static void
btree_xlog_delete(XLogReaderState *record)
{
XLogRecPtr lsn = record->EndRecPtr;
xl_btree_delete *xlrec = (xl_btree_delete *) XLogRecGetData(record);
Buffer buffer;
Page page;
BTPageOpaque opaque;
/*
* If we have any conflict processing to do, it must happen before we
* update the page.
*
* Btree delete records can conflict with standby queries. You might
* think that vacuum records would conflict as well, but we've handled
* that already. XLOG_HEAP2_CLEANUP_INFO records provide the highest xid
* cleaned by the vacuum of the heap and so we can resolve any conflicts
* just once when that arrives. After that we know that no conflicts
* exist from individual btree vacuum records on that index.
*/
if (InHotStandby)
{
TransactionId latestRemovedXid = btree_xlog_delete_get_latestRemovedXid(record);
RelFileNode rnode;
XLogRecGetBlockTag(record, 0, &rnode, NULL, NULL);
ResolveRecoveryConflictWithSnapshot(latestRemovedXid, rnode);
}
/*
* We don't need to take a cleanup lock to apply these changes. See
* nbtree/README for details.
*/
if (XLogReadBufferForRedo(record, 0, &buffer) == BLK_NEEDS_REDO)
{
page = (Page) BufferGetPage(buffer);
if (XLogRecGetDataLen(record) > SizeOfBtreeDelete)
{
OffsetNumber *unused;
unused = (OffsetNumber *) ((char *) xlrec + SizeOfBtreeDelete);
PageIndexMultiDelete(page, unused, xlrec->nitems);
}
/*
* Mark the page as not containing any LP_DEAD items --- see comments
* in _bt_delitems_delete().
*/
opaque = (BTPageOpaque) PageGetSpecialPointer(page);
opaque->btpo_flags &= ~BTP_HAS_GARBAGE;
PageSetLSN(page, lsn);
MarkBufferDirty(buffer);
}
if (BufferIsValid(buffer))
UnlockReleaseBuffer(buffer);
}
static void
spgRedoVacuumRoot(XLogRecPtr lsn, XLogRecord *record)
{
char *ptr = XLogRecGetData(record);
spgxlogVacuumRoot *xldata = (spgxlogVacuumRoot *) ptr;
OffsetNumber *toDelete;
Buffer buffer;
Page page;
ptr += sizeof(spgxlogVacuumRoot);
toDelete = (OffsetNumber *) ptr;
if (!(record->xl_info & XLR_BKP_BLOCK_1))
{
buffer = XLogReadBuffer(xldata->node, SPGIST_HEAD_BLKNO, false);
if (BufferIsValid(buffer))
{
page = BufferGetPage(buffer);
if (!XLByteLE(lsn, PageGetLSN(page)))
{
/* The tuple numbers are in order */
PageIndexMultiDelete(page, toDelete, xldata->nDelete);
PageSetLSN(page, lsn);
PageSetTLI(page, ThisTimeLineID);
MarkBufferDirty(buffer);
}
UnlockReleaseBuffer(buffer);
}
}
}
static void
spgRedoVacuumRoot(XLogRecPtr lsn, XLogRecord *record)
{
char *ptr = XLogRecGetData(record);
spgxlogVacuumRoot *xldata = (spgxlogVacuumRoot *) ptr;
OffsetNumber *toDelete;
Buffer buffer;
Page page;
toDelete = xldata->offsets;
if (XLogReadBufferForRedo(lsn, record, 0, xldata->node, xldata->blkno,
&buffer) == BLK_NEEDS_REDO)
{
page = BufferGetPage(buffer);
/* The tuple numbers are in order */
PageIndexMultiDelete(page, toDelete, xldata->nDelete);
PageSetLSN(page, lsn);
MarkBufferDirty(buffer);
}
if (BufferIsValid(buffer))
UnlockReleaseBuffer(buffer);
}
/*
* Delete item(s) from a btree page.
*
* 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,
* and will write the buffer afterwards. Also, the given itemnos *must*
* appear in increasing order in the array.
*/
void
_bt_delitems(Relation rel, Buffer buf,
OffsetNumber *itemnos, int nitems)
{
Page page = BufferGetPage(buf);
/* No ereport(ERROR) until changes are logged */
START_CRIT_SECTION();
/* Fix the page */
PageIndexMultiDelete(page, itemnos, nitems);
/* XLOG stuff */
if (!rel->rd_istemp)
{
xl_btree_delete xlrec;
XLogRecPtr recptr;
XLogRecData rdata[2];
xlrec.node = rel->rd_node;
xlrec.block = BufferGetBlockNumber(buf);
rdata[0].data = (char *) &xlrec;
rdata[0].len = SizeOfBtreeDelete;
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_DELETE, rdata);
PageSetLSN(page, recptr);
PageSetTLI(page, ThisTimeLineID);
}
END_CRIT_SECTION();
}
/*
* redo delete on gist index page to remove tuples marked as DEAD during index
* tuple insertion
*/
static void
gistRedoDeleteRecord(XLogReaderState *record)
{
XLogRecPtr lsn = record->EndRecPtr;
gistxlogDelete *xldata = (gistxlogDelete *) XLogRecGetData(record);
Buffer buffer;
Page page;
/*
* If we have any conflict processing to do, it must happen before we
* update the page.
*
* GiST delete records can conflict with standby queries. You might think
* that vacuum records would conflict as well, but we've handled that
* already. XLOG_HEAP2_CLEANUP_INFO records provide the highest xid
* cleaned by the vacuum of the heap and so we can resolve any conflicts
* just once when that arrives. After that we know that no conflicts
* exist from individual gist vacuum records on that index.
*/
if (InHotStandby)
{
TransactionId latestRemovedXid = gistRedoDeleteRecordGetLatestRemovedXid(record);
RelFileNode rnode;
XLogRecGetBlockTag(record, 0, &rnode, NULL, NULL);
ResolveRecoveryConflictWithSnapshot(latestRemovedXid, rnode);
}
if (XLogReadBufferForRedo(record, 0, &buffer) == BLK_NEEDS_REDO)
{
page = (Page) BufferGetPage(buffer);
if (XLogRecGetDataLen(record) > SizeOfGistxlogDelete)
{
OffsetNumber *todelete;
todelete = (OffsetNumber *) ((char *) xldata + SizeOfGistxlogDelete);
PageIndexMultiDelete(page, todelete, xldata->ntodelete);
}
GistClearPageHasGarbage(page);
GistMarkTuplesDeleted(page);
PageSetLSN(page, lsn);
MarkBufferDirty(buffer);
}
if (BufferIsValid(buffer))
UnlockReleaseBuffer(buffer);
}
/*
* Clean up redirect and placeholder tuples on the given page
*
* Redirect tuples can be marked placeholder once they're old enough.
* Placeholder tuples can be removed if it won't change the offsets of
* non-placeholder ones.
*
* Unlike the routines above, this works on both leaf and inner pages.
