/* * _hash_splitbucket -- split 'obucket' into 'obucket' and 'nbucket' * * We are splitting a bucket that consists of a base bucket page and zero * or more overflow (bucket chain) pages. We must relocate tuples that * belong in the new bucket, and compress out any free space in the old * bucket. * * The caller must hold exclusive locks on both buckets to ensure that * no one else is trying to access them (see README). * * The caller must hold a pin, but no lock, on the metapage buffer. * The buffer is returned in the same state. (The metapage is only * touched if it becomes necessary to add or remove overflow pages.) */ static void _hash_splitbucket(Relation rel, Buffer metabuf, Bucket obucket, Bucket nbucket, BlockNumber start_oblkno, BlockNumber start_nblkno, uint32 maxbucket, uint32 highmask, uint32 lowmask) { Bucket bucket; Buffer obuf; Buffer nbuf; BlockNumber oblkno; BlockNumber nblkno; bool null; Datum datum; HashPageOpaque oopaque; HashPageOpaque nopaque; IndexTuple itup; Size itemsz; OffsetNumber ooffnum; OffsetNumber noffnum; OffsetNumber omaxoffnum; Page opage; Page npage; TupleDesc itupdesc = RelationGetDescr(rel); MIRROREDLOCK_BUFMGR_MUST_ALREADY_BE_HELD; /* * It should be okay to simultaneously write-lock pages from each bucket, * since no one else can be trying to acquire buffer lock on pages of * either bucket. */ oblkno = start_oblkno; obuf = _hash_getbuf(rel, oblkno, HASH_WRITE); _hash_checkpage(rel, obuf, LH_BUCKET_PAGE); opage = BufferGetPage(obuf); oopaque = (HashPageOpaque) PageGetSpecialPointer(opage); nblkno = start_nblkno; nbuf = _hash_getbuf(rel, nblkno, HASH_WRITE); npage = BufferGetPage(nbuf); /* initialize the new bucket's primary page */ _hash_pageinit(npage, BufferGetPageSize(nbuf)); nopaque = (HashPageOpaque) PageGetSpecialPointer(npage); nopaque->hasho_prevblkno = InvalidBlockNumber; nopaque->hasho_nextblkno = InvalidBlockNumber; nopaque->hasho_bucket = nbucket; nopaque->hasho_flag = LH_BUCKET_PAGE; nopaque->hasho_filler = HASHO_FILL; /* * Partition the tuples in the old bucket between the old bucket and the * new bucket, advancing along the old bucket's overflow bucket chain and * adding overflow pages to the new bucket as needed. */ ooffnum = FirstOffsetNumber; omaxoffnum = PageGetMaxOffsetNumber(opage); for (;;) { /* * at each iteration through this loop, each of these variables should * be up-to-date: obuf opage oopaque ooffnum omaxoffnum */ /* check if we're at the end of the page */ if (ooffnum > omaxoffnum) { /* at end of page, but check for an(other) overflow page */ oblkno = oopaque->hasho_nextblkno; if (!BlockNumberIsValid(oblkno)) break; /* * we ran out of tuples on this particular page, but we have more * overflow pages; advance to next page. */ _hash_wrtbuf(rel, obuf); obuf = _hash_getbuf(rel, oblkno, HASH_WRITE); _hash_checkpage(rel, obuf, LH_OVERFLOW_PAGE); opage = BufferGetPage(obuf); oopaque = (HashPageOpaque) PageGetSpecialPointer(opage); ooffnum = FirstOffsetNumber; omaxoffnum = PageGetMaxOffsetNumber(opage); continue; } /* * Re-hash the tuple to determine which bucket it now belongs in. * * It is annoying to call the hash function while holding locks, but * releasing and relocking the page for each tuple is unappealing too. */ itup = (IndexTuple) PageGetItem(opage, PageGetItemId(opage, ooffnum)); datum = index_getattr(itup, 1, itupdesc, &null); Assert(!null); bucket = _hash_hashkey2bucket(_hash_datum2hashkey(rel, datum), maxbucket, highmask, lowmask); if (bucket == nbucket) { /* * insert the tuple into the new bucket. if it doesn't fit on the * current page in the new bucket, we must allocate a new overflow * page and place the tuple on that page instead. */ itemsz = IndexTupleDSize(*itup); itemsz = MAXALIGN(itemsz); if (PageGetFreeSpace(npage) < itemsz) { /* write out nbuf and drop lock, but keep pin */ _hash_chgbufaccess(rel, nbuf, HASH_WRITE, HASH_NOLOCK); /* chain to a new overflow page */ nbuf = _hash_addovflpage(rel, metabuf, nbuf); _hash_checkpage(rel, nbuf, LH_OVERFLOW_PAGE); npage = BufferGetPage(nbuf); /* we don't need nopaque within the loop */ } noffnum = OffsetNumberNext(PageGetMaxOffsetNumber(npage)); if (PageAddItem(npage, (Item) itup, itemsz, noffnum, LP_USED) == InvalidOffsetNumber) elog(ERROR, "failed to add index item to \"%s\"", RelationGetRelationName(rel)); /* * now delete the tuple from the old bucket. after this section * of code, 'ooffnum' will actually point to the ItemId to which * we would point if we had advanced it before the deletion * (PageIndexTupleDelete repacks the ItemId array). this also * means that 'omaxoffnum' is exactly one less than it used to be, * so we really can just decrement it instead of calling * PageGetMaxOffsetNumber. */ PageIndexTupleDelete(opage, ooffnum); omaxoffnum = OffsetNumberPrev(omaxoffnum); } else { /* * the tuple stays on this page. we didn't move anything, so we * didn't delete anything and therefore we don't have to change * 'omaxoffnum'. */ Assert(bucket == obucket); ooffnum = OffsetNumberNext(ooffnum); } } /* * We're at the end of the old bucket chain, so we're done partitioning * the tuples. Before quitting, call _hash_squeezebucket to ensure the * tuples remaining in the old bucket (including the overflow pages) are * packed as tightly as possible. The new bucket is already tight. */ _hash_wrtbuf(rel, obuf); _hash_wrtbuf(rel, nbuf); _hash_squeezebucket(rel, obucket, start_oblkno); }
Datum ginbulkdelete(PG_FUNCTION_ARGS) { IndexVacuumInfo *info = (IndexVacuumInfo *) PG_GETARG_POINTER(0); IndexBulkDeleteResult *stats = (IndexBulkDeleteResult *) PG_GETARG_POINTER(1); IndexBulkDeleteCallback callback = (IndexBulkDeleteCallback) PG_GETARG_POINTER(2); void *callback_state = (void *) PG_GETARG_POINTER(3); Relation index = info->index; BlockNumber blkno = GIN_ROOT_BLKNO; GinVacuumState gvs; Buffer buffer; BlockNumber rootOfPostingTree[BLCKSZ / (sizeof(IndexTupleData) + sizeof(ItemId))]; uint32 nRoot; gvs.index = index; gvs.callback = callback; gvs.callback_state = callback_state; gvs.strategy = info->strategy; initGinState(&gvs.ginstate, index); /* first time through? */ if (stats == NULL) { /* Yes, so initialize stats to zeroes */ stats = (IndexBulkDeleteResult *) palloc0(sizeof(IndexBulkDeleteResult)); /* and cleanup any pending inserts */ ginInsertCleanup(index, &gvs.ginstate, true, stats); } /* we'll re-count the tuples each time */ stats->num_index_tuples = 0; gvs.result = stats; buffer = ReadBufferExtended(index, MAIN_FORKNUM, blkno, RBM_NORMAL, info->strategy); /* find leaf page */ for (;;) { Page page = BufferGetPage(buffer); IndexTuple itup; LockBuffer(buffer, GIN_SHARE); Assert(!GinPageIsData(page)); if (GinPageIsLeaf(page)) { LockBuffer(buffer, GIN_UNLOCK); LockBuffer(buffer, GIN_EXCLUSIVE); if (blkno == GIN_ROOT_BLKNO && !GinPageIsLeaf(page)) { LockBuffer(buffer, GIN_UNLOCK); continue; /* check it one more */ } break; } Assert(PageGetMaxOffsetNumber(page) >= FirstOffsetNumber); itup = (IndexTuple) PageGetItem(page, PageGetItemId(page, FirstOffsetNumber)); blkno = GinItemPointerGetBlockNumber(&(itup)->t_tid); Assert(blkno != InvalidBlockNumber); UnlockReleaseBuffer(buffer); buffer = ReadBufferExtended(index, MAIN_FORKNUM, blkno, RBM_NORMAL, info->strategy); } /* right now we found leftmost page in entry's BTree */ for (;;) { Page page = BufferGetPage(buffer); Page resPage; uint32 i; Assert(!GinPageIsData(page)); resPage = ginVacuumEntryPage(&gvs, buffer, rootOfPostingTree, &nRoot); blkno = GinPageGetOpaque(page)->rightlink; if (resPage) { START_CRIT_SECTION(); PageRestoreTempPage(resPage, page); MarkBufferDirty(buffer); xlogVacuumPage(gvs.index, buffer); UnlockReleaseBuffer(buffer); END_CRIT_SECTION(); } else { UnlockReleaseBuffer(buffer); } vacuum_delay_point(); for (i = 0; i < nRoot; i++) { ginVacuumPostingTree(&gvs, rootOfPostingTree[i]); vacuum_delay_point(); } if (blkno == InvalidBlockNumber) /* rightmost page */ break; buffer = ReadBufferExtended(index, MAIN_FORKNUM, blkno, RBM_NORMAL, info->strategy); LockBuffer(buffer, GIN_EXCLUSIVE); } PG_RETURN_POINTER(gvs.result); }
/* ---------------- * index_getnext - get the next heap tuple from a scan * * The result is the next heap tuple satisfying the scan keys and the * snapshot, or NULL if no more matching tuples exist. On success, * the buffer containing the heap tuple is pinned (the pin will be dropped * at the next index_getnext or index_endscan). * * Note: caller must check scan->xs_recheck, and perform rechecking of the * scan keys if required. We do not do that here because we don't have * enough information to do it efficiently in the general case. * ---------------- */ HeapTuple index_getnext(IndexScanDesc scan, ScanDirection direction) { HeapTuple heapTuple = &scan->xs_ctup; ItemPointer tid = &heapTuple->t_self; FmgrInfo *procedure; SCAN_CHECKS; GET_SCAN_PROCEDURE(amgettuple); Assert(TransactionIdIsValid(RecentGlobalXmin)); /* * We always reset xs_hot_dead; if we are here then either we are just * starting the scan, or we previously returned a visible tuple, and in * either case it's inappropriate to kill the prior index entry. */ scan->xs_hot_dead = false; for (;;) { OffsetNumber offnum; bool at_chain_start; Page dp; if (scan->xs_next_hot != InvalidOffsetNumber) { /* * We are resuming scan of a HOT chain after having returned an * earlier member. Must still hold pin on current heap page. */ Assert(BufferIsValid(scan->xs_cbuf)); Assert(ItemPointerGetBlockNumber(tid) == BufferGetBlockNumber(scan->xs_cbuf)); Assert(TransactionIdIsValid(scan->xs_prev_xmax)); offnum = scan->xs_next_hot; at_chain_start = false; scan->xs_next_hot = InvalidOffsetNumber; } else { bool found; Buffer prev_buf; /* * If we scanned a whole HOT chain and found only dead tuples, * tell index AM to kill its entry for that TID. We do not do this * when in recovery because it may violate MVCC to do so. see * comments in RelationGetIndexScan(). */ if (!scan->xactStartedInRecovery) scan->kill_prior_tuple = scan->xs_hot_dead; /* * The AM's gettuple proc finds the next index entry matching the * scan keys, and puts the TID in xs_ctup.t_self (ie, *tid). It * should also set scan->xs_recheck, though we pay no attention to * that here. */ found = DatumGetBool(FunctionCall2(procedure, PointerGetDatum(scan), Int32GetDatum(direction))); /* Reset kill flag immediately for safety */ scan->kill_prior_tuple = false; /* If we're out of index entries, break out of outer loop */ if (!found) break; pgstat_count_index_tuples(scan->indexRelation, 1); /* Switch to correct buffer if we don't have it already */ prev_buf = scan->xs_cbuf; scan->xs_cbuf = ReleaseAndReadBuffer(scan->xs_cbuf, scan->heapRelation, ItemPointerGetBlockNumber(tid)); /* * Prune page, but only if we weren't already on this page */ if (prev_buf != scan->xs_cbuf) heap_page_prune_opt(scan->heapRelation, scan->xs_cbuf, RecentGlobalXmin); /* Prepare to scan HOT chain starting at index-referenced offnum */ offnum = ItemPointerGetOffsetNumber(tid); at_chain_start = true; /* We don't know what the first tuple's xmin should be */ scan->xs_prev_xmax = InvalidTransactionId; /* Initialize flag to detect if all entries are dead */ scan->xs_hot_dead = true; } /* Obtain share-lock on the buffer so we can examine visibility */ LockBuffer(scan->xs_cbuf, BUFFER_LOCK_SHARE); dp = (Page) BufferGetPage(scan->xs_cbuf); /* Scan through possible multiple members of HOT-chain */ for (;;) { ItemId lp; ItemPointer ctid; bool valid; /* check for bogus TID */ if (offnum < FirstOffsetNumber || offnum > PageGetMaxOffsetNumber(dp)) break; lp = PageGetItemId(dp, offnum); /* check for unused, dead, or redirected items */ if (!ItemIdIsNormal(lp)) { /* We should only see a redirect at start of chain */ if (ItemIdIsRedirected(lp) && at_chain_start) { /* Follow the redirect */ offnum = ItemIdGetRedirect(lp); at_chain_start = false; continue; } /* else must be end of chain */ break; } /* * We must initialize all of *heapTuple (ie, scan->xs_ctup) since * it is returned to the executor on success. */ heapTuple->t_data = (HeapTupleHeader) PageGetItem(dp, lp); heapTuple->t_len = ItemIdGetLength(lp); ItemPointerSetOffsetNumber(tid, offnum); heapTuple->t_tableOid = RelationGetRelid(scan->heapRelation); ctid = &heapTuple->t_data->t_ctid; /* * Shouldn't see a HEAP_ONLY tuple at chain start. (This test * should be unnecessary, since the chain root can't be removed * while we have pin on the index entry, but let's make it * anyway.) */ if (at_chain_start && HeapTupleIsHeapOnly(heapTuple)) break; /* * The xmin should match the previous xmax value, else chain is * broken. (Note: this test is not optional because it protects * us against the case where the prior chain member's xmax aborted * since we looked at it.) */ if (TransactionIdIsValid(scan->xs_prev_xmax) && !TransactionIdEquals(scan->xs_prev_xmax, HeapTupleHeaderGetXmin(heapTuple->t_data))) break; /* If it's visible per the snapshot, we must return it */ valid = HeapTupleSatisfiesVisibility(heapTuple, scan->xs_snapshot, scan->xs_cbuf); CheckForSerializableConflictOut(valid, scan->heapRelation, heapTuple, scan->xs_cbuf); if (valid) { /* * If the snapshot is MVCC, we know that it could accept at * most one member of the HOT chain, so we can skip examining * any more members. Otherwise, check for continuation of the * HOT-chain, and set state for next time. */ if (IsMVCCSnapshot(scan->xs_snapshot) && !IsolationIsSerializable()) scan->xs_next_hot = InvalidOffsetNumber; else if (HeapTupleIsHotUpdated(heapTuple)) { Assert(ItemPointerGetBlockNumber(ctid) == ItemPointerGetBlockNumber(tid)); scan->xs_next_hot = ItemPointerGetOffsetNumber(ctid); scan->xs_prev_xmax = HeapTupleHeaderGetXmax(heapTuple->t_data); } else scan->xs_next_hot = InvalidOffsetNumber; PredicateLockTuple(scan->heapRelation, heapTuple); LockBuffer(scan->xs_cbuf, BUFFER_LOCK_UNLOCK); pgstat_count_heap_fetch(scan->indexRelation); return heapTuple; } /* * If we can't see it, maybe no one else can either. Check to see * if the tuple is dead to all transactions. If we find that all * the tuples in the HOT chain are dead, we'll signal the index AM * to not return that TID on future indexscans. */ if (scan->xs_hot_dead && HeapTupleSatisfiesVacuum(heapTuple->t_data, RecentGlobalXmin, scan->xs_cbuf) != HEAPTUPLE_DEAD) scan->xs_hot_dead = false; /* * Check to see if HOT chain continues past this tuple; if so * fetch the next offnum (we don't bother storing it into * xs_next_hot, but must store xs_prev_xmax), and loop around. */ if (HeapTupleIsHotUpdated(heapTuple)) { Assert(ItemPointerGetBlockNumber(ctid) == ItemPointerGetBlockNumber(tid)); offnum = ItemPointerGetOffsetNumber(ctid); at_chain_start = false; scan->xs_prev_xmax = HeapTupleHeaderGetXmax(heapTuple->t_data); } else break; /* end of chain */ } /* loop over a single HOT chain */ LockBuffer(scan->xs_cbuf, BUFFER_LOCK_UNLOCK); /* Loop around to ask index AM for another TID */ scan->xs_next_hot = InvalidOffsetNumber; } /* Release any held pin on a heap page */ if (BufferIsValid(scan->xs_cbuf)) { ReleaseBuffer(scan->xs_cbuf); scan->xs_cbuf = InvalidBuffer; } return NULL; /* failure exit */ }
/* * find entry with lowest penalty */ OffsetNumber gistchoose(Relation r, Page p, IndexTuple it, /* it has compressed entry */ GISTSTATE *giststate) { OffsetNumber maxoff; OffsetNumber i; OffsetNumber which; float sum_grow, which_grow[INDEX_MAX_KEYS]; GISTENTRY entry, identry[INDEX_MAX_KEYS]; bool isnull[INDEX_MAX_KEYS]; maxoff = PageGetMaxOffsetNumber(p); *which_grow = -1.0; which = InvalidOffsetNumber; sum_grow = 1; gistDeCompressAtt(giststate, r, it, NULL, (OffsetNumber) 0, identry, isnull); Assert(maxoff >= FirstOffsetNumber); Assert(!GistPageIsLeaf(p)); for (i = FirstOffsetNumber; i <= maxoff && sum_grow; i = OffsetNumberNext(i)) { int j; IndexTuple itup = (IndexTuple) PageGetItem(p, PageGetItemId(p, i)); sum_grow = 0; for (j = 0; j < r->rd_att->natts; j++) { Datum datum; float usize; bool IsNull; datum = index_getattr(itup, j + 1, giststate->tupdesc, &IsNull); gistdentryinit(giststate, j, &entry, datum, r, p, i, FALSE, IsNull); usize = gistpenalty(giststate, j, &entry, IsNull, &identry[j], isnull[j]); if (which_grow[j] < 0 || usize < which_grow[j]) { which = i; which_grow[j] = usize; if (j < r->rd_att->natts - 1 && i == FirstOffsetNumber) which_grow[j + 1] = -1; sum_grow += which_grow[j]; } else if (which_grow[j] == usize) sum_grow += usize; else { sum_grow = 1; break; } } } if (which == InvalidOffsetNumber) which = FirstOffsetNumber; return which; }
