;
 
@@ -225,6 +232,7 @@ typedef struct SWinodwPhysiNode {
 
 typedef struct SIntervalPhysiNode {
   SWinodwPhysiNode window;
+  SNode* pTspk; // timestamp primary key
   int64_t    interval;
   int64_t    offset;
   int64_t    sliding;
@@ -238,6 +246,12 @@ typedef struct SSessionWinodwPhysiNode {
   int64_t    gap;
 } SSessionWinodwPhysiNode;
 
+typedef struct SSortPhysiNode {
+  SPhysiNode node;
+  SNodeList* pExprs;   // these are expression list of order_by_clause and parameter expression of aggregate function
+  SNodeList* pSortKeys; // element is SOrderByExprNode, and SOrderByExprNode::pExpr is SColumnNode
+} SSortPhysiNode;
+
 typedef struct SDataSinkNode {
   ENodeType type;
   SDataBlockDescNode* pInputDataBlockDesc;

include/libs/nodes/querynodes.h

@@ -56,7 +56,7 @@ typedef enum EColumnType {
 typedef struct SColumnNode {
   SExprNode node; // QUERY_NODE_COLUMN
   uint64_t tableId;
-  int16_t colId;
+  col_id_t colId;
   EColumnType colType; // column or tag
   char dbName[TSDB_DB_NAME_LEN];
   char tableName[TSDB_TABLE_NAME_LEN];
@@ -191,12 +191,13 @@ typedef struct SStateWindowNode {
 
 typedef struct SSessionWindowNode {
   ENodeType type; // QUERY_NODE_SESSION_WINDOW
-  SNode* pCol;
+  SNode* pCol; // timestamp primary key
   SNode* pGap; // gap between two session window(in microseconds)
 } SSessionWindowNode;
 
 typedef struct SIntervalWindowNode {
   ENodeType type; // QUERY_NODE_INTERVAL_WINDOW
+  SNode* pCol; // timestamp primary key
   SNode* pInterval; // SValueNode
   SNode* pOffset;   // SValueNode
   SNode* pSliding;  // SValueNode
@@ -231,6 +232,7 @@ typedef struct SSelectStmt {
   SNodeList* pOrderByList; // SOrderByExprNode
   SNode* pLimit;
   SNode* pSlimit;
+  char stmtName[TSDB_TABLE_NAME_LEN];
 } SSelectStmt;
 
 typedef enum ESetOperatorType {

include/libs/qcom/query.h

@@ -43,10 +43,10 @@ typedef enum {
 } ETaskType;
 
 typedef struct STableComInfo {
-  uint8_t numOfTags;     // the number of tags in schema
+  uint8_t  numOfTags;     // the number of tags in schema
-  uint8_t precision;     // the number of precision
+  uint8_t  precision;     // the number of precision
-  int16_t numOfColumns;  // the number of columns
+  col_id_t numOfColumns;  // the number of columns
-  int32_t rowSize;       // row size of the schema
+  int32_t  rowSize;       // row size of the schema
 } STableComInfo;
 
 typedef struct SIndexMeta {
@@ -173,7 +173,7 @@ bool           tIsValidSchema(struct SSchema* pSchema, int32_t numOfCols, int32_
 int32_t queryCreateTableMetaFromMsg(STableMetaRsp* msg, bool isSuperTable, STableMeta** pMeta);
 char *jobTaskStatusStr(int32_t status);
 
-SSchema createSchema(uint8_t type, int32_t bytes, int32_t colId, const char* name);
+SSchema createSchema(int8_t type, int32_t bytes, col_id_t colId, const char* name);
 
 extern int32_t (*queryBuildMsg[TDMT_MAX])(void* input, char** msg, int32_t msgSize, int32_t* msgLen);
 extern int32_t (*queryProcessMsgRsp[TDMT_MAX])(void* output, char* msg, int32_t msgSize);

include/libs/stream/tstream.h

@@ -62,10 +62,11 @@ typedef struct {
 } STaskExec;
 
 typedef struct {
-  int8_t reserved;
+  int32_t taskId;
 } STaskDispatcherInplace;
 
 typedef struct {
+  int32_t taskId;
   int32_t nodeId;
   SEpSet  epSet;
 } STaskDispatcherFixedEp;

include/os/os.h

@@ -75,7 +75,6 @@ extern "C" {
 #include "osDef.h"
 #include "osDir.h"
 #include "osEndian.h"
-#include "osEnv.h"
 #include "osFile.h"
 #include "osLocale.h"
 #include "osLz4.h"
@@ -93,8 +92,9 @@ extern "C" {
 #include "osTime.h"
 #include "osTimer.h"
 #include "osTimezone.h"
+#include "osEnv.h"
 
-void osInit();
+void osDefaultInit();
 
 #ifdef __cplusplus
 }

include/os/osEnv.h

@@ -22,17 +22,18 @@
 extern "C" {
 #endif
 
-extern char    tsOsName[];
+extern char            tsOsName[];
-extern char    tsTimezone[];
+extern char            tsTimezoneStr[];
-extern char    tsCharset[];
+extern enum TdTimezone tsTimezone;
-extern char    tsLocale[];
+extern char            tsCharset[];
-extern int8_t  tsDaylight;
+extern char            tsLocale[];
-extern bool    tsEnableCoreFile;
+extern int8_t          tsDaylight;
-extern int64_t tsPageSizeKB;
+extern bool            tsEnableCoreFile;
-extern int64_t tsOpenMax;
+extern int64_t         tsPageSizeKB;
-extern int64_t tsStreamMax;
+extern int64_t         tsOpenMax;
-extern float   tsNumOfCores;
+extern int64_t         tsStreamMax;
-extern int64_t tsTotalMemoryKB;
+extern float           tsNumOfCores;
+extern int64_t         tsTotalMemoryKB;
 
 extern char configDir[];
 extern char tsDataDir[];
@@ -43,11 +44,12 @@ extern SDiskSpace tsDataSpace;
 extern SDiskSpace tsLogSpace;
 extern SDiskSpace tsTempSpace;
 
-void osInit();
+void osDefaultInit();
 void osUpdate();
 void osCleanup();
 bool osLogSpaceAvailable();
 void osSetTimezone(const char *timezone);
+void osSetSystemLocale(const char *inLocale, const char *inCharSet);
 
 #ifdef __cplusplus
 }

include/os/osTimezone.h

@@ -26,8 +26,37 @@ extern "C" {
     #define tzset TZSET_FUNC_TAOS_FORBID
 #endif
 
-void taosGetSystemTimezone(char *outTimezone);
+enum TdTimezone
-void taosSetSystemTimezone(const char *inTimezone, char *outTimezone, int8_t *outDaylight);
+{
+      TdWestZone12=-12,
+      TdWestZone11,
+      TdWestZone10,
+      TdWestZone9,
+      TdWestZone8,
+      TdWestZone7,
+      TdWestZone6,
+      TdWestZone5,
+      TdWestZone4,
+      TdWestZone3,
+      TdWestZone2,
+      TdWestZone1,
+      TdZeroZone,
+      TdEastZone1,
+      TdEastZone2,
+      TdEastZone3,
+      TdEastZone4,
+      TdEastZone5,
+      TdEastZone6,
+      TdEastZone7,
+      TdEastZone8,
+      TdEastZone9,
+      TdEastZone10,
+      TdEastZone11,
+      TdEastZone12
+};
+
+void taosGetSystemTimezone(char *outTimezone, enum TdTimezone *tsTimezone);
+void taosSetSystemTimezone(const char *inTimezone, char *outTimezone, int8_t *outDaylight, enum TdTimezone *tsTimezone);
 
 #ifdef __cplusplus
 }

include/util/taoserror.h

@@ -16,6 +16,8 @@
 #ifndef _TD_UTIL_TAOS_ERROR_H_
 #define _TD_UTIL_TAOS_ERROR_H_
 
+#include "os.h"
+
 #ifdef __cplusplus
 extern "C" {
 #endif

include/util/tdef.h

@@ -209,7 +209,7 @@ typedef enum ELogicConditionType {
 #define TSDB_FUNC_TYPE_AGGREGATE 2
 #define TSDB_FUNC_MAX_RETRIEVE   1024
 
-#define TSDB_INDEX_NAME_LEN      33  // 32 + 1 '\0'
+#define TSDB_INDEX_NAME_LEN      65  // 64 + 1 '\0'
 #define TSDB_TYPE_STR_MAX_LEN    32
 #define TSDB_TABLE_FNAME_LEN     (TSDB_DB_FNAME_LEN + TSDB_TABLE_NAME_LEN + TSDB_NAME_DELIMITER_LEN)
 #define TSDB_TOPIC_FNAME_LEN     TSDB_TABLE_FNAME_LEN

include/util/tprocess.h

@@ -51,6 +51,7 @@ void      taosProcCleanup(SProcObj *pProc);
 int32_t   taosProcRun(SProcObj *pProc);
 void      taosProcStop(SProcObj *pProc);
 bool      taosProcIsChild(SProcObj *pProc);
+int32_t   taosProcChildId(SProcObj *pProc);
 
 int32_t taosProcPutToChildQueue(SProcObj *pProc, void *pHead, int32_t headLen, void *pBody, int32_t bodyLen);
 int32_t taosProcPutToParentQueue(SProcObj *pProc, void *pHead, int32_t headLen, void *pBody, int32_t bodyLen);

source/client/src/clientEnv.c

@@ -408,10 +408,10 @@ int taos_options_imp(TSDB_OPTION option, const char *str) {
       assert(cfg != NULL);
 
       if (cfg->cfgStatus <= TAOS_CFG_CSTATUS_OPTION) {
-        tstrncpy(tsTimezone, str, TD_TIMEZONE_LEN);
+        tstrncpy(tsTimezoneStr, str, TD_TIMEZONE_LEN);
         tsSetTimeZone();
         cfg->cfgStatus = TAOS_CFG_CSTATUS_OPTION;
-        tscDebug("timezone set:%s, input:%s by taos_options", tsTimezone, str);
+        tscDebug("timezone set:%s, input:%s by taos_options", tsTimezoneStr, str);
       } else {
         tscWarn("config option:%s, input value:%s, is configured by %s, use %s", cfg->option, str,
                 tsCfgStatusStr[cfg->cfgStatus], (char *)cfg->ptr);

source/client/src/clientHb.c

@@ -91,7 +91,7 @@ static int32_t hbProcessStbInfoRsp(void *value, int32_t valueLen, struct SCatalo
     } else {
       tscDebug("hb update stb, db:%s, stb:%s", rsp->dbFName, rsp->stbName);
       if (rsp->pSchemas[0].colId != PRIMARYKEY_TIMESTAMP_COL_ID) {
-        tscError("invalid colId[%d] for the first column in table meta rsp msg", rsp->pSchemas[0].colId);
+        tscError("invalid colId[%" PRIi16 "] for the first column in table meta rsp msg", rsp->pSchemas[0].colId);
         tFreeSTableMetaBatchRsp(&batchMetaRsp);
         return TSDB_CODE_TSC_INVALID_VALUE;
       }

source/client/src/tmq.c

@@ -27,9 +27,7 @@
 #include "tref.h"
 
 struct tmq_list_t {
-  int32_t cnt;
+  SArray container;
-  int32_t tot;
-  char*   elems[];
 };
 
 struct tmq_topic_vgroup_t {
@@ -45,11 +43,14 @@ struct tmq_topic_vgroup_list_t {
 struct tmq_conf_t {
   char           clientId[256];
   char           groupId[TSDB_CGROUP_LEN];
-  int8_t         auto_commit;
+  int8_t         autoCommit;
   int8_t         resetOffset;
+  uint16_t       port;
+  char*          ip;
+  char*          user;
+  char*          pass;
+  char*          db;
   tmq_commit_cb* commit_cb;
-  /*char*          ip;*/
-  /*uint16_t       port;*/
 };
 
 struct tmq_t {
@@ -98,12 +99,13 @@ typedef struct {
 
 typedef struct {
   // subscribe info
-  int32_t sqlLen;
+  int32_t        sqlLen;
-  char*   sql;
+  char*          sql;
-  char*   topicName;
+  char*          topicName;
-  int64_t topicId;
+  int64_t        topicId;
-  int32_t nextVgIdx;
+  int32_t        nextVgIdx;
-  SArray* vgs;  // SArray<SMqClientVg>
+  SArray*        vgs;  // SArray<SMqClientVg>
+  SSchemaWrapper schema;
 } SMqClientTopic;
 
 typedef struct {
@@ -137,7 +139,7 @@ typedef struct {
 
 tmq_conf_t* tmq_conf_new() {
   tmq_conf_t* conf = taosMemoryCalloc(1, sizeof(tmq_conf_t));
-  conf->auto_commit = false;
+  conf->autoCommit = false;
   conf->resetOffset = TMQ_CONF__RESET_OFFSET__EARLIEAST;
   return conf;
 }
@@ -151,21 +153,24 @@ tmq_conf_res_t tmq_conf_set(tmq_conf_t* conf, const char* key, const char* value
     strcpy(conf->groupId, value);
     return TMQ_CONF_OK;
   }
+
   if (strcmp(key, "client.id") == 0) {
     strcpy(conf->clientId, value);
     return TMQ_CONF_OK;
   }
+
   if (strcmp(key, "enable.auto.commit") == 0) {
     if (strcmp(value, "true") == 0) {
-      conf->auto_commit = true;
+      conf->autoCommit = true;
       return TMQ_CONF_OK;
     } else if (strcmp(value, "false") == 0) {
-      conf->auto_commit = false;
+      conf->autoCommit = false;
       return TMQ_CONF_OK;
     } else {
       return TMQ_CONF_INVALID;
     }
   }
+
   if (strcmp(key, "auto.offset.reset") == 0) {
     if (strcmp(value, "none") == 0) {
       conf->resetOffset = TMQ_CONF__RESET_OFFSET__NONE;
@@ -180,28 +185,51 @@ tmq_conf_res_t tmq_conf_set(tmq_conf_t* conf, const char* key, const char* value
       return TMQ_CONF_INVALID;
     }
   }
+
+  if (strcmp(key, "td.connect.ip") == 0) {
+    conf->ip = strdup(value);
+    return TMQ_CONF_OK;
+  }
+  if (strcmp(key, "td.connect.user") == 0) {
+    conf->user = strdup(value);
+    return TMQ_CONF_OK;
+  }
+  if (strcmp(key, "td.connect.pass") == 0) {
+    conf->pass = strdup(value);
+    return TMQ_CONF_OK;
+  }
+  if (strcmp(key, "td.connect.port") == 0) {
+    conf->port = atoi(value);
+    return TMQ_CONF_OK;
+  }
+  if (strcmp(key, "td.connect.db") == 0) {
+    conf->db = strdup(value);
+    return TMQ_CONF_OK;
+  }
+
   return TMQ_CONF_UNKNOWN;
 }
 
 tmq_list_t* tmq_list_new() {
-  tmq_list_t* ptr = taosMemoryMalloc(sizeof(tmq_list_t) + 8 * sizeof(char*));
+  //
-  if (ptr == NULL) {
+  return (tmq_list_t*)taosArrayInit(0, sizeof(void*));
-    return ptr;
-  }
-  ptr->cnt = 0;
-  ptr->tot = 8;
-  return ptr;
 }
 
-int32_t tmq_list_append(tmq_list_t* ptr, const char* src) {
+int32_t tmq_list_append(tmq_list_t* list, const char* src) {
-  if (ptr->cnt >= ptr->tot - 1) return -1;
+  SArray* container = &list->container;
-  ptr->elems[ptr->cnt] = strdup(src);
+  char*   topic = strdup(src);
-  ptr->cnt++;
+  if (taosArrayPush(container, &topic) == NULL) return -1;
   return 0;
 }
 
+void tmq_list_destroy(tmq_list_t* list) {
+  SArray* container = &list->container;
+  /*taosArrayDestroy(container);*/
+  taosArrayDestroyEx(container, (void (*)(void*))taosMemoryFree);
+}
+
 void tmqClearUnhandleMsg(tmq_t* tmq) {
-  tmq_message_t* msg;
+  tmq_message_t* msg = NULL;
   while (1) {
     taosGetQitem(tmq->qall, (void**)&msg);
     if (msg)
@@ -268,17 +296,57 @@ tmq_t* tmq_consumer_new(void* conn, tmq_conf_t* conf, char* errstr, int32_t errs
   // set conf
   strcpy(pTmq->clientId, conf->clientId);
   strcpy(pTmq->groupId, conf->groupId);
-  pTmq->autoCommit = conf->auto_commit;
+  pTmq->autoCommit = conf->autoCommit;
   pTmq->commit_cb = conf->commit_cb;
   pTmq->resetOffsetCfg = conf->resetOffset;
 
-  tsem_init(&pTmq->rspSem, 0, 0);
-
   pTmq->consumerId = generateRequestId() & (((uint64_t)-1) >> 1);
   pTmq->clientTopics = taosArrayInit(0, sizeof(SMqClientTopic));
+  if (pTmq->clientTopics == NULL) {
+    taosMemoryFree(pTmq);
+    return NULL;
+  }
 
   pTmq->mqueue = taosOpenQueue();
   pTmq->qall = taosAllocateQall();
+
+  tsem_init(&pTmq->rspSem, 0, 0);
+
+  return pTmq;
+}
+
+tmq_t* tmq_consumer_new1(tmq_conf_t* conf, char* errstr, int32_t errstrLen) {
+  tmq_t* pTmq = taosMemoryCalloc(1, sizeof(tmq_t));
+  if (pTmq == NULL) {
+    return NULL;
+  }
+  pTmq->pTscObj = taos_connect(conf->ip, conf->user, conf->pass, conf->db, conf->port);
+
+  pTmq->inWaiting = 0;
+  pTmq->status = 0;
+  pTmq->pollCnt = 0;
+  pTmq->epoch = 0;
+  pTmq->waitingRequest = 0;
+  pTmq->readyRequest = 0;
+  // set conf
+  strcpy(pTmq->clientId, conf->clientId);
+  strcpy(pTmq->groupId, conf->groupId);
+  pTmq->autoCommit = conf->autoCommit;
+  pTmq->commit_cb = conf->commit_cb;
+  pTmq->resetOffsetCfg = conf->resetOffset;
+
+  pTmq->consumerId = generateRequestId() & (((uint64_t)-1) >> 1);
+  pTmq->clientTopics = taosArrayInit(0, sizeof(SMqClientTopic));
+  if (pTmq->clientTopics == NULL) {
+    taosMemoryFree(pTmq);
+    return NULL;
+  }
+
+  pTmq->mqueue = taosOpenQueue();
+  pTmq->qall = taosAllocateQall();
+
+  tsem_init(&pTmq->rspSem, 0, 0);
+
   return pTmq;
 }
 
@@ -372,7 +440,8 @@ tmq_resp_err_t tmq_commit(tmq_t* tmq, const tmq_topic_vgroup_list_t* offsets, in
 
 tmq_resp_err_t tmq_subscribe(tmq_t* tmq, tmq_list_t* topic_list) {
   SRequestObj* pRequest = NULL;
-  int32_t      sz = topic_list->cnt;
+  SArray*      container = &topic_list->container;
+  int32_t      sz = taosArrayGetSize(container);
   // destroy ex
   taosArrayDestroy(tmq->clientTopics);
   tmq->clientTopics = taosArrayInit(sz, sizeof(SMqClientTopic));
@@ -384,7 +453,8 @@ tmq_resp_err_t tmq_subscribe(tmq_t* tmq, tmq_list_t* topic_list) {
   req.topicNames = taosArrayInit(sz, sizeof(void*));
 
   for (int i = 0; i < sz; i++) {
-    char* topicName = topic_list->elems[i];
+    /*char* topicName = topic_list->elems[i];*/
+    char* topicName = taosArrayGetP(container, i);
 
     SName name = {0};
     char* dbName = getDbOfConnection(tmq->pTscObj);
@@ -737,7 +807,7 @@ int32_t tmqPollCb(void* param, const SDataBuf* pMsg, int32_t code) {
   SMqClientVg*    pVg = pParam->pVg;
   tmq_t*          tmq = pParam->tmq;
   if (code != 0) {
-    printf("msg discard %x\n", code);
+    printf("msg discard, code:%x\n", code);
     goto WRITE_QUEUE_FAIL;
   }
 
@@ -807,10 +877,10 @@ WRITE_QUEUE_FAIL:
 }
 
 bool tmqUpdateEp(tmq_t* tmq, int32_t epoch, SMqCMGetSubEpRsp* pRsp) {
+  printf("call update ep %d\n", epoch);
   bool    set = false;
   int32_t sz = taosArrayGetSize(pRsp->topics);
-  if (tmq->clientTopics) taosArrayDestroy(tmq->clientTopics);
+  SArray* newTopics = taosArrayInit(sz, sizeof(SMqClientTopic));
-  tmq->clientTopics = taosArrayInit(sz, sizeof(SMqClientTopic));
   for (int32_t i = 0; i < sz; i++) {
     SMqClientTopic topic = {0};
     SMqSubTopicEp* pTopicEp = taosArrayGet(pRsp->topics, i);
@@ -829,8 +899,10 @@ bool tmqUpdateEp(tmq_t* tmq, int32_t epoch, SMqCMGetSubEpRsp* pRsp) {
       taosArrayPush(topic.vgs, &clientVg);
       set = true;
     }
-    taosArrayPush(tmq->clientTopics, &topic);
+    taosArrayPush(newTopics, &topic);
   }
+  if (tmq->clientTopics) taosArrayDestroy(tmq->clientTopics);
+  tmq->clientTopics = newTopics;
   atomic_store_32(&tmq->epoch, epoch);
   return set;
 }
@@ -1149,6 +1221,7 @@ tmq_message_t* tmqHandleAllRsp(tmq_t* tmq, int64_t blockingTime, bool pollIfRese
       if (rspMsg->msg.head.epoch == atomic_load_32(&tmq->epoch)) {
         /*printf("epoch match\n");*/
         SMqClientVg* pVg = rspMsg->vg;
+        /*printf("vg %d offset %ld up to %ld\n", pVg->vgId, pVg->currentOffset, rspMsg->msg.rspOffset);*/
         pVg->currentOffset = rspMsg->msg.rspOffset;
         atomic_store_32(&pVg->vgStatus, TMQ_VG_STATUS__IDLE);
         return rspMsg;

source/common/src/tdataformat.c

@@ -106,12 +106,12 @@ void *tdDecodeSchema(void *buf, STSchema **pRSchema) {
   if (tdInitTSchemaBuilder(&schemaBuilder, version) < 0) return NULL;
 
   for (int i = 0; i < numOfCols; i++) {
-    int8_t  type = 0;
+    col_type_t  type = 0;
-    int16_t colId = 0;
+    col_id_t    colId = 0;
-    int16_t bytes = 0;
+    col_bytes_t bytes = 0;
     buf = taosDecodeFixedI8(buf, &type);
     buf = taosDecodeFixedI16(buf, &colId);
-    buf = taosDecodeFixedI16(buf, &bytes);
+    buf = taosDecodeFixedI32(buf, &bytes);
     if (tdAddColToSchema(&schemaBuilder, type, colId, bytes) < 0) {
       tdDestroyTSchemaBuilder(&schemaBuilder);
       return NULL;
@@ -148,7 +148,7 @@ void tdResetTSchemaBuilder(STSchemaBuilder *pBuilder, int32_t version) {
   pBuilder->version = version;
 }
 
-int tdAddColToSchema(STSchemaBuilder *pBuilder, int8_t type, int16_t colId, int16_t bytes) {
+int tdAddColToSchema(STSchemaBuilder *pBuilder, int8_t type, col_id_t colId, col_bytes_t bytes) {
   if (!isValidDataType(type)) return -1;
 
   if (pBuilder->nCols >= pBuilder->tCols) {

source/common/src/tglobal.c

@@ -303,7 +303,7 @@ static int32_t taosAddClientCfg(SConfig *pCfg) {
 static int32_t taosAddSystemCfg(SConfig *pCfg) {
   SysNameInfo info = taosGetSysNameInfo();
 
-  if (cfgAddTimezone(pCfg, "timezone", tsTimezone) != 0) return -1;
+  if (cfgAddTimezone(pCfg, "timezone", tsTimezoneStr) != 0) return -1;
   if (cfgAddLocale(pCfg, "locale", tsLocale) != 0) return -1;
   if (cfgAddCharset(pCfg, "charset", tsCharset) != 0) return -1;
   if (cfgAddBool(pCfg, "enableCoreFile", 1, 1) != 0) return -1;
@@ -431,12 +431,13 @@ static int32_t taosSetClientCfg(SConfig *pCfg) {
 static void taosSetSystemCfg(SConfig *pCfg) {
   SConfigItem *pItem = cfgGetItem(pCfg, "timezone");
   osSetTimezone(pItem->str);
-  uDebug("timezone format changed from %s to %s", pItem->str, tsTimezone);
+  uDebug("timezone format changed from %s to %s", pItem->str, tsTimezoneStr);
-  cfgSetItem(pCfg, "timezone", tsTimezone, pItem->stype);
+  cfgSetItem(pCfg, "timezone", tsTimezoneStr, pItem->stype);
 
   const char *locale = cfgGetItem(pCfg, "locale")->str;
   const char *charset = cfgGetItem(pCfg, "charset")->str;
   taosSetSystemLocale(locale, charset);
+  osSetSystemLocale(locale, charset);
 
   bool enableCore = cfgGetItem(pCfg, "enableCoreFile")->bval;
   taosSetConsoleEcho(enableCore);
@@ -483,7 +484,7 @@ static int32_t taosSetServerCfg(SConfig *pCfg) {
 
 int32_t taosCreateLog(const char *logname, int32_t logFileNum, const char *cfgDir, const char *envFile,
                       const char *apolloUrl, SArray *pArgs, bool tsc) {
-  osInit();
+  osDefaultInit();
 
   SConfig *pCfg = cfgInit();
   if (pCfg == NULL) return -1;

source/common/src/tmsg.c

@@ -299,14 +299,14 @@ int32_t tSerializeSVCreateTbReq(void **buf, SVCreateTbReq *pReq) {
       tlen += taosEncodeFixedU32(buf, pReq->stbCfg.nCols);
       for (uint32_t i = 0; i < pReq->stbCfg.nCols; i++) {
         tlen += taosEncodeFixedI8(buf, pReq->stbCfg.pSchema[i].type);
-        tlen += taosEncodeFixedI32(buf, pReq->stbCfg.pSchema[i].colId);
+        tlen += taosEncodeFixedI16(buf, pReq->stbCfg.pSchema[i].colId);
         tlen += taosEncodeFixedI32(buf, pReq->stbCfg.pSchema[i].bytes);
         tlen += taosEncodeString(buf, pReq->stbCfg.pSchema[i].name);
       }
       tlen += taosEncodeFixedU32(buf, pReq->stbCfg.nTagCols);
       for (uint32_t i = 0; i < pReq->stbCfg.nTagCols; i++) {
         tlen += taosEncodeFixedI8(buf, pReq->stbCfg.pTagSchema[i].type);
-        tlen += taosEncodeFixedI32(buf, pReq->stbCfg.pTagSchema[i].colId);
+        tlen += taosEncodeFixedI16(buf, pReq->stbCfg.pTagSchema[i].colId);
         tlen += taosEncodeFixedI32(buf, pReq->stbCfg.pTagSchema[i].bytes);
         tlen += taosEncodeString(buf, pReq->stbCfg.pTagSchema[i].name);
       }
@@ -333,7 +333,7 @@ int32_t tSerializeSVCreateTbReq(void **buf, SVCreateTbReq *pReq) {
       tlen += taosEncodeFixedU32(buf, pReq->ntbCfg.nCols);
       for (uint32_t i = 0; i < pReq->ntbCfg.nCols; i++) {
         tlen += taosEncodeFixedI8(buf, pReq->ntbCfg.pSchema[i].type);
-        tlen += taosEncodeFixedI32(buf, pReq->ntbCfg.pSchema[i].colId);
+        tlen += taosEncodeFixedI16(buf, pReq->ntbCfg.pSchema[i].colId);
         tlen += taosEncodeFixedI32(buf, pReq->ntbCfg.pSchema[i].bytes);
         tlen += taosEncodeString(buf, pReq->ntbCfg.pSchema[i].name);
       }
@@ -374,7 +374,7 @@ void *tDeserializeSVCreateTbReq(void *buf, SVCreateTbReq *pReq) {
       pReq->stbCfg.pSchema = (SSchema *)taosMemoryMalloc(pReq->stbCfg.nCols * sizeof(SSchema));
       for (uint32_t i = 0; i < pReq->stbCfg.nCols; i++) {
         buf = taosDecodeFixedI8(buf, &(pReq->stbCfg.pSchema[i].type));
-        buf = taosDecodeFixedI32(buf, &(pReq->stbCfg.pSchema[i].colId));
+        buf = taosDecodeFixedI16(buf, &(pReq->stbCfg.pSchema[i].colId));
         buf = taosDecodeFixedI32(buf, &(pReq->stbCfg.pSchema[i].bytes));
         buf = taosDecodeStringTo(buf, pReq->stbCfg.pSchema[i].name);
       }
@@ -382,7 +382,7 @@ void *tDeserializeSVCreateTbReq(void *buf, SVCreateTbReq *pReq) {
       pReq->stbCfg.pTagSchema = (SSchema *)taosMemoryMalloc(pReq->stbCfg.nTagCols * sizeof(SSchema));
       for (uint32_t i = 0; i < pReq->stbCfg.nTagCols; i++) {
         buf = taosDecodeFixedI8(buf, &(pReq->stbCfg.pTagSchema[i].type));
-        buf = taosDecodeFixedI32(buf, &pReq->stbCfg.pTagSchema[i].colId);
+        buf = taosDecodeFixedI16(buf, &pReq->stbCfg.pTagSchema[i].colId);
         buf = taosDecodeFixedI32(buf, &pReq->stbCfg.pTagSchema[i].bytes);
         buf = taosDecodeStringTo(buf, pReq->stbCfg.pTagSchema[i].name);
       }
@@ -422,7 +422,7 @@ void *tDeserializeSVCreateTbReq(void *buf, SVCreateTbReq *pReq) {
       pReq->ntbCfg.pSchema = (SSchema *)taosMemoryMalloc(pReq->ntbCfg.nCols * sizeof(SSchema));
       for (uint32_t i = 0; i < pReq->ntbCfg.nCols; i++) {
         buf = taosDecodeFixedI8(buf, &pReq->ntbCfg.pSchema[i].type);
-        buf = taosDecodeFixedI32(buf, &pReq->ntbCfg.pSchema[i].colId);
+        buf = taosDecodeFixedI16(buf, &pReq->ntbCfg.pSchema[i].colId);
         buf = taosDecodeFixedI32(buf, &pReq->ntbCfg.pSchema[i].bytes);
         buf = taosDecodeStringTo(buf, pReq->ntbCfg.pSchema[i].name);
       }

source/dnode/mgmt/container/inc/dndInt.h

@@ -56,7 +56,6 @@ void    dndCleanupServer(SDnode *pDnode);
 int32_t dndInitClient(SDnode *pDnode);
 void    dndCleanupClient(SDnode *pDnode);
 int32_t dndInitMsgHandle(SDnode *pDnode);
-void    dndSendRpcRsp(SMgmtWrapper *pWrapper, SRpcMsg *pRsp);
 
 #ifdef __cplusplus
 }

source/dnode/mgmt/container/src/dndExec.c

@@ -20,7 +20,7 @@ static void dndResetLog(SMgmtWrapper *pMgmt) {
   char logname[24] = {0};
   snprintf(logname, sizeof(logname), "%slog", pMgmt->name);
 
-  dInfo("node:%s, reset log to %s", pMgmt->name, logname);
+  dInfo("node:%s, reset log to %s in child process", pMgmt->name, logname);
   taosCloseLog();
   taosInitLog(logname, 1);
 }
@@ -51,6 +51,7 @@ int32_t dndOpenNode(SMgmtWrapper *pWrapper) {
 
 void dndCloseNode(SMgmtWrapper *pWrapper) {
   dDebug("node:%s, start to close", pWrapper->name);
+  pWrapper->required = false;
   taosWLockLatch(&pWrapper->latch);
   if (pWrapper->deployed) {
     (*pWrapper->fp.closeFp)(pWrapper);
@@ -138,7 +139,7 @@ static void dndConsumeChildQueue(SMgmtWrapper *pWrapper, SNodeMsg *pMsg, int32_t
 static void dndConsumeParentQueue(SMgmtWrapper *pWrapper, SRpcMsg *pRsp, int32_t msgLen, void *pCont, int32_t contLen) {
   dTrace("msg:%p, get from parent queue", pRsp);
   pRsp->pCont = pCont;
-  dndSendRpcRsp(pWrapper, pRsp);
+  dndSendRsp(pWrapper, pRsp);
   taosMemoryFree(pRsp);
 }
 
@@ -178,7 +179,6 @@ static int32_t dndRunInMultiProcess(SDnode *pDnode) {
                      .parentFreeHeadFp = (ProcFreeFp)taosMemoryFree,
                      .parentMallocBodyFp = (ProcMallocFp)rpcMallocCont,
                      .parentFreeBodyFp = (ProcFreeFp)rpcFreeCont,
-                     .testFlag = 0,
                      .pParent = pWrapper,
                      .name = pWrapper->name};
     SProcObj *pProc = taosProcInit(&cfg);
@@ -200,7 +200,7 @@ static int32_t dndRunInMultiProcess(SDnode *pDnode) {
       dInfo("node:%s, will be initialized in child process", pWrapper->name);
       dndOpenNode(pWrapper);
     } else {
-      dInfo("node:%s, will not start in parent process", pWrapper->name);
+      dInfo("node:%s, will not start in parent process, child pid:%d", pWrapper->name, taosProcChildId(pProc));
       pWrapper->procType = PROC_PARENT;
     }
 
@@ -210,16 +210,20 @@ static int32_t dndRunInMultiProcess(SDnode *pDnode) {
     }
   }
 
-#if 0
+  dndSetStatus(pDnode, DND_STAT_RUNNING);
-  SMgmtWrapper *pWrapper = dndAcquireWrapper(pDnode, DNODE);
+
-  if (pWrapper->procType == PROC_PARENT && dmStart(pWrapper->pMgmt) != 0) {
+  for (ENodeType n = 0; n < NODE_MAX; ++n) {
-    dndReleaseWrapper(pWrapper);
+    SMgmtWrapper *pWrapper = &pDnode->wrappers[n];
-    dError("failed to start dnode worker since %s", terrstr());
+    if (!pWrapper->required) continue;
-    return -1;
+    if (pWrapper->fp.startFp == NULL) continue;
+    if (pWrapper->procType == PROC_PARENT && n != DNODE) continue;
+    if (pWrapper->procType == PROC_CHILD && n == DNODE) continue;
+    if ((*pWrapper->fp.startFp)(pWrapper) != 0) {
+      dError("node:%s, failed to start since %s", pWrapper->name, terrstr());
+      return -1;
+    }
   }
 
-  dndReleaseWrapper(pWrapper);
-#endif
   return 0;
 }
 

source/dnode/mgmt/container/src/dndFile.c

@@ -16,14 +16,16 @@
 #define _DEFAULT_SOURCE
 #include "dndInt.h"
 
