diff --git a/source/dnode/vnode/src/inc/tsdb.h b/source/dnode/vnode/src/inc/tsdb.h index 7546b0943e..91aacaa328 100644 --- a/source/dnode/vnode/src/inc/tsdb.h +++ b/source/dnode/vnode/src/inc/tsdb.h @@ -65,12 +65,14 @@ typedef struct SBlockInfo SBlockInfo; typedef struct SSmaInfo SSmaInfo; typedef struct SBlockCol SBlockCol; typedef struct SVersionRange SVersionRange; +typedef struct SLDataIter SLDataIter; #define TSDB_FILE_DLMT ((uint32_t)0xF00AFA0F) #define TSDB_MAX_SUBBLOCKS 8 -#define TSDB_MAX_LAST_FILE 16 -#define TSDB_DEFAULT_LAST_FILE 8 +#define TSDB_MAX_SST_FILE 16 +#define TSDB_DEFAULT_SST_FILE 8 #define TSDB_FHDR_SIZE 512 +#define TSDB_DEFAULT_PAGE_SIZE 4096 #define HAS_NONE ((int8_t)0x1) #define HAS_NULL ((int8_t)0x2) @@ -82,6 +84,14 @@ typedef struct SVersionRange SVersionRange; #define TSDBKEY_MIN ((TSDBKEY){.ts = TSKEY_MIN, .version = VERSION_MIN}) #define TSDBKEY_MAX ((TSDBKEY){.ts = TSKEY_MAX, .version = VERSION_MAX}) +#define PAGE_CONTENT_SIZE(PAGE) ((PAGE) - sizeof(TSCKSUM)) +#define LOGIC_TO_FILE_OFFSET(LOFFSET, PAGE) \ + ((LOFFSET) / PAGE_CONTENT_SIZE(PAGE) * (PAGE) + (LOFFSET) % PAGE_CONTENT_SIZE(PAGE)) +#define FILE_TO_LOGIC_OFFSET(OFFSET, PAGE) ((OFFSET) / (PAGE)*PAGE_CONTENT_SIZE(PAGE) + (OFFSET) % (PAGE)) +#define PAGE_OFFSET(PGNO, PAGE) (((PGNO)-1) * (PAGE)) +#define OFFSET_PGNO(OFFSET, PAGE) ((OFFSET) / (PAGE) + 1) +#define LOGIC_TO_FILE_SIZE(LSIZE, PAGE) OFFSET_PGNO(LOGIC_TO_FILE_OFFSET(LSIZE, PAGE), PAGE) * (PAGE) + // tsdbUtil.c ============================================================================================== // TSDBROW #define TSDBROW_TS(ROW) (((ROW)->type == 0) ? (ROW)->pTSRow->ts : (ROW)->pBlockData->aTSKEY[(ROW)->iRow]) @@ -195,7 +205,6 @@ int32_t tsdbCmprColData(SColData *pColData, int8_t cmprAlg, SBlockCol *pBlockCol uint8_t **ppBuf); int32_t tsdbDecmprColData(uint8_t *pIn, SBlockCol *pBlockCol, int8_t cmprAlg, int32_t nVal, SColData *pColData, uint8_t **ppBuf); -int32_t tsdbReadAndCheck(TdFilePtr pFD, int64_t offset, uint8_t **ppOut, int32_t size, int8_t toCheck); // tsdbMemTable ============================================================================================== // SMemTable int32_t tsdbMemTableCreate(STsdb *pTsdb, SMemTable **ppMemTable); @@ -563,7 +572,7 @@ struct SDFileSet { SDataFile *pDataF; SSmaFile *pSmaF; uint8_t nSstF; - SSstFile *aSstF[TSDB_MAX_LAST_FILE]; + SSstFile *aSstF[TSDB_MAX_SST_FILE]; }; struct SRowIter { @@ -578,46 +587,54 @@ struct SRowMerger { SArray *pArray; // SArray }; +typedef struct { + char *path; + int32_t szPage; + int32_t flag; + TdFilePtr pFD; + int64_t pgno; + uint8_t *pBuf; + int64_t szFile; +} STsdbFD; + struct SDelFWriter { - STsdb *pTsdb; - SDelFile fDel; - TdFilePtr pWriteH; - + STsdb *pTsdb; + SDelFile fDel; + STsdbFD *pWriteH; uint8_t *aBuf[1]; }; -struct SDataFWriter { - STsdb *pTsdb; - SDFileSet wSet; - - TdFilePtr pHeadFD; - TdFilePtr pDataFD; - TdFilePtr pSmaFD; - TdFilePtr pLastFD; - - SHeadFile fHead; - SDataFile fData; - SSmaFile fSma; - SSstFile fSst[TSDB_MAX_LAST_FILE]; - - uint8_t *aBuf[4]; -}; - struct STsdbReadSnap { SMemTable *pMem; SMemTable *pIMem; STsdbFS fs; }; +struct SDataFWriter { + STsdb *pTsdb; + SDFileSet wSet; + + STsdbFD *pHeadFD; + STsdbFD *pDataFD; + STsdbFD *pSmaFD; + STsdbFD *pSstFD; + + SHeadFile fHead; + SDataFile fData; + SSmaFile fSma; + SSstFile fSst[TSDB_MAX_SST_FILE]; + + uint8_t *aBuf[4]; +}; + struct SDataFReader { STsdb *pTsdb; SDFileSet *pSet; - TdFilePtr pHeadFD; - TdFilePtr pDataFD; - TdFilePtr pSmaFD; - TdFilePtr aLastFD[TSDB_MAX_LAST_FILE]; - - uint8_t *aBuf[3]; + STsdbFD *pHeadFD; + STsdbFD *pDataFD; + STsdbFD *pSmaFD; + STsdbFD *aSstFD[TSDB_MAX_SST_FILE]; + uint8_t *aBuf[3]; }; typedef struct { @@ -627,15 +644,16 @@ typedef struct { } SRowInfo; typedef struct SMergeTree { - int8_t backward; - SRBTree rbt; - SArray *pIterList; - struct SLDataIter *pIter; + int8_t backward; + SRBTree rbt; + SArray *pIterList; + SLDataIter *pIter; } SMergeTree; -int32_t tMergeTreeOpen(SMergeTree *pMTree, int8_t backward, SDataFReader* pFReader, uint64_t uid, STimeWindow* pTimeWindow, SVersionRange* pVerRange); -void tMergeTreeAddIter(SMergeTree *pMTree, struct SLDataIter *pIter); -bool tMergeTreeNext(SMergeTree *pMTree); +int32_t tMergeTreeOpen(SMergeTree *pMTree, int8_t backward, SDataFReader *pFReader, uint64_t uid, + STimeWindow *pTimeWindow, SVersionRange *pVerRange); +void tMergeTreeAddIter(SMergeTree *pMTree, SLDataIter *pIter); +bool tMergeTreeNext(SMergeTree *pMTree); TSDBROW tMergeTreeGetRow(SMergeTree *pMTree); void tMergeTreeClose(SMergeTree *pMTree); diff --git a/source/dnode/vnode/src/tsdb/tsdbCommit.c b/source/dnode/vnode/src/tsdb/tsdbCommit.c index acbdc65f12..6f7a78ee46 100644 --- a/source/dnode/vnode/src/tsdb/tsdbCommit.c +++ b/source/dnode/vnode/src/tsdb/tsdbCommit.c @@ -71,7 +71,7 @@ typedef struct { SDataIter *pIter; SRBTree rbt; SDataIter dataIter; - SDataIter aDataIter[TSDB_MAX_LAST_FILE]; + SDataIter aDataIter[TSDB_MAX_SST_FILE]; int8_t toLastOnly; }; struct { @@ -92,9 +92,6 @@ typedef struct { SArray *aDelData; // SArray } SCommitter; -extern int32_t tsdbReadSstBlockEx(SDataFReader *pReader, int32_t iSst, SSstBlk *aSstBlk, - SBlockData *pBlockData); // todo - static int32_t tsdbStartCommit(STsdb *pTsdb, SCommitter *pCommitter); static int32_t tsdbCommitData(SCommitter *pCommitter); static int32_t tsdbCommitDel(SCommitter *pCommitter); @@ -445,7 +442,7 @@ static int32_t tsdbOpenCommitIter(SCommitter *pCommitter) { pIter->iSstBlk = 0; SSstBlk *pSstBlk = (SSstBlk *)taosArrayGet(pIter->aSstBlk, 0); - code = tsdbReadSstBlockEx(pCommitter->dReader.pReader, iSst, pSstBlk, &pIter->bData); + code = tsdbReadSstBlock(pCommitter->dReader.pReader, iSst, pSstBlk, &pIter->bData); if (code) goto _err; pIter->iRow = 0; @@ -760,7 +757,7 @@ static int32_t tsdbStartCommit(STsdb *pTsdb, SCommitter *pCommitter) { pCommitter->minRow = pTsdb->pVnode->config.tsdbCfg.minRows; pCommitter->maxRow = pTsdb->pVnode->config.tsdbCfg.maxRows; pCommitter->cmprAlg = pTsdb->pVnode->config.tsdbCfg.compression; - pCommitter->maxLast = TSDB_DEFAULT_LAST_FILE; // TODO: make it as a config + pCommitter->maxLast = TSDB_DEFAULT_SST_FILE; // TODO: make it as a config pCommitter->aTbDataP = tsdbMemTableGetTbDataArray(pTsdb->imem); if (pCommitter->aTbDataP == NULL) { code = TSDB_CODE_OUT_OF_MEMORY; @@ -790,7 +787,7 @@ static int32_t tsdbCommitDataStart(SCommitter *pCommitter) { if (code) goto _exit; // merger - for (int32_t iSst = 0; iSst < TSDB_MAX_LAST_FILE; iSst++) { + for (int32_t iSst = 0; iSst < TSDB_MAX_SST_FILE; iSst++) { SDataIter *pIter = &pCommitter->aDataIter[iSst]; pIter->aSstBlk = taosArrayInit(0, sizeof(SSstBlk)); if (pIter->aSstBlk == NULL) { @@ -832,7 +829,7 @@ static void tsdbCommitDataEnd(SCommitter *pCommitter) { tBlockDataDestroy(&pCommitter->dReader.bData, 1); // merger - for (int32_t iSst = 0; iSst < TSDB_MAX_LAST_FILE; iSst++) { + for (int32_t iSst = 0; iSst < TSDB_MAX_SST_FILE; iSst++) { SDataIter *pIter = &pCommitter->aDataIter[iSst]; taosArrayDestroy(pIter->aSstBlk); tBlockDataDestroy(&pIter->bData, 1); @@ -1059,7 +1056,7 @@ static int32_t tsdbNextCommitRow(SCommitter *pCommitter) { if (pIter->iSstBlk < taosArrayGetSize(pIter->aSstBlk)) { SSstBlk *pSstBlk = (SSstBlk *)taosArrayGet(pIter->aSstBlk, pIter->iSstBlk); - code = tsdbReadSstBlockEx(pCommitter->dReader.pReader, pIter->iSst, pSstBlk, &pIter->bData); + code = tsdbReadSstBlock(pCommitter->dReader.pReader, pIter->iSst, pSstBlk, &pIter->bData); if (code) goto _exit; pIter->iRow = 0; diff --git a/source/dnode/vnode/src/tsdb/tsdbFS.c b/source/dnode/vnode/src/tsdb/tsdbFS.c index e6bc9d9936..14bc1214a6 100644 --- a/source/dnode/vnode/src/tsdb/tsdbFS.c +++ b/source/dnode/vnode/src/tsdb/tsdbFS.c @@ -21,6 +21,9 @@ static int32_t tsdbEncodeFS(uint8_t *p, STsdbFS *pFS) { int8_t hasDel = pFS->pDelFile ? 1 : 0; uint32_t nSet = taosArrayGetSize(pFS->aDFileSet); + // version + n += tPutI8(p ? p + n : p, 0); + // SDelFile n += tPutI8(p ? p + n : p, hasDel); if (hasDel) { @@ -292,7 +295,7 @@ static int32_t tsdbScanAndTryFixFS(STsdb *pTsdb) { code = TAOS_SYSTEM_ERROR(errno); goto _err; } - if (size != pSet->pHeadF->size) { + if (size != LOGIC_TO_FILE_SIZE(pSet->pHeadF->size, TSDB_DEFAULT_PAGE_SIZE)) { code = TSDB_CODE_FILE_CORRUPTED; goto _err; } @@ -303,10 +306,10 @@ static int32_t tsdbScanAndTryFixFS(STsdb *pTsdb) { code = TAOS_SYSTEM_ERROR(errno); goto _err; } - if (size < pSet->pDataF->size) { + if (size < LOGIC_TO_FILE_SIZE(pSet->pDataF->size, TSDB_DEFAULT_PAGE_SIZE)) { code = TSDB_CODE_FILE_CORRUPTED; goto _err; - } else if (size > pSet->pDataF->size) { + } else if (size > LOGIC_TO_FILE_SIZE(pSet->pDataF->size, TSDB_DEFAULT_PAGE_SIZE)) { code = tsdbDFileRollback(pTsdb, pSet, TSDB_DATA_FILE); if (code) goto _err; } @@ -317,10 +320,10 @@ static int32_t tsdbScanAndTryFixFS(STsdb *pTsdb) { code = TAOS_SYSTEM_ERROR(errno); goto _err; } - if (size < pSet->pSmaF->size) { + if (size < LOGIC_TO_FILE_SIZE(pSet->pSmaF->size, TSDB_DEFAULT_PAGE_SIZE)) { code = TSDB_CODE_FILE_CORRUPTED; goto _err; - } else if (size > pSet->pSmaF->size) { + } else if (size > LOGIC_TO_FILE_SIZE(pSet->pSmaF->size, TSDB_DEFAULT_PAGE_SIZE)) { code = tsdbDFileRollback(pTsdb, pSet, TSDB_SMA_FILE); if (code) goto _err; } @@ -332,7 +335,7 @@ static int32_t tsdbScanAndTryFixFS(STsdb *pTsdb) { code = TAOS_SYSTEM_ERROR(errno); goto _err; } - if (size != pSet->aSstF[iSst]->size) { + if (size != LOGIC_TO_FILE_SIZE(pSet->aSstF[iSst]->size, TSDB_DEFAULT_PAGE_SIZE)) { code = TSDB_CODE_FILE_CORRUPTED; goto _err; } @@ -364,10 +367,12 @@ static int32_t tsdbRecoverFS(STsdb *pTsdb, uint8_t *pData, int64_t nData) { int32_t code = 0; int8_t hasDel; uint32_t nSet; - int32_t n; + int32_t