/*
* Deletes pending list pages up to (not including) newHead page.
* If newHead == InvalidBlockNumber then function drops the whole list.
*
* metapage is pinned and exclusive-locked throughout this function.
*
* Returns true if another cleanup process is running concurrently
* (if so, we can just abandon our own efforts)
*/
static bool
shiftList(Relation index, Buffer metabuffer, BlockNumber newHead,
bool fill_fsm, IndexBulkDeleteResult *stats)
{
Page metapage;
GinMetaPageData *metadata;
BlockNumber blknoToDelete;
metapage = BufferGetPage(metabuffer);
metadata = GinPageGetMeta(metapage);
blknoToDelete = metadata->head;
do
{
Page page;
int i;
int64 nDeletedHeapTuples = 0;
ginxlogDeleteListPages data;
Buffer buffers[GIN_NDELETE_AT_ONCE];
BlockNumber freespace[GIN_NDELETE_AT_ONCE];
data.ndeleted = 0;
while (data.ndeleted < GIN_NDELETE_AT_ONCE && blknoToDelete != newHead)
{
freespace[data.ndeleted] = blknoToDelete;
buffers[data.ndeleted] = ReadBuffer(index, blknoToDelete);
LockBuffer(buffers[data.ndeleted], GIN_EXCLUSIVE);
page = BufferGetPage(buffers[data.ndeleted]);
data.ndeleted++;
if (GinPageIsDeleted(page))
{
/* concurrent cleanup process is detected */
for (i = 0; i < data.ndeleted; i++)
UnlockReleaseBuffer(buffers[i]);
return true;
}
nDeletedHeapTuples += GinPageGetOpaque(page)->maxoff;
blknoToDelete = GinPageGetOpaque(page)->rightlink;
}
if (stats)
stats->pages_deleted += data.ndeleted;
/*
* This operation touches an unusually large number of pages, so
* prepare the XLogInsert machinery for that before entering the
* critical section.
*/
if (RelationNeedsWAL(index))
XLogEnsureRecordSpace(data.ndeleted, 0);
START_CRIT_SECTION();
metadata->head = blknoToDelete;
Assert(metadata->nPendingPages >= data.ndeleted);
metadata->nPendingPages -= data.ndeleted;
Assert(metadata->nPendingHeapTuples >= nDeletedHeapTuples);
metadata->nPendingHeapTuples -= nDeletedHeapTuples;
if (blknoToDelete == InvalidBlockNumber)
{
metadata->tail = InvalidBlockNumber;
metadata->tailFreeSize = 0;
metadata->nPendingPages = 0;
metadata->nPendingHeapTuples = 0;
}
MarkBufferDirty(metabuffer);
for (i = 0; i < data.ndeleted; i++)
{
page = BufferGetPage(buffers[i]);
GinPageGetOpaque(page)->flags = GIN_DELETED;
MarkBufferDirty(buffers[i]);
}
if (RelationNeedsWAL(index))
{
XLogRecPtr recptr;
XLogBeginInsert();
XLogRegisterBuffer(0, metabuffer, REGBUF_WILL_INIT);
for (i = 0; i < data.ndeleted; i++)
XLogRegisterBuffer(i + 1, buffers[i], REGBUF_WILL_INIT);
memcpy(&data.metadata, metadata, sizeof(GinMetaPageData));
XLogRegisterData((char *) &data,
sizeof(ginxlogDeleteListPages));
recptr = XLogInsert(RM_GIN_ID, XLOG_GIN_DELETE_LISTPAGE);
PageSetLSN(metapage, recptr);
for (i = 0; i < data.ndeleted; i++)
{
page = BufferGetPage(buffers[i]);
PageSetLSN(page, recptr);
}
}
for (i = 0; i < data.ndeleted; i++)
UnlockReleaseBuffer(buffers[i]);
END_CRIT_SECTION();
for (i = 0; fill_fsm && i < data.ndeleted; i++)
RecordFreeIndexPage(index, freespace[i]);
} while (blknoToDelete != newHead);
return false;
}
/*
* _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 */
}
/*
* 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, NULL, NULL,
BGP_NO_SNAPSHOT_TEST);
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, NULL, NULL,
BGP_NO_SNAPSHOT_TEST));
/* 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, NULL, NULL, BGP_NO_SNAPSHOT_TEST);
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,
NULL, NULL,
BGP_NO_SNAPSHOT_TEST));
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,
NULL, NULL,
BGP_NO_SNAPSHOT_TEST));
dist->page = BufferGetPage(dist->buffer, NULL, NULL,
BGP_NO_SNAPSHOT_TEST);
/* Write the WAL record */
if (RelationNeedsWAL(rel))
recptr = gistXLogSplit(rel->rd_node, blkno, 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 also get here if ntuples==0.
