diff --git a/src/backend/access/gin/ginxlog.c b/src/backend/access/gin/ginxlog.c index ffabf4e259..d0553bb8f7 100644 --- a/src/backend/access/gin/ginxlog.c +++ b/src/backend/access/gin/ginxlog.c @@ -20,25 +20,25 @@ static MemoryContext opCtx; /* working memory for operations */ static void -ginRedoClearIncompleteSplit(XLogRecPtr lsn, RelFileNode node, BlockNumber blkno) +ginRedoClearIncompleteSplit(XLogRecPtr lsn, XLogRecord *record, + int block_index, + RelFileNode node, BlockNumber blkno) { Buffer buffer; Page page; - buffer = XLogReadBuffer(node, blkno, false); - if (!BufferIsValid(buffer)) - return; /* page was deleted, nothing to do */ - page = (Page) BufferGetPage(buffer); - - if (lsn > PageGetLSN(page)) + if (XLogReadBufferForRedo(lsn, record, block_index, node, blkno, &buffer) + == BLK_NEEDS_REDO) { + page = (Page) BufferGetPage(buffer); + GinPageGetOpaque(page)->flags &= ~GIN_INCOMPLETE_SPLIT; PageSetLSN(page, lsn); MarkBufferDirty(buffer); } - - UnlockReleaseBuffer(buffer); + if (BufferIsValid(buffer)) + UnlockReleaseBuffer(buffer); } static void @@ -332,7 +332,6 @@ ginRedoInsert(XLogRecPtr lsn, XLogRecord *record) { ginxlogInsert *data = (ginxlogInsert *) XLogRecGetData(record); Buffer buffer; - Page page; char *payload; BlockNumber leftChildBlkno = InvalidBlockNumber; BlockNumber rightChildBlkno = InvalidBlockNumber; @@ -351,26 +350,14 @@ ginRedoInsert(XLogRecPtr lsn, XLogRecord *record) rightChildBlkno = BlockIdGetBlockNumber((BlockId) payload); payload += sizeof(BlockIdData); - if (record->xl_info & XLR_BKP_BLOCK(0)) - (void) RestoreBackupBlock(lsn, record, 0, false, false); - else - ginRedoClearIncompleteSplit(lsn, data->node, leftChildBlkno); + ginRedoClearIncompleteSplit(lsn, record, 0, data->node, leftChildBlkno); } - /* If we have a full-page image, restore it and we're done */ - if (record->xl_info & XLR_BKP_BLOCK(isLeaf ? 0 : 1)) + if (XLogReadBufferForRedo(lsn, record, isLeaf ? 0 : 1, data->node, + data->blkno, &buffer) == BLK_NEEDS_REDO) { - (void) RestoreBackupBlock(lsn, record, isLeaf ? 0 : 1, false, false); - return; - } + Page page = BufferGetPage(buffer); - buffer = XLogReadBuffer(data->node, data->blkno, false); - if (!BufferIsValid(buffer)) - return; /* page was deleted, nothing to do */ - page = (Page) BufferGetPage(buffer); - - if (lsn > PageGetLSN(page)) - { /* How to insert the payload is tree-type specific */ if (data->flags & GIN_INSERT_ISDATA) { @@ -386,8 +373,8 @@ ginRedoInsert(XLogRecPtr lsn, XLogRecord *record) PageSetLSN(page, lsn); MarkBufferDirty(buffer); } - - UnlockReleaseBuffer(buffer); + if (BufferIsValid(buffer)) + UnlockReleaseBuffer(buffer); } static void @@ -476,12 +463,7 @@ ginRedoSplit(XLogRecPtr lsn, XLogRecord *record) * split */ if (!isLeaf) - { - if (record->xl_info & XLR_BKP_BLOCK(0)) - (void) RestoreBackupBlock(lsn, record, 0, false, false); - else - ginRedoClearIncompleteSplit(lsn, data->node, data->leftChildBlkno); - } + ginRedoClearIncompleteSplit(lsn, record, 0, data->node, data->leftChildBlkno); flags = 0; if (isLeaf) @@ -605,31 +587,21 @@ ginRedoVacuumDataLeafPage(XLogRecPtr lsn, XLogRecord *record) { ginxlogVacuumDataLeafPage *xlrec = (ginxlogVacuumDataLeafPage *) XLogRecGetData(record); Buffer buffer; - Page page; - /* If we have a full-page image, restore it and we're done */ - if (record->xl_info & XLR_BKP_BLOCK(0)) + if (XLogReadBufferForRedo(lsn, record, 0, xlrec->node, xlrec->blkno, + &buffer) == BLK_NEEDS_REDO) { - (void) RestoreBackupBlock(lsn, record, 0, false, false); - return; - } + Page page = BufferGetPage(buffer); - buffer = XLogReadBuffer(xlrec->node, xlrec->blkno, false); - if (!BufferIsValid(buffer)) - return; - page = (Page) BufferGetPage(buffer); + Assert(GinPageIsLeaf(page)); + Assert(GinPageIsData(page)); - Assert(GinPageIsLeaf(page)); - Assert(GinPageIsData(page)); - - if (lsn > PageGetLSN(page)) - { ginRedoRecompress(page, &xlrec->data); PageSetLSN(page, lsn); MarkBufferDirty(buffer); } - - UnlockReleaseBuffer(buffer); + if (BufferIsValid(buffer)) + UnlockReleaseBuffer(buffer); } static void @@ -641,62 +613,42 @@ ginRedoDeletePage(XLogRecPtr lsn, XLogRecord *record) Buffer lbuffer; Page page; - if (record->xl_info & XLR_BKP_BLOCK(0)) - dbuffer = RestoreBackupBlock(lsn, record, 0, false, true); - else + if (XLogReadBufferForRedo(lsn, record, 0, data->node, data->blkno, &dbuffer) + == BLK_NEEDS_REDO) { - dbuffer = XLogReadBuffer(data->node, data->blkno, false); - if (BufferIsValid(dbuffer)) - { - page = BufferGetPage(dbuffer); - if (lsn > PageGetLSN(page)) - { - Assert(GinPageIsData(page)); - GinPageGetOpaque(page)->flags = GIN_DELETED; - PageSetLSN(page, lsn); - MarkBufferDirty(dbuffer); - } - } + page = BufferGetPage(dbuffer); + + Assert(GinPageIsData(page)); + GinPageGetOpaque(page)->flags = GIN_DELETED; + PageSetLSN(page, lsn); + MarkBufferDirty(dbuffer); } - if (record->xl_info & XLR_BKP_BLOCK(1)) - pbuffer = RestoreBackupBlock(lsn, record, 1, false, true); - else + if (XLogReadBufferForRedo(lsn, record, 1, data->node, data->parentBlkno, + &pbuffer) == BLK_NEEDS_REDO) { - pbuffer = XLogReadBuffer(data->node, data->parentBlkno, false); - if (BufferIsValid(pbuffer)) - { - page = BufferGetPage(pbuffer); - if (lsn > PageGetLSN(page)) - { - Assert(GinPageIsData(page)); - Assert(!GinPageIsLeaf(page)); - GinPageDeletePostingItem(page, data->parentOffset); - PageSetLSN(page, lsn); - MarkBufferDirty(pbuffer); - } - } + page = BufferGetPage(pbuffer); + + Assert(GinPageIsData(page)); + Assert(!GinPageIsLeaf(page)); + GinPageDeletePostingItem(page, data->parentOffset); + PageSetLSN(page, lsn); + MarkBufferDirty(pbuffer); } - if (record->xl_info & XLR_BKP_BLOCK(2)) - (void) RestoreBackupBlock(lsn, record, 2, false, false); - else if (data->leftBlkno != InvalidBlockNumber) + if (XLogReadBufferForRedo(lsn, record, 2, data->node, data->leftBlkno, + &lbuffer) == BLK_NEEDS_REDO) { - lbuffer = XLogReadBuffer(data->node, data->leftBlkno, false); - if (BufferIsValid(lbuffer)) - { - page = BufferGetPage(lbuffer); - if (lsn > PageGetLSN(page)) - { - Assert(GinPageIsData(page)); - GinPageGetOpaque(page)->rightlink = data->rightLink; - PageSetLSN(page, lsn); - MarkBufferDirty(lbuffer); - } - UnlockReleaseBuffer(lbuffer); - } + page = BufferGetPage(lbuffer); + + Assert(GinPageIsData(page)); + GinPageGetOpaque(page)->rightlink = data->rightLink; + PageSetLSN(page, lsn); + MarkBufferDirty(lbuffer); } + if (BufferIsValid(lbuffer)) + UnlockReleaseBuffer(lbuffer); if (BufferIsValid(pbuffer)) UnlockReleaseBuffer(pbuffer); if (BufferIsValid(dbuffer)) @@ -730,74 +682,64 @@ ginRedoUpdateMetapage(XLogRecPtr lsn, XLogRecord *record) /* * insert into tail page */ - if (record->xl_info & XLR_BKP_BLOCK(0)) - (void) RestoreBackupBlock(lsn, record, 0, false, false); - else + if (XLogReadBufferForRedo(lsn, record, 0, data->node, + data->metadata.tail, &buffer) + == BLK_NEEDS_REDO) { - buffer = XLogReadBuffer(data->node, data->metadata.tail, false); - if (BufferIsValid(buffer)) + Page page = BufferGetPage(buffer); + OffsetNumber off; + int i; + Size tupsize; + IndexTuple tuples; + + tuples = (IndexTuple) (XLogRecGetData(record) + sizeof(ginxlogUpdateMeta)); + + if (PageIsEmpty(page)) + off = FirstOffsetNumber; + else + off = OffsetNumberNext(PageGetMaxOffsetNumber(page)); + + for (i = 0; i < data->ntuples; i++) { - Page page = BufferGetPage(buffer); + tupsize = IndexTupleSize(tuples); - if (lsn > PageGetLSN(page)) - { - OffsetNumber l, - off = (PageIsEmpty(page)) ? FirstOffsetNumber : - OffsetNumberNext(PageGetMaxOffsetNumber(page)); - int i, - tupsize; - IndexTuple tuples = (IndexTuple) (XLogRecGetData(record) + sizeof(ginxlogUpdateMeta)); + if (PageAddItem(page, (Item) tuples, tupsize, off, + false, false) == InvalidOffsetNumber) + elog(ERROR, "failed to add item to index page"); - for (i = 0; i < data->ntuples; i++) - { - tupsize = IndexTupleSize(tuples); + tuples = (IndexTuple) (((char *) tuples) + tupsize); - l = PageAddItem(page, (Item) tuples, tupsize, off, false, false); - - if (l == InvalidOffsetNumber) - elog(ERROR, "failed to add item to index page"); - - tuples = (IndexTuple) (((char *) tuples) + tupsize); - - off++; - } - - /* - * Increase counter of heap tuples - */ - GinPageGetOpaque(page)->maxoff++; - - PageSetLSN(page, lsn); - MarkBufferDirty(buffer); - } - UnlockReleaseBuffer(buffer); + off++; } + + /* + * Increase counter of heap tuples + */ + GinPageGetOpaque(page)->maxoff++; + + PageSetLSN(page, lsn); + MarkBufferDirty(buffer); } + if (BufferIsValid(buffer)) + UnlockReleaseBuffer(buffer); } else if (data->prevTail != InvalidBlockNumber) { /* * New tail */ - if (record->xl_info & XLR_BKP_BLOCK(0)) - (void) RestoreBackupBlock(lsn, record, 0, false, false); - else + if (XLogReadBufferForRedo(lsn, record, 0, data->node, data->prevTail, + &buffer) == BLK_NEEDS_REDO) { - buffer = XLogReadBuffer(data->node, data->prevTail, false); - if (BufferIsValid(buffer)) - { - Page page = BufferGetPage(buffer); + Page page = BufferGetPage(buffer); - if (lsn > PageGetLSN(page)) - { - GinPageGetOpaque(page)->rightlink = data->newRightlink; + GinPageGetOpaque(page)->rightlink = data->newRightlink; - PageSetLSN(page, lsn); - MarkBufferDirty(buffer); - } - UnlockReleaseBuffer(buffer); - } + PageSetLSN(page, lsn); + MarkBufferDirty(buffer); } + if (BufferIsValid(buffer)) + UnlockReleaseBuffer(buffer); } UnlockReleaseBuffer(metabuffer); diff --git a/src/backend/access/gist/gistxlog.c b/src/backend/access/gist/gistxlog.c index 7d36b2ab6a..ecc095671d 100644 --- a/src/backend/access/gist/gistxlog.c +++ b/src/backend/access/gist/gistxlog.c @@ -48,31 +48,26 @@ gistRedoClearFollowRight(XLogRecPtr lsn, XLogRecord *record, int block_index, { Buffer buffer; Page page; - - if (record->xl_info & XLR_BKP_BLOCK(block_index)) - buffer = RestoreBackupBlock(lsn, record, block_index, false, true); - else - { - buffer = XLogReadBuffer(node, childblkno, false); - if (!BufferIsValid(buffer)) - return; /* page was deleted, nothing to do */ - } - page = (Page) BufferGetPage(buffer); + XLogRedoAction action; /* - * Note that we still update the page even if page LSN is equal to the LSN - * of this record, because the updated NSN is not included in the full - * page image. + * Note that we still update the page even if it was restored from a full + * page image, because the updated NSN is not included in the image. */ - if (lsn >= PageGetLSN(page)) + action = XLogReadBufferForRedo(lsn, record, block_index, node, childblkno, + &buffer); + if (action == BLK_NEEDS_REDO || action == BLK_RESTORED) { + page = BufferGetPage(buffer); + GistPageSetNSN(page, lsn); GistClearFollowRight(page); PageSetLSN(page, lsn); MarkBufferDirty(buffer); } - UnlockReleaseBuffer(buffer); + if (BufferIsValid(buffer)) + UnlockReleaseBuffer(buffer); } /* @@ -87,104 +82,86 @@ gistRedoPageUpdateRecord(XLogRecPtr lsn, XLogRecord *record) Page page; char *data; - /* - * We need to acquire and hold lock on target page while updating the left - * child page. If we have a full-page image of target page, getting the - * lock is a side-effect of restoring that image. Note that even if the - * target page no longer exists, we'll still attempt to replay the change - * on the child page. - */ - if (record->xl_info & XLR_BKP_BLOCK(0)) - buffer = RestoreBackupBlock(lsn, record, 0, false, true); - else - buffer = XLogReadBuffer(xldata->node, xldata->blkno, false); + if (XLogReadBufferForRedo(lsn, record, 0, xldata->node, xldata->blkno, + &buffer) == BLK_NEEDS_REDO) + { + page = (Page) BufferGetPage(buffer); - /* Fix follow-right data on left child page */ + data = begin + sizeof(gistxlogPageUpdate); + + /* Delete old tuples */ + if (xldata->ntodelete > 0) + { + int i; + OffsetNumber *todelete = (OffsetNumber *) data; + + data += sizeof(OffsetNumber) * xldata->ntodelete; + + for (i = 0; i < xldata->ntodelete; i++) + PageIndexTupleDelete(page, todelete[i]); + if (GistPageIsLeaf(page)) + GistMarkTuplesDeleted(page); + } + + /* add tuples */ + if (data - begin < record->xl_len) + { + OffsetNumber off = (PageIsEmpty(page)) ? FirstOffsetNumber : + OffsetNumberNext(PageGetMaxOffsetNumber(page)); + + while (data - begin < record->xl_len) + { + IndexTuple itup = (IndexTuple) data; + Size sz = IndexTupleSize(itup); + OffsetNumber l; + + data += sz; + + l = PageAddItem(page, (Item) itup, sz, off, false, false); + if (l == InvalidOffsetNumber) + elog(ERROR, "failed to add item to GiST index page, size %d bytes", + (int) sz); + off++; + } + } + else + { + /* + * special case: leafpage, nothing to insert, nothing to delete, + * then vacuum marks page + */ + if (GistPageIsLeaf(page) && xldata->ntodelete == 0) + GistClearTuplesDeleted(page); + } + + if (!GistPageIsLeaf(page) && + PageGetMaxOffsetNumber(page) == InvalidOffsetNumber && + xldata->blkno == GIST_ROOT_BLKNO) + { + /* + * all links on non-leaf root page was deleted by vacuum full, so + * root page becomes a leaf + */ + GistPageSetLeaf(page); + } + + PageSetLSN(page, lsn); + MarkBufferDirty(buffer); + } + + /* + * Fix follow-right data on left child page + * + * This must be done while still holding the lock on the target page. Note + * that even if the target page no longer exists, we still attempt to + * replay the change on the child page. + */ if (BlockNumberIsValid(xldata->leftchild)) gistRedoClearFollowRight(lsn, record, 1, xldata->node, xldata->leftchild); - /* Done if target page no longer exists */ - if (!BufferIsValid(buffer)) - return; - - /* nothing more to do if page was backed up (and no info to do it with) */ - if (record->xl_info & XLR_BKP_BLOCK(0)) - { + if (BufferIsValid(buffer)) UnlockReleaseBuffer(buffer); - return; - } - - page = (Page) BufferGetPage(buffer); - - /* nothing more to do if change already applied */ - if (lsn <= PageGetLSN(page)) - { - UnlockReleaseBuffer(buffer); - return; - } - - data = begin + sizeof(gistxlogPageUpdate); - - /* Delete old tuples */ - if (xldata->ntodelete > 0) - { - int i; - OffsetNumber *todelete = (OffsetNumber *) data; - - data += sizeof(OffsetNumber) * xldata->ntodelete; - - for (i = 0; i < xldata->ntodelete; i++) - PageIndexTupleDelete(page, todelete[i]); - if (GistPageIsLeaf(page)) - GistMarkTuplesDeleted(page); - } - - /* add tuples */ - if (data - begin < record->xl_len) - { - OffsetNumber off = (PageIsEmpty(page)) ? FirstOffsetNumber : - OffsetNumberNext(PageGetMaxOffsetNumber(page)); - - while (data - begin < record->xl_len) - { - IndexTuple itup = (IndexTuple) data; - Size sz = IndexTupleSize(itup); - OffsetNumber l; - - data += sz; - - l = PageAddItem(page, (Item) itup, sz, off, false, false); - if (l == InvalidOffsetNumber) - elog(ERROR, "failed to add item to GiST index page, size %d