n = 0; + + // version + n += tGetI8(pData + n, NULL); // SDelFile - n = 0; n += tGetI8(pData + n, &hasDel); if (hasDel) { pTsdb->fs.pDelFile = (SDelFile *)taosMemoryMalloc(sizeof(SDelFile)); diff --git a/source/dnode/vnode/src/tsdb/tsdbFile.c b/source/dnode/vnode/src/tsdb/tsdbFile.c index 632a2c827b..2a7966e423 100644 --- a/source/dnode/vnode/src/tsdb/tsdbFile.c +++ b/source/dnode/vnode/src/tsdb/tsdbFile.c @@ -148,7 +148,7 @@ int32_t tsdbDFileRollback(STsdb *pTsdb, SDFileSet *pSet, EDataFileT ftype) { } // ftruncate - if (taosFtruncateFile(pFD, size) < 0) { + if (taosFtruncateFile(pFD, LOGIC_TO_FILE_SIZE(size, TSDB_DEFAULT_PAGE_SIZE)) < 0) { code = TAOS_SYSTEM_ERROR(errno); goto _err; } diff --git a/source/dnode/vnode/src/tsdb/tsdbMergeTree.c b/source/dnode/vnode/src/tsdb/tsdbMergeTree.c index e85b136c2a..171e32007a 100644 --- a/source/dnode/vnode/src/tsdb/tsdbMergeTree.c +++ b/source/dnode/vnode/src/tsdb/tsdbMergeTree.c @@ -16,7 +16,7 @@ #include "tsdb.h" // SLDataIter ================================================= -typedef struct SLDataIter { +struct SLDataIter { SRBTreeNode node; SSstBlk *pSstBlk; SDataFReader *pReader; @@ -31,13 +31,11 @@ typedef struct SLDataIter { uint64_t uid; STimeWindow timeWindow; SVersionRange verRange; -} SLDataIter; +}; -static SBlockData* getCurrentBlock(SLDataIter* pIter) { - return &pIter->bData[pIter->loadIndex]; -} +static SBlockData *getCurrentBlock(SLDataIter *pIter) { return &pIter->bData[pIter->loadIndex]; } -static SBlockData* getNextBlock(SLDataIter* pIter) { +static SBlockData *getNextBlock(SLDataIter *pIter) { pIter->loadIndex ^= 1; return getCurrentBlock(pIter); } @@ -116,8 +114,6 @@ void tLDataIterClose(SLDataIter *pIter) { taosMemoryFree(pIter); } -extern int32_t tsdbReadSstBlockEx(SDataFReader *pReader, int32_t iSst, SSstBlk *pSstBlk, SBlockData *pBlockData); - void tLDataIterNextBlock(SLDataIter *pIter) { int32_t step = pIter->backward ? -1 : 1; pIter->iSstBlk += step; @@ -150,9 +146,9 @@ void tLDataIterNextBlock(SLDataIter *pIter) { static void findNextValidRow(SLDataIter *pIter) { int32_t step = pIter->backward ? -1 : 1; - bool hasVal = false; - int32_t i = pIter->iRow; - SBlockData* pBlockData = getCurrentBlock(pIter); + bool hasVal = false; + int32_t i = pIter->iRow; + SBlockData *pBlockData = getCurrentBlock(pIter); for (; i < pBlockData->nRow && i >= 0; i += step) { if (pBlockData->aUid != NULL) { @@ -220,12 +216,12 @@ bool tLDataIterNextRow(SLDataIter *pIter) { return false; } - int32_t iBlockL = pIter->iSstBlk; - SBlockData* pBlockData = getCurrentBlock(pIter); + int32_t iBlockL = pIter->iSstBlk; + SBlockData *pBlockData = getCurrentBlock(pIter); if (pBlockData->nRow == 0 && pIter->pSstBlk != NULL) { // current block not loaded yet pBlockData = getNextBlock(pIter); - code = tsdbReadSstBlockEx(pIter->pReader, pIter->iSst, pIter->pSstBlk, pBlockData); + code = tsdbReadSstBlock(pIter->pReader, pIter->iSst, pIter->pSstBlk, pBlockData); if (code != TSDB_CODE_SUCCESS) { goto _exit; } @@ -249,7 +245,7 @@ bool tLDataIterNextRow(SLDataIter *pIter) { if (iBlockL != pIter->iSstBlk) { pBlockData = getNextBlock(pIter); - code = tsdbReadSstBlockEx(pIter->pReader, pIter->iSst, pIter->pSstBlk, pBlockData); + code = tsdbReadSstBlock(pIter->pReader, pIter->iSst, pIter->pSstBlk, pBlockData); if (code) { goto _exit; } @@ -306,7 +302,7 @@ int32_t tMergeTreeOpen(SMergeTree *pMTree, int8_t backward, SDataFReader *pFRead tRBTreeCreate(&pMTree->rbt, tLDataIterCmprFn); int32_t code = TSDB_CODE_OUT_OF_MEMORY; - struct SLDataIter *pIterList[TSDB_DEFAULT_LAST_FILE] = {0}; + struct SLDataIter *pIterList[TSDB_DEFAULT_SST_FILE] = {0}; for (int32_t i = 0; i < pFReader->pSet->nSstF; ++i) { // open all last file code = tLDataIterOpen(&pIterList[i], pFReader, i, pMTree->backward, uid, pTimeWindow, pVerRange); if (code != TSDB_CODE_SUCCESS) { diff --git a/source/dnode/vnode/src/tsdb/tsdbReaderWriter.c b/source/dnode/vnode/src/tsdb/tsdbReaderWriter.c index 34bf1215d7..25daec76c6 100644 --- a/source/dnode/vnode/src/tsdb/tsdbReaderWriter.c +++ b/source/dnode/vnode/src/tsdb/tsdbReaderWriter.c @@ -15,912 +15,191 @@ #include "tsdb.h" -// SDelFWriter ==================================================== -int32_t tsdbDelFWriterOpen(SDelFWriter **ppWriter, SDelFile *pFile, STsdb *pTsdb) { - int32_t code = 0; - char fname[TSDB_FILENAME_LEN]; - char hdr[TSDB_FHDR_SIZE] = {0}; - SDelFWriter *pDelFWriter; - int64_t n; - - // alloc - pDelFWriter = (SDelFWriter *)taosMemoryCalloc(1, sizeof(*pDelFWriter)); - if (pDelFWriter == NULL) { - code = TSDB_CODE_OUT_OF_MEMORY; - goto _err; - } - pDelFWriter->pTsdb = pTsdb; - pDelFWriter->fDel = *pFile; - - tsdbDelFileName(pTsdb, pFile, fname); - pDelFWriter->pWriteH = taosOpenFile(fname, TD_FILE_WRITE | TD_FILE_CREATE); - if (pDelFWriter->pWriteH == NULL) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } - - // update header - n = taosWriteFile(pDelFWriter->pWriteH, &hdr, TSDB_FHDR_SIZE); - if (n < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } - - pDelFWriter->fDel.size = TSDB_FHDR_SIZE; - pDelFWriter->fDel.offset = 0; - - *ppWriter = pDelFWriter; - return code; - -_err: - tsdbError("vgId:%d, failed to open del file writer since %s", TD_VID(pTsdb->pVnode), tstrerror(code)); - *ppWriter = NULL; - return code; -} - -int32_t tsdbDelFWriterClose(SDelFWriter **ppWriter, int8_t sync) { - int32_t code = 0; - SDelFWriter *pWriter = *ppWriter; - STsdb *pTsdb = pWriter->pTsdb; - - // sync - if (sync && taosFsyncFile(pWriter->pWriteH) < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } - - // close - if (taosCloseFile(&pWriter->pWriteH) < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } - - for (int32_t iBuf = 0; iBuf < sizeof(pWriter->aBuf) / sizeof(uint8_t *); iBuf++) { - tFree(pWriter->aBuf[iBuf]); - } - taosMemoryFree(pWriter); - - *ppWriter = NULL; - return code; - -_err: - tsdbError("vgId:%d, failed to close del file writer since %s", TD_VID(pTsdb->pVnode), tstrerror(code)); - return code; -} - -int32_t tsdbWriteDelData(SDelFWriter *pWriter, SArray *aDelData, SDelIdx *pDelIdx) { - int32_t code = 0; - int64_t size; - int64_t n; - - // prepare - size = sizeof(uint32_t); - for (int32_t iDelData = 0; iDelData < taosArrayGetSize(aDelData); iDelData++) { - size += tPutDelData(NULL, taosArrayGet(aDelData, iDelData)); - } - size += sizeof(TSCKSUM); - - // alloc - code = tRealloc(&pWriter->aBuf[0], size); - if (code) goto _err; - - // build - n = 0; - n += tPutU32(pWriter->aBuf[0] + n, TSDB_FILE_DLMT); - for (int32_t iDelData = 0; iDelData < taosArrayGetSize(aDelData); iDelData++) { - n += tPutDelData(pWriter->aBuf[0] + n, taosArrayGet(aDelData, iDelData)); - } - taosCalcChecksumAppend(0, pWriter->aBuf[0], size); - - ASSERT(n + sizeof(TSCKSUM) == size); - - // write - n = taosWriteFile(pWriter->pWriteH, pWriter->aBuf[0], size); - if (n < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } - - ASSERT(n == size); - - // update - pDelIdx->offset = pWriter->fDel.size; - pDelIdx->size = size; - pWriter->fDel.size += size; - - return code; - -_err: - tsdbError("vgId:%d, failed to write del data since %s", TD_VID(pWriter->pTsdb->pVnode), tstrerror(code)); - return code; -} - -int32_t tsdbWriteDelIdx(SDelFWriter *pWriter, SArray *aDelIdx) { +// =============== PAGE-WISE FILE =============== +static int32_t tsdbOpenFile(const char *path, int32_t szPage, int32_t flag, STsdbFD **ppFD) { int32_t code = 0; - int64_t size; - int64_t n; - SDelIdx *pDelIdx; + STsdbFD *pFD; - // prepare - size = sizeof(uint32_t); - for (int32_t iDelIdx = 0; iDelIdx < taosArrayGetSize(aDelIdx); iDelIdx++) { - size += tPutDelIdx(NULL, taosArrayGet(aDelIdx, iDelIdx)); - } - size += sizeof(TSCKSUM); + *ppFD = NULL; - // alloc - code = tRealloc(&pWriter->aBuf[0], size); - if (code) goto _err; - - // build - n = 0; - n += tPutU32(pWriter->aBuf[0] + n, TSDB_FILE_DLMT); - for (int32_t iDelIdx = 0; iDelIdx < taosArrayGetSize(aDelIdx); iDelIdx++) { - n += tPutDelIdx(pWriter->aBuf[0] + n, taosArrayGet(aDelIdx, iDelIdx)); - } - taosCalcChecksumAppend(0, pWriter->aBuf[0], size); - - ASSERT(n + sizeof(TSCKSUM) == size); - - // write - n = taosWriteFile(pWriter->pWriteH, pWriter->aBuf[0], size); - if (n < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } - - // update - pWriter->fDel.offset = pWriter->fDel.size; - pWriter->fDel.size += size; - - return code; - -_err: - tsdbError("vgId:%d, write del idx failed since %s", TD_VID(pWriter->pTsdb->pVnode), tstrerror(code)); - return code; -} - -int32_t tsdbUpdateDelFileHdr(SDelFWriter *pWriter) { - int32_t code = 0; - char hdr[TSDB_FHDR_SIZE]; - int64_t size = TSDB_FHDR_SIZE; - int64_t n; - - // build - memset(hdr, 0, size); - tPutDelFile(hdr, &pWriter->fDel); - taosCalcChecksumAppend(0, hdr, size); - - // seek - if (taosLSeekFile(pWriter->pWriteH, 0, SEEK_SET) < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } - - // write - n = taosWriteFile(pWriter->pWriteH, hdr, size); - if (n < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } - - return code; - -_err: - tsdbError("vgId:%d, update del file hdr failed since %s", TD_VID(pWriter->pTsdb->pVnode), tstrerror(code)); - return code; -} - -// SDelFReader ==================================================== -struct SDelFReader { - STsdb *pTsdb; - SDelFile fDel; - TdFilePtr pReadH; - - uint8_t *aBuf[1]; -}; - -int32_t tsdbDelFReaderOpen(SDelFReader **ppReader, SDelFile *pFile, STsdb *pTsdb) { - int32_t code = 0; - char fname[TSDB_FILENAME_LEN]; - SDelFReader *pDelFReader; - int64_t n; - - // alloc - pDelFReader = (SDelFReader *)taosMemoryCalloc(1, sizeof(*pDelFReader)); - if (pDelFReader == NULL) { + pFD = (STsdbFD *)taosMemoryCalloc(1, sizeof(*pFD) + strlen(path) + 1); + if (pFD == NULL) { code = TSDB_CODE_OUT_OF_MEMORY; - goto _err; + goto _exit; } - // open impl - pDelFReader->pTsdb = pTsdb; - pDelFReader->fDel = *pFile; - - tsdbDelFileName(pTsdb, pFile, fname); - pDelFReader->pReadH = taosOpenFile(fname, TD_FILE_READ); - if (pDelFReader->pReadH == NULL) { + pFD->path = (char *)&pFD[1]; + strcpy(pFD->path, path); + pFD->szPage = szPage; + pFD->flag = flag; + pFD->pFD = taosOpenFile(path, flag); + if (pFD->pFD == NULL) { code = TAOS_SYSTEM_ERROR(errno); - taosMemoryFree(pDelFReader); - goto _err; + goto _exit; } + pFD->szPage = szPage; + pFD->pgno = 0; + pFD->pBuf = taosMemoryCalloc(1, szPage); + if (pFD->pBuf == NULL) { + code = TSDB_CODE_OUT_OF_MEMORY; + taosMemoryFree(pFD); + goto _exit; + } + if (taosStatFile(path, &pFD->szFile, NULL) < 0) { + code = TAOS_SYSTEM_ERROR(errno); + goto _exit; + } + ASSERT(pFD->szFile % szPage == 0); + pFD->szFile = pFD->szFile / szPage; + *ppFD = pFD; _exit: - *ppReader = pDelFReader; - return code; - -_err: - tsdbError("vgId:%d, del file reader open failed since %s", TD_VID(pTsdb->pVnode), tstrerror(code)); - *ppReader = NULL; return code; } -int32_t tsdbDelFReaderClose(SDelFReader **ppReader) { - int32_t code = 0; - SDelFReader *pReader = *ppReader; +static void tsdbCloseFile(STsdbFD **ppFD) { + STsdbFD *pFD = *ppFD; + taosMemoryFree(pFD->pBuf); + taosCloseFile(&pFD->pFD); + taosMemoryFree(pFD); + *ppFD = NULL; +} - if (pReader) { - if (taosCloseFile(&pReader->pReadH) < 0) { +static int32_t tsdbWriteFilePage(STsdbFD *pFD) { + int32_t code = 0; + + if (pFD->pgno > 0) { + int64_t n = taosLSeekFile(pFD->pFD, PAGE_OFFSET(pFD->pgno, pFD->szPage), SEEK_SET); + if (n < 0) { code = TAOS_SYSTEM_ERROR(errno); goto _exit; } - for (int32_t iBuf = 0; iBuf < sizeof(pReader->aBuf) / sizeof(uint8_t *); iBuf++) { - tFree(pReader->aBuf[iBuf]); + + taosCalcChecksumAppend(0, pFD->pBuf, pFD->szPage); + + n = taosWriteFile(pFD->pFD, pFD->pBuf, pFD->szPage); + if (n < 0) { + code = TAOS_SYSTEM_ERROR(errno); + goto _exit; + } + + if (pFD->szFile < pFD->pgno) { + pFD->szFile = pFD->pgno; } - taosMemoryFree(pReader); } - *ppReader = NULL; + pFD->pgno = 0; _exit: return code; } -int32_t tsdbReadDelData(SDelFReader *pReader, SDelIdx *pDelIdx, SArray *aDelData) { +static int32_t tsdbReadFilePage(STsdbFD *pFD, int64_t pgno) { int32_t code = 0; - int64_t offset = pDelIdx->offset; - int64_t size = pDelIdx->size; - int64_t n; - taosArrayClear(aDelData); + ASSERT(pgno <= pFD->szFile); // seek - if (taosLSeekFile(pReader->pReadH, offset, SEEK_SET) < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } - - // alloc - code = tRealloc(&pReader->aBuf[0], size); - if (code) goto _err; - - // read - n = taosReadFile(pReader->pReadH, pReader->aBuf[0], size); + int64_t offset = PAGE_OFFSET(pgno, pFD->szPage); + int64_t n = taosLSeekFile(pFD->pFD, offset, SEEK_SET); if (n < 0) { code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } else if (n < size) { - code = TSDB_CODE_FILE_CORRUPTED; - goto _err; - } - - // check - if (!taosCheckChecksumWhole(pReader->aBuf[0], size)) { - code = TSDB_CODE_FILE_CORRUPTED; - goto _err; - } - - // // decode - n = 0; - - uint32_t delimiter; - n += tGetU32(pReader->aBuf[0] + n, &delimiter); - while (n < size - sizeof(TSCKSUM)) { - SDelData delData; - n += tGetDelData(pReader->aBuf[0] + n, &delData); - - if (taosArrayPush(aDelData, &delData) == NULL) { - code = TSDB_CODE_OUT_OF_MEMORY; - goto _err; - } - } - - ASSERT(n == size - sizeof(TSCKSUM)); - - return code; - -_err: - tsdbError("vgId:%d, read del data failed since %s", TD_VID(pReader->pTsdb->pVnode), tstrerror(code)); - return code; -} - -int32_t tsdbReadDelIdx(SDelFReader *pReader, SArray *aDelIdx) { - int32_t code = 0; - int32_t n; - int64_t offset = pReader->fDel.offset; - int64_t size = pReader->fDel.size - offset; - - taosArrayClear(aDelIdx); - - // seek - if (taosLSeekFile(pReader->pReadH, offset, SEEK_SET) < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } - - // alloc - code = tRealloc(&pReader->aBuf[0], size); - if (code) goto _err; - - // read - n = taosReadFile(pReader->pReadH, pReader->aBuf[0], size); - if (n < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } else if (n < size) { - code = TSDB_CODE_FILE_CORRUPTED; - goto _err; - } - - // check - if (!taosCheckChecksumWhole(pReader->aBuf[0], size)) { - code = TSDB_CODE_FILE_CORRUPTED; - goto _err; - } - - // decode - n = 0; - uint32_t delimiter; - n += tGetU32(pReader->aBuf[0] + n, &delimiter); - ASSERT(delimiter == TSDB_FILE_DLMT); - - while (n < size - sizeof(TSCKSUM)) { - SDelIdx delIdx; - - n += tGetDelIdx(pReader->aBuf[0] + n, &delIdx); - - if (taosArrayPush(aDelIdx, &delIdx) == NULL) { - code = TSDB_CODE_OUT_OF_MEMORY; - goto _err; - } - } - - ASSERT(n == size - sizeof(TSCKSUM)); - - return code; - -_err: - tsdbError("vgId:%d, read del idx failed since %s", TD_VID(pReader->pTsdb->pVnode), tstrerror(code)); - return code; -} - -// SDataFReader ==================================================== -int32_t tsdbDataFReaderOpen(SDataFReader **ppReader, STsdb *pTsdb, SDFileSet *pSet) { - int32_t code = 0; - SDataFReader *pReader; - char fname[TSDB_FILENAME_LEN]; - - // alloc - pReader = (SDataFReader *)taosMemoryCalloc(1, sizeof(*pReader)); - if (pReader == NULL) { - code = TSDB_CODE_OUT_OF_MEMORY; - goto _err; - } - pReader->pTsdb = pTsdb; - pReader->pSet = pSet; - - // open impl - // head - tsdbHeadFileName(pTsdb, pSet->diskId, pSet->fid, pSet->pHeadF, fname); - pReader->pHeadFD = taosOpenFile(fname, TD_FILE_READ); - if (pReader->pHeadFD == NULL) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } - - // data - tsdbDataFileName(pTsdb, pSet->diskId, pSet->fid, pSet->pDataF, fname); - pReader->pDataFD = taosOpenFile(fname, TD_FILE_READ); - if (pReader->pDataFD == NULL) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } - - // sma - tsdbSmaFileName(pTsdb, pSet->diskId, pSet->fid, pSet->pSmaF, fname); - pReader->pSmaFD = taosOpenFile(fname, TD_FILE_READ); - if (pReader->pSmaFD == NULL) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } - - // sst - for (int32_t iSst = 0; iSst < pSet->nSstF; iSst++) { - tsdbSstFileName(pTsdb, pSet->diskId, pSet->fid, pSet->aSstF[iSst], fname); - pReader->aLastFD[iSst] = taosOpenFile(fname, TD_FILE_READ); - if (pReader->aLastFD[iSst] == NULL) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } - } - - *ppReader = pReader; - return code; - -_err: - tsdbError("vgId:%d, tsdb data file reader open failed since %s", TD_VID(pTsdb->pVnode), tstrerror(code)); - *ppReader = NULL; - return code; -} - -int32_t tsdbDataFReaderClose(SDataFReader **ppReader) { - int32_t code = 0; - if (*ppReader == NULL) goto _exit; - - // head - if (taosCloseFile(&(*ppReader)->pHeadFD) < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } - - // data - if (taosCloseFile(&(*ppReader)->pDataFD) < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } - - // sma - if (taosCloseFile(&(*ppReader)->pSmaFD) < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } - - // sst - for (int32_t iSst = 0; iSst < (*ppReader)->pSet->nSstF; iSst++) { - if (taosCloseFile(&(*ppReader)->aLastFD[iSst]) < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } - } - - for (int32_t iBuf = 0; iBuf < sizeof((*ppReader)->aBuf) / sizeof(uint8_t *); iBuf++) { - tFree((*ppReader)->aBuf[iBuf]); - } - taosMemoryFree(*ppReader); - -_exit: - *ppReader = NULL; - return code; - -_err: - tsdbError("vgId:%d, data file reader close failed since %s", TD_VID((*ppReader)->pTsdb->pVnode), tstrerror(code)); - return code; -} - -int32_t tsdbReadBlockIdx(SDataFReader *pReader, SArray *aBlockIdx) { - int32_t code = 0; - int64_t offset = pReader->pSet->pHeadF->offset; - int64_t size = pReader->pSet->pHeadF->size - offset; - int64_t n; - uint32_t delimiter; - - taosArrayClear(aBlockIdx); - if (size == 0) { goto _exit; } - // alloc - code = tRealloc(&pReader->aBuf[0], size); - if (code) goto _err; - - // seek - if (taosLSeekFile(pReader->pHeadFD, offset, SEEK_SET) < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } - // read - n = taosReadFile(pReader->pHeadFD, pReader->aBuf[0], size); + n = taosReadFile(pFD->pFD, pFD->pBuf, pFD->szPage); if (n < 0) { code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } else if (n < size) { + goto _exit; + } else if (n < pFD->szPage) { code = TSDB_CODE_FILE_CORRUPTED; - goto _err; - } - - // check - if (!taosCheckChecksumWhole(pReader->aBuf[0], size)) { - code = TSDB_CODE_FILE_CORRUPTED; - goto _err; - } - - // decode - n = 0; - n = tGetU32(pReader->aBuf[0] + n, &delimiter); - ASSERT(delimiter == TSDB_FILE_DLMT); - - while (n < size - sizeof(TSCKSUM)) { - SBlockIdx blockIdx; - n += tGetBlockIdx(pReader->aBuf[0] + n, &blockIdx); - - if (taosArrayPush(aBlockIdx, &blockIdx) == NULL) { - code = TSDB_CODE_OUT_OF_MEMORY; - goto _err; - } - } - - ASSERT(n + sizeof(TSCKSUM) == size); - -_exit: - return code; - -_err: - tsdbError("vgId:%d, read block idx failed since %s", TD_VID(pReader->pTsdb->pVnode), tstrerror(code)); - return code; -} - -int32_t tsdbReadSstBlk(SDataFReader *pReader, int32_t iSst, SArray *aSstBlk) { - int32_t code = 0; - int64_t offset = pReader->pSet->aSstF[iSst]->offset; - int64_t size = pReader->pSet->aSstF[iSst]->size - offset; - int64_t n; - uint32_t delimiter; - - taosArrayClear(aSstBlk); - if (size == 0) { goto _exit; } - // alloc - code = tRealloc(&pReader->aBuf[0], size); - if (code) goto _err; - - // seek - if (taosLSeekFile(pReader->aLastFD[iSst], offset, SEEK_SET) < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } - - // read - n = taosReadFile(pReader->aLastFD[iSst], pReader->aBuf[0], size); - if (n < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } else if (n < size) { - code = TSDB_CODE_FILE_CORRUPTED; - goto _err; - } - // check - if (!taosCheckChecksumWhole(pReader->aBuf[0], size)) { + if (!taosCheckChecksumWhole(pFD->pBuf, pFD->szPage)) { code = TSDB_CODE_FILE_CORRUPTED; - goto _err; + goto _exit; } - // decode - n = 0; - n = tGetU32(pReader->aBuf[0] + n, &delimiter); - ASSERT(delimiter == TSDB_FILE_DLMT); - - while (n < size - sizeof(TSCKSUM)) { - SSstBlk blockl; - n += tGetSstBlk(pReader->aBuf[0] + n, &blockl); - - if (taosArrayPush(aSstBlk, &blockl) == NULL) { - code = TSDB_CODE_OUT_OF_MEMORY; - goto _err; - } - } - - ASSERT(n + sizeof(TSCKSUM) == size); + pFD->pgno = pgno; _exit: return code; - -_err: - tsdbError("vgId:%d read blockl failed since %s", TD_VID(pReader->pTsdb->pVnode), tstrerror(code)); - return code; } -int32_t tsdbReadBlock(SDataFReader *pReader, SBlockIdx *pBlockIdx, SMapData *mBlock) { +static int32_t tsdbWriteFile(STsdbFD *pFD, int64_t offset, uint8_t *pBuf, int64_t size) { int32_t code = 0; - int64_t offset = pBlockIdx->offset; - int64_t size = pBlockIdx->size; - int64_t n; - int64_t tn; + int64_t fOffset = LOGIC_TO_FILE_OFFSET(offset, pFD->szPage); + int64_t pgno = OFFSET_PGNO(fOffset, pFD->szPage); + int64_t bOffset = fOffset % pFD->szPage; + int64_t n = 0; - // alloc - code = tRealloc(&pReader->aBuf[0], size); - if (code) goto _err; + do { + if (pFD->pgno != pgno) { + code = tsdbWriteFilePage(pFD); + if (code) goto _exit; - // seek - if (taosLSeekFile(pReader->pHeadFD, offset, SEEK_SET) < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } - - // read - n = taosReadFile(pReader->pHeadFD, pReader->aBuf[0], size); - if (n < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } else if (n < size) { - code = TSDB_CODE_FILE_CORRUPTED; - goto _err; - } - - // check - if (!taosCheckChecksumWhole(pReader->aBuf[0], size)) { - code = TSDB_CODE_FILE_CORRUPTED; - goto _err; - } - - // decode - n = 0; - - uint32_t delimiter; - n += tGetU32(pReader->aBuf[0] + n, &delimiter); - ASSERT(delimiter == TSDB_FILE_DLMT); - - tn = tGetMapData(pReader->aBuf[0] + n, mBlock); - if (tn < 0) { - code = TSDB_CODE_OUT_OF_MEMORY; - goto _err; - } - n += tn; - ASSERT(n + sizeof(TSCKSUM) == size); - - return code; - -_err: - tsdbError("vgId:%d, read block failed since %s", TD_VID(pReader->pTsdb->pVnode), tstrerror(code)); - return code; -} - -int32_t tsdbReadBlockSma(SDataFReader *pReader, SDataBlk *pDataBlk, SArray *aColumnDataAgg) { - int32_t code = 0; - SSmaInfo *pSmaInfo = &pDataBlk->smaInfo; - - ASSERT(pSmaInfo->size > 0); - - taosArrayClear(aColumnDataAgg); - - // alloc - int32_t size = pSmaInfo->size + sizeof(TSCKSUM); - code = tRealloc(&pReader->aBuf[0], size); - if (code) goto _err; - - // seek - int64_t n = taosLSeekFile(pReader->pSmaFD, pSmaInfo->offset, SEEK_SET); - if (n < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } else if (n < pSmaInfo->offset) { - code = TSDB_CODE_FILE_CORRUPTED; - goto _err; - } - - // read - n = taosReadFile(pReader->pSmaFD, pReader->aBuf[0], size); - if (n < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } else if (n < size) { - code = TSDB_CODE_FILE_CORRUPTED; - goto _err; - } - - // check - if (!taosCheckChecksumWhole(pReader->aBuf[0], size)) { - code = TSDB_CODE_FILE_CORRUPTED; - goto _err; - } - - // decode - n = 0; - while (n < pSmaInfo->size) { - SColumnDataAgg sma; - - n += tGetColumnDataAgg(pReader->aBuf[0] + n, &sma); - if (taosArrayPush(aColumnDataAgg, &sma) == NULL) { - code = TSDB_CODE_OUT_OF_MEMORY; - goto _err; + if (pgno <= pFD->szFile) { + code = tsdbReadFilePage(pFD, pgno); + if (code) goto _exit; + } else { + pFD->pgno = pgno; + } } - } - return code; + int64_t nWrite = TMIN(PAGE_CONTENT_SIZE(pFD->szPage) - bOffset, size - n); + memcpy(pFD->pBuf + bOffset, pBuf + n, nWrite); -_err: - tsdbError("vgId:%d tsdb read block sma failed since %s", TD_VID(pReader->pTsdb->pVnode), tstrerror(code)); + pgno++; + bOffset = 0; + n += nWrite; + } while (n < size); + +_exit: return code; } -static int32_t tsdbReadBlockDataImpl(SDataFReader *pReader, SBlockInfo *pBlkInfo, int8_t fromLast, - SBlockData *pBlockData) { +static int32_t tsdbReadFile(STsdbFD *pFD, int64_t offset, uint8_t *pBuf, int64_t size) { int32_t code = 0; + int64_t n = 0; + int64_t fOffset = LOGIC_TO_FILE_OFFSET(offset, pFD->szPage); + int64_t pgno = OFFSET_PGNO(fOffset, pFD->szPage); + int32_t szPgCont = PAGE_CONTENT_SIZE(pFD->szPage); + int64_t bOffset = fOffset % pFD->szPage; - tBlockDataClear(pBlockData); + ASSERT(pgno && pgno <= pFD->szFile); + ASSERT(bOffset < szPgCont); - TdFilePtr pFD = fromLast ? pReader->aLastFD[0] : pReader->pDataFD; // (todo) - - // uid + version + tskey - code = tsdbReadAndCheck(pFD, pBlkInfo->offset, &pReader->aBuf[0], pBlkInfo->szKey, 1); - if (code) goto _err; - SDiskDataHdr hdr; - uint8_t *p = pReader->aBuf[0] + tGetDiskDataHdr(pReader->aBuf[0], &hdr); - - ASSERT(hdr.delimiter == TSDB_FILE_DLMT); - ASSERT(pBlockData->suid == hdr.suid); - ASSERT(pBlockData->uid == hdr.uid); - - pBlockData->nRow = hdr.nRow; - - // uid - if (hdr.uid == 0) { - ASSERT(hdr.szUid); - code = tsdbDecmprData(p, hdr.szUid, TSDB_DATA_TYPE_BIGINT, hdr.cmprAlg, (uint8_t **)&pBlockData->aUid, - sizeof(int64_t) * hdr.nRow, &pReader->aBuf[1]); - if (code) goto _err; - } else { - ASSERT(!hdr.szUid); - } - p += hdr.szUid; - - // version - code = tsdbDecmprData(p, hdr.szVer, TSDB_DATA_TYPE_BIGINT, hdr.cmprAlg, (uint8_t **)&pBlockData->aVersion, - sizeof(int64_t) * hdr.nRow, &pReader->aBuf[1]); - if (code) goto _err; - p += hdr.szVer; - - // TSKEY - code = tsdbDecmprData(p, hdr.szKey, TSDB_DATA_TYPE_TIMESTAMP, hdr.cmprAlg, (uint8_t **)&pBlockData->aTSKEY, - sizeof(TSKEY) * hdr.nRow, &pReader->aBuf[1]); - if (code) goto _err; - p += hdr.szKey; - - ASSERT(p - pReader->aBuf[0] == pBlkInfo->szKey - sizeof(TSCKSUM)); - - // read and decode columns - if (taosArrayGetSize(pBlockData->aIdx) == 0) goto _exit; - - if (hdr.szBlkCol > 0) { - int64_t offset = pBlkInfo->offset + pBlkInfo->szKey; - code = tsdbReadAndCheck(pFD, offset, &pReader->aBuf[0], hdr.szBlkCol + sizeof(TSCKSUM), 1); - if (code) goto _err; - } - - SBlockCol blockCol = {.cid = 0}; - SBlockCol *pBlockCol = &blockCol; - int32_t n = 0; - - for (int32_t iColData = 0; iColData < taosArrayGetSize(pBlockData->aIdx); iColData++) { - SColData *pColData = tBlockDataGetColDataByIdx(pBlockData, iColData); - - while (pBlockCol && pBlockCol->cid < pColData->cid) { - if (n < hdr.szBlkCol) { - n += tGetBlockCol(pReader->aBuf[0] + n, pBlockCol); - } else { - ASSERT(n == hdr.szBlkCol); - pBlockCol = NULL; - } + while (n < size) { + if (pFD->pgno != pgno) { + code = tsdbReadFilePage(pFD, pgno); + if (code) goto _exit; } - if (pBlockCol == NULL || pBlockCol->cid > pColData->cid) { - // add a lot of NONE - for (int32_t iRow = 0; iRow < hdr.nRow; iRow++) { - code = tColDataAppendValue(pColData, &COL_VAL_NONE(pColData->cid, pColData->type)); - if (code) goto _err; - } - } else { - ASSERT(pBlockCol->type == pColData->type); - ASSERT(pBlockCol->flag && pBlockCol->flag != HAS_NONE); + int64_t nRead = TMIN(szPgCont - bOffset, size - n); + memcpy(pBuf + n, pFD->pBuf + bOffset, nRead); - if (pBlockCol->flag == HAS_NULL) { - // add a lot of NULL - for (int32_t iRow = 0; iRow < hdr.nRow; iRow++) { - code = tColDataAppendValue(pColData, &COL_VAL_NULL(pBlockCol->cid, pBlockCol->type)); - if (code) goto _err; - } - } else { - // decode from binary - int64_t offset = pBlkInfo->offset + pBlkInfo->szKey + hdr.szBlkCol + sizeof(TSCKSUM) + pBlockCol->offset; - int32_t size = pBlockCol->szBitmap + pBlockCol->szOffset + pBlockCol->szValue + sizeof(TSCKSUM); - - code = tsdbReadAndCheck(pFD, offset, &pReader->aBuf[1], size, 0); - if (code) goto _err; - - code = tsdbDecmprColData(pReader->aBuf[1], pBlockCol, hdr.cmprAlg, hdr.nRow, pColData, &pReader->aBuf[2]); - if (code) goto _err; - } - } + n += nRead; + pgno++; + bOffset = 0; } _exit: return code; - -_err: - tsdbError("vgId:%d tsdb read block data impl failed since %s", TD_VID(pReader->pTsdb->pVnode), tstrerror(code)); - return code; } -int32_t tsdbReadDataBlock(SDataFReader *pReader, SDataBlk *pDataBlk, SBlockData *pBlockData) { +static int32_t tsdbFsyncFile(STsdbFD *pFD) { int32_t code = 0; - code = tsdbReadBlockDataImpl(pReader, &pDataBlk->aSubBlock[0], 0, pBlockData); - if (code) goto _err; + code = tsdbWriteFilePage(pFD); + if (code) goto _exit; - if (pDataBlk->nSubBlock > 1) { - SBlockData bData1; - SBlockData bData2; - - // create - code = tBlockDataCreate(&bData1); - if (code) goto _err; - code = tBlockDataCreate(&bData2); - if (code) goto _err; - - // init - tBlockDataInitEx(&bData1, pBlockData); - tBlockDataInitEx(&bData2, pBlockData); - - for (int32_t iSubBlock = 1; iSubBlock < pDataBlk->nSubBlock; iSubBlock++) { - code = tsdbReadBlockDataImpl(pReader, &pDataBlk->aSubBlock[iSubBlock], 0, &bData1); - if (code) { - tBlockDataDestroy(&bData1, 1); - tBlockDataDestroy(&bData2, 1); - goto _err; - } - - code = tBlockDataCopy(pBlockData, &bData2); - if (code) { - tBlockDataDestroy(&bData1, 1); - tBlockDataDestroy(&bData2, 1); - goto _err; - } - - code = tBlockDataMerge(&bData1, &bData2, pBlockData); - if (code) { - tBlockDataDestroy(&bData1, 1); - tBlockDataDestroy(&bData2, 1); - goto _err; - } - } - - tBlockDataDestroy(&bData1, 1); - tBlockDataDestroy(&bData2, 1); + if (taosFsyncFile(pFD->pFD) < 0) { + code = TAOS_SYSTEM_ERROR(errno); + goto _exit; } - return code; - -_err: - tsdbError("vgId:%d tsdb read data block failed since %s", TD_VID(pReader->pTsdb->pVnode), tstrerror(code)); - return code; -} - -int32_t tsdbReadSstBlock(SDataFReader *pReader, int32_t iSst, SSstBlk *pSstBlk, SBlockData *pBlockData) { - int32_t code = 0; - - code = tsdbReadBlockDataImpl(pReader, &pSstBlk->bInfo, 1, pBlockData); - if (code) goto _err; - - return code; - -_err: - tsdbError("vgId:%d tsdb read last block failed since %s", TD_VID(pReader->pTsdb->pVnode), tstrerror(code)); - return code; -} - -int32_t tsdbReadSstBlockEx(SDataFReader *pReader, int32_t iSst, SSstBlk *pSstBlk, SBlockData *pBlockData) { - int32_t code = 0; - - // read - code = tsdbReadAndCheck(pReader->aLastFD[iSst], pSstBlk->bInfo.offset, &pReader->aBuf[0], pSstBlk->bInfo.szBlock, 0); - if (code) goto _exit; - - // decmpr - code = tDecmprBlockData(pReader->aBuf[0], pSstBlk->bInfo.szBlock, pBlockData, &pReader->aBuf[1]); - if (code) goto _exit; - _exit: return code; } @@ -930,6 +209,7 @@ int32_t tsdbDataFWriterOpen(SDataFWriter **ppWriter, STsdb *pTsdb, SDFileSet *pS int32_t code = 0; int32_t flag; int64_t n; + int32_t szPage = TSDB_DEFAULT_PAGE_SIZE; SDataFWriter *pWriter = NULL; char fname[TSDB_FILENAME_LEN]; char hdr[TSDB_FHDR_SIZE] = {0}; @@ -941,14 +221,12 @@ int32_t tsdbDataFWriterOpen(SDataFWriter **ppWriter, STsdb *pTsdb, SDFileSet *pS goto _err; } pWriter->pTsdb = pTsdb; - pWriter->wSet = (SDFileSet){ - .diskId = pSet->diskId, - .fid = pSet->fid, - .pHeadF = &pWriter->fHead, - .pDataF = &pWriter->fData, - .pSmaF = &pWriter->fSma, - .nSstF = pSet->nSstF // - }; + pWriter->wSet = (SDFileSet){.diskId = pSet->diskId, + .fid = pSet->fid, + .pHeadF = &pWriter->fHead, + .pDataF = &pWriter->fData, + .pSmaF = &pWriter->fSma, + .nSstF = pSet->nSstF}; pWriter->fHead = *pSet->pHeadF; pWriter->fData = *pSet->pDataF; pWriter->fSma = *pSet->pSmaF; @@ -958,98 +236,54 @@ int32_t tsdbDataFWriterOpen(SDataFWriter **ppWriter, STsdb *pTsdb, SDFileSet *pS } // head - flag = TD_FILE_WRITE | TD_FILE_CREATE | TD_FILE_TRUNC; + flag = TD_FILE_READ | TD_FILE_WRITE | TD_FILE_CREATE | TD_FILE_TRUNC; tsdbHeadFileName(pTsdb, pWriter->wSet.diskId, pWriter->wSet.fid, &pWriter->fHead, fname); - pWriter->pHeadFD = taosOpenFile(fname, flag); - if (pWriter->pHeadFD == NULL) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } - - n = taosWriteFile(pWriter->pHeadFD, hdr, TSDB_FHDR_SIZE); - if (n < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } - - ASSERT(n == TSDB_FHDR_SIZE); + code = tsdbOpenFile(fname, szPage, flag, &pWriter->pHeadFD); + if (code) goto _err; + code = tsdbWriteFile(pWriter->pHeadFD, 0, hdr, TSDB_FHDR_SIZE); + if (code) goto _err; pWriter->fHead.size += TSDB_FHDR_SIZE; // data if (pWriter->fData.size == 0) { - flag = TD_FILE_WRITE | TD_FILE_CREATE | TD_FILE_TRUNC; + flag = TD_FILE_READ | TD_FILE_WRITE | TD_FILE_CREATE | TD_FILE_TRUNC; } else { - flag = TD_FILE_WRITE; + flag = TD_FILE_READ | TD_FILE_WRITE; } tsdbDataFileName(pTsdb, pWriter->wSet.diskId, pWriter->wSet.fid, &pWriter->fData, fname); - pWriter->pDataFD = taosOpenFile(fname, flag); - if (pWriter->pDataFD == NULL) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } + code = tsdbOpenFile(fname, szPage, flag, &pWriter->pDataFD); + if (code) goto _err; if (pWriter->fData.size == 0) { - n = taosWriteFile(pWriter->pDataFD, hdr, TSDB_FHDR_SIZE); - if (n < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } - + code = tsdbWriteFile(pWriter->pDataFD, 0, hdr, TSDB_FHDR_SIZE); + if (code) goto _err; pWriter->fData.size += TSDB_FHDR_SIZE; - } else { - n = taosLSeekFile(pWriter->pDataFD, 0, SEEK_END); - if (n < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } - - ASSERT(n == pWriter->fData.size); } // sma if (pWriter->fSma.size == 0) { - flag = TD_FILE_WRITE | TD_FILE_CREATE | TD_FILE_TRUNC; + flag = TD_FILE_READ | TD_FILE_WRITE | TD_FILE_CREATE | TD_FILE_TRUNC; } else { - flag = TD_FILE_WRITE; + flag = TD_FILE_READ | TD_FILE_WRITE; } tsdbSmaFileName(pTsdb, pWriter->wSet.diskId, pWriter->wSet.fid, &pWriter->fSma, fname); - pWriter->pSmaFD = taosOpenFile(fname, flag); - if (pWriter->pSmaFD == NULL) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } + code = tsdbOpenFile(fname, szPage, flag, &pWriter->pSmaFD); + if (code) goto _err; if (pWriter->fSma.size == 0) { - n = taosWriteFile(pWriter->pSmaFD, hdr, TSDB_FHDR_SIZE); - if (n < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } + code = tsdbWriteFile(pWriter->pSmaFD, 0, hdr, TSDB_FHDR_SIZE); + if (code) goto _err; pWriter->fSma.size += TSDB_FHDR_SIZE; - } else { - n = taosLSeekFile(pWriter->pSmaFD, 0, SEEK_END); - if (n < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } - - ASSERT(n == pWriter->fSma.size); } // sst ASSERT(pWriter->fSst[pSet->nSstF - 1].size == 0); - flag = TD_FILE_WRITE | TD_FILE_CREATE | TD_FILE_TRUNC; + flag = TD_FILE_READ | TD_FILE_WRITE | TD_FILE_CREATE | TD_FILE_TRUNC; tsdbSstFileName(pTsdb, pWriter->wSet.diskId, pWriter->wSet.fid, &pWriter->fSst[pSet->nSstF - 1], fname); - pWriter->pLastFD = taosOpenFile(fname, flag); - if (pWriter->pLastFD == NULL) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } - n = taosWriteFile(pWriter->pLastFD, hdr, TSDB_FHDR_SIZE); - if (n < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } + code = tsdbOpenFile(fname, szPage, flag, &pWriter->pSstFD); + if (code) goto _err; + code = tsdbWriteFile(pWriter->pSstFD, 0, hdr, TSDB_FHDR_SIZE); + if (code) goto _err; pWriter->fSst[pWriter->wSet.nSstF - 1].size += TSDB_FHDR_SIZE; *ppWriter = pWriter; @@ -1069,46 +303,23 @@ int32_t tsdbDataFWriterClose(SDataFWriter **ppWriter, int8_t sync) { pTsdb = (*ppWriter)->pTsdb; if (sync) { - if (taosFsyncFile((*ppWriter)->pHeadFD) < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } + code = tsdbFsyncFile((*ppWriter)->pHeadFD); + if (code) goto _err; - if (taosFsyncFile((*ppWriter)->pDataFD) < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } + code = tsdbFsyncFile((*ppWriter)->pDataFD); + if (code) goto _err; - if (taosFsyncFile((*ppWriter)->pSmaFD) < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } + code = tsdbFsyncFile((*ppWriter)->pSmaFD); + if (code) goto _err; - if (taosFsyncFile((*ppWriter)->pLastFD) < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } + code = tsdbFsyncFile((*ppWriter)->pSstFD); + if (code) goto _err; } - if (taosCloseFile(&(*ppWriter)->pHeadFD) < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } - - if (taosCloseFile(&(*ppWriter)->pDataFD) < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } - - if (taosCloseFile(&(*ppWriter)->pSmaFD) < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } - - if (taosCloseFile(&(*ppWriter)->pLastFD) < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } + tsdbCloseFile(&(*ppWriter)->pHeadFD); + tsdbCloseFile(&(*ppWriter)->pDataFD); + tsdbCloseFile(&(*ppWriter)->pSmaFD); + tsdbCloseFile(&(*ppWriter)->pSstFD); for (int32_t iBuf = 0; iBuf < sizeof((*ppWriter)->aBuf) / sizeof(uint8_t *); iBuf++) { tFree((*ppWriter)->aBuf[iBuf]); @@ -1131,70 +342,26 @@ int32_t tsdbUpdateDFileSetHeader(SDataFWriter *pWriter) { // head ============== memset(hdr, 0, TSDB_FHDR_SIZE); tPutHeadFile(hdr, &pWriter->fHead); - taosCalcChecksumAppend(0, hdr, TSDB_FHDR_SIZE); - - n = taosLSeekFile(pWriter->pHeadFD, 0, SEEK_SET); - if (n < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } - - n = taosWriteFile(pWriter->pHeadFD, hdr, TSDB_FHDR_SIZE); - if (n < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } + code = tsdbWriteFile(pWriter->pHeadFD, 0, hdr, TSDB_FHDR_SIZE); + if (code) goto _err; // data ============== memset(hdr, 0, TSDB_FHDR_SIZE); tPutDataFile(hdr, &pWriter->fData); - taosCalcChecksumAppend(0, hdr, TSDB_FHDR_SIZE); - - n = taosLSeekFile(pWriter->pDataFD, 0, SEEK_SET); - if (n < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } - - n = taosWriteFile(pWriter->pDataFD, hdr, TSDB_FHDR_SIZE); - if (n < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } + code = tsdbWriteFile(pWriter->pDataFD, 0, hdr, TSDB_FHDR_SIZE); + if (code) goto _err; // sma ============== memset(hdr, 0, TSDB_FHDR_SIZE); tPutSmaFile(hdr, &pWriter->fSma); - taosCalcChecksumAppend(0, hdr, TSDB_FHDR_SIZE); - - n = taosLSeekFile(pWriter->pSmaFD, 0, SEEK_SET); - if (n < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } - - n = taosWriteFile(pWriter->pSmaFD, hdr, TSDB_FHDR_SIZE); - if (n < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } + code = tsdbWriteFile(pWriter->pSmaFD, 0, hdr, TSDB_FHDR_SIZE); + if (code) goto _err; // sst ============== memset(hdr, 0, TSDB_FHDR_SIZE); tPutSstFile(hdr, &pWriter->fSst[pWriter->wSet.nSstF - 1]); - taosCalcChecksumAppend(0, hdr, TSDB_FHDR_SIZE); - - n = taosLSeekFile(pWriter->pLastFD, 0, SEEK_SET); - if (n < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } - - n = taosWriteFile(pWriter->pLastFD, hdr, TSDB_FHDR_SIZE); - if (n < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } + code = tsdbWriteFile(pWriter->pSstFD, 0, hdr, TSDB_FHDR_SIZE); + if (code) goto _err; return code; @@ -1206,7 +373,7 @@ _err: int32_t tsdbWriteBlockIdx(SDataFWriter *pWriter, SArray *aBlockIdx) { int32_t code = 0; SHeadFile *pHeadFile = &pWriter->fHead; - int64_t size = 0; + int64_t size; int64_t n; // check @@ -1216,11 +383,10 @@ int32_t tsdbWriteBlockIdx(SDataFWriter *pWriter, SArray *aBlockIdx) { } // prepare - size = sizeof(uint32_t); + size = 0; for (int32_t iBlockIdx = 0; iBlockIdx < taosArrayGetSize(aBlockIdx); iBlockIdx++) { size += tPutBlockIdx(NULL, taosArrayGet(aBlockIdx, iBlockIdx)); } - size += sizeof(TSCKSUM); // alloc code = tRealloc(&pWriter->aBuf[0], size); @@ -1228,28 +394,23 @@ int32_t tsdbWriteBlockIdx(SDataFWriter *pWriter, SArray *aBlockIdx) { // build n = 0; - n = tPutU32(pWriter->aBuf[0] + n, TSDB_FILE_DLMT); for (int32_t iBlockIdx = 0; iBlockIdx < taosArrayGetSize(aBlockIdx); iBlockIdx++) { n += tPutBlockIdx(pWriter->aBuf[0] + n, taosArrayGet(aBlockIdx, iBlockIdx)); } - taosCalcChecksumAppend(0, pWriter->aBuf[0], size); - - ASSERT(n + sizeof(TSCKSUM) == size); + ASSERT(n == size); // write - n = taosWriteFile(pWriter->pHeadFD, pWriter->aBuf[0], size); - if (n < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } + code = tsdbWriteFile(pWriter->pHeadFD, pHeadFile->size, pWriter->aBuf[0], size); + if (code) goto _err; // update pHeadFile->offset = pHeadFile->size; pHeadFile->size += size; _exit: - tsdbTrace("vgId:%d write block idx, offset:%" PRId64 " size:%" PRId64 " nBlockIdx:%d", TD_VID(pWriter->pTsdb->pVnode), - pHeadFile->offset, size, taosArrayGetSize(aBlockIdx)); + // tsdbTrace("vgId:%d write block idx, offset:%" PRId64 " size:%" PRId64 " nBlockIdx:%d", + // TD_VID(pWriter->pTsdb->pVnode), + // pHeadFile->offset, size, taosArrayGetSize(aBlockIdx)); return code; _err: @@ -1266,24 +427,16 @@ int32_t tsdbWriteBlock(SDataFWriter *pWriter, SMapData *mBlock, SBlockIdx *pBloc ASSERT(mBlock->nItem > 0); // alloc - size = sizeof(uint32_t) + tPutMapData(NULL, mBlock) + sizeof(TSCKSUM); + size = tPutMapData(NULL, mBlock); code = tRealloc(&pWriter->aBuf[0], size); if (code) goto _err; // build - n = 0; - n += tPutU32(pWriter->aBuf[0] + n, TSDB_FILE_DLMT); - n += tPutMapData(pWriter->aBuf[0] + n, mBlock); - taosCalcChecksumAppend(0, pWriter->aBuf[0], size); - - ASSERT(n + sizeof(TSCKSUM) == size); + n = tPutMapData(pWriter->aBuf[0], mBlock); // write - n = taosWriteFile(pWriter->pHeadFD, pWriter->aBuf[0], size); - if (n < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } + code = tsdbWriteFile(pWriter->pHeadFD, pHeadFile->size, pWriter->aBuf[0], size); + if (code) goto _err; // update pBlockIdx->offset = pHeadFile->size; @@ -1304,7 +457,7 @@ _err: int32_t tsdbWriteSstBlk(SDataFWriter *pWriter, SArray *aSstBlk) { int32_t code = 0; SSstFile *pSstFile = &pWriter->fSst[pWriter->wSet.nSstF - 1]; - int64_t size = 0; + int64_t size; int64_t n; // check @@ -1314,11 +467,10 @@ int32_t tsdbWriteSstBlk(SDataFWriter *pWriter, SArray *aSstBlk) { } // size - size = sizeof(uint32_t); // TSDB_FILE_DLMT + size = 0; for (int32_t iBlockL = 0; iBlockL < taosArrayGetSize(aSstBlk); iBlockL++) { size += tPutSstBlk(NULL, taosArrayGet(aSstBlk, iBlockL)); } - size += sizeof(TSCKSUM); // alloc code = tRealloc(&pWriter->aBuf[0], size); @@ -1326,27 +478,20 @@ int32_t tsdbWriteSstBlk(SDataFWriter *pWriter, SArray *aSstBlk) { // encode n = 0; - n += tPutU32(pWriter->aBuf[0] + n, TSDB_FILE_DLMT); for (int32_t iBlockL = 0; iBlockL < taosArrayGetSize(aSstBlk); iBlockL++) { n += tPutSstBlk(pWriter->aBuf[0] + n, taosArrayGet(aSstBlk, iBlockL)); } - taosCalcChecksumAppend(0, pWriter->aBuf[0], size); - - ASSERT(n + sizeof(TSCKSUM) == size); // write - n = taosWriteFile(pWriter->pLastFD, pWriter->aBuf[0], size); - if (n < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } + code = tsdbWriteFile(pWriter->pSstFD, pSstFile->size, pWriter->aBuf[0], size); + if (code) goto _err; // update pSstFile->offset = pSstFile->size; pSstFile->size += size; _exit: - tsdbTrace("vgId:%d tsdb write blockl, loffset:%" PRId64 " size:%" PRId64, TD_VID(pWriter->pTsdb->pVnode), + tsdbTrace("vgId:%d tsdb write sst block, loffset:%" PRId64 " size:%" PRId64, TD_VID(pWriter->pTsdb->pVnode), pSstFile->offset, size); return code; @@ -1355,30 +500,6 @@ _err: return code; } -static void tsdbUpdateBlockInfo(SBlockData *pBlockData, SDataBlk *pDataBlk) { - for (int32_t iRow = 0; iRow < pBlockData->nRow; iRow++) { - TSDBKEY key = {.ts = pBlockData->aTSKEY[iRow], .version = pBlockData->aVersion[iRow]}; - - if (iRow == 0) { - if (tsdbKeyCmprFn(&pDataBlk->minKey, &key) > 0) { - pDataBlk->minKey = key; - } - } else { - if (pBlockData->aTSKEY[iRow] == pBlockData->aTSKEY[iRow - 1]) { - pDataBlk->hasDup = 1; - } - } - - if (iRow == pBlockData->nRow - 1 && tsdbKeyCmprFn(&pDataBlk->maxKey, &key) < 0) { - pDataBlk->maxKey = key; - } - - pDataBlk->minVer = TMIN(pDataBlk->minVer, key.version); - pDataBlk->maxVer = TMAX(pDataBlk->maxVer, key.version); - } - pDataBlk->nRow += pBlockData->nRow; -} - static int32_t tsdbWriteBlockSma(SDataFWriter *pWriter, SBlockData *pBlockData, SSmaInfo *pSmaInfo) { int32_t code = 0; @@ -1401,21 +522,14 @@ static int32_t tsdbWriteBlockSma(SDataFWriter *pWriter, SBlockData *pBlockData, // write if (pSmaInfo->size) { - int32_t size = pSmaInfo->size + sizeof(TSCKSUM); - - code = tRealloc(&pWriter->aBuf[0], size); + code = tRealloc(&pWriter->aBuf[0], pSmaInfo->size); if (code) goto _err; - taosCalcChecksumAppend(0, pWriter->aBuf[0], size); - - int64_t n = taosWriteFile(pWriter->pSmaFD, pWriter->aBuf[0], size); - if (n < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } + code = tsdbWriteFile(pWriter->pSmaFD, pWriter->fSma.size, pWriter->aBuf[0], pSmaInfo->size); + if (code) goto _err; pSmaInfo->offset = pWriter->fSma.size; - pWriter->fSma.size += size; + pWriter->fSma.size += pSmaInfo->size; } return code; @@ -1431,7 +545,11 @@ int32_t tsdbWriteBlockData(SDataFWriter *pWriter, SBlockData *pBlockData, SBlock ASSERT(pBlockData->nRow > 0); - pBlkInfo->offset = toLast ? pWriter->fSst[pWriter->wSet.nSstF - 1].size : pWriter->fData.size; + if (toLast) { + pBlkInfo->offset = pWriter->fSst[pWriter->wSet.nSstF - 1].size; + } else { + pBlkInfo->offset = pWriter->fData.size; + } pBlkInfo->szBlock = 0; pBlkInfo->szKey = 0; @@ -1440,37 +558,29 @@ int32_t tsdbWriteBlockData(SDataFWriter *pWriter, SBlockData *pBlockData, SBlock if (code) goto _err; // write ================= - TdFilePtr pFD = toLast ? pWriter->pLastFD : pWriter->pDataFD; + STsdbFD *pFD = toLast ? pWriter->pSstFD : pWriter->pDataFD; pBlkInfo->szKey = aBufN[3] + aBufN[2]; pBlkInfo->szBlock = aBufN[0] + aBufN[1] + aBufN[2] + aBufN[3]; - int64_t n = taosWriteFile(pFD, pWriter->aBuf[3], aBufN[3]); - if (n < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } + int64_t offset = pBlkInfo->offset; + code = tsdbWriteFile(pFD, offset, pWriter->aBuf[3], aBufN[3]); + if (code) goto _err; + offset += aBufN[3]; - n = taosWriteFile(pFD, pWriter->aBuf[2], aBufN[2]); - if (n < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } + code = tsdbWriteFile(pFD, offset, pWriter->aBuf[2], aBufN[2]); + if (code) goto _err; + offset += aBufN[2]; if (aBufN[1]) { - n = taosWriteFile(pFD, pWriter->aBuf[1], aBufN[1]); - if (n < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } + code = tsdbWriteFile(pFD, offset, pWriter->aBuf[1], aBufN[1]); + if (code) goto _err; + offset += aBufN[1]; } if (aBufN[0]) { - n = taosWriteFile(pFD, pWriter->aBuf[0], aBufN[0]); - if (n < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _err; - } + code = tsdbWriteFile(pFD, offset, pWriter->aBuf[0], aBufN[0]); + if (code) goto _err; } // update info @@ -1608,3 +718,702 @@ _err: tsdbError("vgId:%d, tsdb DFileSet copy failed since %s", TD_VID(pTsdb->pVnode), tstrerror(code)); return code; } + +// SDataFReader ==================================================== +int32_t tsdbDataFReaderOpen(SDataFReader **ppReader, STsdb *pTsdb, SDFileSet *pSet) { + int32_t code = 0; + SDataFReader *pReader; + int32_t szPage = TSDB_DEFAULT_PAGE_SIZE; + char fname[TSDB_FILENAME_LEN]; + + // alloc + pReader = (SDataFReader *)taosMemoryCalloc(1, sizeof(*pReader)); + if (pReader == NULL) { + code = TSDB_CODE_OUT_OF_MEMORY; + goto _err; + } + pReader->pTsdb = pTsdb; + pReader->pSet = pSet; + + // head + tsdbHeadFileName(pTsdb, pSet->diskId, pSet->fid, pSet->pHeadF, fname); + code = tsdbOpenFile(fname, szPage, TD_FILE_READ, &pReader->pHeadFD); + if (code) goto _err; + + // data + tsdbDataFileName(pTsdb, pSet->diskId, pSet->fid, pSet->pDataF, fname); + code = tsdbOpenFile(fname, szPage, TD_FILE_READ, &pReader->pDataFD); + if (code) goto _err; + + // sma + tsdbSmaFileName(pTsdb, pSet->diskId, pSet->fid, pSet->pSmaF, fname); + code = tsdbOpenFile(fname, szPage, TD_FILE_READ, &pReader->pSmaFD); + if (code) goto _err; + + // sst + for (int32_t iSst = 0; iSst < pSet->nSstF; iSst++) { + tsdbSstFileName(pTsdb, pSet->diskId, pSet->fid, pSet->aSstF[iSst], fname); + code = tsdbOpenFile(fname, szPage, TD_FILE_READ, &pReader->aSstFD[iSst]); + if (code) goto _err; + } + + *ppReader = pReader; + return code; + +_err: + tsdbError("vgId:%d, tsdb data file reader open failed since %s", TD_VID(pTsdb->pVnode), tstrerror(code)); + *ppReader = NULL; + return code; +} + +int32_t tsdbDataFReaderClose(SDataFReader **ppReader) { + int32_t code = 0; + if (*ppReader == NULL) return code; + + // head + tsdbCloseFile(&(*ppReader)->pHeadFD); + + // data + tsdbCloseFile(&(*ppReader)->pDataFD); + + // sma + tsdbCloseFile(&(*ppReader)->pSmaFD); + + // sst + for (int32_t iSst = 0; iSst < TSDB_MAX_SST_FILE; iSst++) { + if ((*ppReader)->aSstFD[iSst]) { + tsdbCloseFile(&(*ppReader)->aSstFD[iSst]); + } + } + + for (int32_t iBuf = 0; iBuf < sizeof((*ppReader)->aBuf) / sizeof(uint8_t *); iBuf++) { + tFree((*ppReader)->aBuf[iBuf]); + } + taosMemoryFree(*ppReader); + *ppReader = NULL; + return code; + +_err: + tsdbError("vgId:%d, data file reader close failed since %s", TD_VID((*ppReader)->pTsdb->pVnode), tstrerror(code)); + return code; +} + +int32_t tsdbReadBlockIdx(SDataFReader *pReader, SArray *aBlockIdx) { + int32_t code = 0; + SHeadFile *pHeadFile = pReader->pSet->pHeadF; + int64_t offset = pHeadFile->offset; + int64_t size = pHeadFile->size - offset; + + taosArrayClear(aBlockIdx); + if (size == 0) return code; + + // alloc + code = tRealloc(&pReader->aBuf[0], size); + if (code) goto _err; + + // read + code = tsdbReadFile(pReader->pHeadFD, offset, pReader->aBuf[0], size); + if (code) goto _err; + + // decode + int64_t n = 0; + while (n < size) { + SBlockIdx blockIdx; + n += tGetBlockIdx(pReader->aBuf[0] + n, &blockIdx); + + if (taosArrayPush(aBlockIdx, &blockIdx) == NULL) { + code = TSDB_CODE_OUT_OF_MEMORY; + goto _err; + } + } + ASSERT(n == size); + + return code; + +_err: + tsdbError("vgId:%d, read block