/* * Fetch a tuples that matchs the search key; this can be invoked * either to fetch the first such tuple or subsequent matching * tuples. Returns true iff a matching tuple was found. */ static int gistnext(IndexScanDesc scan, ScanDirection dir, ItemPointer tids, int maxtids, bool ignore_killed_tuples) { MIRROREDLOCK_BUFMGR_DECLARE; Page p; OffsetNumber n; GISTScanOpaque so; GISTSearchStack *stk; IndexTuple it; GISTPageOpaque opaque; int ntids = 0; so = (GISTScanOpaque) scan->opaque; // -------- MirroredLock ---------- MIRROREDLOCK_BUFMGR_LOCK; if ( so->qual_ok == false ) return 0; if (ItemPointerIsValid(&so->curpos) == false) { /* Being asked to fetch the first entry, so start at the root */ Assert(so->curbuf == InvalidBuffer); Assert(so->stack == NULL); so->curbuf = ReadBuffer(scan->indexRelation, GIST_ROOT_BLKNO); stk = so->stack = (GISTSearchStack *) palloc0(sizeof(GISTSearchStack)); stk->next = NULL; stk->block = GIST_ROOT_BLKNO; pgstat_count_index_scan(scan->indexRelation); } else if (so->curbuf == InvalidBuffer) { MIRROREDLOCK_BUFMGR_UNLOCK; // -------- MirroredLock ---------- return 0; } /* * check stored pointers from last visit */ if ( so->nPageData > 0 ) { while( ntids < maxtids && so->curPageData < so->nPageData ) { tids[ ntids ] = scan->xs_ctup.t_self = so->pageData[ so->curPageData ].heapPtr; ItemPointerSet(&(so->curpos), BufferGetBlockNumber(so->curbuf), so->pageData[ so->curPageData ].pageOffset); so->curPageData ++; ntids++; } if ( ntids == maxtids ) { MIRROREDLOCK_BUFMGR_UNLOCK; // -------- MirroredLock ---------- return ntids; } /* * Go to the next page */ stk = so->stack->next; pfree(so->stack); so->stack = stk; /* If we're out of stack entries, we're done */ if (so->stack == NULL) { ReleaseBuffer(so->curbuf); so->curbuf = InvalidBuffer; MIRROREDLOCK_BUFMGR_UNLOCK; // -------- MirroredLock ---------- return ntids; } so->curbuf = ReleaseAndReadBuffer(so->curbuf, scan->indexRelation, stk->block); } for (;;) { /* First of all, we need lock buffer */ Assert(so->curbuf != InvalidBuffer); LockBuffer(so->curbuf, GIST_SHARE); gistcheckpage(scan->indexRelation, so->curbuf); p = BufferGetPage(so->curbuf); opaque = GistPageGetOpaque(p); /* remember lsn to identify page changed for tuple's killing */ so->stack->lsn = PageGetLSN(p); /* check page split, occured from last visit or visit to parent */ if (!XLogRecPtrIsInvalid(so->stack->parentlsn) && XLByteLT(so->stack->parentlsn, opaque->nsn) && opaque->rightlink != InvalidBlockNumber /* sanity check */ && (so->stack->next == NULL || so->stack->next->block != opaque->rightlink) /* check if already added */ ) { /* detect page split, follow right link to add pages */ stk = (GISTSearchStack *) palloc(sizeof(GISTSearchStack)); stk->next = so->stack->next; stk->block = opaque->rightlink; stk->parentlsn = so->stack->parentlsn; memset(&(stk->lsn), 0, sizeof(GistNSN)); so->stack->next = stk; } /* if page is empty, then just skip it */ if (PageIsEmpty(p)) { LockBuffer(so->curbuf, GIST_UNLOCK); stk = so->stack->next; pfree(so->stack); so->stack = stk; if (so->stack == NULL) { ReleaseBuffer(so->curbuf); so->curbuf = InvalidBuffer; MIRROREDLOCK_BUFMGR_UNLOCK; // -------- MirroredLock ---------- return ntids; } so->curbuf = ReleaseAndReadBuffer(so->curbuf, scan->indexRelation, stk->block); continue; } if (ScanDirectionIsBackward(dir)) n = PageGetMaxOffsetNumber(p); else n = FirstOffsetNumber; /* wonderful, we can look at page */ so->nPageData = so->curPageData = 0; for (;;) { n = gistfindnext(scan, n, dir); if (!OffsetNumberIsValid(n)) { while( ntids < maxtids && so->curPageData < so->nPageData ) { tids[ ntids ] = scan->xs_ctup.t_self = so->pageData[ so->curPageData ].heapPtr; ItemPointerSet(&(so->curpos), BufferGetBlockNumber(so->curbuf), so->pageData[ so->curPageData ].pageOffset); so->curPageData ++; ntids++; } if ( ntids == maxtids ) { LockBuffer(so->curbuf, GIST_UNLOCK); MIRROREDLOCK_BUFMGR_UNLOCK; // -------- MirroredLock ---------- return ntids; } /* * We ran out of matching index entries on the current page, * so pop the top stack entry and use it to continue the * search. */ LockBuffer(so->curbuf, GIST_UNLOCK); stk = so->stack->next; pfree(so->stack); so->stack = stk; /* If we're out of stack entries, we're done */ if (so->stack == NULL) { ReleaseBuffer(so->curbuf); so->curbuf = InvalidBuffer; MIRROREDLOCK_BUFMGR_UNLOCK; // -------- MirroredLock ---------- return ntids; } so->curbuf = ReleaseAndReadBuffer(so->curbuf, scan->indexRelation, stk->block); /* XXX go up */ break; } if (GistPageIsLeaf(p)) { /* * We've found a matching index entry in a leaf page, so * return success. Note that we keep "curbuf" pinned so that * we can efficiently resume the index scan later. */ if (!(ignore_killed_tuples && ItemIdIsDead(PageGetItemId(p, n)))) { it = (IndexTuple) PageGetItem(p, PageGetItemId(p, n)); so->pageData[ so->nPageData ].heapPtr = it->t_tid; so->pageData[ so->nPageData ].pageOffset = n; so->nPageData ++; } } else { /* * We've found an entry in an internal node whose key is * consistent with the search key, so push it to stack */ stk = (GISTSearchStack *) palloc(sizeof(GISTSearchStack)); it = (IndexTuple) PageGetItem(p, PageGetItemId(p, n)); stk->block = ItemPointerGetBlockNumber(&(it->t_tid)); memset(&(stk->lsn), 0, sizeof(GistNSN)); stk->parentlsn = so->stack->lsn; stk->next = so->stack->next; so->stack->next = stk; } if (ScanDirectionIsBackward(dir)) n = OffsetNumberPrev(n); else n = OffsetNumberNext(n); } } MIRROREDLOCK_BUFMGR_UNLOCK; // -------- MirroredLock ---------- return ntids; }
/* * btvacuumpage --- VACUUM one page * * This processes a single page for btvacuumscan(). In some cases we * must go back and re-examine previously-scanned pages; this routine * recurses when necessary to handle that case. * * blkno is the page to process. orig_blkno is the highest block number * reached by the outer btvacuumscan loop (the same as blkno, unless we * are recursing to re-examine a previous page). */ static void btvacuumpage(BTVacState *vstate, BlockNumber blkno, BlockNumber orig_blkno) { IndexVacuumInfo *info = vstate->info; IndexBulkDeleteResult *stats = vstate->stats; IndexBulkDeleteCallback callback = vstate->callback; void *callback_state = vstate->callback_state; Relation rel = info->index; bool delete_now; BlockNumber recurse_to; Buffer buf; Page page; BTPageOpaque opaque = NULL; restart: delete_now = false; recurse_to = P_NONE; /* call vacuum_delay_point while not holding any buffer lock */ vacuum_delay_point(); /* * We can't use _bt_getbuf() here because it always applies * _bt_checkpage(), which will barf on an all-zero page. We want to * recycle all-zero pages, not fail. Also, we want to use a nondefault * buffer access strategy. */ buf = ReadBufferExtended(rel, MAIN_FORKNUM, blkno, RBM_NORMAL, info->strategy); LockBuffer(buf, BT_READ); page = BufferGetPage(buf); if (!PageIsNew(page)) { _bt_checkpage(rel, buf); opaque = (BTPageOpaque) PageGetSpecialPointer(page); } /* * If we are recursing, the only case we want to do anything with is a * live leaf page having the current vacuum cycle ID. Any other state * implies we already saw the page (eg, deleted it as being empty). */ if (blkno != orig_blkno) { if (_bt_page_recyclable(page) || P_IGNORE(opaque) || !P_ISLEAF(opaque) || opaque->btpo_cycleid != vstate->cycleid) { _bt_relbuf(rel, buf); return; } } /* Page is valid, see what to do with it */ if (_bt_page_recyclable(page)) { /* Okay to recycle this page */ RecordFreeIndexPage(rel, blkno); vstate->totFreePages++; stats->pages_deleted++; } else if (P_ISDELETED(opaque)) { /* Already deleted, but can't recycle yet */ stats->pages_deleted++; } else if (P_ISHALFDEAD(opaque)) { /* Half-dead, try to delete */ delete_now = true; } else if (P_ISLEAF(opaque)) { OffsetNumber deletable[MaxOffsetNumber]; int ndeletable; OffsetNumber offnum, minoff, maxoff; /* * Trade in the initial read lock for a super-exclusive write lock on * this page. We must get such a lock on every leaf page over the * course of the vacuum scan, whether or not it actually contains any * deletable tuples --- see nbtree/README. */ LockBuffer(buf, BUFFER_LOCK_UNLOCK); LockBufferForCleanup(buf); /* * Remember highest leaf page number we've taken cleanup lock on; see * notes in btvacuumscan */ if (blkno > vstate->lastBlockLocked) vstate->lastBlockLocked = blkno; /* * Check whether we need to recurse back to earlier pages. What we * are concerned about is a page split that happened since we started * the vacuum scan. If the split moved some tuples to a lower page * then we might have missed 'em. If so, set up for tail recursion. * (Must do this before possibly clearing btpo_cycleid below!) */ if (vstate->cycleid != 0 && opaque->btpo_cycleid == vstate->cycleid && !(opaque->btpo_flags & BTP_SPLIT_END) && !P_RIGHTMOST(opaque) && opaque->btpo_next < orig_blkno) recurse_to = opaque->btpo_next; /* * Scan over all items to see which ones need deleted according to the * callback function. */ ndeletable = 0; minoff = P_FIRSTDATAKEY(opaque); maxoff = PageGetMaxOffsetNumber(page); if (callback) { for (offnum = minoff; offnum <= maxoff; offnum = OffsetNumberNext(offnum)) { IndexTuple itup; ItemPointer htup; itup = (IndexTuple) PageGetItem(page, PageGetItemId(page, offnum)); htup = &(itup->t_tid); /* * During Hot Standby we currently assume that * XLOG_BTREE_VACUUM records do not produce conflicts. That is * only true as long as the callback function depends only * upon whether the index tuple refers to heap tuples removed * in the initial heap scan. When vacuum starts it derives a * value of OldestXmin. Backends taking later snapshots could * have a RecentGlobalXmin with a later xid than the vacuum's * OldestXmin, so it is possible that row versions deleted * after OldestXmin could be marked as killed by other * backends. The callback function *could* look at the index * tuple state in isolation and decide to delete the index * tuple, though currently it does not. If it ever did, we * would need to reconsider whether XLOG_BTREE_VACUUM records * should cause conflicts. If they did cause conflicts they * would be fairly harsh conflicts, since we haven't yet * worked out a way to pass a useful value for * latestRemovedXid on the XLOG_BTREE_VACUUM records. This * applies to *any* type of index that marks index tuples as * killed. */ if (callback(htup, callback_state)) deletable[ndeletable++] = offnum; } } /* * Apply any needed deletes. We issue just one _bt_delitems_vacuum() * call per page, so as to minimize WAL traffic. */ if (ndeletable > 0) { BlockNumber lastBlockVacuumed = InvalidBlockNumber; /* * We may need to record the lastBlockVacuumed for use when * non-MVCC scans might be performed on the index on a * hot standby. See explanation in btree_xlog_vacuum(). * * On a hot standby, a non-MVCC scan can only take place * when we access a Toast Index, so we need only record * the lastBlockVacuumed if we are vacuuming a Toast Index. */ if (rel->rd_rel->relnamespace == PG_TOAST_NAMESPACE) lastBlockVacuumed = vstate->lastBlockVacuumed; /* * Notice that the issued XLOG_BTREE_VACUUM WAL record includes an * instruction to the replay code to get cleanup lock on all pages * between the previous lastBlockVacuumed and this page. This * ensures that WAL replay locks all leaf pages at some point. * * Since we can visit leaf pages out-of-order when recursing, * replay might end up locking such pages an extra time, but it * doesn't seem worth the amount of bookkeeping it'd take to avoid * that. */ _bt_delitems_vacuum(rel, buf, deletable, ndeletable, lastBlockVacuumed); /* * Remember highest leaf page number we've issued a * XLOG_BTREE_VACUUM WAL record for. */ if (blkno > vstate->lastBlockVacuumed) vstate->lastBlockVacuumed = blkno; stats->tuples_removed += ndeletable; /* must recompute maxoff */ maxoff = PageGetMaxOffsetNumber(page); } else { /* * If the page has been split during this vacuum cycle, it seems * worth expending a write to clear btpo_cycleid even if we don't * have any deletions to do. (If we do, _bt_delitems_vacuum takes * care of this.) This ensures we won't process the page again. * * We treat this like a hint-bit update because there's no need to * WAL-log it. */ if (vstate->cycleid != 0 && opaque->btpo_cycleid == vstate->cycleid) { opaque->btpo_cycleid = 0; MarkBufferDirtyHint(buf, true); } } /* * If it's now empty, try to delete; else count the live tuples. We * don't delete when recursing, though, to avoid putting entries into * freePages out-of-order (doesn't seem worth any extra code to handle * the case). */ if (minoff > maxoff) delete_now = (blkno == orig_blkno); else stats->num_index_tuples += maxoff - minoff + 1; } if (delete_now) { MemoryContext oldcontext; int ndel; /* Run pagedel in a temp context to avoid memory leakage */ MemoryContextReset(vstate->pagedelcontext); oldcontext = MemoryContextSwitchTo(vstate->pagedelcontext); ndel = _bt_pagedel(rel, buf); /* count only this page, else may double-count parent */ if (ndel) stats->pages_deleted++; MemoryContextSwitchTo(oldcontext); /* pagedel released buffer, so we shouldn't */ } else _bt_relbuf(rel, buf); /* * This is really tail recursion, but if the compiler is too stupid to * optimize it as such, we'd eat an uncomfortably large amount of stack * space per recursion level (due to the deletable[] array). A failure is * improbable since the number of levels isn't likely to be large ... but * just in case, let's hand-optimize into a loop. */ if (recurse_to != P_NONE) { blkno = recurse_to; goto restart; } }
static void gistFindCorrectParent(Relation r, GISTInsertStack *child) { GISTInsertStack *parent = child->parent; LockBuffer(parent->buffer, GIST_EXCLUSIVE); gistcheckpage(r, parent->buffer); parent->page = (Page) BufferGetPage(parent->buffer); /* here we don't need to distinguish between split and page update */ if (parent->childoffnum == InvalidOffsetNumber || !XLByteEQ(parent->lsn, PageGetLSN(parent->page))) { /* parent is changed, look child in right links until found */ OffsetNumber i, maxoff; ItemId iid; IndexTuple idxtuple; GISTInsertStack *ptr; while (true) { maxoff = PageGetMaxOffsetNumber(parent->page); for (i = FirstOffsetNumber; i <= maxoff; i = OffsetNumberNext(i)) { iid = PageGetItemId(parent->page, i); idxtuple = (IndexTuple) PageGetItem(parent->page, iid); if (ItemPointerGetBlockNumber(&(idxtuple->t_tid)) == child->blkno) { /* yes!!, found */ parent->childoffnum = i; return; } } parent->blkno = GistPageGetOpaque(parent->page)->rightlink; UnlockReleaseBuffer(parent->buffer); if (parent->blkno == InvalidBlockNumber) /* * end of chain and still didn't found parent, It's very-very * rare situation when root splited */ break; parent->buffer = ReadBuffer(r, parent->blkno); LockBuffer(parent->buffer, GIST_EXCLUSIVE); gistcheckpage(r, parent->buffer); parent->page = (Page) BufferGetPage(parent->buffer); } /* * awful!!, we need search tree to find parent ... , but before we * should release all old parent */ ptr = child->parent->parent; /* child->parent already released * above */ while (ptr) { ReleaseBuffer(ptr->buffer); ptr = ptr->parent; } /* ok, find new path */ ptr = parent = gistFindPath(r, child->blkno); Assert(ptr != NULL); /* read all buffers as expected by caller */ /* note we don't lock them or gistcheckpage them here! */ while (ptr) { ptr->buffer = ReadBuffer(r, ptr->blkno); ptr->page = (Page) BufferGetPage(ptr->buffer); ptr = ptr->parent; } /* install new chain of parents to stack */ child->parent = parent; parent->child = child; /* make recursive call to normal processing */ gistFindCorrectParent(r, child); } return; }