+#define MAXLEN 1024
+
 int32_t dndReadFile(SMgmtWrapper *pWrapper, bool *pDeployed) {
-  int32_t   code = TSDB_CODE_NODE_PARSE_FILE_ERROR;
+  int32_t       code = TSDB_CODE_NODE_PARSE_FILE_ERROR;
-  int32_t   len = 0;
+  int32_t       len = 0;
-  int32_t   maxLen = 1024;
+  const int32_t maxLen = MAXLEN;
-  char     *content = taosMemoryCalloc(1, maxLen + 1);
+  char          content[MAXLEN + 1] = {0};
-  cJSON    *root = NULL;
+  cJSON        *root = NULL;
-  char      file[PATH_MAX];
+  char          file[PATH_MAX];
-  TdFilePtr pFile = NULL;
+  TdFilePtr     pFile = NULL;
 
   snprintf(file, sizeof(file), "%s%s%s.json", pWrapper->path, TD_DIRSEP, pWrapper->name);
   pFile = taosOpenFile(file, TD_FILE_READ);
@@ -57,7 +59,6 @@ int32_t dndReadFile(SMgmtWrapper *pWrapper, bool *pDeployed) {
   dDebug("succcessed to read file %s, deployed:%d", file, *pDeployed);
 
 _OVER:
-  if (content != NULL) taosMemoryFree(content);
   if (root != NULL) cJSON_Delete(root);
   if (pFile != NULL) taosCloseFile(&pFile);
 
@@ -66,7 +67,7 @@ _OVER:
 }
 
 int32_t dndWriteFile(SMgmtWrapper *pWrapper, bool deployed) {
-  char file[PATH_MAX];
+  char file[PATH_MAX] = {0};
   snprintf(file, sizeof(file), "%s%s%s.json", pWrapper->path, TD_DIRSEP, pWrapper->name);
 
   TdFilePtr pFile = taosOpenFile(file, TD_FILE_CTEATE | TD_FILE_WRITE | TD_FILE_TRUNC);
@@ -76,9 +77,9 @@ int32_t dndWriteFile(SMgmtWrapper *pWrapper, bool deployed) {
     return -1;
   }
 
-  int32_t len = 0;
+  int32_t       len = 0;
-  int32_t maxLen = 1024;
+  const int32_t maxLen = MAXLEN;
-  char   *content = taosMemoryCalloc(1, maxLen + 1);
+  char          content[MAXLEN + 1] = {0};
 
   len += snprintf(content + len, maxLen - len, "{\n");
   len += snprintf(content + len, maxLen - len, "  \"deployed\": %d\n", deployed);
@@ -87,9 +88,8 @@ int32_t dndWriteFile(SMgmtWrapper *pWrapper, bool deployed) {
   taosWriteFile(pFile, content, len);
   taosFsyncFile(pFile);
   taosCloseFile(&pFile);
-  taosMemoryFree(content);
 
-  char realfile[PATH_MAX];
+  char realfile[PATH_MAX] = {0};
   snprintf(realfile, sizeof(realfile), "%s%s%s.json", pWrapper->path, TD_DIRSEP, pWrapper->name);
 
   if (taosRenameFile(file, realfile) != 0) {

source/dnode/mgmt/container/src/dndMsg.c

@@ -43,36 +43,40 @@ static inline int32_t dndBuildMsg(SNodeMsg *pMsg, SRpcMsg *pRpc) {
 
   memcpy(pMsg->user, connInfo.user, TSDB_USER_LEN);
   memcpy(&pMsg->rpcMsg, pRpc, sizeof(SRpcMsg));
-
   return 0;
 }
 
 void dndProcessRpcMsg(SMgmtWrapper *pWrapper, SRpcMsg *pRpc, SEpSet *pEpSet) {
-  if (pEpSet && pEpSet->numOfEps > 0 && pRpc->msgType == TDMT_MND_STATUS_RSP) {
-    dndUpdateMnodeEpSet(pWrapper->pDnode, pEpSet);
-  }
-
   int32_t   code = -1;
   SNodeMsg *pMsg = NULL;
   NodeMsgFp msgFp = NULL;
 
+  if (pEpSet && pEpSet->numOfEps > 0 && pRpc->msgType == TDMT_MND_STATUS_RSP) {
+    dndUpdateMnodeEpSet(pWrapper->pDnode, pEpSet);
+  }
+
   if (dndMarkWrapper(pWrapper) != 0) goto _OVER;
   if ((msgFp = dndGetMsgFp(pWrapper, pRpc)) == NULL) goto _OVER;
   if ((pMsg = taosAllocateQitem(sizeof(SNodeMsg))) == NULL) goto _OVER;
   if (dndBuildMsg(pMsg, pRpc) != 0) goto _OVER;
 
-  dTrace("msg:%p, is created, handle:%p app:%p user:%s", pMsg, pRpc->handle, pRpc->ahandle, pMsg->user);
   if (pWrapper->procType == PROC_SINGLE) {
+    dTrace("msg:%p, is created, handle:%p app:%p user:%s", pMsg, pRpc->handle, pRpc->ahandle, pMsg->user);
     code = (*msgFp)(pWrapper->pMgmt, pMsg);
   } else if (pWrapper->procType == PROC_PARENT) {
+    dTrace("msg:%p, is created and will put into child queue, handle:%p app:%p user:%s", pMsg, pRpc->handle,
+           pRpc->ahandle, pMsg->user);
     code = taosProcPutToChildQueue(pWrapper->pProc, pMsg, sizeof(SNodeMsg), pRpc->pCont, pRpc->contLen);
   } else {
+    dTrace("msg:%p, should not processed in child process, handle:%p app:%p user:%s", pMsg, pRpc->handle, pRpc->ahandle,
+           pMsg->user);
+    ASSERT(1);
   }
 
 _OVER:
   if (code == 0) {
     if (pWrapper->procType == PROC_PARENT) {
-      dTrace("msg:%p, is freed", pMsg);
+      dTrace("msg:%p, is freed in parent process", pMsg);
       taosFreeQitem(pMsg);
       rpcFreeCont(pRpc->pCont);
     }

source/dnode/mgmt/container/src/dndObj.c

@@ -175,7 +175,7 @@ int32_t dndMarkWrapper(SMgmtWrapper *pWrapper) {
   int32_t code = 0;
 
   taosRLockLatch(&pWrapper->latch);
-  if (pWrapper->deployed) {
+  if (pWrapper->deployed || (pWrapper->procType == PROC_PARENT && pWrapper->required)) {
     int32_t refCount = atomic_add_fetch_32(&pWrapper->refCount, 1);
     dTrace("node:%s, is marked, refCount:%d", pWrapper->name, refCount);
   } else {

source/dnode/mgmt/container/src/dndTransport.c

@@ -348,7 +348,7 @@ int32_t dndSendReqToMnode(SMgmtWrapper *pWrapper, SRpcMsg *pReq) {
   }
 }
 
-void dndSendRpcRsp(SMgmtWrapper *pWrapper, SRpcMsg *pRsp) {
+static void dndSendRpcRsp(SMgmtWrapper *pWrapper, SRpcMsg *pRsp) {
   if (pRsp->code == TSDB_CODE_APP_NOT_READY) {
     SMgmtWrapper *pDnodeWrapper = dndAcquireWrapper(pWrapper->pDnode, DNODE);
     if (pDnodeWrapper != NULL) {

source/dnode/mgmt/dnode/src/dmMsg.c

@@ -36,7 +36,7 @@ void dmSendStatusReq(SDnodeMgmt *pMgmt) {
   req.clusterCfg.checkTime = 0;
   char timestr[32] = "1970-01-01 00:00:00.00";
   (void)taosParseTime(timestr, &req.clusterCfg.checkTime, (int32_t)strlen(timestr), TSDB_TIME_PRECISION_MILLI, 0);
-  memcpy(req.clusterCfg.timezone, tsTimezone, TD_TIMEZONE_LEN);
+  memcpy(req.clusterCfg.timezone, tsTimezoneStr, TD_TIMEZONE_LEN);
   memcpy(req.clusterCfg.locale, tsLocale, TD_LOCALE_LEN);
   memcpy(req.clusterCfg.charset, tsCharset, TD_LOCALE_LEN);
   taosRUnLockLatch(&pMgmt->latch);

source/dnode/mgmt/vnode/src/vmWorker.c

@@ -160,6 +160,24 @@ static void vmProcessSyncQueue(SQueueInfo *pInfo, STaosQall *qall, int32_t numOf
   }
 }
 
+static void vmProcessMergeQueue(SQueueInfo *pInfo, STaosQall *qall, int32_t numOfMsgs) {
+  SVnodeObj *pVnode = pInfo->ahandle;
+  SNodeMsg  *pMsg = NULL;
+
+  for (int32_t i = 0; i < numOfMsgs; ++i) {
+    taosGetQitem(qall, (void **)&pMsg);
+
+    dTrace("msg:%p, will be processed in vnode-merge queue", pMsg);
+    int32_t code = vnodeProcessFetchMsg(pVnode->pImpl, &pMsg->rpcMsg);
+    if (code != 0) {
+      vmSendRsp(pVnode->pWrapper, pMsg, code);
+      dTrace("msg:%p, is freed, result:0x%04x:%s", pMsg, code & 0XFFFF, tstrerror(code));
+      rpcFreeCont(pMsg->rpcMsg.pCont);
+      taosFreeQitem(pMsg);
+    }
+  }
+}
+
 static int32_t vmPutNodeMsgToQueue(SVnodesMgmt *pMgmt, SNodeMsg *pMsg, EQueueType qtype) {
   SRpcMsg *pRpc = &pMsg->rpcMsg;
   int32_t  code = -1;
@@ -308,7 +326,7 @@ int32_t vmGetQueueSize(SMgmtWrapper *pWrapper, int32_t vgId, EQueueType qtype) {
 int32_t vmAllocQueue(SVnodesMgmt *pMgmt, SVnodeObj *pVnode) {
   pVnode->pWriteQ = tWWorkerAllocQueue(&pMgmt->writePool, pVnode, (FItems)vmProcessWriteQueue);
   pVnode->pApplyQ = tWWorkerAllocQueue(&pMgmt->writePool, pVnode, (FItems)vmProcessApplyQueue);
-  pVnode->pMergeQ = tWWorkerAllocQueue(&pMgmt->mergePool, pVnode, (FItems)vmProcessMergeMsg);
+  pVnode->pMergeQ = tWWorkerAllocQueue(&pMgmt->mergePool, pVnode, (FItems)vmProcessMergeQueue);
   pVnode->pSyncQ = tWWorkerAllocQueue(&pMgmt->syncPool, pVnode, (FItems)vmProcessSyncQueue);
   pVnode->pFetchQ = tQWorkerAllocQueue(&pMgmt->fetchPool, pVnode, (FItem)vmProcessFetchQueue);
   pVnode->pQueryQ = tQWorkerAllocQueue(&pMgmt->queryPool, pVnode, (FItem)vmProcessQueryQueue);

source/dnode/mnode/impl/inc/mndDef.h

@@ -633,18 +633,19 @@ static FORCE_INLINE void tDeleteSMqSubscribeObj(SMqSubscribeObj* pSub) {
 }
 
 typedef struct {
-  char     name[TSDB_TOPIC_FNAME_LEN];
+  char           name[TSDB_TOPIC_FNAME_LEN];
-  char     db[TSDB_DB_FNAME_LEN];
+  char           db[TSDB_DB_FNAME_LEN];
-  int64_t  createTime;
+  int64_t        createTime;
-  int64_t  updateTime;
+  int64_t        updateTime;
-  int64_t  uid;
+  int64_t        uid;
-  int64_t  dbUid;
+  int64_t        dbUid;
-  int32_t  version;
+  int32_t        version;
-  SRWLatch lock;
+  SRWLatch       lock;
-  int32_t  sqlLen;
+  int32_t        sqlLen;
-  char*    sql;
+  char*          sql;
-  char*    logicalPlan;
+  char*          logicalPlan;
-  char*    physicalPlan;
+  char*          physicalPlan;
+  SSchemaWrapper schema;
 } SMqTopicObj;
 
 typedef struct {
@@ -733,12 +734,12 @@ typedef struct {
   int8_t   sourceType;
   int8_t   sinkType;
   // int32_t  sqlLen;
-  int32_t sinkVgId;  // 0 for automatic
+  int32_t        sinkVgId;  // 0 for automatic
-  char*   sql;
+  char*          sql;
-  char*   logicalPlan;
+  char*          logicalPlan;
-  char*   physicalPlan;
+  char*          physicalPlan;
-  SArray* tasks;     // SArray<SArray<SStreamTask>>
+  SArray*        tasks;  // SArray<SArray<SStreamTask>>
-  SArray* ColAlias;  // SArray<char*>
+  SSchemaWrapper outputSchema;
 } SStreamObj;
 
 int32_t tEncodeSStreamObj(SCoder* pEncoder, const SStreamObj* pObj);

source/dnode/mnode/impl/src/mndDef.c

@@ -17,7 +17,7 @@
 
 int32_t tEncodeSStreamObj(SCoder *pEncoder, const SStreamObj *pObj) {
   int32_t sz = 0;
-  int32_t outputNameSz = 0;
+  /*int32_t outputNameSz = 0;*/
   if (tEncodeCStr(pEncoder, pObj->name) < 0) return -1;
   if (tEncodeCStr(pEncoder, pObj->db) < 0) return -1;
   if (tEncodeI64(pEncoder, pObj->createTime) < 0) return -1;
@@ -45,6 +45,9 @@ int32_t tEncodeSStreamObj(SCoder *pEncoder, const SStreamObj *pObj) {
     }
   }
 
+  if (tEncodeSSchemaWrapper(pEncoder, &pObj->outputSchema) < 0) return -1;
+
+#if 0
   if (pObj->ColAlias != NULL) {
     outputNameSz = taosArrayGetSize(pObj->ColAlias);
   }
@@ -53,6 +56,7 @@ int32_t tEncodeSStreamObj(SCoder *pEncoder, const SStreamObj *pObj) {
     char *name = taosArrayGetP(pObj->ColAlias, i);
     if (tEncodeCStr(pEncoder, name) < 0) return -1;
   }
+#endif
   return pEncoder->pos;
 }
 
@@ -85,6 +89,9 @@ int32_t tDecodeSStreamObj(SCoder *pDecoder, SStreamObj *pObj) {
       taosArrayPush(pObj->tasks, pArray);
     }
   }
+
+  if (tDecodeSSchemaWrapper(pDecoder, &pObj->outputSchema) < 0) return -1;
+#if 0
   int32_t outputNameSz;
   if (tDecodeI32(pDecoder, &outputNameSz) < 0) return -1;
   if (outputNameSz != 0) {
@@ -98,5 +105,6 @@ int32_t tDecodeSStreamObj(SCoder *pDecoder, SStreamObj *pObj) {
     if (tDecodeCStrAlloc(pDecoder, &name) < 0) return -1;
     taosArrayPush(pObj->ColAlias, &name);
   }
+#endif
   return 0;
 }

source/dnode/mnode/impl/src/mndDnode.c

@@ -277,8 +277,8 @@ static int32_t mndCheckClusterCfgPara(SMnode *pMnode, const SClusterCfg *pCfg) {
     return DND_REASON_STATUS_INTERVAL_NOT_MATCH;
   }
 
-  if ((0 != strcasecmp(pCfg->timezone, tsTimezone)) && (pMnode->checkTime != pCfg->checkTime)) {
+  if ((0 != strcasecmp(pCfg->timezone, tsTimezoneStr)) && (pMnode->checkTime != pCfg->checkTime)) {
-    mError("timezone [%s - %s] [%" PRId64 " - %" PRId64 "] cfg inconsistent", pCfg->timezone, tsTimezone,
+    mError("timezone [%s - %s] [%" PRId64 " - %" PRId64 "] cfg inconsistent", pCfg->timezone, tsTimezoneStr,
            pCfg->checkTime, pMnode->checkTime);
     return DND_REASON_TIME_ZONE_NOT_MATCH;
   }
@@ -677,7 +677,7 @@ static int32_t mndRetrieveConfigs(SNodeMsg *pReq, SShowObj *pShow, char *data, i
   totalRows++;
 
   cfgOpts[totalRows] = "timezone";
-  snprintf(cfgVals[totalRows], TSDB_CONIIG_VALUE_LEN, "%s", tsTimezone);
+  snprintf(cfgVals[totalRows], TSDB_CONIIG_VALUE_LEN, "%s", tsTimezoneStr);
   totalRows++;
 
   cfgOpts[totalRows] = "locale";

source/dnode/mnode/impl/src/mndScheduler.c

@@ -185,6 +185,7 @@ int32_t mndScheduleStream(SMnode* pMnode, STrans* pTrans, SStreamObj* pStream) {
           pTask->dispatchMsgType = TDMT_VND_TASK_MERGE_EXEC;
           pTask->dispatchType = TASK_DISPATCH__FIXED;
 
+          pTask->fixedEpDispatcher.taskId = lastLevelTask->taskId;
           pTask->fixedEpDispatcher.nodeId = lastLevelTask->nodeId;
           pTask->fixedEpDispatcher.epSet = lastLevelTask->epSet;
         }

source/dnode/mnode/impl/src/mndStb.c

@@ -98,7 +98,7 @@ SSdbRaw *mndStbActionEncode(SStbObj *pStb) {
   for (int32_t i = 0; i < pStb->numOfColumns; ++i) {
     SSchema *pSchema = &pStb->pColumns[i];
     SDB_SET_INT8(pRaw, dataPos, pSchema->type, STB_ENCODE_OVER)
-    SDB_SET_INT32(pRaw, dataPos, pSchema->colId, STB_ENCODE_OVER)
+    SDB_SET_INT16(pRaw, dataPos, pSchema->colId, STB_ENCODE_OVER)
     SDB_SET_INT32(pRaw, dataPos, pSchema->bytes, STB_ENCODE_OVER)
     SDB_SET_BINARY(pRaw, dataPos, pSchema->name, TSDB_COL_NAME_LEN, STB_ENCODE_OVER)
   }
@@ -106,7 +106,7 @@ SSdbRaw *mndStbActionEncode(SStbObj *pStb) {
   for (int32_t i = 0; i < pStb->numOfTags; ++i) {
     SSchema *pSchema = &pStb->pTags[i];
     SDB_SET_INT8(pRaw, dataPos, pSchema->type, STB_ENCODE_OVER)
-    SDB_SET_INT32(pRaw, dataPos, pSchema->colId, STB_ENCODE_OVER)
+    SDB_SET_INT16(pRaw, dataPos, pSchema->colId, STB_ENCODE_OVER)
     SDB_SET_INT32(pRaw, dataPos, pSchema->bytes, STB_ENCODE_OVER)
     SDB_SET_BINARY(pRaw, dataPos, pSchema->name, TSDB_COL_NAME_LEN, STB_ENCODE_OVER)
   }
@@ -114,7 +114,7 @@ SSdbRaw *mndStbActionEncode(SStbObj *pStb) {
   for (int32_t i = 0; i < pStb->numOfSmas; ++i) {
     SSchema *pSchema = &pStb->pSmas[i];
     SDB_SET_INT8(pRaw, dataPos, pSchema->type, STB_ENCODE_OVER)
-    SDB_SET_INT32(pRaw, dataPos, pSchema->colId, STB_ENCODE_OVER)
+    SDB_SET_INT16(pRaw, dataPos, pSchema->colId, STB_ENCODE_OVER)
     SDB_SET_INT32(pRaw, dataPos, pSchema->bytes, STB_ENCODE_OVER)
     SDB_SET_BINARY(pRaw, dataPos, pSchema->name, TSDB_COL_NAME_LEN, STB_ENCODE_OVER)
   }
@@ -185,7 +185,7 @@ static SSdbRow *mndStbActionDecode(SSdbRaw *pRaw) {
   for (int32_t i = 0; i < pStb->numOfColumns; ++i) {
     SSchema *pSchema = &pStb->pColumns[i];
     SDB_GET_INT8(pRaw, dataPos, &pSchema->type, STB_DECODE_OVER)
-    SDB_GET_INT32(pRaw, dataPos, &pSchema->colId, STB_DECODE_OVER)
+    SDB_GET_INT16(pRaw, dataPos, &pSchema->colId, STB_DECODE_OVER)
     SDB_GET_INT32(pRaw, dataPos, &pSchema->bytes, STB_DECODE_OVER)
     SDB_GET_BINARY(pRaw, dataPos, pSchema->name, TSDB_COL_NAME_LEN, STB_DECODE_OVER)
   }
@@ -193,7 +193,7 @@ static SSdbRow *mndStbActionDecode(SSdbRaw *pRaw) {
   for (int32_t i = 0; i < pStb->numOfTags; ++i) {
     SSchema *pSchema = &pStb->pTags[i];
     SDB_GET_INT8(pRaw, dataPos, &pSchema->type, STB_DECODE_OVER)
-    SDB_GET_INT32(pRaw, dataPos, &pSchema->colId, STB_DECODE_OVER)
+    SDB_GET_INT16(pRaw, dataPos, &pSchema->colId, STB_DECODE_OVER)
     SDB_GET_INT32(pRaw, dataPos, &pSchema->bytes, STB_DECODE_OVER)
     SDB_GET_BINARY(pRaw, dataPos, pSchema->name, TSDB_COL_NAME_LEN, STB_DECODE_OVER)
   }
@@ -201,7 +201,7 @@ static SSdbRow *mndStbActionDecode(SSdbRaw *pRaw) {
   for (int32_t i = 0; i < pStb->numOfSmas; ++i) {
     SSchema *pSchema = &pStb->pSmas[i];
     SDB_GET_INT8(pRaw, dataPos, &pSchema->type, STB_DECODE_OVER)
-    SDB_GET_INT32(pRaw, dataPos, &pSchema->colId, STB_DECODE_OVER)
+    SDB_GET_INT16(pRaw, dataPos, &pSchema->colId, STB_DECODE_OVER)
     SDB_GET_INT32(pRaw, dataPos, &pSchema->bytes, STB_DECODE_OVER)
     SDB_GET_BINARY(pRaw, dataPos, pSchema->name, TSDB_COL_NAME_LEN, STB_DECODE_OVER)
   }

source/dnode/mnode/impl/src/mndStream.c

@@ -14,7 +14,6 @@
  */
 
 #include "mndStream.h"
-#include "parser.h"
 #include "mndAuth.h"
 #include "mndDb.h"
 #include "mndDnode.h"
@@ -26,6 +25,7 @@
 #include "mndTrans.h"
 #include "mndUser.h"
 #include "mndVgroup.h"
+#include "parser.h"
 #include "tname.h"
 
 #define MND_STREAM_VER_NUMBER   1
@@ -248,23 +248,22 @@ static int32_t mndStreamGetPlanString(const char *ast, char **pStr) {
 
 int32_t mndAddStreamToTrans(SMnode *pMnode, SStreamObj *pStream, const char *ast, STrans *pTrans) {
   SNode *pAst = NULL;
-#if 1 // TODO: remove debug info later
+
-  printf("ast = %s\n", ast); 
-#endif
   if (nodesStringToNode(ast, &pAst) < 0) {
     return -1;
   }
-#if 1
-  SSchemaWrapper sw = {0};
-  qExtractResultSchema(pAst, (int32_t*)&sw.nCols, &sw.pSchema);
 
+  if (qExtractResultSchema(pAst, (int32_t *)&pStream->outputSchema.nCols, &pStream->outputSchema.pSchema) != 0) {
+    return -1;
+  }
+
+#if 1
   printf("|");
-  for (int i = 0; i < sw.nCols; i++) {
+  for (int i = 0; i < pStream->outputSchema.nCols; i++) {
-    printf(" %15s |", (char *)sw.pSchema[i].name);
+    printf(" %15s |", (char *)pStream->outputSchema.pSchema[i].name);
   }
   printf("\n=======================================================\n");
 
-  pStream->ColAlias = NULL;
 #endif
 
   if (TSDB_CODE_SUCCESS != mndStreamGetPlanString(ast, &pStream->physicalPlan)) {

source/dnode/mnode/impl/src/mndTopic.c

@@ -23,6 +23,7 @@
 #include "mndTrans.h"
 #include "mndUser.h"
 #include "mndVgroup.h"
+#include "parser.h"
 #include "tname.h"
 
 #define MND_TOPIC_VER_NUMBER   1
@@ -85,6 +86,16 @@ SSdbRaw *mndTopicActionEncode(SMqTopicObj *pTopic) {
   SDB_SET_INT32(pRaw, dataPos, physicalPlanLen, TOPIC_ENCODE_OVER);
   SDB_SET_BINARY(pRaw, dataPos, pTopic->physicalPlan, physicalPlanLen, TOPIC_ENCODE_OVER);
 
+  int32_t swLen = taosEncodeSSchemaWrapper(NULL, &pTopic->schema);
+  void   *swBuf = taosMemoryMalloc(swLen);
+  if (swBuf == NULL) {
+    goto TOPIC_ENCODE_OVER;
+  }
+  void *aswBuf = swBuf;
+  taosEncodeSSchemaWrapper(&aswBuf, &pTopic->schema);
+  SDB_SET_INT32(pRaw, dataPos, swLen, TOPIC_ENCODE_OVER);
+  SDB_SET_BINARY(pRaw, dataPos, swBuf, swLen, TOPIC_ENCODE_OVER);
+
   SDB_SET_RESERVE(pRaw, dataPos, MND_TOPIC_RESERVE_SIZE, TOPIC_ENCODE_OVER);
   SDB_SET_DATALEN(pRaw, dataPos, TOPIC_ENCODE_OVER);
 
@@ -149,6 +160,17 @@ SSdbRow *mndTopicActionDecode(SSdbRaw *pRaw) {
   }
   SDB_GET_BINARY(pRaw, dataPos, pTopic->physicalPlan, len, TOPIC_DECODE_OVER);
 
+  SDB_GET_INT32(pRaw, dataPos, &len, TOPIC_DECODE_OVER);
+  void *buf = taosMemoryMalloc(len);
+  if (buf == NULL) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    goto TOPIC_DECODE_OVER;
+  }
+  SDB_GET_BINARY(pRaw, dataPos, buf, len, TOPIC_DECODE_OVER);
+  if (taosDecodeSSchemaWrapper(buf, &pTopic->schema) == NULL) {
+    goto TOPIC_DECODE_OVER;
+  }
+
   SDB_GET_RESERVE(pRaw, dataPos, MND_TOPIC_RESERVE_SIZE, TOPIC_DECODE_OVER);
 
   terrno = TSDB_CODE_SUCCESS;
@@ -283,6 +305,14 @@ static int32_t mndCreateTopic(SMnode *pMnode, SNodeMsg *pReq, SCMCreateTopicReq
     topicObj.physicalPlan = pPlanStr;
   }
 
+  SNode *pAst = NULL;
+  if (nodesStringToNode(pCreate->ast, &pAst) < 0) {
+    return -1;
+  }
+  if (qExtractResultSchema(pAst, &topicObj.schema.nCols, &topicObj.schema.pSchema) != 0) {
+    return -1;
+  }
+
   STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_ROLLBACK, TRN_TYPE_CREATE_TOPIC, &pReq->rpcMsg);
   if (pTrans == NULL) {
     mError("topic:%s, failed to create since %s", pCreate->name, terrstr());

source/dnode/mnode/impl/src/mndUser.c

@@ -21,7 +21,7 @@
 #include "mndTrans.h"
 #include "tbase64.h"
 
-#define TSDB_USER_VER_NUMBER 1
+#define TSDB_USER_VER_NUMBER   1
 #define TSDB_USER_RESERVE_SIZE 64
 
 static int32_t  mndCreateDefaultUsers(SMnode *pMnode);
@@ -270,7 +270,7 @@ static int32_t mndCreateUser(SMnode *pMnode, char *acct, SCreateUserReq *pCreate
   userObj.updateTime = userObj.createdTime;
   userObj.superUser = pCreate->superUser;
 
-  STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_ROLLBACK,TRN_TYPE_CREATE_USER, &pReq->rpcMsg);
+  STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_ROLLBACK, TRN_TYPE_CREATE_USER, &pReq->rpcMsg);
   if (pTrans == NULL) {
     mError("user:%s, failed to create since %s", pCreate->user, terrstr());
     return -1;
@@ -350,7 +350,7 @@ CREATE_USER_OVER:
 }
 
 static int32_t mndUpdateUser(SMnode *pMnode, SUserObj *pOld, SUserObj *pNew, SNodeMsg *pReq) {
-  STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_ROLLBACK, TRN_TYPE_ALTER_USER,&pReq->rpcMsg);
+  STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_ROLLBACK, TRN_TYPE_ALTER_USER, &pReq->rpcMsg);
   if (pTrans == NULL) {
     mError("user:%s, failed to update since %s", pOld->user, terrstr());
     return -1;
@@ -511,7 +511,7 @@ ALTER_USER_OVER:
 }
 
 static int32_t mndDropUser(SMnode *pMnode, SNodeMsg *pReq, SUserObj *pUser) {
-  STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_ROLLBACK,TRN_TYPE_DROP_USER, &pReq->rpcMsg);
+  STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_ROLLBACK, TRN_TYPE_DROP_USER, &pReq->rpcMsg);
   if (pTrans == NULL) {
     mError("user:%s, failed to drop since %s", pUser->user, terrstr());
     return -1;

source/dnode/vnode/src/meta/metaBDBImpl.c

@@ -233,7 +233,7 @@ int metaSaveSmaToDB(SMeta *pMeta, STSma *pSmaCfg) {
 
   // save sma info
   int32_t len = tEncodeTSma(NULL, pSmaCfg);
-  pBuf = taosMemoryCalloc(len, 1);
+  pBuf = taosMemoryCalloc(1, len);
   if (pBuf == NULL) {
     terrno = TSDB_CODE_OUT_OF_MEMORY;
     return -1;
@@ -285,7 +285,7 @@ static int metaEncodeSchema(void **buf, SSchemaWrapper *pSW) {
   for (int i = 0; i < pSW->nCols; i++) {
     pSchema = pSW->pSchema + i;
     tlen += taosEncodeFixedI8(buf, pSchema->type);
-    tlen += taosEncodeFixedI32(buf, pSchema->colId);
+    tlen += taosEncodeFixedI16(buf, pSchema->colId);
     tlen += taosEncodeFixedI32(buf, pSchema->bytes);
     tlen += taosEncodeString(buf, pSchema->name);
   }
@@ -301,7 +301,7 @@ static void *metaDecodeSchema(void *buf, SSchemaWrapper *pSW) {
   for (int i = 0; i < pSW->nCols; i++) {
     pSchema = pSW->pSchema + i;
     buf = taosDecodeFixedI8(buf, &pSchema->type);
-    buf = taosDecodeFixedI32(buf, &pSchema->colId);
+    buf = taosDecodeFixedI16(buf, &pSchema->colId);
     buf = taosDecodeFixedI32(buf, &pSchema->bytes);
     buf = taosDecodeStringTo(buf, pSchema->name);
   }
@@ -516,6 +516,7 @@ static int metaEncodeTbInfo(void **buf, STbCfg *pTbCfg) {
     tsize += taosEncodeFixedU64(buf, pTbCfg->ctbCfg.suid);
     tsize += tdEncodeKVRow(buf, pTbCfg->ctbCfg.pTag);
   } else if (pTbCfg->type == META_NORMAL_TABLE) {
+    // TODO
   } else {
     ASSERT(0);
   }
@@ -538,6 +539,7 @@ static void *metaDecodeTbInfo(void *buf, STbCfg *pTbCfg) {
     buf = taosDecodeFixedU64(buf, &(pTbCfg->ctbCfg.suid));
     buf = tdDecodeKVRow(buf, &(pTbCfg->ctbCfg.pTag));
   } else if (pTbCfg->type == META_NORMAL_TABLE) {
+    // TODO
   } else {
     ASSERT(0);
   }

source/dnode/vnode/src/tq/tq.c

@@ -42,8 +42,8 @@ STQ* tqOpen(const char* path, SVnode* pVnode, SWal* pWal, SMeta* pVnodeMeta, STq
     // TODO: error code of buffer pool
   }
 #endif
-  pTq->tqMeta =
+  pTq->tqMeta = tqStoreOpen(pTq, path, (FTqSerialize)tqSerializeConsumer, (FTqDeserialize)tqDeserializeConsumer,
-      tqStoreOpen(pTq, path, (FTqSerialize)tqSerializeConsumer, (FTqDeserialize)tqDeserializeConsumer, (FTqDelete)taosMemoryFree, 0);
+                            (FTqDelete)taosMemoryFree, 0);
   if (pTq->tqMeta == NULL) {
     taosMemoryFree(pTq);
 #if 0
@@ -498,12 +498,16 @@ int32_t tqProcessStreamTrigger(STQ* pTq, void* data, int32_t dataLen) {
 }
 
 int32_t tqProcessTaskExec(STQ* pTq, SRpcMsg* msg) {
-  SStreamTaskExecReq* pReq = msg->pCont;
+  char* msgstr = POINTER_SHIFT(msg->pCont, sizeof(SMsgHead));
-  int32_t             taskId = pReq->taskId;
+
-  SStreamTask*        pTask = taosHashGet(pTq->pStreamTasks, &taskId, sizeof(int32_t));
+  SStreamTaskExecReq req;
+  tDecodeSStreamTaskExecReq(msgstr, &req);
+
+  int32_t      taskId = req.taskId;
+  SStreamTask* pTask = taosHashGet(pTq->pStreamTasks, &taskId, sizeof(int32_t));
   ASSERT(pTask);
 
-  if (streamExecTask(pTask, &pTq->pVnode->msgCb, pReq->data, STREAM_DATA_TYPE_SSDATA_BLOCK, 0) < 0) {
+  if (streamExecTask(pTask, &pTq->pVnode->msgCb, req.data, STREAM_DATA_TYPE_SSDATA_BLOCK, 0) < 0) {
     // TODO
   }
   return 0;

source/dnode/vnode/src/tq/tqRead.c

@@ -130,8 +130,8 @@ SArray* tqRetrieveDataBlock(STqReadHandle* pHandle) {
   int32_t colNeed = 0;
   while (colMeta < pSchemaWrapper->nCols && colNeed < colNumNeed) {
     SSchema* pColSchema = &pSchemaWrapper->pSchema[colMeta];
-    int16_t  colIdSchema = pColSchema->colId;
+    col_id_t colIdSchema = pColSchema->colId;
-    int16_t  colIdNeed = *(int16_t*)taosArrayGet(pHandle->pColIdList, colNeed);
+    col_id_t colIdNeed = *(col_id_t*)taosArrayGet(pHandle->pColIdList, colNeed);
     if (colIdSchema < colIdNeed) {
       colMeta++;
     } else if (colIdSchema > colIdNeed) {
@@ -159,7 +159,7 @@ SArray* tqRetrieveDataBlock(STqReadHandle* pHandle) {
 
   int j = 0;
   for (int32_t i = 0; i < colNumNeed; i++) {
-    int16_t colId = *(int16_t*)taosArrayGet(pHandle->pColIdList, i);
+    col_id_t colId = *(col_id_t*)taosArrayGet(pHandle->pColIdList, i);
     while (j < pSchemaWrapper->nCols && pSchemaWrapper->pSchema[j].colId < colId) {
       j++;
     }

source/dnode/vnode/src/tsdb/tsdbCommit.c

@@ -1369,7 +1369,7 @@ int tsdbWriteBlockImpl(STsdb *pRepo, STable *pTable, SDFile *pDFile, SDFile *pDF
     }
   }
 
-  // Update pBlock membership vairables
+  // Update pBlock membership variables
   pBlock->last = isLast;
   pBlock->offset = offset;
   pBlock->algorithm = pCfg->compression;

source/dnode/vnode/src/tsdb/tsdbMemTable.c

@@ -255,7 +255,7 @@ int32_t tdScanAndConvertSubmitMsg(SSubmitReq *pMsg) {
   return 0;
 }
 