*/
static void
vacuumRedirectAndPlaceholder(Relation index, Buffer buffer)
{
Page page = BufferGetPage(buffer);
SpGistPageOpaque opaque = SpGistPageGetOpaque(page);
OffsetNumber i,
max = PageGetMaxOffsetNumber(page),
firstPlaceholder = InvalidOffsetNumber;
bool hasNonPlaceholder = false;
bool hasUpdate = false;
OffsetNumber itemToPlaceholder[MaxIndexTuplesPerPage];
OffsetNumber itemnos[MaxIndexTuplesPerPage];
spgxlogVacuumRedirect xlrec;
XLogRecData rdata[3];
xlrec.node = index->rd_node;
xlrec.blkno = BufferGetBlockNumber(buffer);
xlrec.nToPlaceholder = 0;
xlrec.newestRedirectXid = InvalidTransactionId;
START_CRIT_SECTION();
/*
* Scan backwards to convert old redirection tuples to placeholder tuples,
* and identify location of last non-placeholder tuple while at it.
*/
for (i = max;
i >= FirstOffsetNumber &&
(opaque->nRedirection > 0 || !hasNonPlaceholder);
i--)
{
SpGistDeadTuple dt;
dt = (SpGistDeadTuple) PageGetItem(page, PageGetItemId(page, i));
if (dt->tupstate == SPGIST_REDIRECT &&
TransactionIdPrecedes(dt->xid, RecentGlobalXmin))
{
dt->tupstate = SPGIST_PLACEHOLDER;
Assert(opaque->nRedirection > 0);
opaque->nRedirection--;
opaque->nPlaceholder++;
/* remember newest XID among the removed redirects */
if (!TransactionIdIsValid(xlrec.newestRedirectXid) ||
TransactionIdPrecedes(xlrec.newestRedirectXid, dt->xid))
xlrec.newestRedirectXid = dt->xid;
ItemPointerSetInvalid(&dt->pointer);
itemToPlaceholder[xlrec.nToPlaceholder] = i;
xlrec.nToPlaceholder++;
hasUpdate = true;
}
if (dt->tupstate == SPGIST_PLACEHOLDER)
{
if (!hasNonPlaceholder)
firstPlaceholder = i;
}
else
{
hasNonPlaceholder = true;
}
}
/*
* Any placeholder tuples at the end of page can safely be removed. We
* can't remove ones before the last non-placeholder, though, because we
* can't alter the offset numbers of non-placeholder tuples.
*/
if (firstPlaceholder != InvalidOffsetNumber)
{
/*
* We do not store this array to rdata because it's easy to recreate.
*/
for (i = firstPlaceholder; i <= max; i++)
itemnos[i - firstPlaceholder] = i;
i = max - firstPlaceholder + 1;
Assert(opaque->nPlaceholder >= i);
opaque->nPlaceholder -= i;
/* The array is surely sorted, so can use PageIndexMultiDelete */
PageIndexMultiDelete(page, itemnos, i);
hasUpdate = true;
}
xlrec.firstPlaceholder = firstPlaceholder;
if (hasUpdate)
MarkBufferDirty(buffer);
if (hasUpdate && RelationNeedsWAL(index))
{
XLogRecPtr recptr;
ACCEPT_RDATA_DATA(&xlrec, SizeOfSpgxlogVacuumRedirect, 0);
ACCEPT_RDATA_DATA(itemToPlaceholder, sizeof(OffsetNumber) * xlrec.nToPlaceholder, 1);
ACCEPT_RDATA_BUFFER(buffer, 2);
recptr = XLogInsert(RM_SPGIST_ID, XLOG_SPGIST_VACUUM_REDIRECT, rdata);
PageSetLSN(page, recptr);
}
END_CRIT_SECTION();
}
static void
spgRedoVacuumRedirect(XLogRecPtr lsn, XLogRecord *record)
{
char *ptr = XLogRecGetData(record);
spgxlogVacuumRedirect *xldata = (spgxlogVacuumRedirect *) ptr;
OffsetNumber *itemToPlaceholder;
Buffer buffer;
itemToPlaceholder = xldata->offsets;
/*
* If any redirection tuples are being removed, make sure there are no
* live Hot Standby transactions that might need to see them.
*/
if (InHotStandby)
{
if (TransactionIdIsValid(xldata->newestRedirectXid))
ResolveRecoveryConflictWithSnapshot(xldata->newestRedirectXid,
xldata->node);
}
if (XLogReadBufferForRedo(lsn, record, 0, xldata->node, xldata->blkno,
&buffer) == BLK_NEEDS_REDO)
{
Page page = BufferGetPage(buffer);
SpGistPageOpaque opaque = SpGistPageGetOpaque(page);
int i;
/* Convert redirect pointers to plain placeholders */
for (i = 0; i < xldata->nToPlaceholder; i++)
{
SpGistDeadTuple dt;
dt = (SpGistDeadTuple) PageGetItem(page,
PageGetItemId(page, itemToPlaceholder[i]));
Assert(dt->tupstate == SPGIST_REDIRECT);
dt->tupstate = SPGIST_PLACEHOLDER;
ItemPointerSetInvalid(&dt->pointer);
}
Assert(opaque->nRedirection >= xldata->nToPlaceholder);
opaque->nRedirection -= xldata->nToPlaceholder;
opaque->nPlaceholder += xldata->nToPlaceholder;
/* Remove placeholder tuples at end of page */
if (xldata->firstPlaceholder != InvalidOffsetNumber)
{
int max = PageGetMaxOffsetNumber(page);
OffsetNumber *toDelete;
toDelete = palloc(sizeof(OffsetNumber) * max);
for (i = xldata->firstPlaceholder; i <= max; i++)
toDelete[i - xldata->firstPlaceholder] = i;
i = max - xldata->firstPlaceholder + 1;
Assert(opaque->nPlaceholder >= i);
opaque->nPlaceholder -= i;
/* The array is sorted, so can use PageIndexMultiDelete */
PageIndexMultiDelete(page, toDelete, i);
pfree(toDelete);
}
PageSetLSN(page, lsn);
MarkBufferDirty(buffer);
}
if (BufferIsValid(buffer))
UnlockReleaseBuffer(buffer);
}
/*
* Vacuum a root page when it is also a leaf
*
* On the root, we just delete any dead leaf tuples; no fancy business
*/
static void
vacuumLeafRoot(spgBulkDeleteState *bds, Relation index, Buffer buffer)
{
Page page = BufferGetPage(buffer);
spgxlogVacuumRoot xlrec;
XLogRecData rdata[3];
OffsetNumber toDelete[MaxIndexTuplesPerPage];
OffsetNumber i,
max = PageGetMaxOffsetNumber(page);
xlrec.blkno = BufferGetBlockNumber(buffer);
xlrec.nDelete = 0;
/* Scan page, identify tuples to delete, accumulate stats */
for (i = FirstOffsetNumber; i <= max; i++)
{
SpGistLeafTuple lt;
lt = (SpGistLeafTuple) PageGetItem(page,
PageGetItemId(page, i));
if (lt->tupstate == SPGIST_LIVE)
{
Assert(ItemPointerIsValid(<->heapPtr));
if (bds->callback(<->heapPtr, bds->callback_state))
{
bds->stats->tuples_removed += 1;
toDelete[xlrec.nDelete] = i;
xlrec.nDelete++;
}
else
{
bds->stats->num_index_tuples += 1;
}
}
else
{
/* all tuples on root should be live */
elog(ERROR, "unexpected SPGiST tuple state: %d",
lt->tupstate);
}
}
if (xlrec.nDelete == 0)
return; /* nothing more to do */
/* Prepare WAL record */
xlrec.node = index->rd_node;
STORE_STATE(&bds->spgstate, xlrec.stateSrc);
ACCEPT_RDATA_DATA(&xlrec, SizeOfSpgxlogVacuumRoot, 0);
/* sizeof(xlrec) should be a multiple of sizeof(OffsetNumber) */
ACCEPT_RDATA_DATA(toDelete, sizeof(OffsetNumber) * xlrec.nDelete, 1);
ACCEPT_RDATA_BUFFER(buffer, 2);
/* Do the update */
START_CRIT_SECTION();
/* The tuple numbers are in order, so we can use PageIndexMultiDelete */
PageIndexMultiDelete(page, toDelete, xlrec.nDelete);
MarkBufferDirty(buffer);
if (RelationNeedsWAL(index))
{
XLogRecPtr recptr;
recptr = XLogInsert(RM_SPGIST_ID, XLOG_SPGIST_VACUUM_ROOT, rdata);
PageSetLSN(page, recptr);
}
END_CRIT_SECTION();
}
/*
* replay delete operation of hash index
*/
static void
hash_xlog_delete(XLogReaderState *record)
{
XLogRecPtr lsn = record->EndRecPtr;
xl_hash_delete *xldata = (xl_hash_delete *) XLogRecGetData(record);
Buffer bucketbuf = InvalidBuffer;
Buffer deletebuf;
Page page;
XLogRedoAction action;
/*
* Ensure we have a cleanup lock on primary bucket page before we start
* with the actual replay operation. This is to ensure that neither a
* scan can start nor a scan can be already-in-progress during the replay
* of this operation. If we allow scans during this operation, then they
* can miss some records or show the same record multiple times.