/* * Helper function to perform deletion of index entries from a bucket. * * This function expects that the caller has acquired a cleanup lock on the * primary bucket page, and will return with a write lock again held on the * primary bucket page. The lock won't necessarily be held continuously, * though, because we'll release it when visiting overflow pages. * * It would be very bad if this function cleaned a page while some other * backend was in the midst of scanning it, because hashgettuple assumes * that the next valid TID will be greater than or equal to the current * valid TID. There can't be any concurrent scans in progress when we first * enter this function because of the cleanup lock we hold on the primary * bucket page, but as soon as we release that lock, there might be. We * handle that by conspiring to prevent those scans from passing our cleanup * scan. To do that, we lock the next page in the bucket chain before * releasing the lock on the previous page. (This type of lock chaining is * not ideal, so we might want to look for a better solution at some point.) * * We need to retain a pin on the primary bucket to ensure that no concurrent * split can start. */ void hashbucketcleanup(Relation rel, Bucket cur_bucket, Buffer bucket_buf, BlockNumber bucket_blkno, BufferAccessStrategy bstrategy, uint32 maxbucket, uint32 highmask, uint32 lowmask, double *tuples_removed, double *num_index_tuples, bool split_cleanup, IndexBulkDeleteCallback callback, void *callback_state) { BlockNumber blkno; Buffer buf; Bucket new_bucket PG_USED_FOR_ASSERTS_ONLY = InvalidBucket; bool bucket_dirty = false; blkno = bucket_blkno; buf = bucket_buf; if (split_cleanup) new_bucket = _hash_get_newbucket_from_oldbucket(rel, cur_bucket, lowmask, maxbucket); /* Scan each page in bucket */ for (;;) { HashPageOpaque opaque; OffsetNumber offno; OffsetNumber maxoffno; Buffer next_buf; Page page; OffsetNumber deletable[MaxOffsetNumber]; int ndeletable = 0; bool retain_pin = false; bool clear_dead_marking = false; vacuum_delay_point(); page = BufferGetPage(buf); opaque = (HashPageOpaque) PageGetSpecialPointer(page); /* Scan each tuple in page */ maxoffno = PageGetMaxOffsetNumber(page); for (offno = FirstOffsetNumber; offno <= maxoffno; offno = OffsetNumberNext(offno)) { ItemPointer htup; IndexTuple itup; Bucket bucket; bool kill_tuple = false; itup = (IndexTuple) PageGetItem(page, PageGetItemId(page, offno)); htup = &(itup->t_tid); /* * To remove the dead tuples, we strictly want to rely on results * of callback function. refer btvacuumpage for detailed reason. */ if (callback && callback(htup, callback_state)) { kill_tuple = true; if (tuples_removed) *tuples_removed += 1; } else if (split_cleanup) { /* delete the tuples that are moved by split. */ bucket = _hash_hashkey2bucket(_hash_get_indextuple_hashkey(itup), maxbucket, highmask, lowmask); /* mark the item for deletion */ if (bucket != cur_bucket) { /* * We expect tuples to either belong to curent bucket or * new_bucket. This is ensured because we don't allow * further splits from bucket that contains garbage. See * comments in _hash_expandtable. */ Assert(bucket == new_bucket); kill_tuple = true; } } if (kill_tuple) { /* mark the item for deletion */ deletable[ndeletable++] = offno; } else { /* we're keeping it, so count it */ if (num_index_tuples) *num_index_tuples += 1; } } /* retain the pin on primary bucket page till end of bucket scan */ if (blkno == bucket_blkno) retain_pin = true; else retain_pin = false; blkno = opaque->hasho_nextblkno; /* * Apply deletions, advance to next page and write page if needed. */ if (ndeletable > 0) { /* No ereport(ERROR) until changes are logged */ START_CRIT_SECTION(); PageIndexMultiDelete(page, deletable, ndeletable); bucket_dirty = true; /* * Let us mark the page as clean if vacuum removes the DEAD tuples * from an index page. We do this by clearing LH_PAGE_HAS_DEAD_TUPLES * flag. */ if (tuples_removed && *tuples_removed > 0 && opaque->hasho_flag & LH_PAGE_HAS_DEAD_TUPLES) { opaque->hasho_flag &= ~LH_PAGE_HAS_DEAD_TUPLES; clear_dead_marking = true; } MarkBufferDirty(buf); /* XLOG stuff */ if (RelationNeedsWAL(rel)) { xl_hash_delete xlrec; XLogRecPtr recptr; xlrec.clear_dead_marking = clear_dead_marking; xlrec.is_primary_bucket_page = (buf == bucket_buf) ? true : false; XLogBeginInsert(); XLogRegisterData((char *) &xlrec, SizeOfHashDelete); /* * bucket buffer needs to be registered to ensure that we can * acquire a cleanup lock on it during replay. */ if (!xlrec.is_primary_bucket_page) XLogRegisterBuffer(0, bucket_buf, REGBUF_STANDARD | REGBUF_NO_IMAGE); XLogRegisterBuffer(1, buf, REGBUF_STANDARD); XLogRegisterBufData(1, (char *) deletable, ndeletable * sizeof(OffsetNumber)); recptr = XLogInsert(RM_HASH_ID, XLOG_HASH_DELETE); PageSetLSN(BufferGetPage(buf), recptr); } END_CRIT_SECTION(); } /* bail out if there are no more pages to scan. */ if (!BlockNumberIsValid(blkno)) break; next_buf = _hash_getbuf_with_strategy(rel, blkno, HASH_WRITE, LH_OVERFLOW_PAGE, bstrategy); /* * release the lock on previous page after acquiring the lock on next * page */ if (retain_pin) LockBuffer(buf, BUFFER_LOCK_UNLOCK); else _hash_relbuf(rel, buf); buf = next_buf; } /* * lock the bucket page to clear the garbage flag and squeeze the bucket. * if the current buffer is same as bucket buffer, then we already have * lock on bucket page. */ if (buf != bucket_buf) { _hash_relbuf(rel, buf); LockBuffer(bucket_buf, BUFFER_LOCK_EXCLUSIVE); } /* * Clear the garbage flag from bucket after deleting the tuples that are * moved by split. We purposefully clear the flag before squeeze bucket, * so that after restart, vacuum shouldn't again try to delete the moved * by split tuples. */ if (split_cleanup) { HashPageOpaque bucket_opaque; Page page; page = BufferGetPage(bucket_buf); bucket_opaque = (HashPageOpaque) PageGetSpecialPointer(page); /* No ereport(ERROR) until changes are logged */ START_CRIT_SECTION(); bucket_opaque->hasho_flag &= ~LH_BUCKET_NEEDS_SPLIT_CLEANUP; MarkBufferDirty(bucket_buf); /* XLOG stuff */ if (RelationNeedsWAL(rel)) { XLogRecPtr recptr; XLogBeginInsert(); XLogRegisterBuffer(0, bucket_buf, REGBUF_STANDARD); recptr = XLogInsert(RM_HASH_ID, XLOG_HASH_SPLIT_CLEANUP); PageSetLSN(page, recptr); } END_CRIT_SECTION(); } /* * If we have deleted anything, try to compact free space. For squeezing * the bucket, we must have a cleanup lock, else it can impact the * ordering of tuples for a scan that has started before it. */ if (bucket_dirty && IsBufferCleanupOK(bucket_buf)) _hash_squeezebucket(rel, cur_bucket, bucket_blkno, bucket_buf, bstrategy); else LockBuffer(bucket_buf, BUFFER_LOCK_UNLOCK); }
/* * Prune and repair fragmentation in the specified page. * * Caller must have pin and buffer cleanup lock on the page. * * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). * * If report_stats is true then we send the number of reclaimed heap-only * tuples to pgstats. (This must be FALSE during vacuum, since vacuum will * send its own new total to pgstats, and we don't want this delta applied * on top of that.) * * Returns the number of tuples deleted from the page and sets * latestRemovedXid. */ int heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, bool report_stats, TransactionId *latestRemovedXid) { int ndeleted = 0; Page page = BufferGetPage(buffer); OffsetNumber offnum, maxoff; PruneState prstate; /* * Our strategy is to scan the page and make lists of items to change, * then apply the changes within a critical section. This keeps as much * logic as possible out of the critical section, and also ensures that * WAL replay will work the same as the normal case. * * First, initialize the new pd_prune_xid value to zero (indicating no * prunable tuples). If we find any tuples which may soon become * prunable, we will save the lowest relevant XID in new_prune_xid. Also * initialize the rest of our working state. */ prstate.new_prune_xid = InvalidTransactionId; prstate.latestRemovedXid = *latestRemovedXid; prstate.nredirected = prstate.ndead = prstate.nunused = 0; memset(prstate.marked, 0, sizeof(prstate.marked)); /* Scan the page */ maxoff = PageGetMaxOffsetNumber(page); for (offnum = FirstOffsetNumber; offnum <= maxoff; offnum = OffsetNumberNext(offnum)) { ItemId itemid; /* Ignore items already processed as part of an earlier chain */ if (prstate.marked[offnum]) continue; /* Nothing to do if slot is empty or already dead */ itemid = PageGetItemId(page, offnum); if (!ItemIdIsUsed(itemid) || ItemIdIsDead(itemid)) continue; /* Process this item or chain of items */ ndeleted += heap_prune_chain(relation, buffer, offnum, OldestXmin, &prstate); } /* Any error while applying the changes is critical */ START_CRIT_SECTION(); /* Have we found any prunable items? */ if (prstate.nredirected > 0 || prstate.ndead > 0 || prstate.nunused > 0) { /* * Apply the planned item changes, then repair page fragmentation, and * update the page's hint bit about whether it has free line pointers. */ heap_page_prune_execute(buffer, prstate.redirected, prstate.nredirected, prstate.nowdead, prstate.ndead, prstate.nowunused, prstate.nunused); /* * Update the page's pd_prune_xid field to either zero, or the lowest * XID of any soon-prunable tuple. */ ((PageHeader) page)->pd_prune_xid = prstate.new_prune_xid; /* * Also clear the "page is full" flag, since there's no point in * repeating the prune/defrag process until something else happens to * the page. */ PageClearFull(page); MarkBufferDirty(buffer); /* * Emit a WAL HEAP_CLEAN record showing what we did */ if (RelationNeedsWAL(relation)) { XLogRecPtr recptr; recptr = log_heap_clean(relation, buffer, prstate.redirected, prstate.nredirected, prstate.nowdead, prstate.ndead, prstate.nowunused, prstate.nunused, prstate.latestRemovedXid); PageSetLSN(BufferGetPage(buffer), recptr); } } else { /* * If we didn't prune anything, but have found a new value for the * pd_prune_xid field, update it and mark the buffer dirty. This is * treated as a non-WAL-logged hint. * * Also clear the "page is full" flag if it is set, since there's no * point in repeating the prune/defrag process until something else * happens to the page. */ if (((PageHeader) page)->pd_prune_xid != prstate.new_prune_xid || PageIsFull(page)) { ((PageHeader) page)->pd_prune_xid = prstate.new_prune_xid; PageClearFull(page); MarkBufferDirtyHint(buffer, true); } } END_CRIT_SECTION(); /* * If requested, report the number of tuples reclaimed to pgstats. This is * ndeleted minus ndead, because we don't want to count a now-DEAD root * item as a deletion for this purpose. */ if (report_stats && ndeleted > prstate.ndead) pgstat_update_heap_dead_tuples(relation, ndeleted - prstate.ndead); *latestRemovedXid = prstate.latestRemovedXid; /* * XXX Should we update the FSM information of this page ? * * There are two schools of thought here. We may not want to update FSM * information so that the page is not used for unrelated UPDATEs/INSERTs * and any free space in this page will remain available for further * UPDATEs in *this* page, thus improving chances for doing HOT updates. * * But for a large table and where a page does not receive further UPDATEs * for a long time, we might waste this space by not updating the FSM * information. The relation may get extended and fragmented further. * * One possibility is to leave "fillfactor" worth of space in this page * and update FSM with the remaining space. */ return ndeleted; }
/* * Traverse the tree to find path from root page to specified "child" block. * * returns a new insertion stack, starting from the parent of "child", up * to the root. *downlinkoffnum is set to the offset of the downlink in the * direct parent of child. * * To prevent deadlocks, this should lock only one page at a time. */ static GISTInsertStack * gistFindPath(Relation r, BlockNumber child, OffsetNumber *downlinkoffnum) { Page page; Buffer buffer; OffsetNumber i, maxoff; ItemId iid; IndexTuple idxtuple; List *fifo; GISTInsertStack *top, *ptr; BlockNumber blkno; top = (GISTInsertStack *) palloc0(sizeof(GISTInsertStack)); top->blkno = GIST_ROOT_BLKNO; top->downlinkoffnum = InvalidOffsetNumber; fifo = list_make1(top); while (fifo != NIL) { /* Get next page to visit */ top = linitial(fifo); fifo = list_delete_first(fifo); buffer = ReadBuffer(r, top->blkno); LockBuffer(buffer, GIST_SHARE); gistcheckpage(r, buffer); page = (Page) BufferGetPage(buffer); if (GistPageIsLeaf(page)) { /* * Because we scan the index top-down, all the rest of the pages * in the queue must be leaf pages as well. */ UnlockReleaseBuffer(buffer); break; } top->lsn = PageGetLSN(page); /* * If F_FOLLOW_RIGHT is set, the page to the right doesn't have a * downlink. This should not normally happen.. */ if (GistFollowRight(page)) elog(ERROR, "concurrent GiST page split was incomplete"); if (top->parent && top->parent->lsn < GistPageGetNSN(page) && GistPageGetOpaque(page)->rightlink != InvalidBlockNumber /* sanity check */ ) { /* * Page was split while we looked elsewhere. We didn't see the * downlink to the right page when we scanned the parent, so add * it to the queue now. * * Put the right page ahead of the queue, so that we visit it * next. That's important, because if this is the lowest internal * level, just above leaves, we might already have queued up some * leaf pages, and we assume that there can't be any non-leaf * pages behind leaf pages. */ ptr = (GISTInsertStack *) palloc0(sizeof(GISTInsertStack)); ptr->blkno = GistPageGetOpaque(page)->rightlink; ptr->downlinkoffnum = InvalidOffsetNumber; ptr->parent = top->parent; fifo = lcons(ptr, fifo); } maxoff = PageGetMaxOffsetNumber(page); for (i = FirstOffsetNumber; i <= maxoff; i = OffsetNumberNext(i)) { iid = PageGetItemId(page, i); idxtuple = (IndexTuple) PageGetItem(page, iid); blkno = ItemPointerGetBlockNumber(&(idxtuple->t_tid)); if (blkno == child) { /* Found it! */ UnlockReleaseBuffer(buffer); *downlinkoffnum = i; return top; } else { /* Append this child to the list of pages to visit later */ ptr = (GISTInsertStack *) palloc0(sizeof(GISTInsertStack)); ptr->blkno = blkno; ptr->downlinkoffnum = i; ptr->parent = top; fifo = lappend(fifo, ptr); } } UnlockReleaseBuffer(buffer); } elog(ERROR, "failed to re-find parent of a page in index \"%s\", block %u", RelationGetRelationName(r), child); return NULL; /* keep compiler quiet */ }
/* * hashgettuple() -- Get the next tuple in the scan. */ bool hashgettuple(IndexScanDesc scan, ScanDirection dir) { HashScanOpaque so = (HashScanOpaque) scan->opaque; Relation rel = scan->indexRelation; Buffer buf; Page page; OffsetNumber offnum; ItemPointer current; bool res; /* Hash indexes are always lossy since we store only the hash code */ scan->xs_recheck = true; /* * We hold pin but not lock on current buffer while outside the hash AM. * Reacquire the read lock here. */ if (BufferIsValid(so->hashso_curbuf)) LockBuffer(so->hashso_curbuf, BUFFER_LOCK_SHARE); /* * If we've already initialized this scan, we can just advance it in the * appropriate direction. If we haven't done so yet, we call a routine to * get the first item in the scan. */ current = &(so->hashso_curpos); if (ItemPointerIsValid(current)) { /* * An insertion into the current index page could have happened while * we didn't have read lock on it. Re-find our position by looking * for the TID we previously returned. (Because we hold a pin on the * primary bucket page, no deletions or splits could have occurred; * therefore we can expect that the TID still exists in the current * index page, at an offset >= where we were.) */ OffsetNumber maxoffnum; buf = so->hashso_curbuf; Assert(BufferIsValid(buf)); page = BufferGetPage(buf); /* * We don't need test for old snapshot here as the current buffer is * pinned, so vacuum can't clean the page. */ maxoffnum = PageGetMaxOffsetNumber(page); for (offnum = ItemPointerGetOffsetNumber(current); offnum <= maxoffnum; offnum = OffsetNumberNext(offnum)) { IndexTuple itup; itup = (IndexTuple) PageGetItem(page, PageGetItemId(page, offnum)); if (ItemPointerEquals(&(so->hashso_heappos), &(itup->t_tid))) break; } if (offnum > maxoffnum) elog(ERROR, "failed to re-find scan position within index \"%s\"", RelationGetRelationName(rel)); ItemPointerSetOffsetNumber(current, offnum); /* * Check to see if we should kill the previously-fetched tuple. */ if (scan->kill_prior_tuple) { /* * Yes, so remember it for later. (We'll deal with all such * tuples at once right after leaving the index page or at * end of scan.) In case if caller reverses the indexscan * direction it is quite possible that the same item might * get entered multiple times. But, we don't detect that; * instead, we just forget any excess entries. */ if (so->killedItems == NULL) so->killedItems = palloc(MaxIndexTuplesPerPage * sizeof(HashScanPosItem)); if (so->numKilled < MaxIndexTuplesPerPage) { so->killedItems[so->numKilled].heapTid = so->hashso_heappos; so->killedItems[so->numKilled].indexOffset = ItemPointerGetOffsetNumber(&(so->hashso_curpos)); so->numKilled++; } } /* * Now continue the scan. */ res = _hash_next(scan, dir); } else res = _hash_first(scan, dir); /* * Skip killed tuples if asked to. */ if (scan->ignore_killed_tuples) { while (res) { offnum = ItemPointerGetOffsetNumber(current); page = BufferGetPage(so->hashso_curbuf); if (!ItemIdIsDead(PageGetItemId(page, offnum))) break; res = _hash_next(scan, dir); } } /* Release read lock on current buffer, but keep it pinned */ if (BufferIsValid(so->hashso_curbuf)) LockBuffer(so->hashso_curbuf, BUFFER_LOCK_UNLOCK); /* Return current heap TID on success */ scan->xs_ctup.t_self = so->hashso_heappos; return res; }
