static int64
nextval_internal(Oid relid)
{
SeqTable elm;
Relation seqrel;
Buffer buf;
Page page;
HeapTupleData seqtuple;
Form_pg_sequence seq;
int64 incby,
maxv,
minv,
cache,
log,
fetch,
last;
int64 result,
next,
rescnt = 0;
bool logit = false;
/* open and AccessShareLock sequence */
init_sequence(relid, &elm, &seqrel);
if (pg_class_aclcheck(elm->relid, GetUserId(),
ACL_USAGE | ACL_UPDATE) != ACLCHECK_OK)
ereport(ERROR,
(errcode(ERRCODE_INSUFFICIENT_PRIVILEGE),
errmsg("permission denied for sequence %s",
RelationGetRelationName(seqrel))));
/* read-only transactions may only modify temp sequences */
if (!seqrel->rd_islocaltemp)
PreventCommandIfReadOnly("nextval()");
/*
* Forbid this during parallel operation because, to make it work,
* the cooperating backends would need to share the backend-local cached
* sequence information. Currently, we don't support that.
*/
PreventCommandIfParallelMode("nextval()");
if (elm->last != elm->cached) /* some numbers were cached */
{
Assert(elm->last_valid);
Assert(elm->increment != 0);
elm->last += elm->increment;
relation_close(seqrel, NoLock);
last_used_seq = elm;
return elm->last;
}
/* lock page' buffer and read tuple */
seq = read_seq_tuple(elm, seqrel, &buf, &seqtuple);
page = BufferGetPage(buf);
last = next = result = seq->last_value;
incby = seq->increment_by;
maxv = seq->max_value;
minv = seq->min_value;
fetch = cache = seq->cache_value;
log = seq->log_cnt;
if (!seq->is_called)
{
rescnt++; /* return last_value if not is_called */
fetch--;
}
/*
* Decide whether we should emit a WAL log record. If so, force up the
* fetch count to grab SEQ_LOG_VALS more values than we actually need to
* cache. (These will then be usable without logging.)
*
* If this is the first nextval after a checkpoint, we must force a new___
* WAL record to be written anyway, else replay starting from the
* checkpoint would fail to advance the sequence past the logged values.
* In this case we may as well fetch extra values.
*/
if (log < fetch || !seq->is_called)
{
/* forced log to satisfy local demand for values */
fetch = log = fetch + SEQ_LOG_VALS;
logit = true;
}
else
{
XLogRecPtr redoptr = GetRedoRecPtr();
if (PageGetLSN(page) <= redoptr)
{
/* last update of seq was before checkpoint */
fetch = log = fetch + SEQ_LOG_VALS;
logit = true;
}
}
while (fetch) /* try to fetch cache [+ log ] numbers */
{
/*
* Check MAXVALUE for ascending sequences and MINVALUE for descending
* sequences
*/
if (incby > 0)
{
/* ascending sequence */
if ((maxv >= 0 && next > maxv - incby) ||
(maxv < 0 && next + incby > maxv))
{
if (rescnt > 0)
break; /* stop fetching */
if (!seq->is_cycled)
{
char buf[100];
snprintf(buf, sizeof(buf), INT64_FORMAT, maxv);
ereport(ERROR,
(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
errmsg("nextval: reached maximum value of sequence \"%s\" (%s)",
RelationGetRelationName(seqrel), buf)));
}
next = minv;
}
else
next += incby;
}
else
{
/* descending sequence */
if ((minv < 0 && next < minv - incby) ||
(minv >= 0 && next + incby < minv))
{
if (rescnt > 0)
break; /* stop fetching */
if (!seq->is_cycled)
{
char buf[100];
snprintf(buf, sizeof(buf), INT64_FORMAT, minv);
ereport(ERROR,
(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
errmsg("nextval: reached minimum value of sequence \"%s\" (%s)",
RelationGetRelationName(seqrel), buf)));
}
next = maxv;
}
else
next += incby;
}
fetch--;
if (rescnt < cache)
{
log--;
rescnt++;
last = next;
if (rescnt == 1) /* if it's first result - */
result = next; /* it's what to return */
}
}
log -= fetch; /* adjust for any unfetched numbers */
Assert(log >= 0);
/* save info in local cache */
elm->last = result; /* last returned number */
elm->cached = last; /* last fetched number */
elm->last_valid = true;
last_used_seq = elm;
/*
* If something needs to be WAL logged, acquire an xid, so this
* transaction's commit will trigger a WAL flush and wait for
* syncrep. It's sufficient to ensure the toplevel transaction has an xid,
* no need to assign xids subxacts, that'll already trigger an appropriate
* wait. (Have to do that here, so we're outside the critical section)
*/
if (logit && RelationNeedsWAL(seqrel))
GetTopTransactionId();
/* ready to change the on-disk (or really, in-buffer) tuple */
START_CRIT_SECTION();
/*
* We must mark the buffer dirty before doing XLogInsert(); see notes in
* SyncOneBuffer(). However, we don't apply the desired changes just yet.
* This looks like a violation of the buffer update protocol, but it is in
* fact safe because we hold exclusive lock on the buffer. Any other
* process, including a checkpoint, that tries to examine the buffer
* contents will block until we release the lock, and then will see the
* final state that we install below.
*/
MarkBufferDirty(buf);
/* XLOG stuff */
if (logit && RelationNeedsWAL(seqrel))
{
xl_seq_rec xlrec;
XLogRecPtr recptr;
/*
* We don't log the current state of the tuple, but rather the state
* as it would appear after "log" more fetches. This lets us skip
* that many future WAL records, at the cost that we lose those
* sequence values if we crash.
*/
XLogBeginInsert();
XLogRegisterBuffer(0, buf, REGBUF_WILL_INIT);
/* set values that will be saved in xlog */
seq->last_value = next;
seq->is_called = true;
seq->log_cnt = 0;
xlrec.node = seqrel->rd_node;
XLogRegisterData((char *) &xlrec, sizeof(xl_seq_rec));
XLogRegisterData((char *) seqtuple.t_data, seqtuple.t_len);
recptr = XLogInsert(RM_SEQ_ID, XLOG_SEQ_LOG);
PageSetLSN(page, recptr);
}
/* Now update sequence tuple to the intended final state */
seq->last_value = last; /* last fetched number */
seq->is_called = true;
seq->log_cnt = log; /* how much is logged */
END_CRIT_SECTION();
UnlockReleaseBuffer(buf);
relation_close(seqrel, NoLock);
return result;
}
/*
* _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 */
}
static bool
gistplacetopage(GISTInsertState *state, GISTSTATE *giststate)
{
bool is_splitted = false;
bool is_leaf = (GistPageIsLeaf(state->stack->page)) ? true : false;
/*
* if (!is_leaf) remove old key: This node's key has been modified, either
* because a child split occurred or because we needed to adjust our key
* for an insert in a child node. Therefore, remove the old version of
* this node's key.
*
* for WAL replay, in the non-split case we handle this by setting up a
* one-element todelete array; in the split case, it's handled implicitly
* because the tuple vector passed to gistSplit won't include this tuple.
*
* XXX: If we want to change fillfactors between node and leaf, fillfactor
* = (is_leaf ? state->leaf_fillfactor : state->node_fillfactor)
*/
if (gistnospace(state->stack->page, state->itup, state->ituplen,
is_leaf ? InvalidOffsetNumber : state->stack->childoffnum,
state->freespace))
{
/* no space for insertion */
IndexTuple *itvec;
int tlen;
SplitedPageLayout *dist = NULL,
*ptr;
BlockNumber rrlink = InvalidBlockNumber;
GistNSN oldnsn;
is_splitted = true;
/*
* Form index tuples vector to split: remove old tuple if t's needed
* and add new tuples to vector
*/
itvec = gistextractpage(state->stack->page, &tlen);
if (!is_leaf)
{
/* on inner page we should remove old tuple */
int pos = state->stack->childoffnum - FirstOffsetNumber;
tlen--;
if (pos != tlen)
memmove(itvec + pos, itvec + pos + 1, sizeof(IndexTuple) * (tlen - pos));
}
itvec = gistjoinvector(itvec, &tlen, state->itup, state->ituplen);
dist = gistSplit(state->r, state->stack->page, itvec, tlen, giststate);
state->itup = (IndexTuple *) palloc(sizeof(IndexTuple) * tlen);
state->ituplen = 0;
if (state->stack->blkno != GIST_ROOT_BLKNO)
{
/*
* if non-root split then we should not allocate new buffer, but
* we must create temporary page to operate
*/
dist->buffer = state->stack->buffer;
dist->page = PageGetTempPage(BufferGetPage(dist->buffer), sizeof(GISTPageOpaqueData));
/* clean all flags except F_LEAF */
GistPageGetOpaque(dist->page)->flags = (is_leaf) ? F_LEAF : 0;
}
/* make new pages and fills them */
for (ptr = dist; ptr; ptr = ptr->next)
{
int i;
char *data;
/* get new page */
if (ptr->buffer == InvalidBuffer)
{
ptr->buffer = gistNewBuffer(state->r);
GISTInitBuffer(ptr->buffer, (is_leaf) ? F_LEAF : 0);
ptr->page = BufferGetPage(ptr->buffer);
}
ptr->block.blkno = BufferGetBlockNumber(ptr->buffer);
/*
* fill page, we can do it because all these pages are new
* (ie not linked in tree or masked by temp page
*/
data = (char *) (ptr->list);
for (i = 0; i < ptr->block.num; i++)
{
if (PageAddItem(ptr->page, (Item) data, IndexTupleSize((IndexTuple) data), i + FirstOffsetNumber, LP_USED) == InvalidOffsetNumber)
elog(ERROR, "failed to add item to index page in \"%s\"", RelationGetRelationName(state->r));
data += IndexTupleSize((IndexTuple) data);
}
/* set up ItemPointer and remember it for parent */
ItemPointerSetBlockNumber(&(ptr->itup->t_tid), ptr->block.blkno);
state->itup[state->ituplen] = ptr->itup;
state->ituplen++;
}
/* saves old rightlink */
if (state->stack->blkno != GIST_ROOT_BLKNO)
rrlink = GistPageGetOpaque(dist->page)->rightlink;
START_CRIT_SECTION();
/*
* must mark buffers dirty before XLogInsert, even though we'll still
* be changing their opaque fields below. set up right links.
*/
for (ptr = dist; ptr; ptr = ptr->next)
{
MarkBufferDirty(ptr->buffer);
GistPageGetOpaque(ptr->page)->rightlink = (ptr->next) ?
ptr->next->block.blkno : rrlink;
}
/* restore splitted non-root page */
if (state->stack->blkno != GIST_ROOT_BLKNO)
{
PageRestoreTempPage(dist->page, BufferGetPage(dist->buffer));
dist->page = BufferGetPage(dist->buffer);
}
if (!state->r->rd_istemp)
{
XLogRecPtr recptr;
XLogRecData *rdata;
rdata = formSplitRdata(state->r->rd_node, state->stack->blkno,
is_leaf, &(state->key), dist);
recptr = XLogInsert(RM_GIST_ID, XLOG_GIST_PAGE_SPLIT, rdata);
for (ptr = dist; ptr; ptr = ptr->next)
{
PageSetLSN(ptr->page, recptr);
PageSetTLI(ptr->page, ThisTimeLineID);
}
}
else
{
for (ptr = dist; ptr; ptr = ptr->next)
{
PageSetLSN(ptr->page, XLogRecPtrForTemp);
}
}
/* set up NSN */
oldnsn = GistPageGetOpaque(dist->page)->nsn;
if (state->stack->blkno == GIST_ROOT_BLKNO)
/* if root split we should put initial value */
oldnsn = PageGetLSN(dist->page);
for (ptr = dist; ptr; ptr = ptr->next)
{
/* only for last set oldnsn */
GistPageGetOpaque(ptr->page)->nsn = (ptr->next) ?
PageGetLSN(ptr->page) : oldnsn;
}
/*
* release buffers, if it was a root split then release all buffers
* because we create all buffers
*/
ptr = (state->stack->blkno == GIST_ROOT_BLKNO) ? dist : dist->next;
for (; ptr; ptr = ptr->next)
UnlockReleaseBuffer(ptr->buffer);
if (state->stack->blkno == GIST_ROOT_BLKNO)
{
gistnewroot(state->r, state->stack->buffer, state->itup, state->ituplen, &(state->key));
state->needInsertComplete = false;
}
END_CRIT_SECTION();
}
else
{
/* enough space */
START_CRIT_SECTION();
if (!is_leaf)
PageIndexTupleDelete(state->stack->page, state->stack->childoffnum);
gistfillbuffer(state->r, state->stack->page, state->itup, state->ituplen, InvalidOffsetNumber);
MarkBufferDirty(state->stack->buffer);
if (!state->r->rd_istemp)
{
OffsetNumber noffs = 0,
offs[1];
XLogRecPtr recptr;
XLogRecData *rdata;
if (!is_leaf)
{
/* only on inner page we should delete previous version */
offs[0] = state->stack->childoffnum;
noffs = 1;
}
rdata = formUpdateRdata(state->r->rd_node, state->stack->buffer,
offs, noffs,
state->itup, state->ituplen,
&(state->key));
recptr = XLogInsert(RM_GIST_ID, XLOG_GIST_PAGE_UPDATE, rdata);
PageSetLSN(state->stack->page, recptr);
PageSetTLI(state->stack->page, ThisTimeLineID);
}
else
PageSetLSN(state->stack->page, XLogRecPtrForTemp);
if (state->stack->blkno == GIST_ROOT_BLKNO)
state->needInsertComplete = false;
END_CRIT_SECTION();
if (state->ituplen > 1)
{ /* previous is_splitted==true */
/*
* child was splited, so we must form union for insertion in
* parent
*/
IndexTuple newtup = gistunion(state->r, state->itup, state->ituplen, giststate);
ItemPointerSetBlockNumber(&(newtup->t_tid), state->stack->blkno);
state->itup[0] = newtup;
state->ituplen = 1;
}
else if (is_leaf)
{
/*
* itup[0] store key to adjust parent, we set it to valid to
* correct check by GistTupleIsInvalid macro in gistgetadjusted()
*/
ItemPointerSetBlockNumber(&(state->itup[0]->t_tid), state->stack->blkno);
GistTupleSetValid(state->itup[0]);
}
}
return is_splitted;
}
/*
* _bt_getroot() -- Get the root page of the btree.
*
* Since the root page can move around the btree file, we have to read
* its location from the metadata page, and then read the root page
* itself. If no root page exists yet, we have to create one. The
* standard class of race conditions exists here; I think I covered
* them all in the Hopi Indian rain dance of lock requests below.
*
* The access type parameter (BT_READ or BT_WRITE) controls whether
* a new root page will be created or not. If access = BT_READ,
* and no root page exists, we just return InvalidBuffer. For
* BT_WRITE, we try to create the root page if it doesn't exist.
* NOTE that the returned root page will have only a read lock set
* on it even if access = BT_WRITE!
*
* The returned page is not necessarily the true root --- it could be
* a "fast root" (a page that is alone in its level due to deletions).
* Also, if the root page is split while we are "in flight" to it,
* what we will return is the old root, which is now just the leftmost
* page on a probably-not-very-wide level. For most purposes this is
* as good as or better than the true root, so we do not bother to
* insist on finding the true root. We do, however, guarantee to
* return a live (not deleted or half-dead) page.
*
* On successful return, the root page is pinned and read-locked.
* The metadata page is not locked or pinned on exit.
*/
Buffer
_bt_getroot(Relation rel, int access)
{
Buffer metabuf;
Page metapg;
BTPageOpaque metaopaque;
Buffer rootbuf;
Page rootpage;
BTPageOpaque rootopaque;
BlockNumber rootblkno;
uint32 rootlevel;
BTMetaPageData *metad;
/*
* Try to use previously-cached metapage data to find the root. This
* normally saves one buffer access per index search, which is a very
* helpful savings in bufmgr traffic and hence contention.
*/
if (rel->rd_amcache != NULL)
{
metad = (BTMetaPageData *) rel->rd_amcache;
/* We shouldn't have cached it if any of these fail */
Assert(metad->btm_magic == BTREE_MAGIC);
Assert(metad->btm_version == BTREE_VERSION);
Assert(metad->btm_root != P_NONE);
rootblkno = metad->btm_fastroot;
Assert(rootblkno != P_NONE);
rootlevel = metad->btm_fastlevel;
rootbuf = _bt_getbuf(rel, rootblkno, BT_READ);
rootpage = BufferGetPage(rootbuf);
rootopaque = (BTPageOpaque) PageGetSpecialPointer(rootpage);
/*
* Since the cache might be stale, we check the page more carefully
* here than normal. We *must* check that it's not deleted. If it's
* not alone on its level, then we reject too --- this may be overly
* paranoid but better safe than sorry. Note we don't check P_ISROOT,
* because that's not set in a "fast root".