*/
if (xldata->is_primary_bucket_page)
action = XLogReadBufferForRedoExtended(record, 1, RBM_NORMAL, true, &deletebuf);
else
{
/*
* we don't care for return value as the purpose of reading bucketbuf
* is to ensure a cleanup lock on primary bucket page.
*/
(void) XLogReadBufferForRedoExtended(record, 0, RBM_NORMAL, true, &bucketbuf);
action = XLogReadBufferForRedo(record, 1, &deletebuf);
}
/* replay the record for deleting entries in bucket page */
if (action == BLK_NEEDS_REDO)
{
char *ptr;
Size len;
ptr = XLogRecGetBlockData(record, 1, &len);
page = (Page) BufferGetPage(deletebuf);
if (len > 0)
{
OffsetNumber *unused;
OffsetNumber *unend;
unused = (OffsetNumber *) ptr;
unend = (OffsetNumber *) ((char *) ptr + len);
if ((unend - unused) > 0)
PageIndexMultiDelete(page, unused, unend - unused);
}
/*
* Mark the page as not containing any LP_DEAD items only if
* clear_dead_marking flag is set to true. See comments in
* hashbucketcleanup() for details.
*/
if (xldata->clear_dead_marking)
{
HashPageOpaque pageopaque;
pageopaque = (HashPageOpaque) PageGetSpecialPointer(page);
pageopaque->hasho_flag &= ~LH_PAGE_HAS_DEAD_TUPLES;
}
PageSetLSN(page, lsn);
MarkBufferDirty(deletebuf);
}
if (BufferIsValid(deletebuf))
UnlockReleaseBuffer(deletebuf);
if (BufferIsValid(bucketbuf))
UnlockReleaseBuffer(bucketbuf);
}
static void
btree_xlog_vacuum(XLogReaderState *record)
{
XLogRecPtr lsn = record->EndRecPtr;
xl_btree_vacuum *xlrec = (xl_btree_vacuum *) XLogRecGetData(record);
Buffer buffer;
Page page;
BTPageOpaque opaque;
/*
* If queries might be active then we need to ensure every leaf page is
* unpinned between the lastBlockVacuumed and the current block, if there
* are any. This prevents replay of the VACUUM from reaching the stage of
* removing heap tuples while there could still be indexscans "in flight"
* to those particular tuples (see nbtree/README).
*
* It might be worth checking if there are actually any backends running;
* if not, we could just skip this.
*
* Since VACUUM can visit leaf pages out-of-order, it might issue records
* with lastBlockVacuumed >= block; that's not an error, it just means
* nothing to do now.
*
* Note: since we touch all pages in the range, we will lock non-leaf
* pages, and also any empty (all-zero) pages that may be in the index. It
* doesn't seem worth the complexity to avoid that. But it's important
* that HotStandbyActiveInReplay() will not return true if the database
* isn't yet consistent; so we need not fear reading still-corrupt blocks
* here during crash recovery.
*/
if (HotStandbyActiveInReplay())
{
RelFileNode thisrnode;
BlockNumber thisblkno;
BlockNumber blkno;
XLogRecGetBlockTag(record, 0, &thisrnode, NULL, &thisblkno);
for (blkno = xlrec->lastBlockVacuumed + 1; blkno < thisblkno; blkno++)
{
/*
* We use RBM_NORMAL_NO_LOG mode because it's not an error
* condition to see all-zero pages. The original btvacuumpage
* scan would have skipped over all-zero pages, noting them in FSM
* but not bothering to initialize them just yet; so we mustn't
* throw an error here. (We could skip acquiring the cleanup lock
* if PageIsNew, but it's probably not worth the cycles to test.)
*
* XXX we don't actually need to read the block, we just need to
* confirm it is unpinned. If we had a special call into the
* buffer manager we could optimise this so that if the block is
* not in shared_buffers we confirm it as unpinned.
*/
buffer = XLogReadBufferExtended(thisrnode, MAIN_FORKNUM, blkno,
RBM_NORMAL_NO_LOG);
if (BufferIsValid(buffer))
{
LockBufferForCleanup(buffer);
UnlockReleaseBuffer(buffer);
}
}
}
/*
* Like in btvacuumpage(), we need to take a cleanup lock on every leaf
* page. See nbtree/README for details.
*/
if (XLogReadBufferForRedoExtended(record, 0, RBM_NORMAL, true, &buffer)
== BLK_NEEDS_REDO)
{
char *ptr;
Size len;
ptr = XLogRecGetBlockData(record, 0, &len);
page = (Page) BufferGetPage(buffer);
if (len > 0)
{
OffsetNumber *unused;
OffsetNumber *unend;
unused = (OffsetNumber *) ptr;
unend = (OffsetNumber *) ((char *) ptr + len);
if ((unend - unused) > 0)
PageIndexMultiDelete(page, unused, unend - unused);
}
/*
* Mark the page as not containing any LP_DEAD items --- see comments
* in _bt_delitems_vacuum().