/* * _hash_finish_split() -- Finish the previously interrupted split operation * * To complete the split operation, we form the hash table of TIDs in new * bucket which is then used by split operation to skip tuples that are * already moved before the split operation was previously interrupted. * * The caller must hold a pin, but no lock, on the metapage and old bucket's * primary page buffer. The buffers are returned in the same state. (The * metapage is only touched if it becomes necessary to add or remove overflow * pages.) */ void _hash_finish_split(Relation rel, Buffer metabuf, Buffer obuf, Bucket obucket, uint32 maxbucket, uint32 highmask, uint32 lowmask) { HASHCTL hash_ctl; HTAB *tidhtab; Buffer bucket_nbuf = InvalidBuffer; Buffer nbuf; Page npage; BlockNumber nblkno; BlockNumber bucket_nblkno; HashPageOpaque npageopaque; Bucket nbucket; bool found; /* Initialize hash tables used to track TIDs */ memset(&hash_ctl, 0, sizeof(hash_ctl)); hash_ctl.keysize = sizeof(ItemPointerData); hash_ctl.entrysize = sizeof(ItemPointerData); hash_ctl.hcxt = CurrentMemoryContext; tidhtab = hash_create("bucket ctids", 256, /* arbitrary initial size */ &hash_ctl, HASH_ELEM | HASH_BLOBS | HASH_CONTEXT); bucket_nblkno = nblkno = _hash_get_newblock_from_oldbucket(rel, obucket); /* * Scan the new bucket and build hash table of TIDs */ for (;;) { OffsetNumber noffnum; OffsetNumber nmaxoffnum; nbuf = _hash_getbuf(rel, nblkno, HASH_READ, LH_BUCKET_PAGE | LH_OVERFLOW_PAGE); /* remember the primary bucket buffer to acquire cleanup lock on it. */ if (nblkno == bucket_nblkno) bucket_nbuf = nbuf; npage = BufferGetPage(nbuf); npageopaque = (HashPageOpaque) PageGetSpecialPointer(npage); /* Scan each tuple in new page */ nmaxoffnum = PageGetMaxOffsetNumber(npage); for (noffnum = FirstOffsetNumber; noffnum <= nmaxoffnum; noffnum = OffsetNumberNext(noffnum)) { IndexTuple itup; /* Fetch the item's TID and insert it in hash table. */ itup = (IndexTuple) PageGetItem(npage, PageGetItemId(npage, noffnum)); (void) hash_search(tidhtab, &itup->t_tid, HASH_ENTER, &found); Assert(!found); } nblkno = npageopaque->hasho_nextblkno; /* * release our write lock without modifying buffer and ensure to * retain the pin on primary bucket. */ if (nbuf == bucket_nbuf) LockBuffer(nbuf, BUFFER_LOCK_UNLOCK); else _hash_relbuf(rel, nbuf); /* Exit loop if no more overflow pages in new bucket */ if (!BlockNumberIsValid(nblkno)) break; } /* * Conditionally get the cleanup lock on old and new buckets to perform * the split operation. If we don't get the cleanup locks, silently give * up and next insertion on old bucket will try again to complete the * split. */ if (!ConditionalLockBufferForCleanup(obuf)) { hash_destroy(tidhtab); return; } if (!ConditionalLockBufferForCleanup(bucket_nbuf)) { LockBuffer(obuf, BUFFER_LOCK_UNLOCK); hash_destroy(tidhtab); return; } npage = BufferGetPage(bucket_nbuf); npageopaque = (HashPageOpaque) PageGetSpecialPointer(npage); nbucket = npageopaque->hasho_bucket; _hash_splitbucket(rel, metabuf, obucket, nbucket, obuf, bucket_nbuf, tidhtab, maxbucket, highmask, lowmask); _hash_dropbuf(rel, bucket_nbuf); hash_destroy(tidhtab); }
/* * _hash_splitbucket -- split 'obucket' into 'obucket' and 'nbucket' * * This routine is used to partition the tuples between old and new bucket and * is used to finish the incomplete split operations. To finish the previously * interrupted split operation, the caller needs to fill htab. If htab is set, * then we skip the movement of tuples that exists in htab, otherwise NULL * value of htab indicates movement of all the tuples that belong to the new * bucket. * * We are splitting a bucket that consists of a base bucket page and zero * or more overflow (bucket chain) pages. We must relocate tuples that * belong in the new bucket. * * The caller must hold cleanup locks on both buckets to ensure that * no one else is trying to access them (see README). * * The caller must hold a pin, but no lock, on the metapage buffer. * The buffer is returned in the same state. (The metapage is only * touched if it becomes necessary to add or remove overflow pages.) * * Split needs to retain pin on primary bucket pages of both old and new * buckets till end of operation. This is to prevent vacuum from starting * while a split is in progress. * * In addition, the caller must have created the new bucket's base page, * which is passed in buffer nbuf, pinned and write-locked. The lock will be * released here and pin must be released by the caller. (The API is set up * this way because we must do _hash_getnewbuf() before releasing the metapage * write lock. So instead of passing the new bucket's start block number, we * pass an actual buffer.) */ static void _hash_splitbucket(Relation rel, Buffer metabuf, Bucket obucket, Bucket nbucket, Buffer obuf, Buffer nbuf, HTAB *htab, uint32 maxbucket, uint32 highmask, uint32 lowmask) { Buffer bucket_obuf; Buffer bucket_nbuf; Page opage; Page npage; HashPageOpaque oopaque; HashPageOpaque nopaque; OffsetNumber itup_offsets[MaxIndexTuplesPerPage]; IndexTuple itups[MaxIndexTuplesPerPage]; Size all_tups_size = 0; int i; uint16 nitups = 0; bucket_obuf = obuf; opage = BufferGetPage(obuf); oopaque = (HashPageOpaque) PageGetSpecialPointer(opage); bucket_nbuf = nbuf; npage = BufferGetPage(nbuf); nopaque = (HashPageOpaque) PageGetSpecialPointer(npage); /* * Partition the tuples in the old bucket between the old bucket and the * new bucket, advancing along the old bucket's overflow bucket chain and * adding overflow pages to the new bucket as needed. Outer loop iterates * once per page in old bucket. */ for (;;) { BlockNumber oblkno; OffsetNumber ooffnum; OffsetNumber omaxoffnum; /* Scan each tuple in old page */ omaxoffnum = PageGetMaxOffsetNumber(opage); for (ooffnum = FirstOffsetNumber; ooffnum <= omaxoffnum; ooffnum = OffsetNumberNext(ooffnum)) { IndexTuple itup; Size itemsz; Bucket bucket; bool found = false; /* skip dead tuples */ if (ItemIdIsDead(PageGetItemId(opage, ooffnum))) continue; /* * Before inserting a tuple, probe the hash table containing TIDs * of tuples belonging to new bucket, if we find a match, then * skip that tuple, else fetch the item's hash key (conveniently * stored in the item) and determine which bucket it now belongs * in. */ itup = (IndexTuple) PageGetItem(opage, PageGetItemId(opage, ooffnum)); if (htab) (void) hash_search(htab, &itup->t_tid, HASH_FIND, &found); if (found) continue; bucket = _hash_hashkey2bucket(_hash_get_indextuple_hashkey(itup), maxbucket, highmask, lowmask); if (bucket == nbucket) { IndexTuple new_itup; /* * make a copy of index tuple as we have to scribble on it. */ new_itup = CopyIndexTuple(itup); /* * mark the index tuple as moved by split, such tuples are * skipped by scan if there is split in progress for a bucket. */ new_itup->t_info |= INDEX_MOVED_BY_SPLIT_MASK; /* * insert the tuple into the new bucket. if it doesn't fit on * the current page in the new bucket, we must allocate a new * overflow page and place the tuple on that page instead. */ itemsz = IndexTupleDSize(*new_itup); itemsz = MAXALIGN(itemsz); if (PageGetFreeSpaceForMultipleTuples(npage, nitups + 1) < (all_tups_size + itemsz)) { /* * Change the shared buffer state in critical section, * otherwise any error could make it unrecoverable. */ START_CRIT_SECTION(); _hash_pgaddmultitup(rel, nbuf, itups, itup_offsets, nitups); MarkBufferDirty(nbuf); /* log the split operation before releasing the lock */ log_split_page(rel, nbuf); END_CRIT_SECTION(); /* drop lock, but keep pin */ LockBuffer(nbuf, BUFFER_LOCK_UNLOCK); /* be tidy */ for (i = 0; i < nitups; i++) pfree(itups[i]); nitups = 0; all_tups_size = 0; /* chain to a new overflow page */ nbuf = _hash_addovflpage(rel, metabuf, nbuf, (nbuf == bucket_nbuf) ? true : false); npage = BufferGetPage(nbuf); nopaque = (HashPageOpaque) PageGetSpecialPointer(npage); } itups[nitups++] = new_itup; all_tups_size += itemsz; } else { /* * the tuple stays on this page, so nothing to do. */ Assert(bucket == obucket); } } oblkno = oopaque->hasho_nextblkno; /* retain the pin on the old primary bucket */ if (obuf == bucket_obuf) LockBuffer(obuf, BUFFER_LOCK_UNLOCK); else _hash_relbuf(rel, obuf); /* Exit loop if no more overflow pages in old bucket */ if (!BlockNumberIsValid(oblkno)) { /* * Change the shared buffer state in critical section, otherwise * any error could make it unrecoverable. */ START_CRIT_SECTION(); _hash_pgaddmultitup(rel, nbuf, itups, itup_offsets, nitups); MarkBufferDirty(nbuf); /* log the split operation before releasing the lock */ log_split_page(rel, nbuf); END_CRIT_SECTION(); if (nbuf == bucket_nbuf) LockBuffer(nbuf, BUFFER_LOCK_UNLOCK); else _hash_relbuf(rel, nbuf); /* be tidy */ for (i = 0; i < nitups; i++) pfree(itups[i]); break; } /* Else, advance to next old page */ obuf = _hash_getbuf(rel, oblkno, HASH_READ, LH_OVERFLOW_PAGE); opage = BufferGetPage(obuf); oopaque = (HashPageOpaque) PageGetSpecialPointer(opage); } /* * We're at the end of the old bucket chain, so we're done partitioning * the tuples. Mark the old and new buckets to indicate split is * finished. * * To avoid deadlocks due to locking order of buckets, first lock the old * bucket and then the new bucket. */ LockBuffer(bucket_obuf, BUFFER_LOCK_EXCLUSIVE); opage = BufferGetPage(bucket_obuf); oopaque = (HashPageOpaque) PageGetSpecialPointer(opage); LockBuffer(bucket_nbuf, BUFFER_LOCK_EXCLUSIVE); npage = BufferGetPage(bucket_nbuf); nopaque = (HashPageOpaque) PageGetSpecialPointer(npage); START_CRIT_SECTION(); oopaque->hasho_flag &= ~LH_BUCKET_BEING_SPLIT; nopaque->hasho_flag &= ~LH_BUCKET_BEING_POPULATED; /* * After the split is finished, mark the old bucket to indicate that it * contains deletable tuples. We will clear split-cleanup flag after * deleting such tuples either at the end of split or at the next split * from old bucket or at the time of vacuum. */ oopaque->hasho_flag |= LH_BUCKET_NEEDS_SPLIT_CLEANUP; /* * now write the buffers, here we don't release the locks as caller is * responsible to release locks. */ MarkBufferDirty(bucket_obuf); MarkBufferDirty(bucket_nbuf); if (RelationNeedsWAL(rel)) { XLogRecPtr recptr; xl_hash_split_complete xlrec; xlrec.old_bucket_flag = oopaque->hasho_flag; xlrec.new_bucket_flag = nopaque->hasho_flag; XLogBeginInsert(); XLogRegisterData((char *) &xlrec, SizeOfHashSplitComplete); XLogRegisterBuffer(0, bucket_obuf, REGBUF_STANDARD); XLogRegisterBuffer(1, bucket_nbuf, REGBUF_STANDARD); recptr = XLogInsert(RM_HASH_ID, XLOG_HASH_SPLIT_COMPLETE); PageSetLSN(BufferGetPage(bucket_obuf), recptr); PageSetLSN(BufferGetPage(bucket_nbuf), recptr); } END_CRIT_SECTION(); /* * If possible, clean up the old bucket. We might not be able to do this * if someone else has a pin on it, but if not then we can go ahead. This * isn't absolutely necessary, but it reduces bloat; if we don't do it * now, VACUUM will do it eventually, but maybe not until new overflow * pages have been allocated. Note that there's no need to clean up the * new bucket. */ if (IsBufferCleanupOK(bucket_obuf)) { LockBuffer(bucket_nbuf, BUFFER_LOCK_UNLOCK); hashbucketcleanup(rel, obucket, bucket_obuf, BufferGetBlockNumber(bucket_obuf), NULL, maxbucket, highmask, lowmask, NULL, NULL, true, NULL, NULL); } else { LockBuffer(bucket_nbuf, BUFFER_LOCK_UNLOCK); LockBuffer(bucket_obuf, BUFFER_LOCK_UNLOCK); } }
static void bitmap_xlog_insert_lovitem(bool redo, XLogRecPtr lsn, XLogRecord* record) { xl_bm_lovitem *xlrec = (xl_bm_lovitem*) XLogRecGetData(record); Relation reln; reln = XLogOpenRelation(xlrec->bm_node); if (!RelationIsValid(reln)) return; if (redo) { Buffer lovBuffer; Page lovPage; #ifdef BM_DEBUG ereport(LOG, (errcode(LOG), errmsg("call bitmap_xlog_insert_lovitem: redo=%d, blkno=%d\n", redo, xlrec->bm_lov_blkno))); #endif lovBuffer = XLogReadBuffer(false, reln, xlrec->bm_lov_blkno); if (!BufferIsValid(lovBuffer)) elog(PANIC, "bm_insert_redo: block unfound: %d", xlrec->bm_lov_blkno); lovPage = BufferGetPage(lovBuffer); if (XLByteLT(PageGetLSN(lovPage), lsn)) { if(xlrec->bm_isNewItem) { OffsetNumber newOffset, itemSize; newOffset = OffsetNumberNext(PageGetMaxOffsetNumber(lovPage)); if (newOffset != xlrec->bm_lov_offset) elog(PANIC, "bm_insert_redo: LOV item is not inserted in pos %d(requested %d)", newOffset, xlrec->bm_lov_offset); itemSize = sizeof(BMLOVItemData); if (itemSize > PageGetFreeSpace(lovPage)) elog(PANIC, "bm_insert_redo: not enough space in LOV page %d", xlrec->bm_lov_blkno); if (PageAddItem(lovPage, (Item)&(xlrec->bm_lovItem), itemSize, newOffset, LP_USED) == InvalidOffsetNumber) ereport(ERROR, (errcode(ERRCODE_INTERNAL_ERROR), errmsg("failed to add LOV item to \"%s\"", RelationGetRelationName(reln)))); } else{ BMLOVItem oldLovItem; oldLovItem = (BMLOVItem) PageGetItem(lovPage, PageGetItemId(lovPage, xlrec->bm_lov_offset)); memcpy(oldLovItem, &(xlrec->bm_lovItem), sizeof(BMLOVItemData)); } PageSetLSN(lovPage, lsn); PageSetTLI(lovPage, ThisTimeLineID); _bitmap_wrtbuf(lovBuffer); } else { _bitmap_relbuf(lovBuffer); } } else elog(PANIC, "bm_insert_undo: not implemented."); }
/* * Bulk deletion of all index entries pointing to a set of heap tuples. * The set of target tuples is specified via a callback routine that tells * whether any given heap tuple (identified by ItemPointer) is being deleted. * * Result: a palloc'd struct containing statistical info for VACUUM displays. */ Datum hashbulkdelete(PG_FUNCTION_ARGS) { IndexVacuumInfo *info = (IndexVacuumInfo *) PG_GETARG_POINTER(0); IndexBulkDeleteResult *stats = (IndexBulkDeleteResult *) PG_GETARG_POINTER(1); IndexBulkDeleteCallback callback = (IndexBulkDeleteCallback) PG_GETARG_POINTER(2); void *callback_state = (void *) PG_GETARG_POINTER(3); Relation rel = info->index; double tuples_removed; double num_index_tuples; double orig_ntuples; Bucket orig_maxbucket; Bucket cur_maxbucket; Bucket cur_bucket; Buffer metabuf; HashMetaPage metap; HashMetaPageData local_metapage; tuples_removed = 0; num_index_tuples = 0; /* * Read the metapage to fetch original bucket and tuple counts. Also, we * keep a copy of the last-seen metapage so that we can use its * hashm_spares[] values to compute bucket page addresses. This is a bit * hokey but perfectly safe, since the interesting entries in the spares * array cannot change under us; and it beats rereading the metapage for * each bucket. */ metabuf = _hash_getbuf(rel, HASH_METAPAGE, HASH_READ, LH_META_PAGE); metap = HashPageGetMeta(BufferGetPage(metabuf)); orig_maxbucket = metap->hashm_maxbucket; orig_ntuples = metap->hashm_ntuples; memcpy(&local_metapage, metap, sizeof(local_metapage)); _hash_relbuf(rel, metabuf); /* Scan the buckets that we know exist */ cur_bucket = 0; cur_maxbucket = orig_maxbucket; loop_top: while (cur_bucket <= cur_maxbucket) { BlockNumber bucket_blkno; BlockNumber blkno; bool bucket_dirty = false; /* Get address of bucket's start page */ bucket_blkno = BUCKET_TO_BLKNO(&local_metapage, cur_bucket); /* Exclusive-lock the bucket so we can shrink it */ _hash_getlock(rel, bucket_blkno, HASH_EXCLUSIVE); /* Shouldn't have any active scans locally, either */ if (_hash_has_active_scan(rel, cur_bucket)) elog(ERROR, "hash index has active scan during VACUUM"); /* Scan each page in bucket */ blkno = bucket_blkno; while (BlockNumberIsValid(blkno)) { Buffer buf; Page page; HashPageOpaque opaque; OffsetNumber offno; OffsetNumber maxoffno; OffsetNumber deletable[MaxOffsetNumber]; int ndeletable = 0; vacuum_delay_point(); buf = _hash_getbuf_with_strategy(rel, blkno, HASH_WRITE, LH_BUCKET_PAGE | LH_OVERFLOW_PAGE, info->strategy); page = BufferGetPage(buf); opaque = (HashPageOpaque) PageGetSpecialPointer(page); Assert(opaque->hasho_bucket == cur_bucket); /* Scan each tuple in page */ maxoffno = PageGetMaxOffsetNumber(page); for (offno = FirstOffsetNumber; offno <= maxoffno; offno = OffsetNumberNext(offno)) { IndexTuple itup; ItemPointer htup; itup = (IndexTuple) PageGetItem(page, PageGetItemId(page, offno)); htup = &(itup->t_tid); if (callback(htup, callback_state)) { /* mark the item for deletion */ deletable[ndeletable++] = offno; tuples_removed += 1; } else num_index_tuples += 1; } /* * Apply deletions and write page if needed, advance to next page. */ blkno = opaque->hasho_nextblkno; if (ndeletable > 0) { PageIndexMultiDelete(page, deletable, ndeletable); _hash_wrtbuf(rel, buf); bucket_dirty = true; } else _hash_relbuf(rel, buf); } /* If we deleted anything, try to compact free space */ if (bucket_dirty) _hash_squeezebucket(rel, cur_bucket, bucket_blkno, info->strategy); /* Release bucket lock */ _hash_droplock(rel, bucket_blkno, HASH_EXCLUSIVE); /* Advance to next bucket */ cur_bucket++; } /* Write-lock metapage and check for split since we started */ metabuf = _hash_getbuf(rel, HASH_METAPAGE, HASH_WRITE, LH_META_PAGE); metap = HashPageGetMeta(BufferGetPage(metabuf)); if (cur_maxbucket != metap->hashm_maxbucket) { /* There's been a split, so process the additional bucket(s) */ cur_maxbucket = metap->hashm_maxbucket; memcpy(&local_metapage, metap, sizeof(local_metapage)); _hash_relbuf(rel, metabuf); goto loop_top; } /* Okay, we're really done. Update tuple count in metapage. */ if (orig_maxbucket == metap->hashm_maxbucket && orig_ntuples == metap->hashm_ntuples) { /* * No one has split or inserted anything since start of scan, so * believe our count as gospel. */ metap->hashm_ntuples = num_index_tuples; } else { /* * Otherwise, our count is untrustworthy since we may have * double-scanned tuples in split buckets. Proceed by dead-reckoning. * (Note: we still return estimated_count = false, because using this * count is better than not updating reltuples at all.) */ if (metap->hashm_ntuples > tuples_removed) metap->hashm_ntuples -= tuples_removed; else metap->hashm_ntuples = 0; num_index_tuples = metap->hashm_ntuples; } _hash_wrtbuf(rel, metabuf); /* return statistics */ if (stats == NULL) stats = (IndexBulkDeleteResult *) palloc0(sizeof(IndexBulkDeleteResult)); stats->estimated_count = false; stats->num_index_tuples = num_index_tuples; stats->tuples_removed += tuples_removed; /* hashvacuumcleanup will fill in num_pages */ PG_RETURN_POINTER(stats); }