*/
if (!P_IGNORE(rootopaque) &&
rootopaque->btpo.level == rootlevel &&
P_LEFTMOST(rootopaque) &&
P_RIGHTMOST(rootopaque))
{
/* OK, accept cached page as the root */
return rootbuf;
}
_bt_relbuf(rel, rootbuf);
/* Cache is stale, throw it away */
if (rel->rd_amcache)
pfree(rel->rd_amcache);
rel->rd_amcache = NULL;
}
metabuf = _bt_getbuf(rel, BTREE_METAPAGE, BT_READ);
metapg = BufferGetPage(metabuf);
metaopaque = (BTPageOpaque) PageGetSpecialPointer(metapg);
metad = BTPageGetMeta(metapg);
/* sanity-check the metapage */
if (!(metaopaque->btpo_flags & BTP_META) ||
metad->btm_magic != BTREE_MAGIC)
ereport(ERROR,
(errcode(ERRCODE_INDEX_CORRUPTED),
errmsg("index \"%s\" is not a btree",
RelationGetRelationName(rel))));
if (metad->btm_version != BTREE_VERSION)
ereport(ERROR,
(errcode(ERRCODE_INDEX_CORRUPTED),
errmsg("version mismatch in index \"%s\": file version %d, code version %d",
RelationGetRelationName(rel),
metad->btm_version, BTREE_VERSION)));
/* if no root page initialized yet, do it */
if (metad->btm_root == P_NONE)
{
/* If access = BT_READ, caller doesn't want us to create root yet */
if (access == BT_READ)
{
_bt_relbuf(rel, metabuf);
return InvalidBuffer;
}
/* trade in our read lock for a write lock */
LockBuffer(metabuf, BUFFER_LOCK_UNLOCK);
LockBuffer(metabuf, BT_WRITE);
/*
* Race condition: if someone else initialized the metadata between
* the time we released the read lock and acquired the write lock, we
* must avoid doing it again.
*/
if (metad->btm_root != P_NONE)
{
/*
* Metadata initialized by someone else. In order to guarantee no
* deadlocks, we have to release the metadata page and start all
* over again. (Is that really true? But it's hardly worth trying
* to optimize this case.)
*/
_bt_relbuf(rel, metabuf);
return _bt_getroot(rel, access);
}
/*
* Get, initialize, write, and leave a lock of the appropriate type on
* the new root page. Since this is the first page in the tree, it's
* a leaf as well as the root.
*/
rootbuf = _bt_getbuf(rel, P_NEW, BT_WRITE);
rootblkno = BufferGetBlockNumber(rootbuf);
rootpage = BufferGetPage(rootbuf);
rootopaque = (BTPageOpaque) PageGetSpecialPointer(rootpage);
rootopaque->btpo_prev = rootopaque->btpo_next = P_NONE;
rootopaque->btpo_flags = (BTP_LEAF | BTP_ROOT);
rootopaque->btpo.level = 0;
rootopaque->btpo_cycleid = 0;
/* NO ELOG(ERROR) till meta is updated */
START_CRIT_SECTION();
metad->btm_root = rootblkno;
metad->btm_level = 0;
metad->btm_fastroot = rootblkno;
metad->btm_fastlevel = 0;
MarkBufferDirty(rootbuf);
MarkBufferDirty(metabuf);
/* XLOG stuff */
if (RelationNeedsWAL(rel))
{
xl_btree_newroot xlrec;
XLogRecPtr recptr;
XLogRecData rdata;
xlrec.node = rel->rd_node;
xlrec.rootblk = rootblkno;
xlrec.level = 0;
rdata.data = (char *) &xlrec;
rdata.len = SizeOfBtreeNewroot;
rdata.buffer = InvalidBuffer;
rdata.next = NULL;
recptr = XLogInsert(RM_BTREE_ID, XLOG_BTREE_NEWROOT, &rdata);
PageSetLSN(rootpage, recptr);
PageSetTLI(rootpage, ThisTimeLineID);
PageSetLSN(metapg, recptr);
PageSetTLI(metapg, ThisTimeLineID);
}
END_CRIT_SECTION();
/*
* Send out relcache inval for metapage change (probably unnecessary
* here, but let's be safe).
*/
CacheInvalidateRelcache(rel);
/*
* swap root write lock for read lock. There is no danger of anyone
* else accessing the new root page while it's unlocked, since no one
* else knows where it is yet.
*/
LockBuffer(rootbuf, BUFFER_LOCK_UNLOCK);
LockBuffer(rootbuf, BT_READ);
/* okay, metadata is correct, release lock on it */
_bt_relbuf(rel, metabuf);
}
else
{
rootblkno = metad->btm_fastroot;
Assert(rootblkno != P_NONE);
rootlevel = metad->btm_fastlevel;
/*
* Cache the metapage data for next time
*/
rel->rd_amcache = MemoryContextAlloc(rel->rd_indexcxt,
sizeof(BTMetaPageData));
memcpy(rel->rd_amcache, metad, sizeof(BTMetaPageData));
/*
* We are done with the metapage; arrange to release it via first
* _bt_relandgetbuf call
*/
rootbuf = metabuf;
for (;;)
{
rootbuf = _bt_relandgetbuf(rel, rootbuf, rootblkno, BT_READ);
rootpage = BufferGetPage(rootbuf);
rootopaque = (BTPageOpaque) PageGetSpecialPointer(rootpage);
if (!P_IGNORE(rootopaque))
break;
/* it's dead, Jim. step right one page */
if (P_RIGHTMOST(rootopaque))
elog(ERROR, "no live root page found in index \"%s\"",
RelationGetRelationName(rel));
rootblkno = rootopaque->btpo_next;
}
/* Note: can't check btpo.level on deleted pages */
if (rootopaque->btpo.level != rootlevel)
elog(ERROR, "root page %u of index \"%s\" has level %u, expected %u",
rootblkno, RelationGetRelationName(rel),
rootopaque->btpo.level, rootlevel);
}
/*
* By here, we have a pin and read lock on the root page, and no lock set
* on the metadata page. Return the root page's buffer.
*/
return rootbuf;
}
/*
* _hash_addovflpage
*
* Add an overflow page to the bucket whose last page is pointed to by 'buf'.
*
* On entry, the caller must hold a pin but no lock on 'buf'. The pin is
* dropped before exiting (we assume the caller is not interested in 'buf'
* anymore) if not asked to retain. The pin will be retained only for the
* primary bucket. The returned overflow page will be pinned and
* write-locked; it is guaranteed to be empty.
*
* The caller must hold a pin, but no lock, on the metapage buffer.
* That buffer is returned in the same state.
*
* NB: since this could be executed concurrently by multiple processes,
* one should not assume that the returned overflow page will be the
* immediate successor of the originally passed 'buf'. Additional overflow
* pages might have been added to the bucket chain in between.
*/
Buffer
_hash_addovflpage(Relation rel, Buffer metabuf, Buffer buf, bool retain_pin)
{
Buffer ovflbuf;
Page page;
Page ovflpage;
HashPageOpaque pageopaque;
HashPageOpaque ovflopaque;
HashMetaPage metap;
Buffer mapbuf = InvalidBuffer;
Buffer newmapbuf = InvalidBuffer;
BlockNumber blkno;
uint32 orig_firstfree;
uint32 splitnum;
uint32 *freep = NULL;
uint32 max_ovflpg;
uint32 bit;
uint32 bitmap_page_bit;
uint32 first_page;
uint32 last_bit;
uint32 last_page;
uint32 i,
j;
bool page_found = false;
/*
* Write-lock the tail page. Here, we need to maintain locking order such
* that, first acquire the lock on tail page of bucket, then on meta page
* to find and lock the bitmap page and if it is found, then lock on meta
* page is released, then finally acquire the lock on new overflow buffer.
* We need this locking order to avoid deadlock with backends that are
* doing inserts.
*
* Note: We could have avoided locking many buffers here if we made two
* WAL records for acquiring an overflow page (one to allocate an overflow
* page and another to add it to overflow bucket chain). However, doing
* so can leak an overflow page, if the system crashes after allocation.
* Needless to say, it is better to have a single record from a
* performance point of view as well.
*/
LockBuffer(buf, BUFFER_LOCK_EXCLUSIVE);
/* probably redundant... */
_hash_checkpage(rel, buf, LH_BUCKET_PAGE | LH_OVERFLOW_PAGE);
/* loop to find current tail page, in case someone else inserted too */
for (;;)
{
BlockNumber nextblkno;
page = BufferGetPage(buf);
pageopaque = (HashPageOpaque) PageGetSpecialPointer(page);
nextblkno = pageopaque->hasho_nextblkno;
if (!BlockNumberIsValid(nextblkno))
break;
/* we assume we do not need to write the unmodified page */
if (retain_pin)
{
/* pin will be retained only for the primary bucket page */
Assert((pageopaque->hasho_flag & LH_PAGE_TYPE) == LH_BUCKET_PAGE);
LockBuffer(buf, BUFFER_LOCK_UNLOCK);
}
else
_hash_relbuf(rel, buf);
retain_pin = false;
buf = _hash_getbuf(rel, nextblkno, HASH_WRITE, LH_OVERFLOW_PAGE);
}
/* Get exclusive lock on the meta page */
LockBuffer(metabuf, BUFFER_LOCK_EXCLUSIVE);
_hash_checkpage(rel, metabuf, LH_META_PAGE);
metap = HashPageGetMeta(BufferGetPage(metabuf));
/* start search at hashm_firstfree */
orig_firstfree = metap->hashm_firstfree;
first_page = orig_firstfree >> BMPG_SHIFT(metap);
bit = orig_firstfree & BMPG_MASK(metap);
i = first_page;
j = bit / BITS_PER_MAP;
bit &= ~(BITS_PER_MAP - 1);
/* outer loop iterates once per bitmap page */
for (;;)
{
BlockNumber mapblkno;
Page mappage;
uint32 last_inpage;
/* want to end search with the last existing overflow page */
splitnum = metap->hashm_ovflpoint;
max_ovflpg = metap->hashm_spares[splitnum] - 1;
last_page = max_ovflpg >> BMPG_SHIFT(metap);
last_bit = max_ovflpg & BMPG_MASK(metap);
if (i > last_page)
break;
Assert(i < metap->hashm_nmaps);
mapblkno = metap->hashm_mapp[i];
if (i == last_page)
last_inpage = last_bit;
else
last_inpage = BMPGSZ_BIT(metap) - 1;
/* Release exclusive lock on metapage while reading bitmap page */
LockBuffer(metabuf, BUFFER_LOCK_UNLOCK);
mapbuf = _hash_getbuf(rel, mapblkno, HASH_WRITE, LH_BITMAP_PAGE);
mappage = BufferGetPage(mapbuf);
freep = HashPageGetBitmap(mappage);
for (; bit <= last_inpage; j++, bit += BITS_PER_MAP)
{
if (freep[j] != ALL_SET)
{
page_found = true;
/* Reacquire exclusive lock on the meta page */
LockBuffer(metabuf, BUFFER_LOCK_EXCLUSIVE);
/* convert bit to bit number within page */
bit += _hash_firstfreebit(freep[j]);
bitmap_page_bit = bit;
/* convert bit to absolute bit number */
bit += (i << BMPG_SHIFT(metap));
/* Calculate address of the recycled overflow page */
blkno = bitno_to_blkno(metap, bit);
/* Fetch and init the recycled page */
ovflbuf = _hash_getinitbuf(rel, blkno);
goto found;
}
}
/* No free space here, try to advance to next map page */
_hash_relbuf(rel, mapbuf);
mapbuf = InvalidBuffer;
i++;
j = 0; /* scan from start of next map page */
bit = 0;
/* Reacquire exclusive lock on the meta page */
LockBuffer(metabuf, BUFFER_LOCK_EXCLUSIVE);
}
/*
* No free pages --- have to extend the relation to add an overflow page.
* First, check to see if we have to add a new bitmap page too.
*/
if (last_bit == (uint32) (BMPGSZ_BIT(metap) - 1))
{
/*
* We create the new bitmap page with all pages marked "in use".
* Actually two pages in the new bitmap's range will exist
* immediately: the bitmap page itself, and the following page which
* is the one we return to the caller. Both of these are correctly
* marked "in use". Subsequent pages do not exist yet, but it is
* convenient to pre-mark them as "in use" too.
*/
bit = metap->hashm_spares[splitnum];
/* metapage already has a write lock */
if (metap->hashm_nmaps >= HASH_MAX_BITMAPS)
ereport(ERROR,
(errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED),
errmsg("out of overflow pages in hash index \"%s\"",
RelationGetRelationName(rel))));
newmapbuf = _hash_getnewbuf(rel, bitno_to_blkno(metap, bit), MAIN_FORKNUM);
}
else
{
/*
* Nothing to do here; since the page will be past the last used page,
* we know its bitmap bit was preinitialized to "in use".
*/
}
/* Calculate address of the new overflow page */
bit = BufferIsValid(newmapbuf) ?
metap->hashm_spares[splitnum] + 1 : metap->hashm_spares[splitnum];
blkno = bitno_to_blkno(metap, bit);
/*
* Fetch the page with _hash_getnewbuf to ensure smgr's idea of the
* relation length stays in sync with ours. XXX It's annoying to do this
* with metapage write lock held; would be better to use a lock that
* doesn't block incoming searches.
*
* It is okay to hold two buffer locks here (one on tail page of bucket
* and other on new overflow page) since there cannot be anyone else
* contending for access to ovflbuf.
*/
ovflbuf = _hash_getnewbuf(rel, blkno, MAIN_FORKNUM);
found:
/*
* Do the update. No ereport(ERROR) until changes are logged. We want to
* log the changes for bitmap page and overflow page together to avoid
* loss of pages in case the new page is added.
*/
START_CRIT_SECTION();
if (page_found)
{
Assert(BufferIsValid(mapbuf));
/* mark page "in use" in the bitmap */
SETBIT(freep, bitmap_page_bit);
MarkBufferDirty(mapbuf);
}
else
{
/* update the count to indicate new overflow page is added */
metap->hashm_spares[splitnum]++;
if (BufferIsValid(newmapbuf))
{
_hash_initbitmapbuffer(newmapbuf, metap->hashm_bmsize, false);
MarkBufferDirty(newmapbuf);
/* add the new bitmap page to the metapage's list of bitmaps */
metap->hashm_mapp[metap->hashm_nmaps] = BufferGetBlockNumber(newmapbuf);
metap->hashm_nmaps++;
metap->hashm_spares[splitnum]++;
MarkBufferDirty(metabuf);
}
/*
* for new overflow page, we don't need to explicitly set the bit in
* bitmap page, as by default that will be set to "in use".
*/
}
/*
* Adjust hashm_firstfree to avoid redundant searches. But don't risk
* changing it if someone moved it while we were searching bitmap pages.
*/
if (metap->hashm_firstfree == orig_firstfree)
{
metap->hashm_firstfree = bit + 1;
MarkBufferDirty(metabuf);
}
/* initialize new overflow page */
ovflpage = BufferGetPage(ovflbuf);
ovflopaque = (HashPageOpaque) PageGetSpecialPointer(ovflpage);
ovflopaque->hasho_prevblkno = BufferGetBlockNumber(buf);
ovflopaque->hasho_nextblkno = InvalidBlockNumber;
ovflopaque->hasho_bucket = pageopaque->hasho_bucket;
ovflopaque->hasho_flag = LH_OVERFLOW_PAGE;
ovflopaque->hasho_page_id = HASHO_PAGE_ID;
MarkBufferDirty(ovflbuf);
/* logically chain overflow page to previous page */
pageopaque->hasho_nextblkno = BufferGetBlockNumber(ovflbuf);
MarkBufferDirty(buf);
/* XLOG stuff */
if (RelationNeedsWAL(rel))
{
XLogRecPtr recptr;
xl_hash_add_ovfl_page xlrec;
xlrec.bmpage_found = page_found;
xlrec.bmsize = metap->hashm_bmsize;
XLogBeginInsert();
XLogRegisterData((char *) &xlrec, SizeOfHashAddOvflPage);
XLogRegisterBuffer(0, ovflbuf, REGBUF_WILL_INIT);
XLogRegisterBufData(0, (char *) &pageopaque->hasho_bucket, sizeof(Bucket));
XLogRegisterBuffer(1, buf, REGBUF_STANDARD);
if (BufferIsValid(mapbuf))
{
XLogRegisterBuffer(2, mapbuf, REGBUF_STANDARD);
XLogRegisterBufData(2, (char *) &bitmap_page_bit, sizeof(uint32));
}
if (BufferIsValid(newmapbuf))
XLogRegisterBuffer(3, newmapbuf, REGBUF_WILL_INIT);
XLogRegisterBuffer(4, metabuf, REGBUF_STANDARD);
XLogRegisterBufData(4, (char *) &metap->hashm_firstfree, sizeof(uint32));
recptr = XLogInsert(RM_HASH_ID, XLOG_HASH_ADD_OVFL_PAGE);
PageSetLSN(BufferGetPage(ovflbuf), recptr);
PageSetLSN(BufferGetPage(buf), recptr);
if (BufferIsValid(mapbuf))
PageSetLSN(BufferGetPage(mapbuf), recptr);
if (BufferIsValid(newmapbuf))
PageSetLSN(BufferGetPage(newmapbuf), recptr);
PageSetLSN(BufferGetPage(metabuf), recptr);
}
END_CRIT_SECTION();
if (retain_pin)
LockBuffer(buf, BUFFER_LOCK_UNLOCK);
else
_hash_relbuf(rel, buf);
if (BufferIsValid(mapbuf))
_hash_relbuf(rel, mapbuf);
LockBuffer(metabuf, BUFFER_LOCK_UNLOCK);
if (BufferIsValid(newmapbuf))
_hash_relbuf(rel, newmapbuf);
return ovflbuf;
}
/*
* 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;
}
/*
* Delete item(s) from a btree page during VACUUM.