-int tsdbScanAndConvertSubmitMsg(STsdb *pTsdb, SSubmitReq *pMsg) {
+static int tsdbScanAndConvertSubmitMsg(STsdb *pTsdb, SSubmitReq *pMsg) {
   ASSERT(pMsg != NULL);
   // STsdbMeta *    pMeta = pTsdb->tsdbMeta;
   SSubmitMsgIter msgIter = {0};

source/dnode/vnode/src/tsdb/tsdbSma.c

@@ -38,9 +38,10 @@ typedef enum {
 } ESmaStorageLevel;
 
 typedef struct {
-  STsdb  *pTsdb;
+  STsdb       *pTsdb;
-  SDBFile dFile;
+  SDBFile      dFile;
-  int32_t interval;  // interval with the precision of DB
+  SSDataBlock *pData;     // sma data
+  int32_t      interval;  // interval with the precision of DB
 } STSmaWriteH;
 
 typedef struct {
@@ -98,7 +99,8 @@ static int32_t tsdbGetTSmaDataImpl(STsdb *pTsdb, STSmaDataWrapper *pData, int64_
                                    int32_t nMaxResult);
 
 // insert data
-static int32_t tsdbInitTSmaWriteH(STSmaWriteH *pSmaH, STsdb *pTsdb, STSmaDataWrapper *pData);
+static int32_t tsdbInitTSmaWriteH(STSmaWriteH *pSmaH, STsdb *pTsdb, SSDataBlock *pData, int64_t interval,
+                                  int8_t intervalUnit);
 static void    tsdbDestroyTSmaWriteH(STSmaWriteH *pSmaH);
 static int32_t tsdbInitTSmaReadH(STSmaReadH *pSmaH, STsdb *pTsdb, int64_t interval, int8_t intervalUnit);
 static int32_t tsdbGetSmaStorageLevel(int64_t interval, int8_t intervalUnit);
@@ -800,9 +802,10 @@ static int32_t tsdbInsertTSmaDataSection(STSmaWriteH *pSmaH, STSmaDataWrapper *p
   return TSDB_CODE_SUCCESS;
 }
 
-static int32_t tsdbInitTSmaWriteH(STSmaWriteH *pSmaH, STsdb *pTsdb, STSmaDataWrapper *pData) {
+static int32_t tsdbInitTSmaWriteH(STSmaWriteH *pSmaH, STsdb *pTsdb, SSDataBlock *pData, int64_t interval, int8_t intervalUnit) {
   pSmaH->pTsdb = pTsdb;
-  pSmaH->interval = tsdbGetIntervalByPrecision(pData->interval, pData->intervalUnit, REPO_CFG(pTsdb)->precision);
+  pSmaH->interval = tsdbGetIntervalByPrecision(interval, intervalUnit, REPO_CFG(pTsdb)->precision);
+  pSmaH->pData = pData;
   return TSDB_CODE_SUCCESS;
 }
 
@@ -857,10 +860,10 @@ static int32_t tsdbGetTSmaDays(STsdb *pTsdb, int64_t interval, int32_t storageLe
  * @return int32_t
  */
 static int32_t tsdbInsertTSmaDataImpl(STsdb *pTsdb, char *msg) {
-  STsdbCfg         *pCfg = REPO_CFG(pTsdb);
+  STsdbCfg    *pCfg = REPO_CFG(pTsdb);
-  STSmaDataWrapper *pData = (STSmaDataWrapper *)msg;
+  SSDataBlock *pData = (SSDataBlock *)msg;
-  SSmaEnv          *pEnv = atomic_load_ptr(&pTsdb->pTSmaEnv);
+  SSmaEnv     *pEnv = atomic_load_ptr(&pTsdb->pTSmaEnv);
-  int64_t           indexUid = SMA_TEST_INDEX_UID;
+  int64_t      indexUid = SMA_TEST_INDEX_UID;
 
   if (pEnv == NULL) {
     terrno = TSDB_CODE_INVALID_PTR;
@@ -868,15 +871,15 @@ static int32_t tsdbInsertTSmaDataImpl(STsdb *pTsdb, char *msg) {
     return terrno;
   }
 
-  if (pData->dataLen <= 0) {
+  if (pData == NULL) {
-    TASSERT(0);
+    terrno = TSDB_CODE_INVALID_PTR;
-    terrno = TSDB_CODE_INVALID_PARA;
+    tsdbWarn("vgId:%d insert tSma data failed since pData is NULL", REPO_ID(pTsdb));
-    return TSDB_CODE_FAILED;
+    return terrno;
   }
 
-  STSmaWriteH tSmaH = {0};
+  if (taosArrayGetSize(pData->pDataBlock) <= 0) {
-
+    terrno = TSDB_CODE_INVALID_PARA;
-  if (tsdbInitTSmaWriteH(&tSmaH, pTsdb, pData) != 0) {
+    tsdbWarn("vgId:%d insert tSma data failed since pDataBlock is empty", REPO_ID(pTsdb));
     return TSDB_CODE_FAILED;
   }
 
@@ -895,6 +898,14 @@ static int32_t tsdbInsertTSmaDataImpl(STsdb *pTsdb, char *msg) {
     return TSDB_CODE_FAILED;
   }
 
+  STSma *pSma = pItem->pSma;
+
+  STSmaWriteH tSmaH = {0};
+
+  if (tsdbInitTSmaWriteH(&tSmaH, pTsdb, pData, pSma->interval, pSma->intervalUnit) != 0) {
+    return TSDB_CODE_FAILED;
+  }
+
   char rPath[TSDB_FILENAME_LEN] = {0};
   char aPath[TSDB_FILENAME_LEN] = {0};
   snprintf(rPath, TSDB_FILENAME_LEN, "%s%s%" PRIi64, SMA_ENV_PATH(pEnv), TD_DIRSEP, indexUid);
@@ -907,8 +918,11 @@ static int32_t tsdbInsertTSmaDataImpl(STsdb *pTsdb, char *msg) {
   }
 
   // Step 1: Judge the storage level and days
-  int32_t storageLevel = tsdbGetSmaStorageLevel(pData->interval, pData->intervalUnit);
+  int32_t storageLevel = tsdbGetSmaStorageLevel(pSma->interval, pSma->intervalUnit);
   int32_t daysPerFile = tsdbGetTSmaDays(pTsdb, tSmaH.interval, storageLevel);
+
+
+#if 0  
   int32_t fid = (int32_t)(TSDB_KEY_FID(pData->skey, daysPerFile, pCfg->precision));
 
   // Step 2: Set the DFile for storage of SMA index, and iterate/split the TSma data and store to B+Tree index file
@@ -933,7 +947,7 @@ static int32_t tsdbInsertTSmaDataImpl(STsdb *pTsdb, char *msg) {
 
   // Step 3: reset the SSmaStat
   tsdbResetExpiredWindow(pTsdb, SMA_ENV_STAT(pTsdb->pTSmaEnv), pData->indexUid, pData->skey);
-
+#endif
   tsdbDestroyTSmaWriteH(&tSmaH);
   tsdbUnRefSmaStat(pTsdb, pStat);
   return TSDB_CODE_SUCCESS;
@@ -999,29 +1013,58 @@ static int32_t tsdbSetRSmaDataFile(STSmaWriteH *pSmaH, STSmaDataWrapper *pData,
 }
 
 static int32_t tsdbInsertRSmaDataImpl(STsdb *pTsdb, char *msg) {
-  STsdbCfg         *pCfg = REPO_CFG(pTsdb);
+  STsdbCfg    *pCfg = REPO_CFG(pTsdb);
-  STSmaDataWrapper *pData = (STSmaDataWrapper *)msg;
+  SSDataBlock *pData = (SSDataBlock *)msg;
-  SSmaEnv          *pEnv = atomic_load_ptr(&pTsdb->pRSmaEnv);
+  SSmaEnv     *pEnv = atomic_load_ptr(&pTsdb->pRSmaEnv);
+  int64_t      indexUid = SMA_TEST_INDEX_UID;
 
   if (pEnv == NULL) {
     terrno = TSDB_CODE_INVALID_PTR;
-    tsdbWarn("vgId:%d insert tSma data failed since pTSmaEnv is NULL", REPO_ID(pTsdb));
+    tsdbWarn("vgId:%d insert rSma data failed since pTSmaEnv is NULL", REPO_ID(pTsdb));
     return terrno;
   }
 
-  if (pData->dataLen <= 0) {
+  if (pEnv == NULL) {
-    TASSERT(0);
+    terrno = TSDB_CODE_INVALID_PTR;
+    tsdbWarn("vgId:%d insert rSma data failed since pTSmaEnv is NULL", REPO_ID(pTsdb));
+    return terrno;
+  }
+
+  if (pData == NULL) {
+    terrno = TSDB_CODE_INVALID_PTR;
+    tsdbWarn("vgId:%d insert rSma data failed since pData is NULL", REPO_ID(pTsdb));
+    return terrno;
+  }
+
+  if (taosArrayGetSize(pData->pDataBlock) <= 0) {
     terrno = TSDB_CODE_INVALID_PARA;
+    tsdbWarn("vgId:%d insert rSma data failed since pDataBlock is empty", REPO_ID(pTsdb));
     return TSDB_CODE_FAILED;
   }
 
+  SSmaStat     *pStat = SMA_ENV_STAT(pTsdb->pTSmaEnv);
+  SSmaStatItem *pItem = NULL;
+
+  tsdbRefSmaStat(pTsdb, pStat);
+
+  if (pStat && pStat->smaStatItems) {
+    pItem = taosHashGet(pStat->smaStatItems, &indexUid, sizeof(indexUid));
+  }
+
+  if ((pItem == NULL) || ((pItem = *(SSmaStatItem **)pItem) == NULL) || tsdbSmaStatIsDropped(pItem)) {
+    terrno = TSDB_CODE_TDB_INVALID_SMA_STAT;
+    tsdbUnRefSmaStat(pTsdb, pStat);
+    return TSDB_CODE_FAILED;
+  }
+
+  STSma *pSma = pItem->pSma;
+
   STSmaWriteH tSmaH = {0};
 
-  if (tsdbInitTSmaWriteH(&tSmaH, pTsdb, pData) != 0) {
+  if (tsdbInitTSmaWriteH(&tSmaH, pTsdb, pData, pSma->interval, pSma->intervalUnit) != 0) {
     return TSDB_CODE_FAILED;
   }
 
-  int64_t indexUid = SMA_TEST_INDEX_UID;
   char    rPath[TSDB_FILENAME_LEN] = {0};
   char    aPath[TSDB_FILENAME_LEN] = {0};
   snprintf(rPath, TSDB_FILENAME_LEN, "%s%s%" PRIi64, SMA_ENV_PATH(pEnv), TD_DIRSEP, indexUid);
@@ -1033,8 +1076,9 @@ static int32_t tsdbInsertRSmaDataImpl(STsdb *pTsdb, char *msg) {
   }
 
   // Step 1: Judge the storage level and days
-  int32_t storageLevel = tsdbGetSmaStorageLevel(pData->interval, pData->intervalUnit);
+  int32_t storageLevel = tsdbGetSmaStorageLevel(pSma->interval, pSma->intervalUnit);
   int32_t daysPerFile = tsdbGetTSmaDays(pTsdb, tSmaH.interval, storageLevel);
+  #if 0
   int32_t fid = (int32_t)(TSDB_KEY_FID(pData->skey, daysPerFile, pCfg->precision));
 
   // Step 2: Set the DFile for storage of SMA index, and iterate/split the TSma data and store to B+Tree index file
@@ -1057,8 +1101,10 @@ static int32_t tsdbInsertRSmaDataImpl(STsdb *pTsdb, char *msg) {
 
   // Step 3: reset the SSmaStat
   tsdbResetExpiredWindow(pTsdb, SMA_ENV_STAT(pTsdb->pTSmaEnv), pData->indexUid, pData->skey);
+#endif
 
   tsdbDestroyTSmaWriteH(&tSmaH);
+  tsdbUnRefSmaStat(pTsdb, pStat);
   return TSDB_CODE_SUCCESS;
 }
 

source/dnode/vnode/src/vnd/vnodeQuery.c

@@ -66,6 +66,8 @@ int vnodeProcessFetchMsg(SVnode *pVnode, SRpcMsg *pMsg) {
     case TDMT_VND_CONSUME:
       return tqProcessPollReq(pVnode->pTq, pMsg);
     case TDMT_VND_TASK_EXEC:
+    case TDMT_VND_TASK_PIPE_EXEC:
+    case TDMT_VND_TASK_MERGE_EXEC:
       return tqProcessTaskExec(pVnode->pTq, pMsg);
     case TDMT_VND_STREAM_TRIGGER:
       return tqProcessStreamTrigger(pVnode->pTq, pMsg->pCont, pMsg->contLen);

source/dnode/vnode/src/vnd/vnodeWrite.c

@@ -74,6 +74,9 @@ int vnodeApplyWMsg(SVnode *pVnode, SRpcMsg *pMsg, SRpcMsg **pRsp) {
       // TODO: maybe need to clear the request struct
       taosMemoryFree(vCreateTbReq.stbCfg.pSchema);
       taosMemoryFree(vCreateTbReq.stbCfg.pTagSchema);
+      taosMemoryFree(vCreateTbReq.stbCfg.pBSmaCols);
+      taosMemoryFree(vCreateTbReq.stbCfg.pRSmaParam);
+      taosMemoryFree(vCreateTbReq.dbFName);
       taosMemoryFree(vCreateTbReq.name);
       break;
     }
@@ -102,13 +105,18 @@ int vnodeApplyWMsg(SVnode *pVnode, SRpcMsg *pMsg, SRpcMsg **pRsp) {
           vError("vgId:%d, failed to create table: %s", pVnode->vgId, pCreateTbReq->name);
         }
         taosMemoryFree(pCreateTbReq->name);
+        taosMemoryFree(pCreateTbReq->dbFName);
         if (pCreateTbReq->type == TD_SUPER_TABLE) {
           taosMemoryFree(pCreateTbReq->stbCfg.pSchema);
           taosMemoryFree(pCreateTbReq->stbCfg.pTagSchema);
+          taosMemoryFree(pCreateTbReq->stbCfg.pBSmaCols);
+          taosMemoryFree(pCreateTbReq->stbCfg.pRSmaParam);
         } else if (pCreateTbReq->type == TD_CHILD_TABLE) {
           taosMemoryFree(pCreateTbReq->ctbCfg.pTag);
         } else {
           taosMemoryFree(pCreateTbReq->ntbCfg.pSchema);
+          taosMemoryFree(pCreateTbReq->ntbCfg.pBSmaCols);
+          taosMemoryFree(pCreateTbReq->ntbCfg.pRSmaParam);
         }
       }
 
@@ -135,6 +143,9 @@ int vnodeApplyWMsg(SVnode *pVnode, SRpcMsg *pMsg, SRpcMsg **pRsp) {
       tDeserializeSVCreateTbReq(POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead)), &vAlterTbReq);
       taosMemoryFree(vAlterTbReq.stbCfg.pSchema);
       taosMemoryFree(vAlterTbReq.stbCfg.pTagSchema);
+      taosMemoryFree(vAlterTbReq.stbCfg.pBSmaCols);
+      taosMemoryFree(vAlterTbReq.stbCfg.pRSmaParam);
+      taosMemoryFree(vAlterTbReq.dbFName);
       taosMemoryFree(vAlterTbReq.name);
       break;
     }
@@ -176,7 +187,7 @@ int vnodeApplyWMsg(SVnode *pVnode, SRpcMsg *pMsg, SRpcMsg **pRsp) {
       }
 
       // record current timezone of server side
-      tstrncpy(vCreateSmaReq.tSma.timezone, tsTimezone, TD_TIMEZONE_LEN);
+      tstrncpy(vCreateSmaReq.tSma.timezone, tsTimezoneStr, TD_TIMEZONE_LEN);
 
       if (metaCreateTSma(pVnode->pMeta, &vCreateSmaReq) < 0) {
         // TODO: handle error

source/dnode/vnode/test/tsdbSmaTest.cpp

@@ -280,7 +280,7 @@ TEST(testCase, tSma_metaDB_Put_Get_Del_Test) {
 }
 #endif
 
-#if 1
+#if 0
 TEST(testCase, tSma_Data_Insert_Query_Test) {
   // step 1: prepare meta
   const char    *smaIndexName1 = "sma_index_test_1";

source/libs/catalog/src/catalog.c

@@ -53,7 +53,6 @@ SCtgAction gCtgAction[CTG_ACT_MAX] = {{
                           }
 };
 
-
 void ctgFreeMetaRent(SCtgRentMgmt *mgmt) {
   if (NULL == mgmt->slots) {
     return;

source/libs/executor/inc/executorimpl.h

@@ -415,6 +415,7 @@ typedef struct STableScanInfo {
   int32_t*        rowCellInfoOffset;
   SExprInfo*      pExpr;
   SSDataBlock     block;
+  SArray*         pColMatchInfo;
   int32_t         numOfOutput;
   int64_t         elapsedTime;
   int32_t         prevGroupId;  // previous table group id
@@ -648,8 +649,8 @@ typedef struct SDistinctOperatorInfo {
 } SDistinctOperatorInfo;
 
 SOperatorInfo* createExchangeOperatorInfo(const SNodeList* pSources, SSDataBlock* pBlock, SExecTaskInfo* pTaskInfo);
-SOperatorInfo* createTableScanOperatorInfo(void* pTsdbReadHandle, int32_t order, int32_t numOfOutput,
+SOperatorInfo* createTableScanOperatorInfo(void* pTsdbReadHandle, int32_t order, int32_t numOfCols, int32_t repeatTime,
-                                           int32_t repeatTime, int32_t reverseTime, SExecTaskInfo* pTaskInfo);
+                                           int32_t reverseTime, SArray* pColMatchInfo, SExecTaskInfo* pTaskInfo);
 SOperatorInfo* createAggregateOperatorInfo(SOperatorInfo* downstream, SExprInfo* pExprInfo, int32_t numOfCols, SSDataBlock* pResultBlock,
                                            SExecTaskInfo* pTaskInfo, const STableGroupInfo* pTableGroupInfo);
 SOperatorInfo* createMultiTableAggOperatorInfo(SOperatorInfo* downstream, SExprInfo* pExprInfo, int32_t numOfCols, SSDataBlock* pResBlock, SExecTaskInfo* pTaskInfo, const STableGroupInfo* pTableGroupInfo);

source/libs/executor/src/executor.c

@@ -97,6 +97,8 @@ qTaskInfo_t qCreateStreamExecTaskInfo(void* msg, void* streamReadHandle) {
   pMsg->contentLen = pMsg->contentLen;
 #endif
 
+  qDebugL("stream task string %s", (const char*)msg);
+
   struct SSubplan* plan = NULL;
   int32_t          code = qStringToSubplan(msg, &plan);
   if (code != TSDB_CODE_SUCCESS) {

source/libs/executor/src/executorimpl.c

@@ -66,6 +66,11 @@ typedef enum SResultTsInterpType {
   RESULT_ROW_END_INTERP   = 2,
 } SResultTsInterpType;
 
+typedef struct SColMatchInfo {
+  int32_t colId;
+  int32_t targetSlotId;
+} SColMatchInfo;
+
 #if 0
 static UNUSED_FUNC void *u_malloc (size_t __size) {
   uint32_t v = taosRand();
@@ -2944,12 +2949,21 @@ int32_t loadDataBlock(SExecTaskInfo *pTaskInfo, STableScanInfo* pTableScanInfo,
 
   *status = BLK_DATA_ALL_NEEDED;
 
-  pBlock->pDataBlock = tsdbRetrieveDataBlock(pTableScanInfo->pTsdbReadHandle, NULL);
+  SArray* pCols = tsdbRetrieveDataBlock(pTableScanInfo->pTsdbReadHandle, NULL);
-  if (pBlock->pDataBlock == NULL) {
+  if (pCols == NULL) {
     return terrno;
-  } else {
-    return TSDB_CODE_SUCCESS;
   }
+
+    int32_t numOfCols = pBlock->info.numOfCols;
+    for(int32_t i = 0; i < numOfCols; ++i) {
+      SColumnInfoData* p = taosArrayGet(pCols, i);
+      SColMatchInfo* pColMatchInfo = taosArrayGet(pTableScanInfo->pColMatchInfo, i);
+      ASSERT(pColMatchInfo->colId == p->info.colId);
+
+      taosArraySet(pBlock->pDataBlock, pColMatchInfo->targetSlotId, p);
+    }
+
+    return TSDB_CODE_SUCCESS;
 }
 
 int32_t loadDataBlockOnDemand(SExecTaskInfo *pTaskInfo, STableScanInfo* pTableScanInfo, SSDataBlock* pBlock, uint32_t* status) {
@@ -5374,7 +5388,8 @@ SSDataBlock* createResultDataBlock(const SArray* pExprInfo) {
   return pResBlock;
 }
 
-SOperatorInfo* createTableScanOperatorInfo(void* pTsdbReadHandle, int32_t order, int32_t numOfOutput, int32_t repeatTime, int32_t reverseTime, SExecTaskInfo* pTaskInfo) {
+SOperatorInfo* createTableScanOperatorInfo(void* pTsdbReadHandle, int32_t order, int32_t numOfOutput, int32_t repeatTime, int32_t reverseTime, SArray* pColMatchInfo,
+    SExecTaskInfo* pTaskInfo) {
   assert(repeatTime > 0);
 
   STableScanInfo* pInfo    = taosMemoryCalloc(1, sizeof(STableScanInfo));
@@ -5387,12 +5402,19 @@ SOperatorInfo* createTableScanOperatorInfo(void* pTsdbReadHandle, int32_t order,
     return NULL;
   }
 
+  pInfo->block.pDataBlock = taosArrayInit(numOfOutput, sizeof(SColumnInfoData));
+  for(int32_t i = 0; i < numOfOutput; ++i) {
+    SColumnInfoData idata = {0};
+    taosArrayPush(pInfo->block.pDataBlock, &idata);
+  }
+
   pInfo->pTsdbReadHandle = pTsdbReadHandle;
   pInfo->times             = repeatTime;
   pInfo->reverseTimes      = reverseTime;
   pInfo->order             = order;
   pInfo->current           = 0;
   pInfo->scanFlag          = MAIN_SCAN;
+  pInfo->pColMatchInfo     = pColMatchInfo;
   pOperator->name          = "TableScanOperator";
   pOperator->operatorType  = QUERY_NODE_PHYSICAL_PLAN_TABLE_SCAN;
   pOperator->blockingOptr  = false;
@@ -8569,6 +8591,7 @@ static int32_t doCreateTableGroup(void* metaHandle, int32_t tableType, uint64_t
 static SArray* extractTableIdList(const STableGroupInfo* pTableGroupInfo);
 static SArray* extractScanColumnId(SNodeList* pNodeList);
 static SArray* extractColumnInfo(SNodeList* pNodeList);
+static SArray* extractColMatchInfo(SNodeList* pNodeList);
 
 SOperatorInfo* doCreateOperatorTreeNode(SPhysiNode* pPhyNode, SExecTaskInfo* pTaskInfo, SReadHandle* pHandle, uint64_t queryId, uint64_t taskId, STableGroupInfo* pTableGroupInfo) {
   if (pPhyNode->pChildren == NULL || LIST_LENGTH(pPhyNode->pChildren) == 0) {
@@ -8577,7 +8600,9 @@ SOperatorInfo* doCreateOperatorTreeNode(SPhysiNode* pPhyNode, SExecTaskInfo* pTa
 
       size_t      numOfCols = LIST_LENGTH(pScanPhyNode->pScanCols);
       tsdbReaderT pDataReader = doCreateDataReader((STableScanPhysiNode*)pPhyNode, pHandle, pTableGroupInfo, (uint64_t)queryId, taskId);
-      return createTableScanOperatorInfo(pDataReader, pScanPhyNode->order, numOfCols, pScanPhyNode->count, pScanPhyNode->reverse, pTaskInfo);
+      SArray* pColList = extractColMatchInfo(pScanPhyNode->pScanCols);
+
+      return createTableScanOperatorInfo(pDataReader, pScanPhyNode->order, numOfCols, pScanPhyNode->count, pScanPhyNode->reverse, pColList, pTaskInfo);
     } else if (QUERY_NODE_PHYSICAL_PLAN_EXCHANGE == nodeType(pPhyNode)) {
       SExchangePhysiNode* pExchange = (SExchangePhysiNode*)pPhyNode;
       SSDataBlock* pResBlock = createOutputBuf_rv1(pExchange->node.pOutputDataBlockDesc);
@@ -8717,9 +8742,14 @@ SArray* extractScanColumnId(SNodeList* pNodeList) {
   }
 
   for(int32_t i = 0; i < numOfCols; ++i) {
-    STargetNode* pNode = (STargetNode*) nodesListGetNode(pNodeList, i);
+    for (int32_t j = 0; j < numOfCols; ++j) {
-    SColumnNode* pColNode = (SColumnNode*) pNode->pExpr;
+      STargetNode* pNode = (STargetNode*) nodesListGetNode(pNodeList, j);
-    taosArrayPush(pList, &pColNode->colId);
+      if (pNode->slotId == i) {
+        SColumnNode* pColNode = (SColumnNode*) pNode->pExpr;
+        taosArrayPush(pList, &pColNode->colId);
+        break;
+      }
+    }
   }
 
   return pList;
@@ -8751,6 +8781,28 @@ SArray* extractColumnInfo(SNodeList* pNodeList) {
   return pList;
 }
 
+SArray* extractColMatchInfo(SNodeList* pNodeList) {
+  size_t numOfCols = LIST_LENGTH(pNodeList);
+  SArray* pList = taosArrayInit(numOfCols, sizeof(SColMatchInfo));
+  if (pList == NULL) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return NULL;
+  }
+
+  for(int32_t i = 0; i < numOfCols; ++i) {
+    STargetNode* pNode = (STargetNode*) nodesListGetNode(pNodeList, i);
+    SColumnNode* pColNode = (SColumnNode*) pNode->pExpr;
+
+    SColMatchInfo c = {0};
+    c.colId = pColNode->colId;
+    c.targetSlotId = pNode->slotId;
+
+    taosArrayPush(pList, &c);
+  }
+
+  return pList;
+}
+
 int32_t doCreateTableGroup(void* metaHandle, int32_t tableType, uint64_t tableUid, STableGroupInfo* pGroupInfo, uint64_t queryId, uint64_t taskId) {
   int32_t code = 0;
   if (tableType == TSDB_SUPER_TABLE) {

source/libs/nodes/src/nodesCloneFuncs.c

@@ -19,6 +19,11 @@
 #include "taos.h"
 #include "taoserror.h"
 
+#define COPY_ALL_SCALAR_FIELDS \
+	do { \
+    memcpy((pDst), (pSrc), sizeof(*pSrc)); \
+	} while (0)
+
 #define COPY_SCALAR_FIELD(fldname) \
 	do { \
     (pDst)->fldname = (pSrc)->fldname; \
@@ -195,6 +200,12 @@ static SNode* groupingSetNodeCopy(const SGroupingSetNode* pSrc, SGroupingSetNode
   return (SNode*)pDst;
 }
 
+static SNode* orderByExprNodeCopy(const SOrderByExprNode* pSrc, SOrderByExprNode* pDst) {
+  COPY_ALL_SCALAR_FIELDS;
+  CLONE_NODE_FIELD(pExpr);
+  return (SNode*)pDst;
+}
+
 static SNode* fillNodeCopy(const SFillNode* pSrc, SFillNode* pDst) {
   COPY_SCALAR_FIELD(mode);
   CLONE_NODE_FIELD(pValues);
@@ -251,6 +262,7 @@ static SNode* logicAggCopy(const SAggLogicNode* pSrc, SAggLogicNode* pDst) {
 static SNode* logicProjectCopy(const SProjectLogicNode* pSrc, SProjectLogicNode* pDst) {
   COPY_BASE_OBJECT_FIELD(node, logicNodeCopy);
   CLONE_NODE_LIST_FIELD(pProjections);
+  COPY_CHAR_ARRAY_FIELD(stmtName);
   return (SNode*)pDst;
 }
 
@@ -267,16 +279,24 @@ static SNode* logicExchangeCopy(const SExchangeLogicNode* pSrc, SExchangeLogicNo
 }
 
 static SNode* logicWindowCopy(const SWindowLogicNode* pSrc, SWindowLogicNode* pDst) {
+  COPY_ALL_SCALAR_FIELDS;
   COPY_BASE_OBJECT_FIELD(node, logicNodeCopy);
-  COPY_SCALAR_FIELD(winType);
+  // COPY_SCALAR_FIELD(winType);
   CLONE_NODE_LIST_FIELD(pFuncs);
-  COPY_SCALAR_FIELD(interval);
+  // COPY_SCALAR_FIELD(interval);
-  COPY_SCALAR_FIELD(offset);
+  // COPY_SCALAR_FIELD(offset);
-  COPY_SCALAR_FIELD(sliding);
+  // COPY_SCALAR_FIELD(sliding);
-  COPY_SCALAR_FIELD(intervalUnit);
+  // COPY_SCALAR_FIELD(intervalUnit);
-  COPY_SCALAR_FIELD(slidingUnit);
+  // COPY_SCALAR_FIELD(slidingUnit);
   CLONE_NODE_FIELD(pFill);
-  COPY_SCALAR_FIELD(sessionGap);
+  // COPY_SCALAR_FIELD(sessionGap);
+  CLONE_NODE_FIELD(pTspk);
+  return (SNode*)pDst;
+}
+
+static SNode* logicSortCopy(const SSortLogicNode* pSrc, SSortLogicNode* pDst) {
+  COPY_BASE_OBJECT_FIELD(node, logicNodeCopy);
+  CLONE_NODE_LIST_FIELD(pSortKeys);
   return (SNode*)pDst;
 }
 
@@ -339,6 +359,7 @@ SNodeptr nodesCloneNode(const SNodeptr pNode) {
     case QUERY_NODE_GROUPING_SET:
       return groupingSetNodeCopy((const SGroupingSetNode*)pNode, (SGroupingSetNode*)pDst);
     case QUERY_NODE_ORDER_BY_EXPR:
+      return orderByExprNodeCopy((const SOrderByExprNode*)pNode, (SOrderByExprNode*)pDst);
     case QUERY_NODE_LIMIT:
       break;
     case QUERY_NODE_FILL:
@@ -361,6 +382,8 @@ SNodeptr nodesCloneNode(const SNodeptr pNode) {
       return logicExchangeCopy((const SExchangeLogicNode*)pNode, (SExchangeLogicNode*)pDst);
     case QUERY_NODE_LOGIC_PLAN_WINDOW:
       return logicWindowCopy((const SWindowLogicNode*)pNode, (SWindowLogicNode*)pDst);
+    case QUERY_NODE_LOGIC_PLAN_SORT:
+      return logicSortCopy((const SSortLogicNode*)pNode, (SSortLogicNode*)pDst);
     case QUERY_NODE_LOGIC_SUBPLAN:
       return logicSubplanCopy((const SLogicSubplan*)pNode, (SLogicSubplan*)pDst);
     default:

source/libs/nodes/src/nodesCodeFuncs.c

@@ -851,9 +851,7 @@ static int32_t jsonToPhysiJoinNode(const SJson* pJson, void* pObj) {
 
   int32_t code = jsonToPhysicPlanNode(pJson, pObj);
   if (TSDB_CODE_SUCCESS == code) {
-    int32_t val;
+    code = tjsonGetNumberValue(pJson, jkJoinPhysiPlanJoinType, pNode->joinType);
-    code = tjsonGetIntValue(pJson, jkJoinPhysiPlanJoinType, &val);
-    pNode->joinType = val;
   }
   if (TSDB_CODE_SUCCESS == code) {
     code = jsonToNodeObject(pJson, jkJoinPhysiPlanOnConditions, &pNode->pOnConditions);
@@ -934,6 +932,37 @@ static int32_t jsonToPhysiExchangeNode(const SJson* pJson, void* pObj) {
   return code;
 }
 
+static const char* jkSortPhysiPlanExprs = "Exprs";
+static const char* jkSortPhysiPlanSortKeys = "SortKeys";
+
+static int32_t physiSortNodeToJson(const void* pObj, SJson* pJson) {
+  const SSortPhysiNode* pNode = (const SSortPhysiNode*)pObj;
+
+  int32_t code = physicPlanNodeToJson(pObj, pJson);
+  if (TSDB_CODE_SUCCESS == code) {
+    code = nodeListToJson(pJson, jkSortPhysiPlanExprs, pNode->pExprs);
+  }
+  if (TSDB_CODE_SUCCESS == code) {
+    code = nodeListToJson(pJson, jkSortPhysiPlanSortKeys, pNode->pSortKeys);
+  }
+
+  return code;
+}
+
+static int32_t jsonToPhysiSortNode(const SJson* pJson, void* pObj) {
+  SSortPhysiNode* pNode = (SSortPhysiNode*)pObj;
+
+  int32_t code = jsonToPhysicPlanNode(pJson, pObj);
+  if (TSDB_CODE_SUCCESS == code) {
+    code = jsonToNodeList(pJson, jkSortPhysiPlanExprs, &pNode->pExprs);
+  }
+  if (TSDB_CODE_SUCCESS == code) {
+    code = jsonToNodeList(pJson, jkSortPhysiPlanSortKeys, &pNode->pSortKeys);
+  }
+
+  return code;
+}
+
 static const char* jkWindowPhysiPlanExprs = "Exprs";
 static const char* jkWindowPhysiPlanFuncs = "Funcs";
 