bytes", - (int) sz); - off++; - } - } - else - { - /* - * special case: leafpage, nothing to insert, nothing to delete, then - * vacuum marks page - */ - if (GistPageIsLeaf(page) && xldata->ntodelete == 0) - GistClearTuplesDeleted(page); - } - - if (!GistPageIsLeaf(page) && - PageGetMaxOffsetNumber(page) == InvalidOffsetNumber && - xldata->blkno == GIST_ROOT_BLKNO) - { - /* - * all links on non-leaf root page was deleted by vacuum full, so root - * page becomes a leaf - */ - GistPageSetLeaf(page); - } - - PageSetLSN(page, lsn); - MarkBufferDirty(buffer); - UnlockReleaseBuffer(buffer); } static void diff --git a/src/backend/access/heap/heapam.c b/src/backend/access/heap/heapam.c index 4d7575bc06..ae15c0b955 100644 --- a/src/backend/access/heap/heapam.c +++ b/src/backend/access/heap/heapam.c @@ -7134,15 +7134,13 @@ heap_xlog_clean(XLogRecPtr lsn, XLogRecord *record) { xl_heap_clean *xlrec = (xl_heap_clean *) XLogRecGetData(record); Buffer buffer; - Page page; - OffsetNumber *end; - OffsetNumber *redirected; - OffsetNumber *nowdead; - OffsetNumber *nowunused; - int nredirected; - int ndead; - int nunused; - Size freespace; + Size freespace = 0; + RelFileNode rnode; + BlockNumber blkno; + XLogRedoAction action; + + rnode = xlrec->node; + blkno = xlrec->block; /* * We're about to remove tuples. In Hot Standby mode, ensure that there's @@ -7153,65 +7151,63 @@ heap_xlog_clean(XLogRecPtr lsn, XLogRecord *record) * latestRemovedXid is invalid, skip conflict processing. */ if (InHotStandby && TransactionIdIsValid(xlrec->latestRemovedXid)) - ResolveRecoveryConflictWithSnapshot(xlrec->latestRemovedXid, - xlrec->node); + ResolveRecoveryConflictWithSnapshot(xlrec->latestRemovedXid, rnode); /* * If we have a full-page image, restore it (using a cleanup lock) and * we're done. */ - if (record->xl_info & XLR_BKP_BLOCK(0)) + action = XLogReadBufferForRedoExtended(lsn, record, 0, + rnode, MAIN_FORKNUM, blkno, + RBM_NORMAL, true, &buffer); + if (action == BLK_NEEDS_REDO) { - (void) RestoreBackupBlock(lsn, record, 0, true, false); - return; + Page page = (Page) BufferGetPage(buffer); + OffsetNumber *end; + OffsetNumber *redirected; + OffsetNumber *nowdead; + OffsetNumber *nowunused; + int nredirected; + int ndead; + int nunused; + + nredirected = xlrec->nredirected; + ndead = xlrec->ndead; + end = (OffsetNumber *) ((char *) xlrec + record->xl_len); + redirected = (OffsetNumber *) ((char *) xlrec + SizeOfHeapClean); + nowdead = redirected + (nredirected * 2); + nowunused = nowdead + ndead; + nunused = (end - nowunused); + Assert(nunused >= 0); + + /* Update all item pointers per the record, and repair fragmentation */ + heap_page_prune_execute(buffer, + redirected, nredirected, + nowdead, ndead, + nowunused, nunused); + + freespace = PageGetHeapFreeSpace(page); /* needed to update FSM below */ + + /* + * Note: we don't worry about updating the page's prunability hints. + * At worst this will cause an extra prune cycle to occur soon. + */ + + PageSetLSN(page, lsn); + MarkBufferDirty(buffer); } - - buffer = XLogReadBufferExtended(xlrec->node, MAIN_FORKNUM, xlrec->block, RBM_NORMAL); - if (!BufferIsValid(buffer)) - return; - LockBufferForCleanup(buffer); - page = (Page) BufferGetPage(buffer); - - if (lsn <= PageGetLSN(page)) - { + if (BufferIsValid(buffer)) UnlockReleaseBuffer(buffer); - return; - } - - nredirected = xlrec->nredirected; - ndead = xlrec->ndead; - end = (OffsetNumber *) ((char *) xlrec + record->xl_len); - redirected = (OffsetNumber *) ((char *) xlrec + SizeOfHeapClean); - nowdead = redirected + (nredirected * 2); - nowunused = nowdead + ndead; - nunused = (end - nowunused); - Assert(nunused >= 0); - - /* Update all item pointers per the record, and repair fragmentation */ - heap_page_prune_execute(buffer, - redirected, nredirected, - nowdead, ndead, - nowunused, nunused); - - freespace = PageGetHeapFreeSpace(page); /* needed to update FSM below */ - - /* - * Note: we don't worry about updating the page's prunability hints. At - * worst this will cause an extra prune cycle to occur soon. - */ - - PageSetLSN(page, lsn); - MarkBufferDirty(buffer); - UnlockReleaseBuffer(buffer); /* * Update the FSM as well. * - * XXX: We don't get here if the page was restored from full page image. - * We don't bother to update the FSM in that case, it doesn't need to be + * XXX: Don't do this if the page was restored from full page image. We + * don't bother to update the FSM in that case, it doesn't need to be * totally accurate anyway. */ - XLogRecordPageWithFreeSpace(xlrec->node, xlrec->block, freespace); + if (action == BLK_NEEDS_REDO) + XLogRecordPageWithFreeSpace(xlrec->node, xlrec->block, freespace); } /* @@ -7226,6 +7222,14 @@ static void heap_xlog_visible(XLogRecPtr lsn, XLogRecord *record) { xl_heap_visible *xlrec = (xl_heap_visible *) XLogRecGetData(record); + Buffer buffer; + Page page; + RelFileNode rnode; + BlockNumber blkno; + XLogRedoAction action; + + rnode = xlrec->node; + blkno = xlrec->block; /* * If there are any Hot Standby transactions running that have an xmin @@ -7237,60 +7241,43 @@ heap_xlog_visible(XLogRecPtr lsn, XLogRecord *record) * rather than killing the transaction outright. */ if (InHotStandby) - ResolveRecoveryConflictWithSnapshot(xlrec->cutoff_xid, xlrec->node); + ResolveRecoveryConflictWithSnapshot(xlrec->cutoff_xid, rnode); /* - * If heap block was backed up, restore it. This can only happen with - * checksums enabled. + * Read the heap page, if it still exists. If the heap file has dropped or + * truncated later in recovery, we don't need to update the page, but we'd + * better still update the visibility map. */ - if (record->xl_info & XLR_BKP_BLOCK(1)) + action = XLogReadBufferForRedo(lsn, record, 1, rnode, blkno, &buffer); + if (action == BLK_NEEDS_REDO) { - Assert(DataChecksumsEnabled()); - (void) RestoreBackupBlock(lsn, record, 1, false, false); - } - else - { - Buffer buffer; - Page page; - /* - * Read the heap page, if it still exists. If the heap file has been - * dropped or truncated later in recovery, we don't need to update the - * page, but we'd better still update the visibility map. + * We don't bump the LSN of the heap page when setting the visibility + * map bit (unless checksums are enabled, in which case we must), + * because that would generate an unworkable volume of full-page + * writes. This exposes us to torn page hazards, but since we're not + * inspecting the existing page contents in any way, we don't care. + * + * However, all operations that clear the visibility map bit *do* bump + * the LSN, and those operations will only be replayed if the XLOG LSN + * follows the page LSN. Thus, if the page LSN has advanced past our + * XLOG record's LSN, we mustn't mark the page all-visible, because + * the subsequent update won't be replayed to clear the flag. */ - buffer = XLogReadBufferExtended(xlrec->node, MAIN_FORKNUM, - xlrec->block, RBM_NORMAL); - if (BufferIsValid(buffer)) - { - LockBuffer(buffer, BUFFER_LOCK_EXCLUSIVE); - - page = (Page) BufferGetPage(buffer); - - /* - * We don't bump the LSN of the heap page when setting the - * visibility map bit (unless checksums are enabled, in which case - * we must), because that would generate an unworkable volume of - * full-page writes. This exposes us to torn page hazards, but - * since we're not inspecting the existing page contents in any - * way, we don't care. - * - * However, all operations that clear the visibility map bit *do* - * bump the LSN, and those operations will only be replayed if the - * XLOG LSN follows the page LSN. Thus, if the page LSN has - * advanced past our XLOG record's LSN, we mustn't mark the page - * all-visible, because the subsequent update won't be replayed to - * clear the flag. - */ - if (lsn > PageGetLSN(page)) - { - PageSetAllVisible(page); - MarkBufferDirty(buffer); - } - - /* Done with heap page. */ - UnlockReleaseBuffer(buffer); - } + page = BufferGetPage(buffer); + PageSetAllVisible(page); + MarkBufferDirty(buffer); } + else if (action == BLK_RESTORED) + { + /* + * If heap block was backed up, restore it. This can only happen with + * checksums enabled. + */ + Assert(DataChecksumsEnabled()); + } + if (BufferIsValid(buffer)) + UnlockReleaseBuffer(buffer); /* * Even if we skipped the heap page update due to the LSN interlock, it's @@ -7305,8 +7292,8 @@ heap_xlog_visible(XLogRecPtr lsn, XLogRecord *record) Relation reln; Buffer vmbuffer = InvalidBuffer; - reln = CreateFakeRelcacheEntry(xlrec->node); - visibilitymap_pin(reln, xlrec->block, &vmbuffer); + reln = CreateFakeRelcacheEntry(rnode); + visibilitymap_pin(reln, blkno, &vmbuffer); /* * Don't set the bit if replay has already passed this point. @@ -7320,7 +7307,7 @@ heap_xlog_visible(XLogRecPtr lsn, XLogRecord *record) * real harm is done; and the next VACUUM will fix it. */ if (lsn > PageGetLSN(BufferGetPage(vmbuffer))) - visibilitymap_set(reln, xlrec->block, InvalidBuffer, lsn, vmbuffer, + visibilitymap_set(reln, blkno, InvalidBuffer, lsn, vmbuffer, xlrec->cutoff_xid); ReleaseBuffer(vmbuffer); @@ -7347,42 +7334,30 @@ heap_xlog_freeze_page(XLogRecPtr lsn, XLogRecord *record) if (InHotStandby) ResolveRecoveryConflictWithSnapshot(cutoff_xid, xlrec->node); - /* If we have a full-page image, restore it and we're done */ - if (record->xl_info & XLR_BKP_BLOCK(0)) + if (XLogReadBufferForRedo(lsn, record, 0, xlrec->node, xlrec->block, + &buffer) == BLK_NEEDS_REDO) { - (void) RestoreBackupBlock(lsn, record, 0, false, false); - return; + page = BufferGetPage(buffer); + + /* now execute freeze plan for each frozen tuple */ + for (ntup = 0; ntup < xlrec->ntuples; ntup++) + { + xl_heap_freeze_tuple *xlrec_tp; + ItemId lp; + HeapTupleHeader tuple; + + xlrec_tp = &xlrec->tuples[ntup]; + lp = PageGetItemId(page, xlrec_tp->offset); /* offsets are one-based */ + tuple = (HeapTupleHeader) PageGetItem(page, lp); + + heap_execute_freeze_tuple(tuple, xlrec_tp); + } + + PageSetLSN(page, lsn); + MarkBufferDirty(buffer); } - - buffer = XLogReadBuffer(xlrec->node, xlrec->block, false); - if (!BufferIsValid(buffer)) - return; - - page = (Page) BufferGetPage(buffer); - - if (lsn <= PageGetLSN(page)) - { + if (BufferIsValid(buffer)) UnlockReleaseBuffer(buffer); - return; - } - - /* now execute freeze plan for each frozen tuple */ - for (ntup = 0; ntup < xlrec->ntuples; ntup++) - { - xl_heap_freeze_tuple *xlrec_tp; - ItemId lp; - HeapTupleHeader tuple; - - xlrec_tp = &xlrec->tuples[ntup]; - lp = PageGetItemId(page, xlrec_tp->offset); /* offsets are one-based */ - tuple = (HeapTupleHeader) PageGetItem(page, lp); - - heap_execute_freeze_tuple(tuple, xlrec_tp); - } - - PageSetLSN(page, lsn); - MarkBufferDirty(buffer); - UnlockReleaseBuffer(buffer); } /* @@ -7422,8 +7397,10 @@ heap_xlog_delete(XLogRecPtr lsn, XLogRecord *record) ItemId lp = NULL; HeapTupleHeader htup; BlockNumber blkno; + RelFileNode target_node; blkno = ItemPointerGetBlockNumber(&(xlrec->target.tid)); + target_node = xlrec->target.node; /* * The visibility map may need to be fixed even if the heap page is @@ -7431,7 +7408,7 @@ heap_xlog_delete(XLogRecPtr lsn, XLogRecord *record) */ if (xlrec->flags & XLOG_HEAP_ALL_VISIBLE_CLEARED) { - Relation reln = CreateFakeRelcacheEntry(xlrec->target.node); + Relation reln = CreateFakeRelcacheEntry(target_node); Buffer vmbuffer = InvalidBuffer; visibilitymap_pin(reln, blkno, &vmbuffer); @@ -7440,52 +7417,41 @@ heap_xlog_delete(XLogRecPtr lsn, XLogRecord *record) FreeFakeRelcacheEntry(reln); } - /* If we have a full-page image, restore it and we're done */ - if (record->xl_info & XLR_BKP_BLOCK(0)) + if (XLogReadBufferForRedo(lsn, record, 0, target_node, blkno, &buffer) + == BLK_NEEDS_REDO) { - (void) RestoreBackupBlock(lsn, record, 0, false, false); - return; + page = (Page) BufferGetPage(buffer); + + offnum = ItemPointerGetOffsetNumber(&(xlrec->target.tid)); + if (PageGetMaxOffsetNumber(page) >= offnum) + lp = PageGetItemId(page, offnum); + + if (PageGetMaxOffsetNumber(page) < offnum || !ItemIdIsNormal(lp)) + elog(PANIC, "heap_delete_redo: invalid lp"); + + htup = (HeapTupleHeader) PageGetItem(page, lp); + + htup->t_infomask &= ~(HEAP_XMAX_BITS | HEAP_MOVED); + htup->t_infomask2 &= ~HEAP_KEYS_UPDATED; + HeapTupleHeaderClearHotUpdated(htup); + fix_infomask_from_infobits(xlrec->infobits_set, + &htup->t_infomask, &htup->t_infomask2); + HeapTupleHeaderSetXmax(htup, xlrec->xmax); + HeapTupleHeaderSetCmax(htup, FirstCommandId, false); + + /* Mark the page as a candidate for pruning */ + PageSetPrunable(page, record->xl_xid); + + if (xlrec->flags & XLOG_HEAP_ALL_VISIBLE_CLEARED) + PageClearAllVisible(page); + + /* Make sure there is no forward chain link in t_ctid */ + htup->t_ctid = xlrec->target.tid; + PageSetLSN(page, lsn); + MarkBufferDirty(buffer); } - - buffer = XLogReadBuffer(xlrec->target.node, blkno, false); - if (!BufferIsValid(buffer)) - return; - page = (Page) BufferGetPage(buffer); - - if (lsn <= PageGetLSN(page)) /* changes are applied */ - { + if (BufferIsValid(buffer)) UnlockReleaseBuffer(buffer); - return; - } - - offnum = ItemPointerGetOffsetNumber(&(xlrec->target.tid)); - if (PageGetMaxOffsetNumber(page) >= offnum) - lp = PageGetItemId(page, offnum); - - if (PageGetMaxOffsetNumber(page) < offnum || !ItemIdIsNormal(lp)) - elog(PANIC, "heap_delete_redo: invalid lp"); - - htup = (HeapTupleHeader) PageGetItem(page, lp); - - htup->t_infomask &= ~(HEAP_XMAX_BITS | HEAP_MOVED); - htup->t_infomask2 &= ~HEAP_KEYS_UPDATED; - HeapTupleHeaderClearHotUpdated(htup); - fix_infomask_from_infobits(xlrec->infobits_set, - &htup->t_infomask, &htup->t_infomask2); - HeapTupleHeaderSetXmax(htup, xlrec->xmax); - HeapTupleHeaderSetCmax(htup, FirstCommandId, false); - - /* Mark the page as a candidate for pruning */ - PageSetPrunable(page, record->xl_xid); - - if (xlrec->flags & XLOG_HEAP_ALL_VISIBLE_CLEARED) - PageClearAllVisible(page); - - /* Make sure there is no forward chain link in t_ctid */ - htup->t_ctid = xlrec->target.tid; - PageSetLSN(page, lsn); - MarkBufferDirty(buffer); - UnlockReleaseBuffer(buffer); } static void @@ -7503,9 +7469,12 @@ heap_xlog_insert(XLogRecPtr lsn, XLogRecord *record) HeapTupleHeader htup; xl_heap_header xlhdr; uint32 newlen; - Size freespace; + Size freespace = 0; + RelFileNode target_node; BlockNumber blkno; + XLogRedoAction action; + target_node = xlrec->target.node; blkno = ItemPointerGetBlockNumber(&(xlrec->target.tid)); /* @@ -7514,7 +7483,7 @@ heap_xlog_insert(XLogRecPtr lsn, XLogRecord *record) */ if (xlrec->flags & XLOG_HEAP_ALL_VISIBLE_CLEARED) { - Relation reln = CreateFakeRelcacheEntry(xlrec->target.node); + Relation reln = CreateFakeRelcacheEntry(target_node); Buffer vmbuffer = InvalidBuffer; visibilitymap_pin(reln, blkno, &vmbuffer); @@ -7523,82 +7492,76 @@ heap_xlog_insert(XLogRecPtr lsn, XLogRecord *record) FreeFakeRelcacheEntry(reln); } - /* If we have a full-page image, restore it and we're done */ - if (record->xl_info & XLR_BKP_BLOCK(0)) - { - (void) RestoreBackupBlock(lsn, record, 0, false, false); - return; - } - + /* + * If we inserted the first and only tuple on the page, re-initialize + * the page from scratch. + */ if (record->xl_info & XLOG_HEAP_INIT_PAGE) { - buffer = XLogReadBuffer(xlrec->target.node, blkno, true); - Assert(BufferIsValid(buffer)); - page = (Page) BufferGetPage(buffer); - + XLogReadBufferForRedoExtended(lsn, record, 0, + target_node, MAIN_FORKNUM, blkno, + RBM_ZERO, false, &buffer); + page = BufferGetPage(buffer); PageInit(page, BufferGetPageSize(buffer), 0); + action = BLK_NEEDS_REDO; } else + action = XLogReadBufferForRedo(lsn, record, 0, target_node, blkno, + &buffer); + + if (action == BLK_NEEDS_REDO) { - buffer = XLogReadBuffer(xlrec->target.node, blkno, false); - if (!BufferIsValid(buffer)) - return; - page = (Page) BufferGetPage(buffer); + page = BufferGetPage(buffer); - if (lsn <= PageGetLSN(page)) /* changes are applied */ - { - UnlockReleaseBuffer(buffer); - return; - } + offnum = ItemPointerGetOffsetNumber(&(xlrec->target.tid)); + if (PageGetMaxOffsetNumber(page) + 1 < offnum) + elog(PANIC, "heap_insert_redo: invalid max offset number"); + + newlen = record->xl_len - SizeOfHeapInsert - SizeOfHeapHeader; + Assert(newlen <= MaxHeapTupleSize); + memcpy((char *) &xlhdr, + (char *) xlrec + SizeOfHeapInsert, + SizeOfHeapHeader); + htup = &tbuf.hdr; + MemSet((char *) htup, 0, sizeof(HeapTupleHeaderData)); + /* PG73FORMAT: get bitmap [+ padding] [+ oid] + data */ + memcpy((char *) htup + offsetof(HeapTupleHeaderData, t_bits), + (char *) xlrec + SizeOfHeapInsert + SizeOfHeapHeader, + newlen); + newlen += offsetof(HeapTupleHeaderData, t_bits); + htup->t_infomask2 = xlhdr.t_infomask2; + htup->t_infomask = xlhdr.t_infomask; + htup->t_hoff = xlhdr.t_hoff; + HeapTupleHeaderSetXmin(htup, record->xl_xid); + HeapTupleHeaderSetCmin(htup, FirstCommandId); + htup->t_ctid = xlrec->target.tid; + + offnum = PageAddItem(page, (Item) htup, newlen, offnum, true, true); + if (offnum == InvalidOffsetNumber) + elog(PANIC, "heap_insert_redo: failed to add tuple"); + + freespace = PageGetHeapFreeSpace(page); /* needed to update FSM below */ + + PageSetLSN(page, lsn); + + if (xlrec->flags & XLOG_HEAP_ALL_VISIBLE_CLEARED) + PageClearAllVisible(page); + + MarkBufferDirty(buffer); } - - offnum = ItemPointerGetOffsetNumber(&(xlrec->target.tid)); - if (PageGetMaxOffsetNumber(page) + 1 < offnum) - elog(PANIC, "heap_insert_redo: invalid max offset number"); - - newlen = record->xl_len - SizeOfHeapInsert - SizeOfHeapHeader; - Assert(newlen <= MaxHeapTupleSize); - memcpy((char *) &xlhdr, - (char *) xlrec + SizeOfHeapInsert, - SizeOfHeapHeader); - htup = &tbuf.hdr; - MemSet((char *) htup, 0, sizeof(HeapTupleHeaderData)); - /* PG73FORMAT: get bitmap [+ padding] [+ oid] + data */ - memcpy((char *) htup + offsetof(HeapTupleHeaderData, t_bits), - (char *) xlrec + SizeOfHeapInsert + SizeOfHeapHeader, - newlen); - newlen += offsetof(HeapTupleHeaderData, t_bits); - htup->t_infomask2 = xlhdr.t_infomask2; - htup->t_infomask = xlhdr.t_infomask; - htup->t_hoff = xlhdr.t_hoff; - HeapTupleHeaderSetXmin(htup, record->xl_xid); - HeapTupleHeaderSetCmin(htup, FirstCommandId); - htup->t_ctid = xlrec->target.tid; - - offnum = PageAddItem(page, (Item) htup, newlen, offnum, true, true); - if (offnum == InvalidOffsetNumber) - elog(PANIC, "heap_insert_redo: failed to add tuple"); - - freespace = PageGetHeapFreeSpace(page); /* needed to update FSM below */ - - PageSetLSN(page, lsn); - - if (xlrec->flags & XLOG_HEAP_ALL_VISIBLE_CLEARED) - PageClearAllVisible(page); - - MarkBufferDirty(buffer); - UnlockReleaseBuffer(buffer); + if (BufferIsValid(buffer)) + UnlockReleaseBuffer(buffer); /* * If the page is running low on free space, update the FSM as well. * Arbitrarily, our definition of "low" is less than 20%. We can't do much * better than that without knowing the fill-factor for the table. * - * XXX: We don't get here if the page was restored from full page image. - * We don't bother to update the FSM in that case, it doesn't need to be + * XXX: Don't do this if the page was restored from full page image. We + * don't bother to update the FSM in that case, it doesn't need to be * totally accurate anyway. */ - if (freespace < BLCKSZ / 5) + if (action == BLK_NEEDS_REDO && freespace < BLCKSZ / 5) XLogRecordPageWithFreeSpace(xlrec->target.node, blkno, freespace); } @@ -7610,6 +7573,8 @@ heap_xlog_multi_insert(XLogRecPtr lsn, XLogRecord *record) { char *recdata = XLogRecGetData(record); xl_heap_multi_insert *xlrec; + RelFileNode rnode; + BlockNumber blkno; Buffer buffer; Page page; struct @@ -7619,10 +7584,10 @@ heap_xlog_multi_insert(XLogRecPtr lsn, XLogRecord *record) } tbuf; HeapTupleHeader htup; uint32 newlen; - Size freespace; - BlockNumber blkno; + Size freespace = 0; int i; bool isinit = (record->xl_info & XLOG_HEAP_INIT_PAGE) != 0; + XLogRedoAction action; /* * Insertion doesn't overwrite MVCC data, so no conflict processing is @@ -7632,6 +7597,9 @@ heap_xlog_multi_insert(XLogRecPtr lsn, XLogRecord *record) xlrec = (xl_heap_multi_insert *) recdata; recdata += SizeOfHeapMultiInsert; + rnode = xlrec->node; + blkno = xlrec->blkno; + /* * If we're reinitializing the page, the tuples are stored in order from * FirstOffsetNumber. Otherwise there's an array of offsets in the WAL @@ -7640,15 +7608,13 @@ heap_xlog_multi_insert(XLogRecPtr lsn, XLogRecord *record) if (!isinit) recdata += sizeof(OffsetNumber) * xlrec->ntuples; - blkno = xlrec->blkno; - /* * The visibility map may need to be fixed even if the heap page is * already up-to-date. */ if (xlrec->flags & XLOG_HEAP_ALL_VISIBLE_CLEARED) { - Relation reln = CreateFakeRelcacheEntry(xlrec->node); + Relation reln = CreateFakeRelcacheEntry(rnode); Buffer vmbuffer = InvalidBuffer; visibilitymap_pin(reln, blkno, &vmbuffer); @@ -7657,94 +7623,82 @@ heap_xlog_multi_insert(XLogRecPtr lsn, XLogRecord *record) FreeFakeRelcacheEntry(reln); } - /* If we have a full-page image, restore it and we're done */ - if (record->xl_info & XLR_BKP_BLOCK(0)) - { - (void) RestoreBackupBlock(lsn, record, 0, false, false); - return; - } - if (isinit) { - buffer = XLogReadBuffer(xlrec->node, blkno, true); - Assert(BufferIsValid(buffer)); - page = (Page) BufferGetPage(buffer); - + XLogReadBufferForRedoExtended(lsn, record, 0, + rnode, MAIN_FORKNUM, blkno, + RBM_ZERO, false, &buffer); + page = BufferGetPage(buffer); PageInit(page, BufferGetPageSize(buffer), 0); + action = BLK_NEEDS_REDO; } else - { - buffer = XLogReadBuffer(xlrec->node, blkno, false); - if (!BufferIsValid(buffer)) - return; - page = (Page) BufferGetPage(buffer); + action = XLogReadBufferForRedo(lsn, record, 0, rnode, blkno, &buffer); - if (lsn <= PageGetLSN(page)) /* changes are applied */ + if (action == BLK_NEEDS_REDO) + { + page = BufferGetPage(buffer); + for (i = 0; i < xlrec->ntuples; i++) { - UnlockReleaseBuffer(buffer); - return; + OffsetNumber offnum; + xl_multi_insert_tuple *xlhdr; + + if (isinit) + offnum = FirstOffsetNumber + i; + else + offnum = xlrec->offsets[i]; + if (PageGetMaxOffsetNumber(page) + 1 < offnum) + elog(PANIC, "heap_multi_insert_redo: invalid max offset number"); + + xlhdr = (xl_multi_insert_tuple *) SHORTALIGN(recdata); + recdata = ((char *) xlhdr) + SizeOfMultiInsertTuple; + + newlen = xlhdr->datalen; + Assert(newlen <= MaxHeapTupleSize); + htup = &tbuf.hdr; + MemSet((char *) htup, 0, sizeof(HeapTupleHeaderData)); + /* PG73FORMAT: get bitmap [+ padding] [+ oid] + data */ + memcpy((char *) htup + offsetof(HeapTupleHeaderData, t_bits), + (char *) recdata, + newlen); + recdata += newlen; + + newlen += offsetof(HeapTupleHeaderData, t_bits); + htup->t_infomask2 = xlhdr->t_infomask2; + htup->t_infomask = xlhdr->t_infomask; + htup->t_hoff = xlhdr->t_hoff; + HeapTupleHeaderSetXmin(htup, record->xl_xid); + HeapTupleHeaderSetCmin(htup, FirstCommandId); + ItemPointerSetBlockNumber(&htup->t_ctid, blkno); + ItemPointerSetOffsetNumber(&htup->t_ctid, offnum); + + offnum = PageAddItem(page, (Item) htup, newlen, offnum, true, true); + if (offnum == InvalidOffsetNumber) + elog(PANIC, "heap_multi_insert_redo: failed to add tuple"); } + + freespace = PageGetHeapFreeSpace(page); /* needed to update FSM below */ + + PageSetLSN(page, lsn); + + if (xlrec->flags & XLOG_HEAP_ALL_VISIBLE_CLEARED) + PageClearAllVisible(page); + + MarkBufferDirty(buffer); } - - for (i = 0; i < xlrec->ntuples; i++) - { - OffsetNumber offnum; - xl_multi_insert_tuple *xlhdr; - - if (isinit) - offnum = FirstOffsetNumber + i; - else - offnum = xlrec->offsets[i]; - if (PageGetMaxOffsetNumber(page) + 1 < offnum) - elog(PANIC, "heap_multi_insert_redo: invalid max offset number"); - - xlhdr = (xl_multi_insert_tuple *) SHORTALIGN(recdata); - recdata = ((char *) xlhdr) + SizeOfMultiInsertTuple; - - newlen = xlhdr->datalen; - Assert(newlen <= MaxHeapTupleSize); - htup = &tbuf.hdr; - MemSet((char *) htup, 0, sizeof(HeapTupleHeaderData)); - /* PG73FORMAT: get bitmap [+ padding] [+ oid] + data */ - memcpy((char *) htup + offsetof(HeapTupleHeaderData, t_bits), - (char *) recdata, - newlen); - recdata += newlen; - - newlen += offsetof(HeapTupleHeaderData, t_bits); - htup->t_infomask2 = xlhdr->t_infomask2; - htup->t_infomask = xlhdr->t_infomask; - htup->t_hoff = xlhdr->t_hoff; - HeapTupleHeaderSetXmin(htup, record->xl_xid); - HeapTupleHeaderSetCmin(htup, FirstCommandId); - ItemPointerSetBlockNumber(&htup->t_ctid, blkno); - ItemPointerSetOffsetNumber(&htup->t_ctid, offnum); - - offnum = PageAddItem(page, (Item) htup, newlen, offnum, true, true); - if (offnum == InvalidOffsetNumber) - elog(PANIC, "heap_multi_insert_redo: failed to add tuple"); - } - - freespace = PageGetHeapFreeSpace(page); /* needed to update FSM below */ - - PageSetLSN(page, lsn); - - if (xlrec->flags & XLOG_HEAP_ALL_VISIBLE_CLEARED) - PageClearAllVisible(page); - - MarkBufferDirty(buffer); - UnlockReleaseBuffer(buffer); + if (BufferIsValid(buffer)) + UnlockReleaseBuffer(buffer); /* * If the page is running low on free space, update the FSM as well. * Arbitrarily, our definition of "low" is less than 20%. We can't do much * better than that without knowing the fill-factor for the table. * - * XXX: We don't get here if the page was restored from full page image. - * We don't bother to update the FSM in that case, it doesn't need to be + * XXX: Don't do this if the page was restored from full page image. We + * don't bother to update the FSM in that case, it doesn't need to be * totally accurate anyway. */ - if (freespace < BLCKSZ / 5) + if (action == BLK_NEEDS_REDO && freespace < BLCKSZ / 5) XLogRecordPageWithFreeSpace(xlrec->node, blkno, freespace); } @@ -7755,8 +7709,9 @@ static void heap_xlog_update(XLogRecPtr lsn, XLogRecord *record, bool hot_update) { xl_heap_update *xlrec = (xl_heap_update *) XLogRecGetData(record); - bool samepage = (ItemPointerGetBlockNumber(&(xlrec->newtid)) == - ItemPointerGetBlockNumber(&(xlrec->target.tid))); + RelFileNode rnode; + BlockNumber oldblk; + BlockNumber newblk; Buffer obuffer, nbuffer; Page page; @@ -7775,24 +7730,29 @@ heap_xlog_update(XLogRecPtr lsn, XLogRecord *record, bool hot_update) } tbuf; xl_heap_header_len xlhdr; uint32 newlen; - Size freespace; + Size freespace = 0; + XLogRedoAction oldaction; + XLogRedoAction newaction; /* initialize to keep the compiler quiet */ oldtup.t_data = NULL; oldtup.t_len = 0; + rnode = xlrec->target.node; + newblk = ItemPointerGetBlockNumber(&xlrec->newtid); + oldblk = ItemPointerGetBlockNumber(&xlrec->target.tid); + /* * The visibility map may need to be fixed even if the heap page is * already up-to-date. */ if (xlrec->flags & XLOG_HEAP_ALL_VISIBLE_CLEARED) { - Relation reln = CreateFakeRelcacheEntry(xlrec->target.node); - BlockNumber block = ItemPointerGetBlockNumber(&xlrec->target.tid); + Relation reln = CreateFakeRelcacheEntry(rnode); Buffer vmbuffer = InvalidBuffer; - visibilitymap_pin(reln, block, &vmbuffer); - visibilitymap_clear(reln, block, vmbuffer); + visibilitymap_pin(reln, oldblk, &vmbuffer); + visibilitymap_clear(reln, oldblk, vmbuffer); ReleaseBuffer(vmbuffer); FreeFakeRelcacheEntry(reln); } @@ -7807,84 +7767,67 @@ heap_xlog_update(XLogRecPtr lsn, XLogRecord *record, bool hot_update) * added the new tuple to the new page. */ - if (record->xl_info & XLR_BKP_BLOCK(0)) - { - obuffer = RestoreBackupBlock(lsn, record, 0, false, true); - if (samepage) - { - /* backup block covered both changes, so we're done */ - UnlockReleaseBuffer(obuffer); - return; - } - goto newt; - } - /* Deal with old tuple version */ - - obuffer = XLogReadBuffer(xlrec->target.node, - ItemPointerGetBlockNumber(&(xlrec->target.tid)), - false); - if (!BufferIsValid(obuffer)) - goto newt; - page = (Page) BufferGetPage(obuffer); - - if (lsn <= PageGetLSN(page)) /* changes are applied */ + oldaction = XLogReadBufferForRedo(lsn, record, 0, rnode, oldblk, &obuffer); + if (oldaction == BLK_NEEDS_REDO) { - if (samepage) - { - UnlockReleaseBuffer(obuffer); - return; - } - goto newt; + page = (Page) BufferGetPage(obuffer); + + offnum = ItemPointerGetOffsetNumber(&(xlrec->target.tid)); + if (PageGetMaxOffsetNumber(page) >= offnum) + lp = PageGetItemId(page, offnum); + + if (PageGetMaxOffsetNumber(page) < offnum || !ItemIdIsNormal(lp)) + elog(PANIC, "heap_update_redo: invalid lp"); + + htup = (HeapTupleHeader) PageGetItem(page, lp); + + oldtup.t_data = htup; + oldtup.t_len = ItemIdGetLength(lp); + + htup->t_infomask &= ~(HEAP_XMAX_BITS | HEAP_MOVED); + htup->t_infomask2 &= ~HEAP_KEYS_UPDATED; + if (hot_update) + HeapTupleHeaderSetHotUpdated(htup); + else + HeapTupleHeaderClearHotUpdated(htup); + fix_infomask_from_infobits(xlrec->old_infobits_set, &htup->t_infomask, + &htup->t_infomask2); + HeapTupleHeaderSetXmax(htup, xlrec->old_xmax); + HeapTupleHeaderSetCmax(htup, FirstCommandId, false); + /* Set forward chain link in t_ctid */ + htup->t_ctid = xlrec->newtid; + + /* Mark the page as a candidate for pruning */ + PageSetPrunable(page, record->xl_xid); + + if (xlrec->flags & XLOG_HEAP_ALL_VISIBLE_CLEARED) + PageClearAllVisible(page); + + PageSetLSN(page, lsn); + MarkBufferDirty(obuffer); } - offnum = ItemPointerGetOffsetNumber(&(xlrec->target.tid)); - if (PageGetMaxOffsetNumber(page) >= offnum) - lp = PageGetItemId(page, offnum); - - if (PageGetMaxOffsetNumber(page) < offnum || !ItemIdIsNormal(lp)) - elog(PANIC, "heap_update_redo: invalid lp"); - - htup = (HeapTupleHeader) PageGetItem(page, lp); - - oldtup.t_data = htup; - oldtup.t_len = ItemIdGetLength(lp); - - htup->t_infomask &= ~(HEAP_XMAX_BITS | HEAP_MOVED); - htup->t_infomask2 &= ~HEAP_KEYS_UPDATED; - if (hot_update) - HeapTupleHeaderSetHotUpdated(htup); - else - HeapTupleHeaderClearHotUpdated(htup); - fix_infomask_from_infobits(xlrec->old_infobits_set, &htup->t_infomask, - &htup->t_infomask2); - HeapTupleHeaderSetXmax(htup, xlrec->old_xmax); - HeapTupleHeaderSetCmax(htup, FirstCommandId, false); - /* Set forward chain link in t_ctid */ - htup->t_ctid = xlrec->newtid; - - /* Mark the page as a candidate for pruning */ - PageSetPrunable(page, record->xl_xid); - - if (xlrec->flags & XLOG_HEAP_ALL_VISIBLE_CLEARED) - PageClearAllVisible(page); - /* - * this test is ugly, but necessary to avoid thinking that insert change - * is already applied + * Read the page the new tuple goes into, if different from old. */ - if (samepage) + if (oldblk == newblk) { nbuffer = obuffer; - goto newsame; + newaction = oldaction; } - - PageSetLSN(page, lsn); - MarkBufferDirty(obuffer); - - /* Deal with new tuple */ - -newt:; + else if (record->xl_info & XLOG_HEAP_INIT_PAGE) + { + XLogReadBufferForRedoExtended(lsn, record, 1, + rnode, MAIN_FORKNUM, newblk, + RBM_ZERO, false, &nbuffer); + page = (Page) BufferGetPage(nbuffer); + PageInit(page, BufferGetPageSize(nbuffer), 0); + newaction = BLK_NEEDS_REDO; + } + else + newaction = XLogReadBufferForRedo(lsn, record, 1, rnode, newblk, + &nbuffer); /* * The visibility map may need to be fixed even if the heap page is @@ -7893,144 +7836,110 @@ newt:; if (xlrec->flags & XLOG_HEAP_NEW_ALL_VISIBLE_CLEARED) { Relation reln = CreateFakeRelcacheEntry(xlrec->target.node); - BlockNumber block = ItemPointerGetBlockNumber(&xlrec->newtid); Buffer vmbuffer = InvalidBuffer; - visibilitymap_pin(reln, block, &vmbuffer); - visibilitymap_clear(reln, block, vmbuffer); + visibilitymap_pin(reln, newblk, &vmbuffer); + visibilitymap_clear(reln, newblk, vmbuffer); ReleaseBuffer(vmbuffer); FreeFakeRelcacheEntry(reln); } - if (record->xl_info & XLR_BKP_BLOCK(1)) + /* Deal with new tuple */ + if (newaction == BLK_NEEDS_REDO) { - (void) RestoreBackupBlock(lsn, record, 1, false, false); - if (BufferIsValid(obuffer)) - UnlockReleaseBuffer(obuffer); - return; - } - - if (record->xl_info & XLOG_HEAP_INIT_PAGE) - { - nbuffer = XLogReadBuffer(xlrec->target.node, - ItemPointerGetBlockNumber(&(xlrec->newtid)), - true); - Assert(BufferIsValid(nbuffer)); page = (Page) BufferGetPage(nbuffer); - PageInit(page, BufferGetPageSize(nbuffer), 0); - } - else - { - nbuffer = XLogReadBuffer(xlrec->target.node, - ItemPointerGetBlockNumber(&(xlrec->newtid)), - false); - if (!BufferIsValid(nbuffer)) + offnum = ItemPointerGetOffsetNumber(&(xlrec->newtid)); + if (PageGetMaxOffsetNumber(page) + 1 < offnum) + elog(PANIC, "heap_update_redo: invalid max offset number"); + + recdata = (char *) xlrec + SizeOfHeapUpdate; + + if (xlrec->flags & XLOG_HEAP_PREFIX_FROM_OLD) { - if (BufferIsValid(obuffer)) - UnlockReleaseBuffer(obuffer); - return; + Assert(newblk == oldblk); + memcpy(&prefixlen, recdata, sizeof(uint16)); + recdata += sizeof(uint16); } - page = (Page) BufferGetPage(nbuffer); - - if (lsn <= PageGetLSN(page)) /* changes are applied */ + if (xlrec->flags & XLOG_HEAP_SUFFIX_FROM_OLD) { - UnlockReleaseBuffer(nbuffer); - if (BufferIsValid(obuffer)) - UnlockReleaseBuffer(obuffer); - return; + Assert(newblk == oldblk); + memcpy(&suffixlen, recdata, sizeof(uint16)); + recdata += sizeof(uint16); } + + memcpy((char *) &xlhdr, recdata, SizeOfHeapHeaderLen); + recdata += SizeOfHeapHeaderLen; + + Assert(xlhdr.t_len + prefixlen + suffixlen <= MaxHeapTupleSize); + htup = &tbuf.hdr; + MemSet((char *) htup, 0, sizeof(HeapTupleHeaderData)); + + /* + * Reconstruct the new tuple using the prefix and/or suffix from the + * old tuple, and the data stored in the WAL record. + */ + newp = (char *) htup + offsetof(HeapTupleHeaderData, t_bits); + if (prefixlen > 0) + { + int len; + + /* copy bitmap [+ padding] [+ oid] from WAL record */ + len = xlhdr.header.t_hoff - offsetof(HeapTupleHeaderData, t_bits); + memcpy(newp, recdata, len); + recdata += len; + newp += len; + + /* copy prefix from old tuple */ + memcpy(newp, (char *) oldtup.t_data + oldtup.t_data->t_hoff, prefixlen); + newp += prefixlen; + + /* copy new tuple data from WAL record */ + len = xlhdr.t_len - (xlhdr.header.t_hoff - offsetof(HeapTupleHeaderData, t_bits)); + memcpy(newp, recdata, len); + recdata += len; + newp += len; + } + else + { + /* + * copy bitmap [+ padding] [+ oid] + data from record, all in one + * go + */ + memcpy(newp, recdata, xlhdr.t_len); + recdata += xlhdr.t_len; + newp += xlhdr.t_len; + } + /* copy suffix from old tuple */ + if (suffixlen > 0) + memcpy(newp, (char *) oldtup.t_data + oldtup.t_len - suffixlen, suffixlen); + + newlen = offsetof(HeapTupleHeaderData, t_bits) + xlhdr.t_len + prefixlen + suffixlen; + htup->t_infomask2 = xlhdr.header.t_infomask2; + htup->t_infomask = xlhdr.header.t_infomask; + htup->t_hoff = xlhdr.header.t_hoff; + + HeapTupleHeaderSetXmin(htup, record->xl_xid); + HeapTupleHeaderSetCmin(htup, FirstCommandId); + HeapTupleHeaderSetXmax(htup, xlrec->new_xmax); + /* Make sure there is no forward chain link in t_ctid */ + htup->t_ctid = xlrec->newtid; + + offnum = PageAddItem(page, (Item) htup, newlen, offnum, true, true); + if (offnum == InvalidOffsetNumber) + elog(PANIC, "heap_update_redo: failed to add tuple"); + + if (xlrec->flags & XLOG_HEAP_NEW_ALL_VISIBLE_CLEARED) + PageClearAllVisible(page); + + freespace = PageGetHeapFreeSpace(page); /* needed to update FSM below */ + + PageSetLSN(page, lsn); + MarkBufferDirty(nbuffer); } - -newsame:; - - offnum = ItemPointerGetOffsetNumber(&(xlrec->newtid)); - if (PageGetMaxOffsetNumber(page) + 1 < offnum) - elog(PANIC, "heap_update_redo: invalid max offset number"); - - recdata = (char *) xlrec + SizeOfHeapUpdate; - - if (xlrec->flags & XLOG_HEAP_PREFIX_FROM_OLD) - { - Assert(samepage); - memcpy(&prefixlen, recdata, sizeof(uint16)); - recdata += sizeof(uint16); - } - if (xlrec->flags & XLOG_HEAP_SUFFIX_FROM_OLD) - { - Assert(samepage); - memcpy(&suffixlen, recdata, sizeof(uint16)); - recdata += sizeof(uint16); - } - - memcpy((char *) &xlhdr, recdata, SizeOfHeapHeaderLen); - recdata += SizeOfHeapHeaderLen; - - Assert(xlhdr.t_len + prefixlen + suffixlen <= MaxHeapTupleSize); - htup = &tbuf.hdr; - MemSet((char *) htup, 0, sizeof(HeapTupleHeaderData)); - - /* - * Reconstruct the new tuple using the prefix and/or suffix from the old - * tuple, and the data stored in the WAL record. - */ - newp = (char *) htup + offsetof(HeapTupleHeaderData, t_bits); - if (prefixlen > 0) - { - int len; - - /* copy bitmap [+ padding] [+ oid] from WAL record */ - len = xlhdr.header.t_hoff - offsetof(HeapTupleHeaderData, t_bits); - memcpy(newp, recdata, len); - recdata += len; - newp += len; - - /* copy prefix from old tuple */ - memcpy(newp, (char *) oldtup.t_data + oldtup.t_data->t_hoff, prefixlen); - newp += prefixlen; - - /* copy new tuple data from WAL record */ - len = xlhdr.t_len - (xlhdr.header.t_hoff - offsetof(HeapTupleHeaderData, t_bits)); - memcpy(newp, recdata, len); - recdata += len; - newp += len; - } - else - { - /* copy bitmap [+ padding] [+ oid] + data from record, all in one go */ - memcpy(newp, recdata, xlhdr.t_len); - recdata += xlhdr.t_len; - newp += xlhdr.t_len; - } - /* copy suffix from old tuple */ - if (suffixlen > 0) - memcpy(newp, (char *) oldtup.t_data + oldtup.t_len - suffixlen, suffixlen); - - newlen = offsetof(HeapTupleHeaderData, t_bits) +xlhdr.t_len + prefixlen + suffixlen; - htup->t_infomask2 = xlhdr.header.t_infomask2; - htup->t_infomask = xlhdr.header.t_infomask; - htup->t_hoff = xlhdr.header.t_hoff; - - HeapTupleHeaderSetXmin(htup, record->xl_xid); - HeapTupleHeaderSetCmin(htup, FirstCommandId); - HeapTupleHeaderSetXmax(htup, xlrec->new_xmax); - /* Make sure there is no forward chain link in t_ctid */ - htup->t_ctid = xlrec->newtid; - - offnum = PageAddItem(page, (Item) htup, newlen, offnum, true, true); - if (offnum == InvalidOffsetNumber) - elog(PANIC, "heap_update_redo: failed to add tuple"); - - if (xlrec->flags & XLOG_HEAP_NEW_ALL_VISIBLE_CLEARED) - PageClearAllVisible(page); - - freespace = PageGetHeapFreeSpace(page); /* needed to update FSM below */ - - PageSetLSN(page, lsn); - MarkBufferDirty(nbuffer); - UnlockReleaseBuffer(nbuffer); - - if (BufferIsValid(obuffer) && obuffer != nbuffer) + if (BufferIsValid(nbuffer) && nbuffer != obuffer) + UnlockReleaseBuffer(nbuffer); + if (BufferIsValid(obuffer)) UnlockReleaseBuffer(obuffer); /* @@ -8044,11 +7953,11 @@ newsame:; * as it did before the update, assuming the new tuple is about the same * size as the old one. * - * XXX: We don't get here if the page was restored from full page image. - * We don't bother to update the FSM in that case, it doesn't need to be + * XXX: Don't do this if the page was restored from full page image. We + * don't bother to update the FSM in that case, it doesn't need to be * totally accurate anyway. */ - if (!hot_update && freespace < BLCKSZ / 5) + if (newaction == BLK_NEEDS_REDO && !hot_update && freespace < BLCKSZ / 5) XLogRecordPageWithFreeSpace(xlrec->target.node, ItemPointerGetBlockNumber(&(xlrec->newtid)), freespace); @@ -8064,53 +7973,41 @@ heap_xlog_lock(XLogRecPtr lsn, XLogRecord *record) ItemId lp = NULL; HeapTupleHeader htup; - /* If we have a full-page image, restore it and we're done */ - if (record->xl_info & XLR_BKP_BLOCK(0)) + if (XLogReadBufferForRedo(lsn, record, 0, xlrec->target.node, + ItemPointerGetBlockNumber(&xlrec->target.tid), + &buffer) == BLK_NEEDS_REDO) { - (void) RestoreBackupBlock(lsn, record, 0, false, false); - return; + page = (Page) BufferGetPage(buffer); + + offnum = ItemPointerGetOffsetNumber(&(xlrec->target.tid)); + if (PageGetMaxOffsetNumber(page) >= offnum) + lp = PageGetItemId(page, offnum); + + if (PageGetMaxOffsetNumber(page) < offnum || !ItemIdIsNormal(lp)) + elog(PANIC, "heap_lock_redo: invalid lp"); + + htup = (HeapTupleHeader) PageGetItem(page, lp); + + fix_infomask_from_infobits(xlrec->infobits_set, &htup->t_infomask, + &htup->t_infomask2); + + /* + * Clear relevant update flags, but only if the modified infomask says + * there's no update. + */ + if (HEAP_XMAX_IS_LOCKED_ONLY(htup->t_infomask)) + { + HeapTupleHeaderClearHotUpdated(htup); + /* Make sure there is no forward chain link in t_ctid */ + htup->t_ctid = xlrec->target.tid; + } + HeapTupleHeaderSetXmax(htup, xlrec->locking_xid); + HeapTupleHeaderSetCmax(htup, FirstCommandId, false); + PageSetLSN(page, lsn); + MarkBufferDirty(buffer); } - - buffer = XLogReadBuffer(xlrec->target.node, - ItemPointerGetBlockNumber(&(xlrec->target.tid)), - false); - if (!BufferIsValid(buffer)) - return; - page = (Page) BufferGetPage(buffer); - - if (lsn <= PageGetLSN(page)) /* changes are applied */ - { + if (BufferIsValid(buffer)) UnlockReleaseBuffer(buffer); - return; - } - - offnum = ItemPointerGetOffsetNumber(&(xlrec->target.tid)); - if (PageGetMaxOffsetNumber(page) >= offnum) - lp = PageGetItemId(page, offnum); - - if (PageGetMaxOffsetNumber(page) < offnum || !ItemIdIsNormal(lp)) - elog(PANIC, "heap_lock_redo: invalid lp"); - - htup = (HeapTupleHeader) PageGetItem(page, lp); - - fix_infomask_from_infobits(xlrec->infobits_set, &htup->t_infomask, - &htup->t_infomask2); - - /* - * Clear relevant update flags, but only if the modified infomask says - * there's no update. - */ - if (HEAP_XMAX_IS_LOCKED_ONLY(htup->t_infomask)) - { - HeapTupleHeaderClearHotUpdated(htup); - /* Make sure there is no forward chain link in t_ctid */ - htup->t_ctid = xlrec->target.tid; - } - HeapTupleHeaderSetXmax(htup, xlrec->locking_xid); - HeapTupleHeaderSetCmax(htup, FirstCommandId, false); - PageSetLSN(page, lsn); - MarkBufferDirty(buffer); - UnlockReleaseBuffer(buffer); } static void @@ -8124,42 +8021,29 @@ heap_xlog_lock_updated(XLogRecPtr lsn, XLogRecord *record) ItemId lp = NULL; HeapTupleHeader htup; - /* If we have a full-page image, restore it and we're done */ - if (record->xl_info & XLR_BKP_BLOCK(0)) + if (XLogReadBufferForRedo(lsn, record, 0, xlrec->target.node, + ItemPointerGetBlockNumber(&(xlrec->target.tid)), + &buffer) == BLK_NEEDS_REDO) { - (void) RestoreBackupBlock(lsn, record, 0, false, false); - return; + page = BufferGetPage(buffer); + offnum = ItemPointerGetOffsetNumber(&(xlrec->target.tid)); + if (PageGetMaxOffsetNumber(page) >= offnum) + lp = PageGetItemId(page, offnum); + + if (PageGetMaxOffsetNumber(page) < offnum || !ItemIdIsNormal(lp)) + elog(PANIC, "heap_xlog_lock_updated: invalid lp"); + + htup = (HeapTupleHeader) PageGetItem(page, lp); + + fix_infomask_from_infobits(xlrec->infobits_set, &htup->t_infomask, + &htup->t_infomask2); + HeapTupleHeaderSetXmax(htup, xlrec->xmax); + + PageSetLSN(page, lsn); + MarkBufferDirty(buffer); } - - buffer = XLogReadBuffer(xlrec->target.node, - ItemPointerGetBlockNumber(&(xlrec->target.tid)), - false); - if (!BufferIsValid(buffer)) - return; - page = (Page) BufferGetPage(buffer); - - if (lsn <= PageGetLSN(page)) /* changes are applied */ - { + if (BufferIsValid(buffer)) UnlockReleaseBuffer(buffer); - return; - } - - offnum = ItemPointerGetOffsetNumber(&(xlrec->target.tid)); - if (PageGetMaxOffsetNumber(page) >= offnum) - lp = PageGetItemId(page, offnum); - - if (PageGetMaxOffsetNumber(page) < offnum || !ItemIdIsNormal(lp)) - elog(PANIC, "heap_xlog_lock_updated: invalid lp"); - - htup = (HeapTupleHeader) PageGetItem(page, lp); - - fix_infomask_from_infobits(xlrec->infobits_set, &htup->t_infomask, - &htup->t_infomask2); - HeapTupleHeaderSetXmax(htup, xlrec->xmax); - - PageSetLSN(page, lsn); - MarkBufferDirty(buffer); - UnlockReleaseBuffer(buffer); } static void @@ -8174,47 +8058,35 @@ heap_xlog_inplace(XLogRecPtr lsn, XLogRecord *record) uint32 oldlen; uint32 newlen; - /* If we have a full-page image, restore it and we're done */ - if (record->xl_info & XLR_BKP_BLOCK(0)) + if (XLogReadBufferForRedo(lsn, record, 0, xlrec->target.node, + ItemPointerGetBlockNumber(&(xlrec->target.tid)), + &buffer) == BLK_NEEDS_REDO) { - (void) RestoreBackupBlock(lsn, record, 0, false, false); - return; + page = BufferGetPage(buffer); + + offnum = ItemPointerGetOffsetNumber(&(xlrec->target.tid)); + if (PageGetMaxOffsetNumber(page) >= offnum) + lp = PageGetItemId(page, offnum); + + if (PageGetMaxOffsetNumber(page) < offnum || !ItemIdIsNormal(lp)) + elog(PANIC, "heap_inplace_redo: invalid lp"); + + htup = (HeapTupleHeader) PageGetItem(page, lp); + + oldlen = ItemIdGetLength(lp) - htup->t_hoff; + newlen = record->xl_len - SizeOfHeapInplace; + if (oldlen != newlen) + elog(PANIC, "heap_inplace_redo: wrong tuple length"); + + memcpy((char *) htup + htup->t_hoff, + (char *) xlrec + SizeOfHeapInplace, + newlen); + + PageSetLSN(page, lsn); + MarkBufferDirty(buffer); } - - buffer = XLogReadBuffer(xlrec->target.node, - ItemPointerGetBlockNumber(&(xlrec->target.tid)), - false); - if (!BufferIsValid(buffer)) - return; - page = (Page) BufferGetPage(buffer); - - if (lsn <= PageGetLSN(page)) /* changes are applied */ - { + if (BufferIsValid(buffer)) UnlockReleaseBuffer(buffer); - return; - } - - offnum = ItemPointerGetOffsetNumber(&(xlrec->target.tid)); - if (PageGetMaxOffsetNumber(page) >= offnum) - lp = PageGetItemId(page, offnum); - - if (PageGetMaxOffsetNumber(page) < offnum || !ItemIdIsNormal(lp)) - elog(PANIC, "heap_inplace_redo: invalid lp"); - - htup = (HeapTupleHeader) PageGetItem(page, lp); - - oldlen = ItemIdGetLength(lp) - htup->t_hoff; - newlen = record->xl_len - SizeOfHeapInplace; - if (oldlen != newlen) - elog(PANIC, "heap_inplace_redo: wrong tuple length"); - - memcpy((char *) htup + htup->t_hoff, - (char *) xlrec + SizeOfHeapInplace, - newlen); - - PageSetLSN(page, lsn); - MarkBufferDirty(buffer); - UnlockReleaseBuffer(buffer); } void diff --git a/src/backend/access/nbtree/nbtxlog.c b/src/backend/access/nbtree/nbtxlog.c index 5f9fc49e78..fac006e297 100644 --- a/src/backend/access/nbtree/nbtxlog.c +++ b/src/backend/access/nbtree/nbtxlog.c @@ -116,27 +116,25 @@ _bt_restore_meta(RelFileNode rnode, XLogRecPtr lsn, */ static void _bt_clear_incomplete_split(XLogRecPtr lsn, XLogRecord *record, + int block_index, RelFileNode rnode, BlockNumber cblock) { Buffer buf; - buf = XLogReadBuffer(rnode, cblock, false); - if (BufferIsValid(buf)) + if (XLogReadBufferForRedo(lsn, record, block_index, rnode, cblock, &buf) + == BLK_NEEDS_REDO) { Page page = (Page) BufferGetPage(buf); + BTPageOpaque pageop = (BTPageOpaque) PageGetSpecialPointer(page); - if (lsn > PageGetLSN(page)) - { - BTPageOpaque pageop = (BTPageOpaque) PageGetSpecialPointer(page); + Assert((pageop->btpo_flags & BTP_INCOMPLETE_SPLIT) != 0); + pageop->btpo_flags &= ~BTP_INCOMPLETE_SPLIT; - Assert((pageop->btpo_flags & BTP_INCOMPLETE_SPLIT) != 0); - pageop->btpo_flags &= ~BTP_INCOMPLETE_SPLIT; - - PageSetLSN(page, lsn); - MarkBufferDirty(buf); - } - UnlockReleaseBuffer(buf); + PageSetLSN(page, lsn); + MarkBufferDirty(buf); } + if (BufferIsValid(buf)) + UnlockReleaseBuffer(buf); } static void @@ -184,39 +182,28 @@ btree_xlog_insert(bool isleaf, bool ismeta, */ if (!isleaf) { - if (record->xl_info & XLR_BKP_BLOCK(0)) - (void) RestoreBackupBlock(lsn, record, 0, false, false); - else - _bt_clear_incomplete_split(lsn, record, xlrec->target.node, cblkno); + _bt_clear_incomplete_split(lsn, record, 0, xlrec->target.node, cblkno); main_blk_index = 1; } else main_blk_index = 0; - if (record->xl_info & XLR_BKP_BLOCK(main_blk_index)) - (void) RestoreBackupBlock(lsn, record, main_blk_index, false, false); - else + if (XLogReadBufferForRedo(lsn, record, main_blk_index, xlrec->target.node, + ItemPointerGetBlockNumber(&(xlrec->target.tid)), + &buffer) == BLK_NEEDS_REDO) { - buffer = XLogReadBuffer(xlrec->target.node, - ItemPointerGetBlockNumber(&(xlrec->target.tid)), - false); - if (BufferIsValid(buffer)) - { - page = (Page) BufferGetPage(buffer); + page = BufferGetPage(buffer); - if (lsn > PageGetLSN(page)) - { - if (PageAddItem(page, (Item) datapos, datalen, - ItemPointerGetOffsetNumber(&(xlrec->target.tid)), - false, false) == InvalidOffsetNumber) - elog(PANIC, "btree_insert_redo: failed to add item"); + if (PageAddItem(page, (Item) datapos, datalen, + ItemPointerGetOffsetNumber(&(xlrec->target.tid)), + false, false) == InvalidOffsetNumber) + elog(PANIC, "btree_insert_redo: failed to add item"); - PageSetLSN(page, lsn); - MarkBufferDirty(buffer); - } - UnlockReleaseBuffer(buffer); - } + PageSetLSN(page, lsn); + MarkBufferDirty(buffer); } + if (BufferIsValid(buffer)) + UnlockReleaseBuffer(buffer); /* * Note: in normal operation, we'd update the metapage while still holding @@ -299,12 +286,7 @@ btree_xlog_split(bool onleft, bool isroot, * before locking the other pages) */ if (!isleaf) - { - if (record->xl_info & XLR_BKP_BLOCK(1)) - (void) RestoreBackupBlock(lsn, record, 1, false, false); - else - _bt_clear_incomplete_split(lsn, record, xlrec->node, cblkno); - } + _bt_clear_incomplete_split(lsn, record, 1, xlrec->node, cblkno); /* Reconstruct right (new) sibling page from scratch */ rbuf = XLogReadBuffer(xlrec->node, xlrec->rightsib, true); @@ -340,87 +322,77 @@ btree_xlog_split(bool onleft, bool isroot, /* don't release the buffer yet; we touch right page's first item below */ /* Now reconstruct left (original) sibling page */ - if (record->xl_info & XLR_BKP_BLOCK(0)) - lbuf = RestoreBackupBlock(lsn, record, 0, false, true); - else + if (XLogReadBufferForRedo(lsn, record, 0, xlrec->node, xlrec->leftsib, + &lbuf) == BLK_NEEDS_REDO) { - lbuf = XLogReadBuffer(xlrec->node, xlrec->leftsib, false); + /* + * To retain the same physical order of the tuples that they had, we + * initialize a temporary empty page for the left page and add all the + * items to that in item number order. This mirrors how _bt_split() + * works. It's not strictly required to retain the same physical + * order, as long as the items are in the correct item number order, + * but it helps debugging. See also _bt_restore_page(), which does + * the same for the right page. + */ + Page lpage = (Page) BufferGetPage(lbuf); + BTPageOpaque lopaque = (BTPageOpaque) PageGetSpecialPointer(lpage); + OffsetNumber off; + Page newlpage; + OffsetNumber leftoff; - if (BufferIsValid(lbuf)) + newlpage = PageGetTempPageCopySpecial(lpage); + + /* Set high key */ + leftoff = P_HIKEY; + if (PageAddItem(newlpage, left_hikey, left_hikeysz, + P_HIKEY, false, false) == InvalidOffsetNumber) + elog(PANIC, "failed to add high key to left page after split"); + leftoff = OffsetNumberNext(leftoff); + + for (off = P_FIRSTDATAKEY(lopaque); off < xlrec->firstright; off++) { - /* - * To retain the same physical order of the tuples that they had, - * we initialize a temporary empty page for the left page and add - * all the items to that in item number order. This mirrors how - * _bt_split() works. It's not strictly required to retain the - * same physical order, as long as the items are in the correct - * item number order, but it helps debugging. See also - * _bt_restore_page(), which does the same for the right page. - */ - Page lpage = (Page) BufferGetPage(lbuf); - BTPageOpaque lopaque = (BTPageOpaque) PageGetSpecialPointer(lpage); + ItemId itemid; + Size itemsz; + Item item; - if (lsn > PageGetLSN(lpage)) + /* add the new item if it was inserted on left page */ + if (onleft && off == newitemoff) { - OffsetNumber off; - Page newlpage; - OffsetNumber leftoff; - - newlpage = PageGetTempPageCopySpecial(lpage); - - /* Set high key */ - leftoff = P_HIKEY; - if (PageAddItem(newlpage, left_hikey, left_hikeysz, - P_HIKEY, false, false) == InvalidOffsetNumber) - elog(PANIC, "failed to add high key to left page after split"); + if (PageAddItem(newlpage, newitem, newitemsz, leftoff, + false, false) == InvalidOffsetNumber) + elog(ERROR, "failed to add new item to left page after split"); leftoff = OffsetNumberNext(leftoff); - - for (off = P_FIRSTDATAKEY(lopaque); off < xlrec->firstright; off++) - { - ItemId itemid; - Size itemsz; - Item item; - - /* add the new item if it was inserted on left page */ - if (onleft && off == newitemoff) - { - if (PageAddItem(newlpage, newitem, newitemsz, leftoff, - false, false) == InvalidOffsetNumber) - elog(ERROR, "failed to add new item to left page after split"); - leftoff = OffsetNumberNext(leftoff); - } - - itemid = PageGetItemId(lpage, off); - itemsz = ItemIdGetLength(itemid); - item = PageGetItem(lpage, itemid); - if (PageAddItem(newlpage, item, itemsz, leftoff, - false, false) == InvalidOffsetNumber) - elog(ERROR, "failed to add old item to left page after split"); - leftoff = OffsetNumberNext(leftoff); - } - - /* cope with possibility that newitem goes at the end */ - if (onleft && off == newitemoff) - { - if (PageAddItem(newlpage, newitem, newitemsz, leftoff, - false, false) == InvalidOffsetNumber) - elog(ERROR, "failed to add new item to left page after split"); - leftoff = OffsetNumberNext(leftoff); - } - - PageRestoreTempPage(newlpage, lpage); - - /* Fix opaque fields */ - lopaque->btpo_flags = BTP_INCOMPLETE_SPLIT; - if (isleaf) - lopaque->btpo_flags |= BTP_LEAF; - lopaque->btpo_next = xlrec->rightsib; - lopaque->btpo_cycleid = 0; - - PageSetLSN(lpage, lsn); - MarkBufferDirty(lbuf); } + + itemid = PageGetItemId(lpage, off); + itemsz = ItemIdGetLength(itemid); + item = PageGetItem(lpage, itemid); + if (PageAddItem(newlpage, item, itemsz, leftoff, + false, false) == InvalidOffsetNumber) + elog(ERROR, "failed to add old item to left page after split"); + leftoff = OffsetNumberNext(leftoff); } + + /* cope with possibility that newitem goes at the end */ + if (onleft && off == newitemoff) + { + if (PageAddItem(newlpage, newitem, newitemsz, leftoff, + false, false) == InvalidOffsetNumber) + elog(ERROR, "failed to add new item to left page after split"); + leftoff = OffsetNumberNext(leftoff); + } + + PageRestoreTempPage(newlpage, lpage); + + /* Fix opaque fields */ + lopaque->btpo_flags = BTP_INCOMPLETE_SPLIT; + if (isleaf) + lopaque->btpo_flags |= BTP_LEAF; + lopaque->btpo_next = xlrec->rightsib; + lopaque->btpo_cycleid = 0; + + PageSetLSN(lpage, lsn); + MarkBufferDirty(lbuf); } /* We no longer need the buffers */ @@ -443,31 +415,21 @@ btree_xlog_split(bool onleft, bool isroot, * whether this was a leaf or internal page. */ int rnext_index = isleaf ? 1 : 2; + Buffer buffer; - if (record->xl_info & XLR_BKP_BLOCK(rnext_index)) - (void) RestoreBackupBlock(lsn, record, rnext_index, false, false); - else + if (XLogReadBufferForRedo(lsn, record, rnext_index, xlrec->node, + xlrec->rnext, &buffer) == BLK_NEEDS_REDO) { - Buffer buffer; + Page page = (Page) BufferGetPage(buffer); + BTPageOpaque pageop = (BTPageOpaque) PageGetSpecialPointer(page); - buffer = XLogReadBuffer(xlrec->node, xlrec->rnext, false); + pageop->btpo_prev = xlrec->rightsib; - if (BufferIsValid(buffer)) - { - Page page = (Page) BufferGetPage(buffer); - - if (lsn > PageGetLSN(page)) - { - BTPageOpaque pageop = (BTPageOpaque) PageGetSpecialPointer(page); - - pageop->btpo_prev = xlrec->rightsib; - - PageSetLSN(page, lsn); - MarkBufferDirty(buffer); - } - UnlockReleaseBuffer(buffer); - } + PageSetLSN(page, lsn); + MarkBufferDirty(buffer); } + if (BufferIsValid(buffer)) + UnlockReleaseBuffer(buffer); } } @@ -529,54 +491,41 @@ btree_xlog_vacuum(XLogRecPtr lsn, XLogRecord *record) } } - /* - * If we have a full-page image, restore it (using a cleanup lock) and - * we're done. - */ - if (record->xl_info & XLR_BKP_BLOCK(0)) - { - (void) RestoreBackupBlock(lsn, record, 0, true, false); - return; - } - /* * Like in btvacuumpage(), we need to take a cleanup lock on every leaf * page. See nbtree/README for details. */ - buffer = XLogReadBufferExtended(xlrec->node, MAIN_FORKNUM, xlrec->block, RBM_NORMAL); - if (!BufferIsValid(buffer)) - return; - LockBufferForCleanup(buffer); - page = (Page) BufferGetPage(buffer); - - if (lsn <= PageGetLSN(page)) + if (XLogReadBufferForRedoExtended(lsn, record, 0, + xlrec->node, MAIN_FORKNUM, xlrec->block, + RBM_NORMAL, true, &buffer) + == BLK_NEEDS_REDO) { + page = (Page) BufferGetPage(buffer); + + if (record->xl_len > SizeOfBtreeVacuum) + { + OffsetNumber *unused; + OffsetNumber *unend; + + unused = (OffsetNumber *) ((char *) xlrec + SizeOfBtreeVacuum); + unend = (OffsetNumber *) ((char *) xlrec + record->xl_len); + + if ((unend - unused) > 0) + PageIndexMultiDelete(page, unused, unend - unused); + } + + /* + * Mark the page as not containing any LP_DEAD items --- see comments + * in _bt_delitems_vacuum(). + */ + opaque = (BTPageOpaque) PageGetSpecialPointer(page); + opaque->btpo_flags &= ~BTP_HAS_GARBAGE; + + PageSetLSN(page, lsn); + MarkBufferDirty(buffer); + } + if (BufferIsValid(buffer)) UnlockReleaseBuffer(buffer); - return; - } - - if (record->xl_len > SizeOfBtreeVacuum) - { - OffsetNumber *unused; - OffsetNumber *unend; - - unused = (OffsetNumber *) ((char *) xlrec + SizeOfBtreeVacuum); - unend = (OffsetNumber *) ((char *) xlrec + record->xl_len); - - if ((unend - unused) > 0) - PageIndexMultiDelete(page, unused, unend - unused); - } - - /* - * Mark the page as not containing any LP_DEAD items --- see comments in - * _bt_delitems_vacuum(). - */ - opaque = (BTPageOpaque) PageGetSpecialPointer(page); - opaque->btpo_flags &= ~BTP_HAS_GARBAGE; - - PageSetLSN(page, lsn); - MarkBufferDirty(buffer); - UnlockReleaseBuffer(buffer); } /* @@ -752,47 +701,36 @@ btree_xlog_delete(XLogRecPtr lsn, XLogRecord *record) ResolveRecoveryConflictWithSnapshot(latestRemovedXid, xlrec->node); } - /* If we have a full-page image, restore it and we're done */ - if (record->xl_info & XLR_BKP_BLOCK(0)) - { - (void) RestoreBackupBlock(lsn, record, 0, false, false); - return; - } - /* * We don't need to take a cleanup lock to apply these changes. See * nbtree/README for details. */ - buffer = XLogReadBuffer(xlrec->node, xlrec->block, false); - if (!BufferIsValid(buffer)) - return; - page = (Page) BufferGetPage(buffer); - - if (lsn <= PageGetLSN(page)) + if (XLogReadBufferForRedo(lsn, record, 0, xlrec->node, xlrec->block, + &buffer) == BLK_NEEDS_REDO) { + page = (Page) BufferGetPage(buffer); + + if (record->xl_len > SizeOfBtreeDelete) + { + OffsetNumber *unused; + + unused = (OffsetNumber *) ((char *) xlrec + SizeOfBtreeDelete); + + PageIndexMultiDelete(page, unused, xlrec->nitems); + } + + /* + * Mark the page as not containing any LP_DEAD items --- see comments + * in _bt_delitems_delete(). + */ + opaque = (BTPageOpaque) PageGetSpecialPointer(page); + opaque->btpo_flags &= ~BTP_HAS_GARBAGE; + + PageSetLSN(page, lsn); + MarkBufferDirty(buffer); + } + if (BufferIsValid(buffer)) UnlockReleaseBuffer(buffer); - return; - } - - if (record->xl_len > SizeOfBtreeDelete) - { - OffsetNumber *unused; - - unused = (OffsetNumber *) ((char *) xlrec + SizeOfBtreeDelete); - - PageIndexMultiDelete(page, unused, xlrec->nitems); - } - - /* - * Mark the page as not containing any LP_DEAD items --- see comments in - * _bt_delitems_delete(). - */ - opaque = (BTPageOpaque) PageGetSpecialPointer(page); - opaque->btpo_flags &= ~BTP_HAS_GARBAGE; - - PageSetLSN(page, lsn); - MarkBufferDirty(buffer); - UnlockReleaseBuffer(buffer); } static void @@ -816,42 +754,36 @@ btree_xlog_mark_page_halfdead(uint8 info, XLogRecPtr lsn, XLogRecord *record) */ /* parent page */ - if (record->xl_info & XLR_BKP_BLOCK(0)) - (void) RestoreBackupBlock(lsn, record, 0, false, false); - else + if (XLogReadBufferForRedo(lsn, record, 