/* * Prune specified item pointer or a HOT chain originating at that item. * * If the item is an index-referenced tuple (i.e. not a heap-only tuple), * the HOT chain is pruned by removing all DEAD tuples at the start of the HOT * chain. We also prune any RECENTLY_DEAD tuples preceding a DEAD tuple. * This is OK because a RECENTLY_DEAD tuple preceding a DEAD tuple is really * DEAD, the OldestXmin test is just too coarse to detect it. * * The root line pointer is redirected to the tuple immediately after the * latest DEAD tuple. If all tuples in the chain are DEAD, the root line * pointer is marked LP_DEAD. (This includes the case of a DEAD simple * tuple, which we treat as a chain of length 1.) * * OldestXmin is the cutoff XID used to identify dead tuples. * * We don't actually change the page here, except perhaps for hint-bit updates * caused by HeapTupleSatisfiesVacuum. We just add entries to the arrays in * prstate showing the changes to be made. Items to be redirected are added * to the redirected[] array (two entries per redirection); items to be set to * LP_DEAD state are added to nowdead[]; and items to be set to LP_UNUSED * state are added to nowunused[]. * * Returns the number of tuples (to be) deleted from the page. */ static int heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, TransactionId OldestXmin, PruneState *prstate) { int ndeleted = 0; Page dp = (Page) BufferGetPage(buffer); TransactionId priorXmax = InvalidTransactionId; ItemId rootlp; HeapTupleHeader htup; OffsetNumber latestdead = InvalidOffsetNumber, maxoff = PageGetMaxOffsetNumber(dp), offnum; OffsetNumber chainitems[MaxHeapTuplesPerPage]; int nchain = 0, i; HeapTupleData tup; tup.t_tableOid = RelationGetRelid(relation); rootlp = PageGetItemId(dp, rootoffnum); /* * If it's a heap-only tuple, then it is not the start of a HOT chain. */ if (ItemIdIsNormal(rootlp)) { htup = (HeapTupleHeader) PageGetItem(dp, rootlp); tup.t_data = htup; tup.t_len = ItemIdGetLength(rootlp); ItemPointerSet(&(tup.t_self), BufferGetBlockNumber(buffer), rootoffnum); if (HeapTupleHeaderIsHeapOnly(htup)) { /* * If the tuple is DEAD and doesn't chain to anything else, mark * it unused immediately. (If it does chain, we can only remove * it as part of pruning its chain.) * * We need this primarily to handle aborted HOT updates, that is, * XMIN_INVALID heap-only tuples. Those might not be linked to by * any chain, since the parent tuple might be re-updated before * any pruning occurs. So we have to be able to reap them * separately from chain-pruning. (Note that * HeapTupleHeaderIsHotUpdated will never return true for an * XMIN_INVALID tuple, so this code will work even when there were * sequential updates within the aborted transaction.) * * Note that we might first arrive at a dead heap-only tuple * either here or while following a chain below. Whichever path * gets there first will mark the tuple unused. */ if (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer) == HEAPTUPLE_DEAD && !HeapTupleHeaderIsHotUpdated(htup)) { heap_prune_record_unused(prstate, rootoffnum); HeapTupleHeaderAdvanceLatestRemovedXid(htup, &prstate->latestRemovedXid); ndeleted++; } /* Nothing more to do */ return ndeleted; } } /* Start from the root tuple */ offnum = rootoffnum; /* while not end of the chain */ for (;;) { ItemId lp; bool tupdead, recent_dead; /* Some sanity checks */ if (offnum < FirstOffsetNumber || offnum > maxoff) break; /* If item is already processed, stop --- it must not be same chain */ if (prstate->marked[offnum]) break; lp = PageGetItemId(dp, offnum); /* Unused item obviously isn't part of the chain */ if (!ItemIdIsUsed(lp)) break; /* * If we are looking at the redirected root line pointer, jump to the * first normal tuple in the chain. If we find a redirect somewhere * else, stop --- it must not be same chain. */ if (ItemIdIsRedirected(lp)) { if (nchain > 0) break; /* not at start of chain */ chainitems[nchain++] = offnum; offnum = ItemIdGetRedirect(rootlp); continue; } /* * Likewise, a dead item pointer can't be part of the chain. (We * already eliminated the case of dead root tuple outside this * function.) */ if (ItemIdIsDead(lp)) break; Assert(ItemIdIsNormal(lp)); htup = (HeapTupleHeader) PageGetItem(dp, lp); tup.t_data = htup; tup.t_len = ItemIdGetLength(lp); ItemPointerSet(&(tup.t_self), BufferGetBlockNumber(buffer), offnum); /* * Check the tuple XMIN against prior XMAX, if any */ if (TransactionIdIsValid(priorXmax) && !TransactionIdEquals(HeapTupleHeaderGetXmin(htup), priorXmax)) break; /* * OK, this tuple is indeed a member of the chain. */ chainitems[nchain++] = offnum; /* * Check tuple's visibility status. */ tupdead = recent_dead = false; switch (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer)) { case HEAPTUPLE_DEAD: tupdead = true; break; case HEAPTUPLE_RECENTLY_DEAD: recent_dead = true; /* * This tuple may soon become DEAD. Update the hint field so * that the page is reconsidered for pruning in future. */ heap_prune_record_prunable(prstate, HeapTupleHeaderGetUpdateXid(htup)); break; case HEAPTUPLE_DELETE_IN_PROGRESS: /* * This tuple may soon become DEAD. Update the hint field so * that the page is reconsidered for pruning in future. */ heap_prune_record_prunable(prstate, HeapTupleHeaderGetUpdateXid(htup)); break; case HEAPTUPLE_LIVE: case HEAPTUPLE_INSERT_IN_PROGRESS: /* * If we wanted to optimize for aborts, we might consider * marking the page prunable when we see INSERT_IN_PROGRESS. * But we don't. See related decisions about when to mark the * page prunable in heapam.c. */ break; default: elog(ERROR, "unexpected HeapTupleSatisfiesVacuum result"); break; } /* * Remember the last DEAD tuple seen. We will advance past * RECENTLY_DEAD tuples just in case there's a DEAD one after them; * but we can't advance past anything else. (XXX is it really worth * continuing to scan beyond RECENTLY_DEAD? The case where we will * find another DEAD tuple is a fairly unusual corner case.) */ if (tupdead) { latestdead = offnum; HeapTupleHeaderAdvanceLatestRemovedXid(htup, &prstate->latestRemovedXid); } else if (!recent_dead) break; /* * If the tuple is not HOT-updated, then we are at the end of this * HOT-update chain. */ if (!HeapTupleHeaderIsHotUpdated(htup)) break; /* * Advance to next chain member. */ Assert(ItemPointerGetBlockNumber(&htup->t_ctid) == BufferGetBlockNumber(buffer)); offnum = ItemPointerGetOffsetNumber(&htup->t_ctid); priorXmax = HeapTupleHeaderGetUpdateXid(htup); } /* * If we found a DEAD tuple in the chain, adjust the HOT chain so that all * the DEAD tuples at the start of the chain are removed and the root line * pointer is appropriately redirected. */ if (OffsetNumberIsValid(latestdead)) { /* * Mark as unused each intermediate item that we are able to remove * from the chain. * * When the previous item is the last dead tuple seen, we are at the * right candidate for redirection. */ for (i = 1; (i < nchain) && (chainitems[i - 1] != latestdead); i++) { heap_prune_record_unused(prstate, chainitems[i]); ndeleted++; } /* * If the root entry had been a normal tuple, we are deleting it, so * count it in the result. But changing a redirect (even to DEAD * state) doesn't count. */ if (ItemIdIsNormal(rootlp)) ndeleted++; /* * If the DEAD tuple is at the end of the chain, the entire chain is * dead and the root line pointer can be marked dead. Otherwise just * redirect the root to the correct chain member. */ if (i >= nchain) heap_prune_record_dead(prstate, rootoffnum); else heap_prune_record_redirect(prstate, rootoffnum, chainitems[i]); } else if (nchain < 2 && ItemIdIsRedirected(rootlp)) { /* * We found a redirect item that doesn't point to a valid follow-on * item. This can happen if the loop in heap_page_prune caused us to * visit the dead successor of a redirect item before visiting the * redirect item. We can clean up by setting the redirect item to * DEAD state. */ heap_prune_record_dead(prstate, rootoffnum); } return ndeleted; }
/* * _hash_squeezebucket(rel, bucket) * * Try to squeeze the tuples onto pages occurring earlier in the * bucket chain in an attempt to free overflow pages. When we start * the "squeezing", the page from which we start taking tuples (the * "read" page) is the last bucket in the bucket chain and the page * onto which we start squeezing tuples (the "write" page) is the * first page in the bucket chain. The read page works backward and * the write page works forward; the procedure terminates when the * read page and write page are the same page. * * At completion of this procedure, it is guaranteed that all pages in * the bucket are nonempty, unless the bucket is totally empty (in * which case all overflow pages will be freed). The original implementation * required that to be true on entry as well, but it's a lot easier for * callers to leave empty overflow pages and let this guy clean it up. * * Caller must acquire cleanup lock on the primary page of the target * bucket to exclude any scans that are in progress, which could easily * be confused into returning the same tuple more than once or some tuples * not at all by the rearrangement we are performing here. To prevent * any concurrent scan to cross the squeeze scan we use lock chaining * similar to hasbucketcleanup. Refer comments atop hashbucketcleanup. * * We need to retain a pin on the primary bucket to ensure that no concurrent * split can start. * * Since this function is invoked in VACUUM, we provide an access strategy * parameter that controls fetches of the bucket pages. */ void _hash_squeezebucket(Relation rel, Bucket bucket, BlockNumber bucket_blkno, Buffer bucket_buf, BufferAccessStrategy bstrategy) { BlockNumber wblkno; BlockNumber rblkno; Buffer wbuf; Buffer rbuf; Page wpage; Page rpage; HashPageOpaque wopaque; HashPageOpaque ropaque; /* * start squeezing into the primary bucket page. */ wblkno = bucket_blkno; wbuf = bucket_buf; wpage = BufferGetPage(wbuf); wopaque = (HashPageOpaque) PageGetSpecialPointer(wpage); /* * if there aren't any overflow pages, there's nothing to squeeze. caller * is responsible for releasing the pin on primary bucket page. */ if (!BlockNumberIsValid(wopaque->hasho_nextblkno)) { LockBuffer(wbuf, BUFFER_LOCK_UNLOCK); return; } /* * Find the last page in the bucket chain by starting at the base bucket * page and working forward. Note: we assume that a hash bucket chain is * usually smaller than the buffer ring being used by VACUUM, else using * the access strategy here would be counterproductive. */ rbuf = InvalidBuffer; ropaque = wopaque; do { rblkno = ropaque->hasho_nextblkno; if (rbuf != InvalidBuffer) _hash_relbuf(rel, rbuf); rbuf = _hash_getbuf_with_strategy(rel, rblkno, HASH_WRITE, LH_OVERFLOW_PAGE, bstrategy); rpage = BufferGetPage(rbuf); ropaque = (HashPageOpaque) PageGetSpecialPointer(rpage); Assert(ropaque->hasho_bucket == bucket); } while (BlockNumberIsValid(ropaque->hasho_nextblkno)); /* * squeeze the tuples. */ for (;;) { OffsetNumber roffnum; OffsetNumber maxroffnum; OffsetNumber deletable[MaxOffsetNumber]; IndexTuple itups[MaxIndexTuplesPerPage]; Size tups_size[MaxIndexTuplesPerPage]; OffsetNumber itup_offsets[MaxIndexTuplesPerPage]; uint16 ndeletable = 0; uint16 nitups = 0; Size all_tups_size = 0; int i; bool retain_pin = false; readpage: /* Scan each tuple in "read" page */ maxroffnum = PageGetMaxOffsetNumber(rpage); for (roffnum = FirstOffsetNumber; roffnum <= maxroffnum; roffnum = OffsetNumberNext(roffnum)) { IndexTuple itup; Size itemsz; /* skip dead tuples */ if (ItemIdIsDead(PageGetItemId(rpage, roffnum))) continue; itup = (IndexTuple) PageGetItem(rpage, PageGetItemId(rpage, roffnum)); itemsz = IndexTupleDSize(*itup); itemsz = MAXALIGN(itemsz); /* * Walk up the bucket chain, looking for a page big enough for * this item and all other accumulated items. Exit if we reach * the read page. */ while (PageGetFreeSpaceForMultipleTuples(wpage, nitups + 1) < (all_tups_size + itemsz)) { Buffer next_wbuf = InvalidBuffer; bool tups_moved = false; Assert(!PageIsEmpty(wpage)); if (wblkno == bucket_blkno) retain_pin = true; wblkno = wopaque->hasho_nextblkno; Assert(BlockNumberIsValid(wblkno)); /* don't need to move to next page if we reached the read page */ if (wblkno != rblkno) next_wbuf = _hash_getbuf_with_strategy(rel, wblkno, HASH_WRITE, LH_OVERFLOW_PAGE, bstrategy); if (nitups > 0) { Assert(nitups == ndeletable); /* * This operation needs to log multiple tuples, prepare * WAL for that. */ if (RelationNeedsWAL(rel)) XLogEnsureRecordSpace(0, 3 + nitups); START_CRIT_SECTION(); /* * we have to insert tuples on the "write" page, being * careful to preserve hashkey ordering. (If we insert * many tuples into the same "write" page it would be * worth qsort'ing them). */ _hash_pgaddmultitup(rel, wbuf, itups, itup_offsets, nitups); MarkBufferDirty(wbuf); /* Delete tuples we already moved off read page */ PageIndexMultiDelete(rpage, deletable, ndeletable); MarkBufferDirty(rbuf); /* XLOG stuff */ if (RelationNeedsWAL(rel)) { XLogRecPtr recptr; xl_hash_move_page_contents xlrec; xlrec.ntups = nitups; xlrec.is_prim_bucket_same_wrt = (wbuf == bucket_buf) ? true : false; XLogBeginInsert(); XLogRegisterData((char *) &xlrec, SizeOfHashMovePageContents); /* * bucket buffer needs to be registered to ensure that * we can acquire a cleanup lock on it during replay. */ if (!xlrec.is_prim_bucket_same_wrt) XLogRegisterBuffer(0, bucket_buf, REGBUF_STANDARD | REGBUF_NO_IMAGE); XLogRegisterBuffer(1, wbuf, REGBUF_STANDARD); XLogRegisterBufData(1, (char *) itup_offsets, nitups * sizeof(OffsetNumber)); for (i = 0; i < nitups; i++) XLogRegisterBufData(1, (char *) itups[i], tups_size[i]); XLogRegisterBuffer(2, rbuf, REGBUF_STANDARD); XLogRegisterBufData(2, (char *) deletable, ndeletable * sizeof(OffsetNumber)); recptr = XLogInsert(RM_HASH_ID, XLOG_HASH_MOVE_PAGE_CONTENTS); PageSetLSN(BufferGetPage(wbuf), recptr); PageSetLSN(BufferGetPage(rbuf), recptr); } END_CRIT_SECTION(); tups_moved = true; } /* * release the lock on previous page after acquiring the lock * on next page */ if (retain_pin) LockBuffer(wbuf, BUFFER_LOCK_UNLOCK); else _hash_relbuf(rel, wbuf); /* nothing more to do if we reached the read page */ if (rblkno == wblkno) { _hash_relbuf(rel, rbuf); return; } wbuf = next_wbuf; wpage = BufferGetPage(wbuf); wopaque = (HashPageOpaque) PageGetSpecialPointer(wpage); Assert(wopaque->hasho_bucket == bucket); retain_pin = false; /* be tidy */ for (i = 0; i < nitups; i++) pfree(itups[i]); nitups = 0; all_tups_size = 0; ndeletable = 0; /* * after moving the tuples, rpage would have been compacted, * so we need to rescan it. */ if (tups_moved) goto readpage; } /* remember tuple for deletion from "read" page */ deletable[ndeletable++] = roffnum; /* * we need a copy of index tuples as they can be freed as part of * overflow page, however we need them to write a WAL record in * _hash_freeovflpage. */ itups[nitups] = CopyIndexTuple(itup); tups_size[nitups++] = itemsz; all_tups_size += itemsz; } /* * If we reach here, there are no live tuples on the "read" page --- * it was empty when we got to it, or we moved them all. So we can * just free the page without bothering with deleting tuples * individually. Then advance to the previous "read" page. * * Tricky point here: if our read and write pages are adjacent in the * bucket chain, our write lock on wbuf will conflict with * _hash_freeovflpage's attempt to update the sibling links of the * removed page. In that case, we don't need to lock it again. */ rblkno = ropaque->hasho_prevblkno; Assert(BlockNumberIsValid(rblkno)); /* free this overflow page (releases rbuf) */ _hash_freeovflpage(rel, bucket_buf, rbuf, wbuf, itups, itup_offsets, tups_size, nitups, bstrategy); /* be tidy */ for (i = 0; i < nitups; i++) pfree(itups[i]); /* are we freeing the page adjacent to wbuf? */ if (rblkno == wblkno) { /* retain the pin on primary bucket page till end of bucket scan */ if (wblkno == bucket_blkno) LockBuffer(wbuf, BUFFER_LOCK_UNLOCK); else _hash_relbuf(rel, wbuf); return; } rbuf = _hash_getbuf_with_strategy(rel, rblkno, HASH_WRITE, LH_OVERFLOW_PAGE, bstrategy); rpage = BufferGetPage(rbuf); ropaque = (HashPageOpaque) PageGetSpecialPointer(rpage); Assert(ropaque->hasho_bucket == bucket); } /* NOTREACHED */ }