@@ -971,6 +1000,7 @@ static const char* jkIntervalPhysiPlanSliding = "Sliding";
 static const char* jkIntervalPhysiPlanIntervalUnit = "intervalUnit";
 static const char* jkIntervalPhysiPlanSlidingUnit  = "slidingUnit";
 static const char* jkIntervalPhysiPlanFill = "Fill";
+static const char* jkIntervalPhysiPlanTsPk = "TsPk";
 
 static int32_t physiIntervalNodeToJson(const void* pObj, SJson* pJson) {
   const SIntervalPhysiNode* pNode = (const SIntervalPhysiNode*)pObj;
@@ -994,6 +1024,9 @@ static int32_t physiIntervalNodeToJson(const void* pObj, SJson* pJson) {
   if (TSDB_CODE_SUCCESS == code) {
     code = tjsonAddObject(pJson, jkIntervalPhysiPlanFill, nodeToJson, pNode->pFill);
   }
+  if (TSDB_CODE_SUCCESS == code) {
+    code = tjsonAddObject(pJson, jkIntervalPhysiPlanTsPk, nodeToJson, pNode->pTspk);
+  }
 
   return code;
 }
@@ -1020,6 +1053,9 @@ static int32_t jsonToPhysiIntervalNode(const SJson* pJson, void* pObj) {
   if (TSDB_CODE_SUCCESS == code) {
     code = jsonToNodeObject(pJson, jkIntervalPhysiPlanFill, (SNode**)&pNode->pFill);
   }
+  if (TSDB_CODE_SUCCESS == code) {
+    code = jsonToNodeObject(pJson, jkIntervalPhysiPlanTsPk, (SNode**)&pNode->pTspk);
+  }
 
   return code;
 }
@@ -1178,9 +1214,7 @@ static int32_t jsonToSubplan(const SJson* pJson, void* pObj) {
 
   int32_t code = tjsonToObject(pJson, jkSubplanId, jsonToSubplanId, &pNode->id);
   if (TSDB_CODE_SUCCESS == code) {
-    int32_t val;
+    code = tjsonGetNumberValue(pJson, jkSubplanType, pNode->subplanType);
-    code = tjsonGetIntValue(pJson, jkSubplanType, &val);
-    pNode->subplanType = val;
   }
   if (TSDB_CODE_SUCCESS == code) {
     code = tjsonGetIntValue(pJson, jkSubplanMsgType, &pNode->msgType);
@@ -1370,9 +1404,7 @@ static int32_t jsonToColumnNode(const SJson* pJson, void* pObj) {
     code = tjsonGetSmallIntValue(pJson, jkColumnColId, &pNode->colId);
   }
   if (TSDB_CODE_SUCCESS == code) {
-    int32_t tmp;
+    code = tjsonGetNumberValue(pJson, jkColumnColType, pNode->colType);
-    code = tjsonGetIntValue(pJson, jkColumnColType, &tmp);
-    pNode->colType = tmp;
   }
   if (TSDB_CODE_SUCCESS == code) {
     code = tjsonGetStringValue(pJson, jkColumnDbName, pNode->dbName);
@@ -1560,9 +1592,7 @@ static int32_t jsonToOperatorNode(const SJson* pJson, void* pObj) {
 
   int32_t code = jsonToExprNode(pJson, pObj);
   if (TSDB_CODE_SUCCESS == code) {
-    int32_t val;
+    code = tjsonGetNumberValue(pJson, jkOperatorType, pNode->opType);
-    code = tjsonGetIntValue(pJson, jkOperatorType, &val);
-    pNode->opType = val;
   }
   if (TSDB_CODE_SUCCESS == code) {
     code = jsonToNodeObject(pJson, jkOperatorLeft, &pNode->pLeft);
@@ -1596,9 +1626,7 @@ static int32_t jsonToLogicConditionNode(const SJson* pJson, void* pObj) {
 
   int32_t code = jsonToExprNode(pJson, pObj);
   if (TSDB_CODE_SUCCESS == code) {
-    int32_t val;
+    code = tjsonGetNumberValue(pJson, jkLogicCondType, pNode->condType);
-    code = tjsonGetIntValue(pJson, jkLogicCondType, &val);
-    pNode->condType = val;
   }
   if (TSDB_CODE_SUCCESS == code) {
     code = jsonToNodeList(pJson, jkLogicCondParameters, &pNode->pParameterList);
@@ -1821,10 +1849,43 @@ static int32_t groupingSetNodeToJson(const void* pObj, SJson* pJson) {
   return code;
 }
 
+static const char* jkOrderByExprExpr = "Expr";
+static const char* jkOrderByExprOrder = "Order";
+static const char* jkOrderByExprNullOrder = "NullOrder";
+
+static int32_t orderByExprNodeToJson(const void* pObj, SJson* pJson) {
+  const SOrderByExprNode* pNode = (const SOrderByExprNode*)pObj;
+
+  int32_t code = tjsonAddObject(pJson, jkOrderByExprExpr, nodeToJson, pNode->pExpr);
+  if (TSDB_CODE_SUCCESS == code) {
+    code = tjsonAddIntegerToObject(pJson, jkOrderByExprOrder, pNode->order);
+  }
+  if (TSDB_CODE_SUCCESS == code) {
+    code = tjsonAddIntegerToObject(pJson, jkOrderByExprNullOrder, pNode->nullOrder);
+  }
+
+  return code;
+}
+
+static int32_t jsonToOrderByExprNode(const SJson* pJson, void* pObj) {
+  SOrderByExprNode* pNode = (SOrderByExprNode*)pObj;
+
+  int32_t code = jsonToNodeObject(pJson, jkOrderByExprExpr, &pNode->pExpr);
+  if (TSDB_CODE_SUCCESS == code) {
+    code = tjsonGetNumberValue(pJson, jkOrderByExprOrder, pNode->order);
+  }
+  if (TSDB_CODE_SUCCESS == code) {
+    code = tjsonGetNumberValue(pJson, jkOrderByExprNullOrder, pNode->nullOrder);
+  }
+
+  return code;
+}
+
 static const char* jkIntervalWindowInterval = "Interval";
 static const char* jkIntervalWindowOffset = "Offset";
 static const char* jkIntervalWindowSliding = "Sliding";
 static const char* jkIntervalWindowFill = "Fill";
+static const char* jkIntervalWindowTsPk = "TsPk";
 
 static int32_t intervalWindowNodeToJson(const void* pObj, SJson* pJson) {
   const SIntervalWindowNode* pNode = (const SIntervalWindowNode*)pObj;
@@ -1839,6 +1900,9 @@ static int32_t intervalWindowNodeToJson(const void* pObj, SJson* pJson) {
   if (TSDB_CODE_SUCCESS == code) {
     code = tjsonAddObject(pJson, jkIntervalWindowFill, nodeToJson, pNode->pFill);
   }
+  if (TSDB_CODE_SUCCESS == code) {
+    code = tjsonAddObject(pJson, jkIntervalWindowTsPk, nodeToJson, pNode->pCol);
+  }
 
   return code;
 }
@@ -1856,6 +1920,9 @@ static int32_t jsonToIntervalWindowNode(const SJson* pJson, void* pObj) {
   if (TSDB_CODE_SUCCESS == code) {
     code = jsonToNodeObject(pJson, jkIntervalWindowFill, &pNode->pFill);
   }
+  if (TSDB_CODE_SUCCESS == code) {
+    code = jsonToNodeObject(pJson, jkIntervalWindowTsPk, &pNode->pCol);
+  }
 
   return code;
 }
@@ -2033,6 +2100,7 @@ static const char* jkSelectStmtHaving = "Having";
 static const char* jkSelectStmtOrderBy = "OrderBy";
 static const char* jkSelectStmtLimit = "Limit";
 static const char* jkSelectStmtSlimit = "Slimit";
+static const char* jkSelectStmtStmtName = "StmtName";
 
 static int32_t selectStmtTojson(const void* pObj, SJson* pJson) {
   const SSelectStmt* pNode = (const SSelectStmt*)pObj;
@@ -2068,6 +2136,9 @@ static int32_t selectStmtTojson(const void* pObj, SJson* pJson) {
   if (TSDB_CODE_SUCCESS == code) {
     code = tjsonAddObject(pJson, jkSelectStmtSlimit, nodeToJson, pNode->pSlimit);
   }
+  if (TSDB_CODE_SUCCESS == code) {
+    code = tjsonAddStringToObject(pJson, jkSelectStmtStmtName, pNode->stmtName);
+  }
 
   return code;
 }
@@ -2106,6 +2177,9 @@ static int32_t jsonToSelectStmt(const SJson* pJson, void* pObj) {
   if (TSDB_CODE_SUCCESS == code) {
     code = jsonToNodeObject(pJson, jkSelectStmtSlimit, &pNode->pSlimit);
   }
+  if (TSDB_CODE_SUCCESS == code) {
+    code = tjsonGetStringValue(pJson, jkSelectStmtStmtName, pNode->stmtName);
+  }
 
   return code;
 }
@@ -2169,6 +2243,7 @@ static int32_t specificNodeToJson(const void* pObj, SJson* pJson) {
     case QUERY_NODE_GROUPING_SET:
       return groupingSetNodeToJson(pObj, pJson);
     case QUERY_NODE_ORDER_BY_EXPR:
+      return orderByExprNodeToJson(pObj, pJson);
     case QUERY_NODE_LIMIT:
     case QUERY_NODE_STATE_WINDOW:
     case QUERY_NODE_SESSION_WINDOW:
@@ -2232,7 +2307,7 @@ static int32_t specificNodeToJson(const void* pObj, SJson* pJson) {
     case QUERY_NODE_PHYSICAL_PLAN_EXCHANGE:
       return physiExchangeNodeToJson(pObj, pJson);
     case QUERY_NODE_PHYSICAL_PLAN_SORT:
-      break;
+      return physiSortNodeToJson(pObj, pJson);
     case QUERY_NODE_PHYSICAL_PLAN_INTERVAL:
       return physiIntervalNodeToJson(pObj, pJson);
     case QUERY_NODE_PHYSICAL_PLAN_SESSION_WINDOW:
@@ -2272,7 +2347,8 @@ static int32_t jsonToSpecificNode(const SJson* pJson, void* pObj) {
     //   break;
     // case QUERY_NODE_GROUPING_SET:
     //   return jsonToGroupingSetNode(pJson, pObj);
-    // case QUERY_NODE_ORDER_BY_EXPR:
+    case QUERY_NODE_ORDER_BY_EXPR:
+      return jsonToOrderByExprNode(pJson, pObj);
     // case QUERY_NODE_LIMIT:
     // case QUERY_NODE_STATE_WINDOW:
     // case QUERY_NODE_SESSION_WINDOW:
@@ -2321,6 +2397,8 @@ static int32_t jsonToSpecificNode(const SJson* pJson, void* pObj) {
       return jsonToPhysiAggNode(pJson, pObj);
     case QUERY_NODE_PHYSICAL_PLAN_EXCHANGE:
       return jsonToPhysiExchangeNode(pJson, pObj);
+    case QUERY_NODE_PHYSICAL_PLAN_SORT:
+      return jsonToPhysiSortNode(pJson, pObj);
     case QUERY_NODE_PHYSICAL_PLAN_INTERVAL:
       return jsonToPhysiIntervalNode(pJson, pObj);
     case QUERY_NODE_PHYSICAL_PLAN_SESSION_WINDOW:
@@ -2362,9 +2440,7 @@ static int32_t nodeToJson(const void* pObj, SJson* pJson) {
 static int32_t jsonToNode(const SJson* pJson, void* pObj) {
   SNode* pNode = (SNode*)pObj;
 
-  int32_t val = 0;
+  int32_t code = tjsonGetNumberValue(pJson, jkNodeType, pNode->type);
-  int32_t code = tjsonGetIntValue(pJson, jkNodeType, &val);
-  pNode->type = val;
   if (TSDB_CODE_SUCCESS == code) {
     code = tjsonToObject(pJson, nodesNodeName(pNode->type), jsonToSpecificNode, pNode);
     if (TSDB_CODE_SUCCESS != code) {

source/libs/nodes/src/nodesTraverseFuncs.c

@@ -99,6 +99,9 @@ static EDealRes walkNode(SNode* pNode, ETraversalOrder order, FNodeWalker walker
       if (DEAL_RES_ERROR != res) {
         res = walkNode(pInterval->pFill, order, walker, pContext);
       }
+      if (DEAL_RES_ERROR != res) {
+        res = walkNode(pInterval->pCol, order, walker, pContext);
+      }
       break;
     }
     case QUERY_NODE_NODE_LIST:
@@ -225,6 +228,9 @@ static EDealRes rewriteNode(SNode** pRawNode, ETraversalOrder order, FNodeRewrit
       if (DEAL_RES_ERROR != res) {
         res = rewriteNode(&(pInterval->pFill), order, rewriter, pContext);
       }
+      if (DEAL_RES_ERROR != res) {
+        res = rewriteNode(&(pInterval->pCol), order, rewriter, pContext);
+      }
       break;
     }
     case QUERY_NODE_NODE_LIST:
@@ -294,10 +300,10 @@ void nodesWalkSelectStmt(SSelectStmt* pSelect, ESqlClause clause, FNodeWalker wa
     case SQL_CLAUSE_GROUP_BY:
       nodesWalkNode(pSelect->pHaving, walker, pContext);
     case SQL_CLAUSE_HAVING:
-      nodesWalkList(pSelect->pProjectionList, walker, pContext);
-    case SQL_CLAUSE_SELECT:
       nodesWalkList(pSelect->pOrderByList, walker, pContext);
     case SQL_CLAUSE_ORDER_BY:
+      nodesWalkList(pSelect->pProjectionList, walker, pContext);
+    case SQL_CLAUSE_SELECT:
     default:
       break;
   }

source/libs/nodes/src/nodesUtilFuncs.c

@@ -159,6 +159,8 @@ SNodeptr nodesMakeNode(ENodeType type) {
       return makeNode(type, sizeof(SExchangeLogicNode));
     case QUERY_NODE_LOGIC_PLAN_WINDOW:
       return makeNode(type, sizeof(SWindowLogicNode));
+    case QUERY_NODE_LOGIC_PLAN_SORT:
+      return makeNode(type, sizeof(SSortLogicNode));
     case QUERY_NODE_LOGIC_SUBPLAN:
       return makeNode(type, sizeof(SLogicSubplan));
     case QUERY_NODE_LOGIC_PLAN:
@@ -182,7 +184,7 @@ SNodeptr nodesMakeNode(ENodeType type) {
     case QUERY_NODE_PHYSICAL_PLAN_EXCHANGE:
       return makeNode(type, sizeof(SExchangePhysiNode));
     case QUERY_NODE_PHYSICAL_PLAN_SORT:
-      return makeNode(type, sizeof(SNode));
+      return makeNode(type, sizeof(SSortPhysiNode));
     case QUERY_NODE_PHYSICAL_PLAN_INTERVAL:
       return makeNode(type, sizeof(SIntervalPhysiNode));
     case QUERY_NODE_PHYSICAL_PLAN_SESSION_WINDOW:
@@ -555,7 +557,7 @@ static EDealRes collectColumns(SNode* pNode, void* pContext) {
   if (QUERY_NODE_COLUMN == nodeType(pNode)) {
     SColumnNode* pCol = (SColumnNode*)pNode;
     int32_t colId = pCol->colId;
-    if (0 == strcmp(pCxt->pTableAlias, pCol->tableAlias)) {
+    if (NULL == pCxt->pTableAlias || 0 == strcmp(pCxt->pTableAlias, pCol->tableAlias)) {
       return doCollect(pCxt, colId, pNode);
     }
   }

source/libs/parser/inc/parInsertData.h

@@ -41,26 +41,26 @@ typedef struct SBoundColumn {
 } SBoundColumn;
 
 typedef struct {
-  uint16_t schemaColIdx;
+  col_id_t schemaColIdx;
-  uint16_t boundIdx;
+  col_id_t boundIdx;
-  uint16_t finalIdx;
+  col_id_t finalIdx;
 } SBoundIdxInfo;
 
 typedef struct SParsedDataColInfo {
-  int16_t        numOfCols;
+  col_id_t       numOfCols;
-  int16_t        numOfBound;
+  col_id_t       numOfBound;
   uint16_t       flen;        // TODO: get from STSchema
   uint16_t       allNullLen;  // TODO: get from STSchema(base on SDataRow)
   uint16_t       extendedVarLen;
   uint16_t       boundNullLen;    // bound column len with all NULL value(without VarDataOffsetT/SColIdx part)
-  int32_t *      boundedColumns;  // bound column idx according to schema
+  col_id_t      *boundColumns;    // bound column idx according to schema
-  SBoundColumn * cols;
+  SBoundColumn  *cols;
   SBoundIdxInfo *colIdxInfo;
   int8_t         orderStatus;  // bound columns
 } SParsedDataColInfo;
 
 typedef struct {
-  uint8_t memRowType;  // default is 0, that is SDataRow
+  uint8_t rowType;  // default is 0, that is SDataRow
   int32_t rowSize;
 } SMemRowBuilder;
 
@@ -92,11 +92,11 @@ static FORCE_INLINE int32_t getExtendedRowSize(STableDataBlocks *pBlock) {
          (int32_t)TD_BITMAP_BYTES(pTableInfo->numOfColumns - 1);
 }
 
-static FORCE_INLINE void getMemRowAppendInfo(SSchema *pSchema, uint8_t rowType, SParsedDataColInfo *spd,
+static FORCE_INLINE void getSTSRowAppendInfo(SSchema *pSchema, uint8_t rowType, SParsedDataColInfo *spd, col_id_t idx,
-                                                int32_t idx, int32_t *toffset, int32_t *colIdx) {
+                                             int32_t *toffset, col_id_t *colIdx) {
-  int32_t schemaIdx = 0;
+  col_id_t schemaIdx = 0;
   if (IS_DATA_COL_ORDERED(spd)) {
-    schemaIdx = spd->boundedColumns[idx] - PRIMARYKEY_TIMESTAMP_COL_ID;
+    schemaIdx = spd->boundColumns[idx] - PRIMARYKEY_TIMESTAMP_COL_ID;
     if (TD_IS_TP_ROW_T(rowType)) {
       *toffset = (spd->cols + schemaIdx)->toffset;  // the offset of firstPart
       *colIdx = schemaIdx;
@@ -132,7 +132,7 @@ static FORCE_INLINE int32_t setBlockInfo(SSubmitBlk *pBlocks, STableDataBlocks*
 
 int32_t schemaIdxCompar(const void *lhs, const void *rhs);
 int32_t boundIdxCompar(const void *lhs, const void *rhs);
-void setBoundColumnInfo(SParsedDataColInfo* pColList, SSchema* pSchema, int32_t numOfCols);
+void    setBoundColumnInfo(SParsedDataColInfo *pColList, SSchema *pSchema, col_id_t numOfCols);
 void destroyBoundColumnInfo(SParsedDataColInfo* pColList);
 void destroyBlockArrayList(SArray* pDataBlockList);
 void destroyBlockHashmap(SHashObj* pDataBlockHash);

source/libs/parser/inc/parUtil.h

@@ -30,6 +30,8 @@ extern "C" {
 #define parserDebug(param, ...)  qDebug("PARSER: " param, __VA_ARGS__)
 #define parserTrace(param, ...)  qTrace("PARSER: " param, __VA_ARGS__)
 
+#define PK_TS_COL_INTERNAL_NAME "_rowts"
+
 typedef struct SMsgBuf {
   int32_t len;
   char   *buf;

source/libs/parser/src/parAstCreater.c

@@ -645,6 +645,11 @@ SNode* createTempTableNode(SAstCreateContext* pCxt, SNode* pSubquery, const STok
   tempTable->pSubquery = pSubquery;
   if (NULL != pTableAlias && TK_NK_NIL != pTableAlias->type) {
     strncpy(tempTable->table.tableAlias, pTableAlias->z, pTableAlias->n);
+  } else {
+    sprintf(tempTable->table.tableAlias, "%p", tempTable);
+  }
+  if (QUERY_NODE_SELECT_STMT == nodeType(pSubquery)) {
+    strcpy(((SSelectStmt*)pSubquery)->stmtName, tempTable->table.tableAlias);
   }
   return (SNode*)tempTable;
 }
@@ -697,6 +702,13 @@ SNode* createStateWindowNode(SAstCreateContext* pCxt, SNode* pCol) {
 SNode* createIntervalWindowNode(SAstCreateContext* pCxt, SNode* pInterval, SNode* pOffset, SNode* pSliding, SNode* pFill) {
   SIntervalWindowNode* interval = (SIntervalWindowNode*)nodesMakeNode(QUERY_NODE_INTERVAL_WINDOW);
   CHECK_OUT_OF_MEM(interval);
+  interval->pCol = nodesMakeNode(QUERY_NODE_COLUMN);
+  if (NULL == interval->pCol) {
+    nodesDestroyNode(interval);
+    CHECK_OUT_OF_MEM(interval->pCol);
+  }
+  ((SColumnNode*)interval->pCol)->colId = PRIMARYKEY_TIMESTAMP_COL_ID;
+  strcpy(((SColumnNode*)interval->pCol)->colName, PK_TS_COL_INTERNAL_NAME);
   interval->pInterval = pInterval;
   interval->pOffset = pOffset;
   interval->pSliding = pSliding;
@@ -792,6 +804,7 @@ SNode* createSelectStmt(SAstCreateContext* pCxt, bool isDistinct, SNodeList* pPr
   select->isDistinct = isDistinct;
   select->pProjectionList = pProjectionList;
   select->pFromTable = pTable;
+  sprintf(select->stmtName, "%p", select);
   return (SNode*)select;
 }
 

source/libs/parser/src/parInsert.c

@@ -600,9 +600,9 @@ static int32_t parseValueToken(char** end, SToken* pToken, SSchema* pSchema, int
 
 typedef struct SMemParam {
   SRowBuilder* rb;
-  SSchema* schema;
+  SSchema*     schema;
-  int32_t toffset;
+  int32_t      toffset;
-  int32_t      colIdx;
+  col_id_t     colIdx;
 } SMemParam;
 
 static FORCE_INLINE int32_t MemRowAppend(const void* value, int32_t len, void* param) {
@@ -623,9 +623,11 @@ static FORCE_INLINE int32_t MemRowAppend(const void* value, int32_t len, void* p
     tdAppendColValToRow(rb, pa->schema->colId, pa->schema->type, TD_VTYPE_NORM, rowEnd, false, pa->toffset, pa->colIdx);
   } else {
     if (value == NULL) {  // it is a null data
-      tdAppendColValToRow(rb, pa->schema->colId, pa->schema->type, TD_VTYPE_NULL, value, false, pa->toffset, pa->colIdx);
+      tdAppendColValToRow(rb, pa->schema->colId, pa->schema->type, TD_VTYPE_NULL, value, false, pa->toffset,
+                          pa->colIdx);
     } else {
-      tdAppendColValToRow(rb, pa->schema->colId, pa->schema->type, TD_VTYPE_NORM, value, false, pa->toffset, pa->colIdx);
+      tdAppendColValToRow(rb, pa->schema->colId, pa->schema->type, TD_VTYPE_NORM, value, false, pa->toffset,
+                          pa->colIdx);
     }
   }
   return TSDB_CODE_SUCCESS;
@@ -633,18 +635,18 @@ static FORCE_INLINE int32_t MemRowAppend(const void* value, int32_t len, void* p
 
 // pSql -> tag1_name, ...)
 static int32_t parseBoundColumns(SInsertParseContext* pCxt, SParsedDataColInfo* pColList, SSchema* pSchema) {
-  int32_t nCols = pColList->numOfCols;
+  col_id_t nCols = pColList->numOfCols;
 
   pColList->numOfBound = 0; 
   pColList->boundNullLen = 0;
-  memset(pColList->boundedColumns, 0, sizeof(int32_t) * nCols);
+  memset(pColList->boundColumns, 0, sizeof(col_id_t) * nCols);
-  for (int32_t i = 0; i < nCols; ++i) {
+  for (col_id_t i = 0; i < nCols; ++i) {
     pColList->cols[i].valStat = VAL_STAT_NONE;
   }
 
   SToken sToken;
   bool    isOrdered = true;
-  int32_t lastColIdx = -1;  // last column found
+  col_id_t lastColIdx = -1;  // last column found
   while (1) {
     NEXT_TOKEN(pCxt->pSql, sToken);
 
@@ -652,8 +654,8 @@ static int32_t parseBoundColumns(SInsertParseContext* pCxt, SParsedDataColInfo*
       break;
     }
 
-    int32_t t = lastColIdx + 1;
+    col_id_t t = lastColIdx + 1;
-    int32_t index = findCol(&sToken, t, nCols, pSchema);
+    col_id_t index = findCol(&sToken, t, nCols, pSchema);
     if (index < 0 && t > 0) {
       index = findCol(&sToken, 0, t, pSchema);
       isOrdered = false;
@@ -666,7 +668,7 @@ static int32_t parseBoundColumns(SInsertParseContext* pCxt, SParsedDataColInfo*
     }
     lastColIdx = index;
     pColList->cols[index].valStat = VAL_STAT_HAS;
-    pColList->boundedColumns[pColList->numOfBound] = index + PRIMARYKEY_TIMESTAMP_COL_ID;
+    pColList->boundColumns[pColList->numOfBound] = index + PRIMARYKEY_TIMESTAMP_COL_ID;
     ++pColList->numOfBound;
     switch (pSchema[t].type) {
       case TSDB_DATA_TYPE_BINARY:
@@ -689,18 +691,19 @@ static int32_t parseBoundColumns(SInsertParseContext* pCxt, SParsedDataColInfo*
       return TSDB_CODE_TSC_OUT_OF_MEMORY;
     }
     SBoundIdxInfo* pColIdx = pColList->colIdxInfo;
-    for (uint16_t i = 0; i < pColList->numOfBound; ++i) {
+    for (col_id_t i = 0; i < pColList->numOfBound; ++i) {
-      pColIdx[i].schemaColIdx = (uint16_t)pColList->boundedColumns[i];
+      pColIdx[i].schemaColIdx = pColList->boundColumns[i];
       pColIdx[i].boundIdx = i;
     }
     qsort(pColIdx, pColList->numOfBound, sizeof(SBoundIdxInfo), schemaIdxCompar);
-    for (uint16_t i = 0; i < pColList->numOfBound; ++i) {
+    for (col_id_t i = 0; i < pColList->numOfBound; ++i) {
       pColIdx[i].finalIdx = i;
     }
     qsort(pColIdx, pColList->numOfBound, sizeof(SBoundIdxInfo), boundIdxCompar);
   }
 
-  memset(&pColList->boundedColumns[pColList->numOfBound], 0, sizeof(int32_t) * (pColList->numOfCols - pColList->numOfBound));
+  memset(&pColList->boundColumns[pColList->numOfBound], 0,
+         sizeof(col_id_t) * (pColList->numOfCols - pColList->numOfBound));
 
   return TSDB_CODE_SUCCESS;
 }
@@ -714,8 +717,8 @@ typedef struct SKvParam {
 static int32_t KvRowAppend(const void *value, int32_t len, void *param) {
   SKvParam* pa = (SKvParam*) param;
 
-  int32_t type  = pa->schema->type;
+  int8_t  type = pa->schema->type;
-  int32_t colId = pa->schema->colId;
+  int16_t colId = pa->schema->colId;
 
   if (TSDB_DATA_TYPE_BINARY == type) {
     STR_WITH_SIZE_TO_VARSTR(pa->buf, value, len);
@@ -747,7 +750,7 @@ static int32_t parseTagsClause(SInsertParseContext* pCxt, SSchema* pTagsSchema,
   char tmpTokenBuf[TSDB_MAX_BYTES_PER_ROW] = {0};  // used for deleting Escape character: \\, \', \"
   for (int i = 0; i < pCxt->tags.numOfBound; ++i) {
     NEXT_TOKEN_WITH_PREV(pCxt->pSql, sToken);
-    SSchema* pSchema = &pTagsSchema[pCxt->tags.boundedColumns[i]];
+    SSchema* pSchema = &pTagsSchema[pCxt->tags.boundColumns[i]];
     param.schema = pSchema;
     CHECK_CODE(parseValueToken(&pCxt->pSql, &sToken, pSchema, precision, tmpTokenBuf, KvRowAppend, &param, &pCxt->msg));
   }
@@ -813,9 +816,9 @@ static int parseOneRow(SInsertParseContext* pCxt, STableDataBlocks* pDataBlocks,
   // 1. set the parsed value from sql string
   for (int i = 0; i < spd->numOfBound; ++i) {
     NEXT_TOKEN_WITH_PREV(pCxt->pSql, sToken);
-    SSchema *pSchema = &schema[spd->boundedColumns[i] - 1];
+    SSchema* pSchema = &schema[spd->boundColumns[i] - 1];
     param.schema = pSchema;
-    getMemRowAppendInfo(schema, pBuilder->rowType, spd, i, &param.toffset, &param.colIdx);
+    getSTSRowAppendInfo(schema, pBuilder->rowType, spd, i, &param.toffset, &param.colIdx);
     CHECK_CODE(parseValueToken(&pCxt->pSql, &sToken, pSchema, timePrec, tmpTokenBuf, MemRowAppend, &param, &pCxt->msg));
 
     if (PRIMARYKEY_TIMESTAMP_COL_ID == pSchema->colId) {

source/libs/parser/src/parInsertData.c

@@ -43,11 +43,11 @@ static int32_t rowDataCompar(const void *lhs, const void *rhs) {
   }
 }
 
-void setBoundColumnInfo(SParsedDataColInfo* pColList, SSchema* pSchema, int32_t numOfCols) {
+void setBoundColumnInfo(SParsedDataColInfo* pColList, SSchema* pSchema, col_id_t numOfCols) {
   pColList->numOfCols = numOfCols;
   pColList->numOfBound = numOfCols;
   pColList->orderStatus = ORDER_STATUS_ORDERED;  // default is ORDERED for non-bound mode
-  pColList->boundedColumns = taosMemoryCalloc(pColList->numOfCols, sizeof(int32_t));
+  pColList->boundColumns = taosMemoryCalloc(pColList->numOfCols, sizeof(col_id_t));
   pColList->cols = taosMemoryCalloc(pColList->numOfCols, sizeof(SBoundColumn));
   pColList->colIdxInfo = NULL;
   pColList->flen = 0;
@@ -73,7 +73,7 @@ void setBoundColumnInfo(SParsedDataColInfo* pColList, SSchema* pSchema, int32_t
       default:
         break;
     }
-    pColList->boundedColumns[i] = pSchema[i].colId;
+    pColList->boundColumns[i] = pSchema[i].colId;
   }
   pColList->allNullLen += pColList->flen;
   pColList->boundNullLen = pColList->allNullLen;  // default set allNullLen
@@ -103,7 +103,7 @@ int32_t boundIdxCompar(const void *lhs, const void *rhs) {
 }
 
 void destroyBoundColumnInfo(SParsedDataColInfo* pColList) {
-  taosMemoryFreeClear(pColList->boundedColumns);
+  taosMemoryFreeClear(pColList->boundColumns);
   taosMemoryFreeClear(pColList->cols);
   taosMemoryFreeClear(pColList->colIdxInfo);
 }

source/libs/parser/src/parTranslater.c

@@ -271,6 +271,10 @@ static bool findAndSetColumn(SColumnNode* pCol, const STableNode* pTable) {
   bool found = false;
   if (QUERY_NODE_REAL_TABLE == nodeType(pTable)) {
     const STableMeta* pMeta = ((SRealTableNode*)pTable)->pMeta;
+    if (PRIMARYKEY_TIMESTAMP_COL_ID == pCol->colId && 0 == strcmp(pCol->colName, PK_TS_COL_INTERNAL_NAME)) {
+      setColumnInfoBySchema((SRealTableNode*)pTable, pMeta->schema, false, pCol);
+      return true;
+    }
     int32_t nums = pMeta->tableInfo.numOfTags + pMeta->tableInfo.numOfColumns;
     for (int32_t i = 0; i < nums; ++i) {
       if (0 == strcmp(pCol->colName, pMeta->schema[i].name)) {
@@ -1448,6 +1452,7 @@ static int32_t getSmaIndexBuildAst(STranslateContext* pCxt, SCreateIndexStmt* pS
   if (NULL == pSelect) {
     return TSDB_CODE_OUT_OF_MEMORY;
   }
+  sprintf(pSelect->stmtName, "%p", pSelect);
 
   SRealTableNode* pTable = nodesMakeNode(QUERY_NODE_REAL_TABLE);
   if (NULL == pTable) {
@@ -1463,6 +1468,10 @@ static int32_t getSmaIndexBuildAst(STranslateContext* pCxt, SCreateIndexStmt* pS
     nodesDestroyNode(pSelect);
     return TSDB_CODE_OUT_OF_MEMORY;
   }
+  SNode* pProject = NULL;
+  FOREACH(pProject, pSelect->pProjectionList) {
+    sprintf(((SExprNode*)pProject)->aliasName, "#sma_%p", pProject);
+  }
 
   SIntervalWindowNode* pInterval = nodesMakeNode(QUERY_NODE_INTERVAL_WINDOW);
   if (NULL == pInterval) {
@@ -1470,14 +1479,18 @@ static int32_t getSmaIndexBuildAst(STranslateContext* pCxt, SCreateIndexStmt* pS
     return TSDB_CODE_OUT_OF_MEMORY;
   }
   pSelect->pWindow = (SNode*)pInterval;
+  pInterval->pCol = nodesMakeNode(QUERY_NODE_COLUMN);
   pInterval->pInterval = nodesCloneNode(pStmt->pOptions->pInterval);
   pInterval->pOffset = nodesCloneNode(pStmt->pOptions->pOffset);
   pInterval->pSliding = nodesCloneNode(pStmt->pOptions->pSliding);
-  if (NULL == pInterval->pInterval || (NULL != pStmt->pOptions->pOffset && NULL == pInterval->pOffset) ||
+  if (NULL == pInterval->pCol || NULL == pInterval->pInterval || 
+      (NULL != pStmt->pOptions->pOffset && NULL == pInterval->pOffset) ||
       (NULL != pStmt->pOptions->pSliding && NULL == pInterval->pSliding)) {
     nodesDestroyNode(pSelect);
     return TSDB_CODE_OUT_OF_MEMORY;
   }
+  ((SColumnNode*)pInterval->pCol)->colId = PRIMARYKEY_TIMESTAMP_COL_ID;
+  strcpy(((SColumnNode*)pInterval->pCol)->colName, PK_TS_COL_INTERNAL_NAME);
 
   int32_t code = translateQuery(pCxt, (SNode*)pSelect);
   if (TSDB_CODE_SUCCESS == code) {
@@ -1787,7 +1800,7 @@ static int32_t translateSubquery(STranslateContext* pCxt, SNode* pNode) {
 }
 
 int32_t qExtractResultSchema(const SNode* pRoot, int32_t* numOfCols, SSchema** pSchema) {
-  if (QUERY_NODE_SELECT_STMT == nodeType(pRoot)) {
+  if (NULL != pRoot && QUERY_NODE_SELECT_STMT == nodeType(pRoot)) {
     SSelectStmt* pSelect = (SSelectStmt*) pRoot;
     *numOfCols = LIST_LENGTH(pSelect->pProjectionList);
     *pSchema = taosMemoryCalloc((*numOfCols), sizeof(SSchema));
@@ -1865,6 +1878,7 @@ static int32_t createSelectStmtForShow(ENodeType showType, SSelectStmt** pStmt)
   if (NULL == pSelect) {
     return TSDB_CODE_OUT_OF_MEMORY;
   }
+  sprintf(pSelect->stmtName, "%p", pSelect);
 
   SRealTableNode* pTable = nodesMakeNode(QUERY_NODE_REAL_TABLE);
   if (NULL == pTable) {
@@ -1873,6 +1887,7 @@ static int32_t createSelectStmtForShow(ENodeType showType, SSelectStmt** pStmt)
   }
   strcpy(pTable->table.dbName, TSDB_INFORMATION_SCHEMA_DB);
   strcpy(pTable->table.tableName, getSysTableName(showType));
+  strcpy(pTable->table.tableAlias, pTable->table.tableName);
   pSelect->pFromTable = (SNode*)pTable;
 
   *pStmt = pSelect;
@@ -1975,7 +1990,7 @@ typedef struct SVgroupTablesBatch {
   char               dbName[TSDB_DB_NAME_LEN];
 } SVgroupTablesBatch;
 
-static void toSchema(const SColumnDefNode* pCol, int32_t colId, SSchema* pSchema) {
+static void toSchema(const SColumnDefNode* pCol, col_id_t colId, SSchema* pSchema) {
   pSchema->colId = colId;
   pSchema->type = pCol->dataType.type;
   pSchema->bytes = pCol->dataType.bytes;
@@ -2385,13 +2400,14 @@ static int32_t rewriteQuery(STranslateContext* pCxt, SQuery* pQuery) {
 }
 
 static int32_t setQuery(STranslateContext* pCxt, SQuery* pQuery) {
-  int32_t code = TSDB_CODE_SUCCESS;
   switch (nodeType(pQuery->pRoot)) {
     case QUERY_NODE_SELECT_STMT:
       pQuery->haveResultSet = true;
       pQuery->directRpc = false;
       pQuery->msgType = TDMT_VND_QUERY;
-      code = qExtractResultSchema(pQuery->pRoot, &pQuery->numOfResCols, &pQuery->pResSchema);
+      if (TSDB_CODE_SUCCESS != qExtractResultSchema(pQuery->pRoot, &pQuery->numOfResCols, &pQuery->pResSchema)) {
+        return TSDB_CODE_OUT_OF_MEMORY;
+      }
       break;
     case QUERY_NODE_VNODE_MODIF_STMT:
       pQuery->haveResultSet = false;
@@ -2431,7 +2447,7 @@ static int32_t setQuery(STranslateContext* pCxt, SQuery* pQuery) {
     }
   }
 