*
* This must only be used for deleting leaf items. Deleting an item on a
* non-leaf page has to be done as part of an atomic action that includes
* deleting the page it points to.
*
* This routine assumes that the caller has pinned and locked the buffer.
* Also, the given itemnos *must* appear in increasing order in the array.
*
* We record VACUUMs and b-tree deletes differently in WAL. InHotStandby
* we need to be able to pin all of the blocks in the btree in physical
* order when replaying the effects of a VACUUM, just as we do for the
* original VACUUM itself. lastBlockVacuumed allows us to tell whether an
* intermediate range of blocks has had no changes at all by VACUUM,
* and so must be scanned anyway during replay. We always write a WAL record
* for the last block in the index, whether or not it contained any items
* to be removed. This allows us to scan right up to end of index to
* ensure correct locking.
*/
void
_bt_delitems_vacuum(Relation rel, Buffer buf,
OffsetNumber *itemnos, int nitems,
BlockNumber lastBlockVacuumed)
{
Page page = BufferGetPage(buf);
BTPageOpaque opaque;
/* No ereport(ERROR) until changes are logged */
START_CRIT_SECTION();
/* Fix the page */
if (nitems > 0)
PageIndexMultiDelete(page, itemnos, nitems);
/*
* We can clear the vacuum cycle ID since this page has certainly been
* processed by the current vacuum scan.
*/
opaque = (BTPageOpaque) PageGetSpecialPointer(page);
opaque->btpo_cycleid = 0;
/*
* Mark the page as not containing any LP_DEAD items. This is not
* certainly true (there might be some that have recently been marked, but
* weren't included in our target-item list), but it will almost always be
* true and it doesn't seem worth an additional page scan to check it.
* Remember that BTP_HAS_GARBAGE is only a hint anyway.
*/
opaque->btpo_flags &= ~BTP_HAS_GARBAGE;
MarkBufferDirty(buf);
/* XLOG stuff */
if (RelationNeedsWAL(rel))
{
XLogRecPtr recptr;
XLogRecData rdata[2];
xl_btree_vacuum xlrec_vacuum;
xlrec_vacuum.node = rel->rd_node;
xlrec_vacuum.block = BufferGetBlockNumber(buf);
xlrec_vacuum.lastBlockVacuumed = lastBlockVacuumed;
rdata[0].data = (char *) &xlrec_vacuum;
rdata[0].len = SizeOfBtreeVacuum;
rdata[0].buffer = InvalidBuffer;
rdata[0].next = &(rdata[1]);
/*
* The target-offsets array is not in the buffer, but pretend that it
* is. When XLogInsert stores the whole buffer, the offsets array
* need not be stored too.
*/
if (nitems > 0)
{
rdata[1].data = (char *) itemnos;
rdata[1].len = nitems * sizeof(OffsetNumber);
}
else
{
rdata[1].data = NULL;
rdata[1].len = 0;
}
rdata[1].buffer = buf;
rdata[1].buffer_std = true;
rdata[1].next = NULL;
recptr = XLogInsert(RM_BTREE_ID, XLOG_BTREE_VACUUM, rdata);
PageSetLSN(page, recptr);
PageSetTLI(page, ThisTimeLineID);
}
END_CRIT_SECTION();
}
static void
bitmap_xlog_insert_bitmap(bool redo, XLogRecPtr lsn, XLogRecord* record)
{
xl_bm_bitmappage *xlrec = (xl_bm_bitmappage*) XLogRecGetData(record);
Relation reln;
reln = XLogOpenRelation(xlrec->bm_node);
if (!RelationIsValid(reln))
return;
if (redo)
{
Buffer bitmapBuffer;
Page bitmapPage;
BMBitmapOpaque bitmapPageOpaque ;
bitmapBuffer = XLogReadBuffer(false, reln, xlrec->bm_bitmap_blkno);
if (!BufferIsValid(bitmapBuffer))
elog(PANIC, "bm_insert_redo: block unfound: %d",
xlrec->bm_bitmap_blkno);
bitmapPage = BufferGetPage(bitmapBuffer);
if (XLByteLT(PageGetLSN(bitmapPage), lsn))
{
bitmapPageOpaque = (BMBitmapOpaque)PageGetSpecialPointer(bitmapPage);;
#ifdef BM_DEBUG
ereport(LOG, (errcode(LOG),
errmsg("call bitmap_xlog_insert_bitmap: redo=%d, blkno=%d, isOpaque=%d, words_used=%d, lastword=%d, next_blkno=%d\n", redo, xlrec->bm_bitmap_blkno, xlrec->bm_isOpaque, xlrec->bm_lastword_pos, xlrec->bm_lastword_in_block, xlrec->bm_next_blkno)));
#endif
if (xlrec->bm_isOpaque)
{
if (bitmapPageOpaque->bm_bitmap_next != InvalidBlockNumber)
elog(PANIC,
"%s next bitmap page for blkno %d is already set",
"bm_insert_redo: ",
xlrec->bm_bitmap_blkno);
Assert(bitmapPageOpaque->bm_hrl_words_used ==
BM_NUM_OF_HRL_WORDS_PER_PAGE);
bitmapPageOpaque->bm_bitmap_next = xlrec->bm_next_blkno;
}
else
{
BMBitmap bitmap;
if (bitmapPageOpaque->bm_hrl_words_used !=
xlrec->bm_lastword_pos - 1)
elog(PANIC,
"bm_insert_redo: a bit has been inserted in the pos %d",
xlrec->bm_lastword_pos);
Assert (xlrec->bm_lastword_in_block != 0);
bitmap = (BMBitmap) PageGetContents(bitmapPage);
bitmap->bm_headerWords
[(bitmapPageOpaque->bm_hrl_words_used/BM_HRL_WORD_SIZE)] |=
(1<<(BM_HRL_WORD_SIZE-1-
(bitmapPageOpaque->bm_hrl_words_used%BM_HRL_WORD_SIZE)));
bitmap->bm_contentWords[bitmapPageOpaque->bm_hrl_words_used] =
xlrec->bm_lastword_in_block;
bitmapPageOpaque->bm_hrl_words_used ++;
}
PageSetLSN(bitmapPage, lsn);
PageSetTLI(bitmapPage, ThisTimeLineID);
_bitmap_wrtbuf(bitmapBuffer);
}
else
_bitmap_relbuf(bitmapBuffer);
}
else
elog(PANIC, "bm_insert_undo: not implemented.");
}
/*
* _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_newpage(bool redo, XLogRecPtr lsn, XLogRecord *record)
{
xl_bm_newpage *xlrec = (xl_bm_newpage*) XLogRecGetData(record);
Relation reln;
Page page;
uint8 info;
/* xl_bm_metapage *xlrecMeta = (xl_bm_metapage*)
((char*)xlrec+sizeof(xl_bm_newpage)); */
info = record->xl_info & ~XLR_INFO_MASK;
ereport(DEBUG1, (errmsg_internal("into --> XLogOpenRelation")));
reln = XLogOpenRelation(xlrec->bm_node);
ereport(DEBUG1, (errmsg_internal("done --> XLogOpenRelation")));
if (!RelationIsValid(reln))
return;
ereport(DEBUG1, (errmsg_internal("crash1")));
if (redo)
{
Buffer buffer;
#ifdef BM_DEBUG
ereport(LOG, (errcode(LOG),
errmsg("call bitmap_xlog_newpage: redo=%d, info=%x\n", redo, info)));
#endif
buffer = XLogReadBuffer(true, reln, xlrec->bm_new_blkno);
if (!BufferIsValid(buffer))
elog(PANIC, "bm_insert_redo: block unfound: %d",
xlrec->bm_new_blkno);
page = BufferGetPage(buffer);
if (XLByteLT(PageGetLSN(page), lsn))
{
Buffer metabuf;
BMMetaPage metapage;
switch (info)
{
case XLOG_BITMAP_INSERT_NEWLOV:
_bitmap_lovpageinit(reln, buffer);
break;
case XLOG_BITMAP_INSERT_NEWLOVMETA:
_bitmap_lovmetapageinit(reln, buffer);
break;
case XLOG_BITMAP_INSERT_NEWBITMAP:
_bitmap_bitmappageinit(reln, buffer);
break;
default:
elog(PANIC, "bitmap_redo: unknown newpage op code %u", info);
}
PageSetLSN(page, lsn);
PageSetTLI(page, ThisTimeLineID);
_bitmap_wrtbuf(buffer);
metabuf = XLogReadBuffer(true, reln, BM_METAPAGE);
if (!BufferIsValid(metabuf))
elog(PANIC, "bm_insert_redo: block unfound: %d", BM_METAPAGE);
metapage = (BMMetaPage)BufferGetPage(metabuf);
if (XLByteLT(PageGetLSN(metapage), lsn))
{
PageSetLSN(metapage, lsn);
PageSetTLI(metapage, ThisTimeLineID);
_bitmap_wrtbuf(metabuf);
}
else
_bitmap_relbuf(metabuf);
}
else {
_bitmap_relbuf(buffer);
}
}
else
elog(PANIC, "bm_insert_undo: not implemented.");
/* elog(PANIC, "call completely done for _bitmap_lovmetapageinit from bitmap_xlog_newpage[src/backend/access/bitmap/bitmapxlog.c]", info); */
}
static void
bitmap_xlog_insert_lovmeta(bool redo, XLogRecPtr lsn, XLogRecord* record)
{
xl_bm_lovmetapage *xlrec = (xl_bm_lovmetapage*)XLogRecGetData(record);
Relation reln;
reln = XLogOpenRelation(xlrec->bm_node);
/* reln = XLogOpenRelation(redo, RM_BITMAP_ID, xlrec->bm_node);*/
if (!RelationIsValid(reln))
return;
if (redo)
{
Buffer lovMetabuf;
Page lovMetapage;
BMLOVMetaItem copyMetaItems, metaItems;
#ifdef BM_DEBUG
ereport(LOG, (errcode(LOG),
errmsg("call bitmap_xlog_insert_lovmeta: redo=%d\n", redo)));
#endif
lovMetabuf = XLogReadBuffer(false, reln, BM_LOV_STARTPAGE-1);
if (!BufferIsValid(lovMetabuf))
elog(PANIC, "bm_insert_redo: block unfound: %d -- at (%d,%d,%d)",
BM_LOV_STARTPAGE-1, xlrec->bm_node.spcNode,
xlrec->bm_node.dbNode, xlrec->bm_node.relNode);
lovMetapage = BufferGetPage(lovMetabuf);
if (XLByteLT(PageGetLSN(lovMetapage), lsn))
{
#ifdef BM_DEBUG
uint32 attno;
#endif
copyMetaItems = (BMLOVMetaItem)PageGetContents(lovMetapage);
metaItems = (BMLOVMetaItem)
((char*)xlrec + sizeof(xl_bm_lovmetapage));
memcpy(copyMetaItems, metaItems,
xlrec->bm_num_of_attrs * sizeof(BMLOVMetaItemData));
#ifdef BM_DEBUG
for(attno=0; attno<xlrec->bm_num_of_attrs; attno++)
elog(LOG, "metaItems=%d, %d, %d",
copyMetaItems[attno].bm_lov_heapId,
copyMetaItems[attno].bm_lov_indexId,
copyMetaItems[attno].bm_lov_lastpage);
#endif
PageSetLSN(lovMetapage, lsn);
PageSetTLI(lovMetapage, ThisTimeLineID);
_bitmap_wrtbuf(lovMetabuf);
}
else
_bitmap_relbuf(lovMetabuf);
}
else
elog(PANIC, "bm_insert_undo: not implemented.");
}
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.");
}
/*
* 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 ... */
}
/*
* Main internal procedure that handles 2 & 3 arg forms of SETVAL.
*
* Note that the 3 arg version (which sets the is_called flag) is
* only for use in pg_dump, and setting the is_called flag may not
* work if multiple users are attached to the database and referencing
* the sequence (unlikely if pg_dump is restoring it).
*
* It is necessary to have the 3 arg version so that pg_dump can
* restore the state of a sequence exactly during data-only restores -
* it is the only way to clear the is_called flag in an existing
* sequence.
*/
static void
do_setval(Oid relid, int64 next, bool iscalled)
{
SeqTable elm;
Relation seqrel;
Buffer buf;
HeapTupleData seqtuple;
Form_pg_sequence seq;
/* open and AccessShareLock sequence */
init_sequence(relid, &elm, &seqrel);
if (pg_class_aclcheck(elm->relid, GetUserId(), ACL_UPDATE) != ACLCHECK_OK)
ereport(ERROR,
(errcode(ERRCODE_INSUFFICIENT_PRIVILEGE),
errmsg("permission denied for sequence %s",
RelationGetRelationName(seqrel))));
/* read-only transactions may only modify temp sequences */
if (!seqrel->rd_islocaltemp)
PreventCommandIfReadOnly("setval()");
/*
* Forbid this during parallel operation because, to make it work,
* the cooperating backends would need to share the backend-local cached
* sequence information. Currently, we don't support that.
*/
PreventCommandIfParallelMode("setval()");
/* lock page' buffer and read tuple */
seq = read_seq_tuple(elm, seqrel, &buf, &seqtuple);
if ((next < seq->min_value) || (next > seq->max_value))
{
char bufv[100],
bufm[100],
bufx[100];
snprintf(bufv, sizeof(bufv), INT64_FORMAT, next);
snprintf(bufm, sizeof(bufm), INT64_FORMAT, seq->min_value);
snprintf(bufx, sizeof(bufx), INT64_FORMAT, seq->max_value);
ereport(ERROR,
(errcode(ERRCODE_NUMERIC_VALUE_OUT_OF_RANGE),
errmsg("setval: value %s is out of bounds for sequence \"%s\" (%s..%s)",
bufv, RelationGetRelationName(seqrel),
bufm, bufx)));
}
/* Set the currval() state only if iscalled = true */
if (iscalled)
{
elm->last = next; /* last returned number */
elm->last_valid = true;
}
/* In any case, forget any future cached numbers */
elm->cached = elm->last;
/* check the comment above nextval_internal()'s equivalent call. */
if (RelationNeedsWAL(seqrel))
GetTopTransactionId();
/* ready to change the on-disk (or really, in-buffer) tuple */
START_CRIT_SECTION();
seq->last_value = next; /* last fetched number */
seq->is_called = iscalled;
seq->log_cnt = 0;
MarkBufferDirty(buf);
/* XLOG stuff */
if (RelationNeedsWAL(seqrel))
{
xl_seq_rec xlrec;
XLogRecPtr recptr;
Page page = BufferGetPage(buf);
XLogBeginInsert();
XLogRegisterBuffer(0, buf, REGBUF_WILL_INIT);
xlrec.node = seqrel->rd_node;
XLogRegisterData((char *) &xlrec, sizeof(xl_seq_rec));
XLogRegisterData((char *) seqtuple.t_data, seqtuple.t_len);
recptr = XLogInsert(RM_SEQ_ID, XLOG_SEQ_LOG);
PageSetLSN(page, recptr);
}
END_CRIT_SECTION();
UnlockReleaseBuffer(buf);
relation_close(seqrel, NoLock);
}
/*
* 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.
*
* This function also deletes the tuples that are moved by split to other
* bucket.