*/
opaque = (BTPageOpaque) PageGetSpecialPointer(page);
opaque->btpo_flags &= ~BTP_HAS_GARBAGE;
PageSetLSN(page, lsn);
MarkBufferDirty(buffer);
}
if (BufferIsValid(buffer))
UnlockReleaseBuffer(buffer);
}
/*
* redo any page update (except page split)
*/
static void
gistRedoPageUpdateRecord(XLogReaderState *record)
{
XLogRecPtr lsn = record->EndRecPtr;
gistxlogPageUpdate *xldata = (gistxlogPageUpdate *) XLogRecGetData(record);
Buffer buffer;
Page page;
if (XLogReadBufferForRedo(record, 0, &buffer) == BLK_NEEDS_REDO)
{
char *begin;
char *data;
Size datalen;
int ninserted = 0;
data = begin = XLogRecGetBlockData(record, 0, &datalen);
page = (Page) BufferGetPage(buffer);
/* Delete old tuples */
if (xldata->ntodelete > 0)
{
OffsetNumber *todelete = (OffsetNumber *) data;
data += sizeof(OffsetNumber) * xldata->ntodelete;
PageIndexMultiDelete(page, todelete, xldata->ntodelete);
if (GistPageIsLeaf(page))
GistMarkTuplesDeleted(page);
}
/* add tuples */
if (data - begin < datalen)
{
OffsetNumber off = (PageIsEmpty(page)) ? FirstOffsetNumber :
OffsetNumberNext(PageGetMaxOffsetNumber(page));
while (data - begin < datalen)
{
IndexTuple itup = (IndexTuple) data;
Size sz = IndexTupleSize(itup);
OffsetNumber l;
data += sz;
l = PageAddItem(page, (Item) itup, sz, off, false, false);
if (l == InvalidOffsetNumber)
elog(ERROR, "failed to add item to GiST index page, size %d bytes",
(int) sz);
off++;
ninserted++;
}
}
Assert(ninserted == xldata->ntoinsert);
PageSetLSN(page, lsn);
MarkBufferDirty(buffer);
}
/*
* Fix follow-right data on left child page
*
* This must be done while still holding the lock on the target page. Note
* that even if the target page no longer exists, we still attempt to
* replay the change on the child page.
*/
if (XLogRecHasBlockRef(record, 1))
gistRedoClearFollowRight(record, 1);
if (BufferIsValid(buffer))
UnlockReleaseBuffer(buffer);
}
/*
* replay move of page contents for squeeze operation of hash index
*/
static void
hash_xlog_move_page_contents(XLogReaderState *record)
{
XLogRecPtr lsn = record->EndRecPtr;
xl_hash_move_page_contents *xldata = (xl_hash_move_page_contents *) XLogRecGetData(record);
Buffer bucketbuf = InvalidBuffer;
Buffer writebuf = InvalidBuffer;
Buffer deletebuf = InvalidBuffer;
XLogRedoAction action;
/*
* Ensure we have a cleanup lock on primary bucket page before we start
* with the actual replay operation. This is to ensure that neither a
* scan can start nor a scan can be already-in-progress during the replay
* of this operation. If we allow scans during this operation, then they
* can miss some records or show the same record multiple times.
*/
if (xldata->is_prim_bucket_same_wrt)
action = XLogReadBufferForRedoExtended(record, 1, RBM_NORMAL, true, &writebuf);
else
{
/*
* we don't care for return value as the purpose of reading bucketbuf
* is to ensure a cleanup lock on primary bucket page.
*/
(void) XLogReadBufferForRedoExtended(record, 0, RBM_NORMAL, true, &bucketbuf);
action = XLogReadBufferForRedo(record, 1, &writebuf);
}
/* replay the record for adding entries in overflow buffer */
if (action == BLK_NEEDS_REDO)
{
Page writepage;
char *begin;
char *data;
Size datalen;
uint16 ninserted = 0;
data = begin = XLogRecGetBlockData(record, 1, &datalen);
writepage = (Page) BufferGetPage(writebuf);
if (xldata->ntups > 0)
{
OffsetNumber *towrite = (OffsetNumber *) data;
data += sizeof(OffsetNumber) * xldata->ntups;
while (data - begin < datalen)
{
IndexTuple itup = (IndexTuple) data;
Size itemsz;
OffsetNumber l;
itemsz = IndexTupleDSize(*itup);
itemsz = MAXALIGN(itemsz);
data += itemsz;
l = PageAddItem(writepage, (Item) itup, itemsz, towrite[ninserted], false, false);
if (l == InvalidOffsetNumber)
elog(ERROR, "hash_xlog_move_page_contents: failed to add item to hash index page, size %d bytes",
(int) itemsz);
ninserted++;
}
}
/*
* number of tuples inserted must be same as requested in REDO record.
*/
Assert(ninserted == xldata->ntups);
PageSetLSN(writepage, lsn);
MarkBufferDirty(writebuf);
}
/* replay the record for deleting entries from overflow buffer */
if (XLogReadBufferForRedo(record, 2, &deletebuf) == BLK_NEEDS_REDO)
{
Page page;
char *ptr;
Size len;
ptr = XLogRecGetBlockData(record, 2, &len);
page = (Page) BufferGetPage(deletebuf);
if (len > 0)
{
OffsetNumber *unused;
OffsetNumber *unend;
unused = (OffsetNumber *) ptr;
unend = (OffsetNumber *) ((char *) ptr + len);
if ((unend - unused) > 0)
PageIndexMultiDelete(page, unused, unend - unused);
}
PageSetLSN(page, lsn);
MarkBufferDirty(deletebuf);
}
/*
* Replay is complete, now we can release the buffers. We release locks at
* end of replay operation to ensure that we hold lock on primary bucket
* page till end of operation. We can optimize by releasing the lock on
* write buffer as soon as the operation for same is complete, if it is
* not same as primary bucket page, but that doesn't seem to be worth
* complicating the code.
*/
if (BufferIsValid(deletebuf))
UnlockReleaseBuffer(deletebuf);
if (BufferIsValid(writebuf))
UnlockReleaseBuffer(writebuf);
if (BufferIsValid(bucketbuf))
UnlockReleaseBuffer(bucketbuf);
}
/*
* Delete item(s) from a btree page.
*
* 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.
*/
void
_bt_delitems(Relation rel, Buffer buf,
OffsetNumber *itemnos, int nitems,
bool inVacuum)
{
Page page;
BTPageOpaque opaque;
MIRROREDLOCK_BUFMGR_MUST_ALREADY_BE_HELD;
page = BufferGetPage(buf);
// Fetch gp_persistent_relation_node information that will be added to XLOG record.
RelationFetchGpRelationNodeForXLog(rel);
/* No ereport(ERROR) until changes are logged */
START_CRIT_SECTION();
/* Fix the page */
PageIndexMultiDelete(page, itemnos, nitems);
/*
* If this is within VACUUM, we can clear the vacuum cycle ID since this
* page has certainly been processed by the current vacuum scan.
*/
opaque = (BTPageOpaque) PageGetSpecialPointer(page);
if (inVacuum)
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 (!rel->rd_istemp)
{
xl_btree_delete xlrec;
XLogRecPtr recptr;
XLogRecData rdata[2];
xl_btreenode_set(&(xlrec.btreenode), rel);
xlrec.block = BufferGetBlockNumber(buf);
rdata[0].data = (char *) &xlrec;
rdata[0].len = SizeOfBtreeDelete;
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_DELETE, rdata);
PageSetLSN(page, recptr);
PageSetTLI(page, ThisTimeLineID);
}
END_CRIT_SECTION();
}
/*
* _hash_squeezebucket(rel, bucket)
*
* Try to squeeze the tuples onto pages occurring earlier in the
* bucket chain in an attempt to free overflow pages. When we start
* the "squeezing", the page from which we start taking tuples (the
* "read" page) is the last bucket in the bucket chain and the page
* onto which we start squeezing tuples (the "write" page) is the
* first page in the bucket chain. The read page works backward and
* the write page works forward; the procedure terminates when the
* read page and write page are the same page.