/* * For all items in this page, find their respective root line pointers. * If item k is part of a HOT-chain with root at item j, then we set * root_offsets[k - 1] = j. * * The passed-in root_offsets array must have MaxHeapTuplesPerPage entries. * We zero out all unused entries. * * The function must be called with at least share lock on the buffer, to * prevent concurrent prune operations. * * Note: The information collected here is valid only as long as the caller * holds a pin on the buffer. Once pin is released, a tuple might be pruned * and reused by a completely unrelated tuple. */ void heap_get_root_tuples(Page page, OffsetNumber *root_offsets) { OffsetNumber offnum, maxoff; MemSet(root_offsets, 0, MaxHeapTuplesPerPage * sizeof(OffsetNumber)); maxoff = PageGetMaxOffsetNumber(page); for (offnum = FirstOffsetNumber; offnum <= maxoff; offnum = OffsetNumberNext(offnum)) { ItemId lp = PageGetItemId(page, offnum); HeapTupleHeader htup; OffsetNumber nextoffnum; TransactionId priorXmax; /* skip unused and dead items */ if (!ItemIdIsUsed(lp) || ItemIdIsDead(lp)) continue; if (ItemIdIsNormal(lp)) { htup = (HeapTupleHeader) PageGetItem(page, lp); /* * Check if this tuple is part of a HOT-chain rooted at some other * tuple. If so, skip it for now; we'll process it when we find * its root. */ if (HeapTupleHeaderIsHeapOnly(htup)) continue; /* * This is either a plain tuple or the root of a HOT-chain. * Remember it in the mapping. */ root_offsets[offnum - 1] = offnum; /* If it's not the start of a HOT-chain, we're done with it */ if (!HeapTupleHeaderIsHotUpdated(htup)) continue; /* Set up to scan the HOT-chain */ nextoffnum = ItemPointerGetOffsetNumber(&htup->t_ctid); priorXmax = HeapTupleHeaderGetUpdateXid(htup); } else { /* Must be a redirect item. We do not set its root_offsets entry */ Assert(ItemIdIsRedirected(lp)); /* Set up to scan the HOT-chain */ nextoffnum = ItemIdGetRedirect(lp); priorXmax = InvalidTransactionId; } /* * Now follow the HOT-chain and collect other tuples in the chain. * * Note: Even though this is a nested loop, the complexity of the * function is O(N) because a tuple in the page should be visited not * more than twice, once in the outer loop and once in HOT-chain * chases. */ for (;;) { lp = PageGetItemId(page, nextoffnum); /* Check for broken chains */ if (!ItemIdIsNormal(lp)) break; htup = (HeapTupleHeader) PageGetItem(page, lp); if (TransactionIdIsValid(priorXmax) && !TransactionIdEquals(priorXmax, HeapTupleHeaderGetXmin(htup))) break; /* Remember the root line pointer for this item */ root_offsets[nextoffnum - 1] = offnum; /* Advance to next chain member, if any */ if (!HeapTupleHeaderIsHotUpdated(htup)) break; nextoffnum = ItemPointerGetOffsetNumber(&htup->t_ctid); priorXmax = HeapTupleHeaderGetUpdateXid(htup); } } }
/* * Traverse the tree to find path from root page to specified "child" block. * * returns from the beginning of closest parent; * * To prevent deadlocks, this should lock only one page simultaneously. */ GISTInsertStack * gistFindPath(Relation r, BlockNumber child) { Page page; Buffer buffer; OffsetNumber i, maxoff; ItemId iid; IndexTuple idxtuple; GISTInsertStack *top, *tail, *ptr; BlockNumber blkno; top = tail = (GISTInsertStack *) palloc0(sizeof(GISTInsertStack)); top->blkno = GIST_ROOT_BLKNO; while (top && top->blkno != child) { buffer = ReadBuffer(r, top->blkno); LockBuffer(buffer, GIST_SHARE); gistcheckpage(r, buffer); page = (Page) BufferGetPage(buffer); if (GistPageIsLeaf(page)) { /* we can safety go away, follows only leaf pages */ UnlockReleaseBuffer(buffer); return NULL; } top->lsn = PageGetLSN(page); if (top->parent && XLByteLT(top->parent->lsn, GistPageGetOpaque(page)->nsn) && GistPageGetOpaque(page)->rightlink != InvalidBlockNumber /* sanity check */ ) { /* page splited while we thinking of... */ ptr = (GISTInsertStack *) palloc0(sizeof(GISTInsertStack)); ptr->blkno = GistPageGetOpaque(page)->rightlink; ptr->childoffnum = InvalidOffsetNumber; ptr->parent = top; ptr->next = NULL; tail->next = ptr; tail = ptr; } maxoff = PageGetMaxOffsetNumber(page); for (i = FirstOffsetNumber; i <= maxoff; i = OffsetNumberNext(i)) { iid = PageGetItemId(page, i); idxtuple = (IndexTuple) PageGetItem(page, iid); blkno = ItemPointerGetBlockNumber(&(idxtuple->t_tid)); if (blkno == child) { OffsetNumber poff = InvalidOffsetNumber; /* make childs links */ ptr = top; while (ptr->parent) { /* set child link */ ptr->parent->child = ptr; /* move childoffnum.. */ if (ptr == top) { /* first iteration */ poff = ptr->parent->childoffnum; ptr->parent->childoffnum = ptr->childoffnum; } else { OffsetNumber tmp = ptr->parent->childoffnum; ptr->parent->childoffnum = poff; poff = tmp; } ptr = ptr->parent; } top->childoffnum = i; UnlockReleaseBuffer(buffer); return top; } else { /* Install next inner page to the end of stack */ ptr = (GISTInsertStack *) palloc0(sizeof(GISTInsertStack)); ptr->blkno = blkno; ptr->childoffnum = i; /* set offsetnumber of child to child * !!! */ ptr->parent = top; ptr->next = NULL; tail->next = ptr; tail = ptr; } } UnlockReleaseBuffer(buffer); top = top->next; } return NULL; }
/* * Find correct tuple in non-leaf page. It supposed that * page correctly chosen and searching value SHOULD be on page */ static BlockNumber entryLocateEntry(GinBtree btree, GinBtreeStack *stack) { OffsetNumber low, high, maxoff; IndexTuple itup = NULL; int result; Page page = BufferGetPage(stack->buffer); Assert(!GinPageIsLeaf(page)); Assert(!GinPageIsData(page)); if (btree->fullScan) { stack->off = FirstOffsetNumber; stack->predictNumber *= PageGetMaxOffsetNumber(page); return btree->getLeftMostChild(btree, page); } low = FirstOffsetNumber; maxoff = high = PageGetMaxOffsetNumber(page); Assert(high >= low); high++; while (high > low) { OffsetNumber mid = low + ((high - low) / 2); if (mid == maxoff && GinPageRightMost(page)) { /* Right infinity */ result = -1; } else { OffsetNumber attnum; Datum key; GinNullCategory category; itup = (IndexTuple) PageGetItem(page, PageGetItemId(page, mid)); attnum = gintuple_get_attrnum(btree->ginstate, itup); key = gintuple_get_key(btree->ginstate, itup, &category); result = ginCompareAttEntries(btree->ginstate, btree->entryAttnum, btree->entryKey, btree->entryCategory, attnum, key, category); } if (result == 0) { stack->off = mid; Assert(GinGetDownlink(itup) != GIN_ROOT_BLKNO); return GinGetDownlink(itup); } else if (result > 0) low = mid + 1; else high = mid; } Assert(high >= FirstOffsetNumber && high <= maxoff); stack->off = high; itup = (IndexTuple) PageGetItem(page, PageGetItemId(page, high)); Assert(GinGetDownlink(itup) != GIN_ROOT_BLKNO); return GinGetDownlink(itup); }
/* * This function takes an already open relation and scans its pages, * skipping those that have the corresponding visibility map bit set. * For pages we skip, we find the free space from the free space map * and approximate tuple_len on that basis. For the others, we count * the exact number of dead tuples etc. * * This scan is loosely based on vacuumlazy.c:lazy_scan_heap(), but * we do not try to avoid skipping single pages. */ static void statapprox_heap(Relation rel, output_type *stat) { BlockNumber scanned, nblocks, blkno; Buffer vmbuffer = InvalidBuffer; BufferAccessStrategy bstrategy; TransactionId OldestXmin; uint64 misc_count = 0; OldestXmin = GetOldestXmin(rel, true); bstrategy = GetAccessStrategy(BAS_BULKREAD); nblocks = RelationGetNumberOfBlocks(rel); scanned = 0; for (blkno = 0; blkno < nblocks; blkno++) { Buffer buf; Page page; OffsetNumber offnum, maxoff; Size freespace; CHECK_FOR_INTERRUPTS(); /* * If the page has only visible tuples, then we can find out the * free space from the FSM and move on. */ if (visibilitymap_test(rel, blkno, &vmbuffer)) { freespace = GetRecordedFreeSpace(rel, blkno); stat->tuple_len += BLCKSZ - freespace; stat->free_space += freespace; continue; } buf = ReadBufferExtended(rel, MAIN_FORKNUM, blkno, RBM_NORMAL, bstrategy); LockBuffer(buf, BUFFER_LOCK_SHARE); page = BufferGetPage(buf); /* * It's not safe to call PageGetHeapFreeSpace() on new pages, so * we treat them as being free space for our purposes. */ if (!PageIsNew(page)) stat->free_space += PageGetHeapFreeSpace(page); else stat->free_space += BLCKSZ - SizeOfPageHeaderData; if (PageIsNew(page) || PageIsEmpty(page)) { UnlockReleaseBuffer(buf); continue; } scanned++; /* * Look at each tuple on the page and decide whether it's live * or dead, then count it and its size. Unlike lazy_scan_heap, * we can afford to ignore problems and special cases. */ maxoff = PageGetMaxOffsetNumber(page); for (offnum = FirstOffsetNumber; offnum <= maxoff; offnum = OffsetNumberNext(offnum)) { ItemId itemid; HeapTupleData tuple; itemid = PageGetItemId(page, offnum); if (!ItemIdIsUsed(itemid) || ItemIdIsRedirected(itemid) || ItemIdIsDead(itemid)) { continue; } Assert(ItemIdIsNormal(itemid)); ItemPointerSet(&(tuple.t_self), blkno, offnum); tuple.t_data = (HeapTupleHeader) PageGetItem(page, itemid); tuple.t_len = ItemIdGetLength(itemid); tuple.t_tableOid = RelationGetRelid(rel); /* * We count live and dead tuples, but we also need to add up * others in order to feed vac_estimate_reltuples. */ switch (HeapTupleSatisfiesVacuum(&tuple, OldestXmin, buf)) { case HEAPTUPLE_RECENTLY_DEAD: misc_count++; /* Fall through */ case HEAPTUPLE_DEAD: stat->dead_tuple_len += tuple.t_len; stat->dead_tuple_count++; break; case HEAPTUPLE_LIVE: stat->tuple_len += tuple.t_len; stat->tuple_count++; break; case HEAPTUPLE_INSERT_IN_PROGRESS: case HEAPTUPLE_DELETE_IN_PROGRESS: misc_count++; break; default: elog(ERROR, "unexpected HeapTupleSatisfiesVacuum result"); break; } } UnlockReleaseBuffer(buf); } stat->table_len = (uint64) nblocks * BLCKSZ; stat->tuple_count = vac_estimate_reltuples(rel, false, nblocks, scanned, stat->tuple_count+misc_count); /* * Calculate percentages if the relation has one or more pages. */ if (nblocks != 0) { stat->scanned_percent = 100 * scanned / nblocks; stat->tuple_percent = 100.0 * stat->tuple_len / stat->table_len; stat->dead_tuple_percent = 100.0 * stat->dead_tuple_len / stat->table_len; stat->free_percent = 100.0 * stat->free_space / stat->table_len; } if (BufferIsValid(vmbuffer)) { ReleaseBuffer(vmbuffer); vmbuffer = InvalidBuffer; } }
/* * Split entry page and insert new data. * * Returns new temp pages to *newlpage and *newrpage. * The original buffer is left untouched. */ static void entrySplitPage(GinBtree btree, Buffer origbuf, GinBtreeStack *stack, GinBtreeEntryInsertData *insertData, BlockNumber updateblkno, Page *newlpage, Page *newrpage) { OffsetNumber off = stack->off; OffsetNumber i, maxoff, separator = InvalidOffsetNumber; Size totalsize = 0; Size lsize = 0, size; char *ptr; IndexTuple itup; Page page; Page lpage = PageGetTempPageCopy(BufferGetPage(origbuf)); Page rpage = PageGetTempPageCopy(BufferGetPage(origbuf)); Size pageSize = PageGetPageSize(lpage); char tupstore[2 * BLCKSZ]; entryPreparePage(btree, lpage, off, insertData, updateblkno); /* * First, append all the existing tuples and the new tuple we're inserting * one after another in a temporary workspace. */ maxoff = PageGetMaxOffsetNumber(lpage); ptr = tupstore; for (i = FirstOffsetNumber; i <= maxoff; i++) { if (i == off) { size = MAXALIGN(IndexTupleSize(insertData->entry)); memcpy(ptr, insertData->entry, size); ptr += size; totalsize += size + sizeof(ItemIdData); } itup = (IndexTuple) PageGetItem(lpage, PageGetItemId(lpage, i)); size = MAXALIGN(IndexTupleSize(itup)); memcpy(ptr, itup, size); ptr += size; totalsize += size + sizeof(ItemIdData); } if (off == maxoff + 1) { size = MAXALIGN(IndexTupleSize(insertData->entry)); memcpy(ptr, insertData->entry, size); ptr += size; totalsize += size + sizeof(ItemIdData); } /* * Initialize the left and right pages, and copy all the tuples back to * them. */ GinInitPage(rpage, GinPageGetOpaque(lpage)->flags, pageSize); GinInitPage(lpage, GinPageGetOpaque(rpage)->flags, pageSize); ptr = tupstore; maxoff++; lsize = 0; page = lpage; for (i = FirstOffsetNumber; i <= maxoff; i++) { itup = (IndexTuple) ptr; /* * Decide where to split. We try to equalize the pages' total data * size, not number of tuples. */ if (lsize > totalsize / 2) { if (separator == InvalidOffsetNumber) separator = i - 1; page = rpage; } else { lsize += MAXALIGN(IndexTupleSize(itup)) + sizeof(ItemIdData); } if (PageAddItem(page, (Item) itup, IndexTupleSize(itup), InvalidOffsetNumber, false, false) == InvalidOffsetNumber) elog(ERROR, "failed to add item to index page in \"%s\"", RelationGetRelationName(btree->index)); ptr += MAXALIGN(IndexTupleSize(itup)); } /* return temp pages to caller */ *newlpage = lpage; *newrpage = rpage; }
/* * _hash_splitbucket -- split 'obucket' into 'obucket' and 'nbucket' * * We are splitting a bucket that consists of a base bucket page and zero * or more overflow (bucket chain) pages. We must relocate tuples that * belong in the new bucket, and compress out any free space in the old * bucket. * * The caller must hold exclusive locks on both buckets to ensure that * no one else is trying to access them (see README). * * The caller must hold a pin, but no lock, on the metapage buffer. * The buffer is returned in the same state. (The metapage is only * touched if it becomes necessary to add or remove overflow pages.) */ static void _hash_splitbucket(Relation rel, Buffer metabuf, Bucket obucket, Bucket nbucket, BlockNumber start_oblkno, BlockNumber start_nblkno, uint32 maxbucket, uint32 highmask, uint32 lowmask) { BlockNumber oblkno; BlockNumber nblkno; Buffer obuf; Buffer nbuf; Page opage; Page npage; HashPageOpaque oopaque; HashPageOpaque nopaque; /* * It should be okay to simultaneously write-lock pages from each bucket, * since no one else can be trying to acquire buffer lock on pages of * either bucket. */ oblkno = start_oblkno; obuf = _hash_getbuf(rel, oblkno, HASH_WRITE, LH_BUCKET_PAGE); opage = BufferGetPage(obuf); oopaque = (HashPageOpaque) PageGetSpecialPointer(opage); nblkno = start_nblkno; nbuf = _hash_getnewbuf(rel, nblkno); npage = BufferGetPage(nbuf); /* initialize the new bucket's primary page */ nopaque = (HashPageOpaque) PageGetSpecialPointer(npage); nopaque->hasho_prevblkno = InvalidBlockNumber; nopaque->hasho_nextblkno = InvalidBlockNumber; nopaque->hasho_bucket = nbucket; nopaque->hasho_flag = LH_BUCKET_PAGE; nopaque->hasho_page_id = HASHO_PAGE_ID; /* * Partition the tuples in the old bucket between the old bucket and the * new bucket, advancing along the old bucket's overflow bucket chain and * adding overflow pages to the new bucket as needed. Outer loop iterates * once per page in old bucket. */ for (;;) { OffsetNumber ooffnum; OffsetNumber omaxoffnum; OffsetNumber deletable[MaxOffsetNumber]; int ndeletable = 0; /* Scan each tuple in old page */ omaxoffnum = PageGetMaxOffsetNumber(opage); for (ooffnum = FirstOffsetNumber; ooffnum <= omaxoffnum; ooffnum = OffsetNumberNext(ooffnum)) { IndexTuple itup; Size itemsz; Bucket bucket; /* * Fetch the item's hash key (conveniently stored in the item) and * determine which bucket it now belongs in. */ itup = (IndexTuple) PageGetItem(opage, PageGetItemId(opage, ooffnum)); bucket = _hash_hashkey2bucket(_hash_get_indextuple_hashkey(itup), maxbucket, highmask, lowmask); if (bucket == nbucket) { /* * insert the tuple into the new bucket. if it doesn't fit on * the current page in the new bucket, we must allocate a new * overflow page and place the tuple on that page instead. */ itemsz = IndexTupleDSize(*itup); itemsz = MAXALIGN(itemsz); if (PageGetFreeSpace(npage) < itemsz) { /* write out nbuf and drop lock, but keep pin */ _hash_chgbufaccess(rel, nbuf, HASH_WRITE, HASH_NOLOCK); /* chain to a new overflow page */ nbuf = _hash_addovflpage(rel, metabuf, nbuf); npage = BufferGetPage(nbuf); /* we don't need nblkno or nopaque within the loop */ } /* * Insert tuple on new page, using _hash_pgaddtup to ensure * correct ordering by hashkey. This is a tad inefficient * since we may have to shuffle itempointers repeatedly. * Possible future improvement: accumulate all the items for * the new page and qsort them before insertion. */ (void) _hash_pgaddtup(rel, nbuf, itemsz, itup); /* * Mark tuple for deletion from old page. */ deletable[ndeletable++] = ooffnum; } else { /* * the tuple stays on this page, so nothing to do. */ Assert(bucket == obucket); } } oblkno = oopaque->hasho_nextblkno; /* * Done scanning this old page. If we moved any tuples, delete them * from the old page. */ if (ndeletable > 0) { PageIndexMultiDelete(opage, deletable, ndeletable); _hash_wrtbuf(rel, obuf); } else _hash_relbuf(rel, obuf); /* Exit loop if no more overflow pages in old bucket */ if (!BlockNumberIsValid(oblkno)) break; /* Else, advance to next old page */ obuf = _hash_getbuf(rel, oblkno, HASH_WRITE, LH_OVERFLOW_PAGE); opage = BufferGetPage(obuf); oopaque = (HashPageOpaque) PageGetSpecialPointer(opage); } /* * We're at the end of the old bucket chain, so we're done partitioning * the tuples. Before quitting, call _hash_squeezebucket to ensure the * tuples remaining in the old bucket (including the overflow pages) are * packed as tightly as possible. The new bucket is already tight. */ _hash_wrtbuf(rel, nbuf); _hash_squeezebucket(rel, obucket, start_oblkno, NULL); }