*
* Result: a palloc'd struct containing statistical info for VACUUM displays.
*/
IndexBulkDeleteResult *
hashbulkdelete(IndexVacuumInfo *info, IndexBulkDeleteResult *stats,
IndexBulkDeleteCallback callback, void *callback_state)
{
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 = InvalidBuffer;
HashMetaPage metap;
HashMetaPage cachedmetap;
tuples_removed = 0;
num_index_tuples = 0;
/*
* We need a copy of the metapage so that we can use its hashm_spares[]
* values to compute bucket page addresses, but a cached copy should be
* good enough. (If not, we'll detect that further down and refresh the
* cache as necessary.)
*/
cachedmetap = _hash_getcachedmetap(rel, &metabuf, false);
Assert(cachedmetap != NULL);
orig_maxbucket = cachedmetap->hashm_maxbucket;
orig_ntuples = cachedmetap->hashm_ntuples;
/* 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;
Buffer bucket_buf;
Buffer buf;
HashPageOpaque bucket_opaque;
Page page;
bool split_cleanup = false;
/* Get address of bucket's start page */
bucket_blkno = BUCKET_TO_BLKNO(cachedmetap, cur_bucket);
blkno = bucket_blkno;
/*
* We need to acquire a cleanup lock on the primary bucket page to out
* wait concurrent scans before deleting the dead tuples.
*/
buf = ReadBufferExtended(rel, MAIN_FORKNUM, blkno, RBM_NORMAL, info->strategy);
LockBufferForCleanup(buf);
_hash_checkpage(rel, buf, LH_BUCKET_PAGE);
page = BufferGetPage(buf);
bucket_opaque = (HashPageOpaque) PageGetSpecialPointer(page);
/*
* If the bucket contains tuples that are moved by split, then we need
* to delete such tuples. We can't delete such tuples if the split
* operation on bucket is not finished as those are needed by scans.
*/
if (!H_BUCKET_BEING_SPLIT(bucket_opaque) &&
H_NEEDS_SPLIT_CLEANUP(bucket_opaque))
{
split_cleanup = true;
/*
* This bucket might have been split since we last held a lock on
* the metapage. If so, hashm_maxbucket, hashm_highmask and
* hashm_lowmask might be old enough to cause us to fail to remove
* tuples left behind by the most recent split. To prevent that,
* now that the primary page of the target bucket has been locked
* (and thus can't be further split), check whether we need to
* update our cached metapage data.
*
* NB: The check for InvalidBlockNumber is only needed for
* on-disk compatibility with indexes created before we started
* storing hashm_maxbucket in the primary page's hasho_prevblkno.
*/
if (bucket_opaque->hasho_prevblkno != InvalidBlockNumber &&
bucket_opaque->hasho_prevblkno > cachedmetap->hashm_maxbucket)
{
cachedmetap = _hash_getcachedmetap(rel, &metabuf, true);
Assert(cachedmetap != NULL);
}
}
bucket_buf = buf;
hashbucketcleanup(rel, cur_bucket, bucket_buf, blkno, info->strategy,
cachedmetap->hashm_maxbucket,
cachedmetap->hashm_highmask,
cachedmetap->hashm_lowmask, &tuples_removed,
&num_index_tuples, split_cleanup,
callback, callback_state);
_hash_dropbuf(rel, bucket_buf);
/* Advance to next bucket */
cur_bucket++;
}
if (BufferIsInvalid(metabuf))
metabuf = _hash_getbuf(rel, HASH_METAPAGE, HASH_NOLOCK, LH_META_PAGE);
/* Write-lock metapage and check for split since we started */
LockBuffer(metabuf, BUFFER_LOCK_EXCLUSIVE);
metap = HashPageGetMeta(BufferGetPage(metabuf));
if (cur_maxbucket != metap->hashm_maxbucket)
{
/* There's been a split, so process the additional bucket(s) */
LockBuffer(metabuf, BUFFER_LOCK_UNLOCK);
cachedmetap = _hash_getcachedmetap(rel, &metabuf, true);
Assert(cachedmetap != NULL);
cur_maxbucket = cachedmetap->hashm_maxbucket;
goto loop_top;
}
/* Okay, we're really done. Update tuple count in metapage. */
START_CRIT_SECTION();
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;
}
MarkBufferDirty(metabuf);
/* XLOG stuff */
if (RelationNeedsWAL(rel))
{
xl_hash_update_meta_page xlrec;
XLogRecPtr recptr;
xlrec.ntuples = metap->hashm_ntuples;
XLogBeginInsert();
XLogRegisterData((char *) &xlrec, sizeof(SizeOfHashUpdateMetaPage));
XLogRegisterBuffer(0, metabuf, REGBUF_STANDARD);
recptr = XLogInsert(RM_HASH_ID, XLOG_HASH_UPDATE_META_PAGE);
PageSetLSN(BufferGetPage(metabuf), recptr);
}
END_CRIT_SECTION();
_hash_relbuf(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 */
return stats;
}
/*
* _hash_init() -- Initialize the metadata page of a hash index,
* the initial buckets, and the initial bitmap page.
*
* The initial number of buckets is dependent on num_tuples, an estimate
* of the number of tuples to be loaded into the index initially. The
* chosen number of buckets is returned.
*
* We are fairly cavalier about locking here, since we know that no one else
* could be accessing this index. In particular the rule about not holding
* multiple buffer locks is ignored.
*/
uint32
_hash_init(Relation rel, double num_tuples, ForkNumber forkNum)
{
Buffer metabuf;
Buffer buf;
Buffer bitmapbuf;
Page pg;
HashMetaPage metap;
RegProcedure procid;
int32 data_width;
int32 item_width;
int32 ffactor;
uint32 num_buckets;
uint32 i;
bool use_wal;
/* safety check */
if (RelationGetNumberOfBlocksInFork(rel, forkNum) != 0)
elog(ERROR, "cannot initialize non-empty hash index \"%s\"",
RelationGetRelationName(rel));
/*
* WAL log creation of pages if the relation is persistent, or this is the
* init fork. Init forks for unlogged relations always need to be WAL
* logged.
*/
use_wal = RelationNeedsWAL(rel) || forkNum == INIT_FORKNUM;
/*
* Determine the target fill factor (in tuples per bucket) for this index.
* The idea is to make the fill factor correspond to pages about as full
* as the user-settable fillfactor parameter says. We can compute it
* exactly since the index datatype (i.e. uint32 hash key) is fixed-width.
*/
data_width = sizeof(uint32);
item_width = MAXALIGN(sizeof(IndexTupleData)) + MAXALIGN(data_width) +
sizeof(ItemIdData); /* include the line pointer */
ffactor = RelationGetTargetPageUsage(rel, HASH_DEFAULT_FILLFACTOR) / item_width;
/* keep to a sane range */
if (ffactor < 10)
ffactor = 10;
procid = index_getprocid(rel, 1, HASHSTANDARD_PROC);
/*
* We initialize the metapage, the first N bucket pages, and the first
* bitmap page in sequence, using _hash_getnewbuf to cause smgrextend()
* calls to occur. This ensures that the smgr level has the right idea of
* the physical index length.
*
* Critical section not required, because on error the creation of the
* whole relation will be rolled back.
*/
metabuf = _hash_getnewbuf(rel, HASH_METAPAGE, forkNum);
_hash_init_metabuffer(metabuf, num_tuples, procid, ffactor, false);
MarkBufferDirty(metabuf);
pg = BufferGetPage(metabuf);
metap = HashPageGetMeta(pg);
/* XLOG stuff */
if (use_wal)
{
xl_hash_init_meta_page xlrec;
XLogRecPtr recptr;
xlrec.num_tuples = num_tuples;
xlrec.procid = metap->hashm_procid;
xlrec.ffactor = metap->hashm_ffactor;
XLogBeginInsert();
XLogRegisterData((char *) &xlrec, SizeOfHashInitMetaPage);
XLogRegisterBuffer(0, metabuf, REGBUF_WILL_INIT);
recptr = XLogInsert(RM_HASH_ID, XLOG_HASH_INIT_META_PAGE);
PageSetLSN(BufferGetPage(metabuf), recptr);
}
num_buckets = metap->hashm_maxbucket + 1;
/*
* Release buffer lock on the metapage while we initialize buckets.
* Otherwise, we'll be in interrupt holdoff and the CHECK_FOR_INTERRUPTS
* won't accomplish anything. It's a bad idea to hold buffer locks for
* long intervals in any case, since that can block the bgwriter.
*/
LockBuffer(metabuf, BUFFER_LOCK_UNLOCK);
/*
* Initialize and WAL Log the first N buckets
*/
for (i = 0; i < num_buckets; i++)
{
BlockNumber blkno;
/* Allow interrupts, in case N is huge */
CHECK_FOR_INTERRUPTS();
blkno = BUCKET_TO_BLKNO(metap, i);
buf = _hash_getnewbuf(rel, blkno, forkNum);
_hash_initbuf(buf, metap->hashm_maxbucket, i, LH_BUCKET_PAGE, false);
MarkBufferDirty(buf);
if (use_wal)
log_newpage(&rel->rd_node,
forkNum,
blkno,
BufferGetPage(buf),
true);
_hash_relbuf(rel, buf);
}
/* Now reacquire buffer lock on metapage */
LockBuffer(metabuf, BUFFER_LOCK_EXCLUSIVE);
/*
* Initialize bitmap page
*/
bitmapbuf = _hash_getnewbuf(rel, num_buckets + 1, forkNum);
_hash_initbitmapbuffer(bitmapbuf, metap->hashm_bmsize, false);
MarkBufferDirty(bitmapbuf);
/* add the new bitmap page to the metapage's list of bitmaps */
/* metapage already has a write lock */
if (metap->hashm_nmaps >= HASH_MAX_BITMAPS)
ereport(ERROR,
(errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED),
errmsg("out of overflow pages in hash index \"%s\"",
RelationGetRelationName(rel))));
metap->hashm_mapp[metap->hashm_nmaps] = num_buckets + 1;
metap->hashm_nmaps++;
MarkBufferDirty(metabuf);
/* XLOG stuff */
if (use_wal)
{
xl_hash_init_bitmap_page xlrec;
XLogRecPtr recptr;
xlrec.bmsize = metap->hashm_bmsize;
XLogBeginInsert();
XLogRegisterData((char *) &xlrec, SizeOfHashInitBitmapPage);
XLogRegisterBuffer(0, bitmapbuf, REGBUF_WILL_INIT);
/*
* This is safe only because nobody else can be modifying the index at
* this stage; it's only visible to the transaction that is creating
* it.
*/
XLogRegisterBuffer(1, metabuf, REGBUF_STANDARD);
recptr = XLogInsert(RM_HASH_ID, XLOG_HASH_INIT_BITMAP_PAGE);
PageSetLSN(BufferGetPage(bitmapbuf), recptr);
PageSetLSN(BufferGetPage(metabuf), recptr);
}
/* all done */
_hash_relbuf(rel, bitmapbuf);
_hash_relbuf(rel, metabuf);
return num_buckets;
}
/*
* 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);
}
/*
* Attempt to expand the hash table by creating one new bucket.
*
* This will silently do nothing if we don't get cleanup lock on old or
* new bucket.
*
* Complete the pending splits and remove the tuples from old bucket,
* if there are any left over from the previous split.
*
* The caller must hold a pin, but no lock, on the metapage buffer.
* The buffer is returned in the same state.
*/
void
_hash_expandtable(Relation rel, Buffer metabuf)
{
HashMetaPage metap;
Bucket old_bucket;
Bucket new_bucket;
uint32 spare_ndx;
BlockNumber start_oblkno;
BlockNumber start_nblkno;
Buffer buf_nblkno;
Buffer buf_oblkno;
Page opage;
Page npage;
HashPageOpaque oopaque;
HashPageOpaque nopaque;
uint32 maxbucket;
uint32 highmask;
uint32 lowmask;
bool metap_update_masks = false;
bool metap_update_splitpoint = false;
restart_expand:
/*
* Write-lock the meta page. It used to be necessary to acquire a
* heavyweight lock to begin a split, but that is no longer required.
*/
LockBuffer(metabuf, BUFFER_LOCK_EXCLUSIVE);
_hash_checkpage(rel, metabuf, LH_META_PAGE);
metap = HashPageGetMeta(BufferGetPage(metabuf));
/*
* Check to see if split is still needed; someone else might have already
* done one while we waited for the lock.
*
* Make sure this stays in sync with _hash_doinsert()
*/
if (metap->hashm_ntuples <=
(double) metap->hashm_ffactor * (metap->hashm_maxbucket + 1))
goto fail;
/*
* Can't split anymore if maxbucket has reached its maximum possible
* value.
*
* Ideally we'd allow bucket numbers up to UINT_MAX-1 (no higher because
* the calculation maxbucket+1 mustn't overflow). Currently we restrict
* to half that because of overflow looping in _hash_log2() and
* insufficient space in hashm_spares[]. It's moot anyway because an
* index with 2^32 buckets would certainly overflow BlockNumber and hence
* _hash_alloc_buckets() would fail, but if we supported buckets smaller
* than a disk block then this would be an independent constraint.
*
* If you change this, see also the maximum initial number of buckets in
* _hash_init().
*/
if (metap->hashm_maxbucket >= (uint32) 0x7FFFFFFE)
goto fail;
/*
* Determine which bucket is to be split, and attempt to take cleanup lock
* on the old bucket. If we can't get the lock, give up.
*
* The cleanup lock protects us not only against other backends, but
* against our own backend as well.
*
* The cleanup lock is mainly to protect the split from concurrent
* inserts. See src/backend/access/hash/README, Lock Definitions for
* further details. Due to this locking restriction, if there is any
* pending scan, the split will give up which is not good, but harmless.
*/
new_bucket = metap->hashm_maxbucket + 1;
old_bucket = (new_bucket & metap->hashm_lowmask);
start_oblkno = BUCKET_TO_BLKNO(metap, old_bucket);
buf_oblkno = _hash_getbuf_with_condlock_cleanup(rel, start_oblkno, LH_BUCKET_PAGE);
if (!buf_oblkno)
goto fail;
opage = BufferGetPage(buf_oblkno);
oopaque = (HashPageOpaque) PageGetSpecialPointer(opage);
/*
* We want to finish the split from a bucket as there is no apparent
* benefit by not doing so and it will make the code complicated to finish
* the split that involves multiple buckets considering the case where new
* split also fails. We don't need to consider the new bucket for
* completing the split here as it is not possible that a re-split of new
* bucket starts when there is still a pending split from old bucket.
*/
if (H_BUCKET_BEING_SPLIT(oopaque))
{
/*
* Copy bucket mapping info now; refer the comment in code below where
* we copy this information before calling _hash_splitbucket to see
* why this is okay.
*/
maxbucket = metap->hashm_maxbucket;
highmask = metap->hashm_highmask;
lowmask = metap->hashm_lowmask;
/*
* Release the lock on metapage and old_bucket, before completing the
* split.
*/
LockBuffer(metabuf, BUFFER_LOCK_UNLOCK);
LockBuffer(buf_oblkno, BUFFER_LOCK_UNLOCK);
_hash_finish_split(rel, metabuf, buf_oblkno, old_bucket, maxbucket,
highmask, lowmask);
/* release the pin on old buffer and retry for expand. */
_hash_dropbuf(rel, buf_oblkno);
goto restart_expand;
}
/*
* Clean the tuples remained from the previous split. This operation
* requires cleanup lock and we already have one on the old bucket, so
* let's do it. We also don't want to allow further splits from the bucket
* till the garbage of previous split is cleaned. This has two
* advantages; first, it helps in avoiding the bloat due to garbage and
* second is, during cleanup of bucket, we are always sure that the
* garbage tuples belong to most recently split bucket. On the contrary,
* if we allow cleanup of bucket after meta page is updated to indicate
* the new split and before the actual split, the cleanup operation won't
* be able to decide whether the tuple has been moved to the newly created
* bucket and ended up deleting such tuples.
*/
if (H_NEEDS_SPLIT_CLEANUP(oopaque))
{
/*
* Copy bucket mapping info now; refer to the comment in code below
* where we copy this information before calling _hash_splitbucket to
* see why this is okay.
*/
maxbucket = metap->hashm_maxbucket;
highmask = metap->hashm_highmask;
lowmask = metap->hashm_lowmask;
/* Release the metapage lock. */
LockBuffer(metabuf, BUFFER_LOCK_UNLOCK);
hashbucketcleanup(rel, old_bucket, buf_oblkno, start_oblkno, NULL,
maxbucket, highmask, lowmask, NULL, NULL, true,
NULL, NULL);
_hash_dropbuf(rel, buf_oblkno);
goto restart_expand;
}
/*
* There shouldn't be any active scan on new bucket.