*
* At completion of this procedure, it is guaranteed that all pages in
* the bucket are nonempty, unless the bucket is totally empty (in
* which case all overflow pages will be freed). The original implementation
* required that to be true on entry as well, but it's a lot easier for
* callers to leave empty overflow pages and let this guy clean it up.
*
* Caller must hold exclusive lock on the target bucket. This allows
* us to safely lock multiple pages in the bucket.
*
* 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,
BufferAccessStrategy bstrategy)
{
BlockNumber wblkno;
BlockNumber rblkno;
Buffer wbuf;
Buffer rbuf;
Page wpage;
Page rpage;
HashPageOpaque wopaque;
HashPageOpaque ropaque;
bool wbuf_dirty;
/*
* start squeezing into the base bucket page.
*/
wblkno = bucket_blkno;
wbuf = _hash_getbuf_with_strategy(rel,
wblkno,
HASH_WRITE,
LH_BUCKET_PAGE,
bstrategy);
wpage = BufferGetPage(wbuf);
wopaque = (HashPageOpaque) PageGetSpecialPointer(wpage);
/*
* if there aren't any overflow pages, there's nothing to squeeze.
*/
if (!BlockNumberIsValid(wopaque->hasho_nextblkno))
{
_hash_relbuf(rel, wbuf);
return;
}
/*
* Find the last page in the bucket chain by starting at the base bucket
* page and working forward. 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.
*/
wbuf_dirty = false;
for (;;)
{
OffsetNumber roffnum;
OffsetNumber maxroffnum;
OffsetNumber deletable[MaxOffsetNumber];
int ndeletable = 0;
/* Scan each tuple in "read" page */
maxroffnum = PageGetMaxOffsetNumber(rpage);
for (roffnum = FirstOffsetNumber;
roffnum <= maxroffnum;
roffnum = OffsetNumberNext(roffnum))
{
IndexTuple itup;
Size itemsz;
itup = (IndexTuple) PageGetItem(rpage,
PageGetItemId(rpage, roffnum));
itemsz = IndexTupleDSize(*itup);
itemsz = MAXALIGN(itemsz);
/*
* Walk up the bucket chain, looking for a page big enough for
* this item. Exit if we reach the read page.
*/
while (PageGetFreeSpace(wpage) < itemsz)
{
Assert(!PageIsEmpty(wpage));
wblkno = wopaque->hasho_nextblkno;
Assert(BlockNumberIsValid(wblkno));
if (wbuf_dirty)
_hash_wrtbuf(rel, wbuf);
else
_hash_relbuf(rel, wbuf);
/* nothing more to do if we reached the read page */
if (rblkno == wblkno)
{
if (ndeletable > 0)
{
/* Delete tuples we already moved off read page */
PageIndexMultiDelete(rpage, deletable, ndeletable);
_hash_wrtbuf(rel, rbuf);
}
else
_hash_relbuf(rel, rbuf);
return;
}
wbuf = _hash_getbuf_with_strategy(rel,
wblkno,
HASH_WRITE,
LH_OVERFLOW_PAGE,
bstrategy);
wpage = BufferGetPage(wbuf);
wopaque = (HashPageOpaque) PageGetSpecialPointer(wpage);
Assert(wopaque->hasho_bucket == bucket);
wbuf_dirty = false;
}
/*
* we have found room so insert 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 instead of
* doing repeated _hash_pgaddtup.)
*/
(void) _hash_pgaddtup(rel, wbuf, itemsz, itup);
wbuf_dirty = true;
/* remember tuple for deletion from "read" page */
deletable[ndeletable++] = roffnum;
}
/*
* 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. However, in that case we are done anyway, so we can
* simply drop the write lock before calling _hash_freeovflpage.
*/
rblkno = ropaque->hasho_prevblkno;
Assert(BlockNumberIsValid(rblkno));
/* are we freeing the page adjacent to wbuf? */
if (rblkno == wblkno)
{
/* yes, so release wbuf lock first */
if (wbuf_dirty)
_hash_wrtbuf(rel, wbuf);
else
_hash_relbuf(rel, wbuf);
/* free this overflow page (releases rbuf) */
_hash_freeovflpage(rel, rbuf, bstrategy);
/* done */
return;
}
/* free this overflow page, then get the previous one */
_hash_freeovflpage(rel, rbuf, bstrategy);
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 */
}
/*
* 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 current 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 &&
H_HAS_DEAD_TUPLES(opaque))
{
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);
}
void
_bt_delitems_delete(Relation rel, Buffer buf,
OffsetNumber *itemnos, int nitems, Relation heapRel)
{
Page page = BufferGetPage(buf);
BTPageOpaque opaque;
Assert(nitems > 0);
/* No ereport(ERROR) until changes are logged */
START_CRIT_SECTION();
/* Fix the page */
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[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 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();
}
/*
* Delete item(s) from a btree page.
*
* 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
spgRedoVacuumRedirect(XLogRecPtr lsn, XLogRecord *record)
{
char *ptr = XLogRecGetData(record);
spgxlogVacuumRedirect *xldata = (spgxlogVacuumRedirect *) ptr;
OffsetNumber *itemToPlaceholder;
Buffer buffer;
Page page;
ptr += sizeof(spgxlogVacuumRedirect);
itemToPlaceholder = (OffsetNumber *) ptr;
if (!(record->xl_info & XLR_BKP_BLOCK_1))
{
buffer = XLogReadBuffer(xldata->node, xldata->blkno, false);
if (BufferIsValid(buffer))
{
page = BufferGetPage(buffer);
if (!XLByteLE(lsn, PageGetLSN(page)))
{
SpGistPageOpaque opaque = SpGistPageGetOpaque(page);
int i;
/* Convert redirect pointers to plain placeholders */
for (i = 0; i < xldata->nToPlaceholder; i++)
{
SpGistDeadTuple dt;
dt = (SpGistDeadTuple) PageGetItem(page,
PageGetItemId(page, itemToPlaceholder[i]));
Assert(dt->tupstate == SPGIST_REDIRECT);
dt->tupstate = SPGIST_PLACEHOLDER;
ItemPointerSetInvalid(&dt->pointer);
}
Assert(opaque->nRedirection >= xldata->nToPlaceholder);
opaque->nRedirection -= xldata->nToPlaceholder;
opaque->nPlaceholder += xldata->nToPlaceholder;
/* Remove placeholder tuples at end of page */
if (xldata->firstPlaceholder != InvalidOffsetNumber)
{
int max = PageGetMaxOffsetNumber(page);
OffsetNumber *toDelete;
toDelete = palloc(sizeof(OffsetNumber) * max);
for (i = xldata->firstPlaceholder; i <= max; i++)
toDelete[i - xldata->firstPlaceholder] = i;
i = max - xldata->firstPlaceholder + 1;
Assert(opaque->nPlaceholder >= i);
opaque->nPlaceholder -= i;
/* The array is sorted, so can use PageIndexMultiDelete */
PageIndexMultiDelete(page, toDelete, i);
pfree(toDelete);
}
PageSetLSN(page, lsn);
PageSetTLI(page, ThisTimeLineID);
MarkBufferDirty(buffer);
}
UnlockReleaseBuffer(buffer);
}
}
}
/*
* _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 */
}
/*
* Bulk deletion of all index entries pointing to a set of heap tuples and
* check invalid tuples left after upgrade.