-  return code;
+  return TSDB_CODE_SUCCESS;
 }
 
 int32_t doTranslate(SParseContext* pParseCxt, SQuery* pQuery) {

source/libs/parser/test/mockCatalogService.cpp

@@ -87,7 +87,7 @@ private:
     return meta_;
   }
 
-  int32_t colId_;
+  col_id_t                       colId_;
   int32_t rowsize_;
   std::shared_ptr<MockTableMeta> meta_;
 };

source/libs/planner/inc/planInt.h

@@ -22,32 +22,6 @@ extern "C" {
 
 #include "planner.h"
 
-#define CHECK_ALLOC(p, res) \
-  do { \
-    if (NULL == (p)) { \
-      pCxt->errCode = TSDB_CODE_OUT_OF_MEMORY; \
-      return (res); \
-    } \
-  } while (0)
-
-#define CHECK_CODE(exec, res) \
-  do { \
-    int32_t code = (exec); \
-    if (TSDB_CODE_SUCCESS != code) { \
-      pCxt->errCode = code; \
-      return (res); \
-    } \
-  } while (0)
-
-#define CHECK_CODE_EXT(exec) \
-  do { \
-    int32_t code = (exec); \
-    if (TSDB_CODE_SUCCESS != code) { \
-      pCxt->errCode = code; \
-      return code; \
-    } \
-  } while (0)
-
 #define planFatal(param, ...)  qFatal("PLAN: " param, __VA_ARGS__)
 #define planError(param, ...)  qError("PLAN: " param, __VA_ARGS__)
 #define planWarn(param, ...)   qWarn("PLAN: " param, __VA_ARGS__)

source/libs/planner/src/planLogicCreater.c

@@ -45,7 +45,9 @@ static EDealRes doRewriteExpr(SNode** pNode, void* pContext) {
         }
         if (nodesEqualNode(pExpr, *pNode)) {
           SColumnNode* pCol = (SColumnNode*)nodesMakeNode(QUERY_NODE_COLUMN);
-          CHECK_ALLOC(pCol, DEAL_RES_ERROR);
+          if (NULL == pCol) {
+            return DEAL_RES_ERROR;
+          }
           SExprNode* pToBeRewrittenExpr = (SExprNode*)(*pNode);
           pCol->node.resType = pToBeRewrittenExpr->resType;
           strcpy(pCol->node.aliasName, pToBeRewrittenExpr->aliasName);
@@ -65,17 +67,12 @@ static EDealRes doRewriteExpr(SNode** pNode, void* pContext) {
   return DEAL_RES_CONTINUE;
 }
 
-typedef struct SNameExprCxt {
-  int32_t rewriteId;
-} SNameExprCxt;
-
 static EDealRes doNameExpr(SNode* pNode, void* pContext) {
   switch (nodeType(pNode)) {
     case QUERY_NODE_OPERATOR:
     case QUERY_NODE_LOGIC_CONDITION:
     case QUERY_NODE_FUNCTION: {
-      SNameExprCxt* pCxt = (SNameExprCxt*)pContext;
+      sprintf(((SExprNode*)pNode)->aliasName, "#expr_%p", pNode);
-      sprintf(((SExprNode*)pNode)->aliasName, "#expr_%d", pCxt->rewriteId++);
       return DEAL_RES_IGNORE_CHILD;
     }
     default:
@@ -86,9 +83,7 @@ static EDealRes doNameExpr(SNode* pNode, void* pContext) {
 }
 
 static int32_t rewriteExpr(SNodeList* pExprs, SSelectStmt* pSelect, ESqlClause clause) {
-  static int32_t rewriteId = 1;
+  nodesWalkList(pExprs, doNameExpr, NULL);
-  SNameExprCxt nameCxt = { .rewriteId = rewriteId };
-  nodesWalkList(pExprs, doNameExpr, &nameCxt);
   SRewriteExprCxt cxt = { .errCode = TSDB_CODE_SUCCESS, .pExprs = pExprs };
   nodesRewriteSelectStmt(pSelect, clause, doRewriteExpr, &cxt);
   return cxt.errCode;
@@ -291,13 +286,14 @@ static int32_t createLogicNodeByTable(SLogicPlanContext* pCxt, SSelectStmt* pSel
   return code;
 }
 
-static SColumnNode* createColumnByExpr(SExprNode* pExpr) {
+static SColumnNode* createColumnByExpr(const char* pStmtName, SExprNode* pExpr) {
   SColumnNode* pCol = nodesMakeNode(QUERY_NODE_COLUMN);
   if (NULL == pCol) {
     return NULL;
   }
   pCol->node.resType = pExpr->resType;
   strcpy(pCol->colName, pExpr->aliasName);
+  strcpy(pCol->tableAlias, pStmtName);
   return pCol;
 }
 
@@ -311,20 +307,22 @@ static EDealRes doCreateColumn(SNode* pNode, void* pContext) {
   switch (nodeType(pNode)) {
     case QUERY_NODE_COLUMN: {
       SNode* pCol = nodesCloneNode(pNode);
-      CHECK_ALLOC(pCol, DEAL_RES_ERROR);
+      if (NULL == pCol) {
-      CHECK_CODE(nodesListAppend(pCxt->pList, pCol), DEAL_RES_ERROR);
+        return DEAL_RES_ERROR;
-      return DEAL_RES_IGNORE_CHILD;
+      }
+      return (TSDB_CODE_SUCCESS == nodesListAppend(pCxt->pList, pCol) ? DEAL_RES_IGNORE_CHILD : DEAL_RES_ERROR);
     }
     case QUERY_NODE_OPERATOR:
     case QUERY_NODE_LOGIC_CONDITION:
     case QUERY_NODE_FUNCTION: {
       SExprNode* pExpr = (SExprNode*)pNode;
       SColumnNode* pCol = (SColumnNode*)nodesMakeNode(QUERY_NODE_COLUMN);
-      CHECK_ALLOC(pCol, DEAL_RES_ERROR);
+      if (NULL == pCol) {
+        return DEAL_RES_ERROR;
+      }
       pCol->node.resType = pExpr->resType;
       strcpy(pCol->colName, pExpr->aliasName);
-      CHECK_CODE(nodesListAppend(pCxt->pList, (SNode*)pCol), DEAL_RES_ERROR);
+      return (TSDB_CODE_SUCCESS == nodesListAppend(pCxt->pList, pCol) ? DEAL_RES_IGNORE_CHILD : DEAL_RES_ERROR);
-      return DEAL_RES_IGNORE_CHILD;
     }
     default:
       break;
@@ -457,6 +455,12 @@ static int32_t createWindowLogicNodeByInterval(SLogicPlanContext* pCxt, SInterva
   pWindow->sliding = (NULL != pInterval->pSliding ? ((SValueNode*)pInterval->pSliding)->datum.i : pWindow->interval);
   pWindow->slidingUnit = (NULL != pInterval->pSliding ? ((SValueNode*)pInterval->pSliding)->unit : pWindow->intervalUnit);
 
+  pWindow->pTspk = nodesCloneNode(pInterval->pCol);
+  if (NULL == pWindow->pTspk) {
+    nodesDestroyNode(pWindow);
+    return TSDB_CODE_OUT_OF_MEMORY;
+  }
+
   if (NULL != pInterval->pFill) {
     pWindow->pFill = nodesCloneNode(pInterval->pFill);
     if (NULL == pWindow->pFill) {
@@ -485,7 +489,42 @@ static int32_t createWindowLogicNode(SLogicPlanContext* pCxt, SSelectStmt* pSele
   return TSDB_CODE_FAILED;
 }
 
-static int32_t createColumnByProjections(SLogicPlanContext* pCxt, SNodeList* pExprs, SNodeList** pCols) {
+static int32_t createSortLogicNode(SLogicPlanContext* pCxt, SSelectStmt* pSelect, SLogicNode** pLogicNode) {
+  if (NULL == pSelect->pOrderByList) {
+    return TSDB_CODE_SUCCESS;
+  }
+
+  SSortLogicNode* pSort = nodesMakeNode(QUERY_NODE_LOGIC_PLAN_SORT);
+  if (NULL == pSort) {
+    return TSDB_CODE_OUT_OF_MEMORY;
+  }
+
+  SNodeList* pCols = NULL;
+  int32_t code = nodesCollectColumns(pSelect, SQL_CLAUSE_ORDER_BY, NULL, &pCols);
+  if (TSDB_CODE_SUCCESS == code && NULL != pCols) {
+    pSort->node.pTargets = nodesCloneList(pCols);
+    if (NULL == pSort->node.pTargets) {
+      code = TSDB_CODE_OUT_OF_MEMORY;
+    }
+  }
+
+  if (TSDB_CODE_SUCCESS == code) {
+    pSort->pSortKeys = nodesCloneList(pSelect->pOrderByList);
+    if (NULL == pSort->pSortKeys) {
+      code = TSDB_CODE_OUT_OF_MEMORY;
+    }
+  }
+
+  if (TSDB_CODE_SUCCESS == code) {
+    *pLogicNode = (SLogicNode*)pSort;
+  } else {
+    nodesDestroyNode(pSort);
+  }
+
+  return code;
+}
+
+static int32_t createColumnByProjections(SLogicPlanContext* pCxt, const char* pStmtName, SNodeList* pExprs, SNodeList** pCols) {
   SNodeList* pList = nodesMakeList();
   if (NULL == pList) {
     return TSDB_CODE_OUT_OF_MEMORY;
@@ -493,7 +532,7 @@ static int32_t createColumnByProjections(SLogicPlanContext* pCxt, SNodeList* pEx
 
   SNode* pNode;
   FOREACH(pNode, pExprs) {    
-    if (TSDB_CODE_SUCCESS != nodesListAppend(pList, createColumnByExpr((SExprNode*)pNode))) {
+    if (TSDB_CODE_SUCCESS != nodesListAppend(pList, createColumnByExpr(pStmtName, (SExprNode*)pNode))) {
       nodesDestroyList(pList);
       return TSDB_CODE_OUT_OF_MEMORY;
     }
@@ -515,9 +554,10 @@ static int32_t createProjectLogicNode(SLogicPlanContext* pCxt, SSelectStmt* pSel
   if (NULL == pProject->pProjections) {
     code = TSDB_CODE_OUT_OF_MEMORY;
   }
+  strcpy(pProject->stmtName, pSelect->stmtName);
 
   if (TSDB_CODE_SUCCESS == code) {
-    code = createColumnByProjections(pCxt,pSelect->pProjectionList, &pProject->node.pTargets);
+    code = createColumnByProjections(pCxt, pSelect->stmtName, pSelect->pProjectionList, &pProject->node.pTargets);
   }
 
   if (TSDB_CODE_SUCCESS == code) {
@@ -538,6 +578,9 @@ static int32_t createSelectLogicNode(SLogicPlanContext* pCxt, SSelectStmt* pSele
   if (TSDB_CODE_SUCCESS == code) {
     code = createChildLogicNode(pCxt, pSelect, createAggLogicNode, &pRoot);
   }
+  if (TSDB_CODE_SUCCESS == code) {
+    code = createChildLogicNode(pCxt, pSelect, createSortLogicNode, &pRoot);
+  }
   if (TSDB_CODE_SUCCESS == code) {
     code = createChildLogicNode(pCxt, pSelect, createProjectLogicNode, &pRoot);
   }

source/libs/planner/src/planPhysiCreater.c

@@ -17,9 +17,14 @@
 
 #include "functionMgt.h"
 
+typedef struct SSlotIdInfo {
+  int16_t slotId;
+  bool set;
+} SSlotIdInfo;
+
 typedef struct SSlotIndex {
   int16_t dataBlockId;
-  int16_t slotId;
+  SArray* pSlotIdsInfo; // duplicate name slot
 } SSlotIndex;
 
 typedef struct SPhysiPlanContext {
@@ -30,74 +35,197 @@ typedef struct SPhysiPlanContext {
   SArray* pExecNodeList;
 } SPhysiPlanContext;
 
-static int32_t getSlotKey(SNode* pNode, char* pKey) {
+static int32_t getSlotKey(SNode* pNode, const char* pStmtName, char* pKey) {
+  if (QUERY_NODE_ORDER_BY_EXPR == nodeType(pNode)) {
+    return getSlotKey(((SOrderByExprNode*)pNode)->pExpr, pStmtName, pKey);
+  }
+
   if (QUERY_NODE_COLUMN == nodeType(pNode)) {
     SColumnNode* pCol = (SColumnNode*)pNode;
+    if (NULL != pStmtName) {
+      return sprintf(pKey, "%s.%s", pStmtName, pCol->node.aliasName);
+    }
     if ('\0' == pCol->tableAlias[0]) {
       return sprintf(pKey, "%s", pCol->colName);
     }
     return sprintf(pKey, "%s.%s", pCol->tableAlias, pCol->colName);
   }
+
+  if (NULL != pStmtName) {
+    return sprintf(pKey, "%s.%s", pStmtName, ((SExprNode*)pNode)->aliasName);
+  }
   return sprintf(pKey, "%s", ((SExprNode*)pNode)->aliasName);
 }
 
-static SNode* createSlotDesc(SPhysiPlanContext* pCxt, const SNode* pNode, int16_t slotId) {
+static SNode* createSlotDesc(SPhysiPlanContext* pCxt, const SNode* pNode, int16_t slotId, bool output) {
   SSlotDescNode* pSlot = (SSlotDescNode*)nodesMakeNode(QUERY_NODE_SLOT_DESC);
-  CHECK_ALLOC(pSlot, NULL);
+  if (NULL == pSlot) {
+    return NULL;
+  }
   pSlot->slotId = slotId;
   pSlot->dataType = ((SExprNode*)pNode)->resType;
   pSlot->reserve = false;
-  pSlot->output = true;
+  pSlot->output = output;
   return (SNode*)pSlot;
 }
 
-static SNode* createTarget(SNode* pNode, int16_t dataBlockId, int16_t slotId) {
+static int32_t createTarget(SNode* pNode, int16_t dataBlockId, int16_t slotId, SNode** pOutput) {
   STargetNode* pTarget = (STargetNode*)nodesMakeNode(QUERY_NODE_TARGET);
   if (NULL == pTarget) {
-    return NULL;
+    return TSDB_CODE_OUT_OF_MEMORY;
   }
+
   pTarget->dataBlockId = dataBlockId;
   pTarget->slotId = slotId;
   pTarget->pExpr = pNode;
-  return (SNode*)pTarget;
+
+  *pOutput = (SNode*)pTarget;
+  return TSDB_CODE_SUCCESS;
 }
 
-static int32_t addDataBlockDesc(SPhysiPlanContext* pCxt, SNodeList* pList, SDataBlockDescNode* pDataBlockDesc) {
+static int32_t putSlotToHashImpl(int16_t dataBlockId, int16_t slotId, const char* pName, int32_t len, SHashObj* pHash) {
-  SHashObj* pHash = NULL;
+  SSlotIndex* pIndex = taosHashGet(pHash, pName, len);
-  if (NULL == pDataBlockDesc->pSlots) {
+  if (NULL != pIndex) {
-    pDataBlockDesc->pSlots = nodesMakeList();
+    SSlotIdInfo info = { .slotId = slotId, .set = false };
-    CHECK_ALLOC(pDataBlockDesc->pSlots, TSDB_CODE_OUT_OF_MEMORY);
+    taosArrayPush(pIndex->pSlotIdsInfo, &info);
-
+    return TSDB_CODE_SUCCESS;
-    pHash = taosHashInit(LIST_LENGTH(pList), taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY), true, HASH_NO_LOCK);
-    CHECK_ALLOC(pHash, TSDB_CODE_OUT_OF_MEMORY);
-    if (NULL == taosArrayInsert(pCxt->pLocationHelper, pDataBlockDesc->dataBlockId, &pHash)) {
-      taosHashCleanup(pHash);
-      return TSDB_CODE_OUT_OF_MEMORY;
-    }
-  } else {
-    pHash = taosArrayGetP(pCxt->pLocationHelper, pDataBlockDesc->dataBlockId);
   }
-  
-  SNode* pNode = NULL;
-  int16_t slotId = taosHashGetSize(pHash);
-  FOREACH(pNode, pList) {
-    CHECK_CODE_EXT(nodesListStrictAppend(pDataBlockDesc->pSlots, createSlotDesc(pCxt, pNode, slotId)));
 
-    SSlotIndex index = { .dataBlockId = pDataBlockDesc->dataBlockId, .slotId = slotId };
+  SSlotIndex index = { .dataBlockId = dataBlockId, .pSlotIdsInfo = taosArrayInit(TARRAY_MIN_SIZE, sizeof(SSlotIdInfo)) };
-    char name[TSDB_TABLE_NAME_LEN + TSDB_COL_NAME_LEN];
+  if (NULL == index.pSlotIdsInfo) {
-    int32_t len = getSlotKey(pNode, name);
+    return TSDB_CODE_OUT_OF_MEMORY;
-    CHECK_CODE(taosHashPut(pHash, name, len, &index, sizeof(SSlotIndex)), TSDB_CODE_OUT_OF_MEMORY);
-
-    SNode* pTarget = createTarget(pNode, pDataBlockDesc->dataBlockId, slotId);
-    CHECK_ALLOC(pTarget, TSDB_CODE_OUT_OF_MEMORY);
-    REPLACE_NODE(pTarget);
-
-    pDataBlockDesc->resultRowSize += ((SExprNode*)pNode)->resType.bytes;
-    ++slotId;
   }
+  SSlotIdInfo info = { .slotId = slotId, .set = false };
+  taosArrayPush(index.pSlotIdsInfo, &info);
+  return taosHashPut(pHash, pName, len, &index, sizeof(SSlotIndex));
+}
+
+static int32_t putSlotToHash(int16_t dataBlockId, int16_t slotId, SNode* pNode, SHashObj* pHash) {
+  char name[TSDB_TABLE_NAME_LEN + TSDB_COL_NAME_LEN];
+  int32_t len = getSlotKey(pNode, NULL, name);
+  return putSlotToHashImpl(dataBlockId, slotId, name, len, pHash);
+}
+
+static int32_t createDataBlockDescHash(SPhysiPlanContext* pCxt, int32_t capacity, int16_t dataBlockId, SHashObj** pDescHash) {
+  SHashObj* pHash = taosHashInit(capacity, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY), false, HASH_NO_LOCK);
+  if (NULL == pHash) {
+    return TSDB_CODE_OUT_OF_MEMORY;
+  }
+  if (NULL == taosArrayInsert(pCxt->pLocationHelper, dataBlockId, &pHash)) {
+    taosHashCleanup(pHash);
+    return TSDB_CODE_OUT_OF_MEMORY;
+  }
+
+  *pDescHash = pHash;
   return TSDB_CODE_SUCCESS;
 }
 
+static int32_t buildDataBlockSlots(SPhysiPlanContext* pCxt, SNodeList* pList, SDataBlockDescNode* pDataBlockDesc, SHashObj* pHash) {
+  pDataBlockDesc->pSlots = nodesMakeList();
+  if (NULL == pDataBlockDesc->pSlots) {
+    return TSDB_CODE_OUT_OF_MEMORY;
+  }
+
+  int32_t code = TSDB_CODE_SUCCESS;
+  int16_t slotId = 0;
+  SNode* pNode = NULL;
+  FOREACH(pNode, pList) {
+    code = nodesListStrictAppend(pDataBlockDesc->pSlots, createSlotDesc(pCxt, pNode, slotId, true));
+    if (TSDB_CODE_SUCCESS == code) {
+      code = putSlotToHash(pDataBlockDesc->dataBlockId, slotId, pNode, pHash);
+    }
+    if (TSDB_CODE_SUCCESS == code) {
+      pDataBlockDesc->resultRowSize += ((SExprNode*)pNode)->resType.bytes;
+      ++slotId;
+    } else {
+      break;
+    }
+  }
+  return code;
+}
+
+static int32_t createDataBlockDesc(SPhysiPlanContext* pCxt, SNodeList* pList, SDataBlockDescNode** pDataBlockDesc) {
+  SDataBlockDescNode* pDesc = nodesMakeNode(QUERY_NODE_DATABLOCK_DESC);
+  if (NULL == pDesc) {
+    return TSDB_CODE_OUT_OF_MEMORY;
+  }
+  pDesc->dataBlockId = pCxt->nextDataBlockId++;
+
+  SHashObj* pHash = NULL;
+  int32_t code = createDataBlockDescHash(pCxt, LIST_LENGTH(pList), pDesc->dataBlockId, &pHash);
+  if (TSDB_CODE_SUCCESS == code) {
+    code = buildDataBlockSlots(pCxt, pList, pDesc, pHash);
+  }
+
+  if (TSDB_CODE_SUCCESS == code) {
+    *pDataBlockDesc = pDesc;
+  } else {
+    nodesDestroyNode(pDesc);
+  }
+
+  return code;
+}
+
+static int16_t getUnsetSlotId(const SArray* pSlotIdsInfo) {
+  int32_t size = taosArrayGetSize(pSlotIdsInfo);
+  for (int32_t i = 0; i < size; ++i) {
+    SSlotIdInfo* pInfo = taosArrayGet(pSlotIdsInfo, i);
+    if (!pInfo->set) {
+      pInfo->set = true;
+      return pInfo->slotId;
+    }
+  }
+  return ((SSlotIdInfo*)taosArrayGet(pSlotIdsInfo, 0))->slotId;
+}
+
+static int32_t addDataBlockSlotsImpl(SPhysiPlanContext* pCxt, SNodeList* pList, SDataBlockDescNode* pDataBlockDesc, const char* pStmtName, bool output) {
+  int32_t code = TSDB_CODE_SUCCESS;
+  SHashObj* pHash = taosArrayGetP(pCxt->pLocationHelper, pDataBlockDesc->dataBlockId);
+  int16_t nextSlotId = taosHashGetSize(pHash), slotId = 0;
+  SNode* pNode = NULL;
+  FOREACH(pNode, pList) {
+    char name[TSDB_TABLE_NAME_LEN + TSDB_COL_NAME_LEN] = {0};
+    int32_t len = getSlotKey(pNode, pStmtName, name);
+    SSlotIndex* pIndex = taosHashGet(pHash, name, len);
+    if (NULL == pIndex) {
+      code = nodesListStrictAppend(pDataBlockDesc->pSlots, createSlotDesc(pCxt, pNode, nextSlotId, output));
+      if (TSDB_CODE_SUCCESS == code) {
+        code = putSlotToHashImpl(pDataBlockDesc->dataBlockId, nextSlotId, name, len, pHash);
+      }
+      pDataBlockDesc->resultRowSize += ((SExprNode*)pNode)->resType.bytes;
+      slotId = nextSlotId;
+      ++nextSlotId;
+    } else {
+      slotId = getUnsetSlotId(pIndex->pSlotIdsInfo);
+    }
+
+    if (TSDB_CODE_SUCCESS == code) {
+      SNode* pTarget = NULL;
+      code = createTarget(pNode, pDataBlockDesc->dataBlockId, slotId, &pTarget);
+      if (TSDB_CODE_SUCCESS == code) {
+        REPLACE_NODE(pTarget);
+      }
+    }
+  
+    if (TSDB_CODE_SUCCESS != code) {
+      break;
+    }
+  }
+  return code;
+}
+
+static int32_t addDataBlockSlots(SPhysiPlanContext* pCxt, SNodeList* pList, SDataBlockDescNode* pDataBlockDesc) {
+  return addDataBlockSlotsImpl(pCxt, pList, pDataBlockDesc, NULL, false);
+}
+
+static int32_t addDataBlockSlotsForProject(SPhysiPlanContext* pCxt, const char* pStmtName, SNodeList* pList, SDataBlockDescNode* pDataBlockDesc) {
+  return addDataBlockSlotsImpl(pCxt, pList, pDataBlockDesc, pStmtName, true);
+}
+
+static int32_t pushdownDataBlockSlots(SPhysiPlanContext* pCxt, SNodeList* pList, SDataBlockDescNode* pDataBlockDesc) {
+  return addDataBlockSlotsImpl(pCxt, pList, pDataBlockDesc, NULL, true);
+}
+
 typedef struct SSetSlotIdCxt {
   int32_t errCode;
   SHashObj* pLeftHash;
@@ -108,16 +236,17 @@ static EDealRes doSetSlotId(SNode* pNode, void* pContext) {
   if (QUERY_NODE_COLUMN == nodeType(pNode) && 0 != strcmp(((SColumnNode*)pNode)->colName, "*")) {
     SSetSlotIdCxt* pCxt = (SSetSlotIdCxt*)pContext;
     char name[TSDB_TABLE_NAME_LEN + TSDB_COL_NAME_LEN];
-    int32_t len = getSlotKey(pNode, name);
+    int32_t len = getSlotKey(pNode, NULL, name);
     SSlotIndex* pIndex = taosHashGet(pCxt->pLeftHash, name, len);
     if (NULL == pIndex) {
       pIndex = taosHashGet(pCxt->pRightHash, name, len);
     }
     // pIndex is definitely not NULL, otherwise it is a bug
-    CHECK_ALLOC(pIndex, DEAL_RES_ERROR);
+    if (NULL == pIndex) {
+      return DEAL_RES_ERROR;
+    }
     ((SColumnNode*)pNode)->dataBlockId = pIndex->dataBlockId;
-    ((SColumnNode*)pNode)->slotId = pIndex->slotId;
+    ((SColumnNode*)pNode)->slotId = ((SSlotIdInfo*)taosArrayGet(pIndex->pSlotIdsInfo, 0))->slotId;
-    CHECK_ALLOC(pNode, DEAL_RES_ERROR);
     return DEAL_RES_IGNORE_CHILD;
   }
   return DEAL_RES_CONTINUE;
@@ -144,7 +273,7 @@ static int32_t setNodeSlotId(SPhysiPlanContext* pCxt, int16_t leftDataBlockId, i
   return TSDB_CODE_SUCCESS;
 }
 
-static int32_t setListSlotId(SPhysiPlanContext* pCxt, int16_t leftDataBlockId, int16_t rightDataBlockId, SNodeList* pList, SNodeList** pOutput) {
+static int32_t setListSlotId(SPhysiPlanContext* pCxt, int16_t leftDataBlockId, int16_t rightDataBlockId, const SNodeList* pList, SNodeList** pOutput) {
   SNodeList* pRes = nodesCloneList(pList);
   if (NULL == pRes) {
     return TSDB_CODE_OUT_OF_MEMORY;
@@ -164,18 +293,17 @@ static int32_t setListSlotId(SPhysiPlanContext* pCxt, int16_t leftDataBlockId, i
   return TSDB_CODE_SUCCESS;
 }
 
-static SPhysiNode* makePhysiNode(SPhysiPlanContext* pCxt, ENodeType type) {
+static SPhysiNode* makePhysiNode(SPhysiPlanContext* pCxt, SLogicNode* pLogicNode, ENodeType type) {
   SPhysiNode* pPhysiNode = (SPhysiNode*)nodesMakeNode(type);
   if (NULL == pPhysiNode) {
     return NULL;
   }
-  pPhysiNode->pOutputDataBlockDesc = nodesMakeNode(QUERY_NODE_DATABLOCK_DESC);
+
-  if (NULL == pPhysiNode->pOutputDataBlockDesc) {
+  int32_t code = createDataBlockDesc(pCxt, pLogicNode->pTargets, &pPhysiNode->pOutputDataBlockDesc);
+  if (TSDB_CODE_SUCCESS != code) {
     nodesDestroyNode(pPhysiNode);
     return NULL;
   }
-  pPhysiNode->pOutputDataBlockDesc->dataBlockId = pCxt->nextDataBlockId++;
-  pPhysiNode->pOutputDataBlockDesc->type = QUERY_NODE_DATABLOCK_DESC;
   return pPhysiNode;
 }
 
@@ -186,24 +314,11 @@ static int32_t setConditionsSlotId(SPhysiPlanContext* pCxt, const SLogicNode* pL
   return TSDB_CODE_SUCCESS;
 }
 
-static int32_t setSlotOutput(SPhysiPlanContext* pCxt, SNodeList* pTargets, SDataBlockDescNode* pDataBlockDesc) {
-  SHashObj* pHash = taosArrayGetP(pCxt->pLocationHelper, pDataBlockDesc->dataBlockId);
-  char name[TSDB_TABLE_NAME_LEN + TSDB_COL_NAME_LEN];
-  SNode* pNode;
-  FOREACH(pNode, pTargets) {
-    int32_t len = getSlotKey(pNode, name);
-    SSlotIndex* pIndex = taosHashGet(pHash, name, len);
-    // pIndex is definitely not NULL, otherwise it is a bug
-    CHECK_ALLOC(pIndex, TSDB_CODE_FAILED);
-    ((SSlotDescNode*)nodesListGetNode(pDataBlockDesc->pSlots, pIndex->slotId))->output = true;
-  }
-  
-  return TSDB_CODE_SUCCESS;
-}
-
 static SNodeptr createPrimaryKeyCol(SPhysiPlanContext* pCxt, uint64_t tableId) {
   SColumnNode* pCol = nodesMakeNode(QUERY_NODE_COLUMN);
-  CHECK_ALLOC(pCol, NULL);
+  if (NULL == pCol) {
+    return NULL;
+  }
   pCol->node.resType.type = TSDB_DATA_TYPE_TIMESTAMP;
   pCol->node.resType.bytes = tDataTypes[TSDB_DATA_TYPE_TIMESTAMP].bytes;
   pCol->tableId = tableId;
@@ -244,8 +359,12 @@ static int32_t createScanCols(SPhysiPlanContext* pCxt, SScanPhysiNode* pScanPhys
   if (QUERY_NODE_PHYSICAL_PLAN_TABLE_SCAN == nodeType(pScanPhysiNode)
       || QUERY_NODE_PHYSICAL_PLAN_TABLE_SEQ_SCAN == nodeType(pScanPhysiNode)) {
     pScanPhysiNode->pScanCols = nodesMakeList();
-    CHECK_ALLOC(pScanPhysiNode->pScanCols, TSDB_CODE_OUT_OF_MEMORY);
+    if (NULL == pScanPhysiNode->pScanCols) {
-    CHECK_CODE_EXT(nodesListStrictAppend(pScanPhysiNode->pScanCols, createPrimaryKeyCol(pCxt, pScanPhysiNode->uid)));
+      return TSDB_CODE_OUT_OF_MEMORY;
+    }
+    if (TSDB_CODE_SUCCESS != nodesListStrictAppend(pScanPhysiNode->pScanCols, createPrimaryKeyCol(pCxt, pScanPhysiNode->uid))) {
+      return TSDB_CODE_OUT_OF_MEMORY;
+    }
 
     SNode* pNode;
     FOREACH(pNode, pScanCols) {
@@ -255,29 +374,29 @@ static int32_t createScanCols(SPhysiPlanContext* pCxt, SScanPhysiNode* pScanPhys
         strcpy(pCol->colName, ((SColumnNode*)pNode)->colName);
         continue;
       }
-      CHECK_CODE_EXT(nodesListStrictAppend(pScanPhysiNode->pScanCols, nodesCloneNode(pNode)));
+      if (TSDB_CODE_SUCCESS != nodesListStrictAppend(pScanPhysiNode->pScanCols, nodesCloneNode(pNode))) {
+        return TSDB_CODE_OUT_OF_MEMORY;
+      }
     }
   } else {
     pScanPhysiNode->pScanCols = nodesCloneList(pScanCols);
-    CHECK_ALLOC(pScanPhysiNode->pScanCols, TSDB_CODE_OUT_OF_MEMORY);
+    if (NULL == pScanPhysiNode->pScanCols) {
+      return TSDB_CODE_OUT_OF_MEMORY;
+    }
   }
 