/* * Subroutine to pre-check whether a page deletion is safe, that is, its * parent page would be left in a valid or deletable state. * * "target" is the page we wish to delete, and "stack" is a search stack * leading to it (approximately). Note that we will update the stack * entry(s) to reflect current downlink positions --- this is harmless and * indeed saves later search effort in _bt_pagedel. * * Note: it's OK to release page locks after checking, because a safe * deletion can't become unsafe due to concurrent activity. A non-rightmost * page cannot become rightmost unless there's a concurrent page deletion, * but only VACUUM does page deletion and we only allow one VACUUM on an index * at a time. An only child could acquire a sibling (of the same parent) only * by being split ... but that would make it a non-rightmost child so the * deletion is still safe. */ static bool _bt_parent_deletion_safe(Relation rel, BlockNumber target, BTStack stack) { BlockNumber parent; OffsetNumber poffset, maxoff; Buffer pbuf; Page page; BTPageOpaque opaque; /* * In recovery mode, assume the deletion being replayed is valid. We * can't always check it because we won't have a full search stack, and we * should complain if there's a problem, anyway. */ if (InRecovery) return true; /* Locate the parent's downlink (updating the stack entry if needed) */ ItemPointerSet(&(stack->bts_btentry.t_tid), target, P_HIKEY); pbuf = _bt_getstackbuf(rel, stack, BT_READ); if (pbuf == InvalidBuffer) elog(ERROR, "failed to re-find parent key in index \"%s\" for deletion target page %u", RelationGetRelationName(rel), target); parent = stack->bts_blkno; poffset = stack->bts_offset; page = BufferGetPage(pbuf); opaque = (BTPageOpaque) PageGetSpecialPointer(page); maxoff = PageGetMaxOffsetNumber(page); /* * If the target is the rightmost child of its parent, then we can't * delete, unless it's also the only child. */ if (poffset >= maxoff) { /* It's rightmost child... */ if (poffset == P_FIRSTDATAKEY(opaque)) { /* * It's only child, so safe if parent would itself be removable. * We have to check the parent itself, and then recurse to test * the conditions at the parent's parent. */ if (P_RIGHTMOST(opaque) || P_ISROOT(opaque)) { _bt_relbuf(rel, pbuf); return false; } _bt_relbuf(rel, pbuf); return _bt_parent_deletion_safe(rel, parent, stack->bts_parent); } else { /* Unsafe to delete */ _bt_relbuf(rel, pbuf); return false; } } else { /* Not rightmost child, so safe to delete */ _bt_relbuf(rel, pbuf); return true; } }
/* * returns modified page or NULL if page isn't modified. * Function works with original page until first change is occurred, * then page is copied into temporary one. */ static Page ginVacuumEntryPage(GinVacuumState *gvs, Buffer buffer, BlockNumber *roots, uint32 *nroot) { Page origpage = BufferGetPage(buffer), tmppage; OffsetNumber i, maxoff = PageGetMaxOffsetNumber(origpage); tmppage = origpage; *nroot = 0; for (i = FirstOffsetNumber; i <= maxoff; i++) { IndexTuple itup = (IndexTuple) PageGetItem(tmppage, PageGetItemId(tmppage, i)); if (GinIsPostingTree(itup)) { /* * store posting tree's roots for further processing, we can't * vacuum it just now due to risk of deadlocks with scans/inserts */ roots[*nroot] = GinItemPointerGetBlockNumber(&itup->t_tid); (*nroot)++; } else if (GinGetNPosting(itup) > 0) { /* * if we already create temporary page, we will make changes in * place */ ItemPointerData *cleaned = (tmppage == origpage) ? NULL : GinGetPosting(itup); uint32 newN = ginVacuumPostingList(gvs, GinGetPosting(itup), GinGetNPosting(itup), &cleaned); if (GinGetNPosting(itup) != newN) { Datum value; OffsetNumber attnum; /* * Some ItemPointers was deleted, so we should remake our * tuple */ if (tmppage == origpage) { /* * On first difference we create temporary page in memory * and copies content in to it. */ tmppage = PageGetTempPageCopy(origpage); if (newN > 0) { Size pos = ((char *) GinGetPosting(itup)) - ((char *) origpage); memcpy(tmppage + pos, cleaned, sizeof(ItemPointerData) * newN); } pfree(cleaned); /* set itup pointer to new page */ itup = (IndexTuple) PageGetItem(tmppage, PageGetItemId(tmppage, i)); } value = gin_index_getattr(&gvs->ginstate, itup); attnum = gintuple_get_attrnum(&gvs->ginstate, itup); itup = GinFormTuple(gvs->index, &gvs->ginstate, attnum, value, GinGetPosting(itup), newN, true); PageIndexTupleDelete(tmppage, i); if (PageAddItem(tmppage, (Item) itup, IndexTupleSize(itup), i, false, false) != i) elog(ERROR, "failed to add item to index page in \"%s\"", RelationGetRelationName(gvs->index)); pfree(itup); } } } return (tmppage == origpage) ? NULL : tmppage; }
/* * _bt_pagedel() -- Delete a page from the b-tree, if legal to do so. * * This action unlinks the page from the b-tree structure, removing all * pointers leading to it --- but not touching its own left and right links. * The page cannot be physically reclaimed right away, since other processes * may currently be trying to follow links leading to the page; they have to * be allowed to use its right-link to recover. See nbtree/README. * * On entry, the target buffer must be pinned and locked (either read or write * lock is OK). This lock and pin will be dropped before exiting. * * The "stack" argument can be a search stack leading (approximately) to the * target page, or NULL --- outside callers typically pass NULL since they * have not done such a search, but internal recursion cases pass the stack * to avoid duplicated search effort. * * Returns the number of pages successfully deleted (zero if page cannot * be deleted now; could be more than one if parent pages were deleted too). * * NOTE: this leaks memory. Rather than trying to clean up everything * carefully, it's better to run it in a temp context that can be reset * frequently. */ int _bt_pagedel(Relation rel, Buffer buf, BTStack stack) { int result; BlockNumber target, leftsib, rightsib, parent; OffsetNumber poffset, maxoff; uint32 targetlevel, ilevel; ItemId itemid; IndexTuple targetkey, itup; ScanKey itup_scankey; Buffer lbuf, rbuf, pbuf; bool parent_half_dead; bool parent_one_child; bool rightsib_empty; Buffer metabuf = InvalidBuffer; Page metapg = NULL; BTMetaPageData *metad = NULL; Page page; BTPageOpaque opaque; /* * We can never delete rightmost pages nor root pages. While at it, check * that page is not already deleted and is empty. */ page = BufferGetPage(buf); opaque = (BTPageOpaque) PageGetSpecialPointer(page); if (P_RIGHTMOST(opaque) || P_ISROOT(opaque) || P_ISDELETED(opaque) || P_FIRSTDATAKEY(opaque) <= PageGetMaxOffsetNumber(page)) { /* Should never fail to delete a half-dead page */ Assert(!P_ISHALFDEAD(opaque)); _bt_relbuf(rel, buf); return 0; } /* * Save info about page, including a copy of its high key (it must have * one, being non-rightmost). */ target = BufferGetBlockNumber(buf); targetlevel = opaque->btpo.level; leftsib = opaque->btpo_prev; itemid = PageGetItemId(page, P_HIKEY); targetkey = CopyIndexTuple((IndexTuple) PageGetItem(page, itemid)); /* * To avoid deadlocks, we'd better drop the target page lock before going * further. */ _bt_relbuf(rel, buf); /* * We need an approximate pointer to the page's parent page. We use the * standard search mechanism to search for the page's high key; this will * give us a link to either the current parent or someplace to its left * (if there are multiple equal high keys). In recursion cases, the * caller already generated a search stack and we can just re-use that * work. */ if (stack == NULL) { if (!InRecovery) { /* we need an insertion scan key to do our search, so build one */ itup_scankey = _bt_mkscankey(rel, targetkey); /* find the leftmost leaf page containing this key */ stack = _bt_search(rel, rel->rd_rel->relnatts, itup_scankey, false, &lbuf, BT_READ); /* don't need a pin on that either */ _bt_relbuf(rel, lbuf); /* * If we are trying to delete an interior page, _bt_search did * more than we needed. Locate the stack item pointing to our * parent level. */ ilevel = 0; for (;;) { if (stack == NULL) elog(ERROR, "not enough stack items"); if (ilevel == targetlevel) break; stack = stack->bts_parent; ilevel++; } } else { /* * During WAL recovery, we can't use _bt_search (for one reason, * it might invoke user-defined comparison functions that expect * facilities not available in recovery mode). Instead, just set * up a dummy stack pointing to the left end of the parent tree * level, from which _bt_getstackbuf will walk right to the parent * page. Painful, but we don't care too much about performance in * this scenario. */ pbuf = _bt_get_endpoint(rel, targetlevel + 1, false); stack = (BTStack) palloc(sizeof(BTStackData)); stack->bts_blkno = BufferGetBlockNumber(pbuf); stack->bts_offset = InvalidOffsetNumber; /* bts_btentry will be initialized below */ stack->bts_parent = NULL; _bt_relbuf(rel, pbuf); } } /* * We cannot delete a page that is the rightmost child of its immediate * parent, unless it is the only child --- in which case the parent has to * be deleted too, and the same condition applies recursively to it. We * have to check this condition all the way up before trying to delete. We * don't need to re-test when deleting a non-leaf page, though. */ if (targetlevel == 0 && !_bt_parent_deletion_safe(rel, target, stack)) return 0; /* * We have to lock the pages we need to modify in the standard order: * moving right, then up. Else we will deadlock against other writers. * * So, we need to find and write-lock the current left sibling of the * target page. The sibling that was current a moment ago could have * split, so we may have to move right. This search could fail if either * the sibling or the target page was deleted by someone else meanwhile; * if so, give up. (Right now, that should never happen, since page * deletion is only done in VACUUM and there shouldn't be multiple VACUUMs * concurrently on the same table.) */ if (leftsib != P_NONE) { lbuf = _bt_getbuf(rel, leftsib, BT_WRITE); page = BufferGetPage(lbuf); opaque = (BTPageOpaque) PageGetSpecialPointer(page); while (P_ISDELETED(opaque) || opaque->btpo_next != target) { /* step right one page */ leftsib = opaque->btpo_next; _bt_relbuf(rel, lbuf); if (leftsib == P_NONE) { elog(LOG, "no left sibling (concurrent deletion?) in \"%s\"", RelationGetRelationName(rel)); return 0; } lbuf = _bt_getbuf(rel, leftsib, BT_WRITE); page = BufferGetPage(lbuf); opaque = (BTPageOpaque) PageGetSpecialPointer(page); } } else lbuf = InvalidBuffer; /* * Next write-lock the target page itself. It should be okay to take just * a write lock not a superexclusive lock, since no scans would stop on an * empty page. */ buf = _bt_getbuf(rel, target, BT_WRITE); page = BufferGetPage(buf); opaque = (BTPageOpaque) PageGetSpecialPointer(page); /* * Check page is still empty etc, else abandon deletion. The empty check * is necessary since someone else might have inserted into it while we * didn't have it locked; the others are just for paranoia's sake. */ if (P_RIGHTMOST(opaque) || P_ISROOT(opaque) || P_ISDELETED(opaque) || P_FIRSTDATAKEY(opaque) <= PageGetMaxOffsetNumber(page)) { _bt_relbuf(rel, buf); if (BufferIsValid(lbuf)) _bt_relbuf(rel, lbuf); return 0; } if (opaque->btpo_prev != leftsib) elog(ERROR, "left link changed unexpectedly in block %u of index \"%s\"", target, RelationGetRelationName(rel)); /* * And next write-lock the (current) right sibling. */ rightsib = opaque->btpo_next; rbuf = _bt_getbuf(rel, rightsib, BT_WRITE); page = BufferGetPage(rbuf); opaque = (BTPageOpaque) PageGetSpecialPointer(page); if (opaque->btpo_prev != target) elog(ERROR, "right sibling's left-link doesn't match: " "block %u links to %u instead of expected %u in index \"%s\"", rightsib, opaque->btpo_prev, target, RelationGetRelationName(rel)); /* * Any insert which would have gone on the target block will now go to the * right sibling block. */ PredicateLockPageCombine(rel, target, rightsib); /* * Next find and write-lock the current parent of the target page. This is * essentially the same as the corresponding step of splitting. */ ItemPointerSet(&(stack->bts_btentry.t_tid), target, P_HIKEY); pbuf = _bt_getstackbuf(rel, stack, BT_WRITE); if (pbuf == InvalidBuffer) elog(ERROR, "failed to re-find parent key in index \"%s\" for deletion target page %u", RelationGetRelationName(rel), target); parent = stack->bts_blkno; poffset = stack->bts_offset; /* * If the target is the rightmost child of its parent, then we can't * delete, unless it's also the only child --- in which case the parent * changes to half-dead status. The "can't delete" case should have been * detected by _bt_parent_deletion_safe, so complain if we see it now. */ page = BufferGetPage(pbuf); opaque = (BTPageOpaque) PageGetSpecialPointer(page); maxoff = PageGetMaxOffsetNumber(page); parent_half_dead = false; parent_one_child = false; if (poffset >= maxoff) { if (poffset == P_FIRSTDATAKEY(opaque)) parent_half_dead = true; else elog(ERROR, "failed to delete rightmost child %u of block %u in index \"%s\"", target, parent, RelationGetRelationName(rel)); } else { /* Will there be exactly one child left in this parent? */ if (OffsetNumberNext(P_FIRSTDATAKEY(opaque)) == maxoff) parent_one_child = true; } /* * If we are deleting the next-to-last page on the target's level, then * the rightsib is a candidate to become the new fast root. (In theory, it * might be possible to push the fast root even further down, but the odds * of doing so are slim, and the locking considerations daunting.) * * We don't support handling this in the case where the parent is becoming * half-dead, even though it theoretically could occur. * * We can safely acquire a lock on the metapage here --- see comments for * _bt_newroot(). */ if (leftsib == P_NONE && !parent_half_dead) { page = BufferGetPage(rbuf); opaque = (BTPageOpaque) PageGetSpecialPointer(page); Assert(opaque->btpo.level == targetlevel); if (P_RIGHTMOST(opaque)) { /* rightsib will be the only one left on the level */ metabuf = _bt_getbuf(rel, BTREE_METAPAGE, BT_WRITE); metapg = BufferGetPage(metabuf); metad = BTPageGetMeta(metapg); /* * The expected case here is btm_fastlevel == targetlevel+1; if * the fastlevel is <= targetlevel, something is wrong, and we * choose to overwrite it to fix it. */ if (metad->btm_fastlevel > targetlevel + 1) { /* no update wanted */ _bt_relbuf(rel, metabuf); metabuf = InvalidBuffer; } } } /* * Check that the parent-page index items we're about to delete/overwrite * contain what we expect. This can fail if the index has become corrupt * for some reason. We want to throw any error before entering the * critical section --- otherwise it'd be a PANIC. * * The test on the target item is just an Assert because _bt_getstackbuf * should have guaranteed it has the expected contents. The test on the * next-child downlink is known to sometimes fail in the field, though. */ page = BufferGetPage(pbuf); opaque = (BTPageOpaque) PageGetSpecialPointer(page); #ifdef USE_ASSERT_CHECKING itemid = PageGetItemId(page, poffset); itup = (IndexTuple) PageGetItem(page, itemid); Assert(ItemPointerGetBlockNumber(&(itup->t_tid)) == target); #endif if (!parent_half_dead) { OffsetNumber nextoffset; nextoffset = OffsetNumberNext(poffset); itemid = PageGetItemId(page, nextoffset); itup = (IndexTuple) PageGetItem(page, itemid); if (ItemPointerGetBlockNumber(&(itup->t_tid)) != rightsib) elog(ERROR, "right sibling %u of block %u is not next child %u of block %u in index \"%s\"", rightsib, target, ItemPointerGetBlockNumber(&(itup->t_tid)), parent, RelationGetRelationName(rel)); } /* * Here we begin doing the