0, xlrec->target.node, parent, + &buffer) == BLK_NEEDS_REDO) { - buffer = XLogReadBuffer(xlrec->target.node, parent, false); - if (BufferIsValid(buffer)) - { - page = (Page) BufferGetPage(buffer); - pageop = (BTPageOpaque) PageGetSpecialPointer(page); - if (lsn > PageGetLSN(page)) - { - OffsetNumber poffset; - ItemId itemid; - IndexTuple itup; - OffsetNumber nextoffset; - BlockNumber rightsib; + OffsetNumber poffset; + ItemId itemid; + IndexTuple itup; + OffsetNumber nextoffset; + BlockNumber rightsib; - poffset = ItemPointerGetOffsetNumber(&(xlrec->target.tid)); + page = (Page) BufferGetPage(buffer); + pageop = (BTPageOpaque) PageGetSpecialPointer(page); - nextoffset = OffsetNumberNext(poffset); - itemid = PageGetItemId(page, nextoffset); - itup = (IndexTuple) PageGetItem(page, itemid); - rightsib = ItemPointerGetBlockNumber(&itup->t_tid); + poffset = ItemPointerGetOffsetNumber(&(xlrec->target.tid)); - itemid = PageGetItemId(page, poffset); - itup = (IndexTuple) PageGetItem(page, itemid); - ItemPointerSet(&(itup->t_tid), rightsib, P_HIKEY); - nextoffset = OffsetNumberNext(poffset); - PageIndexTupleDelete(page, nextoffset); + nextoffset = OffsetNumberNext(poffset); + itemid = PageGetItemId(page, nextoffset); + itup = (IndexTuple) PageGetItem(page, itemid); + rightsib = ItemPointerGetBlockNumber(&itup->t_tid); - PageSetLSN(page, lsn); - MarkBufferDirty(buffer); - } - UnlockReleaseBuffer(buffer); - } + itemid = PageGetItemId(page, poffset); + itup = (IndexTuple) PageGetItem(page, itemid); + ItemPointerSet(&(itup->t_tid), rightsib, P_HIKEY); + nextoffset = OffsetNumberNext(poffset); + PageIndexTupleDelete(page, nextoffset); + + PageSetLSN(page, lsn); + MarkBufferDirty(buffer); } + if (BufferIsValid(buffer)) + UnlockReleaseBuffer(buffer); /* Rewrite the leaf page as a halfdead page */ buffer = XLogReadBuffer(xlrec->target.node, xlrec->leafblk, true); @@ -911,56 +843,34 @@ btree_xlog_unlink_page(uint8 info, XLogRecPtr lsn, XLogRecord *record) */ /* Fix left-link of right sibling */ - if (record->xl_info & XLR_BKP_BLOCK(0)) - (void) RestoreBackupBlock(lsn, record, 0, false, false); - else + if (XLogReadBufferForRedo(lsn, record, 0, xlrec->node, rightsib, &buffer) + == BLK_NEEDS_REDO) { - buffer = XLogReadBuffer(xlrec->node, rightsib, false); - if (BufferIsValid(buffer)) - { - page = (Page) BufferGetPage(buffer); - if (lsn <= PageGetLSN(page)) - { - UnlockReleaseBuffer(buffer); - } - else - { - pageop = (BTPageOpaque) PageGetSpecialPointer(page); - pageop->btpo_prev = leftsib; + page = (Page) BufferGetPage(buffer); + pageop = (BTPageOpaque) PageGetSpecialPointer(page); + pageop->btpo_prev = leftsib; - PageSetLSN(page, lsn); - MarkBufferDirty(buffer); - UnlockReleaseBuffer(buffer); - } - } + PageSetLSN(page, lsn); + MarkBufferDirty(buffer); } + if (BufferIsValid(buffer)) + UnlockReleaseBuffer(buffer); /* Fix right-link of left sibling, if any */ - if (record->xl_info & XLR_BKP_BLOCK(1)) - (void) RestoreBackupBlock(lsn, record, 1, false, false); - else + if (leftsib != P_NONE) { - if (leftsib != P_NONE) + if (XLogReadBufferForRedo(lsn, record, 1, xlrec->node, leftsib, &buffer) + == BLK_NEEDS_REDO) { - buffer = XLogReadBuffer(xlrec->node, leftsib, false); - if (BufferIsValid(buffer)) - { - page = (Page) BufferGetPage(buffer); - if (lsn <= PageGetLSN(page)) - { - UnlockReleaseBuffer(buffer); - } - else - { - pageop = (BTPageOpaque) PageGetSpecialPointer(page); - pageop->btpo_next = rightsib; + page = (Page) BufferGetPage(buffer); + pageop = (BTPageOpaque) PageGetSpecialPointer(page); + pageop->btpo_next = rightsib; - PageSetLSN(page, lsn); - MarkBufferDirty(buffer); - UnlockReleaseBuffer(buffer); - } - } + PageSetLSN(page, lsn); + MarkBufferDirty(buffer); } + if (BufferIsValid(buffer)) + UnlockReleaseBuffer(buffer); } /* Rewrite target page as empty deleted page */ @@ -1071,10 +981,7 @@ btree_xlog_newroot(XLogRecPtr lsn, XLogRecord *record) Assert(ItemPointerGetOffsetNumber(&(itup->t_tid)) == P_HIKEY); /* Clear the incomplete-split flag in left child */ - if (record->xl_info & XLR_BKP_BLOCK(0)) - (void) RestoreBackupBlock(lsn, record, 0, false, false); - else - _bt_clear_incomplete_split(lsn, record, xlrec->node, cblkno); + _bt_clear_incomplete_split(lsn, record, 0, xlrec->node, cblkno); } PageSetLSN(page, lsn); diff --git a/src/backend/access/spgist/spgxlog.c b/src/backend/access/spgist/spgxlog.c index e4f7fbb53a..911a14828e 100644 --- a/src/backend/access/spgist/spgxlog.c +++ b/src/backend/access/spgist/spgxlog.c @@ -113,6 +113,7 @@ spgRedoAddLeaf(XLogRecPtr lsn, XLogRecord *record) SpGistLeafTupleData leafTupleHdr; Buffer buffer; Page page; + XLogRedoAction action; ptr += sizeof(spgxlogAddLeaf); leafTuple = ptr; @@ -124,82 +125,78 @@ spgRedoAddLeaf(XLogRecPtr lsn, XLogRecord *record) * simultaneously; but in WAL replay it should be safe to update the leaf * page before updating the parent. */ - if (record->xl_info & XLR_BKP_BLOCK(0)) - (void) RestoreBackupBlock(lsn, record, 0, false, false); - else + if (xldata->newPage) { - buffer = XLogReadBuffer(xldata->node, xldata->blknoLeaf, - xldata->newPage); - if (BufferIsValid(buffer)) - { - page = BufferGetPage(buffer); - - if (xldata->newPage) - SpGistInitBuffer(buffer, + buffer = XLogReadBuffer(xldata->node, xldata->blknoLeaf, true); + SpGistInitBuffer(buffer, SPGIST_LEAF | (xldata->storesNulls ? SPGIST_NULLS : 0)); - - if (lsn > PageGetLSN(page)) - { - /* insert new tuple */ - if (xldata->offnumLeaf != xldata->offnumHeadLeaf) - { - /* normal cases, tuple was added by SpGistPageAddNewItem */ - addOrReplaceTuple(page, (Item) leafTuple, leafTupleHdr.size, - xldata->offnumLeaf); - - /* update head tuple's chain link if needed */ - if (xldata->offnumHeadLeaf != InvalidOffsetNumber) - { - SpGistLeafTuple head; - - head = (SpGistLeafTuple) PageGetItem(page, - PageGetItemId(page, xldata->offnumHeadLeaf)); - Assert(head->nextOffset == leafTupleHdr.nextOffset); - head->nextOffset = xldata->offnumLeaf; - } - } - else - { - /* replacing a DEAD tuple */ - PageIndexTupleDelete(page, xldata->offnumLeaf); - if (PageAddItem(page, - (Item) leafTuple, leafTupleHdr.size, - xldata->offnumLeaf, false, false) != xldata->offnumLeaf) - elog(ERROR, "failed to add item of size %u to SPGiST index page", - leafTupleHdr.size); - } - - PageSetLSN(page, lsn); - MarkBufferDirty(buffer); - } - UnlockReleaseBuffer(buffer); - } + action = BLK_NEEDS_REDO; } + else + action = XLogReadBufferForRedo(lsn, record, 0, + xldata->node, xldata->blknoLeaf, + &buffer); + + if (action == BLK_NEEDS_REDO) + { + page = BufferGetPage(buffer); + + /* insert new tuple */ + if (xldata->offnumLeaf != xldata->offnumHeadLeaf) + { + /* normal cases, tuple was added by SpGistPageAddNewItem */ + addOrReplaceTuple(page, (Item) leafTuple, leafTupleHdr.size, + xldata->offnumLeaf); + + /* update head tuple's chain link if needed */ + if (xldata->offnumHeadLeaf != InvalidOffsetNumber) + { + SpGistLeafTuple head; + + head = (SpGistLeafTuple) PageGetItem(page, + PageGetItemId(page, xldata->offnumHeadLeaf)); + Assert(head->nextOffset == leafTupleHdr.nextOffset); + head->nextOffset = xldata->offnumLeaf; + } + } + else + { + /* replacing a DEAD tuple */ + PageIndexTupleDelete(page, xldata->offnumLeaf); + if (PageAddItem(page, (Item) leafTuple, leafTupleHdr.size, + xldata->offnumLeaf, false, false) != xldata->offnumLeaf) + elog(ERROR, "failed to add item of size %u to SPGiST index page", + leafTupleHdr.size); + } + + PageSetLSN(page, lsn); + MarkBufferDirty(buffer); + } + if (BufferIsValid(buffer)) + UnlockReleaseBuffer(buffer); /* update parent downlink if necessary */ - if (record->xl_info & XLR_BKP_BLOCK(1)) - (void) RestoreBackupBlock(lsn, record, 1, false, false); - else if (xldata->blknoParent != InvalidBlockNumber) + if (xldata->blknoParent != InvalidBlockNumber) { - buffer = XLogReadBuffer(xldata->node, xldata->blknoParent, false); - if (BufferIsValid(buffer)) + if (XLogReadBufferForRedo(lsn, record, 1, + xldata->node, xldata->blknoParent, + &buffer) == BLK_NEEDS_REDO) { - page = BufferGetPage(buffer); - if (lsn > PageGetLSN(page)) - { - SpGistInnerTuple tuple; + SpGistInnerTuple tuple; - tuple = (SpGistInnerTuple) PageGetItem(page, + page = BufferGetPage(buffer); + + tuple = (SpGistInnerTuple) PageGetItem(page, PageGetItemId(page, xldata->offnumParent)); - spgUpdateNodeLink(tuple, xldata->nodeI, - xldata->blknoLeaf, xldata->offnumLeaf); + spgUpdateNodeLink(tuple, xldata->nodeI, + xldata->blknoLeaf, xldata->offnumLeaf); - PageSetLSN(page, lsn); - MarkBufferDirty(buffer); - } - UnlockReleaseBuffer(buffer); + PageSetLSN(page, lsn); + MarkBufferDirty(buffer); } + if (BufferIsValid(buffer)) + UnlockReleaseBuffer(buffer); } } @@ -214,6 +211,7 @@ spgRedoMoveLeafs(XLogRecPtr lsn, XLogRecord *record) int nInsert; Buffer buffer; Page page; + XLogRedoAction action; fillFakeState(&state, xldata->stateSrc); @@ -234,98 +232,82 @@ spgRedoMoveLeafs(XLogRecPtr lsn, XLogRecord *record) */ /* Insert tuples on the dest page (do first, so redirect is valid) */ - if (record->xl_info & XLR_BKP_BLOCK(1)) - (void) RestoreBackupBlock(lsn, record, 1, false, false); - else + if (xldata->newPage) { - buffer = XLogReadBuffer(xldata->node, xldata->blknoDst, - xldata->newPage); - if (BufferIsValid(buffer)) - { - page = BufferGetPage(buffer); - - if (xldata->newPage) - SpGistInitBuffer(buffer, + buffer = XLogReadBuffer(xldata->node, xldata->blknoDst, true); + SpGistInitBuffer(buffer, SPGIST_LEAF | (xldata->storesNulls ? SPGIST_NULLS : 0)); - - if (lsn > PageGetLSN(page)) - { - int i; - - for (i = 0; i < nInsert; i++) - { - char *leafTuple; - SpGistLeafTupleData leafTupleHdr; - - /* - * the tuples are not aligned, so must copy to access - * the size field. - */ - leafTuple = ptr; - memcpy(&leafTupleHdr, leafTuple, - sizeof(SpGistLeafTupleData)); - - addOrReplaceTuple(page, (Item) leafTuple, - leafTupleHdr.size, toInsert[i]); - ptr += leafTupleHdr.size; - } - - PageSetLSN(page, lsn); - MarkBufferDirty(buffer); - } - UnlockReleaseBuffer(buffer); - } + action = BLK_NEEDS_REDO; } + else + action = XLogReadBufferForRedo(lsn, record, 1, + xldata->node, xldata->blknoDst, + &buffer); + if (action == BLK_NEEDS_REDO) + { + int i; + + page = BufferGetPage(buffer); + + for (i = 0; i < nInsert; i++) + { + char *leafTuple; + SpGistLeafTupleData leafTupleHdr; + + /* + * the tuples are not aligned, so must copy to access the size + * field. + */ + leafTuple = ptr; + memcpy(&leafTupleHdr, leafTuple, sizeof(SpGistLeafTupleData)); + + addOrReplaceTuple(page, (Item) leafTuple, + leafTupleHdr.size, toInsert[i]); + ptr += leafTupleHdr.size; + } + + PageSetLSN(page, lsn); + MarkBufferDirty(buffer); + } + if (BufferIsValid(buffer)) + UnlockReleaseBuffer(buffer); /* Delete tuples from the source page, inserting a redirection pointer */ - if (record->xl_info & XLR_BKP_BLOCK(0)) - (void) RestoreBackupBlock(lsn, record, 0, false, false); - else + if (XLogReadBufferForRedo(lsn, record, 0, xldata->node, xldata->blknoSrc, + &buffer) == BLK_NEEDS_REDO) { - buffer = XLogReadBuffer(xldata->node, xldata->blknoSrc, false); - if (BufferIsValid(buffer)) - { - page = BufferGetPage(buffer); - if (lsn > PageGetLSN(page)) - { - spgPageIndexMultiDelete(&state, page, toDelete, xldata->nMoves, + page = BufferGetPage(buffer); + spgPageIndexMultiDelete(&state, page, toDelete, xldata->nMoves, state.isBuild ? SPGIST_PLACEHOLDER : SPGIST_REDIRECT, - SPGIST_PLACEHOLDER, - xldata->blknoDst, - toInsert[nInsert - 1]); + SPGIST_PLACEHOLDER, + xldata->blknoDst, + toInsert[nInsert - 1]); - PageSetLSN(page, lsn); - MarkBufferDirty(buffer); - } - UnlockReleaseBuffer(buffer); - } + PageSetLSN(page, lsn); + MarkBufferDirty(buffer); } + if (BufferIsValid(buffer)) + UnlockReleaseBuffer(buffer); /* And update the parent downlink */ - if (record->xl_info & XLR_BKP_BLOCK(2)) - (void) RestoreBackupBlock(lsn, record, 2, false, false); - else + if (XLogReadBufferForRedo(lsn, record, 2, xldata->node, xldata->blknoParent, + &buffer) == BLK_NEEDS_REDO) { - buffer = XLogReadBuffer(xldata->node, xldata->blknoParent, false); - if (BufferIsValid(buffer)) - { - page = BufferGetPage(buffer); - if (lsn > PageGetLSN(page)) - { - SpGistInnerTuple tuple; + SpGistInnerTuple tuple; - tuple = (SpGistInnerTuple) PageGetItem(page, + page = BufferGetPage(buffer); + + tuple = (SpGistInnerTuple) PageGetItem(page, PageGetItemId(page, xldata->offnumParent)); - spgUpdateNodeLink(tuple, xldata->nodeI, - xldata->blknoDst, toInsert[nInsert - 1]); + spgUpdateNodeLink(tuple, xldata->nodeI, + xldata->blknoDst, toInsert[nInsert - 1]); - PageSetLSN(page, lsn); - MarkBufferDirty(buffer); - } - UnlockReleaseBuffer(buffer); - } + PageSetLSN(page, lsn); + MarkBufferDirty(buffer); } + if (BufferIsValid(buffer)) + UnlockReleaseBuffer(buffer); } static void @@ -339,6 +321,7 @@ spgRedoAddNode(XLogRecPtr lsn, XLogRecord *record) Buffer buffer; Page page; int bbi; + XLogRedoAction action; ptr += sizeof(spgxlogAddNode); innerTuple = ptr; @@ -351,29 +334,21 @@ spgRedoAddNode(XLogRecPtr lsn, XLogRecord *record) { /* update in place */ Assert(xldata->blknoParent == InvalidBlockNumber); - if (record->xl_info & XLR_BKP_BLOCK(0)) - (void) RestoreBackupBlock(lsn, record, 0, false, false); - else + if (XLogReadBufferForRedo(lsn, record, 0, xldata->node, xldata->blkno, + &buffer) == BLK_NEEDS_REDO) { - buffer = XLogReadBuffer(xldata->node, xldata->blkno, false); - if (BufferIsValid(buffer)) - { - page = BufferGetPage(buffer); - if (lsn > PageGetLSN(page)) - { - PageIndexTupleDelete(page, xldata->offnum); - if (PageAddItem(page, (Item) innerTuple, innerTupleHdr.size, - xldata->offnum, - false, false) != xldata->offnum) - elog(ERROR, "failed to add item of size %u to SPGiST index page", - innerTupleHdr.size); + page = BufferGetPage(buffer); + PageIndexTupleDelete(page, xldata->offnum); + if (PageAddItem(page, (Item) innerTuple, innerTupleHdr.size, + xldata->offnum, false, false) != xldata->offnum) + elog(ERROR, "failed to add item of size %u to SPGiST index page", + innerTupleHdr.size); - PageSetLSN(page, lsn); - MarkBufferDirty(buffer); - } - UnlockReleaseBuffer(buffer); - } + PageSetLSN(page, lsn); + MarkBufferDirty(buffer); } + if (BufferIsValid(buffer)) + UnlockReleaseBuffer(buffer); } else { @@ -390,90 +365,79 @@ spgRedoAddNode(XLogRecPtr lsn, XLogRecord *record) Assert(xldata->blkno != xldata->blknoNew); /* Install new tuple first so redirect is valid */ - if (record->xl_info & XLR_BKP_BLOCK(1)) - (void) RestoreBackupBlock(lsn, record, 1, false, false); - else + if (xldata->newPage) { - buffer = XLogReadBuffer(xldata->node, xldata->blknoNew, - xldata->newPage); - if (BufferIsValid(buffer)) - { - page = BufferGetPage(buffer); - - /* AddNode is not used for