-  // return sortScanCols(pScanPhysiNode->pScanCols);
+  return sortScanCols(pScanPhysiNode->pScanCols);
-  return TSDB_CODE_SUCCESS;
 }
 
 static int32_t createScanPhysiNodeFinalize(SPhysiPlanContext* pCxt, SScanLogicNode* pScanLogicNode, SScanPhysiNode* pScanPhysiNode, SPhysiNode** pPhyNode) {
   int32_t code = createScanCols(pCxt, pScanPhysiNode, pScanLogicNode->pScanCols);
   if (TSDB_CODE_SUCCESS == code) {
     // Data block describe also needs to be set without scanning column, such as SELECT COUNT(*) FROM t
-    code = addDataBlockDesc(pCxt, pScanPhysiNode->pScanCols, pScanPhysiNode->node.pOutputDataBlockDesc);
+    code = addDataBlockSlots(pCxt, pScanPhysiNode->pScanCols, pScanPhysiNode->node.pOutputDataBlockDesc);
   }
   if (TSDB_CODE_SUCCESS == code) {
     code = setConditionsSlotId(pCxt, (const SLogicNode*)pScanLogicNode, (SPhysiNode*)pScanPhysiNode);
   }
-  if (TSDB_CODE_SUCCESS == code) {
-    code = setSlotOutput(pCxt, pScanLogicNode->node.pTargets, pScanPhysiNode->node.pOutputDataBlockDesc);
-  }
   if (TSDB_CODE_SUCCESS == code) {
     pScanPhysiNode->uid = pScanLogicNode->pMeta->uid;
     pScanPhysiNode->tableType = pScanLogicNode->pMeta->tableType;
@@ -302,7 +421,7 @@ static void vgroupInfoToNodeAddr(const SVgroupInfo* vg, SQueryNodeAddr* pNodeAdd
 }
 
 static int32_t createTagScanPhysiNode(SPhysiPlanContext* pCxt, SScanLogicNode* pScanLogicNode, SPhysiNode** pPhyNode) {
-  STagScanPhysiNode* pTagScan = (STagScanPhysiNode*)makePhysiNode(pCxt, QUERY_NODE_PHYSICAL_PLAN_TAG_SCAN);
+  STagScanPhysiNode* pTagScan = (STagScanPhysiNode*)makePhysiNode(pCxt, (SLogicNode*)pScanLogicNode, QUERY_NODE_PHYSICAL_PLAN_TAG_SCAN);
   if (NULL == pTagScan) {
     return TSDB_CODE_OUT_OF_MEMORY;
   }
@@ -310,7 +429,7 @@ static int32_t createTagScanPhysiNode(SPhysiPlanContext* pCxt, SScanLogicNode* p
 }
 
 static int32_t createTableScanPhysiNode(SPhysiPlanContext* pCxt, SSubplan* pSubplan, SScanLogicNode* pScanLogicNode, SPhysiNode** pPhyNode) {
-  STableScanPhysiNode* pTableScan = (STableScanPhysiNode*)makePhysiNode(pCxt, QUERY_NODE_PHYSICAL_PLAN_TABLE_SCAN);
+  STableScanPhysiNode* pTableScan = (STableScanPhysiNode*)makePhysiNode(pCxt, (SLogicNode*)pScanLogicNode, QUERY_NODE_PHYSICAL_PLAN_TABLE_SCAN);
   if (NULL == pTableScan) {
     return TSDB_CODE_OUT_OF_MEMORY;
   }
@@ -326,7 +445,7 @@ static int32_t createTableScanPhysiNode(SPhysiPlanContext* pCxt, SSubplan* pSubp
 }
 
 static int32_t createSystemTableScanPhysiNode(SPhysiPlanContext* pCxt, SSubplan* pSubplan, SScanLogicNode* pScanLogicNode, SPhysiNode** pPhyNode) {
-  SSystemTableScanPhysiNode* pScan = (SSystemTableScanPhysiNode*)makePhysiNode(pCxt, QUERY_NODE_PHYSICAL_PLAN_SYSTABLE_SCAN);
+  SSystemTableScanPhysiNode* pScan = (SSystemTableScanPhysiNode*)makePhysiNode(pCxt, (SLogicNode*)pScanLogicNode, QUERY_NODE_PHYSICAL_PLAN_SYSTABLE_SCAN);
   if (NULL == pScan) {
     return TSDB_CODE_OUT_OF_MEMORY;
   }
@@ -347,7 +466,7 @@ static int32_t createSystemTableScanPhysiNode(SPhysiPlanContext* pCxt, SSubplan*
 }
 
 static int32_t createStreamScanPhysiNode(SPhysiPlanContext* pCxt, SSubplan* pSubplan, SScanLogicNode* pScanLogicNode, SPhysiNode** pPhyNode) {
-  SStreamScanPhysiNode* pScan = (SStreamScanPhysiNode*)makePhysiNode(pCxt, QUERY_NODE_PHYSICAL_PLAN_STREAM_SCAN);
+  SStreamScanPhysiNode* pScan = (SStreamScanPhysiNode*)makePhysiNode(pCxt, (SLogicNode*)pScanLogicNode, QUERY_NODE_PHYSICAL_PLAN_STREAM_SCAN);
   if (NULL == pScan) {
     return TSDB_CODE_OUT_OF_MEMORY;
   }
@@ -411,7 +530,7 @@ static int32_t createJoinOutputCols(SPhysiPlanContext* pCxt, SDataBlockDescNode*
 }
 
 static int32_t createJoinPhysiNode(SPhysiPlanContext* pCxt, SNodeList* pChildren, SJoinLogicNode* pJoinLogicNode, SPhysiNode** pPhyNode) {
-  SJoinPhysiNode* pJoin = (SJoinPhysiNode*)makePhysiNode(pCxt, QUERY_NODE_PHYSICAL_PLAN_JOIN);
+  SJoinPhysiNode* pJoin = (SJoinPhysiNode*)makePhysiNode(pCxt, (SLogicNode*)pJoinLogicNode, QUERY_NODE_PHYSICAL_PLAN_JOIN);
   if (NULL == pJoin) {
     return TSDB_CODE_OUT_OF_MEMORY;
   }
@@ -424,14 +543,11 @@ static int32_t createJoinPhysiNode(SPhysiPlanContext* pCxt, SNodeList* pChildren
     code = createJoinOutputCols(pCxt, pLeftDesc, pRightDesc, &pJoin->pTargets);
   }
   if (TSDB_CODE_SUCCESS == code) {
-    code = addDataBlockDesc(pCxt, pJoin->pTargets, pJoin->node.pOutputDataBlockDesc);
+    code = addDataBlockSlots(pCxt, pJoin->pTargets, pJoin->node.pOutputDataBlockDesc);
   }
   if (TSDB_CODE_SUCCESS == code) {
     code = setConditionsSlotId(pCxt, (const SLogicNode*)pJoinLogicNode, (SPhysiNode*)pJoin);
   }
-  if (TSDB_CODE_SUCCESS == code) {
-    code = setSlotOutput(pCxt, pJoinLogicNode->node.pTargets, pJoin->node.pOutputDataBlockDesc);
-  }
 
   if (TSDB_CODE_SUCCESS == code) {
     *pPhyNode = (SPhysiNode*)pJoin;
@@ -451,7 +567,9 @@ typedef struct SRewritePrecalcExprsCxt {
 
 static EDealRes collectAndRewrite(SRewritePrecalcExprsCxt* pCxt, SNode** pNode) {
   SNode* pExpr = nodesCloneNode(*pNode);
-  CHECK_ALLOC(pExpr, DEAL_RES_ERROR);
+  if (NULL == pExpr) {
+    return DEAL_RES_ERROR;
+  }
   if (nodesListAppend(pCxt->pPrecalcExprs, pExpr)) {
     nodesDestroyNode(pExpr);
     return DEAL_RES_ERROR;
@@ -499,11 +617,15 @@ static int32_t rewritePrecalcExprs(SPhysiPlanContext* pCxt, SNodeList* pList, SN
 
   if (NULL == *pPrecalcExprs) {
     *pPrecalcExprs = nodesMakeList();
-    CHECK_ALLOC(*pPrecalcExprs, TSDB_CODE_OUT_OF_MEMORY);
+    if (NULL == *pPrecalcExprs) {
+      return TSDB_CODE_OUT_OF_MEMORY;
+    }
   }
   if (NULL == *pRewrittenList) {
     *pRewrittenList = nodesMakeList();
-    CHECK_ALLOC(*pRewrittenList, TSDB_CODE_OUT_OF_MEMORY);
+    if (NULL == *pRewrittenList) {
+      return TSDB_CODE_OUT_OF_MEMORY;
+    }
   }
   SNode* pNode = NULL;
   FOREACH(pNode, pList) {
@@ -513,8 +635,12 @@ static int32_t rewritePrecalcExprs(SPhysiPlanContext* pCxt, SNodeList* pList, SN
     } else {
       pNew = nodesCloneNode(pNode);
     }
-    CHECK_ALLOC(pNew, TSDB_CODE_OUT_OF_MEMORY);
+    if (NULL == pNew) {
-    CHECK_CODE(nodesListAppend(*pRewrittenList, pNew), TSDB_CODE_OUT_OF_MEMORY);
+      return TSDB_CODE_OUT_OF_MEMORY;
+    }
+    if (TSDB_CODE_SUCCESS != nodesListAppend(*pRewrittenList, pNew)) {
+      return TSDB_CODE_OUT_OF_MEMORY;
+    }
   }
   SRewritePrecalcExprsCxt cxt = { .errCode = TSDB_CODE_SUCCESS, .pPrecalcExprs = *pPrecalcExprs };
   nodesRewriteList(*pRewrittenList, doRewritePrecalcExprs, &cxt);
@@ -526,7 +652,7 @@ static int32_t rewritePrecalcExprs(SPhysiPlanContext* pCxt, SNodeList* pList, SN
 }
 
 static int32_t createAggPhysiNode(SPhysiPlanContext* pCxt, SNodeList* pChildren, SAggLogicNode* pAggLogicNode, SPhysiNode** pPhyNode) {
-  SAggPhysiNode* pAgg = (SAggPhysiNode*)makePhysiNode(pCxt, QUERY_NODE_PHYSICAL_PLAN_AGG);
+  SAggPhysiNode* pAgg = (SAggPhysiNode*)makePhysiNode(pCxt, (SLogicNode*)pAggLogicNode, QUERY_NODE_PHYSICAL_PLAN_AGG);
   if (NULL == pAgg) {
     return TSDB_CODE_OUT_OF_MEMORY;
   }
@@ -544,30 +670,27 @@ static int32_t createAggPhysiNode(SPhysiPlanContext* pCxt, SNodeList* pChildren,
   if (TSDB_CODE_SUCCESS == code && NULL != pPrecalcExprs) {
     code = setListSlotId(pCxt, pChildTupe->dataBlockId, -1, pPrecalcExprs, &pAgg->pExprs);
     if (TSDB_CODE_SUCCESS == code) {
-      code = addDataBlockDesc(pCxt, pAgg->pExprs, pChildTupe);
+      code = pushdownDataBlockSlots(pCxt, pAgg->pExprs, pChildTupe);
     }
   }
 
   if (TSDB_CODE_SUCCESS == code && NULL != pGroupKeys) {
     code = setListSlotId(pCxt, pChildTupe->dataBlockId, -1, pGroupKeys, &pAgg->pGroupKeys);
     if (TSDB_CODE_SUCCESS == code) {
-      code = addDataBlockDesc(pCxt, pAgg->pGroupKeys, pAgg->node.pOutputDataBlockDesc);
+      code = addDataBlockSlots(pCxt, pAgg->pGroupKeys, pAgg->node.pOutputDataBlockDesc);
     }
   }
 
   if (TSDB_CODE_SUCCESS == code && NULL != pAggFuncs) {
     code = setListSlotId(pCxt, pChildTupe->dataBlockId, -1, pAggFuncs, &pAgg->pAggFuncs);
     if (TSDB_CODE_SUCCESS == code) {
-      code = addDataBlockDesc(pCxt, pAgg->pAggFuncs, pAgg->node.pOutputDataBlockDesc);
+      code = addDataBlockSlots(pCxt, pAgg->pAggFuncs, pAgg->node.pOutputDataBlockDesc);
     }
   }
 
   if (TSDB_CODE_SUCCESS == code) {
     code = setConditionsSlotId(pCxt, (const SLogicNode*)pAggLogicNode, (SPhysiNode*)pAgg);
   }
-  if (TSDB_CODE_SUCCESS == code) {
-    code = setSlotOutput(pCxt, pAggLogicNode->node.pTargets, pAgg->node.pOutputDataBlockDesc);
-  }
 
   if (TSDB_CODE_SUCCESS == code) {
     *pPhyNode = (SPhysiNode*)pAgg;
@@ -575,18 +698,22 @@ static int32_t createAggPhysiNode(SPhysiPlanContext* pCxt, SNodeList* pChildren,
     nodesDestroyNode(pAgg);
   }
 
+  nodesDestroyList(pPrecalcExprs);
+  nodesDestroyList(pGroupKeys);
+  nodesDestroyList(pAggFuncs);
+
   return code;
 }
 
 static int32_t createProjectPhysiNode(SPhysiPlanContext* pCxt, SNodeList* pChildren, SProjectLogicNode* pProjectLogicNode, SPhysiNode** pPhyNode) {
-  SProjectPhysiNode* pProject = (SProjectPhysiNode*)makePhysiNode(pCxt, QUERY_NODE_PHYSICAL_PLAN_PROJECT);
+  SProjectPhysiNode* pProject = (SProjectPhysiNode*)makePhysiNode(pCxt, (SLogicNode*)pProjectLogicNode, QUERY_NODE_PHYSICAL_PLAN_PROJECT);
   if (NULL == pProject) {
     return TSDB_CODE_OUT_OF_MEMORY;
   }
 
   int32_t code = setListSlotId(pCxt, ((SPhysiNode*)nodesListGetNode(pChildren, 0))->pOutputDataBlockDesc->dataBlockId, -1, pProjectLogicNode->pProjections, &pProject->pProjections);
   if (TSDB_CODE_SUCCESS == code) {
-    code = addDataBlockDesc(pCxt, pProject->pProjections, pProject->node.pOutputDataBlockDesc);
+    code = addDataBlockSlotsForProject(pCxt, pProjectLogicNode->stmtName, pProject->pProjections, pProject->node.pOutputDataBlockDesc);
   }
   if (TSDB_CODE_SUCCESS == code) {
     code = setConditionsSlotId(pCxt, (const SLogicNode*)pProjectLogicNode, (SPhysiNode*)pProject);
@@ -602,34 +729,30 @@ static int32_t createProjectPhysiNode(SPhysiPlanContext* pCxt, SNodeList* pChild
 }
 
 static int32_t doCreateExchangePhysiNode(SPhysiPlanContext* pCxt, SExchangeLogicNode* pExchangeLogicNode, SPhysiNode** pPhyNode) {
-  SExchangePhysiNode* pExchange = (SExchangePhysiNode*)makePhysiNode(pCxt, QUERY_NODE_PHYSICAL_PLAN_EXCHANGE);
+  SExchangePhysiNode* pExchange = (SExchangePhysiNode*)makePhysiNode(pCxt, (SLogicNode*)pExchangeLogicNode, QUERY_NODE_PHYSICAL_PLAN_EXCHANGE);
   if (NULL == pExchange) {
     return TSDB_CODE_OUT_OF_MEMORY;
   }
     
   pExchange->srcGroupId = pExchangeLogicNode->srcGroupId;
-  int32_t code = addDataBlockDesc(pCxt, pExchangeLogicNode->node.pTargets, pExchange->node.pOutputDataBlockDesc);
+  *pPhyNode = (SPhysiNode*)pExchange;
 
-  if (TSDB_CODE_SUCCESS == code) {
+  return TSDB_CODE_SUCCESS;
-    *pPhyNode = (SPhysiNode*)pExchange;
-  } else {
-    nodesDestroyNode(pExchange);
-  }
-
-  return code;
 }
 static int32_t createStreamScanPhysiNodeByExchange(SPhysiPlanContext* pCxt, SExchangeLogicNode* pExchangeLogicNode, SPhysiNode** pPhyNode) {
-  SStreamScanPhysiNode* pScan = (SStreamScanPhysiNode*)makePhysiNode(pCxt, QUERY_NODE_PHYSICAL_PLAN_STREAM_SCAN);
+  SStreamScanPhysiNode* pScan = (SStreamScanPhysiNode*)makePhysiNode(pCxt, (SLogicNode*)pExchangeLogicNode, QUERY_NODE_PHYSICAL_PLAN_STREAM_SCAN);
   if (NULL == pScan) {
     return TSDB_CODE_OUT_OF_MEMORY;
   }
   
-  int32_t code = addDataBlockDesc(pCxt, pExchangeLogicNode->node.pTargets, pScan->node.pOutputDataBlockDesc);
+  int32_t code = TSDB_CODE_SUCCESS;
+
+  pScan->pScanCols = nodesCloneList(pExchangeLogicNode->node.pTargets);
+  if (NULL == pScan->pScanCols) {
+    code = TSDB_CODE_OUT_OF_MEMORY;
+  }
   if (TSDB_CODE_SUCCESS == code) {
-    pScan->pScanCols = nodesCloneList(pExchangeLogicNode->node.pTargets);
+    code = addDataBlockSlots(pCxt, pScan->pScanCols, pScan->node.pOutputDataBlockDesc);
-    if (NULL == pScan->pScanCols) {
-      code = TSDB_CODE_OUT_OF_MEMORY;
-    }
   }
 
   if (TSDB_CODE_SUCCESS == code) {
@@ -659,21 +782,17 @@ static int32_t createWindowPhysiNodeFinalize(SPhysiPlanContext* pCxt, SNodeList*
   if (TSDB_CODE_SUCCESS == code && NULL != pPrecalcExprs) {
     code = setListSlotId(pCxt, pChildTupe->dataBlockId, -1, pPrecalcExprs, &pWindow->pExprs);
     if (TSDB_CODE_SUCCESS == code) {
-      code = addDataBlockDesc(pCxt, pWindow->pExprs, pChildTupe);
+      code = addDataBlockSlots(pCxt, pWindow->pExprs, pChildTupe);
     }
   }
 
   if (TSDB_CODE_SUCCESS == code && NULL != pFuncs) {
     code = setListSlotId(pCxt, pChildTupe->dataBlockId, -1, pFuncs, &pWindow->pFuncs);
     if (TSDB_CODE_SUCCESS == code) {
-      code = addDataBlockDesc(pCxt, pWindow->pFuncs, pWindow->node.pOutputDataBlockDesc);
+      code = addDataBlockSlots(pCxt, pWindow->pFuncs, pWindow->node.pOutputDataBlockDesc);
     }
   }
 
-  if (TSDB_CODE_SUCCESS == code) {
-    code = setSlotOutput(pCxt, pWindowLogicNode->node.pTargets, pWindow->node.pOutputDataBlockDesc);
-  }
-
   if (TSDB_CODE_SUCCESS == code) {
     *pPhyNode = (SPhysiNode*)pWindow;
   } else {
@@ -684,7 +803,7 @@ static int32_t createWindowPhysiNodeFinalize(SPhysiPlanContext* pCxt, SNodeList*
 }
 
 static int32_t createIntervalPhysiNode(SPhysiPlanContext* pCxt, SNodeList* pChildren, SWindowLogicNode* pWindowLogicNode, SPhysiNode** pPhyNode) {
-  SIntervalPhysiNode* pInterval = (SIntervalPhysiNode*)makePhysiNode(pCxt, QUERY_NODE_PHYSICAL_PLAN_INTERVAL);
+  SIntervalPhysiNode* pInterval = (SIntervalPhysiNode*)makePhysiNode(pCxt, (SLogicNode*)pWindowLogicNode, QUERY_NODE_PHYSICAL_PLAN_INTERVAL);
   if (NULL == pInterval) {
     return TSDB_CODE_OUT_OF_MEMORY;
   }
@@ -701,11 +820,18 @@ static int32_t createIntervalPhysiNode(SPhysiPlanContext* pCxt, SNodeList* pChil
     return TSDB_CODE_OUT_OF_MEMORY;
   }
 
+  SDataBlockDescNode* pChildTupe = (((SPhysiNode*)nodesListGetNode(pChildren, 0))->pOutputDataBlockDesc);
+  int32_t code = setNodeSlotId(pCxt, pChildTupe->dataBlockId, -1, pWindowLogicNode->pTspk, &pInterval->pTspk);
+  if (TSDB_CODE_SUCCESS != code) {
+    nodesDestroyNode(pInterval);
+    return code;
+  }
+
   return createWindowPhysiNodeFinalize(pCxt, pChildren, &pInterval->window, pWindowLogicNode, pPhyNode);
 }
 
 static int32_t createSessionWindowPhysiNode(SPhysiPlanContext* pCxt, SNodeList* pChildren, SWindowLogicNode* pWindowLogicNode, SPhysiNode** pPhyNode) {
-  SSessionWinodwPhysiNode* pSession = (SSessionWinodwPhysiNode*)makePhysiNode(pCxt, QUERY_NODE_PHYSICAL_PLAN_SESSION_WINDOW);
+  SSessionWinodwPhysiNode* pSession = (SSessionWinodwPhysiNode*)makePhysiNode(pCxt, (SLogicNode*)pWindowLogicNode, QUERY_NODE_PHYSICAL_PLAN_SESSION_WINDOW);
   if (NULL == pSession) {
     return TSDB_CODE_OUT_OF_MEMORY;
   }
@@ -729,6 +855,41 @@ static int32_t createWindowPhysiNode(SPhysiPlanContext* pCxt, SNodeList* pChildr
   return TSDB_CODE_FAILED;
 }
 
+static int32_t createSortPhysiNode(SPhysiPlanContext* pCxt, SNodeList* pChildren, SSortLogicNode* pSortLogicNode, SPhysiNode** pPhyNode) {
+  SSortPhysiNode* pSort = (SSortPhysiNode*)makePhysiNode(pCxt, (SLogicNode*)pSortLogicNode, QUERY_NODE_PHYSICAL_PLAN_SORT);
+  if (NULL == pSort) {
+    return TSDB_CODE_OUT_OF_MEMORY;
+  }
+
+  SNodeList* pPrecalcExprs = NULL;
+  SNodeList* pSortKeys = NULL;
+  int32_t code = rewritePrecalcExprs(pCxt, pSortLogicNode->pSortKeys, &pPrecalcExprs, &pSortKeys);
+
+  SDataBlockDescNode* pChildTupe = (((SPhysiNode*)nodesListGetNode(pChildren, 0))->pOutputDataBlockDesc);
+  // push down expression to pOutputDataBlockDesc of child node
+  if (TSDB_CODE_SUCCESS == code && NULL != pPrecalcExprs) {
+    code = setListSlotId(pCxt, pChildTupe->dataBlockId, -1, pPrecalcExprs, &pSort->pExprs);
+    if (TSDB_CODE_SUCCESS == code) {
+      code = addDataBlockSlots(pCxt, pSort->pExprs, pChildTupe);
+    }
+  }
+
+  if (TSDB_CODE_SUCCESS == code) {
+    code = setListSlotId(pCxt, pChildTupe->dataBlockId, -1, pSortKeys, &pSort->pSortKeys);
+    if (TSDB_CODE_SUCCESS == code) {
+      code = addDataBlockSlots(pCxt, pSort->pSortKeys, pSort->node.pOutputDataBlockDesc);
+    }
+  }
+
+  if (TSDB_CODE_SUCCESS == code) {
+    *pPhyNode = (SPhysiNode*)pSort;
+  } else {
+    nodesDestroyNode(pSort);
+  }
+
+  return code;
+}
+
 static int32_t doCreatePhysiNode(SPhysiPlanContext* pCxt, SLogicNode* pLogicNode, SSubplan* pSubplan, SNodeList* pChildren, SPhysiNode** pPhyNode) {
   switch (nodeType(pLogicNode)) {
     case QUERY_NODE_LOGIC_PLAN_SCAN:
@@ -743,6 +904,8 @@ static int32_t doCreatePhysiNode(SPhysiPlanContext* pCxt, SLogicNode* pLogicNode
       return createExchangePhysiNode(pCxt, (SExchangeLogicNode*)pLogicNode, pPhyNode);
     case QUERY_NODE_LOGIC_PLAN_WINDOW:
       return createWindowPhysiNode(pCxt, pChildren, (SWindowLogicNode*)pLogicNode, pPhyNode);
+    case QUERY_NODE_LOGIC_PLAN_SORT:
+      return createSortPhysiNode(pCxt, pChildren, (SSortLogicNode*)pLogicNode, pPhyNode);
     default:
       break;
   }
@@ -874,17 +1037,22 @@ static int32_t pushSubplan(SPhysiPlanContext* pCxt, SNodeptr pSubplan, int32_t l
   SNodeListNode* pGroup;
   if (level >= LIST_LENGTH(pSubplans)) {
     pGroup = nodesMakeNode(QUERY_NODE_NODE_LIST);
-    CHECK_ALLOC(pGroup, TSDB_CODE_OUT_OF_MEMORY);
+    if (NULL == pGroup) {
-    CHECK_CODE(nodesListStrictAppend(pSubplans, pGroup), TSDB_CODE_OUT_OF_MEMORY);
+      return TSDB_CODE_OUT_OF_MEMORY;
+    }
+    if (TSDB_CODE_SUCCESS != nodesListStrictAppend(pSubplans, pGroup)) {
+      return TSDB_CODE_OUT_OF_MEMORY;
+    }
   } else {
     pGroup = nodesListGetNode(pSubplans, level);
   }
   if (NULL == pGroup->pNodeList) {
     pGroup->pNodeList = nodesMakeList();
-    CHECK_ALLOC(pGroup->pNodeList, TSDB_CODE_OUT_OF_MEMORY);
+    if (NULL == pGroup->pNodeList) {
+      return TSDB_CODE_OUT_OF_MEMORY;
+    }
   }
-  CHECK_CODE(nodesListStrictAppend(pGroup->pNodeList, pSubplan), TSDB_CODE_OUT_OF_MEMORY);
+  return nodesListStrictAppend(pGroup->pNodeList, pSubplan);
-  return TSDB_CODE_SUCCESS;
 }
 
 static int32_t buildPhysiPlan(SPhysiPlanContext* pCxt, SLogicSubplan* pLogicSubplan, SSubplan* pParent, SQueryPlan* pQueryPlan) {

source/libs/planner/src/planSpliter.c

@@ -65,7 +65,9 @@ static int32_t stsMatch(SSplitContext* pCxt, SLogicSubplan* pSubplan) {
   SLogicNode* pSplitNode = stsMatchByNode(pSubplan->pNode);
   if (NULL != pSplitNode) {
     SStsInfo* pInfo = taosMemoryCalloc(1, sizeof(SStsInfo));
-    CHECK_ALLOC(pInfo, TSDB_CODE_OUT_OF_MEMORY);
+    if (NULL == pInfo) {
+      return TSDB_CODE_OUT_OF_MEMORY;
+    }
     pInfo->pScan = (SScanLogicNode*)pSplitNode;
     pInfo->pSubplan = pSubplan;
     pCxt->pInfo = pInfo;

source/libs/planner/test/plannerTest.cpp

@@ -170,7 +170,7 @@ TEST_F(PlannerTest, groupBy) {
   bind("SELECT count(*) FROM t1");
   ASSERT_TRUE(run());
 
-  bind("SELECT c1, count(*) FROM t1 GROUP BY c1");
+  bind("SELECT c1, max(c3), min(c2), count(*) FROM t1 GROUP BY c1");
   ASSERT_TRUE(run());
 
   bind("SELECT c1 + c3, c1 + count(*) FROM t1 where c2 = 'abc' GROUP BY c1, c3");
@@ -201,10 +201,31 @@ TEST_F(PlannerTest, sessionWindow) {
   ASSERT_TRUE(run());
 }
 
+TEST_F(PlannerTest, orderBy) {
+  setDatabase("root", "test");
+
+  bind("SELECT * FROM t1 order by c1");
+  ASSERT_TRUE(run());
+
+  bind("SELECT c1 FROM t1 order by c2");
+  ASSERT_TRUE(run());
+
+  bind("SELECT * FROM t1 order by c1 + 10, c2");
+  ASSERT_TRUE(run());
+}
+
 TEST_F(PlannerTest, showTables) {
   setDatabase("root", "test");
 
   bind("show tables");
+  ASSERT_TRUE(run());
+}
+
+TEST_F(PlannerTest, showStables) {
+  setDatabase("root", "test");
+
+  bind("show stables");
+  ASSERT_TRUE(run());
 }
 
 TEST_F(PlannerTest, createTopic) {

source/libs/qcom/src/queryUtil.c

@@ -198,7 +198,7 @@ char *jobTaskStatusStr(int32_t status) {
   return "UNKNOWN";
 }
 
-SSchema createSchema(uint8_t type, int32_t bytes, int32_t colId, const char* name) {
+SSchema createSchema(int8_t type, int32_t bytes, col_id_t colId, const char* name) {
   SSchema s = {0};
   s.type  = type;
   s.bytes = bytes;

source/libs/qcom/src/querymsg.c

@@ -188,7 +188,7 @@ static int32_t queryConvertTableMetaMsg(STableMetaRsp *pMetaMsg) {
   }
 
   if (pMetaMsg->pSchemas[0].colId != PRIMARYKEY_TIMESTAMP_COL_ID) {
-    qError("invalid colId[%d] for the first column in table meta rsp msg", pMetaMsg->pSchemas[0].colId);
+    qError("invalid colId[%" PRIi16 "] for the first column in table meta rsp msg", pMetaMsg->pSchemas[0].colId);
     return TSDB_CODE_TSC_INVALID_VALUE;
   }
 

source/libs/qworker/src/qworker.c

@@ -947,10 +947,12 @@ int32_t qwProcessQuery(QW_FPARAMS_DEF, SQWMsg *qwMsg, int8_t taskType) {
   QW_ERR_JRET(qwGetTaskCtx(QW_FPARAMS(), &ctx));
   
   atomic_store_8(&ctx->taskType, taskType);
+
   atomic_store_ptr(&ctx->connInfo.handle, qwMsg->connInfo.handle);
   atomic_store_ptr(&ctx->connInfo.ahandle, qwMsg->connInfo.ahandle);
 
   QW_TASK_DLOGL("subplan json string, len:%d, %s", qwMsg->msgLen, qwMsg->msg);
+
   code = qStringToSubplan(qwMsg->msg, &plan);
   if (TSDB_CODE_SUCCESS != code) {
     QW_TASK_ELOG("task string to subplan failed, code:%x - %s", code, tstrerror(code));

source/libs/qworker/src/qworkerMsg.c

@@ -170,7 +170,7 @@ int32_t qwBuildAndSendShowRsp(SRpcMsg *pMsg, int32_t code) {
     return -1;
   }
 
-  int32_t  cols = 0;
+  col_id_t  cols = 0;
   SSchema *pSchema = showRsp.tableMeta.pSchemas;
 
   const SSchema *s = tGetTbnameColumnSchema();

source/libs/stream/src/tstream.c

@@ -121,7 +121,7 @@ int32_t streamExecTask(SStreamTask* pTask, SMsgCb* pMsgCb, const void* input, in
   } else if (pTask->dispatchType == TASK_DISPATCH__FIXED) {
     SStreamTaskExecReq req = {
         .streamId = pTask->streamId,
-        .taskId = pTask->taskId,
+        .taskId = pTask->fixedEpDispatcher.taskId,
         .data = pRes,
     };
 
@@ -211,8 +211,9 @@ int32_t tEncodeSStreamTask(SCoder* pEncoder, const SStreamTask* pTask) {
   }
 
   if (pTask->dispatchType == TASK_DISPATCH__INPLACE) {
-    if (tEncodeI8(pEncoder, pTask->inplaceDispatcher.reserved) < 0) return -1;
+    if (tEncodeI32(pEncoder, pTask->inplaceDispatcher.taskId) < 0) return -1;
   } else if (pTask->dispatchType == TASK_DISPATCH__FIXED) {
+    if (tEncodeI32(pEncoder, pTask->fixedEpDispatcher.taskId) < 0) return -1;
     if (tEncodeI32(pEncoder, pTask->fixedEpDispatcher.nodeId) < 0) return -1;
     if (tEncodeSEpSet(pEncoder, &pTask->fixedEpDispatcher.epSet) < 0) return -1;
   } else if (pTask->dispatchType == TASK_DISPATCH__SHUFFLE) {
@@ -248,8 +249,9 @@ int32_t tDecodeSStreamTask(SCoder* pDecoder, SStreamTask* pTask) {
   }
 
   if (pTask->dispatchType == TASK_DISPATCH__INPLACE) {
-    if (tDecodeI8(pDecoder, &pTask->inplaceDispatcher.reserved) < 0) return -1;
+    if (tDecodeI32(pDecoder, &pTask->inplaceDispatcher.taskId) < 0) return -1;
   } else if (pTask->dispatchType == TASK_DISPATCH__FIXED) {
+    if (tDecodeI32(pDecoder, &pTask->fixedEpDispatcher.taskId) < 0) return -1;
     if (tDecodeI32(pDecoder, &pTask->fixedEpDispatcher.nodeId) < 0) return -1;
     if (tDecodeSEpSet(pDecoder, &pTask->fixedEpDispatcher.epSet) < 0) return -1;
   } else if (pTask->dispatchType == TASK_DISPATCH__SHUFFLE) {

source/libs/tdb/CMakeLists.txt

@@ -8,7 +8,6 @@ target_sources(tdb
     "src/db/tdbBtree.c"
     "src/db/tdbDb.c"
     "src/db/tdbEnv.c"
-    # "src/db/tdbPage.c"
     "src/page/tdbPage.c"
     "src/page/tdbPageL.c"
 )

source/libs/tdb/src/btree/tdbBtreeBalance.c

@@ -0,0 +1,14 @@
+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */

source/libs/tdb/src/btree/tdbBtreeCommon.c

@@ -0,0 +1,14 @@
+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */

source/libs/tdb/src/btree/tdbBtreeDelete.c

@@ -0,0 +1,14 @@
+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */

source/libs/tdb/src/btree/tdbBtreeInsert.c

@@ -0,0 +1,14 @@
+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */

source/libs/tdb/src/btree/tdbBtreeInt.h

@@ -0,0 +1,27 @@
+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef _TDB_BTREE_INT_H_
+#define _TDB_BTREE_INT_H_
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /*_TDB_BTREE_INT_H_*/

source/libs/tdb/src/btree/tdbBtreeOpen.c

@@ -0,0 +1,14 @@
+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */

source/libs/tdb/src/db/tdbDb.c

@@ -15,13 +15,17 @@
 
 #include "tdbInt.h"
 
-struct STDb {
+struct STDB {
   STEnv  *pEnv;
   SBTree *pBt;
 };
 
-int tdbDbOpen(const char *fname, int keyLen, int valLen, FKeyComparator keyCmprFn, STEnv *pEnv, STDb **ppDb) {
+struct STDBC {
-  STDb   *pDb;
+  SBTC btc;
+};
+
+int tdbDbOpen(const char *fname, int keyLen, int valLen, FKeyComparator keyCmprFn, STEnv *pEnv, STDB **ppDb) {
+  STDB   *pDb;
   SPager *pPager;
   int     ret;
   char    fFullName[TDB_FILENAME_LEN];
@@ -30,7 +34,7 @@ int tdbDbOpen(const char *fname, int keyLen, int valLen, FKeyComparator keyCmprF
 
   *ppDb = NULL;
 
-  pDb = (STDb *)taosMemoryCalloc(1, sizeof(*pDb));
+  pDb = (STDB *)calloc(1, sizeof(*pDb));
   if (pDb == NULL) {
     return -1;
   }
@@ -59,23 +63,23 @@ int tdbDbOpen(const char *fname, int keyLen, int valLen, FKeyComparator keyCmprF
   return 0;
 }
 
-int tdbDbClose(STDb *pDb) {
+int tdbDbClose(STDB *pDb) {
   // TODO
   return 0;
 }
 
-int tdbDbDrop(STDb *pDb) {
+int tdbDbDrop(STDB *pDb) {
   // TODO
   return 0;
 }
 
-int tdbDbInsert(STDb *pDb, const void *pKey, int keyLen, const void *pVal, int valLen) {
+int tdbDbInsert(STDB *pDb, const void *pKey, int keyLen, const void *pVal, int valLen) {
-  SBtCursor  btc;
+  SBTC  btc;
-  SBtCursor *pCur;
+  SBTC *pCur;
-  int        ret;
+  int   ret;
 
   pCur = &btc;
-  ret = tdbBtreeCursor(pCur, pDb->pBt);
+  ret = tdbBtcOpen(pCur, pDb->pBt);
   if (ret < 0) {
     return -1;
   }
@@ -85,5 +89,45 @@ int tdbDbInsert(STDb *pDb, const void *pKey, int keyLen, const void *pVal, int v
     return -1;
   }
 
+  return 0;
+}
+
+int tdbDbGet(STDB *pDb, const void *pKey, int kLen, void **ppVal, int *vLen) {
+  return tdbBtreeGet(pDb->pBt, pKey, kLen, ppVal, vLen);
+}
+
+int tdbDbcOpen(STDB *pDb, STDBC **ppDbc) {
+  int    ret;
+  STDBC *pDbc = NULL;
+
+  *ppDbc = NULL;
+  pDbc = malloc(sizeof(*pDbc));
+  if (pDbc == NULL) {
+    return -1;
+  }
+
+  tdbBtcOpen(&pDbc->btc, pDb->pBt);
+
+  // TODO: move to first now, we can move to any key-value
+  // and in any direction, design new APIs.
+  ret = tdbBtcMoveToFirst(&pDbc->btc);
+  if (ret < 0) {
+    ASSERT(0);
+    return -1;
+  }
+
+  *ppDbc = pDbc;
+  return 0;
+}
+
+int tdbDbNext(STDBC *pDbc, void **ppKey, int *kLen, void **ppVal, int *vLen) {
+  return tdbBtreeNext(&pDbc->btc, ppKey, kLen, ppVal, vLen);
+}
+
+int tdbDbcClose(STDBC *pDbc) {
+  if (pDbc) {
+    free(pDbc);
+  }
+
   return 0;
 }

source/libs/tdb/src/db/tdbEnv.c

@@ -27,7 +27,7 @@ int tdbEnvOpen(const char *rootDir, int pageSize, int cacheSize, STEnv **ppEnv)
   dsize = strlen(rootDir);
   zsize = sizeof(*pEnv) + dsize * 2 + strlen(TDB_JOURNAL_NAME) + 3;
 