*
* Note: it is safe to compute the new bucket's blkno here, even though we
* may still need to update the BUCKET_TO_BLKNO mapping. This is because
* the current value of hashm_spares[hashm_ovflpoint] correctly shows
* where we are going to put a new splitpoint's worth of buckets.
*/
start_nblkno = BUCKET_TO_BLKNO(metap, new_bucket);
/*
* If the split point is increasing we need to allocate a new batch of
* bucket pages.
*/
spare_ndx = _hash_spareindex(new_bucket + 1);
if (spare_ndx > metap->hashm_ovflpoint)
{
uint32 buckets_to_add;
Assert(spare_ndx == metap->hashm_ovflpoint + 1);
/*
* We treat allocation of buckets as a separate WAL-logged action.
* Even if we fail after this operation, won't leak bucket pages;
* rather, the next split will consume this space. In any case, even
* without failure we don't use all the space in one split operation.
*/
buckets_to_add = _hash_get_totalbuckets(spare_ndx) - new_bucket;
if (!_hash_alloc_buckets(rel, start_nblkno, buckets_to_add))
{
/* can't split due to BlockNumber overflow */
_hash_relbuf(rel, buf_oblkno);
goto fail;
}
}
/*
* Physically allocate the new bucket's primary page. We want to do this
* before changing the metapage's mapping info, in case we can't get the
* disk space. Ideally, we don't need to check for cleanup lock on new
* bucket as no other backend could find this bucket unless meta page is
* updated. However, it is good to be consistent with old bucket locking.
*/
buf_nblkno = _hash_getnewbuf(rel, start_nblkno, MAIN_FORKNUM);
if (!IsBufferCleanupOK(buf_nblkno))
{
_hash_relbuf(rel, buf_oblkno);
_hash_relbuf(rel, buf_nblkno);
goto fail;
}
/*
* Since we are scribbling on the pages in the shared buffers, establish a
* critical section. Any failure in this next code leaves us with a big
* problem: the metapage is effectively corrupt but could get written back
* to disk.
*/
START_CRIT_SECTION();
/*
* Okay to proceed with split. Update the metapage bucket mapping info.
*/
metap->hashm_maxbucket = new_bucket;
if (new_bucket > metap->hashm_highmask)
{
/* Starting a new doubling */
metap->hashm_lowmask = metap->hashm_highmask;
metap->hashm_highmask = new_bucket | metap->hashm_lowmask;
metap_update_masks = true;
}
/*
* If the split point is increasing we need to adjust the hashm_spares[]
* array and hashm_ovflpoint so that future overflow pages will be created
* beyond this new batch of bucket pages.
*/
if (spare_ndx > metap->hashm_ovflpoint)
{
metap->hashm_spares[spare_ndx] = metap->hashm_spares[metap->hashm_ovflpoint];
metap->hashm_ovflpoint = spare_ndx;
metap_update_splitpoint = true;
}
MarkBufferDirty(metabuf);
/*
* Copy bucket mapping info now; this saves re-accessing the meta page
* inside _hash_splitbucket's inner loop. Note that once we drop the
* split lock, other splits could begin, so these values might be out of
* date before _hash_splitbucket finishes. That's okay, since all it
* needs is to tell which of these two buckets to map hashkeys into.
*/
maxbucket = metap->hashm_maxbucket;
highmask = metap->hashm_highmask;
lowmask = metap->hashm_lowmask;
opage = BufferGetPage(buf_oblkno);
oopaque = (HashPageOpaque) PageGetSpecialPointer(opage);
/*
* Mark the old bucket to indicate that split is in progress. (At
* operation end, we will clear the split-in-progress flag.) Also, for a
* primary bucket page, hasho_prevblkno stores the number of buckets that
* existed as of the last split, so we must update that value here.
*/
oopaque->hasho_flag |= LH_BUCKET_BEING_SPLIT;
oopaque->hasho_prevblkno = maxbucket;
MarkBufferDirty(buf_oblkno);
npage = BufferGetPage(buf_nblkno);
/*
* initialize the new bucket's primary page and mark it to indicate that
* split is in progress.
*/
nopaque = (HashPageOpaque) PageGetSpecialPointer(npage);
nopaque->hasho_prevblkno = maxbucket;
nopaque->hasho_nextblkno = InvalidBlockNumber;
nopaque->hasho_bucket = new_bucket;
nopaque->hasho_flag = LH_BUCKET_PAGE | LH_BUCKET_BEING_POPULATED;
nopaque->hasho_page_id = HASHO_PAGE_ID;
MarkBufferDirty(buf_nblkno);
/* XLOG stuff */
if (RelationNeedsWAL(rel))
{
xl_hash_split_allocate_page xlrec;
XLogRecPtr recptr;
xlrec.new_bucket = maxbucket;
xlrec.old_bucket_flag = oopaque->hasho_flag;
xlrec.new_bucket_flag = nopaque->hasho_flag;
xlrec.flags = 0;
XLogBeginInsert();
XLogRegisterBuffer(0, buf_oblkno, REGBUF_STANDARD);
XLogRegisterBuffer(1, buf_nblkno, REGBUF_WILL_INIT);
XLogRegisterBuffer(2, metabuf, REGBUF_STANDARD);
if (metap_update_masks)
{
xlrec.flags |= XLH_SPLIT_META_UPDATE_MASKS;
XLogRegisterBufData(2, (char *) &metap->hashm_lowmask, sizeof(uint32));
XLogRegisterBufData(2, (char *) &metap->hashm_highmask, sizeof(uint32));
}
if (metap_update_splitpoint)
{
xlrec.flags |= XLH_SPLIT_META_UPDATE_SPLITPOINT;
XLogRegisterBufData(2, (char *) &metap->hashm_ovflpoint,
sizeof(uint32));
XLogRegisterBufData(2,
(char *) &metap->hashm_spares[metap->hashm_ovflpoint],
sizeof(uint32));
}
XLogRegisterData((char *) &xlrec, SizeOfHashSplitAllocPage);
recptr = XLogInsert(RM_HASH_ID, XLOG_HASH_SPLIT_ALLOCATE_PAGE);
PageSetLSN(BufferGetPage(buf_oblkno), recptr);
PageSetLSN(BufferGetPage(buf_nblkno), recptr);
PageSetLSN(BufferGetPage(metabuf), recptr);
}
END_CRIT_SECTION();
/* drop lock, but keep pin */
LockBuffer(metabuf, BUFFER_LOCK_UNLOCK);
/* Relocate records to the new bucket */
_hash_splitbucket(rel, metabuf,
old_bucket, new_bucket,
buf_oblkno, buf_nblkno, NULL,
maxbucket, highmask, lowmask);
/* all done, now release the pins on primary buckets. */
_hash_dropbuf(rel, buf_oblkno);
_hash_dropbuf(rel, buf_nblkno);
return;
/* Here if decide not to split or fail to acquire old bucket lock */
fail:
/* We didn't write the metapage, so just drop lock */
LockBuffer(metabuf, BUFFER_LOCK_UNLOCK);
}
/*
* Build an SP-GiST index.
*/
Datum
spgbuild(PG_FUNCTION_ARGS)
{
Relation heap = (Relation) PG_GETARG_POINTER(0);
Relation index = (Relation) PG_GETARG_POINTER(1);
IndexInfo *indexInfo = (IndexInfo *) PG_GETARG_POINTER(2);
IndexBuildResult *result;
double reltuples;
SpGistBuildState buildstate;
Buffer metabuffer,
rootbuffer,
nullbuffer;
if (RelationGetNumberOfBlocks(index) != 0)
elog(ERROR, "index \"%s\" already contains data",
RelationGetRelationName(index));
/*
* Initialize the meta page and root pages
*/
metabuffer = SpGistNewBuffer(index);
rootbuffer = SpGistNewBuffer(index);
nullbuffer = SpGistNewBuffer(index);
Assert(BufferGetBlockNumber(metabuffer) == SPGIST_METAPAGE_BLKNO);
Assert(BufferGetBlockNumber(rootbuffer) == SPGIST_ROOT_BLKNO);
Assert(BufferGetBlockNumber(nullbuffer) == SPGIST_NULL_BLKNO);
START_CRIT_SECTION();
SpGistInitMetapage(BufferGetPage(metabuffer));
MarkBufferDirty(metabuffer);
SpGistInitBuffer(rootbuffer, SPGIST_LEAF);
MarkBufferDirty(rootbuffer);
SpGistInitBuffer(nullbuffer, SPGIST_LEAF | SPGIST_NULLS);
MarkBufferDirty(nullbuffer);
if (RelationNeedsWAL(index))
{
XLogRecPtr recptr;
XLogRecData rdata;
/* WAL data is just the relfilenode */
rdata.data = (char *) &(index->rd_node);
rdata.len = sizeof(RelFileNode);
rdata.buffer = InvalidBuffer;
rdata.next = NULL;
recptr = XLogInsert(RM_SPGIST_ID, XLOG_SPGIST_CREATE_INDEX, &rdata);
PageSetLSN(BufferGetPage(metabuffer), recptr);
PageSetLSN(BufferGetPage(rootbuffer), recptr);
PageSetLSN(BufferGetPage(nullbuffer), recptr);
}
END_CRIT_SECTION();
UnlockReleaseBuffer(metabuffer);
UnlockReleaseBuffer(rootbuffer);
UnlockReleaseBuffer(nullbuffer);
/*
* Now insert all the heap data into the index
*/
initSpGistState(&buildstate.spgstate, index);
buildstate.spgstate.isBuild = true;
buildstate.tmpCtx = AllocSetContextCreate(CurrentMemoryContext,
"SP-GiST build temporary context",
ALLOCSET_DEFAULT_MINSIZE,
ALLOCSET_DEFAULT_INITSIZE,
ALLOCSET_DEFAULT_MAXSIZE);
reltuples = IndexBuildHeapScan(heap, index, indexInfo, true,
spgistBuildCallback, (void *) &buildstate);
MemoryContextDelete(buildstate.tmpCtx);
SpGistUpdateMetaPage(index);
result = (IndexBuildResult *) palloc0(sizeof(IndexBuildResult));
result->heap_tuples = result->index_tuples = reltuples;
PG_RETURN_POINTER(result);
}
/*
* Build an SP-GiST index.
*/
IndexBuildResult *
spgbuild(Relation heap, Relation index, IndexInfo *indexInfo)
{
IndexBuildResult *result;
double reltuples;
SpGistBuildState buildstate;
Buffer metabuffer,
rootbuffer,
nullbuffer;
if (RelationGetNumberOfBlocks(index) != 0)
elog(ERROR, "index \"%s\" already contains data",
RelationGetRelationName(index));
/*
* Initialize the meta page and root pages
*/
metabuffer = SpGistNewBuffer(index);
rootbuffer = SpGistNewBuffer(index);
nullbuffer = SpGistNewBuffer(index);
Assert(BufferGetBlockNumber(metabuffer) == SPGIST_METAPAGE_BLKNO);
Assert(BufferGetBlockNumber(rootbuffer) == SPGIST_ROOT_BLKNO);
Assert(BufferGetBlockNumber(nullbuffer) == SPGIST_NULL_BLKNO);
START_CRIT_SECTION();
SpGistInitMetapage(BufferGetPage(metabuffer));
MarkBufferDirty(metabuffer);
SpGistInitBuffer(rootbuffer, SPGIST_LEAF);
MarkBufferDirty(rootbuffer);
SpGistInitBuffer(nullbuffer, SPGIST_LEAF | SPGIST_NULLS);
MarkBufferDirty(nullbuffer);
if (RelationNeedsWAL(index))
{
XLogRecPtr recptr;
XLogBeginInsert();
/*
* Replay will re-initialize the pages, so don't take full pages
* images. No other data to log.
*/
XLogRegisterBuffer(0, metabuffer, REGBUF_WILL_INIT);
XLogRegisterBuffer(1, rootbuffer, REGBUF_WILL_INIT | REGBUF_STANDARD);
XLogRegisterBuffer(2, nullbuffer, REGBUF_WILL_INIT | REGBUF_STANDARD);
recptr = XLogInsert(RM_SPGIST_ID, XLOG_SPGIST_CREATE_INDEX);
PageSetLSN(BufferGetPage(metabuffer), recptr);
PageSetLSN(BufferGetPage(rootbuffer), recptr);
PageSetLSN(BufferGetPage(nullbuffer), recptr);
}
END_CRIT_SECTION();
UnlockReleaseBuffer(metabuffer);
UnlockReleaseBuffer(rootbuffer);
UnlockReleaseBuffer(nullbuffer);
/*
* Now insert all the heap data into the index
*/
initSpGistState(&buildstate.spgstate, index);
buildstate.spgstate.isBuild = true;
buildstate.tmpCtx = AllocSetContextCreate(CurrentMemoryContext,
"SP-GiST build temporary context",
ALLOCSET_DEFAULT_MINSIZE,
ALLOCSET_DEFAULT_INITSIZE,
ALLOCSET_DEFAULT_MAXSIZE);
reltuples = IndexBuildHeapScan(heap, index, indexInfo, true,
spgistBuildCallback, (void *) &buildstate);
MemoryContextDelete(buildstate.tmpCtx);
SpGistUpdateMetaPage(index);
result = (IndexBuildResult *) palloc0(sizeof(IndexBuildResult));
result->heap_tuples = result->index_tuples = reltuples;
return result;
}
/*
* Delete item(s) from a btree page during single-page cleanup.
*
* As above, must only be used on leaf pages.
*
* This routine assumes that the caller has pinned and locked the buffer.
* Also, the given itemnos *must* appear in increasing order in the array.
*
* This is nearly the same as _bt_delitems_vacuum as far as what it does to
* the page, but the WAL logging considerations are quite different. See
* comments for _bt_delitems_vacuum.
*/
void
_bt_delitems_delete(Relation rel, Buffer buf,
OffsetNumber *itemnos, int nitems,
Relation heapRel)
{
Page page = BufferGetPage(buf);
BTPageOpaque opaque;
/* Shouldn't be called unless there's something to do */
Assert(nitems > 0);
/* No ereport(ERROR) until changes are logged */
START_CRIT_SECTION();
/* Fix the page */
PageIndexMultiDelete(page, itemnos, nitems);
/*
* Unlike _bt_delitems_vacuum, we *must not* clear the vacuum cycle ID,
* because this is not called by VACUUM.
*/
/*
* Mark the page as not containing any LP_DEAD items. This is not
* certainly true (there might be some that have recently been marked, but
* weren't included in our target-item list), but it will almost always be
* true and it doesn't seem worth an additional page scan to check it.
* Remember that BTP_HAS_GARBAGE is only a hint anyway.
*/
opaque = (BTPageOpaque) PageGetSpecialPointer(page);
opaque->btpo_flags &= ~BTP_HAS_GARBAGE;
MarkBufferDirty(buf);
/* XLOG stuff */
if (RelationNeedsWAL(rel))
{
XLogRecPtr recptr;
XLogRecData rdata[3];
xl_btree_delete xlrec_delete;
xlrec_delete.node = rel->rd_node;
xlrec_delete.hnode = heapRel->rd_node;
xlrec_delete.block = BufferGetBlockNumber(buf);
xlrec_delete.nitems = nitems;
rdata[0].data = (char *) &xlrec_delete;
rdata[0].len = SizeOfBtreeDelete;
rdata[0].buffer = InvalidBuffer;
rdata[0].next = &(rdata[1]);
/*
* We need the target-offsets array whether or not we store the whole
* buffer, to allow us to find the latestRemovedXid on a standby
* server.
*/
rdata[1].data = (char *) itemnos;
rdata[1].len = nitems * sizeof(OffsetNumber);
rdata[1].buffer = InvalidBuffer;
rdata[1].next = &(rdata[2]);
rdata[2].data = NULL;
rdata[2].len = 0;
rdata[2].buffer = buf;
rdata[2].buffer_std = true;
rdata[2].next = NULL;
recptr = XLogInsert(RM_BTREE_ID, XLOG_BTREE_DELETE, rdata);
PageSetLSN(page, recptr);
PageSetTLI(page, ThisTimeLineID);
}
END_CRIT_SECTION();
}
/*
* Place tuples from 'itup' to 'buffer'. If 'oldoffnum' is valid, the tuple
* at that offset is atomically removed along with inserting the new tuples.
* This is used to replace a tuple with a new one.
*
* If 'leftchildbuf' is valid, we're inserting the downlink for the page
* to the right of 'leftchildbuf', or updating the downlink for 'leftchildbuf'.
* F_FOLLOW_RIGHT flag on 'leftchildbuf' is cleared and NSN is set.