* The set of target tuples is specified via a callback routine that tells
* whether any given heap tuple (identified by ItemPointer) is being deleted.
*
* Result: a palloc'd struct containing statistical info for VACUUM displays.
*/
IndexBulkDeleteResult *
gistbulkdelete(IndexVacuumInfo *info, IndexBulkDeleteResult *stats,
IndexBulkDeleteCallback callback, void *callback_state)
{
Relation rel = info->index;
GistBDItem *stack,
*ptr;
/* first time through? */
if (stats == NULL)
stats = (IndexBulkDeleteResult *) palloc0(sizeof(IndexBulkDeleteResult));
/* we'll re-count the tuples each time */
stats->estimated_count = false;
stats->num_index_tuples = 0;
stack = (GistBDItem *) palloc0(sizeof(GistBDItem));
stack->blkno = GIST_ROOT_BLKNO;
while (stack)
{
Buffer buffer;
Page page;
OffsetNumber i,
maxoff;
IndexTuple idxtuple;
ItemId iid;
buffer = ReadBufferExtended(rel, MAIN_FORKNUM, stack->blkno,
RBM_NORMAL, info->strategy);
LockBuffer(buffer, GIST_SHARE);
gistcheckpage(rel, buffer);
page = (Page) BufferGetPage(buffer);
if (GistPageIsLeaf(page))
{
OffsetNumber todelete[MaxOffsetNumber];
int ntodelete = 0;
LockBuffer(buffer, GIST_UNLOCK);
LockBuffer(buffer, GIST_EXCLUSIVE);
page = (Page) BufferGetPage(buffer);
if (stack->blkno == GIST_ROOT_BLKNO && !GistPageIsLeaf(page))
{
/* only the root can become non-leaf during relock */
UnlockReleaseBuffer(buffer);
/* one more check */
continue;
}
/*
* check for split proceeded after look at parent, we should check
* it after relock
*/
pushStackIfSplited(page, stack);
/*
* Remove deletable tuples from page
*/
maxoff = PageGetMaxOffsetNumber(page);
for (i = FirstOffsetNumber; i <= maxoff; i = OffsetNumberNext(i))
{
iid = PageGetItemId(page, i);
idxtuple = (IndexTuple) PageGetItem(page, iid);
if (callback(&(idxtuple->t_tid), callback_state))
todelete[ntodelete++] = i;
else
stats->num_index_tuples += 1;
}
stats->tuples_removed += ntodelete;
if (ntodelete)
{
START_CRIT_SECTION();
MarkBufferDirty(buffer);
PageIndexMultiDelete(page, todelete, ntodelete);
GistMarkTuplesDeleted(page);
if (RelationNeedsWAL(rel))
{
XLogRecPtr recptr;
recptr = gistXLogUpdate(buffer,
todelete, ntodelete,
NULL, 0, InvalidBuffer);
PageSetLSN(page, recptr);
}
else
PageSetLSN(page, gistGetFakeLSN(rel));
END_CRIT_SECTION();
}
}
else
{
/* check for split proceeded after look at parent */
pushStackIfSplited(page, stack);
maxoff = PageGetMaxOffsetNumber(page);
for (i = FirstOffsetNumber; i <= maxoff; i = OffsetNumberNext(i))
{
iid = PageGetItemId(page, i);
idxtuple = (IndexTuple) PageGetItem(page, iid);
ptr = (GistBDItem *) palloc(sizeof(GistBDItem));
ptr->blkno = ItemPointerGetBlockNumber(&(idxtuple->t_tid));
ptr->parentlsn = BufferGetLSNAtomic(buffer);
ptr->next = stack->next;
stack->next = ptr;
if (GistTupleIsInvalid(idxtuple))
ereport(LOG,
(errmsg("index \"%s\" contains an inner tuple marked as invalid",
RelationGetRelationName(rel)),
errdetail("This is caused by an incomplete page split at crash recovery before upgrading to PostgreSQL 9.1."),
errhint("Please REINDEX it.")));
}
}
UnlockReleaseBuffer(buffer);
ptr = stack->next;
pfree(stack);
stack = ptr;
vacuum_delay_point();
}
return stats;
}
/*
* _hash_splitbucket -- split 'obucket' into 'obucket' and 'nbucket'
*
* We are splitting a bucket that consists of a base bucket page and zero
* or more overflow (bucket chain) pages. We must relocate tuples that
* belong in the new bucket, and compress out any free space in the old
* bucket.
*
* The caller must hold exclusive locks on both buckets to ensure that
* no one else is trying to access them (see README).
*
* The caller must hold a pin, but no lock, on the metapage buffer.
* The buffer is returned in the same state. (The metapage is only
* touched if it becomes necessary to add or remove overflow pages.)
*/
static void
_hash_splitbucket(Relation rel,
Buffer metabuf,
Bucket obucket,
Bucket nbucket,
BlockNumber start_oblkno,
BlockNumber start_nblkno,
uint32 maxbucket,
uint32 highmask,
uint32 lowmask)
{
BlockNumber oblkno;
BlockNumber nblkno;
Buffer obuf;
Buffer nbuf;
Page opage;
Page npage;
HashPageOpaque oopaque;
HashPageOpaque nopaque;
/*
* It should be okay to simultaneously write-lock pages from each bucket,
* since no one else can be trying to acquire buffer lock on pages of
* either bucket.
*/
oblkno = start_oblkno;
obuf = _hash_getbuf(rel, oblkno, HASH_WRITE, LH_BUCKET_PAGE);
opage = BufferGetPage(obuf);
oopaque = (HashPageOpaque) PageGetSpecialPointer(opage);
nblkno = start_nblkno;
nbuf = _hash_getnewbuf(rel, nblkno, MAIN_FORKNUM);
npage = BufferGetPage(nbuf);
/* initialize the new bucket's primary page */
nopaque = (HashPageOpaque) PageGetSpecialPointer(npage);
nopaque->hasho_prevblkno = InvalidBlockNumber;
nopaque->hasho_nextblkno = InvalidBlockNumber;
nopaque->hasho_bucket = nbucket;
nopaque->hasho_flag = LH_BUCKET_PAGE;
nopaque->hasho_page_id = HASHO_PAGE_ID;
/*
* Partition the tuples in the old bucket between the old bucket and the
* new bucket, advancing along the old bucket's overflow bucket chain and
* adding overflow pages to the new bucket as needed. Outer loop iterates
* once per page in old bucket.
*/
for (;;)
{
OffsetNumber ooffnum;
OffsetNumber omaxoffnum;
OffsetNumber deletable[MaxOffsetNumber];
int ndeletable = 0;
/* Scan each tuple in old page */
omaxoffnum = PageGetMaxOffsetNumber(opage);
for (ooffnum = FirstOffsetNumber;
ooffnum <= omaxoffnum;
ooffnum = OffsetNumberNext(ooffnum))
{
IndexTuple itup;
Size itemsz;
Bucket bucket;
/*
* Fetch the item's hash key (conveniently stored in the item) and
* determine which bucket it now belongs in.