deletion. */ /* No ereport(ERROR) until changes are logged */ START_CRIT_SECTION(); /* * Update parent. The normal case is a tad tricky because we want to * delete the target's downlink and the *following* key. Easiest way is * to copy the right sibling's downlink over the target downlink, and then * delete the following item. */ if (parent_half_dead) { PageIndexTupleDelete(page, poffset); opaque->btpo_flags |= BTP_HALF_DEAD; } else { OffsetNumber nextoffset; itemid = PageGetItemId(page, poffset); itup = (IndexTuple) PageGetItem(page, itemid); ItemPointerSet(&(itup->t_tid), rightsib, P_HIKEY); nextoffset = OffsetNumberNext(poffset); PageIndexTupleDelete(page, nextoffset); } /* * Update siblings' side-links. Note the target page's side-links will * continue to point to the siblings. Asserts here are just rechecking * things we already verified above. */ if (BufferIsValid(lbuf)) { page = BufferGetPage(lbuf); opaque = (BTPageOpaque) PageGetSpecialPointer(page); Assert(opaque->btpo_next == target); opaque->btpo_next = rightsib; } page = BufferGetPage(rbuf); opaque = (BTPageOpaque) PageGetSpecialPointer(page); Assert(opaque->btpo_prev == target); opaque->btpo_prev = leftsib; rightsib_empty = (P_FIRSTDATAKEY(opaque) > PageGetMaxOffsetNumber(page)); /* * Mark the page itself deleted. It can be recycled when all current * transactions are gone. Storing GetTopTransactionId() would work, but * we're in VACUUM and would not otherwise have an XID. Having already * updated links to the target, ReadNewTransactionId() suffices as an * upper bound. Any scan having retained a now-stale link is advertising * in its PGXACT an xmin less than or equal to the value we read here. It * will continue to do so, holding back RecentGlobalXmin, for the duration * of that scan. */ page = BufferGetPage(buf); opaque = (BTPageOpaque) PageGetSpecialPointer(page); opaque->btpo_flags &= ~BTP_HALF_DEAD; opaque->btpo_flags |= BTP_DELETED; opaque->btpo.xact = ReadNewTransactionId(); /* And update the metapage, if needed */ if (BufferIsValid(metabuf)) { metad->btm_fastroot = rightsib; metad->btm_fastlevel = targetlevel; MarkBufferDirty(metabuf); } /* Must mark buffers dirty before XLogInsert */ MarkBufferDirty(pbuf); MarkBufferDirty(rbuf); MarkBufferDirty(buf); if (BufferIsValid(lbuf)) MarkBufferDirty(lbuf); /* XLOG stuff */ if (RelationNeedsWAL(rel)) { xl_btree_delete_page xlrec; xl_btree_metadata xlmeta; uint8 xlinfo; XLogRecPtr recptr; XLogRecData rdata[5]; XLogRecData *nextrdata; xlrec.target.node = rel->rd_node; ItemPointerSet(&(xlrec.target.tid), parent, poffset); xlrec.deadblk = target; xlrec.leftblk = leftsib; xlrec.rightblk = rightsib; xlrec.btpo_xact = opaque->btpo.xact; rdata[0].data = (char *) &xlrec; rdata[0].len = SizeOfBtreeDeletePage; rdata[0].buffer = InvalidBuffer; rdata[0].next = nextrdata = &(rdata[1]); if (BufferIsValid(metabuf)) { xlmeta.root = metad->btm_root; xlmeta.level = metad->btm_level; xlmeta.fastroot = metad->btm_fastroot; xlmeta.fastlevel = metad->btm_fastlevel; nextrdata->data = (char *) &xlmeta; nextrdata->len = sizeof(xl_btree_metadata); nextrdata->buffer = InvalidBuffer; nextrdata->next = nextrdata + 1; nextrdata++; xlinfo = XLOG_BTREE_DELETE_PAGE_META; } else if (parent_half_dead) xlinfo = XLOG_BTREE_DELETE_PAGE_HALF; else xlinfo = XLOG_BTREE_DELETE_PAGE; nextrdata->data = NULL; nextrdata->len = 0; nextrdata->next = nextrdata + 1; nextrdata->buffer = pbuf; nextrdata->buffer_std = true; nextrdata++; nextrdata->data = NULL; nextrdata->len = 0; nextrdata->buffer = rbuf; nextrdata->buffer_std = true; nextrdata->next = NULL; if (BufferIsValid(lbuf)) { nextrdata->next = nextrdata + 1; nextrdata++; nextrdata->data = NULL; nextrdata->len = 0; nextrdata->buffer = lbuf; nextrdata->buffer_std = true; nextrdata->next = NULL; } recptr = XLogInsert(RM_BTREE_ID, xlinfo, rdata); if (BufferIsValid(metabuf)) { PageSetLSN(metapg, recptr); PageSetTLI(metapg, ThisTimeLineID); } page = BufferGetPage(pbuf); PageSetLSN(page, recptr); PageSetTLI(page, ThisTimeLineID); page = BufferGetPage(rbuf); PageSetLSN(page, recptr); PageSetTLI(page, ThisTimeLineID); page = BufferGetPage(buf); PageSetLSN(page, recptr); PageSetTLI(page, ThisTimeLineID); if (BufferIsValid(lbuf)) { page = BufferGetPage(lbuf); PageSetLSN(page, recptr); PageSetTLI(page, ThisTimeLineID); } } END_CRIT_SECTION(); /* release metapage; send out relcache inval if metapage changed */ if (BufferIsValid(metabuf)) { CacheInvalidateRelcache(rel); _bt_relbuf(rel, metabuf); } /* can always release leftsib immediately */ if (BufferIsValid(lbuf)) _bt_relbuf(rel, lbuf); /* * If parent became half dead, recurse to delete it. Otherwise, if right * sibling is empty and is now the last child of the parent, recurse to * try to delete it. (These cases cannot apply at the same time, though * the second case might itself recurse to the first.) * * When recursing to parent, we hold the lock on the target page until * done. This delays any insertions into the keyspace that was just * effectively reassigned to the parent's right sibling. If we allowed * that, and there were enough such insertions before we finish deleting * the parent, page splits within that keyspace could lead to inserting * out-of-order keys into the grandparent level. It is thought that that * wouldn't have any serious consequences, but it still seems like a * pretty bad idea. */ if (parent_half_dead) { /* recursive call will release pbuf */ _bt_relbuf(rel, rbuf); result = _bt_pagedel(rel, pbuf, stack->bts_parent) + 1; _bt_relbuf(rel, buf); } else if (parent_one_child && rightsib_empty) { _bt_relbuf(rel, pbuf); _bt_relbuf(rel, buf); /* recursive call will release rbuf */ result = _bt_pagedel(rel, rbuf, stack) + 1; } else { _bt_relbuf(rel, pbuf); _bt_relbuf(rel, buf); _bt_relbuf(rel, rbuf); result = 1; } return result; }
/* * fills WAL record for vacuum leaf page */ static void xlogVacuumPage(Relation index, Buffer buffer) { Page page = BufferGetPage(buffer); XLogRecPtr recptr; XLogRecData rdata[3]; ginxlogVacuumPage data; char *backup; char itups[BLCKSZ]; uint32 len = 0; Assert(GinPageIsLeaf(page)); if (index->rd_istemp) return; data.node = index->rd_node; data.blkno = BufferGetBlockNumber(buffer); if (GinPageIsData(page)) { backup = GinDataPageGetData(page); data.nitem = GinPageGetOpaque(page)->maxoff; if (data.nitem) len = MAXALIGN(sizeof(ItemPointerData) * data.nitem); } else { char *ptr; OffsetNumber i; ptr = backup = itups; for (i = FirstOffsetNumber; i <= PageGetMaxOffsetNumber(page); i++) { IndexTuple itup = (IndexTuple) PageGetItem(page, PageGetItemId(page, i)); memcpy(ptr, itup, IndexTupleSize(itup)); ptr += MAXALIGN(IndexTupleSize(itup)); } data.nitem = PageGetMaxOffsetNumber(page); len = ptr - backup; } rdata[0].buffer = buffer; rdata[0].buffer_std = (GinPageIsData(page)) ? FALSE : TRUE; rdata[0].len = 0; rdata[0].data = NULL; rdata[0].next = rdata + 1; rdata[1].buffer = InvalidBuffer; rdata[1].len = sizeof(ginxlogVacuumPage); rdata[1].data = (char *) &data; if (len == 0) { rdata[1].next = NULL; } else { rdata[1].next = rdata + 2; rdata[2].buffer = InvalidBuffer; rdata[2].len = len; rdata[2].data = backup; rdata[2].next = NULL; } recptr = XLogInsert(RM_GIN_ID, XLOG_GIN_VACUUM_PAGE, rdata); PageSetLSN(page, recptr); PageSetTLI(page, ThisTimeLineID); }
/* * _hash_squeezebucket(rel, bucket) * * Try to squeeze the tuples onto pages occurring earlier in the * bucket chain in an attempt to free overflow pages. When we start * the "squeezing", the page from which we start taking tuples (the * "read" page) is the last bucket in the bucket chain and the page * onto which we start squeezing tuples (the "write" page) is the * first page in the bucket chain. The read page works backward and * the write page works forward; the procedure terminates when the * read page and write page are the same page. * * At completion of this procedure, it is guaranteed that all pages in * the bucket are nonempty, unless the bucket is totally empty (in * which case all overflow pages will be freed). The original implementation * required that to be true on entry as well, but it's a lot easier for * callers to leave empty overflow pages and let this guy clean it up. * * Caller must hold exclusive lock on the target bucket. This allows * us to safely lock multiple pages in the bucket. */ void _hash_squeezebucket(Relation rel, Bucket bucket, BlockNumber bucket_blkno) { Buffer wbuf; Buffer rbuf = 0; BlockNumber wblkno; BlockNumber rblkno; Page wpage; Page rpage; HashPageOpaque wopaque; HashPageOpaque ropaque; OffsetNumber woffnum; OffsetNumber roffnum; IndexTuple itup; Size itemsz; /* * start squeezing into the base bucket page. */ wblkno = bucket_blkno; wbuf = _hash_getbuf(rel, wblkno, HASH_WRITE); _hash_checkpage(rel, wbuf, LH_BUCKET_PAGE); wpage = BufferGetPage(wbuf); wopaque = (HashPageOpaque) PageGetSpecialPointer(wpage); /* * if there aren't any overflow pages, there's nothing to squeeze. */ if (!BlockNumberIsValid(wopaque->hasho_nextblkno)) { _hash_relbuf(rel, wbuf); return; } /* * find the last page in the bucket chain by starting at the base bucket * page and working forward. */ ropaque = wopaque; do { rblkno = ropaque->hasho_nextblkno; if (ropaque != wopaque) _hash_relbuf(rel, rbuf); rbuf = _hash_getbuf(rel, rblkno, HASH_WRITE); _hash_checkpage(rel, rbuf, LH_OVERFLOW_PAGE); rpage = BufferGetPage(rbuf); ropaque = (HashPageOpaque) PageGetSpecialPointer(rpage); Assert(ropaque->hasho_bucket == bucket); } while (BlockNumberIsValid(ropaque->hasho_nextblkno)); /* * squeeze the tuples. */ roffnum = FirstOffsetNumber; for (;;) { /* this test is needed in case page is empty on entry */ if (roffnum <= PageGetMaxOffsetNumber(rpage)) { itup = (IndexTuple) PageGetItem(rpage, PageGetItemId(rpage, roffnum)); itemsz = IndexTupleDSize(*itup); itemsz = MAXALIGN(itemsz); /* * Walk up the bucket chain, looking for a page big enough for * this item. Exit if we reach the read page. */ while (PageGetFreeSpace(wpage) < itemsz) { Assert(!PageIsEmpty(wpage)); wblkno = wopaque->hasho_nextblkno; Assert(BlockNumberIsValid(wblkno)); _hash_wrtbuf(rel, wbuf); if (rblkno == wblkno) { /* wbuf is already released */ _hash_wrtbuf(rel, rbuf); return; } wbuf = _hash_getbuf(rel, wblkno, HASH_WRITE); _hash_checkpage(rel, wbuf, LH_OVERFLOW_PAGE); wpage = BufferGetPage(wbuf); wopaque = (HashPageOpaque) PageGetSpecialPointer(wpage); Assert(wopaque->hasho_bucket == bucket); } /* * we have found room so insert on the "write" page. */ woffnum = OffsetNumberNext(PageGetMaxOffsetNumber(wpage)); if (PageAddItem(wpage, (Item) itup, itemsz, woffnum, LP_USED) == InvalidOffsetNumber) elog(ERROR, "failed to add index item to \"%s\"", RelationGetRelationName(rel)); /* * delete the tuple from the "read" page. PageIndexTupleDelete * repacks the ItemId array, so 'roffnum' will be "advanced" to * the "next" ItemId. */ PageIndexTupleDelete(rpage, roffnum); } /* * if the "read" page is now empty because of the deletion (or because * it was empty when we got to it), free it. * * Tricky point here: if our read and write pages are adjacent in the * bucket chain, our write lock on wbuf will conflict with * _hash_freeovflpage's attempt to update the sibling links of the * removed page. However, in that case we are done anyway, so we can * simply drop the write lock before calling _hash_freeovflpage. */ if (PageIsEmpty(rpage)) { rblkno = ropaque->hasho_prevblkno; Assert(BlockNumberIsValid(rblkno)); /* are we freeing the page adjacent to wbuf? */ if (rblkno == wblkno) { /* yes, so release wbuf lock first */ _hash_wrtbuf(rel, wbuf); /* free this overflow page (releases rbuf) */ _hash_freeovflpage(rel, rbuf); /* done */ return; } /* free this overflow page, then get the previous one */ _hash_freeovflpage(rel, rbuf); rbuf = _hash_getbuf(rel, rblkno, HASH_WRITE); _hash_checkpage(rel, rbuf, LH_OVERFLOW_PAGE); rpage = BufferGetPage(rbuf); ropaque = (HashPageOpaque) PageGetSpecialPointer(rpage); Assert(ropaque->hasho_bucket == bucket); roffnum = FirstOffsetNumber; } } /* NOTREACHED */ }
/* * hashgettuple() -- Get the next tuple in the scan. */ Datum hashgettuple(PG_FUNCTION_ARGS) { IndexScanDesc scan = (IndexScanDesc) PG_GETARG_POINTER(0); ScanDirection dir = (ScanDirection) PG_GETARG_INT32(1); HashScanOpaque so = (HashScanOpaque) scan->opaque; Relation rel = scan->indexRelation; Buffer buf; Page page; OffsetNumber offnum; ItemPointer current; bool res; /* Hash indexes are always lossy since we store only the hash code */ scan->xs_recheck = true; /* * We hold pin but not lock on current buffer while outside the hash AM. * Reacquire the read lock here. */ if (BufferIsValid(so->hashso_curbuf)) _hash_chgbufaccess(rel, so->hashso_curbuf, HASH_NOLOCK, HASH_READ); /* * If we've already initialized this scan, we can just advance it in the * appropriate direction. If we haven't done so yet, we call a routine to * get the first item in the scan. */ current = &(so->hashso_curpos); if (ItemPointerIsValid(current)) { /* * An insertion into the current index page could have happened while * we didn't have read lock on it. Re-find our position by looking * for the TID we previously returned. (Because we hold share lock on * the bucket, no deletions or splits could have occurred; therefore * we can expect that the TID still exists in the current index page, * at an offset >= where we were.) */ OffsetNumber maxoffnum; buf = so->hashso_curbuf; Assert(BufferIsValid(buf)); page = BufferGetPage(buf); maxoffnum = PageGetMaxOffsetNumber(page); for (offnum = ItemPointerGetOffsetNumber(current); offnum <= maxoffnum; offnum = OffsetNumberNext(offnum)) { IndexTuple itup; itup = (IndexTuple) PageGetItem(page, PageGetItemId(page, offnum)); if (ItemPointerEquals(&(so->hashso_heappos), &(itup->t_tid))) break; } if (offnum > maxoffnum) elog(ERROR, "failed to re-find scan position within index \"%s\"", RelationGetRelationName(rel)); ItemPointerSetOffsetNumber(current, offnum); /* * Check to see if we should kill the previously-fetched tuple. */ if (scan->kill_prior_tuple) { /* * Yes, so mark it by setting the LP_DEAD state in the item flags. */ ItemIdMarkDead(PageGetItemId(page, offnum)); /* * Since this can be redone later if needed, mark as a hint. */ MarkBufferDirtyHint(buf, true); } /* * Now continue the scan. */ res = _hash_next(scan, dir); } else res = _hash_first(scan, dir); /* * Skip killed tuples if asked to. */ if (scan->ignore_killed_tuples) { while (res) { offnum = ItemPointerGetOffsetNumber(current); page = BufferGetPage(so->hashso_curbuf); if (!ItemIdIsDead(PageGetItemId(page, offnum))) break; res = _hash_next(scan, dir); } } /* Release read lock on current buffer, but keep it pinned */ if (BufferIsValid(so->hashso_curbuf)) _hash_chgbufaccess(rel, so->hashso_curbuf, HASH_READ, HASH_NOLOCK); /* Return current heap TID on success */ scan->xs_ctup.t_self = so->hashso_heappos; PG_RETURN_BOOL(res); }
/* * Common part of an insert or update. Inserts the new tuple and updates the * revmap. */ static void brin_xlog_insert_update(XLogReaderState *record, xl_brin_insert *xlrec) { XLogRecPtr lsn = record->EndRecPtr; Buffer buffer; BlockNumber regpgno; Page page; XLogRedoAction action; /* * If we inserted the first and only tuple on the page, re-initialize the * page from scratch. */ if (XLogRecGetInfo(record) & XLOG_BRIN_INIT_PAGE) { buffer = XLogInitBufferForRedo(record, 0); page = BufferGetPage(buffer); brin_page_init(page, BRIN_PAGETYPE_REGULAR); action = BLK_NEEDS_REDO; } else { action = XLogReadBufferForRedo(record, 0, &buffer); } /* need this page's blkno to store in revmap */ regpgno = BufferGetBlockNumber(buffer); /* insert the index item into the page */ if (action == BLK_NEEDS_REDO) { OffsetNumber offnum; BrinTuple *tuple; Size tuplen; tuple = (BrinTuple *) XLogRecGetBlockData(record, 0, &tuplen); Assert(tuple->bt_blkno == xlrec->heapBlk); page = (Page) BufferGetPage(buffer); offnum = xlrec->offnum; if (PageGetMaxOffsetNumber(page) + 1 < offnum) elog(PANIC, "brin_xlog_insert_update: invalid max offset number"); offnum = PageAddItem(page, (Item) tuple, tuplen, offnum, true, false); if (offnum == InvalidOffsetNumber) elog(PANIC, "brin_xlog_insert_update: failed to add tuple"); PageSetLSN(page, lsn); MarkBufferDirty(buffer); } if (BufferIsValid(buffer)) UnlockReleaseBuffer(buffer); /* update the revmap */ action = XLogReadBufferForRedo(record, 1, &buffer); if (action == BLK_NEEDS_REDO) { ItemPointerData tid; ItemPointerSet(&tid, regpgno, xlrec->offnum); page = (Page) BufferGetPage(buffer); brinSetHeapBlockItemptr(buffer, xlrec->pagesPerRange, xlrec->heapBlk, tid); PageSetLSN(page, lsn); MarkBufferDirty(buffer); } if (BufferIsValid(buffer)) UnlockReleaseBuffer(buffer); /* XXX no FSM updates here ... */ }