-  pPtr = (uint8_t *)taosMemoryCalloc(1, zsize);
+  pPtr = (uint8_t *)calloc(1, zsize);
   if (pPtr == NULL) {
     return -1;
   }

source/libs/tdb/src/db/tdbPCache.c

@@ -17,7 +17,7 @@
 struct SPCache {
   int             pageSize;
   int             cacheSize;
-  TdThreadMutex mutex;
+  pthread_mutex_t mutex;
   int             nFree;
   SPage          *pFree;
   int             nPage;
@@ -53,19 +53,17 @@ static void   tdbPCacheLock(SPCache *pCache);
 static void   tdbPCacheUnlock(SPCache *pCache);
 static bool   tdbPCacheLocked(SPCache *pCache);
 static SPage *tdbPCacheFetchImpl(SPCache *pCache, const SPgid *pPgid, bool alcNewPage);
-static void   tdbPCachePinPage(SPage *pPage);
+static void   tdbPCachePinPage(SPCache *pCache, SPage *pPage);
-static void   tdbPCacheRemovePageFromHash(SPage *pPage);
+static void   tdbPCacheRemovePageFromHash(SPCache *pCache, SPage *pPage);
-static void   tdbPCacheAddPageToHash(SPage *pPage);
+static void   tdbPCacheAddPageToHash(SPCache *pCache, SPage *pPage);
-static void   tdbPCacheUnpinPage(SPage *pPage);
+static void   tdbPCacheUnpinPage(SPCache *pCache, SPage *pPage);
-static void  *tdbOsMalloc(void *arg, size_t size);
-static void   tdbOsFree(void *arg, void *ptr);
 
 int tdbPCacheOpen(int pageSize, int cacheSize, SPCache **ppCache) {
   SPCache *pCache;
   void    *pPtr;
   SPage   *pPgHdr;
 
-  pCache = (SPCache *)taosMemoryCalloc(1, sizeof(*pCache));
+  pCache = (SPCache *)calloc(1, sizeof(*pCache));
   if (pCache == NULL) {
     return -1;
   }
@@ -74,7 +72,7 @@ int tdbPCacheOpen(int pageSize, int cacheSize, SPCache **ppCache) {
   pCache->cacheSize = cacheSize;
 
   if (tdbPCacheOpenImpl(pCache) < 0) {
-    taosMemoryFree(pCache);
+    free(pCache);
     return -1;
   }
 
@@ -102,7 +100,7 @@ SPage *tdbPCacheFetch(SPCache *pCache, const SPgid *pPgid, bool alcNewPage) {
   return pPage;
 }
 
-void tdbPCacheRelease(SPage *pPage) {
+void tdbPCacheRelease(SPCache *pCache, SPage *pPage) {
   i32 nRef;
 
   nRef = TDB_UNREF_PAGE(pPage);
@@ -110,7 +108,7 @@ void tdbPCacheRelease(SPage *pPage) {
 
   if (nRef == 0) {
     if (1 /*TODO: page still clean*/) {
-      tdbPCacheUnpinPage(pPage);
+      tdbPCacheUnpinPage(pCache, pPage);
     } else {
       // TODO
       ASSERT(0);
@@ -118,13 +116,13 @@ void tdbPCacheRelease(SPage *pPage) {
   }
 }
 
-static void tdbPCacheInitLock(SPCache *pCache) { taosThreadMutexInit(&(pCache->mutex), NULL); }
+static void tdbPCacheInitLock(SPCache *pCache) { pthread_mutex_init(&(pCache->mutex), NULL); }
 
-static void tdbPCacheClearLock(SPCache *pCache) { taosThreadMutexDestroy(&(pCache->mutex)); }
+static void tdbPCacheClearLock(SPCache *pCache) { pthread_mutex_destroy(&(pCache->mutex)); }
 
-static void tdbPCacheLock(SPCache *pCache) { taosThreadMutexLock(&(pCache->mutex)); }
+static void tdbPCacheLock(SPCache *pCache) { pthread_mutex_lock(&(pCache->mutex)); }
 
-static void tdbPCacheUnlock(SPCache *pCache) { taosThreadMutexUnlock(&(pCache->mutex)); }
+static void tdbPCacheUnlock(SPCache *pCache) { pthread_mutex_unlock(&(pCache->mutex)); }
 
 static bool tdbPCacheLocked(SPCache *pCache) {
   assert(0);
@@ -144,7 +142,7 @@ static SPage *tdbPCacheFetchImpl(SPCache *pCache, const SPgid *pPgid, bool alcNe
 
   if (pPage || !alcNewPage) {
     if (pPage) {
-      tdbPCachePinPage(pPage);
+      tdbPCachePinPage(pCache, pPage);
     }
     return pPage;
   }
@@ -160,8 +158,8 @@ static SPage *tdbPCacheFetchImpl(SPCache *pCache, const SPgid *pPgid, bool alcNe
   // 3. Try to Recycle a page
   if (!pPage && !pCache->lru.pLruPrev->isAnchor) {
     pPage = pCache->lru.pLruPrev;
-    tdbPCacheRemovePageFromHash(pPage);
+    tdbPCacheRemovePageFromHash(pCache, pPage);
-    tdbPCachePinPage(pPage);
+    tdbPCachePinPage(pCache, pPage);
   }
 
   // 4. Try a stress allocation (TODO)
@@ -173,16 +171,13 @@ static SPage *tdbPCacheFetchImpl(SPCache *pCache, const SPgid *pPgid, bool alcNe
     memcpy(&(pPage->pgid), pPgid, sizeof(*pPgid));
     pPage->pLruNext = NULL;
     pPage->pPager = NULL;
-    tdbPCacheAddPageToHash(pPage);
+    tdbPCacheAddPageToHash(pCache, pPage);
   }
 
   return pPage;
 }
 
-static void tdbPCachePinPage(SPage *pPage) {
+static void tdbPCachePinPage(SPCache *pCache, SPage *pPage) {
-  SPCache *pCache;
-
-  pCache = pPage->pCache;
   if (!PAGE_IS_PINNED(pPage)) {
     pPage->pLruPrev->pLruNext = pPage->pLruNext;
     pPage->pLruNext->pLruPrev = pPage->pLruPrev;
@@ -192,11 +187,8 @@ static void tdbPCachePinPage(SPage *pPage) {
   }
 }
 
-static void tdbPCacheUnpinPage(SPage *pPage) {
+static void tdbPCacheUnpinPage(SPCache *pCache, SPage *pPage) {
-  SPCache *pCache;
+  i32 nRef;
-  i32      nRef;
-
-  pCache = pPage->pCache;
 
   tdbPCacheLock(pCache);
 
@@ -217,12 +209,10 @@ static void tdbPCacheUnpinPage(SPage *pPage) {
   tdbPCacheUnlock(pCache);
 }
 
-static void tdbPCacheRemovePageFromHash(SPage *pPage) {
+static void tdbPCacheRemovePageFromHash(SPCache *pCache, SPage *pPage) {
-  SPCache *pCache;
+  SPage **ppPage;
-  SPage  **ppPage;
+  int     h;
-  int      h;
 
-  pCache = pPage->pCache;
   h = PCACHE_PAGE_HASH(&(pPage->pgid));
   for (ppPage = &(pCache->pgHash[h % pCache->nHash]); *ppPage != pPage; ppPage = &((*ppPage)->pHashNext))
     ;
@@ -232,11 +222,9 @@ static void tdbPCacheRemovePageFromHash(SPage *pPage) {
   pCache->nPage--;
 }
 
-static void tdbPCacheAddPageToHash(SPage *pPage) {
+static void tdbPCacheAddPageToHash(SPCache *pCache, SPage *pPage) {
-  SPCache *pCache;
+  int h;
-  int      h;
 
-  pCache = pPage->pCache;
   h = PCACHE_PAGE_HASH(&(pPage->pgid)) % pCache->nHash;
 
   pPage->pHashNext = pCache->pgHash[h];
@@ -257,7 +245,7 @@ static int tdbPCacheOpenImpl(SPCache *pCache) {
   pCache->nFree = 0;
   pCache->pFree = NULL;
   for (int i = 0; i < pCache->cacheSize; i++) {
-    ret = tdbPageCreate(pCache->pageSize, &pPage, tdbOsMalloc, NULL);
+    ret = tdbPageCreate(pCache->pageSize, &pPage, NULL, NULL);
     if (ret < 0) {
       // TODO: handle error
       return -1;
@@ -266,7 +254,6 @@ static int tdbPCacheOpenImpl(SPCache *pCache) {
     // pPage->pgid = 0;
     pPage->isAnchor = 0;
     pPage->isLocalPage = 1;
-    pPage->pCache = pCache;
     TDB_INIT_PAGE_REF(pPage);
     pPage->pHashNext = NULL;
     pPage->pLruNext = NULL;
@@ -281,7 +268,7 @@ static int tdbPCacheOpenImpl(SPCache *pCache) {
   // Open the hash table
   pCache->nPage = 0;
   pCache->nHash = pCache->cacheSize;
-  pCache->pgHash = (SPage **)taosMemoryCalloc(pCache->nHash, sizeof(SPage *));
+  pCache->pgHash = (SPage **)calloc(pCache->nHash, sizeof(SPage *));
   if (pCache->pgHash == NULL) {
     // TODO
     return -1;
@@ -297,13 +284,3 @@ static int tdbPCacheOpenImpl(SPCache *pCache) {
 }
 
 int tdbPCacheGetPageSize(SPCache *pCache) { return pCache->pageSize; }
-
-static void *tdbOsMalloc(void *arg, size_t size) {
-  void *ptr;
-
-  ptr = taosMemoryMalloc(size);
-
-  return ptr;
-}
-
-static void tdbOsFree(void *arg, void *ptr) { taosMemoryFree(ptr); }

source/libs/tdb/src/db/tdbPage.c

@@ -1,253 +0,0 @@
-/*
- * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
- *
- * This program is free software: you can use, redistribute, and/or modify
- * it under the terms of the GNU Affero General Public License, version 3
- * or later ("AGPL"), as published by the Free Software Foundation.
- *
- * This program is distributed in the hope that it will be useful, but WITHOUT
- * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
- * FITNESS FOR A PARTICULAR PURPOSE.
- *
- * You should have received a copy of the GNU Affero General Public License
- * along with this program. If not, see <http://www.gnu.org/licenses/>.
- */
-
-#include "tdbInt.h"
-
-typedef struct __attribute__((__packed__)) {
-  u8 szCell[2];
-  u8 nxOffset[2];
-} SFreeCell;
-
-typedef struct __attribute__((__packed__)) {
-  u8 szCell[3];
-  u8 nxOffset[3];
-} SFreeCellL;
-
-/* For small page */
-#define TDB_SPAGE_FREE_CELL_SIZE_PTR(PCELL)     (((SFreeCell *)(PCELL))->szCell)
-#define TDB_SPAGE_FREE_CELL_NXOFFSET_PTR(PCELL) (((SFreeCell *)(PCELL))->nxOffset)
-
-#define TDB_SPAGE_FREE_CELL_SIZE(PCELL)     ((u16 *)TDB_SPAGE_FREE_CELL_SIZE_PTR(PCELL))[0]
-#define TDB_SPAGE_FREE_CELL_NXOFFSET(PCELL) ((u16 *)TDB_SPAGE_FREE_CELL_NXOFFSET_PTR(PCELL))[0]
-
-#define TDB_SPAGE_FREE_CELL_SIZE_SET(PCELL, SIZE)       (TDB_SPAGE_FREE_CELL_SIZE(PCELL) = (SIZE))
-#define TDB_SPAGE_FREE_CELL_NXOFFSET_SET(PCELL, OFFSET) (TDB_SPAGE_FREE_CELL_NXOFFSET(PCELL) = (OFFSET))
-
-/* For large page */
-#define TDB_LPAGE_FREE_CELL_SIZE_PTR(PCELL)     (((SFreeCellL *)(PCELL))->szCell)
-#define TDB_LPAGE_FREE_CELL_NXOFFSET_PTR(PCELL) (((SFreeCellL *)(PCELL))->nxOffset)
-
-#define TDB_LPAGE_FREE_CELL_SIZE(PCELL)     TDB_GET_U24(TDB_LPAGE_FREE_CELL_SIZE_PTR(PCELL))
-#define TDB_LPAGE_FREE_CELL_NXOFFSET(PCELL) TDB_GET_U24(TDB_LPAGE_FREE_CELL_NXOFFSET_PTR(PCELL))
-
-#define TDB_LPAGE_FREE_CELL_SIZE_SET(PCELL, SIZE)       TDB_PUT_U24(TDB_LPAGE_FREE_CELL_SIZE_PTR(PCELL), SIZE)
-#define TDB_LPAGE_FREE_CELL_NXOFFSET_SET(PCELL, OFFSET) TDB_PUT_U24(TDB_LPAGE_FREE_CELL_NXOFFSET_PTR(PCELL), OFFSET)
-
-/* For page */
-#define TDB_PAGE_FREE_CELL_SIZE_PTR(PPAGE, PCELL) \
-  (TDB_IS_LARGE_PAGE(pPage) ? TDB_LPAGE_FREE_CELL_SIZE_PTR(PCELL) : TDB_SPAGE_FREE_CELL_SIZE_PTR(PCELL))
-#define TDB_PAGE_FREE_CELL_NXOFFSET_PTR(PPAGE, PCELL) \
-  (TDB_IS_LARGE_PAGE(pPage) ? TDB_LPAGE_FREE_CELL_NXOFFSET_PTR(PCELL) : TDB_SPAGE_FREE_CELL_NXOFFSET_PTR(PCELL))
-
-#define TDB_PAGE_FREE_CELL_SIZE(PPAGE, PCELL) \
-  (TDB_IS_LARGE_PAGE(pPage) ? TDB_LPAGE_FREE_CELL_SIZE(PCELL) : TDB_SPAGE_FREE_CELL_SIZE(PCELL))
-#define TDB_PAGE_FREE_CELL_NXOFFSET(PPAGE, PCELL) \
-  (TDB_IS_LARGE_PAGE(pPage) ? TDB_LPAGE_FREE_CELL_NXOFFSET(PCELL) : TDB_SPAGE_FREE_CELL_NXOFFSET(PCELL))
-
-#define TDB_PAGE_FREE_CELL_SIZE_SET(PPAGE, PCELL, SIZE) \
-  do {                                                  \
-    if (TDB_IS_LARGE_PAGE(PPAGE)) {                     \
-      TDB_LPAGE_FREE_CELL_SIZE_SET(PCELL, SIZE);        \
-    } else {                                            \
-      TDB_SPAGE_FREE_CELL_SIZE_SET(PCELL, SIZE);        \
-    }                                                   \
-  } while (0)
-#define TDB_PAGE_FREE_CELL_NXOFFSET_SET(PPAGE, PCELL, OFFSET) \
-  do {                                                        \
-    if (TDB_IS_LARGE_PAGE(PPAGE)) {                           \
-      TDB_LPAGE_FREE_CELL_NXOFFSET_SET(PCELL, OFFSET);        \
-    } else {                                                  \
-      TDB_SPAGE_FREE_CELL_NXOFFSET_SET(PCELL, OFFSET);        \
-    }                                                         \
-  } while (0)
-
-static int tdbPageAllocate(SPage *pPage, int size, SCell **ppCell);
-static int tdbPageDefragment(SPage *pPage);
-
-int tdbPageCreate(int pageSize, SPage **ppPage, void *(*xMalloc)(void *, size_t), void *arg) {
-  SPage *pPage;
-  u8    *ptr;
-  int    size;
-
-  ASSERT(TDB_IS_PGSIZE_VLD(pageSize));
-
-  *ppPage = NULL;
-  size = pageSize + sizeof(*pPage);
-
-  ptr = (u8 *)((*xMalloc)(arg, size));
-  if (pPage == NULL) {
-    return -1;
-  }
-
-  memset(ptr, 0, size);
-  pPage = (SPage *)(ptr + pageSize);
-
-  pPage->pData = ptr;
-  pPage->pageSize = pageSize;
-  if (pageSize < 65536) {
-    pPage->szOffset = 2;
-    pPage->szPageHdr = sizeof(SPageHdr);
-    pPage->szFreeCell = sizeof(SFreeCell);
-  } else {
-    pPage->szOffset = 3;
-    pPage->szPageHdr = sizeof(SPageHdrL);
-    pPage->szFreeCell = sizeof(SFreeCellL);
-  }
-  TDB_INIT_PAGE_LOCK(pPage);
-
-  /* TODO */
-
-  *ppPage = pPage;
-  return 0;
-}
-
-int tdbPageDestroy(SPage *pPage, void (*xFree)(void *arg, void *ptr), void *arg) {
-  u8 *ptr;
-
-  ptr = pPage->pData;
-  (*xFree)(arg, ptr);
-
-  return 0;
-}
-
-int tdbPageInsertCell(SPage *pPage, int idx, SCell *pCell, int szCell) {
-  int    ret;
-  SCell *pTarget;
-  u8    *pTmp;
-  int    j;
-
-  if (pPage->nOverflow || szCell + pPage->szOffset > pPage->nFree) {
-    // TODO: need to figure out if pCell may be used by outside of this function
-    j = pPage->nOverflow++;
-
-    pPage->apOvfl[j] = pCell;
-    pPage->aiOvfl[j] = idx;
-  } else {
-    ret = tdbPageAllocate(pPage, szCell, &pTarget);
-    if (ret < 0) {
-      return -1;
-    }
-
-    memcpy(pTarget, pCell, szCell);
-    pTmp = pPage->pCellIdx + idx * pPage->szOffset;
-    memmove(pTmp + pPage->szOffset, pTmp, pPage->pFreeStart - pTmp - pPage->szOffset);
-    TDB_PAGE_CELL_OFFSET_AT_SET(pPage, idx, pTarget - pPage->pData);
-    TDB_PAGE_NCELLS_SET(pPage, TDB_PAGE_NCELLS(pPage) + 1);
-  }
-
-  return 0;
-}
-
-int tdbPageDropCell(SPage *pPage, int idx) {
-  // TODO
-  return 0;
-}
-
-static int tdbPageAllocate(SPage *pPage, int size, SCell **ppCell) {
-  SCell     *pCell;
-  SFreeCell *pFreeCell;
-  u8        *pOffset;
-  int        ret;
-
-  ASSERT(pPage->nFree > size + pPage->szOffset);
-
-  pCell = NULL;
-  *ppCell = NULL;
-
-  // 1. Try to allocate from the free space area
-  if (pPage->pFreeEnd - pPage->pFreeStart > size + pPage->szOffset) {
-    pPage->pFreeEnd -= size;
-    pPage->pFreeStart += pPage->szOffset;
-    pCell = pPage->pFreeEnd;
-  }
-
-  // 2. Try to allocate from the page free list
-  if ((pCell == NULL) && (pPage->pFreeEnd - pPage->pFreeStart >= pPage->szOffset) && TDB_PAGE_FCELL(pPage)) {
-    int szCell;
-    int nxOffset;
-
-    pCell = pPage->pData + TDB_PAGE_FCELL(pPage);
-    pOffset = TDB_IS_LARGE_PAGE(pPage) ? ((SPageHdrL *)(pPage->pPageHdr))[0].fCell
-                                       : (u8 *)&(((SPageHdr *)(pPage->pPageHdr))[0].fCell);
-    szCell = TDB_PAGE_FREE_CELL_SIZE(pPage, pCell);
-    nxOffset = TDB_PAGE_FREE_CELL_NXOFFSET(pPage, pCell);
-
-    for (;;) {
-      // Find a cell
-      if (szCell >= size) {
-        if (szCell - size >= pPage->szFreeCell) {
-          SCell *pTmpCell = pCell + size;
-
-          TDB_PAGE_FREE_CELL_SIZE_SET(pPage, pTmpCell, szCell - size);
-          TDB_PAGE_FREE_CELL_NXOFFSET_SET(pPage, pTmpCell, nxOffset);
-          // TODO: *pOffset = pTmpCell - pPage->pData;
-        } else {
-          TDB_PAGE_NFREE_SET(pPage, TDB_PAGE_NFREE(pPage) + szCell - size);
-          // TODO: *pOffset = nxOffset;
-        }
-        break;
-      }
-
-      // Not find a cell yet
-      if (nxOffset > 0) {
-        pCell = pPage->pData + nxOffset;
-        pOffset = TDB_PAGE_FREE_CELL_NXOFFSET_PTR(pPage, pCell);
-        szCell = TDB_PAGE_FREE_CELL_SIZE(pPage, pCell);
-        nxOffset = TDB_PAGE_FREE_CELL_NXOFFSET(pPage, pCell);
-        continue;
-      } else {
-        pCell = NULL;
-        break;
-      }
-    }
-
-    if (pCell) {
-      pPage->pFreeStart = pPage->pFreeStart + pPage->szOffset;
-    }
-  }
-
-  // 3. Try to dfragment and allocate again
-  if (pCell == NULL) {
-    ret = tdbPageDefragment(pPage);
-    if (ret < 0) {
-      return -1;
-    }
-
-    ASSERT(pPage->pFreeEnd - pPage->pFreeStart > size + pPage->szOffset);
-    ASSERT(pPage->nFree == pPage->pFreeEnd - pPage->pFreeStart);
-
-    // Allocate from the free space area again
-    pPage->pFreeEnd -= size;
-    pPage->pFreeStart += pPage->szOffset;
-    pCell = pPage->pFreeEnd;
-  }
-
-  ASSERT(pCell != NULL);
-
-  pPage->nFree = pPage->nFree - size - pPage->szOffset;
-  *ppCell = pCell;
-  return 0;
-}
-
-static int tdbPageFree(SPage *pPage, int idx, SCell *pCell, int size) {
-  // TODO
-  return 0;
-}
-
-static int tdbPageDefragment(SPage *pPage) {
-  // TODO
-  ASSERT(0);
-  return 0;
-}

source/libs/tdb/src/db/tdbPager.c

@@ -60,7 +60,7 @@ int tdbPagerOpen(SPCache *pCache, const char *fileName, SPager **ppPager) {
   zsize = sizeof(*pPager)  /* SPager */
           + fsize + 1      /* dbFileName */
           + fsize + 8 + 1; /* jFileName */
-  pPtr = (uint8_t *)taosMemoryCalloc(1, zsize);
+  pPtr = (uint8_t *)calloc(1, zsize);
   if (pPtr == NULL) {
     return -1;
   }
@@ -255,6 +255,10 @@ int tdbPagerNewPage(SPager *pPager, SPgno *ppgno, SPage **ppPage, int (*initPage
   return 0;
 }
 
+void tdbPagerReturnPage(SPager *pPager, SPage *pPage) {
+  tdbPCacheRelease(pPager->pCache, pPage);
+}
+
 static int tdbPagerAllocFreePage(SPager *pPager, SPgno *ppgno) {
   // TODO: Allocate a page from the free list
   return 0;

source/libs/tdb/src/inc/tdbBtree.h

@@ -20,10 +20,15 @@
 extern "C" {
 #endif
 
-typedef struct SBTree    SBTree;
+typedef struct SBTree SBTree;
-typedef struct SBtCursor SBtCursor;
+typedef struct SBTC   SBTC;
+typedef struct SBtInfo {
+  SPgno root;
+  int   nLevel;
+  int   nData;
+} SBtInfo;
 
-struct SBtCursor {
+struct SBTC {
   SBTree *pBt;
   i8      iPage;
   SPage  *pPage;
@@ -33,10 +38,19 @@ struct SBtCursor {
   void   *pBuf;
 };
 
+// SBTree
 int tdbBtreeOpen(int keyLen, int valLen, SPager *pFile, FKeyComparator kcmpr, SBTree **ppBt);
 int tdbBtreeClose(SBTree *pBt);
-int tdbBtreeCursor(SBtCursor *pCur, SBTree *pBt);
+int tdbBtCursorInsert(SBTC *pCur, const void *pKey, int kLen, const void *pVal, int vLen);
-int tdbBtCursorInsert(SBtCursor *pCur, const void *pKey, int kLen, const void *pVal, int vLen);
+int tdbBtreeGet(SBTree *pBt, const void *pKey, int kLen, void **ppVal, int *vLen);
+
+// SBTC
+int tdbBtcOpen(SBTC *pCur, SBTree *pBt);
+int tdbBtcMoveToFirst(SBTC *pBtc);
+int tdbBtcMoveToLast(SBTC *pBtc);
+int tdbBtcMoveTo(SBTC *pBtc, const void *pKey, int kLen);
+int tdbBtreeNext(SBTC *pBtc, void **ppKey, int *kLen, void **ppVal, int *vLen);
+int tdbBtcClose(SBTC *pBtc);
 
 #ifdef __cplusplus
 }

source/libs/tdb/src/inc/tdbDb.h

@@ -20,12 +20,20 @@
 extern "C" {
 #endif
 
-typedef struct STDb STDb;
+typedef struct STDB  STDB;
+typedef struct STDBC STDBC;
 
-int tdbDbOpen(const char *fname, int keyLen, int valLen, FKeyComparator keyCmprFn, STEnv *pEnv, STDb **ppDb);
+// STDB
-int tdbDbClose(STDb *pDb);
+int tdbDbOpen(const char *fname, int keyLen, int valLen, FKeyComparator keyCmprFn, STEnv *pEnv, STDB **ppDb);
-int tdbDbDrop(STDb *pDb);
+int tdbDbClose(STDB *pDb);
-int tdbDbInsert(STDb *pDb, const void *pKey, int keyLen, const void *pVal, int valLen);
+int tdbDbDrop(STDB *pDb);
+int tdbDbInsert(STDB *pDb, const void *pKey, int keyLen, const void *pVal, int valLen);
+int tdbDbGet(STDB *pDb, const void *pKey, int kLen, void **ppVal, int *vLen);
+
+// STDBC
+int tdbDbcOpen(STDB *pDb, STDBC **ppDbc);
+int tdbDbNext(STDBC *pDbc, void **ppKey, int *kLen, void **ppVal, int *vLen);
+int tdbDbcClose(STDBC *pDbc);
 
 #ifdef __cplusplus
 }

source/libs/tdb/src/inc/tdbInt.h

@@ -95,7 +95,7 @@ static FORCE_INLINE int tdbCmprPgId(const void *p1, const void *p2) {
 // tdb_log
 #define tdbError(var)
 
-typedef TD_DLIST(STDb) STDbList;
+typedef TD_DLIST(STDB) STDbList;
 typedef TD_DLIST(SPgFile) SPgFileList;
 typedef TD_DLIST_NODE(SPgFile) SPgFileListNode;
 
@@ -141,8 +141,8 @@ typedef int (*FKeyComparator)(const void *pKey1, int kLen1, const void *pKey2, i
 #define TDB_FLAG_IS(flags, flag)     ((flags) == (flag))
 #define TDB_FLAG_HAS(flags, flag)    (((flags) & (flag)) != 0)
 #define TDB_FLAG_NO(flags, flag)     ((flags) & (flag) == 0)
-#define TDB_FLAG_ADD(flags, flag)    ((flags) |= (flag))
+#define TDB_FLAG_ADD(flags, flag)    ((flags) | (flag))
-#define TDB_FLAG_REMOVE(flags, flag) ((flags) &= (~(flag)))
+#define TDB_FLAG_REMOVE(flags, flag) ((flags) & (~(flag)))
 
 typedef struct SPager  SPager;
 typedef struct SPCache SPCache;

source/libs/tdb/src/inc/tdbPCache.h

@@ -21,23 +21,22 @@ extern "C" {
 #endif
 
 #define TDB_PCACHE_PAGE \
-  u8       isAnchor;    \
+  u8      isAnchor;     \
-  u8       isLocalPage; \
+  u8      isLocalPage;  \
-  u8       isDirty;     \
+  u8      isDirty;      \
-  i32      nRef;        \
+  i32     nRef;         \
-  SPCache *pCache;      \
+  SPage  *pFreeNext;    \
-  SPage   *pFreeNext;   \
+  SPage  *pHashNext;    \
-  SPage   *pHashNext;   \
+  SPage  *pLruNext;     \
-  SPage   *pLruNext;    \
+  SPage  *pLruPrev;     \
-  SPage   *pLruPrev;    \
+  SPage  *pDirtyNext;   \
-  SPage   *pDirtyNext;  \
+  SPager *pPager;       \
-  SPager  *pPager;      \
+  SPgid   pgid;
-  SPgid    pgid;
 
 int    tdbPCacheOpen(int pageSize, int cacheSize, SPCache **ppCache);
 int    tdbPCacheClose(SPCache *pCache);
 SPage *tdbPCacheFetch(SPCache *pCache, const SPgid *pPgid, bool alcNewPage);
-void   tdbPCacheRelease(SPage *pPage);
+void   tdbPCacheRelease(SPCache *pCache, SPage *pPage);
 int    tdbPCacheGetPageSize(SPCache *pCache);
 
 #ifdef __cplusplus

source/libs/tdb/src/inc/tdbPage.h

@@ -27,9 +27,6 @@ typedef struct {
   int szOffset;
   int szPageHdr;
   int szFreeCell;
-  // flags
-  u16 (*getFlags)(SPage *);
-  void (*setFlags)(SPage *, u16);
   // cell number
   int (*getCellNum)(SPage *);
   void (*setCellNum)(SPage *, int);
@@ -45,6 +42,9 @@ typedef struct {
   // cell offset at idx
   int (*getCellOffset)(SPage *, int);
   void (*setCellOffset)(SPage *, int, int);
+  // free cell info
+  void (*getFreeCellInfo)(SCell *pCell, int *szCell, int *nxOffset);
+  void (*setFreeCellInfo)(SCell *pCell, int szCell, int nxOffset);
 } SPageMethods;
 
 // Page footer
@@ -53,58 +53,37 @@ typedef struct __attribute__((__packed__)) {
 } SPageFtr;
 
 struct SPage {
-  TdThreadSpinlock lock;
+  pthread_spinlock_t lock;
-  u8                *pData;
   int                pageSize;
+  u8                *pData;
   SPageMethods      *pPageMethods;
   // Fields below used by pager and am
-  u8        szAmHdr;
   u8       *pPageHdr;
-  u8       *pAmHdr;
   u8       *pCellIdx;
   u8       *pFreeStart;
   u8       *pFreeEnd;
   SPageFtr *pPageFtr;
-  int       kLen;  // key length of the page, -1 for unknown
-  int       vLen;  // value length of the page, -1 for unknown
-  int       nFree;
-  int       maxLocal;
-  int       minLocal;
   int       nOverflow;
   SCell    *apOvfl[4];
   int       aiOvfl[4];
+  int       kLen;  // key length of the page, -1 for unknown
+  int       vLen;  // value length of the page, -1 for unknown
+  int       maxLocal;
+  int       minLocal;
+  int (*xCellSize)(const SPage *, SCell *);
   // Fields used by SPCache
   TDB_PCACHE_PAGE
 };
 
-/* For page */
-#define TDB_PAGE_FLAGS(pPage)               (*(pPage)->pPageMethods->getFlags)(pPage)
-#define TDB_PAGE_NCELLS(pPage)              (*(pPage)->pPageMethods->getCellNum)(pPage)
-#define TDB_PAGE_CCELLS(pPage)              (*(pPage)->pPageMethods->getCellBody)(pPage)
-#define TDB_PAGE_FCELL(pPage)               (*(pPage)->pPageMethods->getCellFree)(pPage)
-#define TDB_PAGE_NFREE(pPage)               (*(pPage)->pPageMethods->getFreeBytes)(pPage)
-#define TDB_PAGE_CELL_OFFSET_AT(pPage, idx) (*(pPage)->pPageMethods->getCellOffset)(pPage, idx)
-
-#define TDB_PAGE_FLAGS_SET(pPage, FLAGS)                (*(pPage)->pPageMethods->setFlags)(pPage, FLAGS)
-#define TDB_PAGE_NCELLS_SET(pPage, NCELLS)              (*(pPage)->pPageMethods->setCellNum)(pPage, NCELLS)
-#define TDB_PAGE_CCELLS_SET(pPage, CCELLS)              (*(pPage)->pPageMethods->setCellBody)(pPage, CCELLS)
-#define TDB_PAGE_FCELL_SET(pPage, FCELL)                (*(pPage)->pPageMethods->setCellFree)(pPage, FCELL)
-#define TDB_PAGE_NFREE_SET(pPage, NFREE)                (*(pPage)->pPageMethods->setFreeBytes)(pPage, NFREE)
-#define TDB_PAGE_CELL_OFFSET_AT_SET(pPage, idx, OFFSET) (*(pPage)->pPageMethods->setCellOffset)(pPage, idx, OFFSET)
-
-#define TDB_PAGE_OFFSET_SIZE(pPage) ((pPage)->pPageMethods->szOffset)
-
-#define TDB_PAGE_CELL_AT(pPage, idx) ((pPage)->pData + TDB_PAGE_CELL_OFFSET_AT(pPage, idx))
-
 // For page lock
 #define P_LOCK_SUCC 0
 #define P_LOCK_BUSY 1
 #define P_LOCK_FAIL -1
 
-#define TDB_INIT_PAGE_LOCK(pPage)    taosThreadSpinInit(&((pPage)->lock), 0)
+#define TDB_INIT_PAGE_LOCK(pPage)    pthread_spin_init(&((pPage)->lock), 0)
-#define TDB_DESTROY_PAGE_LOCK(pPage) taosThreadSpinDestroy(&((pPage)->lock))
+#define TDB_DESTROY_PAGE_LOCK(pPage) pthread_spin_destroy(&((pPage)->lock))
-#define TDB_LOCK_PAGE(pPage)         taosThreadSpinLock(&((pPage)->lock))
+#define TDB_LOCK_PAGE(pPage)         pthread_spin_lock(&((pPage)->lock))
-#define TDB_UNLOCK_PAGE(pPage)       taosThreadSpinUnlock(&((pPage)->lock))
+#define TDB_UNLOCK_PAGE(pPage)       pthread_spin_unlock(&((pPage)->lock))
 #define TDB_TRY_LOCK_PAGE(pPage)                       \
   ({                                                   \
     int ret;                                           \
@@ -119,10 +98,43 @@ struct SPage {
   })
 
 // APIs
-int tdbPageCreate(int pageSize, SPage **ppPage, void *(*xMalloc)(void *, size_t), void *arg);
+#define TDB_PAGE_TOTAL_CELLS(pPage)        ((pPage)->nOverflow + (pPage)->pPageMethods->getCellNum(pPage))
-int tdbPageDestroy(SPage *pPage, void (*xFree)(void *arg, void *ptr), void *arg);
+#define TDB_PAGE_USABLE_SIZE(pPage)        ((u8 *)(pPage)->pPageFtr - (pPage)->pCellIdx)
-int tdbPageInsertCell(SPage *pPage, int idx, SCell *pCell, int szCell);
+#define TDB_PAGE_PGNO(pPage)               ((pPage)->pgid.pgno)
-int tdbPageDropCell(SPage *pPage, int idx);
+#define TDB_BYTES_CELL_TAKEN(pPage, pCell) ((*(pPage)->xCellSize)(pPage, pCell) + (pPage)->pPageMethods->szOffset)
+#define TDB_PAGE_OFFSET_SIZE(pPage)        ((pPage)->pPageMethods->szOffset)
+
+int  tdbPageCreate(int pageSize, SPage **ppPage, void *(*xMalloc)(void *, size_t), void *arg);
+int  tdbPageDestroy(SPage *pPage, void (*xFree)(void *arg, void *ptr), void *arg);
+void tdbPageZero(SPage *pPage, u8 szAmHdr, int (*xCellSize)(const SPage *, SCell *));
+void tdbPageInit(SPage *pPage, u8 szAmHdr, int (*xCellSize)(const SPage *, SCell *));
+int  tdbPageInsertCell(SPage *pPage, int idx, SCell *pCell, int szCell, u8 asOvfl);
+int  tdbPageDropCell(SPage *pPage, int idx);
+void tdbPageCopy(SPage *pFromPage, SPage *pToPage);
+
+static inline SCell *tdbPageGetCell(SPage *pPage, int idx) {
+  SCell *pCell;
+  int    iOvfl;
+  int    lidx;
+
+  ASSERT(idx >= 0 && idx < TDB_PAGE_TOTAL_CELLS(pPage));
+
+  iOvfl = 0;
+  for (; iOvfl < pPage->nOverflow; iOvfl++) {
+    if (pPage->aiOvfl[iOvfl] == idx) {
+      pCell = pPage->apOvfl[iOvfl];
+      return pCell;
+    } else if (pPage->aiOvfl[iOvfl] > idx) {
+      break;
+    }
+  }
+
+  lidx = idx - iOvfl;
+  ASSERT(lidx >= 0 && lidx < pPage->pPageMethods->getCellNum(pPage));
+  pCell = pPage->pData + pPage->pPageMethods->getCellOffset(pPage, lidx);
+
+  return pCell;
+}
 