*/
itup = (IndexTuple) PageGetItem(opage,
PageGetItemId(opage, ooffnum));
bucket = _hash_hashkey2bucket(_hash_get_indextuple_hashkey(itup),
maxbucket, highmask, lowmask);
if (bucket == nbucket)
{
/*
* insert the tuple into the new bucket. if it doesn't fit on
* the current page in the new bucket, we must allocate a new
* overflow page and place the tuple on that page instead.
*/
itemsz = IndexTupleDSize(*itup);
itemsz = MAXALIGN(itemsz);
if (PageGetFreeSpace(npage) < itemsz)
{
/* write out nbuf and drop lock, but keep pin */
_hash_chgbufaccess(rel, nbuf, HASH_WRITE, HASH_NOLOCK);
/* chain to a new overflow page */
nbuf = _hash_addovflpage(rel, metabuf, nbuf);
npage = BufferGetPage(nbuf);
/* we don't need nblkno or nopaque within the loop */
}
/*
* Insert tuple on new page, using _hash_pgaddtup to ensure
* correct ordering by hashkey. This is a tad inefficient
* since we may have to shuffle itempointers repeatedly.
* Possible future improvement: accumulate all the items for
* the new page and qsort them before insertion.
*/
(void) _hash_pgaddtup(rel, nbuf, itemsz, itup);
/*
* Mark tuple for deletion from old page.
*/
deletable[ndeletable++] = ooffnum;
}
else
{
/*
* the tuple stays on this page, so nothing to do.
*/
Assert(bucket == obucket);
}
}
oblkno = oopaque->hasho_nextblkno;
/*
* Done scanning this old page. If we moved any tuples, delete them
* from the old page.
*/
if (ndeletable > 0)
{
PageIndexMultiDelete(opage, deletable, ndeletable);
_hash_wrtbuf(rel, obuf);
}
else
_hash_relbuf(rel, obuf);
/* Exit loop if no more overflow pages in old bucket */
if (!BlockNumberIsValid(oblkno))
break;
/* Else, advance to next old page */
obuf = _hash_getbuf(rel, oblkno, HASH_WRITE, LH_OVERFLOW_PAGE);
opage = BufferGetPage(obuf);
oopaque = (HashPageOpaque) PageGetSpecialPointer(opage);
}
/*
* We're at the end of the old bucket chain, so we're done partitioning
* the tuples. Before quitting, call _hash_squeezebucket to ensure the
* tuples remaining in the old bucket (including the overflow pages) are
* packed as tightly as possible. The new bucket is already tight.
*/
_hash_wrtbuf(rel, nbuf);
_hash_squeezebucket(rel, obucket, start_oblkno, NULL);
}
/*
* redo any page update (except page split)
*/
static void
gistRedoPageUpdateRecord(XLogReaderState *record)
{
XLogRecPtr lsn = record->EndRecPtr;
gistxlogPageUpdate *xldata = (gistxlogPageUpdate *) XLogRecGetData(record);
Buffer buffer;
Page page;
if (XLogReadBufferForRedo(record, 0, &buffer) == BLK_NEEDS_REDO)
{
char *begin;
char *data;
Size datalen;
int ninserted = 0;
data = begin = XLogRecGetBlockData(record, 0, &datalen);
page = (Page) BufferGetPage(buffer);
if (xldata->ntodelete == 1 && xldata->ntoinsert == 1)
{
/*
* When replacing one tuple with one other tuple, we must use
* PageIndexTupleOverwrite for consistency with gistplacetopage.
*/
OffsetNumber offnum = *((OffsetNumber *) data);
IndexTuple itup;
Size itupsize;
data += sizeof(OffsetNumber);
itup = (IndexTuple) data;
itupsize = IndexTupleSize(itup);
if (!PageIndexTupleOverwrite(page, offnum, (Item) itup, itupsize))
elog(ERROR, "failed to add item to GiST index page, size %d bytes",
(int) itupsize);
data += itupsize;
/* should be nothing left after consuming 1 tuple */
Assert(data - begin == datalen);
/* update insertion count for assert check below */
ninserted++;
}
else if (xldata->ntodelete > 0)
{
/* Otherwise, delete old tuples if any */
OffsetNumber *todelete = (OffsetNumber *) data;
data += sizeof(OffsetNumber) * xldata->ntodelete;
PageIndexMultiDelete(page, todelete, xldata->ntodelete);
if (GistPageIsLeaf(page))
GistMarkTuplesDeleted(page);
}
/* Add new tuples if any */
if (data - begin < datalen)
{
OffsetNumber off = (PageIsEmpty(page)) ? FirstOffsetNumber :
OffsetNumberNext(PageGetMaxOffsetNumber(page));
while (data - begin < datalen)
{
IndexTuple itup = (IndexTuple) data;
Size sz = IndexTupleSize(itup);
OffsetNumber l;
data += sz;
l = PageAddItem(page, (Item) itup, sz, off, false, false);
if (l == InvalidOffsetNumber)
elog(ERROR, "failed to add item to GiST index page, size %d bytes",
(int) sz);
off++;
ninserted++;
}
}
/* Check that XLOG record contained expected number of tuples */
Assert(ninserted == xldata->ntoinsert);
PageSetLSN(page, lsn);
MarkBufferDirty(buffer);
}
/*
* Fix follow-right data on left child page
*
* This must be done while still holding the lock on the target page. Note
* that even if the target page no longer exists, we still attempt to
* replay the change on the child page.
*/
if (XLogRecHasBlockRef(record, 1))
gistRedoClearFollowRight(record, 1);
if (BufferIsValid(buffer))
UnlockReleaseBuffer(buffer);
}
/*
* 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 curr_page_dirty = 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)
{
PageIndexMultiDelete(page, deletable, ndeletable);
bucket_dirty = true;
curr_page_dirty = true;
}
/* 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 (curr_page_dirty)
{
if (retain_pin)
_hash_chgbufaccess(rel, buf, HASH_WRITE, HASH_NOLOCK);
else
_hash_wrtbuf(rel, buf);
curr_page_dirty = false;
}
else if (retain_pin)
_hash_chgbufaccess(rel, buf, HASH_READ, HASH_NOLOCK);
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);
_hash_chgbufaccess(rel, bucket_buf, HASH_NOLOCK, HASH_WRITE);
}
/*
* 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);
bucket_opaque->hasho_flag &= ~LH_BUCKET_NEEDS_SPLIT_CLEANUP;
}
/*
* 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
_hash_chgbufaccess(rel, bucket_buf, HASH_WRITE, HASH_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.
*
* 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;
HashMetaPage metap;
HashMetaPageData local_metapage;
tuples_removed = 0;
num_index_tuples = 0;
/*
* Read the metapage to fetch original bucket and tuple counts. Also, we
* keep a copy of the last-seen metapage so that we can use its
* hashm_spares[] values to compute bucket page addresses. This is a bit
* hokey but perfectly safe, since the interesting entries in the spares
* array cannot change under us; and it beats rereading the metapage for
* each bucket.