nulls pages */ - if (xldata->newPage) - SpGistInitBuffer(buffer, 0); - - if (lsn > PageGetLSN(page)) - { - addOrReplaceTuple(page, (Item) innerTuple, - innerTupleHdr.size, xldata->offnumNew); - - /* - * If parent is in this same page, don't advance LSN; - * doing so would fool us into not applying the parent - * downlink update below. We'll update the LSN when we - * fix the parent downlink. - */ - if (xldata->blknoParent != xldata->blknoNew) - { - PageSetLSN(page, lsn); - } - MarkBufferDirty(buffer); - } - UnlockReleaseBuffer(buffer); - } + buffer = XLogReadBuffer(xldata->node, xldata->blknoNew, true); + /* AddNode is not used for nulls pages */ + SpGistInitBuffer(buffer, 0); + action = BLK_NEEDS_REDO; } + else + action = XLogReadBufferForRedo(lsn, record, 1, + xldata->node, xldata->blknoNew, + &buffer); + if (action == BLK_NEEDS_REDO) + { + page = BufferGetPage(buffer); + + addOrReplaceTuple(page, (Item) innerTuple, + innerTupleHdr.size, xldata->offnumNew); + + /* + * If parent is in this same page, don't advance LSN; doing so + * would fool us into not applying the parent downlink update + * below. We'll update the LSN when we fix the parent downlink. + */ + if (xldata->blknoParent != xldata->blknoNew) + { + PageSetLSN(page, lsn); + } + MarkBufferDirty(buffer); + } + if (BufferIsValid(buffer)) + UnlockReleaseBuffer(buffer); /* Delete old tuple, replacing it with redirect or placeholder tuple */ - if (record->xl_info & XLR_BKP_BLOCK(0)) - (void) RestoreBackupBlock(lsn, record, 0, false, false); - else + if (XLogReadBufferForRedo(lsn, record, 0, xldata->node, xldata->blkno, + &buffer) == BLK_NEEDS_REDO) { - buffer = XLogReadBuffer(xldata->node, xldata->blkno, false); - if (BufferIsValid(buffer)) + SpGistDeadTuple dt; + + page = BufferGetPage(buffer); + + if (state.isBuild) + dt = spgFormDeadTuple(&state, SPGIST_PLACEHOLDER, + InvalidBlockNumber, + InvalidOffsetNumber); + else + dt = spgFormDeadTuple(&state, SPGIST_REDIRECT, + xldata->blknoNew, + xldata->offnumNew); + + PageIndexTupleDelete(page, xldata->offnum); + if (PageAddItem(page, (Item) dt, dt->size, xldata->offnum, + false, false) != xldata->offnum) + elog(ERROR, "failed to add item of size %u to SPGiST index page", + dt->size); + + if (state.isBuild) + SpGistPageGetOpaque(page)->nPlaceholder++; + else + SpGistPageGetOpaque(page)->nRedirection++; + + /* + * If parent is in this same page, don't advance LSN; doing so + * would fool us into not applying the parent downlink update + * below. We'll update the LSN when we fix the parent downlink. + */ + if (xldata->blknoParent != xldata->blkno) { - page = BufferGetPage(buffer); - if (lsn > PageGetLSN(page)) - { - SpGistDeadTuple dt; - - if (state.isBuild) - dt = spgFormDeadTuple(&state, SPGIST_PLACEHOLDER, - InvalidBlockNumber, - InvalidOffsetNumber); - else - dt = spgFormDeadTuple(&state, SPGIST_REDIRECT, - xldata->blknoNew, - xldata->offnumNew); - - PageIndexTupleDelete(page, xldata->offnum); - if (PageAddItem(page, (Item) dt, dt->size, - xldata->offnum, - false, false) != xldata->offnum) - elog(ERROR, "failed to add item of size %u to SPGiST index page", - dt->size); - - if (state.isBuild) - SpGistPageGetOpaque(page)->nPlaceholder++; - else - SpGistPageGetOpaque(page)->nRedirection++; - - /* - * If parent is in this same page, don't advance LSN; - * doing so would fool us into not applying the parent - * downlink update below. We'll update the LSN when we - * fix the parent downlink. - */ - if (xldata->blknoParent != xldata->blkno) - { - PageSetLSN(page, lsn); - } - MarkBufferDirty(buffer); - } - UnlockReleaseBuffer(buffer); + PageSetLSN(page, lsn); } + MarkBufferDirty(buffer); } + if (BufferIsValid(buffer)) + UnlockReleaseBuffer(buffer); /* * Update parent downlink. Since parent could be in either of the @@ -491,29 +455,32 @@ spgRedoAddNode(XLogRecPtr lsn, XLogRecord *record) { if (bbi == 2) /* else we already did it */ (void) RestoreBackupBlock(lsn, record, bbi, false, false); + action = BLK_RESTORED; + buffer = InvalidBuffer; } else { - buffer = XLogReadBuffer(xldata->node, xldata->blknoParent, false); - if (BufferIsValid(buffer)) - { - page = BufferGetPage(buffer); - if (lsn > PageGetLSN(page)) - { - SpGistInnerTuple innerTuple; + action = XLogReadBufferForRedo(lsn, record, bbi, xldata->node, + xldata->blknoParent, &buffer); + Assert(action != BLK_RESTORED); + } + if (action == BLK_NEEDS_REDO) + { + SpGistInnerTuple innerTuple; - innerTuple = (SpGistInnerTuple) PageGetItem(page, + page = BufferGetPage(buffer); + + innerTuple = (SpGistInnerTuple) PageGetItem(page, PageGetItemId(page, xldata->offnumParent)); - spgUpdateNodeLink(innerTuple, xldata->nodeI, - xldata->blknoNew, xldata->offnumNew); + spgUpdateNodeLink(innerTuple, xldata->nodeI, + xldata->blknoNew, xldata->offnumNew); - PageSetLSN(page, lsn); - MarkBufferDirty(buffer); - } - UnlockReleaseBuffer(buffer); - } + PageSetLSN(page, lsn); + MarkBufferDirty(buffer); } + if (BufferIsValid(buffer)) + UnlockReleaseBuffer(buffer); } } @@ -545,60 +512,56 @@ spgRedoSplitTuple(XLogRecPtr lsn, XLogRecord *record) */ /* insert postfix tuple first to avoid dangling link */ - if (record->xl_info & XLR_BKP_BLOCK(1)) - (void) RestoreBackupBlock(lsn, record, 1, false, false); - else if (xldata->blknoPostfix != xldata->blknoPrefix) + if (xldata->blknoPostfix != xldata->blknoPrefix) { - buffer = XLogReadBuffer(xldata->node, xldata->blknoPostfix, - xldata->newPage); - if (BufferIsValid(buffer)) + XLogRedoAction action; + + if (xldata->newPage) + { + buffer = XLogReadBuffer(xldata->node, xldata->blknoPostfix, true); + /* SplitTuple is not used for nulls pages */ + SpGistInitBuffer(buffer, 0); + action = BLK_NEEDS_REDO; + } + else + action = XLogReadBufferForRedo(lsn, record, 1, + xldata->node, xldata->blknoPostfix, + &buffer); + + if (action == BLK_NEEDS_REDO) { page = BufferGetPage(buffer); - /* SplitTuple is not used for nulls pages */ - if (xldata->newPage) - SpGistInitBuffer(buffer, 0); + addOrReplaceTuple(page, (Item) postfixTuple, + postfixTupleHdr.size, xldata->offnumPostfix); - if (lsn > PageGetLSN(page)) - { - addOrReplaceTuple(page, (Item) postfixTuple, - postfixTupleHdr.size, xldata->offnumPostfix); - - PageSetLSN(page, lsn); - MarkBufferDirty(buffer); - } - UnlockReleaseBuffer(buffer); + PageSetLSN(page, lsn); + MarkBufferDirty(buffer); } + if (BufferIsValid(buffer)) + UnlockReleaseBuffer(buffer); } /* now handle the original page */ - if (record->xl_info & XLR_BKP_BLOCK(0)) - (void) RestoreBackupBlock(lsn, record, 0, false, false); - else + if (XLogReadBufferForRedo(lsn, record, 0, xldata->node, xldata->blknoPrefix, + &buffer) == BLK_NEEDS_REDO) { - buffer = XLogReadBuffer(xldata->node, xldata->blknoPrefix, false); - if (BufferIsValid(buffer)) - { - page = BufferGetPage(buffer); - if (lsn > PageGetLSN(page)) - { - PageIndexTupleDelete(page, xldata->offnumPrefix); - if (PageAddItem(page, (Item) prefixTuple, prefixTupleHdr.size, + page = BufferGetPage(buffer); + PageIndexTupleDelete(page, xldata->offnumPrefix); + if (PageAddItem(page, (Item) prefixTuple, prefixTupleHdr.size, xldata->offnumPrefix, false, false) != xldata->offnumPrefix) - elog(ERROR, "failed to add item of size %u to SPGiST index page", - prefixTupleHdr.size); + elog(ERROR, "failed to add item of size %u to SPGiST index page", + prefixTupleHdr.size); - if (xldata->blknoPostfix == xldata->blknoPrefix) - addOrReplaceTuple(page, (Item) postfixTuple, - postfixTupleHdr.size, - xldata->offnumPostfix); + if (xldata->blknoPostfix == xldata->blknoPrefix) + addOrReplaceTuple(page, (Item) postfixTuple, postfixTupleHdr.size, + xldata->offnumPostfix); - PageSetLSN(page, lsn); - MarkBufferDirty(buffer); - } - UnlockReleaseBuffer(buffer); - } + PageSetLSN(page, lsn); + MarkBufferDirty(buffer); } + if (BufferIsValid(buffer)) + UnlockReleaseBuffer(buffer); } static void @@ -616,9 +579,11 @@ spgRedoPickSplit(XLogRecPtr lsn, XLogRecord *record) Buffer destBuffer; Page srcPage; Page destPage; + Buffer innerBuffer; Page page; int bbi; int i; + XLogRedoAction action; fillFakeState(&state, xldata->stateSrc); @@ -668,46 +633,37 @@ spgRedoPickSplit(XLogRecPtr lsn, XLogRecord *record) * inserting leaf tuples and the new inner tuple, else the added * redirect tuple will be a dangling link.) */ - if (record->xl_info & XLR_BKP_BLOCK(bbi)) + if (XLogReadBufferForRedo(lsn, record, bbi, + xldata->node, xldata->blknoSrc, + &srcBuffer) == BLK_NEEDS_REDO) { - srcBuffer = RestoreBackupBlock(lsn, record, bbi, false, true); - srcPage = NULL; /* don't need to do any page updates */ + srcPage = BufferGetPage(srcBuffer); + + /* + * We have it a bit easier here than in doPickSplit(), because we + * know the inner tuple's location already, so we can inject the + * correct redirection tuple now. + */ + if (!state.isBuild) + spgPageIndexMultiDelete(&state, srcPage, + toDelete, xldata->nDelete, + SPGIST_REDIRECT, + SPGIST_PLACEHOLDER, + xldata->blknoInner, + xldata->offnumInner); + else + spgPageIndexMultiDelete(&state, srcPage, + toDelete, xldata->nDelete, + SPGIST_PLACEHOLDER, + SPGIST_PLACEHOLDER, + InvalidBlockNumber, + InvalidOffsetNumber); + + /* don't update LSN etc till we're done with it */ } else { - srcBuffer = XLogReadBuffer(xldata->node, xldata->blknoSrc, false); - if (BufferIsValid(srcBuffer)) - { - srcPage = BufferGetPage(srcBuffer); - if (lsn > PageGetLSN(srcPage)) - { - /* - * We have it a bit easier here than in doPickSplit(), - * because we know the inner tuple's location already, so - * we can inject the correct redirection tuple now. - */ - if (!state.isBuild) - spgPageIndexMultiDelete(&state, srcPage, - toDelete, xldata->nDelete, - SPGIST_REDIRECT, - SPGIST_PLACEHOLDER, - xldata->blknoInner, - xldata->offnumInner); - else - spgPageIndexMultiDelete(&state, srcPage, - toDelete, xldata->nDelete, - SPGIST_PLACEHOLDER, - SPGIST_PLACEHOLDER, - InvalidBlockNumber, - InvalidOffsetNumber); - - /* don't update LSN etc till we're done with it */ - } - else - srcPage = NULL; /* don't do any page updates */ - } - else - srcPage = NULL; + srcPage = NULL; /* don't do any page updates */ } bbi++; } @@ -735,22 +691,15 @@ spgRedoPickSplit(XLogRecPtr lsn, XLogRecord *record) * We could probably release the page lock immediately in the * full-page-image case, but for safety let's hold it till later. */ - if (record->xl_info & XLR_BKP_BLOCK(bbi)) + if (XLogReadBufferForRedo(lsn, record, bbi, + xldata->node, xldata->blknoDest, + &destBuffer) == BLK_NEEDS_REDO) { - destBuffer = RestoreBackupBlock(lsn, record, bbi, false, true); - destPage = NULL; /* don't need to do any page updates */ + destPage = (Page) BufferGetPage(destBuffer); } else { - destBuffer = XLogReadBuffer(xldata->node, xldata->blknoDest, false); - if (BufferIsValid(destBuffer)) - { - destPage = (Page) BufferGetPage(destBuffer); - if (lsn <= PageGetLSN(destPage)) - destPage = NULL; /* don't do any page updates */ - } - else - destPage = NULL; + destPage = NULL; /* don't do any page updates */ } bbi++; } @@ -787,43 +736,40 @@ spgRedoPickSplit(XLogRecPtr lsn, XLogRecord *record) } /* restore new inner tuple */ - if (record->xl_info & XLR_BKP_BLOCK(bbi)) - (void) RestoreBackupBlock(lsn, record, bbi, false, false); - else + if (xldata->initInner) { - Buffer buffer = XLogReadBuffer(xldata->node, xldata->blknoInner, - xldata->initInner); - - if (BufferIsValid(buffer)) - { - page = BufferGetPage(buffer); - - if (xldata->initInner) - SpGistInitBuffer(buffer, - (xldata->storesNulls ? SPGIST_NULLS : 0)); - - if (lsn > PageGetLSN(page)) - { - addOrReplaceTuple(page, (Item) innerTuple, innerTupleHdr.size, - xldata->offnumInner); - - /* if inner is also parent, update link while we're here */ - if (xldata->blknoInner == xldata->blknoParent) - { - SpGistInnerTuple parent; - - parent = (SpGistInnerTuple) PageGetItem(page, - PageGetItemId(page, xldata->offnumParent)); - spgUpdateNodeLink(parent, xldata->nodeI, - xldata->blknoInner, xldata->offnumInner); - } - - PageSetLSN(page, lsn); - MarkBufferDirty(buffer); - } - UnlockReleaseBuffer(buffer); - } + innerBuffer = XLogReadBuffer(xldata->node, xldata->blknoInner, true); + SpGistInitBuffer(innerBuffer, + (xldata->storesNulls ? SPGIST_NULLS : 0)); + action = BLK_NEEDS_REDO; } + else + action = XLogReadBufferForRedo(lsn, record, bbi, xldata->node, + xldata->blknoInner, &innerBuffer); + + if (action == BLK_NEEDS_REDO) + { + page = BufferGetPage(innerBuffer); + + addOrReplaceTuple(page, (Item) innerTuple, innerTupleHdr.size, + xldata->offnumInner); + + /* if inner is also parent, update link while we're here */ + if (xldata->blknoInner == xldata->blknoParent) + { + SpGistInnerTuple parent; + + parent = (SpGistInnerTuple) PageGetItem(page, + PageGetItemId(page, xldata->offnumParent)); + spgUpdateNodeLink(parent, xldata->nodeI, + xldata->blknoInner, xldata->offnumInner); + } + + PageSetLSN(page, lsn); + MarkBufferDirty(innerBuffer); + } + if (BufferIsValid(innerBuffer)) + UnlockReleaseBuffer(innerBuffer); bbi++; /* @@ -843,31 +789,26 @@ spgRedoPickSplit(XLogRecPtr lsn, XLogRecord *record) } else if (xldata->blknoInner != xldata->blknoParent) { - if (record->xl_info & XLR_BKP_BLOCK(bbi)) - (void) RestoreBackupBlock(lsn, record, bbi, false, false); - else + Buffer parentBuffer; + + if (XLogReadBufferForRedo(lsn, record, bbi, + xldata->node, xldata->blknoParent, + &parentBuffer) == BLK_NEEDS_REDO) { - Buffer buffer = XLogReadBuffer(xldata->node, xldata->blknoParent, false); + SpGistInnerTuple parent; - if (BufferIsValid(buffer)) - { - page = BufferGetPage(buffer); + page = BufferGetPage(parentBuffer); - if (lsn > PageGetLSN(page)) - { - SpGistInnerTuple parent; - - parent = (SpGistInnerTuple) PageGetItem(page, + parent = (SpGistInnerTuple) PageGetItem(page, PageGetItemId(page, xldata->offnumParent)); - spgUpdateNodeLink(parent, xldata->nodeI, - xldata->blknoInner, xldata->offnumInner); + spgUpdateNodeLink(parent, xldata->nodeI, + xldata->blknoInner, xldata->offnumInner); - PageSetLSN(page, lsn); - MarkBufferDirty(buffer); - } - UnlockReleaseBuffer(buffer); - } + PageSetLSN(page, lsn); + MarkBufferDirty(parentBuffer); } + if (BufferIsValid(parentBuffer)) + UnlockReleaseBuffer(parentBuffer); } } @@ -902,62 +843,56 @@ spgRedoVacuumLeaf(XLogRecPtr lsn, XLogRecord *record) ptr += sizeof(OffsetNumber) * xldata->nChain; chainDest = (OffsetNumber *) ptr; - if (record->xl_info & XLR_BKP_BLOCK(0)) - (void) RestoreBackupBlock(lsn, record, 0, false, false); - else + if (XLogReadBufferForRedo(lsn, record, 0, xldata->node, xldata->blkno, + &buffer) == BLK_NEEDS_REDO) { - buffer = XLogReadBuffer(xldata->node, xldata->blkno, false); - if (BufferIsValid(buffer)) + page = BufferGetPage(buffer); + + spgPageIndexMultiDelete(&state, page, + toDead, xldata->nDead, + SPGIST_DEAD, SPGIST_DEAD, + InvalidBlockNumber, + InvalidOffsetNumber); + + spgPageIndexMultiDelete(&state, page, + toPlaceholder, xldata->nPlaceholder, + SPGIST_PLACEHOLDER, SPGIST_PLACEHOLDER, + InvalidBlockNumber, + InvalidOffsetNumber); + + /* see comments in vacuumLeafPage() */ + for (i = 0; i < xldata->nMove; i++) { - page = BufferGetPage(buffer); - if (lsn > PageGetLSN(page)) - { - spgPageIndexMultiDelete(&state, page, - toDead, xldata->nDead, - SPGIST_DEAD, SPGIST_DEAD, - InvalidBlockNumber, - InvalidOffsetNumber); + ItemId idSrc = PageGetItemId(page, moveSrc[i]); + ItemId idDest = PageGetItemId(page, moveDest[i]); + ItemIdData tmp; - spgPageIndexMultiDelete(&state, page, - toPlaceholder, xldata->nPlaceholder, - SPGIST_PLACEHOLDER, SPGIST_PLACEHOLDER, - InvalidBlockNumber, - InvalidOffsetNumber); - - /* see comments in vacuumLeafPage() */ - for (i = 0; i < xldata->nMove; i++) - { - ItemId idSrc = PageGetItemId(page, moveSrc[i]); - ItemId idDest = PageGetItemId(page, moveDest[i]); - ItemIdData tmp; - - tmp = *idSrc; - *idSrc = *idDest; - *idDest = tmp; - } - - spgPageIndexMultiDelete(&state, page, - moveSrc, xldata->nMove, - SPGIST_PLACEHOLDER, SPGIST_PLACEHOLDER, - InvalidBlockNumber, - InvalidOffsetNumber); - - for (i = 0; i < xldata->nChain; i++) - { - SpGistLeafTuple lt; - - lt = (SpGistLeafTuple) PageGetItem(page, - PageGetItemId(page, chainSrc[i])); - Assert(lt->tupstate == SPGIST_LIVE); - lt->nextOffset = chainDest[i]; - } - - PageSetLSN(page, lsn); - MarkBufferDirty(buffer); - } - UnlockReleaseBuffer(buffer); + tmp = *idSrc; + *idSrc = *idDest; + *idDest = tmp; } + + spgPageIndexMultiDelete(&state, page, + moveSrc, xldata->nMove, + SPGIST_PLACEHOLDER, SPGIST_PLACEHOLDER, + InvalidBlockNumber, + InvalidOffsetNumber); + + for (i = 0; i < xldata->nChain; i++) + { + SpGistLeafTuple lt; + + lt = (SpGistLeafTuple) PageGetItem(page, + PageGetItemId(page, chainSrc[i])); + Assert(lt->tupstate == SPGIST_LIVE); + lt->nextOffset = chainDest[i]; + } + + PageSetLSN(page, lsn); + MarkBufferDirty(buffer); } + if (BufferIsValid(buffer)) + UnlockReleaseBuffer(buffer); } static void @@ -971,25 +906,19 @@ spgRedoVacuumRoot(XLogRecPtr lsn, XLogRecord *record) toDelete = xldata->offsets; - if (record->xl_info & XLR_BKP_BLOCK(0)) - (void) RestoreBackupBlock(lsn, record, 0, false, false); - else + if (XLogReadBufferForRedo(lsn, record, 0, xldata->node, xldata->blkno, + &buffer) == BLK_NEEDS_REDO) { - buffer = XLogReadBuffer(xldata->node, xldata->blkno, false); - if (BufferIsValid(buffer)) - { - page = BufferGetPage(buffer); - if (lsn > PageGetLSN(page)) - { - /* The tuple numbers are in order */ - PageIndexMultiDelete(page, toDelete, xldata->nDelete); + page = BufferGetPage(buffer); - PageSetLSN(page, lsn); - MarkBufferDirty(buffer); - } - UnlockReleaseBuffer(buffer); - } + /* The tuple numbers are in order */ + PageIndexMultiDelete(page, toDelete, xldata->nDelete); + + PageSetLSN(page, lsn); + MarkBufferDirty(buffer); } + if (BufferIsValid(buffer)) + UnlockReleaseBuffer(buffer); } static void @@ -999,7 +928,6 @@ spgRedoVacuumRedirect(XLogRecPtr lsn, XLogRecord *record) spgxlogVacuumRedirect *xldata = (spgxlogVacuumRedirect *) ptr; OffsetNumber *itemToPlaceholder; Buffer buffer; - Page page; itemToPlaceholder = xldata->offsets; @@ -1014,64 +942,55 @@ spgRedoVacuumRedirect(XLogRecPtr lsn, XLogRecord *record) xldata->node); } - if (record->xl_info & XLR_BKP_BLOCK(0)) - (void) RestoreBackupBlock(lsn, record, 0, false, false); - else + if (XLogReadBufferForRedo(lsn, record, 0, xldata->node, xldata->blkno, + &buffer) == BLK_NEEDS_REDO) { - buffer = XLogReadBuffer(xldata->node, xldata->blkno, false); + Page page = BufferGetPage(buffer); + SpGistPageOpaque opaque = SpGistPageGetOpaque(page); + int i; - if (BufferIsValid(buffer)) + /* Convert redirect pointers to plain placeholders */ + for (i = 0; i < xldata->nToPlaceholder; i++) { - page = BufferGetPage(buffer); - if (lsn > PageGetLSN(page)) - { - SpGistPageOpaque opaque = SpGistPageGetOpaque(page); - int i; + SpGistDeadTuple dt; - /* Convert redirect pointers to plain placeholders */ - for (i = 0; i < xldata->nToPlaceholder; i++) - { - SpGistDeadTuple dt; - - dt = (SpGistDeadTuple) PageGetItem(page, + dt = (SpGistDeadTuple) PageGetItem(page, PageGetItemId(page, itemToPlaceholder[i])); - Assert(dt->tupstate == SPGIST_REDIRECT); - dt->tupstate = SPGIST_PLACEHOLDER; - ItemPointerSetInvalid(&dt->pointer); - } - - Assert(opaque->nRedirection >= xldata->nToPlaceholder); - opaque->nRedirection -= xldata->nToPlaceholder; - opaque->nPlaceholder += xldata->nToPlaceholder; - - /* Remove placeholder tuples at end of page */ - if (xldata->firstPlaceholder != InvalidOffsetNumber) - { - int max = PageGetMaxOffsetNumber(page); - OffsetNumber *toDelete; - - toDelete = palloc(sizeof(OffsetNumber) * max); - - for (i = xldata->firstPlaceholder; i <= max; i++) - toDelete[i - xldata->firstPlaceholder] = i; - - i = max - xldata->firstPlaceholder + 1; - Assert(opaque->nPlaceholder >= i); - opaque->nPlaceholder -= i; - - /* The array is sorted, so can use PageIndexMultiDelete */ - PageIndexMultiDelete(page, toDelete, i); - - pfree(toDelete); - } - - PageSetLSN(page, lsn); - MarkBufferDirty(buffer); - } - - UnlockReleaseBuffer(buffer); + Assert(dt->tupstate == SPGIST_REDIRECT); + dt->tupstate = SPGIST_PLACEHOLDER; + ItemPointerSetInvalid(&dt->pointer); } + + Assert(opaque->nRedirection >= xldata->nToPlaceholder); + opaque->nRedirection -= xldata->nToPlaceholder; + opaque->nPlaceholder += xldata->nToPlaceholder; + + /* Remove placeholder tuples at end of page */ + if (xldata->firstPlaceholder != InvalidOffsetNumber) + { + int max = PageGetMaxOffsetNumber(page); + OffsetNumber *toDelete; + + toDelete = palloc(sizeof(OffsetNumber) * max); + + for (i = xldata->firstPlaceholder; i <= max; i++) + toDelete[i - xldata->firstPlaceholder] = i; + + i = max - xldata->firstPlaceholder + 1; + Assert(opaque->nPlaceholder >= i); + opaque->nPlaceholder -= i; + + /* The array is sorted, so can use PageIndexMultiDelete */ + PageIndexMultiDelete(page, toDelete, i); + + pfree(toDelete); + } + + PageSetLSN(page, lsn); + MarkBufferDirty(buffer); } + if (BufferIsValid(buffer)) + UnlockReleaseBuffer(buffer); } void diff --git a/src/backend/access/transam/README b/src/backend/access/transam/README index f83526ccc3..92b12fbb6c 100644 --- a/src/backend/access/transam/README +++ b/src/backend/access/transam/README @@ -500,33 +500,28 @@ incrementally update the page, the rdata array *must* mention the buffer ID at least once; otherwise there is no defense against torn-page problems. The standard replay-routine pattern for this case is - if (record->xl_info & XLR_BKP_BLOCK(N)) + if (XLogReadBufferForRedo(lsn, record, N, rnode, blkno, &buffer) == BLK_NEEDS_REDO) { - /* apply the change from the full-page image */ - (void) RestoreBackupBlock(lsn, record, N, false, false); - return; - } + page = (Page) BufferGetPage(buffer); - buffer = XLogReadBuffer(rnode, blkno, false); - if (!BufferIsValid(buffer)) - { - /* page has been deleted, so we need do nothing */ - return; - } - page = (Page) BufferGetPage(buffer); + ... apply the change ... - if (XLByteLE(lsn, PageGetLSN(page))) - { - /* changes are already applied */ + PageSetLSN(page, lsn); + MarkBufferDirty(buffer); + } + if (BufferIsValid(buffer)) UnlockReleaseBuffer(buffer); - return; - } - ... apply the change ... - - PageSetLSN(page, lsn); - MarkBufferDirty(buffer); - UnlockReleaseBuffer(buffer); +XLogReadBufferForRedo reads the page from disk, and checks what action needs to +be taken to the page. If the XLR_BKP_BLOCK(N) flag is set, it restores the +full page image and returns BLK_RESTORED. If there is no full page image, but +page cannot be found or if the change has already been replayed (i.e. the +page's LSN >= the record we're replaying), it returns BLK_NOTFOUND or BLK_DONE, +respectively. Usually, the redo routine only needs to pay attention to the +BLK_NEEDS_REDO return code, which means that the routine should apply the +incremental change. In any case, the caller is responsible for unlocking and +releasing the buffer. Note that XLogReadBufferForRedo returns the buffer +locked even if no redo is required, unless the page does not exist. As noted above, for a multi-page update you need to be able to determine which XLR_BKP_BLOCK(N) flag applies to each page. If a WAL record reflects @@ -539,31 +534,8 @@ per the above discussion, fully-rewritable buffers shouldn't be mentioned in When replaying a WAL record that describes changes on multiple pages, you must be careful to lock the pages properly to prevent concurrent Hot Standby queries from seeing an inconsistent state. If this requires that two -or more buffer locks be held concurrently, the coding pattern shown above -is too simplistic, since it assumes the routine can exit as soon as it's -known the current page requires no modification. Instead, you might have -something like - - if (record->xl_info & XLR_BKP_BLOCK(0)) - { - /* apply the change from the full-page image */ - buffer0 = RestoreBackupBlock(lsn, record, 0, false, true); - } - else - { - buffer0 = XLogReadBuffer(rnode, blkno, false); - if (BufferIsValid(buffer0)) - { - ... apply the change if not already done ... - MarkBufferDirty(buffer0); - } - } - - ... similarly apply the changes for remaining pages ... - - /* and now we can release the lock on the first page */ - if (BufferIsValid(buffer0)) - UnlockReleaseBuffer(buffer0); +or more buffer locks be held concurrently, you must lock the pages in +appropriate order, and not release the locks until all the changes are done. Note that we must only use PageSetLSN/PageGetLSN() when we know the action is serialised. Only Startup process may modify data blocks during recovery, diff --git a/src/backend/access/transam/xlogutils.c b/src/backend/access/transam/xlogutils.c index b7829ff4c6..ef827dbc40 100644 --- a/src/backend/access/transam/xlogutils.c +++ b/src/backend/access/transam/xlogutils.c @@ -242,6 +242,87 @@ XLogCheckInvalidPages(void) invalid_page_tab = NULL; } + +/* + * XLogReadBufferForRedo + * Read a page during XLOG replay + * + * Reads a block referenced by a WAL record into shared buffer cache, and + * determines what needs to be done to redo the changes to it. If the WAL + * record includes a full-page image of the page, it is restored. + * + * 'lsn' is the LSN of the record being replayed. It is compared with the + * page's LSN to determine if the record has already been replayed. + * 'rnode' and 'blkno' point to the block being replayed (main fork number + * is implied, use XLogReadBufferForRedoExtended for other forks). + * 'block_index' identifies the backup block in the record for the page. + * + * Returns one of the following: + * + * BLK_NEEDS_REDO - changes from the WAL record need to be applied + * BLK_DONE - block doesn't need replaying + * BLK_RESTORED - block was restored from a full-page image included in + * the record + * BLK_NOTFOUND - block was not found (because it was truncated away by + * an operation later in the WAL stream) + * + * On return, the buffer is locked in exclusive-mode, and returned in *buf. + * Note that the buffer is locked and returned even if it doesn't need + * replaying. (Getting the buffer lock is not really necessary during + * single-process crash recovery, but some subroutines such as MarkBufferDirty + * will complain if we don't have the lock. In hot standby mode it's + * definitely necessary.) + */ +XLogRedoAction +XLogReadBufferForRedo(XLogRecPtr lsn, XLogRecord *record, int block_index, + RelFileNode rnode, BlockNumber blkno, + Buffer *buf) +{ + return XLogReadBufferForRedoExtended(lsn, record, block_index, + rnode, MAIN_FORKNUM, blkno, + RBM_NORMAL, false, buf); +} + +/* + * XLogReadBufferForRedoExtended + * Like XLogReadBufferForRedo, but with extra options. + * + * If mode is RBM_ZERO or RBM_ZERO_ON_ERROR, if the page doesn't exist, the + * relation is extended with all-zeroes pages up to the referenced block + * number. In RBM_ZERO mode, the return value is always BLK_NEEDS_REDO. + * + * If 'get_cleanup_lock' is true, a "cleanup lock" is acquired on the buffer + * using LockBufferForCleanup(), instead of a regular exclusive lock. + */ +XLogRedoAction +XLogReadBufferForRedoExtended(XLogRecPtr lsn, XLogRecord *record, + int block_index, RelFileNode rnode, + ForkNumber forkno, BlockNumber blkno, + ReadBufferMode mode, bool get_cleanup_lock, + Buffer *buf) +{ + if (record->xl_info & XLR_BKP_BLOCK(block_index)) + { + *buf = RestoreBackupBlock(lsn, record, block_index, + get_cleanup_lock, true); + return BLK_RESTORED; + } + else + { + *buf = XLogReadBufferExtended(rnode, forkno, blkno, mode); + if (BufferIsValid(*buf)) + { + LockBuffer(*buf, BUFFER_LOCK_EXCLUSIVE); + if (lsn <= PageGetLSN(BufferGetPage(*buf))) + return BLK_DONE; + else + return BLK_NEEDS_REDO; + } + else + return BLK_NOTFOUND; + } +} + /* * XLogReadBuffer * Read a page during XLOG replay. diff --git a/src/include/access/xlogutils.h b/src/include/access/xlogutils.h index 58f11d919b..ad579083ab 100644 --- a/src/include/access/xlogutils.h +++ b/src/include/access/xlogutils.h @@ -1,7 +1,7 @@ /* * xlogutils.h * - * PostgreSQL transaction log manager utility routines + * Utilities for replaying WAL records. * * Portions Copyright (c) 1996-2014, PostgreSQL Global Development Group * Portions Copyright (c) 1994, Regents of the University of California @@ -11,6 +11,7 @@ #ifndef XLOG_UTILS_H #define XLOG_UTILS_H +#include "access/xlog.h" #include "storage/bufmgr.h" @@ -22,6 +23,26 @@ extern void XLogDropDatabase(Oid dbid); extern void XLogTruncateRelation(RelFileNode rnode, ForkNumber forkNum, BlockNumber nblocks); +/* Result codes for XLogReadBufferForRedo[Extended] */ +typedef enum +{ + BLK_NEEDS_REDO, /* changes from WAL record need to be applied */ + BLK_DONE, /* block is already up-to-date */ + BLK_RESTORED, /* block was restored from a full-page image */ + BLK_NOTFOUND /* block was not found (and hence does not need to be + * replayed) */ +} XLogRedoAction; + +extern XLogRedoAction XLogReadBufferForRedo(XLogRecPtr lsn, XLogRecord *record, + int block_index, RelFileNode rnode, BlockNumber blkno, + Buffer *buf); +extern XLogRedoAction XLogReadBufferForRedoExtended(XLogRecPtr lsn, + XLogRecord *record, int block_index, + RelFileNode rnode, ForkNumber forkno, + BlockNumber blkno, + ReadBufferMode mode, bool get_cleanup_lock, + Buffer *buf); + extern Buffer XLogReadBuffer(RelFileNode rnode, BlockNumber blkno, bool init); extern Buffer XLogReadBufferExtended(RelFileNode rnode, ForkNumber forknum, BlockNumber blkno, ReadBufferMode mode);