diff options
Diffstat (limited to 'src/backend/access/nbtree/nbtinsert.c')
| -rw-r--r-- | src/backend/access/nbtree/nbtinsert.c | 46 |
1 files changed, 30 insertions, 16 deletions
diff --git a/src/backend/access/nbtree/nbtinsert.c b/src/backend/access/nbtree/nbtinsert.c index 5890f393f67..48d19be3aab 100644 --- a/src/backend/access/nbtree/nbtinsert.c +++ b/src/backend/access/nbtree/nbtinsert.c @@ -1797,7 +1797,6 @@ _bt_insert_parent(Relation rel, stack = &fakestack; stack->bts_blkno = BufferGetBlockNumber(pbuf); stack->bts_offset = InvalidOffsetNumber; - stack->bts_btentry = InvalidBlockNumber; stack->bts_parent = NULL; _bt_relbuf(rel, pbuf); } @@ -1819,8 +1818,7 @@ _bt_insert_parent(Relation rel, * new downlink will be inserted at the correct offset. Even buf's * parent may have changed. */ - stack->bts_btentry = bknum; - pbuf = _bt_getstackbuf(rel, stack); + pbuf = _bt_getstackbuf(rel, stack, bknum); /* * Now we can unlock the right child. The left child will be unlocked @@ -1834,7 +1832,7 @@ _bt_insert_parent(Relation rel, errmsg_internal("failed to re-find parent key in index \"%s\" for split pages %u/%u", RelationGetRelationName(rel), bknum, rbknum))); - /* Recursively update the parent */ + /* Recursively insert into the parent */ _bt_insertonpg(rel, NULL, pbuf, buf, stack->bts_parent, new_item, stack->bts_offset + 1, is_only); @@ -1901,21 +1899,37 @@ _bt_finish_split(Relation rel, Buffer lbuf, BTStack stack) } /* - * _bt_getstackbuf() -- Walk back up the tree one step, and find the item - * we last looked at in the parent. + * _bt_getstackbuf() -- Walk back up the tree one step, and find the pivot + * tuple whose downlink points to child page. * - * This is possible because we save the downlink from the parent item, - * which is enough to uniquely identify it. Insertions into the parent - * level could cause the item to move right; deletions could cause it - * to move left, but not left of the page we previously found it in. + * Caller passes child's block number, which is used to identify + * associated pivot tuple in parent page using a linear search that + * matches on pivot's downlink/block number. The expected location of + * the pivot tuple is taken from the stack one level above the child + * page. This is used as a starting point. Insertions into the + * parent level could cause the pivot tuple to move right; deletions + * could cause it to move left, but not left of the page we previously + * found it on. * - * Adjusts bts_blkno & bts_offset if changed. + * Caller can use its stack to relocate the pivot tuple/downlink for + * any same-level page to the right of the page found by its initial + * descent. This is necessary because of the possibility that caller + * moved right to recover from a concurrent page split. It's also + * convenient for certain callers to be able to step right when there + * wasn't a concurrent page split, while still using their original + * stack. For example, the checkingunique _bt_doinsert() case may + * have to step right when there are many physical duplicates, and its + * scantid forces an insertion to the right of the "first page the + * value could be on". * - * Returns write-locked buffer, or InvalidBuffer if item not found - * (should not happen). + * Returns write-locked parent page buffer, or InvalidBuffer if pivot + * tuple not found (should not happen). Adjusts bts_blkno & + * bts_offset if changed. Page split caller should insert its new + * pivot tuple for its new right sibling page on parent page, at the + * offset number bts_offset + 1. */ Buffer -_bt_getstackbuf(Relation rel, BTStack stack) +_bt_getstackbuf(Relation rel, BTStack stack, BlockNumber child) { BlockNumber blkno; OffsetNumber start; @@ -1977,7 +1991,7 @@ _bt_getstackbuf(Relation rel, BTStack stack) itemid = PageGetItemId(page, offnum); item = (IndexTuple) PageGetItem(page, itemid); - if (BTreeInnerTupleGetDownLink(item) == stack->bts_btentry) + if (BTreeInnerTupleGetDownLink(item) == child) { /* Return accurate pointer to where link is now */ stack->bts_blkno = blkno; @@ -1993,7 +2007,7 @@ _bt_getstackbuf(Relation rel, BTStack stack) itemid = PageGetItemId(page, offnum); item = (IndexTuple) PageGetItem(page, itemid); - if (BTreeInnerTupleGetDownLink(item) == stack->bts_btentry) + if (BTreeInnerTupleGetDownLink(item) == child) { /* Return accurate pointer to where link is now */ stack->bts_blkno = blkno; |