*
* If 'markfollowright' is true and the page is split, the left child is
* marked with F_FOLLOW_RIGHT flag. That is the normal case. During buffered
* index build, however, there is no concurrent access and the page splitting
* is done in a slightly simpler fashion, and false is passed.
*
* If there is not enough room on the page, it is split. All the split
* pages are kept pinned and locked and returned in *splitinfo, the caller
* is responsible for inserting the downlinks for them. However, if
* 'buffer' is the root page and it needs to be split, gistplacetopage()
* performs the split as one atomic operation, and *splitinfo is set to NIL.
* In that case, we continue to hold the root page locked, and the child
* pages are released; note that new tuple(s) are *not* on the root page
* but in one of the new child pages.
*
* If 'newblkno' is not NULL, returns the block number of page the first
* new/updated tuple was inserted to. Usually it's the given page, but could
* be its right sibling if the page was split.
*
* Returns 'true' if the page was split, 'false' otherwise.
*/
bool
gistplacetopage(Relation rel, Size freespace, GISTSTATE *giststate,
Buffer buffer,
IndexTuple *itup, int ntup, OffsetNumber oldoffnum,
BlockNumber *newblkno,
Buffer leftchildbuf,
List **splitinfo,
bool markfollowright)
{
BlockNumber blkno = BufferGetBlockNumber(buffer);
Page page = BufferGetPage(buffer);
bool is_leaf = (GistPageIsLeaf(page)) ? true : false;
XLogRecPtr recptr;
int i;
bool is_split;
/*
* Refuse to modify a page that's incompletely split. This should not
* happen because we finish any incomplete splits while we walk down the
* tree. However, it's remotely possible that another concurrent inserter
* splits a parent page, and errors out before completing the split. We
* will just throw an error in that case, and leave any split we had in
* progress unfinished too. The next insert that comes along will clean up
* the mess.
*/
if (GistFollowRight(page))
elog(ERROR, "concurrent GiST page split was incomplete");
*splitinfo = NIL;
/*
* if isupdate, remove old key: This node's key has been modified, either
* because a child split occurred or because we needed to adjust our key
* for an insert in a child node. Therefore, remove the old version of
* this node's key.
*
* for WAL replay, in the non-split case we handle this by setting up a
* one-element todelete array; in the split case, it's handled implicitly
* because the tuple vector passed to gistSplit won't include this tuple.
*/
is_split = gistnospace(page, itup, ntup, oldoffnum, freespace);
/*
* If leaf page is full, try at first to delete dead tuples. And then
* check again.
*/
if (is_split && GistPageIsLeaf(page) && GistPageHasGarbage(page))
{
gistvacuumpage(rel, page, buffer);
is_split = gistnospace(page, itup, ntup, oldoffnum, freespace);
}
if (is_split)
{
/* no space for insertion */
IndexTuple *itvec;
int tlen;
SplitedPageLayout *dist = NULL,
*ptr;
BlockNumber oldrlink = InvalidBlockNumber;
GistNSN oldnsn = 0;
SplitedPageLayout rootpg;
bool is_rootsplit;
int npage;
is_rootsplit = (blkno == GIST_ROOT_BLKNO);
/*
* Form index tuples vector to split. If we're replacing an old tuple,
* remove the old version from the vector.
*/
itvec = gistextractpage(page, &tlen);
if (OffsetNumberIsValid(oldoffnum))
{
/* on inner page we should remove old tuple */
int pos = oldoffnum - FirstOffsetNumber;
tlen--;
if (pos != tlen)
memmove(itvec + pos, itvec + pos + 1, sizeof(IndexTuple) * (tlen - pos));
}
itvec = gistjoinvector(itvec, &tlen, itup, ntup);
dist = gistSplit(rel, page, itvec, tlen, giststate);
/*
* Check that split didn't produce too many pages.
*/
npage = 0;
for (ptr = dist; ptr; ptr = ptr->next)
npage++;
/* in a root split, we'll add one more page to the list below */
if (is_rootsplit)
npage++;
if (npage > GIST_MAX_SPLIT_PAGES)
elog(ERROR, "GiST page split into too many halves (%d, maximum %d)",
npage, GIST_MAX_SPLIT_PAGES);
/*
* Set up pages to work with. Allocate new buffers for all but the
* leftmost page. The original page becomes the new leftmost page, and
* is just replaced with the new contents.
*
* For a root-split, allocate new buffers for all child pages, the
* original page is overwritten with new root page containing
* downlinks to the new child pages.
*/
ptr = dist;
if (!is_rootsplit)
{
/* save old rightlink and NSN */
oldrlink = GistPageGetOpaque(page)->rightlink;
oldnsn = GistPageGetNSN(page);
dist->buffer = buffer;
dist->block.blkno = BufferGetBlockNumber(buffer);
dist->page = PageGetTempPageCopySpecial(BufferGetPage(buffer));
/* clean all flags except F_LEAF */
GistPageGetOpaque(dist->page)->flags = (is_leaf) ? F_LEAF : 0;
ptr = ptr->next;
}
for (; ptr; ptr = ptr->next)
{
/* Allocate new page */
ptr->buffer = gistNewBuffer(rel);
GISTInitBuffer(ptr->buffer, (is_leaf) ? F_LEAF : 0);
ptr->page = BufferGetPage(ptr->buffer);
ptr->block.blkno = BufferGetBlockNumber(ptr->buffer);
}
/*
* Now that we know which blocks the new pages go to, set up downlink
* tuples to point to them.
*/
for (ptr = dist; ptr; ptr = ptr->next)
{
ItemPointerSetBlockNumber(&(ptr->itup->t_tid), ptr->block.blkno);
GistTupleSetValid(ptr->itup);
}
/*
* If this is a root split, we construct the new root page with the
* downlinks here directly, instead of requiring the caller to insert
* them. Add the new root page to the list along with the child pages.
*/
if (is_rootsplit)
{
IndexTuple *downlinks;
int ndownlinks = 0;
int i;
rootpg.buffer = buffer;
rootpg.page = PageGetTempPageCopySpecial(BufferGetPage(rootpg.buffer));
GistPageGetOpaque(rootpg.page)->flags = 0;
/* Prepare a vector of all the downlinks */
for (ptr = dist; ptr; ptr = ptr->next)
ndownlinks++;
downlinks = palloc(sizeof(IndexTuple) * ndownlinks);
for (i = 0, ptr = dist; ptr; ptr = ptr->next)
downlinks[i++] = ptr->itup;
rootpg.block.blkno = GIST_ROOT_BLKNO;
rootpg.block.num = ndownlinks;
rootpg.list = gistfillitupvec(downlinks, ndownlinks,
&(rootpg.lenlist));
rootpg.itup = NULL;
rootpg.next = dist;
dist = &rootpg;
}
else
{
/* Prepare split-info to be returned to caller */
for (ptr = dist; ptr; ptr = ptr->next)
{
GISTPageSplitInfo *si = palloc(sizeof(GISTPageSplitInfo));
si->buf = ptr->buffer;
si->downlink = ptr->itup;
*splitinfo = lappend(*splitinfo, si);
}
}
/*
* Fill all pages. All the pages are new, ie. freshly allocated empty
* pages, or a temporary copy of the old page.
*/
for (ptr = dist; ptr; ptr = ptr->next)
{
char *data = (char *) (ptr->list);
for (i = 0; i < ptr->block.num; i++)
{
IndexTuple thistup = (IndexTuple) data;
if (PageAddItem(ptr->page, (Item) data, IndexTupleSize(thistup), i + FirstOffsetNumber, false, false) == InvalidOffsetNumber)
elog(ERROR, "failed to add item to index page in \"%s\"", RelationGetRelationName(rel));
/*
* If this is the first inserted/updated tuple, let the caller
* know which page it landed on.
*/
if (newblkno && ItemPointerEquals(&thistup->t_tid, &(*itup)->t_tid))
*newblkno = ptr->block.blkno;
data += IndexTupleSize(thistup);
}
/* Set up rightlinks */
if (ptr->next && ptr->block.blkno != GIST_ROOT_BLKNO)
GistPageGetOpaque(ptr->page)->rightlink =
ptr->next->block.blkno;
else
GistPageGetOpaque(ptr->page)->rightlink = oldrlink;
/*
* Mark the all but the right-most page with the follow-right
* flag. It will be cleared as soon as the downlink is inserted
* into the parent, but this ensures that if we error out before
* that, the index is still consistent. (in buffering build mode,
* any error will abort the index build anyway, so this is not
* needed.)
*/
if (ptr->next && !is_rootsplit && markfollowright)
GistMarkFollowRight(ptr->page);
else
GistClearFollowRight(ptr->page);
/*
* Copy the NSN of the original page to all pages. The
* F_FOLLOW_RIGHT flags ensure that scans will follow the
* rightlinks until the downlinks are inserted.
*/
GistPageSetNSN(ptr->page, oldnsn);
}
/*
* gistXLogSplit() needs to WAL log a lot of pages, prepare WAL
* insertion for that. NB: The number of pages and data segments
* specified here must match the calculations in gistXLogSplit()!
*/
if (RelationNeedsWAL(rel))
XLogEnsureRecordSpace(npage, 1 + npage * 2);
START_CRIT_SECTION();
/*
* Must mark buffers dirty before XLogInsert, even though we'll still
* be changing their opaque fields below.
*/
for (ptr = dist; ptr; ptr = ptr->next)
MarkBufferDirty(ptr->buffer);
if (BufferIsValid(leftchildbuf))
MarkBufferDirty(leftchildbuf);
/*
* The first page in the chain was a temporary working copy meant to
* replace the old page. Copy it over the old page.
*/
PageRestoreTempPage(dist->page, BufferGetPage(dist->buffer));
dist->page = BufferGetPage(dist->buffer);
/* Write the WAL record */
if (RelationNeedsWAL(rel))
recptr = gistXLogSplit(is_leaf,
dist, oldrlink, oldnsn, leftchildbuf,
markfollowright);
else
recptr = gistGetFakeLSN(rel);
for (ptr = dist; ptr; ptr = ptr->next)
{
PageSetLSN(ptr->page, recptr);
}
/*
* Return the new child buffers to the caller.
*
* If this was a root split, we've already inserted the downlink
* pointers, in the form of a new root page. Therefore we can release
* all the new buffers, and keep just the root page locked.
*/
if (is_rootsplit)
{
for (ptr = dist->next; ptr; ptr = ptr->next)
UnlockReleaseBuffer(ptr->buffer);
}
}
else
{
/*
* Enough space. We always get here if ntup==0.
*/
START_CRIT_SECTION();
/*
* Delete old tuple if any, then insert new tuple(s) if any. If
* possible, use the fast path of PageIndexTupleOverwrite.
*/
if (OffsetNumberIsValid(oldoffnum))
{
if (ntup == 1)
{
/* One-for-one replacement, so use PageIndexTupleOverwrite */
if (!PageIndexTupleOverwrite(page, oldoffnum, (Item) *itup,
IndexTupleSize(*itup)))
elog(ERROR, "failed to add item to index page in \"%s\"",
RelationGetRelationName(rel));
}
else
{
/* Delete old, then append new tuple(s) to page */
PageIndexTupleDelete(page, oldoffnum);
gistfillbuffer(page, itup, ntup, InvalidOffsetNumber);
}
}
else
{
/* Just append new tuples at the end of the page */
gistfillbuffer(page, itup, ntup, InvalidOffsetNumber);
}
MarkBufferDirty(buffer);
if (BufferIsValid(leftchildbuf))
MarkBufferDirty(leftchildbuf);
if (RelationNeedsWAL(rel))
{
OffsetNumber ndeloffs = 0,
deloffs[1];
if (OffsetNumberIsValid(oldoffnum))
{
deloffs[0] = oldoffnum;
ndeloffs = 1;
}
recptr = gistXLogUpdate(buffer,
deloffs, ndeloffs, itup, ntup,
leftchildbuf);
PageSetLSN(page, recptr);
}
else
{
recptr = gistGetFakeLSN(rel);
PageSetLSN(page, recptr);
}
if (newblkno)
*newblkno = blkno;
}
/*
* If we inserted the downlink for a child page, set NSN and clear
* F_FOLLOW_RIGHT flag on the left child, so that concurrent scans know to
* follow the rightlink if and only if they looked at the parent page
* before we inserted the downlink.
*
* Note that we do this *after* writing the WAL record. That means that
* the possible full page image in the WAL record does not include these
* changes, and they must be replayed even if the page is restored from
* the full page image. There's a chicken-and-egg problem: if we updated
* the child pages first, we wouldn't know the recptr of the WAL record
* we're about to write.
*/
if (BufferIsValid(leftchildbuf))
{
Page leftpg = BufferGetPage(leftchildbuf);
GistPageSetNSN(leftpg, recptr);
GistClearFollowRight(leftpg);
PageSetLSN(leftpg, recptr);
}
END_CRIT_SECTION();
return is_split;
}
/*
* _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;
}
/*
* Insert an index tuple into the index relation. The revmap is updated to
* mark the range containing the given page as pointing to the inserted entry.
* A WAL record is written.
*
* The buffer, if valid, is first checked for free space to insert the new
* entry; if there isn't enough, a new buffer is obtained and pinned. No
* buffer lock must be held on entry, no buffer lock is held on exit.
*
* Return value is the offset number where the tuple was inserted.
*/
OffsetNumber
brin_doinsert(Relation idxrel, BlockNumber pagesPerRange,
BrinRevmap *revmap, Buffer *buffer, BlockNumber heapBlk,
BrinTuple *tup, Size itemsz)
{
Page page;
BlockNumber blk;
OffsetNumber off;
Buffer revmapbuf;
ItemPointerData tid;
bool extended;
Assert(itemsz == MAXALIGN(itemsz));
/* If the item is oversized, don't even bother. */
if (itemsz > BrinMaxItemSize)
{
ereport(ERROR,
(errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED),
errmsg("index row size %lu exceeds maximum %lu for index \"%s\"",
(unsigned long) itemsz,
(unsigned long) BrinMaxItemSize,
RelationGetRelationName(idxrel))));
return InvalidOffsetNumber; /* keep compiler quiet */
}
/* Make sure the revmap is long enough to contain the entry we need */
brinRevmapExtend(revmap, heapBlk);
/*
* Acquire lock on buffer supplied by caller, if any. If it doesn't have
* enough space, unpin it to obtain a new one below.
*/
if (BufferIsValid(*buffer))
{
/*
* It's possible that another backend (or ourselves!) extended the
* revmap over the page we held a pin on, so we cannot assume that
* it's still a regular page.
*/
LockBuffer(*buffer, BUFFER_LOCK_EXCLUSIVE);
if (br_page_get_freespace(BufferGetPage(*buffer)) < itemsz)
{
UnlockReleaseBuffer(*buffer);
*buffer = InvalidBuffer;
}
}
/*
* If we still don't have a usable buffer, have brin_getinsertbuffer
* obtain one for us.
*/
if (!BufferIsValid(*buffer))
{
do
*buffer = brin_getinsertbuffer(idxrel, InvalidBuffer, itemsz, &extended);
while (!BufferIsValid(*buffer));
}
else
extended = false;
/* Now obtain lock on revmap buffer */
revmapbuf = brinLockRevmapPageForUpdate(revmap, heapBlk);
page = BufferGetPage(*buffer);
blk = BufferGetBlockNumber(*buffer);
/* Execute the actual insertion */
START_CRIT_SECTION();
if (extended)
brin_page_init(BufferGetPage(*buffer), BRIN_PAGETYPE_REGULAR);
off = PageAddItem(page, (Item) tup, itemsz, InvalidOffsetNumber,
false, false);
if (off == InvalidOffsetNumber)
elog(ERROR, "could not insert new index tuple to page");
MarkBufferDirty(*buffer);
BRIN_elog((DEBUG2, "inserted tuple (%u,%u) for range starting at %u",
blk, off, heapBlk));
ItemPointerSet(&tid, blk, off);
brinSetHeapBlockItemptr(revmapbuf, pagesPerRange, heapBlk, tid);
MarkBufferDirty(revmapbuf);
/* XLOG stuff */
if (RelationNeedsWAL(idxrel))
{
xl_brin_insert xlrec;
XLogRecPtr recptr;
uint8 info;
info = XLOG_BRIN_INSERT | (extended ? XLOG_BRIN_INIT_PAGE : 0);
xlrec.heapBlk = heapBlk;
xlrec.pagesPerRange = pagesPerRange;
xlrec.offnum = off;
XLogBeginInsert();
XLogRegisterData((char *) &xlrec, SizeOfBrinInsert);
XLogRegisterBuffer(0, *buffer, REGBUF_STANDARD | (extended ? REGBUF_WILL_INIT : 0));
XLogRegisterBufData(0, (char *) tup, itemsz);
XLogRegisterBuffer(1, revmapbuf, 0);
recptr = XLogInsert(RM_BRIN_ID, info);
PageSetLSN(page, recptr);
PageSetLSN(BufferGetPage(revmapbuf), recptr);
}
END_CRIT_SECTION();
/* Tuple is firmly on buffer; we can release our locks */
LockBuffer(*buffer, BUFFER_LOCK_UNLOCK);
LockBuffer(revmapbuf, BUFFER_LOCK_UNLOCK);
if (extended)
FreeSpaceMapVacuum(idxrel);
return off;
}
/*
* _hash_freeovflpage() -
*
* Remove this overflow page from its bucket's chain, and mark the page as
* free. On entry, ovflbuf is write-locked; it is released before exiting.