*/
START_CRIT_SECTION();
/*
* While we delete only one tuple at once we could mix calls
* PageIndexTupleDelete() here and PageIndexMultiDelete() in
* gistRedoPageUpdateRecord()
*/
if (OffsetNumberIsValid(oldoffnum))
PageIndexTupleDelete(page, oldoffnum);
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(rel->rd_node, 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, NULL, NULL,
BGP_NO_SNAPSHOT_TEST);
GistPageSetNSN(leftpg, recptr);
GistClearFollowRight(leftpg);
PageSetLSN(leftpg, recptr);
}
END_CRIT_SECTION();
return is_split;
}
/*
* _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;
}
/*
* Deletes pending list pages up to (not including) newHead page.
* If newHead == InvalidBlockNumber then function drops the whole list.
*
* metapage is pinned and exclusive-locked throughout this function.
*/
static void
shiftList(Relation index, Buffer metabuffer, BlockNumber newHead,
bool fill_fsm, IndexBulkDeleteResult *stats)
{
Page metapage;
GinMetaPageData *metadata;
BlockNumber blknoToDelete;
metapage = BufferGetPage(metabuffer);
metadata = GinPageGetMeta(metapage);
blknoToDelete = metadata->head;
do
{
Page page;
int i;
int64 nDeletedHeapTuples = 0;
ginxlogDeleteListPages data;
Buffer buffers[GIN_NDELETE_AT_ONCE];
BlockNumber freespace[GIN_NDELETE_AT_ONCE];
data.ndeleted = 0;
while (data.ndeleted < GIN_NDELETE_AT_ONCE && blknoToDelete != newHead)
{
freespace[data.ndeleted] = blknoToDelete;
buffers[data.ndeleted] = ReadBuffer(index, blknoToDelete);
LockBuffer(buffers[data.ndeleted], GIN_EXCLUSIVE);
page = BufferGetPage(buffers[data.ndeleted]);
data.ndeleted++;
Assert(!GinPageIsDeleted(page));
nDeletedHeapTuples += GinPageGetOpaque(page)->maxoff;
blknoToDelete = GinPageGetOpaque(page)->rightlink;
}
if (stats)
stats->pages_deleted += data.ndeleted;
/*
* This operation touches an unusually large number of pages, so
* prepare the XLogInsert machinery for that before entering the
* critical section.
*/
if (RelationNeedsWAL(index))
XLogEnsureRecordSpace(data.ndeleted, 0);
START_CRIT_SECTION();
metadata->head = blknoToDelete;
Assert(metadata->nPendingPages >= data.ndeleted);
metadata->nPendingPages -= data.ndeleted;
Assert(metadata->nPendingHeapTuples >= nDeletedHeapTuples);
metadata->nPendingHeapTuples -= nDeletedHeapTuples;
if (blknoToDelete == InvalidBlockNumber)
{
metadata->tail = InvalidBlockNumber;
metadata->tailFreeSize = 0;
metadata->nPendingPages = 0;
metadata->nPendingHeapTuples = 0;
}
/*
* Set pd_lower just past the end of the metadata. This is essential,
* because without doing so, metadata will be lost if xlog.c
* compresses the page. (We must do this here because pre-v11
* versions of PG did not set the metapage's pd_lower correctly, so a
* pg_upgraded index might contain the wrong value.)
*/
((PageHeader) metapage)->pd_lower =
((char *) metadata + sizeof(GinMetaPageData)) - (char *) metapage;
MarkBufferDirty(metabuffer);
for (i = 0; i < data.ndeleted; i++)
{
page = BufferGetPage(buffers[i]);
GinPageGetOpaque(page)->flags = GIN_DELETED;
MarkBufferDirty(buffers[i]);
}
if (RelationNeedsWAL(index))
{
XLogRecPtr recptr;
XLogBeginInsert();
XLogRegisterBuffer(0, metabuffer,
REGBUF_WILL_INIT | REGBUF_STANDARD);
for (i = 0; i < data.ndeleted; i++)
XLogRegisterBuffer(i + 1, buffers[i], REGBUF_WILL_INIT);
memcpy(&data.metadata, metadata, sizeof(GinMetaPageData));
XLogRegisterData((char *) &data,
sizeof(ginxlogDeleteListPages));
recptr = XLogInsert(RM_GIN_ID, XLOG_GIN_DELETE_LISTPAGE);
PageSetLSN(metapage, recptr);
for (i = 0; i < data.ndeleted; i++)
{
page = BufferGetPage(buffers[i]);
PageSetLSN(page, recptr);
}
}
for (i = 0; i < data.ndeleted; i++)
UnlockReleaseBuffer(buffers[i]);
END_CRIT_SECTION();
for (i = 0; fill_fsm && i < data.ndeleted; i++)
RecordFreeIndexPage(index, freespace[i]);
} while (blknoToDelete != newHead);
}