*/
metabuf = _hash_getbuf(rel, HASH_METAPAGE, HASH_READ, LH_META_PAGE);
metap = HashPageGetMeta(BufferGetPage(metabuf));
orig_maxbucket = metap->hashm_maxbucket;
orig_ntuples = metap->hashm_ntuples;
memcpy(&local_metapage, metap, sizeof(local_metapage));
_hash_relbuf(rel, metabuf);
/* Scan the buckets that we know exist */
cur_bucket = 0;
cur_maxbucket = orig_maxbucket;
loop_top:
while (cur_bucket <= cur_maxbucket)
{
BlockNumber bucket_blkno;
BlockNumber blkno;
bool bucket_dirty = false;
/* Get address of bucket's start page */
bucket_blkno = BUCKET_TO_BLKNO(&local_metapage, cur_bucket);
/* Exclusive-lock the bucket so we can shrink it */
_hash_getlock(rel, bucket_blkno, HASH_EXCLUSIVE);
/* Shouldn't have any active scans locally, either */
if (_hash_has_active_scan(rel, cur_bucket))
elog(ERROR, "hash index has active scan during VACUUM");
/* Scan each page in bucket */
blkno = bucket_blkno;
while (BlockNumberIsValid(blkno))
{
Buffer buf;
Page page;
HashPageOpaque opaque;
OffsetNumber offno;
OffsetNumber maxoffno;
OffsetNumber deletable[MaxOffsetNumber];
int ndeletable = 0;
vacuum_delay_point();
buf = _hash_getbuf_with_strategy(rel, blkno, HASH_WRITE,
LH_BUCKET_PAGE | LH_OVERFLOW_PAGE,
info->strategy);
page = BufferGetPage(buf);
opaque = (HashPageOpaque) PageGetSpecialPointer(page);
Assert(opaque->hasho_bucket == cur_bucket);
/* Scan each tuple in page */
maxoffno = PageGetMaxOffsetNumber(page);
for (offno = FirstOffsetNumber;
offno <= maxoffno;
offno = OffsetNumberNext(offno))
{
IndexTuple itup;
ItemPointer htup;
itup = (IndexTuple) PageGetItem(page,
PageGetItemId(page, offno));
htup = &(itup->t_tid);
if (callback(htup, callback_state))
{
/* mark the item for deletion */
deletable[ndeletable++] = offno;
tuples_removed += 1;
}
else
num_index_tuples += 1;
}
/*
* Apply deletions and write page if needed, advance to next page.
*/
blkno = opaque->hasho_nextblkno;
if (ndeletable > 0)
{
PageIndexMultiDelete(page, deletable, ndeletable);
_hash_wrtbuf(rel, buf);
bucket_dirty = true;
}
else
_hash_relbuf(rel, buf);
}
/* If we deleted anything, try to compact free space */
if (bucket_dirty)
_hash_squeezebucket(rel, cur_bucket, bucket_blkno,
info->strategy);
/* Release bucket lock */
_hash_droplock(rel, bucket_blkno, HASH_EXCLUSIVE);
/* Advance to next bucket */
cur_bucket++;
}
/* Write-lock metapage and check for split since we started */
metabuf = _hash_getbuf(rel, HASH_METAPAGE, HASH_WRITE, LH_META_PAGE);
metap = HashPageGetMeta(BufferGetPage(metabuf));
if (cur_maxbucket != metap->hashm_maxbucket)
{
/* There's been a split, so process the additional bucket(s) */
cur_maxbucket = metap->hashm_maxbucket;
memcpy(&local_metapage, metap, sizeof(local_metapage));
_hash_relbuf(rel, metabuf);
goto loop_top;
}
/* Okay, we're really done. Update tuple count in metapage. */
if (orig_maxbucket == metap->hashm_maxbucket &&
orig_ntuples == metap->hashm_ntuples)
{
/*
* No one has split or inserted anything since start of scan, so
* believe our count as gospel.
*/
metap->hashm_ntuples = num_index_tuples;
}
else
{
/*
* Otherwise, our count is untrustworthy since we may have
* double-scanned tuples in split buckets. Proceed by dead-reckoning.
* (Note: we still return estimated_count = false, because using this
* count is better than not updating reltuples at all.)
*/
if (metap->hashm_ntuples > tuples_removed)
metap->hashm_ntuples -= tuples_removed;
else
metap->hashm_ntuples = 0;
num_index_tuples = metap->hashm_ntuples;
}
_hash_wrtbuf(rel, metabuf);
/* return statistics */
if (stats == NULL)
stats = (IndexBulkDeleteResult *) palloc0(sizeof(IndexBulkDeleteResult));
stats->estimated_count = false;
stats->num_index_tuples = num_index_tuples;
stats->tuples_removed += tuples_removed;
/* hashvacuumcleanup will fill in num_pages */
return stats;
}
/*
* replay delete operation in hash index to remove
* tuples marked as DEAD during index tuple insertion.
*/
static void
hash_xlog_vacuum_one_page(XLogReaderState *record)
{
XLogRecPtr lsn = record->EndRecPtr;
xl_hash_vacuum_one_page *xldata;
Buffer buffer;
Buffer metabuf;
Page page;
XLogRedoAction action;
HashPageOpaque pageopaque;
xldata = (xl_hash_vacuum_one_page *) XLogRecGetData(record);
/*
* If we have any conflict processing to do, it must happen before we
* update the page.
*
* Hash index records that are marked as LP_DEAD and being removed during
* hash index tuple insertion can conflict with standby queries. You might
* think that vacuum records would conflict as well, but we've handled
* that already. XLOG_HEAP2_CLEANUP_INFO records provide the highest xid
* cleaned by the vacuum of the heap and so we can resolve any conflicts
* just once when that arrives. After that we know that no conflicts
* exist from individual hash index vacuum records on that index.
*/
if (InHotStandby)
{
TransactionId latestRemovedXid =
hash_xlog_vacuum_get_latestRemovedXid(record);
RelFileNode rnode;
XLogRecGetBlockTag(record, 0, &rnode, NULL, NULL);
ResolveRecoveryConflictWithSnapshot(latestRemovedXid, rnode);
}
action = XLogReadBufferForRedoExtended(record, 0, RBM_NORMAL, true, &buffer);
if (action == BLK_NEEDS_REDO)
{
char *ptr;
Size len;
ptr = XLogRecGetBlockData(record, 0, &len);
page = (Page) BufferGetPage(buffer);
if (len > 0)
{
OffsetNumber *unused;
OffsetNumber *unend;
unused = (OffsetNumber *) ptr;
unend = (OffsetNumber *) ((char *) ptr + len);
if ((unend - unused) > 0)
PageIndexMultiDelete(page, unused, unend - unused);
}
/*
* Mark the page as not containing any LP_DEAD items. See comments
* in _hash_vacuum_one_page() for details.
*/
pageopaque = (HashPageOpaque) PageGetSpecialPointer(page);
pageopaque->hasho_flag &= ~LH_PAGE_HAS_DEAD_TUPLES;
PageSetLSN(page, lsn);
MarkBufferDirty(buffer);
}
if (BufferIsValid(buffer))
UnlockReleaseBuffer(buffer);
if (XLogReadBufferForRedo(record, 1, &metabuf) == BLK_NEEDS_REDO)
{
Page metapage;
HashMetaPage metap;
metapage = BufferGetPage(metabuf);
metap = HashPageGetMeta(metapage);
metap->hashm_ntuples -= xldata->ntuples;
PageSetLSN(metapage, lsn);
MarkBufferDirty(metabuf);
}
if (BufferIsValid(metabuf))
UnlockReleaseBuffer(metabuf);
}