 #ifdef __cplusplus
 }

source/libs/tdb/src/inc/tdbPager.h

@@ -20,15 +20,16 @@
 extern "C" {
 #endif
 
-int    tdbPagerOpen(SPCache *pCache, const char *fileName, SPager **ppPager);
+int  tdbPagerOpen(SPCache *pCache, const char *fileName, SPager **ppPager);
-int    tdbPagerClose(SPager *pPager);
+int  tdbPagerClose(SPager *pPager);
-int    tdbPagerOpenDB(SPager *pPager, SPgno *ppgno, bool toCreate);
+int  tdbPagerOpenDB(SPager *pPager, SPgno *ppgno, bool toCreate);
-int    tdbPagerWrite(SPager *pPager, SPage *pPage);
+int  tdbPagerWrite(SPager *pPager, SPage *pPage);
-int    tdbPagerBegin(SPager *pPager);
+int  tdbPagerBegin(SPager *pPager);
-int    tdbPagerCommit(SPager *pPager);
+int  tdbPagerCommit(SPager *pPager);
-int    tdbPagerGetPageSize(SPager *pPager);
+int  tdbPagerGetPageSize(SPager *pPager);
-int    tdbPagerFetchPage(SPager *pPager, SPgno pgno, SPage **ppPage, int (*initPage)(SPage *, void *), void *arg);
+int  tdbPagerFetchPage(SPager *pPager, SPgno pgno, SPage **ppPage, int (*initPage)(SPage *, void *), void *arg);
-int    tdbPagerNewPage(SPager *pPager, SPgno *ppgno, SPage **ppPage, int (*initPage)(SPage *, void *), void *arg);
+int  tdbPagerNewPage(SPager *pPager, SPgno *ppgno, SPage **ppPage, int (*initPage)(SPage *, void *), void *arg);
+void tdbPagerReturnPage(SPager *pPager, SPage *pPage);
 
 #ifdef __cplusplus
 }

source/libs/tdb/src/inc/tdbUtil.h

@@ -39,6 +39,38 @@ int tdbGetFileSize(const char *fname, int pgSize, SPgno *pSize);
 
 int tdbPRead(int fd, void *pData, int count, i64 offset);
 
+#define TDB_REALLOC(PTR, SIZE)                                                          \
+  ({                                                                                    \
+    void *nPtr;                                                                         \
+    if ((PTR) == NULL || ((int *)(PTR))[-1] < (SIZE)) {                                 \
+      nPtr = realloc((PTR) ? (char *)(PTR) - sizeof(int) : NULL, (SIZE) + sizeof(int)); \
+      if (nPtr) {                                                                       \
+        ((int *)nPtr)[0] = (SIZE);                                                      \
+        nPtr = (char *)nPtr + sizeof(int);                                              \
+      }                                                                                 \
+    } else {                                                                            \
+      nPtr = (PTR);                                                                     \
+    }                                                                                   \
+    nPtr;                                                                               \
+  })
+
+#define TDB_FREE(PTR)                    \
+  do {                                   \
+    if (PTR) {                           \
+      free((char *)(PTR) - sizeof(int)); \
+    }                                    \
+  } while (0)
+
+static inline void *tdbOsMalloc(void *arg, size_t size) {
+  void *ptr;
+
+  ptr = malloc(size);
+
+  return ptr;
+}
+
+static inline void tdbOsFree(void *arg, void *ptr) { free(ptr); }
+
 static inline int tdbPutVarInt(u8 *p, int v) {
   int n = 0;
 

source/libs/tdb/src/page/tdbPage.c

@@ -18,13 +18,25 @@
 extern SPageMethods pageMethods;
 extern SPageMethods pageLargeMethods;
 
-typedef struct __attribute__((__packed__)) {
+#define TDB_PAGE_HDR_SIZE(pPage)                        ((pPage)->pPageMethods->szPageHdr)
-  u16 szCell;
+#define TDB_PAGE_FREE_CELL_SIZE(pPage)                  ((pPage)->pPageMethods->szFreeCell)
-  u16 nxOffset;
+#define TDB_PAGE_NCELLS(pPage)                          (*(pPage)->pPageMethods->getCellNum)(pPage)
-} SFreeCell;
+#define TDB_PAGE_CCELLS(pPage)                          (*(pPage)->pPageMethods->getCellBody)(pPage)
+#define TDB_PAGE_FCELL(pPage)                           (*(pPage)->pPageMethods->getCellFree)(pPage)
+#define TDB_PAGE_NFREE(pPage)                           (*(pPage)->pPageMethods->getFreeBytes)(pPage)
+#define TDB_PAGE_CELL_OFFSET_AT(pPage, idx)             (*(pPage)->pPageMethods->getCellOffset)(pPage, idx)
+#define TDB_PAGE_NCELLS_SET(pPage, NCELLS)              (*(pPage)->pPageMethods->setCellNum)(pPage, NCELLS)
+#define TDB_PAGE_CCELLS_SET(pPage, CCELLS)              (*(pPage)->pPageMethods->setCellBody)(pPage, CCELLS)
+#define TDB_PAGE_FCELL_SET(pPage, FCELL)                (*(pPage)->pPageMethods->setCellFree)(pPage, FCELL)
+#define TDB_PAGE_NFREE_SET(pPage, NFREE)                (*(pPage)->pPageMethods->setFreeBytes)(pPage, NFREE)
+#define TDB_PAGE_CELL_OFFSET_AT_SET(pPage, idx, OFFSET) (*(pPage)->pPageMethods->setCellOffset)(pPage, idx, OFFSET)
+#define TDB_PAGE_CELL_AT(pPage, idx)                    ((pPage)->pData + TDB_PAGE_CELL_OFFSET_AT(pPage, idx))
+#define TDB_PAGE_MAX_FREE_BLOCK(pPage, szAmHdr) \
+  ((pPage)->pageSize - (szAmHdr)-TDB_PAGE_HDR_SIZE(pPage) - sizeof(SPageFtr))
 
 static int tdbPageAllocate(SPage *pPage, int size, SCell **ppCell);
 static int tdbPageDefragment(SPage *pPage);
+static int tdbPageFree(SPage *pPage, int idx, SCell *pCell, int szCell);
 
 int tdbPageCreate(int pageSize, SPage **ppPage, void *(*xMalloc)(void *, size_t), void *arg) {
   SPage *pPage;
@@ -35,25 +47,26 @@ int tdbPageCreate(int pageSize, SPage **ppPage, void *(*xMalloc)(void *, size_t)
 
   *ppPage = NULL;
   size = pageSize + sizeof(*pPage);
+  if (xMalloc == NULL) {
+    xMalloc = tdbOsMalloc;
+  }
 
   ptr = (u8 *)((*xMalloc)(arg, size));
-  if (pPage == NULL) {
+  if (ptr == NULL) {
     return -1;
   }
 
   memset(ptr, 0, size);
   pPage = (SPage *)(ptr + pageSize);
 
-  pPage->pData = ptr;
+  TDB_INIT_PAGE_LOCK(pPage);
   pPage->pageSize = pageSize;
+  pPage->pData = ptr;
   if (pageSize < 65536) {
     pPage->pPageMethods = &pageMethods;
   } else {
     pPage->pPageMethods = &pageLargeMethods;
   }
-  TDB_INIT_PAGE_LOCK(pPage);
-
-  /* TODO */
 
   *ppPage = pPage;
   return 0;
@@ -62,157 +75,365 @@ int tdbPageCreate(int pageSize, SPage **ppPage, void *(*xMalloc)(void *, size_t)
 int tdbPageDestroy(SPage *pPage, void (*xFree)(void *arg, void *ptr), void *arg) {
   u8 *ptr;
 
+  if (!xFree) {
+    xFree = tdbOsFree;
+  }
+
   ptr = pPage->pData;
   (*xFree)(arg, ptr);
 
   return 0;
 }
 
-int tdbPageInsertCell(SPage *pPage, int idx, SCell *pCell, int szCell) {
+void tdbPageZero(SPage *pPage, u8 szAmHdr, int (*xCellSize)(const SPage *, SCell *)) {
-  int    ret;
+  pPage->pPageHdr = pPage->pData + szAmHdr;
-  SCell *pTarget;
+  TDB_PAGE_NCELLS_SET(pPage, 0);
-  u8    *pTmp;
+  TDB_PAGE_CCELLS_SET(pPage, pPage->pageSize - sizeof(SPageFtr));
-  int    j;
+  TDB_PAGE_FCELL_SET(pPage, 0);
+  TDB_PAGE_NFREE_SET(pPage, TDB_PAGE_MAX_FREE_BLOCK(pPage, szAmHdr));
+  pPage->pCellIdx = pPage->pPageHdr + TDB_PAGE_HDR_SIZE(pPage);
+  pPage->pFreeStart = pPage->pCellIdx;
+  pPage->pFreeEnd = pPage->pData + TDB_PAGE_CCELLS(pPage);
+  pPage->pPageFtr = (SPageFtr *)(pPage->pData + pPage->pageSize - sizeof(SPageFtr));
+  pPage->nOverflow = 0;
+  pPage->xCellSize = xCellSize;
 
-  if (pPage->nOverflow || szCell + TDB_PAGE_OFFSET_SIZE(pPage) > pPage->nFree) {
+  ASSERT((u8 *)pPage->pPageFtr == pPage->pFreeEnd);
-    // TODO: need to figure out if pCell may be used by outside of this function
+}
-    j = pPage->nOverflow++;
 
-    pPage->apOvfl[j] = pCell;
+void tdbPageInit(SPage *pPage, u8 szAmHdr, int (*xCellSize)(const SPage *, SCell *)) {
-    pPage->aiOvfl[j] = idx;
+  pPage->pPageHdr = pPage->pData + szAmHdr;
-  } else {
+  pPage->pCellIdx = pPage->pPageHdr + TDB_PAGE_HDR_SIZE(pPage);
-    ret = tdbPageAllocate(pPage, szCell, &pTarget);
+  pPage->pFreeStart = pPage->pCellIdx + TDB_PAGE_OFFSET_SIZE(pPage) * TDB_PAGE_NCELLS(pPage);
-    if (ret < 0) {
+  pPage->pFreeEnd = pPage->pData + TDB_PAGE_CCELLS(pPage);
-      return -1;
+  pPage->pPageFtr = (SPageFtr *)(pPage->pData + pPage->pageSize - sizeof(SPageFtr));
+  pPage->nOverflow = 0;
+  pPage->xCellSize = xCellSize;
+
+  ASSERT(pPage->pFreeEnd >= pPage->pFreeStart);
+  ASSERT(pPage->pFreeEnd - pPage->pFreeStart <= TDB_PAGE_NFREE(pPage));
+}
+
+int tdbPageInsertCell(SPage *pPage, int idx, SCell *pCell, int szCell, u8 asOvfl) {
+  int    nFree;
+  int    nCells;
+  int    iOvfl;
+  int    lidx;  // local idx
+  SCell *pNewCell;
+
+  ASSERT(szCell <= TDB_PAGE_MAX_FREE_BLOCK(pPage, pPage->pPageHdr - pPage->pData));
+
+  nFree = TDB_PAGE_NFREE(pPage);
+  nCells = TDB_PAGE_NCELLS(pPage);
+  iOvfl = 0;
+
+  for (; iOvfl < pPage->nOverflow; iOvfl++) {
+    if (pPage->aiOvfl[iOvfl] >= idx) {
+      break;
+    }
+  }
+
+  lidx = idx - iOvfl;
+
+  if (asOvfl || nFree < szCell + TDB_PAGE_OFFSET_SIZE(pPage)) {
+    // TODO: make it extensible
+    // add the cell as an overflow cell
+    for (int i = pPage->nOverflow; i > iOvfl; i--) {
+      pPage->apOvfl[i] = pPage->apOvfl[i - 1];
+      pPage->aiOvfl[i] = pPage->aiOvfl[i - 1];
     }
 
-    memcpy(pTarget, pCell, szCell);
+    // TODO: here has memory leak
-    pTmp = pPage->pCellIdx + idx * TDB_PAGE_OFFSET_SIZE(pPage);
+    pNewCell = (SCell *)malloc(szCell);
-    memmove(pTmp + TDB_PAGE_OFFSET_SIZE(pPage), pTmp, pPage->pFreeStart - pTmp - TDB_PAGE_OFFSET_SIZE(pPage));
+    memcpy(pNewCell, pCell, szCell);
-    TDB_PAGE_CELL_OFFSET_AT_SET(pPage, idx, pTarget - pPage->pData);
+
-    TDB_PAGE_NCELLS_SET(pPage, TDB_PAGE_NCELLS(pPage) + 1);
+    pPage->apOvfl[iOvfl] = pNewCell;
+    pPage->aiOvfl[iOvfl] = idx;
+    pPage->nOverflow++;
+    iOvfl++;
+  } else {
+    // page must has enough space to hold the cell locally
+    tdbPageAllocate(pPage, szCell, &pNewCell);
+
+    memcpy(pNewCell, pCell, szCell);
+
+    // no overflow cell exists in this page
+    u8 *src = pPage->pCellIdx + TDB_PAGE_OFFSET_SIZE(pPage) * lidx;
+    u8 *dest = src + TDB_PAGE_OFFSET_SIZE(pPage);
+    memmove(dest, src, pPage->pFreeStart - dest);
+    TDB_PAGE_CELL_OFFSET_AT_SET(pPage, lidx, pNewCell - pPage->pData);
+    TDB_PAGE_NCELLS_SET(pPage, nCells + 1);
+
+    ASSERT(pPage->pFreeStart == pPage->pCellIdx + TDB_PAGE_OFFSET_SIZE(pPage) * (nCells + 1));
+  }
+
+  for (; iOvfl < pPage->nOverflow; iOvfl++) {
+    pPage->aiOvfl[iOvfl]++;
   }
 
   return 0;
 }
 
 int tdbPageDropCell(SPage *pPage, int idx) {
-  // TODO
+  int    lidx;
+  SCell *pCell;
+  int    szCell;
+  int    nCells;
+  int    iOvfl;
+
+  nCells = TDB_PAGE_NCELLS(pPage);
+
+  ASSERT(idx >= 0 && idx < nCells + pPage->nOverflow);
+
+  iOvfl = 0;
+  for (; iOvfl < pPage->nOverflow; iOvfl++) {
+    if (pPage->aiOvfl[iOvfl] == idx) {
+      // remove the over flow cell
+      for (; (++iOvfl) < pPage->nOverflow;) {
+        pPage->aiOvfl[iOvfl - 1] = pPage->aiOvfl[iOvfl] - 1;
+        pPage->apOvfl[iOvfl - 1] = pPage->apOvfl[iOvfl];
+      }
+
+      pPage->nOverflow--;
+      return 0;
+    } else if (pPage->aiOvfl[iOvfl] > idx) {
+      break;
+    }
+  }
+
+  lidx = idx - iOvfl;
+  pCell = TDB_PAGE_CELL_AT(pPage, lidx);
+  szCell = (*pPage->xCellSize)(pPage, pCell);
+  tdbPageFree(pPage, lidx, pCell, szCell);
+  TDB_PAGE_NCELLS_SET(pPage, nCells - 1);
+
+  for (; iOvfl < pPage->nOverflow; iOvfl++) {
+    pPage->aiOvfl[iOvfl]--;
+    ASSERT(pPage->aiOvfl[iOvfl] > 0);
+  }
+
   return 0;
 }
 
-static int tdbPageAllocate(SPage *pPage, int size, SCell **ppCell) {
+void tdbPageCopy(SPage *pFromPage, SPage *pToPage) {
-  SCell     *pCell;
+  int delta, nFree;
-  SFreeCell *pFreeCell;
-  u8        *pOffset;
-  int        ret;
 
-  ASSERT(pPage->nFree > size + TDB_PAGE_OFFSET_SIZE(pPage));
+  pToPage->pFreeStart = pToPage->pPageHdr + (pFromPage->pFreeStart - pFromPage->pPageHdr);
+  pToPage->pFreeEnd = (u8 *)(pToPage->pPageFtr) - ((u8 *)pFromPage->pPageFtr - pFromPage->pFreeEnd);
+
+  ASSERT(pToPage->pFreeEnd >= pToPage->pFreeStart);
+
+  memcpy(pToPage->pPageHdr, pFromPage->pPageHdr, pFromPage->pFreeStart - pFromPage->pPageHdr);
+  memcpy(pToPage->pFreeEnd, pFromPage->pFreeEnd, (u8 *)pFromPage->pPageFtr - pFromPage->pFreeEnd);
+
+  ASSERT(TDB_PAGE_CCELLS(pToPage) == pToPage->pFreeEnd - pToPage->pData);
+
+  delta = (pToPage->pPageHdr - pToPage->pData) - (pFromPage->pPageHdr - pFromPage->pData);
+  if (delta != 0) {
+    nFree = TDB_PAGE_NFREE(pFromPage);
+    TDB_PAGE_NFREE_SET(pToPage, nFree - delta);
+  }
+
+  // Copy the overflow cells
+  for (int iOvfl = 0; iOvfl < pFromPage->nOverflow; iOvfl++) {
+    pToPage->aiOvfl[iOvfl] = pFromPage->aiOvfl[iOvfl];
+    pToPage->apOvfl[iOvfl] = pFromPage->apOvfl[iOvfl];
+  }
+  pToPage->nOverflow = pFromPage->nOverflow;
+}
+
+static int tdbPageAllocate(SPage *pPage, int szCell, SCell **ppCell) {
+  SCell *pFreeCell;
+  u8    *pOffset;
+  int    nFree;
+  int    ret;
+  int    cellFree;
+  SCell *pCell = NULL;
 
-  pCell = NULL;
   *ppCell = NULL;
+  nFree = TDB_PAGE_NFREE(pPage);
 
-  // 1. Try to allocate from the free space area
+  ASSERT(nFree >= szCell + TDB_PAGE_OFFSET_SIZE(pPage));
-  if (pPage->pFreeEnd - pPage->pFreeStart > size + TDB_PAGE_OFFSET_SIZE(pPage)) {
+  ASSERT(TDB_PAGE_CCELLS(pPage) == pPage->pFreeEnd - pPage->pData);
-    pPage->pFreeEnd -= size;
+
-    pPage->pFreeStart += TDB_PAGE_OFFSET_SIZE(pPage);
+  // 1. Try to allocate from the free space block area
+  if (pPage->pFreeEnd - pPage->pFreeStart >= szCell + TDB_PAGE_OFFSET_SIZE(pPage)) {
+    pPage->pFreeEnd -= szCell;
     pCell = pPage->pFreeEnd;
+    TDB_PAGE_CCELLS_SET(pPage, pPage->pFreeEnd - pPage->pData);
+    goto _alloc_finish;
   }
 
   // 2. Try to allocate from the page free list
-  if ((pCell == NULL) && (pPage->pFreeEnd - pPage->pFreeStart >= TDB_PAGE_OFFSET_SIZE(pPage)) &&
+  cellFree = TDB_PAGE_FCELL(pPage);
-      TDB_PAGE_FCELL(pPage)) {
+  ASSERT(cellFree == 0 || cellFree > pPage->pFreeEnd - pPage->pData);
-#if 0
+  if (cellFree && pPage->pFreeEnd - pPage->pFreeStart >= TDB_PAGE_OFFSET_SIZE(pPage)) {
-    int szCell;
+    SCell *pPrevFreeCell = NULL;
-    int nxOffset;
+    int    szPrevFreeCell;
-
+    int    szFreeCell;
-    pCell = pPage->pData + TDB_PAGE_FCELL(pPage);
+    int    nxFreeCell;
-    pOffset = TDB_IS_LARGE_PAGE(pPage) ? ((SPageHdrL *)(pPage->pPageHdr))[0].fCell
+    int    newSize;
-                                       : (u8 *)&(((SPageHdr *)(pPage->pPageHdr))[0].fCell);
-    szCell = TDB_PAGE_FREE_CELL_SIZE(pPage, pCell);
-    nxOffset = TDB_PAGE_FREE_CELL_NXOFFSET(pPage, pCell);
 
     for (;;) {
-      // Find a cell
+      if (cellFree == 0) break;
-      if (szCell >= size) {
-        if (szCell - size >= pPage->szFreeCell) {
-          SCell *pTmpCell = pCell + size;
 
-          TDB_PAGE_FREE_CELL_SIZE_SET(pPage, pTmpCell, szCell - size);
+      pFreeCell = pPage->pData + cellFree;
-          TDB_PAGE_FREE_CELL_NXOFFSET_SET(pPage, pTmpCell, nxOffset);
+      pPage->pPageMethods->getFreeCellInfo(pFreeCell, &szFreeCell, &nxFreeCell);
-          // TODO: *pOffset = pTmpCell - pPage->pData;
+
+      if (szFreeCell >= szCell) {
+        pCell = pFreeCell;
+
+        newSize = szFreeCell - szCell;
+        pFreeCell += szCell;
+        if (newSize >= TDB_PAGE_FREE_CELL_SIZE(pPage)) {
+          pPage->pPageMethods->setFreeCellInfo(pFreeCell, newSize, nxFreeCell);
+          if (pPrevFreeCell) {
+            pPage->pPageMethods->setFreeCellInfo(pPrevFreeCell, szPrevFreeCell, pFreeCell - pPage->pData);
+          } else {
+            TDB_PAGE_FCELL_SET(pPage, pFreeCell - pPage->pData);
+          }
         } else {
-          TDB_PAGE_NFREE_SET(pPage, TDB_PAGE_NFREE(pPage) + szCell - size);
+          if (pPrevFreeCell) {
-          // TODO: *pOffset = nxOffset;
+            pPage->pPageMethods->setFreeCellInfo(pPrevFreeCell, szPrevFreeCell, nxFreeCell);
+          } else {
+            TDB_PAGE_FCELL_SET(pPage, nxFreeCell);
+          }
         }
-        break;
-      }
 
-      // Not find a cell yet
+        goto _alloc_finish;
-      if (nxOffset > 0) {
-        pCell = pPage->pData + nxOffset;
-        pOffset = TDB_PAGE_FREE_CELL_NXOFFSET_PTR(pPage, pCell);
-        szCell = TDB_PAGE_FREE_CELL_SIZE(pPage, pCell);
-        nxOffset = TDB_PAGE_FREE_CELL_NXOFFSET(pPage, pCell);
-        continue;
       } else {
-        pCell = NULL;
+        pPrevFreeCell = pFreeCell;
-        break;
+        szPrevFreeCell = szFreeCell;
+        cellFree = nxFreeCell;
       }
     }
-
-    if (pCell) {
-      pPage->pFreeStart = pPage->pFreeStart + pPage->szOffset;
-    }
-#endif
   }
 
   // 3. Try to dfragment and allocate again
-  if (pCell == NULL) {
+  tdbPageDefragment(pPage);
-    ret = tdbPageDefragment(pPage);
+  ASSERT(pPage->pFreeEnd - pPage->pFreeStart == nFree);
-    if (ret < 0) {
+  ASSERT(nFree == TDB_PAGE_NFREE(pPage));
-      return -1;
+  ASSERT(pPage->pFreeEnd - pPage->pData == TDB_PAGE_CCELLS(pPage));
-    }
 
-    ASSERT(pPage->pFreeEnd - pPage->pFreeStart > size + TDB_PAGE_OFFSET_SIZE(pPage));
+  pPage->pFreeEnd -= szCell;
-    ASSERT(pPage->nFree == pPage->pFreeEnd - pPage->pFreeStart);
+  pCell = pPage->pFreeEnd;
+  TDB_PAGE_CCELLS_SET(pPage, pPage->pFreeEnd - pPage->pData);
 
-    // Allocate from the free space area again
+_alloc_finish:
-    pPage->pFreeEnd -= size;
+  ASSERT(pCell);
-    pPage->pFreeStart += TDB_PAGE_OFFSET_SIZE(pPage);
+  pPage->pFreeStart += TDB_PAGE_OFFSET_SIZE(pPage);
-    pCell = pPage->pFreeEnd;
+  TDB_PAGE_NFREE_SET(pPage, nFree - szCell - TDB_PAGE_OFFSET_SIZE(pPage));
-  }
-
-  ASSERT(pCell != NULL);
-
-  pPage->nFree = pPage->nFree - size - TDB_PAGE_OFFSET_SIZE(pPage);
   *ppCell = pCell;
   return 0;
 }
 
-static int tdbPageFree(SPage *pPage, int idx, SCell *pCell, int size) {
+static int tdbPageFree(SPage *pPage, int idx, SCell *pCell, int szCell) {
-  // TODO
+  int nFree;
+  int cellFree;
+  u8 *dest;
+  u8 *src;
+
+  ASSERT(pCell >= pPage->pFreeEnd);
+  ASSERT(pCell + szCell <= (u8 *)(pPage->pPageFtr));
+  ASSERT(pCell == TDB_PAGE_CELL_AT(pPage, idx));
+
+  nFree = TDB_PAGE_NFREE(pPage);
+
+  if (pCell == pPage->pFreeEnd) {
+    pPage->pFreeEnd += szCell;
+    TDB_PAGE_CCELLS_SET(pPage, pPage->pFreeEnd - pPage->pData);
+  } else {
+    if (szCell >= TDB_PAGE_FREE_CELL_SIZE(pPage)) {
+      cellFree = TDB_PAGE_FCELL(pPage);
+      pPage->pPageMethods->setFreeCellInfo(pCell, szCell, cellFree);
+      TDB_PAGE_FCELL_SET(pPage, pCell - pPage->pData);
+    } else {
+      ASSERT(0);
+    }
+  }
+
+  dest = pPage->pCellIdx + TDB_PAGE_OFFSET_SIZE(pPage) * idx;
+  src = dest + TDB_PAGE_OFFSET_SIZE(pPage);
+  memmove(dest, src, pPage->pFreeStart - src);
+
+  pPage->pFreeStart -= TDB_PAGE_OFFSET_SIZE(pPage);
+  nFree = nFree + szCell + TDB_PAGE_OFFSET_SIZE(pPage);
+  TDB_PAGE_NFREE_SET(pPage, nFree);
   return 0;
 }
 
 static int tdbPageDefragment(SPage *pPage) {
-  // TODO
+  int    nFree;
-  ASSERT(0);
+  int    nCells;
+  SCell *pCell;
+  SCell *pNextCell;
+  SCell *pTCell;
+  int    szCell;
+  int    idx;
+  int    iCell;
+
+  ASSERT(pPage->pFreeEnd - pPage->pFreeStart < nFree);
+
+  nFree = TDB_PAGE_NFREE(pPage);
+  nCells = TDB_PAGE_NCELLS(pPage);
+
+  // Loop to compact the page content
+  // Here we use an O(n^2) algorithm to do the job since
+  // this is a low frequency job.
+  pNextCell = (u8 *)pPage->pPageFtr;
+  pCell = NULL;
+  for (iCell = 0;; iCell++) {
+    // compact over
+    if (iCell == nCells) {
+      pPage->pFreeEnd = pNextCell;
+      break;
+    }
+
+    for (int i = 0; i < nCells; i++) {
+      if (TDB_PAGE_CELL_OFFSET_AT(pPage, i) < pNextCell - pPage->pData) {
+        pTCell = TDB_PAGE_CELL_AT(pPage, i);
+        if (pCell == NULL || pCell < pTCell) {
+          pCell = pTCell;
+          idx = i;
+        }
+      } else {
+        continue;
+      }
+    }
+
+    ASSERT(pCell != NULL);
+
+    szCell = (*pPage->xCellSize)(pPage, pCell);
+
+    ASSERT(pCell + szCell <= pNextCell);
+    if (pCell + szCell < pNextCell) {
+      memmove(pNextCell - szCell, pCell, szCell);
+    }
+
+    pCell = NULL;
+    pNextCell = pNextCell - szCell;
+    TDB_PAGE_CELL_OFFSET_AT_SET(pPage, idx, pNextCell - pPage->pData);
+  }
+
+  ASSERT(pPage->pFreeEnd - pPage->pFreeStart == nFree);
+  TDB_PAGE_CCELLS_SET(pPage, pPage->pFreeEnd - pPage->pData);
+  TDB_PAGE_FCELL_SET(pPage, 0);
+
   return 0;
 }
 
 /* ---------------------------------------------------------------------------------------------------------- */
 typedef struct __attribute__((__packed__)) {
-  u16 flags;
   u16 cellNum;
   u16 cellBody;
   u16 cellFree;
   u16 nFree;
 } SPageHdr;
 
-// flags
+typedef struct __attribute__((__packed__)) {
-static inline u16  getPageFlags(SPage *pPage) { return ((SPageHdr *)(pPage->pPageHdr))[0].flags; }
+  u16 szCell;
-static inline void setPageFlags(SPage *pPage, u16 flags) { ((SPageHdr *)(pPage->pPageHdr))[0].flags = flags; }
+  u16 nxOffset;
+} SFreeCell;
 
 // cellNum
 static inline int  getPageCellNum(SPage *pPage) { return ((SPageHdr *)(pPage->pPageHdr))[0].cellNum; }
@@ -253,20 +474,33 @@ static inline void setPageCellOffset(SPage *pPage, int idx, int offset) {
   ((u16 *)pPage->pCellIdx)[idx] = (u16)offset;
 }
 
+// free cell info
+static inline void getPageFreeCellInfo(SCell *pCell, int *szCell, int *nxOffset) {
+  SFreeCell *pFreeCell = (SFreeCell *)pCell;
+  *szCell = pFreeCell->szCell;
+  *nxOffset = pFreeCell->nxOffset;
+}
+
+static inline void setPageFreeCellInfo(SCell *pCell, int szCell, int nxOffset) {
+  SFreeCell *pFreeCell = (SFreeCell *)pCell;
+  pFreeCell->szCell = szCell;
+  pFreeCell->nxOffset = nxOffset;
+}
+
 SPageMethods pageMethods = {
-    2,                  // szOffset
+    2,                    // szOffset
-    sizeof(SPageHdr),   // szPageHdr
+    sizeof(SPageHdr),     // szPageHdr
-    sizeof(SFreeCell),  // szFreeCell
+    sizeof(SFreeCell),    // szFreeCell
-    getPageFlags,       // getPageFlags
+    getPageCellNum,       // getCellNum
-    setPageFlags,       // setFlagsp
+    setPageCellNum,       // setCellNum
-    getPageCellNum,     // getCellNum
+    getPageCellBody,      // getCellBody
-    setPageCellNum,     // setCellNum
+    setPageCellBody,      // setCellBody
-    getPageCellBody,    // getCellBody
+    getPageCellFree,      // getCellFree
-    setPageCellBody,    // setCellBody
+    setPageCellFree,      // setCellFree
-    getPageCellFree,    // getCellFree
+    getPageNFree,         // getFreeBytes
-    setPageCellFree,    // setCellFree
+    setPageNFree,         // setFreeBytes
-    getPageNFree,       // getFreeBytes
+    getPageCellOffset,    // getCellOffset
-    setPageNFree,       // setFreeBytes
+    setPageCellOffset,    // setCellOffset
-    getPageCellOffset,  // getCellOffset
+    getPageFreeCellInfo,  // getFreeCellInfo
-    setPageCellOffset   // setCellOffset
+    setPageFreeCellInfo   // setFreeCellInfo
 };

source/libs/tdb/src/page/tdbPageL.c

@@ -16,11 +16,10 @@
 #include "tdbInt.h"
 
 typedef struct __attribute__((__packed__)) {
-  u16 flags;
+  u8 cellNum[3];
-  u8  cellNum[3];
+  u8 cellBody[3];
-  u8  cellBody[3];
+  u8 cellFree[3];
-  u8  cellFree[3];
+  u8 nFree[3];
-  u8  nFree[3];
 } SPageHdrL;
 
 typedef struct __attribute__((__packed__)) {
@@ -28,10 +27,6 @@ typedef struct __attribute__((__packed__)) {
   u8 nxOffset[3];
 } SFreeCellL;
 
-// flags
-static inline u16  getPageFlags(SPage *pPage) { return ((SPageHdrL *)(pPage->pPageHdr))[0].flags; }
-static inline void setPageFlags(SPage *pPage, u16 flags) { ((SPageHdrL *)(pPage->pPageHdr))[0].flags = flags; }
-
 // cellNum
 static inline int  getPageCellNum(SPage *pPage) { return TDB_GET_U24(((SPageHdrL *)(pPage->pPageHdr))[0].cellNum); }
 static inline void setPageCellNum(SPage *pPage, int cellNum) {
@@ -66,20 +61,33 @@ static inline void setPageCellOffset(SPage *pPage, int idx, int offset) {
   TDB_PUT_U24(pPage->pCellIdx + 3 * idx, offset);
 }
 
+// free cell info
+static inline void getPageFreeCellInfo(SCell *pCell, int *szCell, int *nxOffset) {
+  SFreeCellL *pFreeCell = (SFreeCellL *)pCell;
+  *szCell = TDB_GET_U24(pFreeCell->szCell);
+  *nxOffset = TDB_GET_U24(pFreeCell->nxOffset);
+}
+
+static inline void setPageFreeCellInfo(SCell *pCell, int szCell, int nxOffset) {
+  SFreeCellL *pFreeCell = (SFreeCellL *)pCell;
+  TDB_PUT_U24(pFreeCell->szCell, szCell);
+  TDB_PUT_U24(pFreeCell->nxOffset, nxOffset);
+}
+
 SPageMethods pageLargeMethods = {
-    3,                   // szOffset
+    3,                    // szOffset
-    sizeof(SPageHdrL),   // szPageHdr
+    sizeof(SPageHdrL),    // szPageHdr
-    sizeof(SFreeCellL),  // szFreeCell
+    sizeof(SFreeCellL),   // szFreeCell
-    getPageFlags,        // getPageFlags
+    getPageCellNum,       // getCellNum
-    setPageFlags,        // setFlagsp
+    setPageCellNum,       // setCellNum
-    getPageCellNum,      // getCellNum
+    getPageCellBody,      // getCellBody
-    setPageCellNum,      // setCellNum
+    setPageCellBody,      // setCellBody
-    getPageCellBody,     // getCellBody
+    getPageCellFree,      // getCellFree
-    setPageCellBody,     // setCellBody
+    setPageCellFree,      // setCellFree
-    getPageCellFree,     // getCellFree
+    getPageNFree,         // getFreeBytes
-    setPageCellFree,     // setCellFree
+    setPageNFree,         // setFreeBytes
-    getPageNFree,        // getFreeBytes
+    getPageCellOffset,    // getCellOffset
-    setPageNFree,        // setFreeBytes
+    setPageCellOffset,    // setCellOffset
-    getPageCellOffset,   // getCellOffset
+    getPageFreeCellInfo,  // getFreeCellInfo
-    setPageCellOffset    // setCellOffset
+    setPageFreeCellInfo   // setFreeCellInfo
 };