*
* Add the tuples (itups) to wbuf in this function. We could do that in the
* caller as well, but the advantage of doing it here is we can easily write
* the WAL for XLOG_HASH_SQUEEZE_PAGE operation. Addition of tuples and
* removal of overflow page has to done as an atomic operation, otherwise
* during replay on standby users might find duplicate records.
*
* Since this function is invoked in VACUUM, we provide an access strategy
* parameter that controls fetches of the bucket pages.
*
* Returns the block number of the page that followed the given page
* in the bucket, or InvalidBlockNumber if no following page.
*
* NB: caller must not hold lock on metapage, nor on page, that's next to
* ovflbuf in the bucket chain. We don't acquire the lock on page that's
* prior to ovflbuf in chain if it is same as wbuf because the caller already
* has a lock on same.
*/
BlockNumber
_hash_freeovflpage(Relation rel, Buffer bucketbuf, Buffer ovflbuf,
Buffer wbuf, IndexTuple *itups, OffsetNumber *itup_offsets,
Size *tups_size, uint16 nitups,
BufferAccessStrategy bstrategy)
{
HashMetaPage metap;
Buffer metabuf;
Buffer mapbuf;
BlockNumber ovflblkno;
BlockNumber prevblkno;
BlockNumber blkno;
BlockNumber nextblkno;
BlockNumber writeblkno;
HashPageOpaque ovflopaque;
Page ovflpage;
Page mappage;
uint32 *freep;
uint32 ovflbitno;
int32 bitmappage,
bitmapbit;
Bucket bucket PG_USED_FOR_ASSERTS_ONLY;
Buffer prevbuf = InvalidBuffer;
Buffer nextbuf = InvalidBuffer;
bool update_metap = false;
/* Get information from the doomed page */
_hash_checkpage(rel, ovflbuf, LH_OVERFLOW_PAGE);
ovflblkno = BufferGetBlockNumber(ovflbuf);
ovflpage = BufferGetPage(ovflbuf);
ovflopaque = (HashPageOpaque) PageGetSpecialPointer(ovflpage);
nextblkno = ovflopaque->hasho_nextblkno;
prevblkno = ovflopaque->hasho_prevblkno;
writeblkno = BufferGetBlockNumber(wbuf);
bucket = ovflopaque->hasho_bucket;
/*
* Fix up the bucket chain. this is a doubly-linked list, so we must fix
* up the bucket chain members behind and ahead of the overflow page being
* deleted. Concurrency issues are avoided by using lock chaining as
* described atop hashbucketcleanup.
*/
if (BlockNumberIsValid(prevblkno))
{
if (prevblkno == writeblkno)
prevbuf = wbuf;
else
prevbuf = _hash_getbuf_with_strategy(rel,
prevblkno,
HASH_WRITE,
LH_BUCKET_PAGE | LH_OVERFLOW_PAGE,
bstrategy);
}
if (BlockNumberIsValid(nextblkno))
nextbuf = _hash_getbuf_with_strategy(rel,
nextblkno,
HASH_WRITE,
LH_OVERFLOW_PAGE,
bstrategy);
/* Note: bstrategy is intentionally not used for metapage and bitmap */
/* Read the metapage so we can determine which bitmap page to use */
metabuf = _hash_getbuf(rel, HASH_METAPAGE, HASH_READ, LH_META_PAGE);
metap = HashPageGetMeta(BufferGetPage(metabuf));
/* Identify which bit to set */
ovflbitno = _hash_ovflblkno_to_bitno(metap, ovflblkno);
bitmappage = ovflbitno >> BMPG_SHIFT(metap);
bitmapbit = ovflbitno & BMPG_MASK(metap);
if (bitmappage >= metap->hashm_nmaps)
elog(ERROR, "invalid overflow bit number %u", ovflbitno);
blkno = metap->hashm_mapp[bitmappage];
/* Release metapage lock while we access the bitmap page */
LockBuffer(metabuf, BUFFER_LOCK_UNLOCK);
/* read the bitmap page to clear the bitmap bit */
mapbuf = _hash_getbuf(rel, blkno, HASH_WRITE, LH_BITMAP_PAGE);
mappage = BufferGetPage(mapbuf);
freep = HashPageGetBitmap(mappage);
Assert(ISSET(freep, bitmapbit));
/* Get write-lock on metapage to update firstfree */
LockBuffer(metabuf, BUFFER_LOCK_EXCLUSIVE);
/* This operation needs to log multiple tuples, prepare WAL for that */
if (RelationNeedsWAL(rel))
XLogEnsureRecordSpace(HASH_XLOG_FREE_OVFL_BUFS, 4 + 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).
*/
if (nitups > 0)
{
_hash_pgaddmultitup(rel, wbuf, itups, itup_offsets, nitups);
MarkBufferDirty(wbuf);
}
/*
* Reinitialize the freed overflow page. Just zeroing the page won't
* work, because WAL replay routines expect pages to be initialized. See
* explanation of RBM_NORMAL mode atop XLogReadBufferExtended. We are
* careful to make the special space valid here so that tools like
* pageinspect won't get confused.
*/
_hash_pageinit(ovflpage, BufferGetPageSize(ovflbuf));
ovflopaque = (HashPageOpaque) PageGetSpecialPointer(ovflpage);
ovflopaque->hasho_prevblkno = InvalidBlockNumber;
ovflopaque->hasho_nextblkno = InvalidBlockNumber;
ovflopaque->hasho_bucket = -1;
ovflopaque->hasho_flag = LH_UNUSED_PAGE;
ovflopaque->hasho_page_id = HASHO_PAGE_ID;
MarkBufferDirty(ovflbuf);
if (BufferIsValid(prevbuf))
{
Page prevpage = BufferGetPage(prevbuf);
HashPageOpaque prevopaque = (HashPageOpaque) PageGetSpecialPointer(prevpage);
Assert(prevopaque->hasho_bucket == bucket);
prevopaque->hasho_nextblkno = nextblkno;
MarkBufferDirty(prevbuf);
}
if (BufferIsValid(nextbuf))
{
Page nextpage = BufferGetPage(nextbuf);
HashPageOpaque nextopaque = (HashPageOpaque) PageGetSpecialPointer(nextpage);
Assert(nextopaque->hasho_bucket == bucket);
nextopaque->hasho_prevblkno = prevblkno;
MarkBufferDirty(nextbuf);
}
/* Clear the bitmap bit to indicate that this overflow page is free */
CLRBIT(freep, bitmapbit);
MarkBufferDirty(mapbuf);
/* if this is now the first free page, update hashm_firstfree */
if (ovflbitno < metap->hashm_firstfree)
{
metap->hashm_firstfree = ovflbitno;
update_metap = true;
MarkBufferDirty(metabuf);
}
/* XLOG stuff */
if (RelationNeedsWAL(rel))
{
xl_hash_squeeze_page xlrec;
XLogRecPtr recptr;
int i;
xlrec.prevblkno = prevblkno;
xlrec.nextblkno = nextblkno;
xlrec.ntups = nitups;
xlrec.is_prim_bucket_same_wrt = (wbuf == bucketbuf);
xlrec.is_prev_bucket_same_wrt = (wbuf == prevbuf);
XLogBeginInsert();
XLogRegisterData((char *) &xlrec, SizeOfHashSqueezePage);
/*
* 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, bucketbuf, REGBUF_STANDARD | REGBUF_NO_IMAGE);
XLogRegisterBuffer(1, wbuf, REGBUF_STANDARD);
if (xlrec.ntups > 0)
{
XLogRegisterBufData(1, (char *) itup_offsets,
nitups * sizeof(OffsetNumber));
for (i = 0; i < nitups; i++)
XLogRegisterBufData(1, (char *) itups[i], tups_size[i]);
}
XLogRegisterBuffer(2, ovflbuf, REGBUF_STANDARD);
/*
* If prevpage and the writepage (block in which we are moving tuples
* from overflow) are same, then no need to separately register
* prevpage. During replay, we can directly update the nextblock in
* writepage.
*/
if (BufferIsValid(prevbuf) && !xlrec.is_prev_bucket_same_wrt)
XLogRegisterBuffer(3, prevbuf, REGBUF_STANDARD);
if (BufferIsValid(nextbuf))
XLogRegisterBuffer(4, nextbuf, REGBUF_STANDARD);
XLogRegisterBuffer(5, mapbuf, REGBUF_STANDARD);
XLogRegisterBufData(5, (char *) &bitmapbit, sizeof(uint32));
if (update_metap)
{
XLogRegisterBuffer(6, metabuf, REGBUF_STANDARD);
XLogRegisterBufData(6, (char *) &metap->hashm_firstfree, sizeof(uint32));
}
recptr = XLogInsert(RM_HASH_ID, XLOG_HASH_SQUEEZE_PAGE);
PageSetLSN(BufferGetPage(wbuf), recptr);
PageSetLSN(BufferGetPage(ovflbuf), recptr);
if (BufferIsValid(prevbuf) && !xlrec.is_prev_bucket_same_wrt)
PageSetLSN(BufferGetPage(prevbuf), recptr);
if (BufferIsValid(nextbuf))
PageSetLSN(BufferGetPage(nextbuf), recptr);
PageSetLSN(BufferGetPage(mapbuf), recptr);
if (update_metap)
PageSetLSN(BufferGetPage(metabuf), recptr);
}
END_CRIT_SECTION();
/* release previous bucket if it is not same as write bucket */
if (BufferIsValid(prevbuf) && prevblkno != writeblkno)
_hash_relbuf(rel, prevbuf);
if (BufferIsValid(ovflbuf))
_hash_relbuf(rel, ovflbuf);
if (BufferIsValid(nextbuf))
_hash_relbuf(rel, nextbuf);
_hash_relbuf(rel, mapbuf);
_hash_relbuf(rel, metabuf);
return nextblkno;
}
/*
* Update tuple origtup (size origsz), located in offset oldoff of buffer
* oldbuf, to newtup (size newsz) as summary tuple for the page range starting
* at heapBlk. oldbuf must not be locked on entry, and is not locked at exit.
*
* If samepage is true, attempt to put the new tuple in the same page, but if
* there's no room, use some other one.
*
* If the update is successful, return true; the revmap is updated to point to
* the new tuple. If the update is not done for whatever reason, return false.
* Caller may retry the update if this happens.
*/
bool
brin_doupdate(Relation idxrel, BlockNumber pagesPerRange,
BrinRevmap *revmap, BlockNumber heapBlk,
Buffer oldbuf, OffsetNumber oldoff,
const BrinTuple *origtup, Size origsz,
const BrinTuple *newtup, Size newsz,
bool samepage)
{
Page oldpage;
ItemId oldlp;
BrinTuple *oldtup;
Size oldsz;
Buffer newbuf;
bool extended;
Assert(newsz == MAXALIGN(newsz));
/* If the item is oversized, don't bother. */
if (newsz > BrinMaxItemSize)
{
ereport(ERROR,
(errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED),
errmsg("index row size %lu exceeds maximum %lu for index \"%s\"",
(unsigned long) newsz,
(unsigned long) BrinMaxItemSize,
RelationGetRelationName(idxrel))));
return false; /* keep compiler quiet */
}
/* make sure the revmap is long enough to contain the entry we need */
brinRevmapExtend(revmap, heapBlk);
if (!samepage)
{
/* need a page on which to put the item */
newbuf = brin_getinsertbuffer(idxrel, oldbuf, newsz, &extended);
if (!BufferIsValid(newbuf))
{
Assert(!extended);
return false;
}
/*
* Note: it's possible (though unlikely) that the returned newbuf is
* the same as oldbuf, if brin_getinsertbuffer determined that the old
* buffer does in fact have enough space.
*/
if (newbuf == oldbuf)
{
Assert(!extended);
newbuf = InvalidBuffer;
}
}
else
{
LockBuffer(oldbuf, BUFFER_LOCK_EXCLUSIVE);
newbuf = InvalidBuffer;
extended = false;
}
oldpage = BufferGetPage(oldbuf);
oldlp = PageGetItemId(oldpage, oldoff);
/*
* Check that the old tuple wasn't updated concurrently: it might have
* moved someplace else entirely ...
*/
if (!ItemIdIsNormal(oldlp))
{
LockBuffer(oldbuf, BUFFER_LOCK_UNLOCK);
/*
* If this happens, and the new buffer was obtained by extending the
* relation, then we need to ensure we don't leave it uninitialized or
* forget about it.
*/
if (BufferIsValid(newbuf))
{
if (extended)
brin_initialize_empty_new_buffer(idxrel, newbuf);
UnlockReleaseBuffer(newbuf);
if (extended)
FreeSpaceMapVacuum(idxrel);
}
return false;
}
oldsz = ItemIdGetLength(oldlp);
oldtup = (BrinTuple *) PageGetItem(oldpage, oldlp);
/*
* ... or it might have been updated in place to different contents.
*/
if (!brin_tuples_equal(oldtup, oldsz, origtup, origsz))
{
LockBuffer(oldbuf, BUFFER_LOCK_UNLOCK);
if (BufferIsValid(newbuf))
{
if (extended)
brin_initialize_empty_new_buffer(idxrel, newbuf);
UnlockReleaseBuffer(newbuf);
if (extended)
FreeSpaceMapVacuum(idxrel);
}
return false;
}
/*
* Great, the old tuple is intact. We can proceed with the update.
*
* If there's enough room in the old page for the new tuple, replace it.
*
* Note that there might now be enough space on the page even though the
* caller told us there isn't, if a concurrent update moved another tuple
* elsewhere or replaced a tuple with a smaller one.
*/
if (((BrinPageFlags(oldpage) & BRIN_EVACUATE_PAGE) == 0) &&
brin_can_do_samepage_update(oldbuf, origsz, newsz))
{
if (BufferIsValid(newbuf))
{
/* as above */
if (extended)
brin_initialize_empty_new_buffer(idxrel, newbuf);
UnlockReleaseBuffer(newbuf);
}
START_CRIT_SECTION();
PageIndexDeleteNoCompact(oldpage, &oldoff, 1);
if (PageAddItem(oldpage, (Item) newtup, newsz, oldoff, true,
false) == InvalidOffsetNumber)
elog(ERROR, "failed to add BRIN tuple");
MarkBufferDirty(oldbuf);
/* XLOG stuff */
if (RelationNeedsWAL(idxrel))
{
xl_brin_samepage_update xlrec;
XLogRecPtr recptr;
uint8 info = XLOG_BRIN_SAMEPAGE_UPDATE;
xlrec.offnum = oldoff;
XLogBeginInsert();
XLogRegisterData((char *) &xlrec, SizeOfBrinSamepageUpdate);
XLogRegisterBuffer(0, oldbuf, REGBUF_STANDARD);
XLogRegisterBufData(0, (char *) newtup, newsz);
recptr = XLogInsert(RM_BRIN_ID, info);
PageSetLSN(oldpage, recptr);
}
END_CRIT_SECTION();
LockBuffer(oldbuf, BUFFER_LOCK_UNLOCK);
if (extended)
FreeSpaceMapVacuum(idxrel);
return true;
}
else if (newbuf == InvalidBuffer)
{
/*
* Not enough space, but caller said that there was. Tell them to
* start over.
*/
LockBuffer(oldbuf, BUFFER_LOCK_UNLOCK);
return false;
}
else
{
/*
* Not enough free space on the oldpage. Put the new tuple on the new
* page, and update the revmap.
*/
Page newpage = BufferGetPage(newbuf);
Buffer revmapbuf;
ItemPointerData newtid;
OffsetNumber newoff;
BlockNumber newblk = InvalidBlockNumber;
Size freespace = 0;
revmapbuf = brinLockRevmapPageForUpdate(revmap, heapBlk);
START_CRIT_SECTION();
/*
* We need to initialize the page if it's newly obtained. Note we
* will WAL-log the initialization as part of the update, so we don't
* need to do that here.