idx failed since %s", TD_VID(pReader->pTsdb->pVnode), tstrerror(code)); + return code; +} + +int32_t tsdbReadSstBlk(SDataFReader *pReader, int32_t iSst, SArray *aSstBlk) { + int32_t code = 0; + SSstFile *pSstFile = pReader->pSet->aSstF[iSst]; + int64_t offset = pSstFile->offset; + int64_t size = pSstFile->size - offset; + + taosArrayClear(aSstBlk); + if (size == 0) return code; + + // alloc + code = tRealloc(&pReader->aBuf[0], size); + if (code) goto _err; + + // read + code = tsdbReadFile(pReader->aSstFD[iSst], offset, pReader->aBuf[0], size); + if (code) goto _err; + + // decode + int64_t n = 0; + while (n < size) { + SSstBlk sstBlk; + n += tGetSstBlk(pReader->aBuf[0] + n, &sstBlk); + + if (taosArrayPush(aSstBlk, &sstBlk) == NULL) { + code = TSDB_CODE_OUT_OF_MEMORY; + goto _err; + } + } + ASSERT(n == size); + + return code; + +_err: + tsdbError("vgId:%d read sst blk failed since %s", TD_VID(pReader->pTsdb->pVnode), tstrerror(code)); + return code; +} + +int32_t tsdbReadBlock(SDataFReader *pReader, SBlockIdx *pBlockIdx, SMapData *mBlock) { + int32_t code = 0; + int64_t offset = pBlockIdx->offset; + int64_t size = pBlockIdx->size; + + // alloc + code = tRealloc(&pReader->aBuf[0], size); + if (code) goto _err; + + // read + code = tsdbReadFile(pReader->pHeadFD, offset, pReader->aBuf[0], size); + if (code) goto _err; + + // decode + int64_t n = tGetMapData(pReader->aBuf[0], mBlock); + if (n < 0) { + code = TSDB_CODE_OUT_OF_MEMORY; + goto _err; + } + ASSERT(n == size); + + return code; + +_err: + tsdbError("vgId:%d, read block failed since %s", TD_VID(pReader->pTsdb->pVnode), tstrerror(code)); + return code; +} + +int32_t tsdbReadBlockSma(SDataFReader *pReader, SDataBlk *pDataBlk, SArray *aColumnDataAgg) { + int32_t code = 0; + SSmaInfo *pSmaInfo = &pDataBlk->smaInfo; + + ASSERT(pSmaInfo->size > 0); + + taosArrayClear(aColumnDataAgg); + + // alloc + code = tRealloc(&pReader->aBuf[0], pSmaInfo->size); + if (code) goto _err; + + // read + code = tsdbReadFile(pReader->pSmaFD, pSmaInfo->offset, pReader->aBuf[0], pSmaInfo->size); + if (code) goto _err; + + // decode + int32_t n = 0; + while (n < pSmaInfo->size) { + SColumnDataAgg sma; + n += tGetColumnDataAgg(pReader->aBuf[0] + n, &sma); + + if (taosArrayPush(aColumnDataAgg, &sma) == NULL) { + code = TSDB_CODE_OUT_OF_MEMORY; + goto _err; + } + } + ASSERT(n == pSmaInfo->size); + return code; + +_err: + tsdbError("vgId:%d tsdb read block sma failed since %s", TD_VID(pReader->pTsdb->pVnode), tstrerror(code)); + return code; +} + +static int32_t tsdbReadBlockDataImpl(SDataFReader *pReader, SBlockInfo *pBlkInfo, SBlockData *pBlockData) { + int32_t code = 0; + + tBlockDataClear(pBlockData); + + STsdbFD *pFD = pReader->pDataFD; + + // uid + version + tskey + code = tRealloc(&pReader->aBuf[0], pBlkInfo->szKey); + if (code) goto _err; + + code = tsdbReadFile(pFD, pBlkInfo->offset, pReader->aBuf[0], pBlkInfo->szKey); + if (code) goto _err; + + SDiskDataHdr hdr; + uint8_t *p = pReader->aBuf[0] + tGetDiskDataHdr(pReader->aBuf[0], &hdr); + + ASSERT(hdr.delimiter == TSDB_FILE_DLMT); + ASSERT(pBlockData->suid == hdr.suid); + ASSERT(pBlockData->uid == hdr.uid); + + pBlockData->nRow = hdr.nRow; + + // uid + if (hdr.uid == 0) { + ASSERT(hdr.szUid); + code = tsdbDecmprData(p, hdr.szUid, TSDB_DATA_TYPE_BIGINT, hdr.cmprAlg, (uint8_t **)&pBlockData->aUid, + sizeof(int64_t) * hdr.nRow, &pReader->aBuf[1]); + if (code) goto _err; + } else { + ASSERT(!hdr.szUid); + } + p += hdr.szUid; + + // version + code = tsdbDecmprData(p, hdr.szVer, TSDB_DATA_TYPE_BIGINT, hdr.cmprAlg, (uint8_t **)&pBlockData->aVersion, + sizeof(int64_t) * hdr.nRow, &pReader->aBuf[1]); + if (code) goto _err; + p += hdr.szVer; + + // TSKEY + code = tsdbDecmprData(p, hdr.szKey, TSDB_DATA_TYPE_TIMESTAMP, hdr.cmprAlg, (uint8_t **)&pBlockData->aTSKEY, + sizeof(TSKEY) * hdr.nRow, &pReader->aBuf[1]); + if (code) goto _err; + p += hdr.szKey; + + ASSERT(p - pReader->aBuf[0] == pBlkInfo->szKey); + + // read and decode columns + if (taosArrayGetSize(pBlockData->aIdx) == 0) goto _exit; + + if (hdr.szBlkCol > 0) { + int64_t offset = pBlkInfo->offset + pBlkInfo->szKey; + + code = tRealloc(&pReader->aBuf[0], hdr.szBlkCol); + if (code) goto _err; + + code = tsdbReadFile(pFD, offset, pReader->aBuf[0], hdr.szBlkCol); + if (code) goto _err; + } + + SBlockCol blockCol = {.cid = 0}; + SBlockCol *pBlockCol = &blockCol; + int32_t n = 0; + + for (int32_t iColData = 0; iColData < taosArrayGetSize(pBlockData->aIdx); iColData++) { + SColData *pColData = tBlockDataGetColDataByIdx(pBlockData, iColData); + + while (pBlockCol && pBlockCol->cid < pColData->cid) { + if (n < hdr.szBlkCol) { + n += tGetBlockCol(pReader->aBuf[0] + n, pBlockCol); + } else { + ASSERT(n == hdr.szBlkCol); + pBlockCol = NULL; + } + } + + if (pBlockCol == NULL || pBlockCol->cid > pColData->cid) { + // add a lot of NONE + for (int32_t iRow = 0; iRow < hdr.nRow; iRow++) { + code = tColDataAppendValue(pColData, &COL_VAL_NONE(pColData->cid, pColData->type)); + if (code) goto _err; + } + } else { + ASSERT(pBlockCol->type == pColData->type); + ASSERT(pBlockCol->flag && pBlockCol->flag != HAS_NONE); + + if (pBlockCol->flag == HAS_NULL) { + // add a lot of NULL + for (int32_t iRow = 0; iRow < hdr.nRow; iRow++) { + code = tColDataAppendValue(pColData, &COL_VAL_NULL(pBlockCol->cid, pBlockCol->type)); + if (code) goto _err; + } + } else { + // decode from binary + int64_t offset = pBlkInfo->offset + pBlkInfo->szKey + hdr.szBlkCol + pBlockCol->offset; + int32_t size = pBlockCol->szBitmap + pBlockCol->szOffset + pBlockCol->szValue; + + code = tRealloc(&pReader->aBuf[1], size); + if (code) goto _err; + + code = tsdbReadFile(pFD, offset, pReader->aBuf[1], size); + if (code) goto _err; + + code = tsdbDecmprColData(pReader->aBuf[1], pBlockCol, hdr.cmprAlg, hdr.nRow, pColData, &pReader->aBuf[2]); + if (code) goto _err; + } + } + } + +_exit: + return code; + +_err: + tsdbError("vgId:%d tsdb read block data impl failed since %s", TD_VID(pReader->pTsdb->pVnode), tstrerror(code)); + return code; +} + +int32_t tsdbReadDataBlock(SDataFReader *pReader, SDataBlk *pDataBlk, SBlockData *pBlockData) { + int32_t code = 0; + + code = tsdbReadBlockDataImpl(pReader, &pDataBlk->aSubBlock[0], pBlockData); + if (code) goto _err; + + if (pDataBlk->nSubBlock > 1) { + SBlockData bData1; + SBlockData bData2; + + // create + code = tBlockDataCreate(&bData1); + if (code) goto _err; + code = tBlockDataCreate(&bData2); + if (code) goto _err; + + // init + tBlockDataInitEx(&bData1, pBlockData); + tBlockDataInitEx(&bData2, pBlockData); + + for (int32_t iSubBlock = 1; iSubBlock < pDataBlk->nSubBlock; iSubBlock++) { + code = tsdbReadBlockDataImpl(pReader, &pDataBlk->aSubBlock[iSubBlock], &bData1); + if (code) { + tBlockDataDestroy(&bData1, 1); + tBlockDataDestroy(&bData2, 1); + goto _err; + } + + code = tBlockDataCopy(pBlockData, &bData2); + if (code) { + tBlockDataDestroy(&bData1, 1); + tBlockDataDestroy(&bData2, 1); + goto _err; + } + + code = tBlockDataMerge(&bData1, &bData2, pBlockData); + if (code) { + tBlockDataDestroy(&bData1, 1); + tBlockDataDestroy(&bData2, 1); + goto _err; + } + } + + tBlockDataDestroy(&bData1, 1); + tBlockDataDestroy(&bData2, 1); + } + + return code; + +_err: + tsdbError("vgId:%d tsdb read data block failed since %s", TD_VID(pReader->pTsdb->pVnode), tstrerror(code)); + return code; +} + +int32_t tsdbReadSstBlock(SDataFReader *pReader, int32_t iSst, SSstBlk *pSstBlk, SBlockData *pBlockData) { + int32_t code = 0; + + // alloc + code = tRealloc(&pReader->aBuf[0], pSstBlk->bInfo.szBlock); + if (code) goto _err; + + // read + code = tsdbReadFile(pReader->aSstFD[iSst], pSstBlk->bInfo.offset, pReader->aBuf[0], pSstBlk->bInfo.szBlock); + if (code) goto _err; + + // decmpr + code = tDecmprBlockData(pReader->aBuf[0], pSstBlk->bInfo.szBlock, pBlockData, &pReader->aBuf[1]); + if (code) goto _err; + + return code; + +_err: + tsdbError("vgId:%d tsdb read sst block failed since %s", TD_VID(pReader->pTsdb->pVnode), tstrerror(code)); + return code; +} + +// SDelFWriter ==================================================== +int32_t tsdbDelFWriterOpen(SDelFWriter **ppWriter, SDelFile *pFile, STsdb *pTsdb) { + int32_t code = 0; + char fname[TSDB_FILENAME_LEN]; + uint8_t hdr[TSDB_FHDR_SIZE] = {0}; + SDelFWriter *pDelFWriter; + int64_t n; + + // alloc + pDelFWriter = (SDelFWriter *)taosMemoryCalloc(1, sizeof(*pDelFWriter)); + if (pDelFWriter == NULL) { + code = TSDB_CODE_OUT_OF_MEMORY; + goto _err; + } + pDelFWriter->pTsdb = pTsdb; + pDelFWriter->fDel = *pFile; + + tsdbDelFileName(pTsdb, pFile, fname); + code = + tsdbOpenFile(fname, TSDB_DEFAULT_PAGE_SIZE, TD_FILE_READ | TD_FILE_WRITE | TD_FILE_CREATE, &pDelFWriter->pWriteH); + if (code) goto _err; + + // update header + code = tsdbWriteFile(pDelFWriter->pWriteH, 0, hdr, TSDB_FHDR_SIZE); + if (code) goto _err; + + pDelFWriter->fDel.size = TSDB_FHDR_SIZE; + pDelFWriter->fDel.offset = 0; + + *ppWriter = pDelFWriter; + return code; + +_err: + tsdbError("vgId:%d, failed to open del file writer since %s", TD_VID(pTsdb->pVnode), tstrerror(code)); + *ppWriter = NULL; + return code; +} + +int32_t tsdbDelFWriterClose(SDelFWriter **ppWriter, int8_t sync) { + int32_t code = 0; + SDelFWriter *pWriter = *ppWriter; + STsdb *pTsdb = pWriter->pTsdb; + + // sync + if (sync) { + code = tsdbFsyncFile(pWriter->pWriteH); + if (code) goto _err; + } + + // close + tsdbCloseFile(&pWriter->pWriteH); + + for (int32_t iBuf = 0; iBuf < sizeof(pWriter->aBuf) / sizeof(uint8_t *); iBuf++) { + tFree(pWriter->aBuf[iBuf]); + } + taosMemoryFree(pWriter); + + *ppWriter = NULL; + return code; + +_err: + tsdbError("vgId:%d, failed to close del file writer since %s", TD_VID(pTsdb->pVnode), tstrerror(code)); + return