*/
if (extended)
brin_page_init(BufferGetPage(newbuf), BRIN_PAGETYPE_REGULAR);
PageIndexDeleteNoCompact(oldpage, &oldoff, 1);
newoff = PageAddItem(newpage, (Item) newtup, newsz,
InvalidOffsetNumber, false, false);
if (newoff == InvalidOffsetNumber)
elog(ERROR, "failed to add BRIN tuple to new page");
MarkBufferDirty(oldbuf);
MarkBufferDirty(newbuf);
/* needed to update FSM below */
if (extended)
{
newblk = BufferGetBlockNumber(newbuf);
freespace = br_page_get_freespace(newpage);
}
ItemPointerSet(&newtid, BufferGetBlockNumber(newbuf), newoff);
brinSetHeapBlockItemptr(revmapbuf, pagesPerRange, heapBlk, newtid);
MarkBufferDirty(revmapbuf);
/* XLOG stuff */
if (RelationNeedsWAL(idxrel))
{
xl_brin_update xlrec;
XLogRecPtr recptr;
uint8 info;
info = XLOG_BRIN_UPDATE | (extended ? XLOG_BRIN_INIT_PAGE : 0);
xlrec.insert.offnum = newoff;
xlrec.insert.heapBlk = heapBlk;
xlrec.insert.pagesPerRange = pagesPerRange;
xlrec.oldOffnum = oldoff;
XLogBeginInsert();
/* new page */
XLogRegisterData((char *) &xlrec, SizeOfBrinUpdate);
XLogRegisterBuffer(0, newbuf, REGBUF_STANDARD | (extended ? REGBUF_WILL_INIT : 0));
XLogRegisterBufData(0, (char *) newtup, newsz);
/* revmap page */
XLogRegisterBuffer(1, revmapbuf, REGBUF_STANDARD);
/* old page */
XLogRegisterBuffer(2, oldbuf, REGBUF_STANDARD);
recptr = XLogInsert(RM_BRIN_ID, info);
PageSetLSN(oldpage, recptr);
PageSetLSN(newpage, recptr);
PageSetLSN(BufferGetPage(revmapbuf), recptr);
}
END_CRIT_SECTION();
LockBuffer(revmapbuf, BUFFER_LOCK_UNLOCK);
LockBuffer(oldbuf, BUFFER_LOCK_UNLOCK);
UnlockReleaseBuffer(newbuf);
if (extended)
{
Assert(BlockNumberIsValid(newblk));
RecordPageWithFreeSpace(idxrel, newblk, freespace);
FreeSpaceMapVacuum(idxrel);
}
return true;
}
}
/*
* Insert value (stored in GinBtree) to tree described by stack
*/
void
ginInsertValue(GinBtree btree, GinBtreeStack *stack)
{
GinBtreeStack *parent = stack;
BlockNumber rootBlkno = InvalidBuffer;
Page page,
rpage,
lpage;
/* remember root BlockNumber */
while (parent)
{
rootBlkno = parent->blkno;
parent = parent->parent;
}
while (stack)
{
XLogRecData *rdata;
BlockNumber savedRightLink;
page = BufferGetPage(stack->buffer);
savedRightLink = GinPageGetOpaque(page)->rightlink;
if (btree->isEnoughSpace(btree, stack->buffer, stack->off))
{
START_CRIT_SECTION();
btree->placeToPage(btree, stack->buffer, stack->off, &rdata);
MarkBufferDirty(stack->buffer);
if (!btree->index->rd_istemp)
{
XLogRecPtr recptr;
recptr = XLogInsert(RM_GIN_ID, XLOG_GIN_INSERT, rdata);
PageSetLSN(page, recptr);
PageSetTLI(page, ThisTimeLineID);
}
UnlockReleaseBuffer(stack->buffer);
END_CRIT_SECTION();
freeGinBtreeStack(stack->parent);
return;
}
else
{
Buffer rbuffer = GinNewBuffer(btree->index);
Page newlpage;
/*
* newlpage is a pointer to memory page, it doesn't associate with
* buffer, stack->buffer should be untouched
*/
newlpage = btree->splitPage(btree, stack->buffer, rbuffer, stack->off, &rdata);
((ginxlogSplit *) (rdata->data))->rootBlkno = rootBlkno;
parent = stack->parent;
if (parent == NULL)
{
/*
* split root, so we need to allocate new left page and place
* pointer on root to left and right page
*/
Buffer lbuffer = GinNewBuffer(btree->index);
((ginxlogSplit *) (rdata->data))->isRootSplit = TRUE;
((ginxlogSplit *) (rdata->data))->rrlink = InvalidBlockNumber;
page = BufferGetPage(stack->buffer);
lpage = BufferGetPage(lbuffer);
rpage = BufferGetPage(rbuffer);
GinPageGetOpaque(rpage)->rightlink = InvalidBlockNumber;
GinPageGetOpaque(newlpage)->rightlink = BufferGetBlockNumber(rbuffer);
((ginxlogSplit *) (rdata->data))->lblkno = BufferGetBlockNumber(lbuffer);
START_CRIT_SECTION();
GinInitBuffer(stack->buffer, GinPageGetOpaque(newlpage)->flags & ~GIN_LEAF);
PageRestoreTempPage(newlpage, lpage);
btree->fillRoot(btree, stack->buffer, lbuffer, rbuffer);
MarkBufferDirty(rbuffer);
MarkBufferDirty(lbuffer);
MarkBufferDirty(stack->buffer);
if (!btree->index->rd_istemp)
{
XLogRecPtr recptr;
recptr = XLogInsert(RM_GIN_ID, XLOG_GIN_SPLIT, rdata);
PageSetLSN(page, recptr);
PageSetTLI(page, ThisTimeLineID);
PageSetLSN(lpage, recptr);
PageSetTLI(lpage, ThisTimeLineID);
PageSetLSN(rpage, recptr);
PageSetTLI(rpage, ThisTimeLineID);
}
UnlockReleaseBuffer(rbuffer);
UnlockReleaseBuffer(lbuffer);
UnlockReleaseBuffer(stack->buffer);
END_CRIT_SECTION();
return;
}
else
{
/* split non-root page */
((ginxlogSplit *) (rdata->data))->isRootSplit = FALSE;
((ginxlogSplit *) (rdata->data))->rrlink = savedRightLink;
lpage = BufferGetPage(stack->buffer);
rpage = BufferGetPage(rbuffer);
GinPageGetOpaque(rpage)->rightlink = savedRightLink;
GinPageGetOpaque(newlpage)->rightlink = BufferGetBlockNumber(rbuffer);
START_CRIT_SECTION();
PageRestoreTempPage(newlpage, lpage);
MarkBufferDirty(rbuffer);
MarkBufferDirty(stack->buffer);
if (!btree->index->rd_istemp)
{
XLogRecPtr recptr;
recptr = XLogInsert(RM_GIN_ID, XLOG_GIN_SPLIT, rdata);
PageSetLSN(lpage, recptr);
PageSetTLI(lpage, ThisTimeLineID);
PageSetLSN(rpage, recptr);
PageSetTLI(rpage, ThisTimeLineID);
}
UnlockReleaseBuffer(rbuffer);
END_CRIT_SECTION();
}
}
btree->isDelete = FALSE;
/* search parent to lock */
LockBuffer(parent->buffer, GIN_EXCLUSIVE);
/* move right if it's needed */
page = BufferGetPage(parent->buffer);
while ((parent->off = btree->findChildPtr(btree, page, stack->blkno, parent->off)) == InvalidOffsetNumber)
{
BlockNumber rightlink = GinPageGetOpaque(page)->rightlink;
LockBuffer(parent->buffer, GIN_UNLOCK);
if (rightlink == InvalidBlockNumber)
{
/*
* rightmost page, but we don't find parent, we should use
* plain search...
*/
findParents(btree, stack, rootBlkno);
parent = stack->parent;
page = BufferGetPage(parent->buffer);
break;
}
parent->blkno = rightlink;
parent->buffer = ReleaseAndReadBuffer(parent->buffer, btree->index, parent->blkno);
LockBuffer(parent->buffer, GIN_EXCLUSIVE);
page = BufferGetPage(parent->buffer);
}
UnlockReleaseBuffer(stack->buffer);
pfree(stack);
stack = parent;
}
}
/*
* visibilitymap_set - set a bit on a previously pinned page
*
* recptr is the LSN of the XLOG record we're replaying, if we're in recovery,
* or InvalidXLogRecPtr in normal running. The page LSN is advanced to the
* one provided; in normal running, we generate a new XLOG record and set the
* page LSN to that value. cutoff_xid is the largest xmin on the page being
* marked all-visible; it is needed for Hot Standby, and can be
* InvalidTransactionId if the page contains no tuples.
*
* Caller is expected to set the heap page's PD_ALL_VISIBLE bit before calling
* this function. Except in recovery, caller should also pass the heap
* buffer. When checksums are enabled and we're not in recovery, we must add
* the heap buffer to the WAL chain to protect it from being torn.
*
* You must pass a buffer containing the correct map page to this function.
* Call visibilitymap_pin first to pin the right one. This function doesn't do
* any I/O.
*/
void
visibilitymap_set(Relation rel, BlockNumber heapBlk, Buffer heapBuf,
XLogRecPtr recptr, Buffer vmBuf, TransactionId cutoff_xid)
{
BlockNumber mapBlock = HEAPBLK_TO_MAPBLOCK(heapBlk);
uint32 mapByte = HEAPBLK_TO_MAPBYTE(heapBlk);
uint8 mapBit = HEAPBLK_TO_MAPBIT(heapBlk);
Page page;
char *map;
#ifdef TRACE_VISIBILITYMAP
elog(DEBUG1, "vm_set %s %d", RelationGetRelationName(rel), heapBlk);
#endif
Assert(InRecovery || XLogRecPtrIsInvalid(recptr));
Assert(InRecovery || BufferIsValid(heapBuf));
/* Check that we have the right heap page pinned, if present */
if (BufferIsValid(heapBuf) && BufferGetBlockNumber(heapBuf) != heapBlk)
elog(ERROR, "wrong heap buffer passed to visibilitymap_set");
/* Check that we have the right VM page pinned */
if (!BufferIsValid(vmBuf) || BufferGetBlockNumber(vmBuf) != mapBlock)
elog(ERROR, "wrong VM buffer passed to visibilitymap_set");
page = BufferGetPage(vmBuf);
map = PageGetContents(page);
LockBuffer(vmBuf, BUFFER_LOCK_EXCLUSIVE);
if (!(map[mapByte] & (1 << mapBit)))
{
START_CRIT_SECTION();
map[mapByte] |= (1 << mapBit);
MarkBufferDirty(vmBuf);
if (RelationNeedsWAL(rel))
{
if (XLogRecPtrIsInvalid(recptr))
{
Assert(!InRecovery);
recptr = log_heap_visible(rel->rd_node, heapBuf, vmBuf,
cutoff_xid);
/*
* If data checksums are enabled (or wal_log_hints=on), we
* need to protect the heap page from being torn.
*/
if (XLogHintBitIsNeeded())
{
Page heapPage = BufferGetPage(heapBuf);
/* caller is expected to set PD_ALL_VISIBLE first */
Assert(PageIsAllVisible(heapPage));
PageSetLSN(heapPage, recptr);
}
}
PageSetLSN(page, recptr);
}
END_CRIT_SECTION();
}
LockBuffer(vmBuf, BUFFER_LOCK_UNLOCK);
}
/*
* Routine to build an index. Basically calls insert over and over.
*
* XXX: it would be nice to implement some sort of bulk-loading
* algorithm, but it is not clear how to do that.
*/
Datum
gistbuild(PG_FUNCTION_ARGS)
{
Relation heap = (Relation) PG_GETARG_POINTER(0);
Relation index = (Relation) PG_GETARG_POINTER(1);
IndexInfo *indexInfo = (IndexInfo *) PG_GETARG_POINTER(2);
IndexBuildResult *result;
double reltuples;
GISTBuildState buildstate;
Buffer buffer;
Page page;
/*
* We expect to be called exactly once for any index relation. If that's
* not the case, big trouble's what we have.
*/
if (RelationGetNumberOfBlocks(index) != 0)
elog(ERROR, "index \"%s\" already contains data",
RelationGetRelationName(index));
/* no locking is needed */
initGISTstate(&buildstate.giststate, index);
/* initialize the root page */
buffer = gistNewBuffer(index);
Assert(BufferGetBlockNumber(buffer) == GIST_ROOT_BLKNO);
page = BufferGetPage(buffer);
START_CRIT_SECTION();
GISTInitBuffer(buffer, F_LEAF);
MarkBufferDirty(buffer);
if (!index->rd_istemp)
{
XLogRecPtr recptr;
XLogRecData rdata;
rdata.data = (char *) &(index->rd_node);
rdata.len = sizeof(RelFileNode);
rdata.buffer = InvalidBuffer;
rdata.next = NULL;
recptr = XLogInsert(RM_GIST_ID, XLOG_GIST_CREATE_INDEX, &rdata);
PageSetLSN(page, recptr);
PageSetTLI(page, ThisTimeLineID);
}
else
PageSetLSN(page, XLogRecPtrForTemp);
UnlockReleaseBuffer(buffer);
END_CRIT_SECTION();
/* build the index */
buildstate.numindexattrs = indexInfo->ii_NumIndexAttrs;
buildstate.indtuples = 0;
/*
* create a temporary memory context that is reset once for each tuple
* inserted into the index
*/
buildstate.tmpCtx = createTempGistContext();
/* do the heap scan */
reltuples = IndexBuildHeapScan(heap, index, indexInfo,
gistbuildCallback, (void *) &buildstate);
/* okay, all heap tuples are indexed */
MemoryContextDelete(buildstate.tmpCtx);
freeGISTstate(&buildstate.giststate);
/*
* Return statistics
*/
result = (IndexBuildResult *) palloc(sizeof(IndexBuildResult));
result->heap_tuples = reltuples;
result->index_tuples = buildstate.indtuples;
PG_RETURN_POINTER(result);
}
/*
* AlterSequence
*
* Modify the definition of a sequence relation
*/
ObjectAddress
AlterSequence(AlterSeqStmt *stmt)
{
Oid relid;
SeqTable elm;
Relation seqrel;
Buffer buf;
HeapTupleData seqtuple;
Form_pg_sequence seq;
FormData_pg_sequence new___;
List *owned_by;
ObjectAddress address;
/* Open and lock sequence. */
relid = RangeVarGetRelid(stmt->sequence, AccessShareLock, stmt->missing_ok);
if (relid == InvalidOid)
{
ereport(NOTICE,
(errmsg("relation \"%s\" does not exist, skipping",
stmt->sequence->relname)));
return InvalidObjectAddress;
}
init_sequence(relid, &elm, &seqrel);
/* allow ALTER to sequence owner only */
if (!pg_class_ownercheck(relid, GetUserId()))
aclcheck_error(ACLCHECK_NOT_OWNER, ACL_KIND_CLASS,
stmt->sequence->relname);
/* lock page' buffer and read tuple into new___ sequence structure */
seq = read_seq_tuple(elm, seqrel, &buf, &seqtuple);
/* Copy old values of options into workspace */
memcpy(&new___, seq, sizeof(FormData_pg_sequence));
/* Check and set new___ values */
init_params(stmt->options, false, &new___, &owned_by);
/* Clear local cache so that we don't think we have cached numbers */
/* Note that we do not change the currval() state */
elm->cached = elm->last;
/* check the comment above nextval_internal()'s equivalent call. */
if (RelationNeedsWAL(seqrel))
GetTopTransactionId();
/* Now okay to update the on-disk tuple */
START_CRIT_SECTION();
memcpy(seq, &new___, sizeof(FormData_pg_sequence));
MarkBufferDirty(buf);
/* XLOG stuff */
if (RelationNeedsWAL(seqrel))
{
xl_seq_rec xlrec;
XLogRecPtr recptr;
Page page = BufferGetPage(buf);
XLogBeginInsert();
XLogRegisterBuffer(0, buf, REGBUF_WILL_INIT);
xlrec.node = seqrel->rd_node;
XLogRegisterData((char *) &xlrec, sizeof(xl_seq_rec));
XLogRegisterData((char *) seqtuple.t_data, seqtuple.t_len);
recptr = XLogInsert(RM_SEQ_ID, XLOG_SEQ_LOG);
PageSetLSN(page, recptr);
}
END_CRIT_SECTION();
UnlockReleaseBuffer(buf);
/* process OWNED BY if given */
if (owned_by)
process_owned_by(seqrel, owned_by);
InvokeObjectPostAlterHook(RelationRelationId, relid, 0);
ObjectAddressSet(address, RelationRelationId, relid);
relation_close(seqrel, NoLock);
return address;
}