code; +} + +int32_t tsdbWriteDelData(SDelFWriter *pWriter, SArray *aDelData, SDelIdx *pDelIdx) { + int32_t code = 0; + int64_t size; + int64_t n; + + // prepare + size = 0; + for (int32_t iDelData = 0; iDelData < taosArrayGetSize(aDelData); iDelData++) { + size += tPutDelData(NULL, taosArrayGet(aDelData, iDelData)); + } + + // alloc + code = tRealloc(&pWriter->aBuf[0], size); + if (code) goto _err; + + // build + n = 0; + for (int32_t iDelData = 0; iDelData < taosArrayGetSize(aDelData); iDelData++) { + n += tPutDelData(pWriter->aBuf[0] + n, taosArrayGet(aDelData, iDelData)); + } + ASSERT(n == size); + + // write + code = tsdbWriteFile(pWriter->pWriteH, pWriter->fDel.size, pWriter->aBuf[0], size); + if (code) goto _err; + + // update + pDelIdx->offset = pWriter->fDel.size; + pDelIdx->size = size; + pWriter->fDel.size += size; + + return code; + +_err: + tsdbError("vgId:%d, failed to write del data since %s", TD_VID(pWriter->pTsdb->pVnode), tstrerror(code)); + return code; +} + +int32_t tsdbWriteDelIdx(SDelFWriter *pWriter, SArray *aDelIdx) { + int32_t code = 0; + int64_t size; + int64_t n; + SDelIdx *pDelIdx; + + // prepare + size = 0; + for (int32_t iDelIdx = 0; iDelIdx < taosArrayGetSize(aDelIdx); iDelIdx++) { + size += tPutDelIdx(NULL, taosArrayGet(aDelIdx, iDelIdx)); + } + + // alloc + code = tRealloc(&pWriter->aBuf[0], size); + if (code) goto _err; + + // build + n = 0; + for (int32_t iDelIdx = 0; iDelIdx < taosArrayGetSize(aDelIdx); iDelIdx++) { + n += tPutDelIdx(pWriter->aBuf[0] + n, taosArrayGet(aDelIdx, iDelIdx)); + } + ASSERT(n == size); + + // write + code = tsdbWriteFile(pWriter->pWriteH, pWriter->fDel.size, pWriter->aBuf[0], size); + if (code) goto _err; + + // update + pWriter->fDel.offset = pWriter->fDel.size; + pWriter->fDel.size += size; + + return code; + +_err: + tsdbError("vgId:%d, write del idx failed since %s", TD_VID(pWriter->pTsdb->pVnode), tstrerror(code)); + return code; +} + +int32_t tsdbUpdateDelFileHdr(SDelFWriter *pWriter) { + int32_t code = 0; + char hdr[TSDB_FHDR_SIZE] = {0}; + int64_t size = TSDB_FHDR_SIZE; + int64_t n; + + // build + tPutDelFile(hdr, &pWriter->fDel); + + // write + code = tsdbWriteFile(pWriter->pWriteH, 0, hdr, size); + if (code) goto _err; + + return code; + +_err: + tsdbError("vgId:%d, update del file hdr failed since %s", TD_VID(pWriter->pTsdb->pVnode), tstrerror(code)); + return code; +} +// SDelFReader ==================================================== +struct SDelFReader { + STsdb *pTsdb; + SDelFile fDel; + STsdbFD *pReadH; + uint8_t *aBuf[1]; +}; + +int32_t tsdbDelFReaderOpen(SDelFReader **ppReader, SDelFile *pFile, STsdb *pTsdb) { + int32_t code = 0; + char fname[TSDB_FILENAME_LEN]; + SDelFReader *pDelFReader; + int64_t n; + + // alloc + pDelFReader = (SDelFReader *)taosMemoryCalloc(1, sizeof(*pDelFReader)); + if (pDelFReader == NULL) { + code = TSDB_CODE_OUT_OF_MEMORY; + goto _err; + } + + // open impl + pDelFReader->pTsdb = pTsdb; + pDelFReader->fDel = *pFile; + + tsdbDelFileName(pTsdb, pFile, fname); + code = tsdbOpenFile(fname, TSDB_DEFAULT_PAGE_SIZE, TD_FILE_READ, &pDelFReader->pReadH); + if (code) goto _err; + + *ppReader = pDelFReader; + return code; + +_err: + tsdbError("vgId:%d, del file reader open failed since %s", TD_VID(pTsdb->pVnode), tstrerror(code)); + *ppReader = NULL; + return code; +} + +int32_t tsdbDelFReaderClose(SDelFReader **ppReader) { + int32_t code = 0; + SDelFReader *pReader = *ppReader; + + if (pReader) { + tsdbCloseFile(&pReader->pReadH); + for (int32_t iBuf = 0; iBuf < sizeof(pReader->aBuf) / sizeof(uint8_t *); iBuf++) { + tFree(pReader->aBuf[iBuf]); + } + taosMemoryFree(pReader); + } + *ppReader = NULL; + +_exit: + return code; +} + +int32_t tsdbReadDelData(SDelFReader *pReader, SDelIdx *pDelIdx, SArray *aDelData) { + int32_t code = 0; + int64_t offset = pDelIdx->offset; + int64_t size = pDelIdx->size; + int64_t n; + + taosArrayClear(aDelData); + + // alloc + code = tRealloc(&pReader->aBuf[0], size); + if (code) goto _err; + + // read + code = tsdbReadFile(pReader->pReadH, offset, pReader->aBuf[0], size); + if (code) goto _err; + + // // decode + n = 0; + while (n < size) { + SDelData delData; + n += tGetDelData(pReader->aBuf[0] + n, &delData); + + if (taosArrayPush(aDelData, &delData) == NULL) { + code = TSDB_CODE_OUT_OF_MEMORY; + goto _err; + } + } + ASSERT(n == size); + + return code; + +_err: + tsdbError("vgId:%d, read del data failed since %s", TD_VID(pReader->pTsdb->pVnode), tstrerror(code)); + return code; +} + +int32_t tsdbReadDelIdx(SDelFReader *pReader, SArray *aDelIdx) { + int32_t code = 0; + int32_t n; + int64_t offset = pReader->fDel.offset; + int64_t size = pReader->fDel.size - offset; + + taosArrayClear(aDelIdx); + + // alloc + code = tRealloc(&pReader->aBuf[0], size); + if (code) goto _err; + + // read + code = tsdbReadFile(pReader->pReadH, offset, pReader->aBuf[0], size); + if (code) goto _err; + + // decode + n = 0; + while (n < size) { + SDelIdx delIdx; + + n += tGetDelIdx(pReader->aBuf[0] + n, &delIdx); + + if (taosArrayPush(aDelIdx, &delIdx) == NULL) { + code = TSDB_CODE_OUT_OF_MEMORY; + goto _err; + } + } + + ASSERT(n == size); + + return code; + +_err: + tsdbError("vgId:%d, read del idx failed since %s", TD_VID(pReader->pTsdb->pVnode), tstrerror(code)); + return code; +} \ No newline at end of file diff --git a/source/dnode/vnode/src/tsdb/tsdbUtil.c b/source/dnode/vnode/src/tsdb/tsdbUtil.c index 6937a27fe4..8509c0c759 100644 --- a/source/dnode/vnode/src/tsdb/tsdbUtil.c +++ b/source/dnode/vnode/src/tsdb/tsdbUtil.c @@ -1548,7 +1548,7 @@ int32_t tCmprBlockData(SBlockData *pBlockData, int8_t cmprAlg, uint8_t **ppOut, if (code) goto _exit; blockCol.offset = aBufN[0]; - aBufN[0] = aBufN[0] + blockCol.szBitmap + blockCol.szOffset + blockCol.szValue + sizeof(TSCKSUM); + aBufN[0] = aBufN[0] + blockCol.szBitmap + blockCol.szOffset + blockCol.szValue; } code = tRealloc(&aBuf[1], hdr.szBlkCol + tPutBlockCol(NULL, &blockCol)); @@ -1556,15 +1556,8 @@ int32_t tCmprBlockData(SBlockData *pBlockData, int8_t cmprAlg, uint8_t **ppOut, hdr.szBlkCol += tPutBlockCol(aBuf[1] + hdr.szBlkCol, &blockCol); } - aBufN[1] = 0; - if (hdr.szBlkCol > 0) { - aBufN[1] = hdr.szBlkCol + sizeof(TSCKSUM); - - code = tRealloc(&aBuf[1], aBufN[1]); - if (code) goto _exit; - - taosCalcChecksumAppend(0, aBuf[1], aBufN[1]); - } + // SBlockCol + aBufN[1] = hdr.szBlkCol; // uid + version + tskey aBufN[2] = 0; @@ -1585,16 +1578,11 @@ int32_t tCmprBlockData(SBlockData *pBlockData, int8_t cmprAlg, uint8_t **ppOut, if (code) goto _exit; aBufN[2] += hdr.szKey; - aBufN[2] += sizeof(TSCKSUM); - code = tRealloc(&aBuf[2], aBufN[2]); - if (code) goto _exit; - // hdr aBufN[3] = tPutDiskDataHdr(NULL, &hdr); code = tRealloc(&aBuf[3], aBufN[3]); if (code) goto _exit; tPutDiskDataHdr(aBuf[3], &hdr); - taosCalcChecksumAppend(taosCalcChecksum(0, aBuf[3], aBufN[3]), aBuf[2], aBufN[2]); // aggragate if (ppOut) { @@ -1626,10 +1614,6 @@ int32_t tDecmprBlockData(uint8_t *pIn, int32_t szIn, SBlockData *pBlockData, uin // SDiskDataHdr n += tGetDiskDataHdr(pIn + n, &hdr); - if (!taosCheckChecksumWhole(pIn, n + hdr.szUid + hdr.szVer + hdr.szKey + sizeof(TSCKSUM))) { - code = TSDB_CODE_FILE_CORRUPTED; - goto _exit; - } ASSERT(hdr.delimiter == TSDB_FILE_DLMT); pBlockData->suid = hdr.suid; @@ -1657,7 +1641,7 @@ int32_t tDecmprBlockData(uint8_t *pIn, int32_t szIn, SBlockData *pBlockData, uin code = tsdbDecmprData(pIn + n, hdr.szKey, TSDB_DATA_TYPE_TIMESTAMP, hdr.cmprAlg, (uint8_t **)&pBlockData->aTSKEY, sizeof(TSKEY) * hdr.nRow, &aBuf[0]); if (code) goto _exit; - n = n + hdr.szKey + sizeof(TSCKSUM); + n += hdr.szKey; // loop to decode each column data if (hdr.szBlkCol == 0) goto _exit; @@ -1679,8 +1663,8 @@ int32_t tDecmprBlockData(uint8_t *pIn, int32_t szIn, SBlockData *pBlockData, uin if (code) goto _exit; } } else { - code = tsdbDecmprColData(pIn + n + hdr.szBlkCol + sizeof(TSCKSUM) + blockCol.offset, &blockCol, hdr.cmprAlg, - hdr.nRow, pColData, &aBuf[0]); + code = tsdbDecmprColData(pIn + n + hdr.szBlkCol + blockCol.offset, &blockCol, hdr.cmprAlg, hdr.nRow, pColData, + &aBuf[0]); if (code) goto _exit; } } @@ -2062,12 +2046,6 @@ int32_t tsdbCmprColData(SColData *pColData, int8_t cmprAlg, SBlockCol *pBlockCol } size += pBlockCol->szValue; - // checksum - size += sizeof(TSCKSUM); - code = tRealloc(ppOut, nOut + size); - if (code) goto _exit; - taosCalcChecksumAppend(0, *ppOut + nOut, size); - _exit: return code; } @@ -2076,12 +2054,6 @@ int32_t tsdbDecmprColData(uint8_t *pIn, SBlockCol *pBlockCol, int8_t cmprAlg, in uint8_t **ppBuf) { int32_t code = 0; - int32_t size = pBlockCol->szBitmap + pBlockCol->szOffset + pBlockCol->szValue + sizeof(TSCKSUM); - if (!taosCheckChecksumWhole(pIn, size)) { - code = TSDB_CODE_FILE_CORRUPTED; - goto _exit; - } - ASSERT(pColData->cid == pBlockCol->cid); ASSERT(pColData->type == pBlockCol->type); pColData->smaOn = pBlockCol->smaOn; @@ -2153,37 +2125,3 @@ int32_t tsdbDecmprColData(uint8_t *pIn, SBlockCol *pBlockCol, int8_t cmprAlg, in _exit: return code; } - -int32_t tsdbReadAndCheck(TdFilePtr pFD, int64_t offset, uint8_t **ppOut, int32_t size, int8_t toCheck) { - int32_t code = 0; - - // alloc - code = tRealloc(ppOut, size); - if (code) goto _exit; - - // seek - int64_t n = taosLSeekFile(pFD, offset, SEEK_SET); - if (n < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _exit; - } - - // read - n = taosReadFile(pFD, *ppOut, size); - if (n < 0) { - code = TAOS_SYSTEM_ERROR(errno); - goto _exit; - } else if (n < size) { - code = TSDB_CODE_FILE_CORRUPTED; - goto _exit; - } - - // check - if (toCheck && !taosCheckChecksumWhole(*ppOut, size)) { - code = TSDB_CODE_FILE_CORRUPTED; - goto _exit; - } - -_exit: - return code; -}